User's Guide

Distance 6.0 Beta 5

Research Unit for Wildlife Population Assessment

Contents
Chapter 1 - Introduction 1
Welcome.................................................................................................................................... 1
Conventions in This Documentation ........................................................................... 2
Where to Go Next........................................................................................................ 2
Citation for Distance.................................................................................................... 3
Distance Sampling Reference Books........................................................................... 3
Staying in Touch........................................................................................................................ 3
Distance-sampling Email List ..................................................................................... 3
Program Distance Web Site......................................................................................... 4
Sending Suggestions and Reporting Problems ............................................................ 4

Chapter 2 - About Distance 5

What is Distance? ...................................................................................................................... 5
Use Agreement .......................................................................................................................... 5
Sponsors..................................................................................................................................... 5
Distance Development Team..................................................................................................... 6
Acknowledgements.................................................................................................................... 6
History of Distance .................................................................................................................... 7
New in Distance......................................................................................................................... 8
New Features of Distance 6.0...................................................................................... 8
New Features of Distance 5.0...................................................................................... 8
New Features of Distance 4.1...................................................................................... 8
New Features of Distance 4.0...................................................................................... 8
New Features of Distance 3.5.................................................................................... 10
New Features Planned for Future Versions ............................................................... 12

Chapter 3 - Getting Started 13

Objective.................................................................................................................................. 13
Example 1: Using Distance to Analyze Simple Data............................................................... 13
Example 1 - Preparing the data for import ................................................................ 13
Example 1 - Creating the Distance project ................................................................ 14
Example 1 - Importing the data ................................................................................. 15
Example 1 - Studying the data in Distance................................................................ 16
Example 1 – Running the first analysis ..................................................................... 16
Example 1 - Creating a new analysis......................................................................... 17
Example 1 - Further investigations............................................................................ 18
Example 2: More Complex Data Import.................................................................................. 19
Example 2 - Preparing to import the data.................................................................. 19
Example 2 - Importing the data ................................................................................. 20
Example 2 - Analysis................................................................................................. 20
Example 3: Using Distance to Design a Survey ...................................................................... 21
Example 3 - Preparing the data for import ................................................................ 21
Example 3 - Creating the Distance project ................................................................ 22
Example 3 - Importing the Geographic Data............................................................. 22
Example 3 - Checking the Data on a Map................................................................. 22
Example 3 - Adding a Coverage Layer ..................................................................... 23
Example 3 - Creating a Survey Design...................................................................... 24
Example 3 - Automated Generation of New Surveys................................................ 24

User's Guide Distance 6.0 Beta 5 Contents • iii

 Example 3 - Design Statistics.................................................................................... 25
Example 3 - Further Investigations ........................................................................... 26
Example 4 - A Second Survey Design Project ........................................................................ 26
Example 4 - Opening the Mexico Project in Distance............................................... 26
Example 4 - Reviewing the Project Properties .......................................................... 26
Example 4 - Examining the Data............................................................................... 27
Example 4 - Creating a New Design ......................................................................... 28
Example 4 - Automated Generation of New Surveys................................................ 29
Example 4 - Design Statistics.................................................................................... 31
Example 4 - Further Investigations ........................................................................... 31
Sample Projects ....................................................................................................................... 31

Chapter 4 - Distance Projects 33

Introduction to Distance Projects............................................................................................. 33
Creating a New Project............................................................................................................ 34
Using an Existing Project as a Template ................................................................... 35
Opening an Existing Project .................................................................................................... 36
Saving and Backing Up a Project ............................................................................................ 36
Saving Projects .......................................................................................................... 36
Backing up Projects................................................................................................... 36
Exporting, Transporting and Archiving Projects ..................................................................... 38
Viewing and Editing Project Properties................................................................................... 39
Compacting a Project............................................................................................................... 39
Importing from Previous Versions of Distance ....................................................................... 39
Importing Distance 4 and 5 Projects.......................................................................... 39
Importing Distance 3.5 Projects ................................................................................ 40
Importing Distance 2.2 - 3.0 Command Files............................................................ 40

Chapter 5 - Data in Distance 41

Data Structure .......................................................................................................................... 41
Data Layers................................................................................................................ 41
Data Fields................................................................................................................. 43
Changing the Data Structure...................................................................................... 45
Getting Data Into Distance ...................................................................................................... 45
Data Entry ................................................................................................................. 46
Data Import................................................................................................................ 46
Geographic (GIS) Data ............................................................................................................ 52
Viewing and Manipulating Geographic Data ............................................................ 52
Coordinate Systems and Projections ......................................................................... 53
GIS Data Format ....................................................................................................... 56
Importing Existing GIS Data..................................................................................... 57
Advanced Data Topics............................................................................................................. 61
Linking to Data From Other Databases ..................................................................... 61

Chapter 6 - Survey Design in Distance 63

Introduction to Survey Design in Distance .............................................................................. 63
Design Classes Available in Distance ....................................................................... 64
Survey Design Concepts.......................................................................................................... 65
Concept: Coverage Probability.................................................................................. 65
Concept: Edge Effects ............................................................................................... 65
Concept: Zigzag Sampling Designs .......................................................................... 66
Setting Up a New Project for Survey Design .......................................................................... 68

Chapter 7 - Analysis in Distance 71

Introduction to Analysis in Distance........................................................................................ 71
Introduction to the Analysis Browser ........................................................................ 71
Introduction to Analysis Details Windows................................................................ 72

iv • Contents User's Guide Distance 6.0 Beta 5

 Analysis Components .............................................................................................................. 72
Data Filters, Model Definitions and Survey Objects................................................. 72
Working with Data Filters and Model Definitions .................................................... 74
Working with Surveys during Analysis..................................................................... 78
Analysis Engines ..................................................................................................................... 80
Conventional Distance Sampling (CDS) Engine....................................................... 80
Multiple Covariate Distance Sampling (MCDS) Engine........................................... 81
Mark Recapture Distance Sampling (MRDS) Engine ............................................... 81
Density Surface Modelling (DSM) Engine ............................................................... 81
Running Analyses.................................................................................................................... 81
Locking the Data Sheet ............................................................................................. 82
Cleaning the Temp folder.......................................................................................... 83
R Statistical Software .............................................................................................................. 83
Installing and Configuring R ..................................................................................... 84
Updating the Version of R That Distance Uses......................................................... 84
Contents of the R folder ............................................................................................ 84
Images produced by R ............................................................................................... 84

Chapter 8 - Conventional Distance Sampling Analysis 87

Introduction to CDS Analysis.................................................................................................. 87
Setting up a Project for CDS Analysis..................................................................................... 88
CDS Analysis Guidelines ........................................................................................................ 88
Output from CDS Analyses ..................................................................................................... 89
CDS Results Details Listing ...................................................................................... 90
About CDS Detection Function Formulae ................................................................ 95
CDS Analysis Browser Results ................................................................................. 98
Exporting CDS Results.............................................................................................. 98
Miscellaneous CDS Analysis Topics..................................................................................... 100
Interval (Binned/Grouped) Data .............................................................................. 100
Missing Data in CDS Analysis................................................................................ 101
Clusters of Objects .................................................................................................. 102
Stratification and Post-stratification ........................................................................ 103
Variance Estimation in CDS.................................................................................... 107
Multipliers in CDS Analysis ................................................................................... 108
Model Averaging in CDS Analysis ......................................................................... 111
Sample Definition in CDS Analysis ........................................................................ 112
Unknown Study Area Size ...................................................................................... 112
Restricting Inference to Density or Abundance in the Covered Region in CDS
Analysis................................................................................................................... 112
Analysis of Data from a Single Transect in CDS .................................................... 113
Running CDS Analyses From Outside Distance ..................................................... 114

Chapter 9 - Multiple Covariates Distance Sampling Analysis 115

Introduction to MCDS Analysis ............................................................................................ 115
Introducing the MCDS Engine ................................................................................ 115
Setting up a Project for MCDS Analysis ............................................................................... 116
Defining MCDS Models........................................................................................................ 116
MCDS Analysis Guidelines................................................................................................... 120
Choosing Covariates to Include in the Model ......................................................... 120
Specifying the Model .............................................................................................. 121
Truncation for MCDS Analyses .............................................................................. 122
Output from MCDS Analyses................................................................................................ 122
MCDS Results Details Listing ................................................................................ 122
Exporting MCDS Results ........................................................................................ 124
Miscellaneous MCDS Analysis Topics ................................................................................. 124
Missing Data in MCDS Analysis ............................................................................ 124
Stratification and Post-stratification in MCDS........................................................ 125
Running MCDS Analyses from Outside Distance .................................................. 125
Analysis of Double Observer Data with the MCDS Engine.................................... 125

User's Guide Distance 6.0 Beta 5 Contents • v

 Chapter 10 - Mark Recapture Distance Sampling 127
Introduction to Mark Recapture Distance Sampling.............................................................. 127
Setting up a Project for MRDS Analysis ............................................................................... 129
Setting up Your Data for MRDS Analysis .............................................................. 129
Defining MRDS Models........................................................................................................ 130
Introduction to MRDS Models ................................................................................ 130
DS and MR Models ................................................................................................. 131
Specifying DS and MR Model Formulae ................................................................ 131
MRDS Analysis Guidelines................................................................................................... 134
Output from MRDS Analyses................................................................................................ 135
MRDS Results Details Listing ................................................................................ 135
MRDS Analysis Browser Results............................................................................ 136
Exporting MRDS Results ........................................................................................ 136
Miscellaneous MRDS Analysis Topics ................................................................................. 137
Interval Data in MRDS............................................................................................ 137
Clusters of Objects in MRDS .................................................................................. 137
Stratification and Post-stratification in MRDS........................................................ 137
Variance Estimation in MRDS ................................................................................ 137
Multipliers in MRDS Analysis ................................................................................ 138
Model Averaging in MRDS Analysis ..................................................................... 138
Sample Definition in MRDS Analysis .................................................................... 139
Using a Previously Fitted Detection Function to Estimate Density in MRDS ........ 139
Restricting Inference to Density or Abundance in the Covered Region in MRDS
Analysis................................................................................................................... 140
Running the MRDS Analysis Engine from Outside Distance ................................. 140
Installing an Updated Version of the MRDS Engine .............................................. 141
Checking Which Version of the MRDS Engine is Being Used............................... 141
Fine-tuning an MRDS Analysis .............................................................................. 142
Single Observer Configuration in the MRDS Engine ............................................. 143

Chapter 11 – Density Surface Modelling 145

Introduction to Density Surface Modelling ........................................................................... 145
Setting up a Project for DSM Analysis.................................................................................. 146
Setting up Your Data for DSM Analysis................................................................. 147
Defining DSM Models .......................................................................................................... 151
Introduction to DSM Models................................................................................... 151
Specifying DSM Model Formulae .......................................................................... 151
DSM Analysis Guidelines ..................................................................................................... 152
Density surface model ............................................................................................. 152
Prediction of abundance to unsurveyed areas.......................................................... 153
Variance estimation using parametric bootstrap...................................................... 154
Output from DSM Analyses .................................................................................................. 154
DSM Results Details Listing ................................................................................... 155
DSM Analysis Browser Results .............................................................................. 156
Exporting DSM Results........................................................................................... 156
Miscellaneous DSM Analysis Topics.................................................................................... 156
Clusters of Objects in DSM..................................................................................... 156
Stratification and Post-stratification in DSM .......................................................... 157
Running the DSM Analysis Engine from Outside Distance.................................... 157
Installing an Updated Version of the DSM Engine ................................................. 158
Checking Which Version of the DSM Engine is Being Used ................................. 158
Fine-tuning a DSM Analysis ................................................................................... 159

Chapter 12 - Troubleshooting 161

Known Problems ................................................................................................................... 161
Internal Errors in the Interface............................................................................................... 161
Problems with the Analysis Engines...................................................................................... 162
Errors and Warnings in the CDS and MCDS Analysis Engines.............................. 162

vi • Contents User's Guide Distance 6.0 Beta 5

 Problems with the MRDS Engine ........................................................................... 162
Stopping an Analysis............................................................................................... 163
GIS Problems......................................................................................................................... 163
Recovering from Unexpected Program Exit.......................................................................... 164
Fixing a corrupted project ....................................................................................... 164

Appendix - Program Reference 165

Introduction to Program Reference........................................................................................ 165
Setup Project Wizard ............................................................................................................. 165
Setup for Analyzing a Survey.................................................................................. 166
Setup for Designing Surveys ................................................................................... 169
Use Another Project as Template ............................................................................ 169
Import from Previous Version of Distance.............................................................. 169
Data Entry Wizard ................................................................................................................. 170
Global Layer Wizard Page ...................................................................................... 171
Stratum Layer Wizard Page..................................................................................... 171
Sample Layer Wizard Page ..................................................................................... 171
Observation Layer Wizard Page.............................................................................. 171
Finished Data Entry Wizard Page ........................................................................... 172
Import Data Wizard ............................................................................................................... 172
Data Source Wizard Page........................................................................................ 172
Data Destination Wizard Page................................................................................. 173
Data File Format Wizard Page ................................................................................ 174
Data File Structure Wizard Page ............................................................................. 174
Finished Import Data Wizard .................................................................................. 176
Troubleshooting the Import Data Wizard................................................................ 176
Project Properties Dialog ....................................................................................................... 177
General Project Properties Tab................................................................................ 177
Geographic Project Properties Tab.......................................................................... 178
Preferences Dialog................................................................................................................. 178
General Preferences Tab.......................................................................................... 178
Geographic Preferences Tab.................................................................................... 180
Survey Design Preferences Tab............................................................................... 180
Analysis Preferences Tab ........................................................................................ 181
Project Browser ..................................................................................................................... 183
Data Explorer......................................................................................................................... 183
Data Layers Viewer................................................................................................. 185
Data Sheet ............................................................................................................... 187
Map Browser ......................................................................................................................... 192
Design Browser ..................................................................................................................... 193
Survey Browser ..................................................................................................................... 195
Analysis Browser................................................................................................................... 197
Map Window ......................................................................................................................... 199
Design Details Window ......................................................................................................... 201
Design Details Inputs Tab ....................................................................................... 201
Design Details Log Tab........................................................................................... 202
Design Details Results Tab ..................................................................................... 203
Survey Details Window ......................................................................................................... 209
Survey Details Inputs Tab ....................................................................................... 209
Survey Details Log Tab........................................................................................... 210
Survey Details Results Tab ..................................................................................... 210
Analysis Details Window ...................................................................................................... 210
Analysis Details Inputs Tab..................................................................................... 210
Analysis Details Log Tab ........................................................................................ 213
Analysis Details Results Tab................................................................................... 214
Design Properties Dialog ....................................................................................................... 215
General Design Properties Tab................................................................................ 216
Coverage Probability Design Properties Tab .......................................................... 217
Sampler Design Properties Tab ............................................................................... 218

User's Guide Distance 6.0 Beta 5 Contents • vii

 Effort Allocation Design Properties Tab ................................................................. 219
Survey Properties Dialog ....................................................................................................... 230
Survey Methods Survey Properties Tab .................................................................. 230
Data Layers Survey Properties Tab ......................................................................... 231
Data Fields Survey Properties Tab .......................................................................... 231
Data Filter Properties Dialog ................................................................................................. 231
Data Selection Tab .................................................................................................. 232
Intervals Tab............................................................................................................ 233
Truncation Tab ........................................................................................................ 235
Units Tab ................................................................................................................. 236
Model Definition Properties Dialog....................................................................................... 236
Model Definition Properties - CDS and MCDS ...................................................... 237
Model Definition Properties - MRDS...................................................................... 250
Model Definition Properties – DSM ....................................................................... 253
Analysis Components Window ............................................................................................. 253
Other Windows...................................................................................................................... 254
About Distance Dialog ............................................................................................ 254
Export Project Dialog .............................................................................................. 254
Projection Parameters Dialog .................................................................................. 255
Create New Layer Dialog........................................................................................ 255
Grid Properties Dialog............................................................................................. 255
Insert or Append Field Dialog ................................................................................. 256
Data Layer Properties Dialog .................................................................................. 256
Shape Properties Dialog .......................................................................................... 257
New Coordinate System Dialog .............................................................................. 257
Column Manager Dialog ......................................................................................... 258
Arrange Sets Dialog ................................................................................................ 258
Map Properties Dialog............................................................................................. 258
Add Map Layer Dialog............................................................................................ 258
Run Design Dialog .................................................................................................. 259
Confirm Change Dialog .......................................................................................... 259
R Image Properties Dialog ...................................................................................... 259
Data Selection Zoom Dialog ................................................................................... 260

Appendix - Inside Distance 261

Introduction to Inside Distance Appendix ............................................................................. 261
Distance components ............................................................................................................. 261
The Distance 6 Database API ................................................................................................ 261
Data File Reference ............................................................................................................... 262
How Distance Stores Data....................................................................................... 262
Linking to External Data from Distdata.mdb .......................................................... 267
Valid Names ............................................................................................................ 276
Miscellaneous topics.............................................................................................................. 278
Random number generation..................................................................................... 278

Appendix - MCDS Engine Reference 279

Introduction to MCDS Engine Reference.............................................................................. 279
Some history .......................................................................................................................... 279
Running the MCDS engine.................................................................................................... 280
MCDS Command Language.................................................................................................. 281
Header Section ........................................................................................................ 282
Options Section ....................................................................................................... 283
Data section ............................................................................................................. 291
Estimate section....................................................................................................... 293
MCDS Engine Required Data Format ................................................................................... 308
Output From the MCDS Engine ............................................................................................ 309
MCDS Engine Command Line Output.................................................................... 310
MCDS Engine Output File ...................................................................................... 310
MCDS Engine Log File........................................................................................... 311

viii • Contents User's Guide Distance 6.0 Beta 5

 MCDS Engine Stats File ......................................................................................... 311
MCDS Engine Plot File........................................................................................... 312
MCDS Engine Bootstrap File.................................................................................. 313
MCDS Engine Bootstrap Progress File ................................................................... 313
MCDS Engine Limitations .................................................................................................... 314
MCDS Engine Fitting Algorithms ......................................................................................... 314
MCDS Engine Error and Warning Messages ........................................................................ 315
MCDS Engine Warning Messages .......................................................................... 315
MCDS Engine Error Messages................................................................................ 319
MCDS Engine Internal Error Messages .................................................................. 323
Changes in MCDS Engine Since Distance 2.2 ...................................................................... 324

Appendix – HT estimation of density when probability of coverage is

unequal 327
Review of the Horvitz-Thompson-like estimator .................................................................. 327
Setting up a Project for DHT Analysis .................................................................................. 328
Output from DHT Analyses................................................................................................... 329
DHT Results Details Listing ................................................................................... 329
Exporting DHT Results ........................................................................................... 329
Miscellaneous DHT Analysis Topics .................................................................................... 330
Clusters of Objects in DHT ..................................................................................... 330
Stratification and Post-stratification in DHT........................................................... 330
Running the DHT Analysis Engine from Outside Distance .................................... 330
Checking Which Version of the DHT Engine is Being Used.................................. 331

Bibliography 333

Glossary of Terms 335

Index 343

User's Guide Distance 6.0 Beta 5 Contents • ix

Chapter 1 - Introduction

Welcome

This is a beta release of Distance 6.0, and the contents of

the manual are still under construction!

Welcome to Distance 6.0!

Distance software allows you to design and analyze distance sampling surveys,
where the aim is to estimate the density and abundance of a biological
population. The survey methodologies covered include line transects, point
transects (variable circular plots), cue counts and trapping webs.
The aim of this documentation is to tell you how to use the program, given that
you already understand the concepts. Conventional distance sampling methods
are described in detail by Buckland et al. (1993, 2001), and advanced methods
are described by Buckland et al. (2004). These books are essential companions
to the software. (See Distance Sampling Reference Books).
This documentation is available in two formats:
• a standard help version, in Microsoft HTML-Help format. This is
what you see when you press the F1 button, or choose Help |
Contents and Index from the main Distance menu.
• a print-ready version, in Adobe Acrobat (.pdf) format. You can
view this version by choosing Help | Online Manuals | Users
Guide from the main Distance menu, or from the Windows Start
Menu by choosing Programs | Distance | Users’ Guide.
For the online version to display properly, you need HTML-Help software
installed on your computer, and for the print-ready version Adobe Acrobat must
be installed. See the System Requirements part of the ReadMe.txt file for more
details.
The documentation is divided into two main parts: Users Guide and Program
Reference.
The Users Guide is designed to be read (or at least scanned!) page by page. It is
divided into chapters, each of which describes a different aspect of the program.
It is probably easiest to read the Users Guide in Adobe Acrobat format, because
some of the sections are quite long.
The Program Reference can be referred to whenever you need to know how to
use a specific part of the interface. It is probably easiest to access by pressing F1

User's Guide Distance 6.0 Beta 5 Chapter 1 - Introduction • 1

 from within Distance. This automatically takes you to a page of information
about the window you are currently viewing. The Program Reference is
currently included as an appendix of the Users Guide.
There are also two other appendices. The first gives an overview of the internal
workings of the Distance software, and contains some reference material for
users wishing to tinker “under the hood” of the program. The second is the
command language reference for the CDS and MCDS analysis engines.
In conjunction with the documentation, you may like to try out the sample
projects. These are located in the “Sample Projects” folder, below the Distance
program folder (usually “C:\Program Files\Distance 5”). These projects are
referred to at various points throughout the text, and they are used as the basis
for the tutorials in Chapter 3. An overview of the sample projects is also given
at the end of Chapter 3.

Conventions in This Documentation

Bold text indicates a reference to an element of the user interface of Distance
(for example the Project Browser, or New Analysis button), and is also used
occasionally for emphasis.

Indicates a paragraph in the text that makes an important point

relevant to the subject being discussed.

Indicates a paragraph that contains advice about how to do something

in Distance. Often the advice highlights a shortcut of some kind, or explains
how to do something that is quite complicated.

Indicates a paragraph that contains some incidental information.

Indicates a topic that contains some advanced material.

These topics can be skipped on first reading. You should come back to them
when you are familiar with the basics of Distance.

Placed above or alongside text that contains an important

warning. You should definitely read this text!

Occurs before text that describes a feature of Distance that was not
present in previous versions. We hope it helps users of the old software
familiarize themselves with the new version.
When describing keyboard actions, CTRL is short for the control key. For
example CTRL-X means hold the control and x keys at the same time.
When describing menu selection the symbol | means “and then select” – for
example Help | About Distance… means select the Help menu and then
select About Distance….

Where to Go Next
We recommend that everyone start by following the guided tour in Chapter 3.
New users should then read the Users Guide Chapters 4-7 – if you do this you
will find using the program much less confusing! People familiar with Distance
3.5 or later should check out the section New in Distance (in Chapter 2) and
should at least glance at Chapters 4-7, as required.
Advanced users will want to check out Chapters 6, 9, 10 and the appendices.
If you have yet to install Distance, you should read the release notes in the file
ReadMe.rtf that accompanies the Distance setup program.

2 • Chapter 1 - Introduction User's Guide Distance 6.0 Beta 5

 Before you start using Distance, please make sure you have read and agreed to
the Use Agreement. If you wish to cite Distance in your next Nature paper,
please use the Citation for Distance.

Citation for Distance

The suggested citation for both the manual and software is:
Thomas, L., Laake, J.L., Rexstad, E., Strindberg, S., Marques, F.F.C., Buckland,
S.T., Borchers, D.L., Anderson, D.R., Burnham, K.P., Burt, M.L., Hedley, S.L.,
Pollard, J.H., Bishop, J.R.B. and Marques, T.A. 2006. Distance 6.0. Release
“x”1. Research Unit for Wildlife Population Assessment, University of St.
Andrews, UK. http://www.ruwpa.st-and.ac.uk/distance/
1
Instead of “x”, substitute the release number of the software you are using –
e.g., “Distance 6.0 Release 1”. You can find the release number by selecting
Help | About Distance... from the main menu. It is important to include the
release number in the citation, because results may not be absolutely identical
between releases.

Distance Sampling Reference Books

The standard reference for conventional distance sampling methods is:
Buckland, S.T., Anderson, D.R., Burnham, K.P., Laake, J.L., Borchers, D.L. and
Thomas, L. 2001. Introduction to Distance Sampling. Oxford University Press,
London.
The book describes the methods in detail, as well as covering aspects of survey
design and giving several worked examples. It updates the previous standard
work:
Buckland, S.T., Anderson, D.R., Burnham, K.P. and Laake, J.L. 1993. Distance
Sampling: Estimating Abundance of Biological Populations. Chapman and Hall,
London, reprinted 1999 by RUWPA, University of St. Andrews, Scotland.
The 1993 book is still available for download over the internet at no charge, at:
http://www.ruwpa.st-and.ac.uk/distance.book/
The advanced methods in this software are described in the following book:
Buckland, S.T., Anderson, D.R., Burnham, K.P., Laake, J.L., Borchers, D.L. and
Thomas, L. (editors) 2004. Advanced Distance Sampling. Oxford University
Press, London.
This book describes automated survey design methods, multiple covariate
distance sampling and mark-recapture distance sampling. In addition methods
are described that are not currently in Distance but which we hope to include in
the future, such as spatial modelling of density and adaptive distance sampling.

Staying in Touch

Distance-sampling Email List

The purpose of this list is to promote the sharing of ideas and information among
researchers and practitioners interested in distance sampling techniques. It is a
relatively low-volume list, and has been running since 1998.
Suitable topics for posting include:
• questions about survey design and analysis
• new methodological developments

User's Guide Distance 6.0 Beta 5 Chapter 1 - Introduction • 3

 • use of software tools (program Distance and other software)
• news about upcoming meetings, workshops and conferences where
distance methods will be discussed
• jobs in distance sampling-related fields
To join this distance-sampling list, send an email to [email protected]
with the following in the body of the message (not in the subject line):
join distance-sampling yourfirstname yourlastname
Replace the text “yourfirstname” with your first name and the text
“yourlastname” with your last name (e.g., join distance-sampling Joan Smith)
In response, you will receive a message back that explains how to use the
listserver. More information about the listserver, and an archive of messages
sent to the list, are available at the list's home page
http://www.jiscmail.ac.uk/lists/distance-sampling.html
Please check the archive of previous messages before posting! Assuming you
have an internet connection, you can access the archives directly from within
Distance by choosing Help | Distance on the Web | distance-sampling
List Archive from the main menu.

Program Distance Web Site

Program updates, etc. will be posted to the program Distance web site. The web
address is
http://www.ruwpa.st-and.ac.uk/distance/
Assuming that you have an internet connection, you can access the web site
directly from within Distance by choosing Help | Distance on the Web |
Distance Home Page from the main menu.

Sending Suggestions and Reporting Problems

If you are having problems with Distance, please check the chapter on
Troubleshooting, and also the release notes file ReadMe.rtf (Help | Release
Notes). An up-to-date list of Known Problems is in the Updates section of the
Program Distance Web Site. In addition, you should check the archives of the
distance-sampling email list.
Once you have exhausted these possibilities, please send a message to the
program authors at [email protected]. If a project file is required to
reproduce the problem, please export it to a zip file (File | Export Project…)
and send it with your email. Please remember that Distance is free, so technical
support is given on a “best effort” basis (i.e., we'll do the best we can, given our
other commitments).

4 • Chapter 1 - Introduction User's Guide Distance 6.0 Beta 5

Chapter 2 - About Distance

What is Distance?
Distance is a Windows-based computer package that allows you to design and
analyze distance sampling surveys of wildlife populations (for more about
distance sampling, see the distance sampling reference books).
Automated Survey Design
Using Distance, you can enter or import information about your study area into
the built-in GIS. You can then try out different types of survey design to see
which might be most feasible. Distance can look at overall properties of the
design such as probability of coverage. It can also generate survey plans from
the design. For more details, see Chapter 6 - Survey Design in Distance.
Data Analysis
Once you have collected your survey data, Distance can be used to analyze it.
Analysis is done in Distance using analysis engines, of which there are currently
three: the conventional distance sampling (CDS) engine, the multiple covariate
distance sampling (MCDS) engine and the mark-recapture distance sampling
(MRDS) engine. For more details, see Chapters 7-10.

Use Agreement

When you install Distance, you agree to abide by the Use Agreement. A copy of
this agreement is in the Distance program directory, in the file
UseAgreement.txt. This can be accessed from within Distance by selecting
Help | About Distance…, and clicking on the Use Agreement tab.

Sponsors
Distance is currently free to all users. Nevertheless, it is not free to develop and
maintain! If you use Distance on a regular basis, please consider sponsoring the
software. You could either make a donation towards program development and
maintenance or you could finance a specific new feature that you would find of
use. More details can be obtained by contacting the program authors (see
Sending suggestions and reporting problems).
A list of sponsors of this release of Distance can be seen by selecting Help |
About Distance…, and clicking on the Sponsors tab. See also the
Acknowledgements section, below.

User's Guide Distance 6.0 Beta 5 Chapter 2 - About Distance • 5

Distance Development Team
Development project coordinator
Len Thomas is an academic fellow with the Centre for Research into Ecological
and Environmental Modelling (CREEM) at the University of St Andrews,
Scotland.
Programmers
Len Thomas (see above)
Jeffrey L. Laake is a statistician at the National Marine Mammal Laboratory in
Seattle, USA.
Samantha Strindberg is a quantitative ecologist working in the Living
Landscapes Program at the Wildlife Conservation Society, New York, USA.
She was formerly a graduate student in the School of Mathematics and Statistics,
University of St. Andrews, Scotland.
Eric Rexstad is a research fellow with Centre for Research into Ecological and
Environmental Modelling (CREEM) at the University of St Andrews, Scotland.
Fernanda F. C. Marques is a wildlife and wildland monitoring specialist with
the Amazon-Andes Conservation Program of the Wildlife Conservation Society,
based in Rio de Genero, Brazil. She was formerly a graduate student in the
School of Mathematics and Statistics at the University of St. Andrews.
M. Louise Burt is a research fellow with Centre for Research into Ecological and
Environmental Modelling (CREEM) at the University of St Andrews, Scotland.
Jon R. B. Bishop is a graduate student in the School of Mathematics and
Statistics at the University of St. Andrews.
Principal investigators
Stephen T. Buckland holds the Chair in Statistics in the School of Mathematics
and Statistics at the University of St. Andrews.
David L. Borchers is the head of the Research Unit for Wildlife Population
Assessment, a research unit within CREEM at the University of St. Andrews.
Jeffrey L. Laake (see above)
David R. Anderson is Professor of Fishery and Wildlife Biology at Colorado
State University, USA, and was formerly Leader of the Colorado Cooperative
Fist and Wildlife Research Unit before retiring in May 2003.
Kenneth P. Burnham is an Assistant Leader of the Colorado Cooperative Fish
and Wildlife Research Unit and Professor of Biological Statistics at Colorado
State University, Fort Collins, USA.
Development team members
Sharon L. Hedley is a consultant statistician based in Fife, Scotland.
John H. Pollard is a software engineer and former graduate student in the School
of Mathematics and Statistics at the University of St. Andrews.
Tiago A. Marques is a graduate student in the School of Mathematics and
Statistics at the University of St. Andrews.

Acknowledgements
We are grateful to our respective agencies and institutions for their support
during the production of this and the previous versions of Distance. In particular,
the Biotechnology & Biological Sciences Research Council (BBSRC) and
Engineering & Physical Sciences Research Council (EPSRC) provided full

6 • Chapter 2 - About Distance User's Guide Distance 6.0 Beta 5

 funding to Len Thomas to coordinate the development of Distance 4. Additional
financial support for the production of Distance 4 and 5 came from the Wildlife
Conservation Society, the US National Marine Fisheries Service, the US
National Parks Service, Department of Fisheries and Oceans Canada and the
Colorado Division of Wildlife, and for Distance 6 came from Geo-Marine Inc.
We thank Brad Martinez, of the Common Controls Replacement Project, and
Steve McMahon, author of vbAccelerator.com, for common dialog code and
components. Thanks also to David Liske of Delmark Computing Services for
his open source HTML help class, Francesco Balena, editor of Visual Basic
Programmers Journal, for the extended collection class, Phil Fresle of Frez
Systems for the compression wrapper class, and Karl E. Peterson, VB MVP, for
various published code snippets.
Julian Derry, from the University of Edinburgh, helped with the programming of
the first windows-based version of Distance. Katy Clarke helped to improve the
manual and provide additional sample projects. Greg Fulling, Carter Watterson
and Anurag Kumar, all of Geo-Marine Inc., helped test the DSM engine.

History of Distance
Distance evolved from program TRANSECT (Burnham et al. 1980). However,
Distance is quite different from its predecessor as a result of changes in analysis
methods and expanded capabilities. The name Distance was chosen because it
can be used to analyze several forms of distance sampling data: line transect,
point transect (variable circular plot) and cue-counts. By contrast TRANSECT
was designed only to analyze line transect data.
Distance versions 1.0 - 2.2 were DOS-based applications that were programmed
using a relatively simple command language. Version 3.0 was a windows
console application, but retained the command language structure. All of these
versions were principally programmed by Jeff Laake of the National Marine
Mammal Laboratory, US Fisheries Service.
In 1997, Steve Buckland and David Borchers, from the University of St
Andrews, obtained funding from two British research councils to proceed with
an ambitious three-year project to develop new distance sampling software. The
new software, which became known as Distance 4, was designed to be fully
windows-based, and be capable of incorporating new features such as
geographic survey design, multiple covariate distance sampling models, spatial
estimation of abundance, and dual observer mark-recapture line transect
methods. A Distance 4 project development team was assembled, coordinated by
Len Thomas. In autumn 1997, it was decided to produce an intermediate version
of Distance: fully windows based, but with the same analysis capabilities as the
current version 3.0. This new program, Distance 3.5, took one full year to
develop, and was released in November 1998, with various updates through to
February 1999. Distance 3.5 was downloaded by over 4000 users, from around
120 countries.
Extension of Distance 3.5 to become Distance 4 began in 1999, and the software
was first previewed at training workshops in summer 2000. After various public
beta versions, Distance 4.0 was released in 2002, followed by Distance 4.1 in
2003 and Distance 5.0 in 2006. This last version has a major new feature in the
form of a link to the free statistical software R, thereby facilitating a major
expansion in the analytical capabilities potentially available to Distance users.
We have also released a beta version of Distance 6.0, which contains a new
density surface modeling engine
We are still actively developing the software, incorporating new features and
extending current ones. If you have any comments or suggestions about the
program, we’d love to hear from you!

User's Guide Distance 6.0 Beta 5 Chapter 2 - About Distance • 7

New in Distance

New Features of Distance 6.0

New Analysis Capabilities

• New analysis engine: density surface modeling (DSM).

New Features of Distance 5.0

New Analysis Capabilities
• Interface with the free statistics software R
• New analysis engine: Mark-recapture distance sampling (MRDS).
Allows analysis of dual observer distance sampling surveys, where
probability of detection on the trackline can be estimated.
Currently restricted to line transects. See Chapter 10 - Mark
Recapture Distance Sampling in the Users Guide for details.
Other Imrovements
• Better bootstrapping in the CDS and MCDS engine: progress is
given on the main toolbar; bootstrap point estimates are given in
the output and analysis browser columns (useful for model
averaging – see Model averaging in CDS Analysis).
• Better documentation (e.g., MCDS engine command reference in
an appendix) and many more sample projects.

New Features of Distance 4.1

New Analysis Capabilities
• Better goodness-of-fit testing – q-q plots, K-S tests and Cramer-
von Mises statistics for exact data in both CDS and MCDS engines
• When calculating adjustment terms, you can now scale distance by
either truncation distance w or estimated scale parameter σ. More
details are in Chapter 9, under Defining MCDS models (since the
option mostly applies to the multiple covariate engine).
• When combining stratum estimates to form a global density
estimate, if you weight by effort you can now treat the effort strata
as either fixed or random effects (see Stratification and Post-
stratification, Chapter 8 - Conventional Distance Sampling
Analysis).
Graphical User Interface
• GIS has improved capabilities for coping with projections: the list
of candidate projections is much longer and you can now specify
projection parameters.

New Features of Distance 4.0

Survey Design
• Ability to enter study area details into built-in GIS

8 • Chapter 2 - About Distance User's Guide Distance 6.0 Beta 5

 • Can then examine properties of user-specified designs, from a wide
range of different design classes (e.g., random points, grid of
points, line segments, zig-zag lines, etc)
• Can generate survey plans for the different designs.
New Analysis Capabilities
• New MCDS engine allows multiple covariates in the detection
function.
• Data selection function of the Data Filter now much more
powerful.
• For surveys with nested sample layers (e.g., clusters of points along
a line), the user can choose which level to treat as the independent
sample for determining encounter rate variance.
• Analysis engines now calculate AICc, as well as other model
selection measures. More of these statistics are summed across
strata for stratified analyses.
• For stratified analyses, MCDS engine allows estimation of
probability of detection by stratum when detection function
estimated globally – can reduce bias when there isn’t enough data
to estimate detection function by stratum.
• For stratified analyses, starting values and bounds on parameters
are now specified independently for each stratum.
Graphical User Interface
• Built-in GIS, based on MapObjects by ESRI (compatible with
ArcView)
• Data Explorer and Analysis Browser consolidated into the Project
Browser. New tabs for Designs, Surveys and Maps added.
• New Analysis Components window makes it easier to keep track of
Data Filter and Model Definitions.
• Lots of small refinements, e.g.:
• Data Filter and Model Definition properties remember which
tab they were last on, making setting up analyses quicker.
• Naming objects is now easier – double click on the names in
the Project Browser, or in the Details windows to edit them.
Distance Projects
• Can now export projects to zip archive files, and open projects
directly from the archive file. Zip files are automatically spanned
across removable disks (e.g., floppy disks) if they are too big to fit
on one disk.
• Can set up projects using another project as a template
Data Storage
• Much more flexible data structure – data can be linked from
external databases in a variety of formats (although there is no user
interface for this yet – has to be done by direct editing of the
database).
• Internal data now stored in DistData.mdb file within separate data
folder. The data folder is also used by default to store the GIS
information (ESRI shapefiles). Distance 4 project is therefore a
project file (.dst file) and the associated project folder.

User's Guide Distance 6.0 Beta 5 Chapter 2 - About Distance • 9

 • More layer types, including coverage probability, nested strata and
effort layers, etc. Can now have unlimited number of data layers in
the project
• New survey object allows multiple surveys to be stored together in
the same project file, but analyzed separately. Surveys can have
different survey methods and different data layers.
• Units for data and analysis can be changed after the project is set
up.
New Utilities
• Data import now more flexible – can import one or more data
layers, and can use ID or Label fields to link to the parent layer.
Various limitations in previous version caused by MS database
engine not correctly recognizing the data types of text fields have
been worked around.

New Features of Distance 3.5

Graphical User Interface
• Well defined menu structure and button-bars allow user to navigate
through program
• Interactive "Wizards" guide the user through the process of setting
up a new distance project
• Spreadsheet-based Data Explorer for entering data
• Summary table (Analysis Browser) allows the user to view and
compare analyses, and is the starting point for creating and running
new analyses
• Analyses can be grouped into sets for convenient archiving
• Analysis specification is completely graphical - users do not need
to learn a command language to use the program
• Each analysis is split into two components: Data Filter and Model
Definition, allowing for easy reuse of the components to create new
analyses
• Results of multiple analyses can be compared side-by-side in
Analysis Details windows
• Any error and warning messages generated during the analysis are
clearly displayed
• Detailed results output is split into pages for ease of viewing
• Fitted detection functions are displayed as high resolution plots
• Extensive windows-based help; context sensitive help available at
any point in the program
Robust Data Storage
• Data and analyses stored in single file (a distance project file),
which has a robust, industry standard database structure
New Utilities
• Import of data from text files - "flat file" format allows easy export
from common database and spreadsheet applications
• Import of command files from previous versions of distance

10 • Chapter 2 - About Distance User's Guide Distance 6.0 Beta 5

 • Click one button to copy high-resolution plots, results text, analysis
tables or data from Distance to the windows clipboard. From there
paste into any word processor or spreadsheet.
New Analysis Capabilities
• Additional information can be stored in the Data Explorer, and this
can be used to subset or post-stratify the data during later analysis
• Data Filter allows selection of subsets of the data for analysis
• Data can be post-stratified (for example by sex or species) for
estimation of components of the analysis
• Multiple analyses can be run at one time (e.g., bootstrap analyses
can be run in the background)
• Multipliers offer a flexible way to scale the density estimate to
account for indirect counts, g(0) < 1, etc.
• Analysis Engine now fully 32-bit, making it significantly faster and
allowing analysis of larger datasets
• Numerous small improvements and fixes have been made in the
Analysis Engine
Changes in Distance Analysis Engine
• This section lists changes in the analysis engine's capabilities
between Distance 2.2 and 3.5 (although we don't list all the minor
fixes). There will undoubtedly be some differences in results
between the versions, due to our switch from the old 16-bit NAG
fitting routines to the new 32-bit IMSL routines. The switch was
made for performance (speed and memory) and licensing reasons,
and we expect the new routines to perform (i.e. Converge) as well,
if not better, than the old ones.
• AIC is summed over strata for stratified analyses
• AIC is now the default for automatic selection of adjustment terms.
This should make little difference in practice, but was done for
consistency with the use of AIC to select among multiple models.
• Can set upper and lower bounds on key function parameter
estimates
• Can select among multiple models using BIC
• Estimation algorithm tweaking options (EPSILON and
ITERATIONS) are now gone. (The new fitting routines by IMSL
don't have these options.)
• Parametric bootstrap for F0 (old VARF command) gone. (It was
not very useful.)
• Cluster size estimate is now based on regression of g(x) vs log(x)
by default (used to be test first and use regression only if
statistically significant). Nothing is lost by using the regression by
default, and the method is now more consistent among analyses.
We also don't like hypothesis tests much and would rather avoid
them.
• The user can no longer use different adjustment term selection
methods in different models within the same estimate (ie
ESTIMATOR /SELECT command has gone).
• Only one ESTIMATOR /CRITERION switch is allowed in an
ESTIMATE section

User's Guide Distance 6.0 Beta 5 Chapter 2 - About Distance • 11

 • The default number of bootstraps is now 999 (used to be 1000 -
should make almost no difference in practice). (For justification,
see Buckland, S.T. 1984. Monte Carlo confidence intervals.
Biometrics 40:811-817)
Other Minor Changes
• The plot file format is now slightly different (see Model Definition
Diagnostics)
• The stats file has some additional statistics (see Model Definition
Diagnostics)

New Features Planned for Future Versions

Survey Design
• More design classes
• More design statistics – index of spatial spread of sample units
• Adaptive survey design
Analysis
• Analysis of adaptive survey designs
• Minor improvements:
• Some refinements in the Model Definition properties to make
the exploratory stage of analysis easier to do
Simulation
• Ability to simulate animal populations, lay surveys down upon
these simulated populations and do analyses. Can then look at
relative efficiency of different survey designs, evaluate the bias of
estimators, etc.
Graphical User Interface
• Exploratory data analysis engine
• Easy-to-use interface for linking GIS data, and attribute data
without importing it.
• Import data from non-text data sources
• Ability to manipulate display properties of maps
and more ...

12 • Chapter 2 - About Distance User's Guide Distance 6.0 Beta 5

Chapter 3 - Getting Started

Objective
The aim of this chapter is to provide a gentle introduction to Distance, and an
overview of its capabilities. We don’t go into much detail, but focus instead on
giving you an impression of how the software works and where to find things.
Please note that this is not a substitute for reading the rest of the
manual. You will need to know about Distance projects (Chapter 4) and how
data is stored in Distance (Chapter 5) before you can use the program effectively
for either survey design (Chapter 6) or analysis (Chapter 7).
This chapter gives step-by-step instructions to walk you through four examples.
In the first, the goal is to perform a preliminary analysis of some straightforward
line transect data. We create a new Distance project, import the data, and do
some preliminary analyses. In the second, we deal with import of slightly more
complex survey data. In the third, we look at creating a geographic project and
using it for survey design, and in the forth we look at more complex geographic
data using one of the sample projects.
Note that this Users Guide is available in both on-line and print-ready formats.
If you’re currently reading the on-line version, you may find it easier to follow
this chapter in the other format – see the Welcome topic for more about the
different formats available.
To start with, we’ll assume that you’ve downloaded and installed Distance. If
not, go to the Program Distance Web Site for instructions.

Example 1: Using Distance to Analyze Simple Data

In this section, we will create a new Distance project, import some simple line
transect data and perform some initial analyses. The data are the simulated line
transect data used as a running example in Chapter 4 of Introduction to Distance
Sampling (see Distance Sampling Reference Books). These data are also
available as the sample project “Line Transect Example” (see Sample Projects).

Example 1 - Preparing the data for import

In general, we do not recommend that survey data be entered directly into
Distance. Instead, it should be stored in some purpose-written data management
software, such as a spreadsheet or database, and then imported into Distance.
The data for this example are stored in a Microsoft Excel file. Distance cannot
import directly from Excel (or any other package); you have to first export the
data to a text file. (If you do not have a copy of Excel on your computer, see the
end of this topic for instructions.)

User's Guide Distance 6.0 Beta 5 Chapter 3 - Getting Started • 13

 • Use Excel to open the file Example1.xls. This file is located in the
Sample Projects folder, which is below the Distance program
folder (usually “C:\Program Files\Distance 6\Sample projects”).
• Note the layout of the file:
• There is one column for each of: stratum name, stratum area,
transect name, transect length, distance (perpendicular distance
in this case) and cluster size. (The exact columns required
depends on the survey – e.g., cluster size would not be
required if objects were individual animals.)
• There is one row for each observation. Look at row 102 -- the
distance and cluster size entries are missing. This represents a
transect (“Line 11”) where there were no observations.
• In Excel, choose File | Save As… Under Save as type: choose
Text (Tab delimited) (*.txt). Click Save, and now close Excel.
• You should now have a text file Example1.txt in the same folder as
Example1.xls. You can open the text file in a text editor (e.g.,
Notepad) to examine it if you like.
If you do not have a copy of Excel on your computer, you can copy the file
Example1Backup.txt to Example1.txt and continue.

Example 1 - Creating the Distance project

In Distance, survey data and analyses are stored in a project. Before you can
import the data, you first need to create a new project.
• From the Windows Start Menu, choose Programs (or All
Programs), then Distance, and click on Distance 6.0.
• In Distance, choose File | New Project. A window opens asking
for the name of the project to create. Under File name, type
“Example1”, and click on Create.
• The New Project Setup Wizard now starts. This is designed to
guide you through creating a new project. It will ask you what you
want to do and give a list of options. The first option, which is
already selected is to analyze a survey that has been completed.
This is what you want to do, so click on Next at the bottom of the
window.
• The next window (Step 2) confirms the selection you have made
and gives some introductory information. Once you have read it
click on Next.
• The next window (Step 3) asks about the Survey Methods. Firstly
it asks what type of survey has been carried out. Make sure Line
transect is selected. The observer configuration should be
Single observer (double observer methods are for estimating
g(0) and are covered in the Advanced Distance Sampling book).
Measurement type for this example is Perpendicular distance,
and the observations are Clusters of objects. When the correct
options are selected, click on Next.
• Step 4 asks about the measurement units that were used. In this
example, distances were measured in Metres, the transect length
was measured in Kilometres and the area was measured in
Square kilometres. Choose the appropriate options and then
click Next.

14 • Chapter 3 - Getting Started User's Guide Distance 6.0 Beta 5

 • Step 5 asks if you wish to add in any multipliers. No multipliers
are required for this data set, so click Next.
• Distance is now ready to set up the project, and in the last window
(Step 6) asks you where to go next. We want to import some data,
so select the option to Proceed to Data Import Wizard, and
click Finish. The project is now created and the Import Data
Wizard is started.

Example 1 - Importing the data

Before you import any of your own data, you need to understand how data is
stored in Distance, by reading Chapter 5 of this Users Guide. But for now,
you’ll get an idea of how import works by following the steps below.
• If you have followed the previous steps, the Import Data Wizard
is now open at the Introduction page (Step 1). Once you have
read the text there, click on Next.
• A window will open prompting you for the file to import. Select
Example1.txt by clicking on it and then click on OK.
• The Data Destination page (Step 3) of the wizard will open. This
page asks you about where the data from your text file should be
placed in the Distance database. Before you import your own data,
you’ll need to find out more about these options (e.g., by pressing
F1), but for now the default options here are appropriate for this
example, so click Next.
• The Data File Format page (Step 4) is now opened. In our case,
the first row contains the column labels, so click on the option Do
not import first row, then click Next.
• The Data File Structure page (Step 5) will now open up. Here you
have several choice as to how you wish to proceed.
• If, like in this example your data is set up in columns in the
same order as you wish them to appear in the Distance data
sheet checking the option Columns are in the same order
as they will appear in the data sheet means that
Distance will automatically assign a layer name, field name
and field type to each row. Care should be taken however to
check that Distance has done this correctly, as if additional
columns are included in the data Distance may mis-label
columns.
• Another other option is to manually assign the names and type.
This is a procedure that is worth knowing, as you will need to
use it when the automatic assignment is incorrect or your data
is not in the correct order to use that option. By default, the
Layer name: row above each of the columns shows
[Ignore]. Double-click on the [Ignore] corresponding to the
first column (above the stratum name column). A drop down
menu will appear. Select Region. Now double-click on the
box below, which is in the row Field name. Again a drop
down menu appears. Select Label. Now double-click on the
box underneath in the field type row, you will see that the
word Label automatically appears there. Next go to the
Layer name for the Area column. Select Region, then
select Area. Note that here area will be the only option to
choose from as the only other option for region has already
been used. Now click on the Field type box, and this will be
automatically filled with the word Decimal. Continue doing

User's Guide Distance 6.0 Beta 5 Chapter 3 - Getting Started • 15

 this process until each column has been assigned a field name,
field type and layer name. The boxes should look like this:

Completed data file structure for Example 1

• Now click on Next, and then click Finish to import the data.

Example 1 - Studying the data in Distance

Now the data has hopefully imported successfully, and the project you created
should be open with the Data tab of the Project Browser selected. This is the
main interface to your data in Distance, and is called the Data Explorer.
• In the left-hand pane, under Data layers, Click on Observation.
This expands the view in the right-hand pane to show you all the
data. Scroll down to verify that all 12 transect lines and 105
observations have been imported. Note that line 11 has no
observations associated with it, as we wanted.

Example 1 – Running the first analysis

Now you’ve created a new project and imported your data, it’s time to do some
analyses.
Before you run any of your own analyses, you should read Chapter 7 of this
Users Guide, which contains more information about how analyses work in
Distance, and lots of pictures of the various parts of the interface. You should
also look at the subsequent chapters which give details of each analysis engine.
For now, however, you’ll get an idea of how analysis works by following gthe
steps below.
• Click on the Analyses tab of the Project Browser. This brings
up a table called the Analysis Browser. This is where the
analyses that are carried out will be listed. In the left-hand pane is
a line highlighted that says New Analysis. This is the default
analysis that Distance creates. A grey ball (the status ball) also
on that line indicates that the analysis has yet to be run.
• Double click on the grey ball (or choose Analyses | Analysis
Details). A new window titled Analysis 1 will open up. This
window is the Analysis Details window, and you are on the Inputs
page (see tabs along the side).
• At the bottom of the Inputs page is a section called Model
Definition. The selected model definition is called Default
Model Definition and this is the only one that has been defined so
far. Click on Properties… beside the model definition to open up
the Model Definition Properties window. This window gives
all the options available to change a model in Distance.
• Click on the Detection Function tab .This will show what key
function and adjustment term are currently selected for use.
Explore the other options if you wish, then click OK to exit the
Model Definition Properties.
• Click on Run in the Analysis Details Inputs page. Once the
analysis has run the Results tab will turn green indicating that
there were no problems with the analysis, and you will be taken to
the Results page. If there had been a problem, then you would be
taken to the Log page of the Analysis Details window. Then the

16 • Chapter 3 - Getting Started User's Guide Distance 6.0 Beta 5

 log tab would be coloured either amber for a warning, or red for an
error.
• In the Results tab click Next several times to view each page
results. Once you have finished looking at the results, close the
Analysis Browser window by clicking on the close button in the
top right hand corner of the window, or choosing Analysis -
Results | Close.
• In the Analyses Browser, the grey ball should now have turned
green, to indicate that the corresponding analysis ran OK.. A
summary the results is given in the right hand window pane. This
is useful for when you wish to compare various analyses, without
having to scan through all the results pages. (The columns shown
can be configured by choosing Analyses | Arrange
columns….)

Example 1 - Creating a new analysis

The default analysis we ran in the last section uses a half-normal key function
with cosine adjustments. Here, we’ll create another analysis that uses a hazard
rate key function with simple polynomial adjustments.
• Return to the Analysis Browser (Analysis tab of the Project
Browser) and click on the New Analysis button on the
Analysis Browser toolbar (or choose Analyses | New
Analysis…).
• Before you go any further, give the analysis a sensible name that
reflects what it does by clicking on the current name (“New
Analysis 1”) and then typing in the new name (e.g., “Untruncated
hr+poly”). You might want to give a sensible name to the default
analysis too (e.g., “Untruncated hn+cos”). This might not seem
important when you only have two analyses, but if you don’t name
new analyses as you create them you’ll soon find yourself loosing
track of which is which!
• Double-click on the status ball for the new analysis (or make sure
it’s highlighted and choose Analyses | Analysis Details) to go
to the Analysis Details window for this analysis. Because the
analysis is not run, you are taken to the Inputs page.
• Click on New… in the Model Definition section. The Model
Definition Properties window will open up. Choose the
Detection Function tab and under Models, choose Hazard
rate key and Simple polynomial adjustments.
• This new model definition is currently called “Default Model
Definition 1”, so give it a sensible name e.g., “hr+poly” by clicking
on Name: at the bottom of the Model Definition Properties.
Then click OK.
• You might also want to give a sensible name to the other model
definition (currently “Default Model Definition”). One way to do
this is to select that model definition in the Inputs page, open the
Model Definition Properties, and change the name there (e.g., to
“hn+cos”). Another, easier, way is to double-click the name on the
Inputs page and change it there.
• Now run the analysis and compare the results with the previous
one.

User's Guide Distance 6.0 Beta 5 Chapter 3 - Getting Started • 17

 Example 1 - Further investigations
As we mentioned in the introduction, these data form the running example in
Chapter 4 of Introduction to Distance Sampling. In the book the various analysis
options are explored including truncation, grouping into intervals, various key
function + adjustment combinations and different methods of estimating mean
cluster size in the population.
To truncate the data a new Data Filter needs to be created. To understand
properly what Data Filters and Model definitions are, you need to read Chapter 7
- Analysis in Distance, but for now we will show you how to create two new
analyses with a new Data Filter.
• In the Analysis Browser, click on the New Analysis button
to create a new analysis.
• Name this analysis something like “19m trunc hn+cos”
• Double click on the status ball to open the Analysis Details for
this analysis.
• Under Data Filter, click New….
• In the Truncation tab, select Discard all observations
beyond and type in 19.
• Give the new data filter a sensible name (e.g., “19m truncation”),
and then click OK.
• Check that the highlighted Model Definition for this analysis is the
“hn+cos” one. Note that two analyses are now using the same
Model Definition – this analysis and the first analysis we ran,
which was without truncation.
Let’s not run this analysis yet – instead we’ll create a second analysis with 19m
truncation and hazard-rate + simple polynomical model.
• Choose Window | Project Browser or View | Project
Browser (or click on the View Project Browser button) to put the
Project Browser window on top.
• Create another analysis, and open the Analysis Details for this
analysis. (You now have two Analysis Details windows open.)
• Name the new analysis “19m trunc hr+poly”.
• Make sure the “19m truncation” Data Filter is selected for this
analysis, and the “hr+poly” Model Definition.
We could run this analysis by clicking the Run… button, but here’s a different
way to run more than one analysis at once. It’s useful if you want to run a
number of time-consuming analyses, so you can go off and do something else
while they are running.
• Choose Window | Project Browser (or click on the View
Project Browser button) to put the Project Browser window on top.
• Click on the “19m trunc hn+cos” analysis and then hold down the
CTRL key and click on the “19m trunc hr+poly” analysis. This
should select both of them. Click on the Run Analysis button
on the Analysis Browser toolbar or choose Analyses | Run
Analysis. The two analyses will be run once after another.
• When they are finished, you can look at the results in the two
Analysis Details windows. (Note: the Analysis Details windows
do not have to be open to run an analysis – you can just highlight it
in the Analysis Browser and click the Run Analysis button.)

18 • Chapter 3 - Getting Started User's Guide Distance 6.0 Beta 5

 Try creating more analyses that replicate the analyses in Chapter 4 of
Introduction to Distance Sampling (such as grouping the data into intervals
before analysis – to do this you need to create a new Data Filter and change the
options in the Intervals tab).
As you create more Data Filters and Model Definitions, you may find that you
want to change their order, delete them or rename several at a time. A
convenient way to do this is using the Analysis Components window, which
can be opened by choosing View | Analysis Components or clicking the
button on the main toolbar. In the Analysis Components window, clicking the
first button lists the Data Filters and clicking the second button lists the
Model Definitions.
An example of a set of Analyses, Model Definitions and Data Filters that have
already been set up is given in the Ducknest sample project. To open this,
choose File | Open Project… and choose “Ducknest”.

Example 2: More Complex Data Import

In this section, we will import point transect data on house wrens (Troglodytes
aedon). The study is used an illustrative example in section 8.6 of Introduction
to Distance Sampling (see Distance Sampling Reference Books), and there is a
sample project that already contains the data (House Wrens.dst - see Sample
Projects) if you want to skip the import part of the section.
The data were collected from 155 points, with between 14-16 points within each
of 10 study blocks. Each block was of size 16-ha. The blocks were situated in
riparian vegetation along 30km of South Platte River bottomland near Crook,
Colarado. The data were collected by 4 observers, who each visited each point.
In the analysis, we will post-stratify by observer to recreate the results from the
Distance book, but another possibility would be to use observer as a covariate in
a multiple covariate distance sampling analysis.
We assume the user has already worked through Example 1: Using Distance to
Analyze Simple Data, so is familiar with the basics of project creation, data
import and data analysis.

Example 2 - Preparing to import the data

The data are again stored in a Microsoft Excel file: Example2.xls. Use the
instructions from the previous example to save this to a tab-delimited text file
Example2.txt. If you don’t have a copy of Excel on your computer, copy the file
Example2Backup.txt to Example2.txt.
Now create a new Distance project, Example2.dst:
• In Distance, choose File | New Project. A window opens asking
for the name of the project to create. Under File name, type
“Example2”, and click on Create.
• Click through the New Project Setup Wizard.
• In Step 3, you need to specify that this is a Point transect
survey, and the measurements are therefore Radial distance.
• In Step 4, the units of measurement are Metres, and area is
measured in Hectares.
• In Step 5, we do want to define a multiplier. We need to tell
Distance that each point was visited 4 times. One way to do
this would be to have a column in the data for survey effort,
and have a value of 4 in each row. This would be most useful
if the survey effort (number of visits) varied between points.

User's Guide Distance 6.0 Beta 5 Chapter 3 - Getting Started • 19

 In this case, there were 4 visits to each point, so we will define
a multiplier called “Visits” that divides the density estimate,
and later we will put a value of 4 for this multiplier. (For more
on Multipliers, see Multipliers in CDS Analysis). To define
the multiplier, tick Other and give it the field name “Visits”.
Choose N for Create Fields for SE and DF (there is no
standard error or degrees of freedom associated with this
multiplier).
• In Step 6, choose Proceed to Data Import Wizard, and
click Finish.

Example 2 - Importing the data

You can pretty-much follow the instructions from the previous example to
import the data. The only change is in the Data File Structure page (Step 5) of
the Import Data Wizard. Here there are two differences: (1) you can’t use the
option that Columns are in the same order as they will appear in the data sheet
(because the Survey Effort field is missing), and (2) there is an extra field for
Observer. To get around (1) you should manually assign the field names to the
columns, as outlined in Example 1. The following are instructions for creating a
new field to put the Observer data in.
• First manually assign all the other columns to the appropriate
fields.
• Then in the Layer name: entry corresponding to the Observer
column, choose Observation (the observer data is an observation-
level column). Click on the entry corresponding to the Field
name: row and the text “New Field” will appear. Type in
“Observer” instead, and hit Enter to confirm this name. Now in
the entry corresponding to the Field type: row, choose Integer
(you could equally well choose Text as the field is likely only
going to be used for data selection and possibly as a factor
covariate in the MCDS engine).
• Click Next to continue to the final Step, and then click Finish.
The new field will be created and the data imported.
• Once the data have been imported, use the Data Explorer (the
Data tab of the Project Browser) to check the new field had
been created correctly.

Example 2 - Analysis
Before analyzing any data, we need to enter the number of visits in the multiplier
field that has been created in the project.
• Click on the Data tab of the Project Browser.
• Under Visits, enter the number 4.
We also need to set up the multiplier correctly in the Multipliers tab of the first
Model Definition.
• Choose View | Analysis Components and then Analysis
Components | Model Definitions.
• Open the default model definition’s properties by choosing
Analysis Components | Item Properties… or by double-
clicking on the ID 1 in the Analysis Components Window.
• Under Multipliers, there should be one multiplier defined, with
field name Visits. However, the Operator is wrong – it is set to

20 • Chapter 3 - Getting Started User's Guide Distance 6.0 Beta 5

 “*” (multiply) by default, while we want the density estimate to be
divided by the number of visits, so we choose “/”.
• Click OK. Each time you define a new model definition, it is
based on the one you have currently selected, so all future model
definitions will automatically now have the correct operator.
If you wish, you can now set about recreating the analyses of section 8.6 of
Introduction to Distance Sampling. However, note that you’ll need to learn
about stratification (by area), post stratification (by observer) and data selection
(selecting out each observer) to recreate all the analyses in the book. These are
all outlined in this Users Guide in Chapter 7 - Analysis in Distance and Chapter
8 - Conventional Distance Sampling Analysis.
Note also that you’re likely to obtain slightly different results from those in the
book, as the data are slightly different and the CDS analysis engine has been
modified since the book was written.
You can also try using observer as a factor covariate using the multiple covariate
distance sampling (MCDS) analysis engine – for more on this see Chapter 9 -
Multiple Covariates Distance Sampling Analysis.
Have fun!

Example 3: Using Distance to Design a Survey

In this section, we will design an aerial line transect survey of marine mammals
in an area of approximately 100km2 adjacent to part of the Scottish coastline,
called St. Andrews Bay. We will use a systematic grid of parallel lines. The
small survey plane permits a total flight length of approximately 250km,
excluding the flight time to and from the landing strip further down the coast.
We would like to create a design that uses the maximum possible proportion of
the flight length moving along transect lines (as opposed to moving between the
lines), while at the same time respecting the overall constraint of 250km.

Example 3 - Preparing the data for import

Before you can start designing a survey, you need to create a new Distance
project and enter the coordinates of the boundary of your study area. There are
several ways to get geographic data into Distance– here we’re going to import
the co-ordinates from a text file. You can also type the coordinates in manually
or import them from a Geographic Information System (GIS) – see Chapter 5 -
Data in Distance for more on these.
You should also read the information in that chapter about coordinate systems
before deciding whether to use a GIS to apply a geographic projection to your
data before importing it. In this example, we have projected the data (from the
OSGB 1936 geo-coordinate system using the transverse mercator projection)
into meters. Distance expects data in meters by default, so this way we won’t
have to deal with any of the geographic coordinate system or projection facilities
in Distance.
You can examine the coordinate data in the file StAndrewsBay.txt in the Sample
Projects folder, which is below the Distance program folder (usually
“C:\Program Files\Distance 6\Sample projects”).

If you want to skip the steps involved in creating the project and
importing the data, you can open the sample project StAndrewsBay.dst (choose
File | Open Project… and, select StAndrewsBay and click Open). Then, go
to the section entitled Example 3 - Creating a Survey Design.

User's Guide Distance 6.0 Beta 5 Chapter 3 - Getting Started • 21

 Example 3 - Creating the Distance project
To create a new project to contain our geographic data and survey designs,
follow these steps.
• From the Windows Start Menu, choose Programs (or All
Programs), then Distance, and click on Distance 6.0.
• In Distance, choose File | New Project. A window opens asking
for the name of the project to create. Under File name, type
“Example3”, and click on Create.
• The New Project Setup Wizard now starts. This is designed to
guide you through creating a new project. It will ask you what you
want to do and give a list of options. You want to choose the
second option, to design a new survey. Then click on Next at the
bottom of the window.
• The next screen contains some information about what Distance
will do next. Read the text and click Finish.
• The Distance project is now created and you are taken to the Data
tab of the project browser. On the left-hand pane, under Data
Layers, you can see that one data layer has been created called
“Study area”. In the right-hand pane, under Contents of Global
layer ‘Study Area’, you can see that the “Study area” layer has
one record, with ID 1 and Shape “Polygon”.

Example 3 - Importing the Geographic Data

• Use a text file editor to open the StAndrewsBay.txt file which
contains the coordinates of the study area. Highlight all of the rows
of data in the file and use the text editor’s copy facility to copy
them to the Windows clipboard. (If you are using Windows
Notepad, you would choose Edit | Select All and then Edit |
Copy).
• Switch back to Distance. Double-click on the word “Polygon”.
This opens a window called the Shape Properties dialog.
• Chose Paste from Clipboard. The X and Y columns should
now contain the rows you copied from the text file, and the top
right hand window that says “# vertices” should now read 257 (a
vertex is a corner, so this tells you how many coordinates you have
pasted in).
• Click OK to close the Shape Properties Dialog.

Example 3 - Checking the Data on a Map

• Click on the Maps tab of the Project Browser.
• From the menu, choose Maps | New Map.
• Double-click on the Name “New Map” and rename the map “St
Andrews Bay”. Hit Enter to save the new name.
• Choose Maps | View Map. A Map windows opens, with title
“Map 1: [St Andrews Bay].
• Choose Map | Add Layer. A window appears asking which layer
to add. Since the project currently contains only one layer, click
OK to add the Study Area to the map. You should see a window
something like the following:

22 • Chapter 3 - Getting Started User's Guide Distance 6.0 Beta 5

 Map of the St Andrews Bay study area

• Click the button in the top right corner of the Map window to
close the map. You will be asked whether you want to save the
changes to have made to the map (you added a new layer to it) –
choose Yes.

Example 3 - Adding a Coverage Layer

Before we can design any surveys, we need to add a second data layer to the
project called a Coverage layer. This is a grid of points at which we assess
coverage probability – i.e., the average probability of being covered by a survey
line. For more on coverage probability and associated concepts, see Chapter 6 -
Survey Design in Distance.
• Click on the Data tab of the Project Browser.
• Choose Data | Create Data Layer. The Create Data Layer
dialog opens.
• Under Layer Name, type Grid, and under Layer type, choose
Coverage.
• We now need to set the grid spacing. To do this, click on the
Properties… button to open the Grid Properties dialog. As a
rough rule of thumb when assessing coverage probability, you want
a grid spacing approximately equal to the transect strip width. In
our case, the truncation distance will be approximately 2km either
side of the plane, giving a strip width of 4km. So, we type in 4
under Distance between grid points and choose Kilometre as
the Units of distance.
• Click OK to close the Grid Properties dialog, and OK again to
close the Create Data Layer dialog. Distance tells you that the grid
spacing you have chosen will lead to approximately 62 grid points
being created. Click OK to confirm.

In larger study areas, choosing a grid spacing of about the transect

width causes a very large number of grid points to be created (10’s of
thousands). In these cases, you would want to compromise and choose a larger
grid spacing.

• You can now view the coverage grid on your map. Click on the
Maps tab of the Project Browser.
• Choose Maps | View Map to open the map of St Andrews Bay.
• Choose Map | Add Layer, choose layer name “Grid” and click
OK. You should now see the coverage probability grid points on
the map.

User's Guide Distance 6.0 Beta 5 Chapter 3 - Getting Started • 23

 Example 3 - Creating a Survey Design
In this section we will create a survey design for a systematic parallel line
transect survey with line spacing of 5 km.
• Click on the Designs tab of the Project Browser.
• Choose Designs | New Design.
• Double-click on the name “New Design” and type “5 km spacing”
or some other suitable name for the design we will create.
• Now choose Designs | Design Details. The design details window
for this design opens on the Inputs tab.
• Under Type of design, choose the Line sampler, and the
Systematic Random Sampling class.
• Now click on Properties… to open the properties dialog for the
systematic random line design.
• We can ignore the General properties for this example.
• Click on the Effort Allocation tab. Under Length Units,
choose Kilometre, and under Spacing, choose 5. Note that
Distance estimates on average 8 lines will be created and that the
average on-effort length will be 226.2 km. Note also that an angle
of 0 means that the lines will run west-east (90 would mean south-
north).
• Note from the Edge spacing option at the top of the Effort
Allocation page that we are using “minus sampling” - this means
that sample lines will be placed only within the study area (plus
sampling means that lines area also placed within a buffer around
the outside of the study area, of width equal to the truncation
distance). What effect do you think using minus sampling might
have on the coverage probability at the edge of the study area? We
will return to this later.
• Click on the Sampler tab. Choose Units of kilometre and Width
of 2.
• Click on the Coverage probability tab. Choose Estimate by
simulation, with 100 simulations for now.
• Click OK to close the Design Properties window.

Example 3 - Automated Generation of New

Surveys
A design is an algorithm for laying down samplers, and a survey is a random
realization of this design. In this section, we will create a random survey, based
on the design we created in the previous section (i.e., with a random start point
and 5km spacing between lines).
• You should be looking at the Inputs tab of the Design Details
window for the design you created - the window is called “Design
1: [5km grid] Set: [Set 1]”.
• Click the Run… button, and choose the second option, to Create
a new Survey. Click OK.
• A new Survey is created, the Survey Details window opens, and
you are taken to the Results tab.

24 • Chapter 3 - Getting Started User's Guide Distance 6.0 Beta 5

 • The first page “Design engine output”, gives you information about
this survey – for example the number of lines generated, the
amount of “on effort” length (transect length) and total line length
(on effort length + distance moving between the end of one line
and the beginning of the next). Scroll down to look at all of the
output.
• Click the Next > button to view a map of the samplers (lines).
• Click the Next > button again to get a list of the samplers’ start
and end points. These could easily be copied to the windows
clipboard and pasted into another medium (such as a text file ready
for upload to a GPS system).
• Click the button to close the survey details.
Note that the new samplers have been added to the project as a new data layer,
and this new data layer has an associated GIS shapefile, so this may be a more
convenient format for exporting to other systems such as GPS.
• You can verify that there is a new data layer by choosing View |
Project Browser, and then clicking on the Data tab. You should
see the data layer 5km grid under the Data layers pane.
• You can also click on the Maps tab, and add the new data layer to
your map (or create a new map for this layer).

Example 3 - Design Statistics

So far, we have created a single realization of the design. However, since
surveys are randomly generated from the design, properties such as total line
length and proportion of time on effort are random properties with some
unknown distribution. We are interested in knowing the average proportion of
time on effort (we want to find a design where this is high), as well as the
minimum total line length (we can only use designs where this is greater then
250km). To find these things out, we will ask Distance to simulate many
instances from the design and record what happens each time.
• Bring the Design Details window to the foreground by choosing
Window | Design 1: [5km grid] Set: [Set 1]
• Click on Run… again. This time, choose the option to Calculate
coverage probability statistics, and click OK.
• We selected 100 simulations in the Design properties, so Distance
does 100 simulations. The progress bar tells you what percentage
have been achieved. Wait for this to reach 100%.
• The Design Details Results tab now goes green, and we can access
the results.
• The first page is called Design engine output and contains
results in text format. Note the maximum total trackline length – is
it more than 250km? Note the mean on/total trackline length (this
is the proportion of time on effort) – we will need this later to
compare with the other designs.
• Click Next > to see the coverage probability map for this design.
Using minus sampling, there is actually a slightly lower coverage
probability within one truncation width of the north and south
edges of the study area than elsewhere. However with only 100
simulations this will not be evident. Instead, you will just see
random variation in coverage probability – “stripes” of high and
low coverage. If you wanted to examine the edge effect in detail,
you should re-run the design using many more simulations (5000,

User's Guide Distance 6.0 Beta 5 Chapter 3 - Getting Started • 25

 say) and probably a finer grid spacing. If you wanted perfectly
even coverage probability, you should use the “plus sampling”
option for the design, in the Design Properties, under Effort
Allocation.

Example 3 - Further Investigations

You can now set up more designs, with different between-transect spacings, and
see if they stay within the 250km total length criterion and perhaps have a better
ratio of on-effort to total line length. As a suggestion, try 4.5, 5.5 and 6km.
For more background about the survey design features of Distance, see this
Users Guide, Chapter 6 - Survey Design in Distance. See also Chapter 7 of
Buckland et al. (2004).

Example 4 - A Second Survey Design Project

This example provides an introduction to more of the geographic information
features in Distance, using an example of survey design for 4 northwestern states
in Mexico. The states form a stratum layer, so different survey effort can be
allocated to the different strata. The geographic information files are not
projected, so we need to specify how to project them in Distance.
The previous example (Example 3: Using Distance to Design a Survey) is
simpler, and should probably be worked through first. It also demonstrates one
method of geographic data import. This example is also introductory, and
demonstrates different ways to access the design windows than the previous
example. The two are therefore complementary.

Example 4 - Opening the Mexico Project in

Distance
To start Distance:
• From the Windows Start Menu, choose Programs (or All
Programs), then Distance, and click on Distance 6.0.
• Once Distance has started, on the top menu bar, select File, then
Open Project.
• Select the project file “Mexico” in the dialog box, and click Open.
The project will take a few seconds to open. Once it’s opened, you will see a
window called the Project Browser.
The project browser is the main interface for getting things done in Distance.
There are 6 tabs along the top: Data, Maps, Designs, Surveys, Analyses
and Simulations. During the course of this chapter, we will examine the
contents of each of these tabs, except the Simulation tab, which is disabled in
this version of Distance.

Example 4 - Reviewing the Project Properties

Let’s begin by examining the distance project itself. In Distance, you store all of
the information about one study area in a project. Projects are made up of a
project file, which is the file you clicked on to open the project, and a data
folder (directory). We can find out more about the project in the Project
Properties window.
• From the top menu bar, select File, then Project properties.
This opens the Project Properties window.

26 • Chapter 3 - Getting Started User's Guide Distance 6.0 Beta 5

 The General tab gives you information about the location of the
project file and its associated data folder (Mexico.dat).
• Click on the Geographic tab.
The Geographic tab gives you information about the default geo-coordinate
system of the geographic data, and the default map projection. The geo-
coordinate system is used to locate the geographic data (which is stored in
decimal degrees of latitude and longitude) on the earth's surface. The projection
is used to convert these data from the curved surface of the earth into a flat plane
that can be used for displaying maps and designing surveys. If you are planning
a survey that will take place over a small geographic area, and you are inputting
your data by hand, then you don't need to worry about geo-coordinate systems or
projections and can set both these options to [None]. In this example, however,
the survey area is quite large and the projection chosen will make some
difference to the results.
• Click on Cancel to close the Project Properties window
without saving any changes you may have made.

Example 4 - Examining the Data

• Click on the Data tab of the Project Browser.
This tab contains the Data Explorer. In the left-hand pane, under
Data Layers, you can see that there are four layers in the project:
“Mex”, “MexStrat”, “Grid1” and “Grid2”. You can tell the layer types
by looking at the icons beside the names: Mex is a Global layer,
MexStrat is a Stratum layer and Grid1 and Grid2 are Coverage layers.
When you open a new distance project, the Global layer is selected by
default, so the layer Mex is now selected.
• Click on the Data Layer Properties button on this tab to find
out more about this layer.
The Layer Properties window opens, and under the Geographic
data tab, you can see that the geographic data is stored in a shapefile
(geographic data file) called Mex.shp in the data folder, and that the
shapes in this layer are polygons (i.e. solid, multi-sided shapes).
• Click Cancel to return to the Data Explorer.
In the right-hand pane of the Data Explorer, you can see its’ fields: ID,
Label, Area and Shape. There is one record, with ID = 1. The Shape
field is new in Distance 4.0 – it holds the geographic information for
that record. Because this layer holds polygons, the shape record has the
word “Polygon” in it.
• Double click on this word to open the Shape Properties
window.
This is where you edit the geographic information inside Distance (an
alternative is to edit the shapefile Mex.shp from outside of distance,
using a GIS package such as ArcView).
• •Click Cancel to return to the Data Explorer.
• The coverage layers Grid1 and Grid2 contain a grid of points that
will be used for determining probability of coverage for our survey
designs.
• Click on “Grid 1” in the left-hand pane of the Data Explorer.
Its records open in the right-hand pane, and you can see that it has 177
records. A better way to look at the grid points is to view them in a
map.

User's Guide Distance 6.0 Beta 5 Chapter 3 - Getting Started • 27

 • Click on the Maps tab in the Project Browser.
• Click on the New Map button (3rd button along) to create a new
map.
• Double-click on the words “New Map” to edit the name of the
map, and call it “Coverage Grids”.
• To view the map, click the View Map button , or double click
on the map's ID. A Map Window opens.
The map starts life blank – you need to add some layers to it.
• Click the Add Layer to Map button and select “Mex” from
the list of layers.
• Click the Add Layer to Map button again and select “Grid 1”
from the list. Now you can see the grid points.
You can also add the points from Grid 2 to the same map.
• Click the Add Layer to Map button again and select “Grid 2”
from the list. Now you can see the grid points.
You can see that the grid points for Grid 2 are much closer together
than those for Grid 1. (Grid 2 was generated with a spacing of 20 km,
while for Grid 1 the spacing was 50km.) The points for Grid 2 obscure
those from Grid 1 – to see both, you can change the order of the map
layers:
• Press and hold the left mouse button on the legend “Grid 2” (left
hand side of the map), and drag it to below “Grid 1”.
When you have finished looking at the map, you can close it:
• Click the button in the top left corner of the Map Window to
close the map. (Say “Yes” if it asks you to save changes.)
Let’s examine the MexStrat data layer.
• Click on the Data tab of the Project Browser
• In the Data Explorer, click on the MexStrat layer to see those
data.
You can see that there are 4 strata. If you want to see where they look
like, you could create a new map in the Maps tab and add the MexStrat
layer to the map.
When you've finished exploring the data, move on to create a new
design.

Example 4 - Creating a New Design

A design is the template from which surveys can be created. When you specify
a new design, you have to choose things like what kind of sampler to use
(points or lines), how to place them in the survey area (e.g., random, systematic),
how much effort to allocate, and how to distribute it among strata, etc. Once
these are set, you use the design to generate surveys – realizations of the
design.
Let’s create a new design.
• Click on the Designs tab of the Project Browser.
• To create a new design, click the New Design button . A new
record appears in the left-hand pane, called “New Design”.

28 • Chapter 3 - Getting Started User's Guide Distance 6.0 Beta 5

 • Double click on this, and edit it to call the new design “150 random
points”.
• Click the Show Details button to open the Design Details
window.
Look under “Type of design” to see the sampler and design class; the
default sampler is “Point” and the default design class is “Simple
Random Sampling”.
Click the Properties button to set the properties for this design.
The Design Properties window opens. The options you see on the design
properties tabs depend on the type of design. In this example, choose the
following options:
• Under Stratum layer, choose the stratum layer “MexStrat”.
• Under Design coordinate system, make sure the box Same
coordinate system as stratum is unchecked. The projection
should say “Plate Carree” and the units Kilometres.
• In the Effort Allocation tab, under Edge Sampling select the
Plus option. Uncheck the box Same effort for all strata. A list
of the four strata in the MexStrat layer appears. Under Allocation
by stratum, click the Percentage from radio button, and enter
"150" as the number of points. In this example, we will put most
of our effort into the two Baja strata (perhaps because this is where
we think most of the animals of interest live). Under Effort %
enter 10 for Sinaloa, 10 for Sonora, 40 for Baja Sur (south) and 40
for Baja Norte (north).
• In the Sampler tab, select “Kilometre” (or “Kilometer”) for the
units. Let's imagine we're surveying for a very vocal species and
that our truncation distance is 5km, so we enter 5 under radius (for
this example we'll assume same sampler properties for all strata).
• Lastly, in the Coverage Probability tab, click on Estimate by
simulation and enter 50 as the number of simulations. This is far
too few for an accurate simulation, but will do for the purposes of
demonstration. Under Grid layer: choose “Grid 2”, which is the
one with the grid points closer together.
• Now click OK to close the Design Properties window. The
properties window closes and we are back with the Design
Details.

Example 4 - Automated Generation of New

Surveys
• Click the Run button on the Design Details window.
A window pops up offering you two choices: (1) Calculate coverage
probability statistics, and (2) Generate a new Survey.
• Choose the second option, and give the new Survey a useful name
like “150 points survey” and the new layer a name like “150
points”.
• Then click OK.
A Survey Details window opens, and the status bar at the top of the
Distance window says “Running Survey”. At this point you have to be
patient while the survey runs. Distance is creating a set of randomly
located survey points, based on the design. When it is finished, the

User's Guide Distance 6.0 Beta 5 Chapter 3 - Getting Started • 29

 Survey Details Results tab opens, and you can review some
statistics about the new survey.
• Click the Next button to see a map of the points – you should be
able to see that there are more in the Western strata (Baja) than the
eastern.
• Click Next again to see a list of the points. You can copy this into
the windows clipboard by typing CTRL-A (control key and A key
at the same time – select all), then CTRL-INS (control and insert –
copy to clipboard). You can then paste this list into a spreadsheet,
document, etc.
• Click on to close the Survey Details window
• Click on the Surveys tab of the Project Browser. You can see
that your new survey has been added in the table of surveys.
• Select the survey and click the Show Details button you get
back to the Survey Details window’s Results tab. You are
automatically taken to the Results tab of a details window that has
a green status light.
• Click on the Inputs tab and then Properties button.
• Click on the Data Layers tab, and you can see that the new
Sample data layer “150 points” has been entered as the lowest
sample layer.
• Close the Survey Properties and Survey Details windows,
and click on the Data tab of the Project Browser. You can see
that the new sample data layer “150 points” has been added below
the “MexStrat” data layer.
There is another way to generate a new survey from a design. You can do it
from inside the Survey tab of the Project Browser.
• Click on the Survey tab of the Project Browser.
• Click on the New Survey button .
A new survey is created in the Survey Browser table, with the name “150
points survey 1”. The first column of the survey table shows a status light,
which is grey: the survey has not been “run” yet. The second column gives the
ID number of the survey. The third shows which Design the survey is related to
– if you hold your mouse over that number it will give you the name of the
design.
Now, you want to run the survey.
• Click the Run selected surveys button .
The status light turns to a running person, and after a while turns green to show
that the survey has been run. You can now compare the results of this survey
with the last one:
• Highlight both surveys in the Survey Browser. To do this click
on one of them, hold the CTRL key and click on the other.
• Now click on the Show Details button . The Survey Details
for both surveys open, on the Results pages.
• Resize and move the windows so that they are lined up side by
side, and click the Next buttons so that both maps are showing.
As you can see, both surveys are realizations of the random survey design.

30 • Chapter 3 - Getting Started User's Guide Distance 6.0 Beta 5

 Example 4 - Design Statistics
• Go back to the Design Details window for your design, and click
Run again.
• This time, choose the top option (Calculate probability of
coverage statistics) and click OK.
You have to wait while Distance generates multiple simulated surveys and uses
these to work out the probability that each grid point will be covered by the
survey. This takes much longer than generating a single survey. When it has
finished, you can see the results in the Results tab.
• Click Next to see a map of the estimated coverage probabilities.
In theory, this design should produce an even probability of coverage within
stratum. However, you can see that there is considerable variation. Why is this?
What would happen if you repeated the run with more simulation runs (say 500,
or 5000)?

Example 4 - Further Investigations

If you want to, go ahead and try out some of the other designs. For example, try
a systematic grid of points. Systematic designs produce more even distribution
of samplers than simple random designs, and we usually recommend them for
that reason. This, and other survey design issues are discussed in Chapter 7 of
both Buckland et al. (2001) and Buckland et al. (2004).
For more background about the survey design features of Distance, see this
Users Guide, Chapter 6 - Survey Design in Distance.

Sample Projects
Distance comes with a number of sample projects, listed in the table below. The
projects located in the “Sample Projects” folder, below the Distance program
folder (usually “C:\Program Files\Distance 6”, or the equivalent in your
language). They demonstrate various aspects of the program – for more
information about a project, open it in Distance, and choose the menu item File |
Project Properties.
Feel free to use the sample projects as a test bed for learning about the program –
try adding and deleting data, creating and running designs and analyses. Have
fun!
Project name Description
Line transect example Simulated line transect data from Chapter 4 of Introduction
to Distance Sampling. Exact data, individuals as clusters
and no stratification. Step-by-step instructions for setting
up a project identical to this and importing the data are
given here in Example 1: Using Distance to Analyze Simple
Data.
Point transect example Simulated point transect data from Chapter 5 of Introduction
to Distance Sampling. Exact data, no clusters or
stratification.
Stratify example An example of stratified data. Two strata, one with high
sample coverage and one with low. Distances are exact and
objects are clusters. The example is based on cetacean data,
and there is also a multiplier defined to account for g(0)<1,
based on a separate experiment to estimate g(0).
Ducknest An example of how to enter and analyze interval data.
There are also 8 example analyses set up, in two sets. Data
are a subset of the Monte Vista duck nest data used as in

User's Guide Distance 6.0 Beta 5 Chapter 3 - Getting Started • 31

 illustrative example in section 8.4 of Introduction to
Distance Sampling. The project is also set up with a suite of
analyses, data filter and model definitions to illustrate a
possible approach to naming and organizing these analysis
components.
House wren Point transect data on house wrens (Troglodytes aedon) in
Colorado used in section 8.6 of Introduction to Distance
Sampling. The project illustrates stratification, use of
additional fields (observer) and a multiplier (number of
visits) – import of these data is covered here in Example 2:
More Complex Data Import.
Songbird Point transect data from a multi-species survey of songbirds
in Colorado used in section 8.7 of Introduction to Distance
Sampling.
Golftees St Andrews golf tee data in the format used as a running
example to illustrate double-observer data in Chapter 6 of
Advanced Distance Sampling.
Amakihi An example of multiple covariate distance sampling data
and analyses. Point transect data on Hawaii Amakihi, a
generalist Hawaiian honeycreeper. There are three potential
covariates: observer (obs), time in minutes (MAS) and
hours (HAS) after sunrise. See the Project Properties for
more information. This data set is the illustration data used
in Marques et al. (2007), and the project contains the
analyses presented in Table 2 of that paper.
StAndrewsBay An example geographic project that can be used for survey
design, comprising geographic data for the waters just off St
Andrews, Scotland. Step-by-step instructions for creating a
project like this one from scratch, including importing the
geographic data, and also for creating example survey
designs are given in this Chapter under Example 3: Using
Distance to Design a Survey.
Mexico An example geographic project that can be used for survey
design, comprising geographic data for 4 states in North-
Western Mexico.
LinkingExample An example of how to link external databases and text files
to a Distance project – an advanced technique outlined in
Linking to Data from Other Databases.

Some of the projects contain data used in the distance sampling text
books, and you can use these to recreate analyses in the books as a learning
exercise. Note, however, that you may find minor differences in results between
the book and the distance projects. In some cases the data in the projects are
slightly different; in others differences will be due to changes in the Distance
analysis engine since the books were published.

If the sample projects become mangled beyond recognition, you can

restore them to their original state by re-installing them from the Distance setup
program. Run the setup program, and in the window called “Select
Components”, uncheck all the options except Sample projects.

32 • Chapter 3 - Getting Started User's Guide Distance 6.0 Beta 5

Chapter 4 - Distance Projects

Introduction to Distance Projects

In Distance, you store all of the information about one study area in a project.
There are two types of project: geographic projects, which contain geographic
(GIS) data, and non-geographic projects, which do not. If you want to work with
the survey design facilities in Distance then your project must be geographic; if
you’re only interested in data analysis then it can be either.
A project is made up of two parts:
• a project file, which contains information about the project
settings, survey designs, analysis specifications and results.
Project file names always end in .dst – e.g., Ducknest.dst. You can
open a project file in Distance by double-clicking on it.
• a data folder (directory), which contains the survey data, effort
data, and other related information. Data folder names always have
the same beginning as the associated project file, but end in .dat –
e.g., Ducknest.dat.
The data folder in turn contains one or more files, and optionally one folder:
• the data file, DistData.mdb. This file contains information about
how the data is stored, and may contain some or all of the data
itself.
• in addition, some of the data may be stored in other external data
files
• if the project is geographic then the data folder will contain one or
more shapefiles. A shapefile is a standard format for storing
geographic information, invented by the GIS company ESRI. Each
shapefile is actually 3 separate files: an .shp file, a .shx file and a
.dbf file. (In addition, there may be other files such as .prj files.)
• if the project contains analyses that have been run using the
statistical software R (e.g., those using the MRDS analysis engine),
it there will be an R folder. This folder contains a R object file
(called .Rdata), and files of images produced by R. (For more about
R and the R folder, see R Statistical Software in Chapter 7).

User's Guide Distance 6.0 Beta 5 Chapter 4 - Distance Projects • 33

 The structure of a Distance project.

In the Windows interface, project files have the following icon:

Distance project icon

Double clicking on files with this icon in Windows opens the associated project
in a Distance session. Windows also stores a list of your recently used projects
in the Windows taskbar Start menu, under Documents.
You can use Windows to rename, move, copy and delete project files just like
any other file – but if you do this you should do the same to the data folder. For
example, if you want to copy a project from one computer to another, you should
copy both the project file and the data folder.

If you’re moving projects between computers, it’s best to pack them

up into one file first, using the export facility in Distance (see Exporting,
Transporting and Archiving Projects.)

Creating a New Project

To create a new project in Distance, choose File | New Project on the main
window menu. Alternatively, press the New Project button on the toolbar, or
use the keyboard shortcut CTRL-N.
The New Project dialog will open, prompting you to choose a filename for
saving the Distance project file. This filename also forms the basis for the title
of the distance project, so it is best to choose a short, descriptive name such as
“Yellow Warbler Site 1”. Any characters that make an acceptable Windows
filename (e.g., spaces and numbers) are allowed.

You can change the default folder that Distance uses to create and
open projects by selecting this folder in a New Project or Open Project

34 • Chapter 4 - Distance Projects User's Guide Distance 6.0 Beta 5

 dialog and then checking the box Save this folder as default for Distance
projects.

After you have chosen a filename, click the Create button. The new project file
is created and the Setup Project Wizard opens. This wizard guides you through
the process of setting up the new project ready for use. Using the wizard, you
can:
• set the project up ready for doing data analysis
• set the project up ready for survey design
• use an existing Distance project as a template
• import data and options from a previous version of Distance
• bypass the wizard and set up the project manually
See the Setup Project Wizard section in the Appendix - Program Reference for
more about these options. Use of an existing project as a template is also
described in the next section. For more about project import, see Importing from
Previous Versions of Distance.
If you choose to set the project up ready for survey design or data analysis, or
use an existing project as a template, then once you complete the wizard, you are
ready to start entering or importing data. See Chapter 5 - Data in Distance for
more information about how this is done.

Using an Existing Project as a Template

Distance can automatically set up a new project by copying the project settings,
data structure, and design, survey and analysis specifications from an existing
project. All that is then needed are the new data, which can be brought in by the
Import Data Wizard.
Examples where this facility is useful include:
• where you want to set up multiple projects with an unusual data
structure (such as extra data fields not created automatically in the
Setup Project Wizard).
• where you repeatedly use a standard set of analysis specifications
in your projects
• where you want to make it easy for naive users to set up projects
with a standard data structure and, and give them some example
analyses specifications

You can use any Distance project as a template, but you may find it
easier to save the projects used as templates to a special folder, to make it easy to
distinguish them from your other projects. You can save a project to another
folder using File | Export Project - see Exporting, Transporting and Archiving
Projects. When exporting projects to the templates folder, you can save space by
choosing the options to exclude the data and results from the exported project.

By default, when you choose Use an existing Distance project

as template, Distance looks in the folder “Templates” under the Distance
program folder. You can change this by selecting the new folder you want to
use, and checking the box Save this folder as default for opening
template files.

User's Guide Distance 6.0 Beta 5 Chapter 4 - Distance Projects • 35

Opening an Existing Project
To open an existing project, choose File | Open Project on the main window
menu. Alternatively, press the Open Project button on the toolbar or use the
keyboard shortcut CTRL-O. The Open Project dialog will open, prompting
you to choose the project to open.

You can change the default folder that distance uses to create and open
projects by selecting this folder in a New Project or Open Project dialog and
then checking the box Save this folder as default for Distance projects.

When you open a Distance project, the Project Browser is automatically restored
to the same size, position and tab as when you last closed the project.
You can also open projects that have been archived in a zip file (see Exporting,
Transporting and Archiving Projects for more about archiving projects). In the
Open Project dialog, select Zip archive files (*.zip) from Files of type:.
Distance will then prompt you for the folder to extract the project files into, and
will then open the extracted files.

You cannot open a project if any of the files in it are read-only. If

you try to do this you will get an error message. This means that you cannot
open a project directly from a CD – copy it to your hard drive first.

While you’re working with a project, Distance reads and writes to the
project and data files constantly. It is therefore much better to keep your project
on your local hard drive, where access times are much faster. We don’t
recommend accessing projects over a network if you can help it, and we
certainly don’t recommend working with projects stored on floppy disks, zip
disks, etc. In all cases, you’re best to copy the project to your local hard drive
before opening it.

Saving and Backing Up a Project

Saving Projects
In Distance, almost all of the changes you make are “live”- that is, they are saved
in the Distance Project the instant you make them. For example, if you add
some new data or create a new analysis, this data or analysis is instantly recorded
in the project file. Because of this, there is no need to “save” distance projects in
the same way that you might save word processor files. Everything is
automatically saved for you as you go along.
Even though you don't need to save your work, it is important to make backup
copies. This is discussed in the next section.

Backing up Projects
Why Back Up?
Having backup copies of your Distance projects can be useful for two reasons:

36 • Chapter 4 - Distance Projects User's Guide Distance 6.0 Beta 5

 1. If the Distance program quits suddenly, for example if there is a
power cut or a program crash, there is a small chance that your
original project may become corrupted and unusable.
2. If you make a mistake in the Data Explorer, such as deleting a
column of data, there is no way to get this data back. It is instantly
deleted from your project, and there is no undo facility.
In both cases if you have a backup copy of your project, such mishaps need not
become disasters!
Distance has an automatic backup facility that is designed to deal with the first
type of need. There is also a facility for making manual backups and restores
that can be used for the second case. These are detailed below.
Automatic backups in Distance
When you open a project, Distance automatically creates a backup copy of the
project file and project folder. These are created in the same directory as the
original and has the same names, except that the symbol “~$” is appended. For
example if your project file is “MyProject.dst”, the backup project file will be
“$~MyProject.dst”, and the backup data folder “$~MyProject.dat”. (Next time
you open a project in Distance, you can use the Windows Explorer to see this
backup file being created.)
When you close a project, Distance automatically deletes the backup copy of
your project file. If, however, Distance exits suddenly without closing the
project, the backup copy is not deleted. This means that even if your original
project file is corrupted, the most work you can loose is the work you did in that
session.

You can turn off this auto-backup feature in the Preferences dialog.

Automatic restore from backup

Whenever you ask Distance to open a project, it first checks for the existence of
a left-over backup file. For example if you ask to open the file “MyProject.dst”,
Distance will check for the existence of “$~MyProject.dst”. If it finds a backup,
it will display the following message:
[picture of backup file message goes here]
If you choose Yes, Distance will open the backup (and, in the course of opening
the backup will make a backup of that, called “$~$~MyProject.dst”!).
If you choose No, Distance will try to open the original project file. During the
opening procedure, it will overwrite the old backup file. If the original project
file is corrupted you will therefore have no backup. Because of this, it is not
recommended that you choose No unless you are sure that the original file is not
corrupted.
Rather than using the automatic restore feature in Distance, see Recovering From
Unexpected Program Exit in Chapter 12 for a better recovery strategy.
Manual backup
Distance has no undo facility. If you accidentally delete some data or analyses,
they are gone for good. Because of this, it is wise to make regular backups of
your project, especially before you make any major changes.
To backup your project in its current state, choose the menu option File | Copy
project to backup, or press the icon, or use the keyboard shortcut CTRL-S.
Manual restore from backup

User's Guide Distance 6.0 Beta 5 Chapter 4 - Distance Projects • 37

 If you made a mistake, such as accidentally deleting some data, and you have a
recent backup, you may want to revert to the backup copy. To do this, choose
the menu option File | Revert to Backup Copy.
Permanent backups
As explained above, the backups created automatically by Distance, or manually
using Copy project to backup, are destroyed when the project is closed.
Therefore it is worth occasionally making an archive backup of the project, and
possibly storing this archive on another computer or semi-permanent medium
such as CD. To do this, choose the menu option File | Export Project. See
the next section Exporting, Transporting and Archiving Projects for details.
A final note on backups
At the risk of stating the obvious... Don't rely on Distance (or any other single
piece of software) to keep the only copy of your precious data! If you have
paper data sheets, make a photocopy and keep them at another location. If you
entered your Distance data via the keyboard, use the Copy to Clipboard
facility in Distance to copy the data into a spreadsheet or text file, and keep that
file on another computer at a different location.

Exporting, Transporting and Archiving Projects

Distance allows you to export projects to another location on your computer. As
well as simply copying the project file and associated data folder, Distance can
export the project to a single, compressed archive (zip file). In addition, you can
choose to exclude certain parts of the project from your export - for example to
exclude the data, or the results.
This facility is useful in three contexts:
1. Making a permanent backup of a project. In this case, you would
probably want to export the project to a zip file, to save disk space.
You can then archive the zip file onto a CD, tape, disk or backup
computer.
2. Transporting projects between computers. Again, you will
probably want to export the project to a zip file – both to save disk
space and because a single file is more convenient to move around
than a distance project.
3. Making templates for setting up new projects. Here, you will want
to export the project as a Distance project. For more about
templates Using an Existing Project as a Template.
To export a distance project, select the menu item File | Export Project. For
more about the options, see the Export Project Dialog section of the Program
Reference.

Only the distance project file and files in the data folder are
exported. If your project links to GIS files or other database files outside the
data folder, these will not be included

To transport a project between non-networked computers, you can

export it direct to floppy disks. As projects tend to be large, it is best to export to
zip archive. If the zip archive is too large to fit on one disk, Distance will
automatically span the archive across multiple disks.

38 • Chapter 4 - Distance Projects User's Guide Distance 6.0 Beta 5

 When transporting projects (e.g., on floppy or over the internet), you
can save even more space by excluding the results from the export as these can
be re-created on the new machine – although if your results include time-
consuming operations such as simulations or bootstrapping, you probably don’t
want to do them again!

Viewing and Editing Project Properties

You can view general information about your project in the Project
Properties Dialog. This information includes the name, location and size of
the project file, and the contents of the data folder. There is also space for you to
make notes about the project. The Geographic tab allows you to turn a non-
geographic project into a geographic one (but not visa-versa), and to set the
default coordinate systems for the project.
To open the Project Properties Dialog, select the menu item File | Project
Properties. For more information about the dialog options, see Project
Properties Dialog in the Program Reference.

Compacting a Project

In a Distance project, the Project File and Data File are actually database files.
Like all database files, they tend to grow as you use them. This is because
records and queries that Distance no longer needs are marked as deleted, but are
not actually removed from the database. Permanently removing deleted records
is called “compacting” a database. To do this, the database must be closed.
Distance projects are automatically compacted when you close the project.
However, if you work with an open project for a long time, you might want to
compact it occasionally. To do this, choose Tools | Compact Project.
Distance will close the project, compact it, and open it again.

Compacting a project is also a way of fixing database corruption. For

more details, see Fixing a Corrupted Project in Chapter 12.

Importing from Previous Versions of Distance

Importing Distance 4 and 5 Projects

Distance 6 can import all the information contained in Distance 4.0 – 5.0 project
files, including the data, designs, surveys, analyses and results.
To import a Distance 4 or 5 project, either:
• Open the Distance 4 or 5 project from inside Distance 6. A dialog
box asks you whether you would like to upgrade the project, and if
so whether you would like to save the upgraded project under a
new name. If you click Yes, you are prompted for the name of the
new project file. The new project is created and the information
from the old project is imported. The old Distance 4 project
remains unchanged in its original location. If you click No, the old

User's Guide Distance 6.0 Beta 5 Chapter 4 - Distance Projects • 39

 Distance 4 or 5 project is upgraded in situ, and can no longer be
opened in Distance 4 or 5.
• Create a new project from inside Distance 6. In the project setup
wizard, choose the option to Import a project or command file
created in a previous version of Distance, and press Next.
Distance then prompts you for the location of the old project file. If
you select a Distance 4 project and click Finish, all information
from that project is imported into the new project.

Importing Distance 3.5 Projects

Distance 6 can import data and project settings from Distance 3.5 project files.
It will not import the data filters, model definitions or analyses.
To import a Distance 3.5 project, either:
• Open the Distance 3.5 project from inside Distance 6. A dialog
box prompts you for a filename for the new Distance 6 project.
When you press OK, the project data and settings are imported into
the new project, and this new project is opened. The old Distance
3.5 project remains unchanged in its original location.
• Create a new project from inside Distance 4. In the project setup
wizard, choose the option to Import a project or command file
created in a previous version of Distance, and press Next.
Distance then prompts you for the location of old project file. If
you select a Distance 3.5 project and click Finish, the project data
and settings are imported into the new project.

Importing Distance 2.2 - 3.0 Command Files

This version of Distance cannot import Distance 3.0 or earlier command files.
This facility will be added to the full version before release. Meanwhile, to
import old command files, you first have to import them into Distance 3.5, and
then import the Distance 3.5 command file.
Alternatively, if the data is already in flat-file format, or is stored in a separate
spreadsheet or database, you could consider setting up a new project and using
the data import facilities in Distance – see Data Import in Chapter 5.

40 • Chapter 4 - Distance Projects User's Guide Distance 6.0 Beta 5

Chapter 5 - Data in Distance

Data Structure
This section describes how your survey and associated data are represented in
Distance. It is essential reading for anyone using the program. There is quite a
lot of material and jargon to absorb, but it is important to understand the
concepts here in order to make efficient use of the software.

Data Layers
Introduction
Data in Distance are divided into a set of nested data layers. Each data layer can
be thought of as a database table, with records (rows) and fields (columns). Data
layers have three attributes associated with them:
• Layer Name (e.g., Study area, Point transect, New Layer 1) – this
is a description of the layer. You can change this from the default
to make it more relevant to your study.
• Layer Type (e.g., Global, Sample, Coverage) – this is a
description of the function of the layer, and its place in the
hierarchy of layers. The layer type is used internally by Distance
and once it is set you cannot change it.
• Geographic (Yes/No). If the project is geographic, then each
data layer can contain geographic information, although not all
layers have to. You can tell if a layer is geographic because it will
contain a Shape field (see Data Fields, below).
Hierarchy of Data Layers
Data layers are linked together in a hierarchy, with a layer of type Global at the
top, and other layers below it. Here’s a simple example:

Example set of data layers from a Distance project

In this picture, the icon (symbol) tells you the layer type, while the text tells you
the data layer name. The top layer, Study Area is of type Global. It has one
child layer, Region of type Stratum. This in turn has one child layer, Line
transect, of type Sample. Lastly, line transect has one child layer, Observation
of type Observation.

User's Guide Distance 6.0 Beta 5 Chapter 5 - Data in Distance • 41

 Multiple Child Layers
Data layers can have more than one child layer. In the following example, the
Global layer, called North Pacific, has two child layers. The first, Management
Area, is of type Stratum. The second, Coverage Grid, is of type grid. The
Management Area layer in turn has two child layers, 1999 Transect and 2000
Transect, both of type Sample.

More complex example data layers from a Distance project

Nested Stratum and Sample Layers

In the above example, the two Sample layers, 1999 Transect and 2000 Transect,
both have layers below them of type SubSample1. This is because in this case
the transect lines are divided up into segments. Each segment represents one day
of shipboard surveys. By storing the effort data in separate segments, we can
choose at the analysis stage whether to treat each transect as an independent
sample, or each segment. For more on this, see Chapter 7 - Analysis in Distance.
In general, Sample layers can have up to 5 sub-sample layers below them:
SubSample1 – SubSample5. Similarly, Stratum layers can have up to 5 sub-
stratum layers below them. We hope to incorporate the ability to analyze survey
designs with multiple nested stratum and effort layers into a future version of
Distance.

List of Data Layer Types

A full list of layer types, together with their associated icon and the allowable
child layer types is given below:

Layer type Icon Allowable child layer types

Global – has only one Stratum, Sample, Coverage, Other
record; contains information
that applies to the whole
study
Stratum – contains one SubStratum1, Sample, Other
record for each stratum
SubStratum1 SubStratum2, Sample, Other
SubStratum2 SubStratum3, Sample, Other
SubStratum3 SubStratum4, Sample, Other
SubStratum4 SubStratum5, Sample, Other
SubStratum5 Sample, Other
Sample – contains one Observation, SubSample1, Other
record per sample (e.g.,
transect)
SubSample1 Observation, SubSample2, Other
SubSample2 Observation, SubSample3, Other
SubSample3 Observation, SubSample4, Other
SubSample4 Observation, SubSample5, Other

42 • Chapter 5 - Data in Distance User's Guide Distance 6.0 Beta 5

 SubSample5 Observation, Other
Observation – contains one Other
record per observation
Coverage – used in survey Other
design module, for storing
probability of coverage.
Contains one record per
coverage grid point.
Other – omnibus layer type Other
used to store miscellaneous
data

Data Fields
Introduction
As we said earlier, each data layer can be though of as a database table, with a
number of fields (columns) and records (rows):

Example Stratum layer table

Each field has four attributes associated with it:

• Field Name (e.g., Area) – this is a description of the field. You
can change this to almost anything you like (although some words
and characters are not allowed – see Valid Field Names in the
Inside Distance Appendix). (Note – you can only change the field
names of internal fields – see Source Database Type, below).
• Field Type (e.g., Integer) – the type of data contained in records
in that field. This is set when the field is created and you cannot
change it. The possible field types for fields you can create are:
Text, Integer, Decimal, Label. Other field types you can see are ID
and Shape. More on these later.
• Units (e.g., Hectare) – the units of measurement of the data, where
appropriate. A large number of units are available. For fields
where there are no units, this should be left as n/a (not applicable).
• Source Database Type – tells you where the data field is
located. Int (internal) means its in the project Data File; Geog
(geographic) means its in a shapefile, and Ext (external) means its
located in some external database.

The ID Field Type

Each layer has a field of type (and name) ID. This is used to identify each record
in the layer. ID fields cannot be edited or deleted by the user.
A second function of the ID fields are to link together records in different layers.
In the following example, the sighting with ID 40 in the Observation layer is

User's Guide Distance 6.0 Beta 5 Chapter 5 - Data in Distance • 43

 contained within the transect with ID 14 in the Line transect (Sample) layer,
which in turn is contained in the region with ID 2 in the Region (Stratum) layer.
The Global layer always contains just 1 record with ID 1, which encompasses
the whole study site.

Example data sheet.

Notice that there are no Observation records opposite the Sample

(Line transect) records 16, and 17. This is because no animals were sighted on
those transects.

The Label Field Type

You may also notice in the above picture that all of the data layers except the
Observation layer have a data field with field type (and name) Label. This field
is always created by default and can be used to name each record (e.g., "Ideal
habitat" and "Marginal habitat" in the Stratum data layer). If you manually add
new data layers to the project, you can add a new field of type Label if you want.
The Shape Field Type
If the data layer is geographic, then it will contain a field of type (and name)
Shape. The records for this field contain a word describing the type of shape
that is stored in them (Polygon, Line or Point). You can double-click on the
cells in the data sheet to edit the shape in the Shape Properties Dialog.
Other Hidden Field Types

There are two field types that you won’t see from the Distance data sheet:
LinkID and ParentID. LinkID fields are used to join geographic and external
data to that stored in the Data File DistData.mdb. ParentID fields are used,
together with the ID field, to link together records from different layers. You
don’t normally have to worry about either of these fields, unless you are editing
the distance database by hand from outside of Distance – for more on this see the
Appendix - Inside Distance.

44 • Chapter 5 - Data in Distance User's Guide Distance 6.0 Beta 5

 Changing the Data Structure
If all you want to do is analyze a straightforward data set, using conventional
distance sampling methods, then you will probably never need to add new data
layers or fields to your project file, beyond those created automatically by the
Project Setup Wizard. You may want to rename the layers and fields, which you
can do easily by double clicking on the names in the Data Explorer.
However, in most other cases, you will want to change the data structure, by
adding new layers and fields, and sometimes deleting layers and fields. Here are
some examples:
In analyzing a multi-species survey, you may want to add an additional field for
Species into the Observation layer. You can then do different analyses, selecting
out only the species of interest.
If you are using the Multiple Covariates Distance Sampling (MCDS) analysis
engine, you will want to have extra fields for the covariables you will be adding
to the detection function.
Before you can use the survey design module to examine probability of
coverage, you need to add a layer of type Coverage to the project, and populate it
with a grid of points.
The survey design module can be used to create new survey plans, which are
stored as new data layers. After a while you may find you have too many layers
lying around, and want to delete some to clean up.
All of the work of adding, renaming and deleting data layers and fields can be
done from the Data Explorer. Advanced users can also change the data structure
from outside of Distance, by directly editing the Data File, DistData.mdb, using
a database package.

Getting Data Into Distance

There are currently five ways to get data into Distance:
• Enter data from the keyboard, using the Data Explorer.
• Enter data from the keyboard, using the Data Entry Wizard.
• Import data from a text file, using the Import Data Wizard.
• Import data from a previous version of Distance, using the Setup
Project Wizard.
• Link an existing database file (including GIS shapefile) to the
Distance database, by editing the Data File by hand.
While it is possible to enter data from the keyboard, it is relatively slow and
inefficient for all but very small datasets. We did not have the time or resources
to completely reinvent the spreadsheet! Therefore, for anything other than small
datasets, you will likely want to enter and store your data in a separate
spreadsheet or database application and use the Data Import facility to bring the
data into Distance. This has the added advantage that your data is backed up
elsewhere should things go wrong!
The following two pages cover Data Entry and Data Import in more detail. For
more information about Importing data from previous versions of distance, see
Importing from Previous Versions of Distance in Chapter 4. For more about
linking external data, see Advanced Data Topics, below.

User's Guide Distance 6.0 Beta 5 Chapter 5 - Data in Distance • 45

 Data Entry
Distance provides two very similar forms for data entry via the keyboard: the
Data Entry Wizard and the Data Explorer.
The Data Entry Wizard guides the user through the process of data entry. It
provides on-screen advice via a text window at the top of the wizard, and moves
through the data one layer at a time. It is most suitable for beginning users, but
its limitation is that it can only be used on simple datasets – those with four data
layers (Global, Stratum, Sample and Observation). The Data Entry Wizard
opens by default at the end of the Setup Project Wizard, if you are setting up a
project for data analysis.
The Data Explorer is not so structured - the user is free to jump among the data
layers, and there is no panel of advice. It is more suited for checking data, and
for data entry by more advanced users.
For more specific information about the layout of these forms, and how to use
them, see the Project Reference pages on the Data Explorer and Data Entry
Wizard.

Data Import

This section contains information about importing non-geographic

data. For information about GIS data in Distance, see Geographic (GIS) data,
and for information about importing this data see Importing Existing GIS data.

Introduction to Data Import

Distance can import data from text files straight into your Distance project. The
data in the text file should be in “flat file format” - i.e., arranged in rows and
columns.
The simplest case is where you want to import survey data into a new project,
after the project has been set up for analysis by the Setup Project Wizard. In this
case, the project will contain four data layers: Global, Stratum, Sample and
Observation (these are the layer types, the layer names will be different – e.g.,
the Global layer is called “Study area” by default). At a minimum, your text file
should contain the following columns:
• stratum label column (you can get away without this – see below)
• sample (transect) label column
• distance column
However, under most circumstances you will want to include other columns
such as:
• stratum area
• sample effort (transect length) - for line transects
• angle - when radial distance and angle are measured
• cluster size - when objects are clusters
The columns should be separated by a delimiter (ASCII character), which can be
either a tab, semicolon, comma or space. The order of the columns is not
important, as you tell Distance which column is which during the import
process. Each row should finish in a Carriage-return + Line-feed combination.
This is the default end-of-line indicator used by most windows-based
applications, so you usually don't have to worry about this.

46 • Chapter 5 - Data in Distance User's Guide Distance 6.0 Beta 5

 While the order of the columns is not important, the order of the rows is. Before
importing data into Distance, you should sort by stratum label (if you’re
importing more than one stratum), then sample (transect) label. This ensures
that all data from the same strata are together, and within this all data from the
same sample (transect) within strata are together. The order of observations
within samples is not important.
The following is an example of a semicolon delimited input file. The first
column is the stratum label, the next is the stratum area, then the transect label,
transect length and finally the perpendicular distance.
Stratum A;100;Line 1A;10;14
Stratum A;100;Line 1A;10;8
Stratum A;100;Line 1A;10;22
Stratum A;100;Line 2A;10.3;7
Stratum A;100;Line 2A;10.3;37
Stratum A;100;Line 2A;10.3;13
Stratum B;123;Line 1B;5.7;
Stratum B;123;Line 2B;8.4;27
Stratum B;123;Line 2B;8.4;76
Stratum B;123;Line 2B;8.4;44
Stratum B;123;Line 2B;8.4;7

Notice that the record “Line 1B” has no distance in the final column - this is a
transect where no objects were seen.
Notice also that all transects from the same stratum are grouped together, and all
observations from the same transect are grouped together. If you accidentally
forgot to sort the data before importing it, so that for example the first four lines
looked like this:

Stratum A;100;Line 1A;10;14

Stratum A;100;Line 1A;10;8
Stratum A;100;Line 2A;10.3;7
Stratum A;100;Line 1A;10;22

then Distance would interpret this as three transects with labels “Line 1A”, “Line
2A” and “Line 1A” again.

Distance treats each successive delimiter as indicating a new

column. This means that you cannot import fixed-format data in the current
version of Distance. Here's an example of some fixed-format data, where the
stratum field is columns 1-10, stratum area is columns 10-20, etc.

Stratum_A 100 Line_1A 10 14

Stratum_A 100 Line_1A 10 8

If you wanted to import this data, you would have to find some way to delete the
multiple spaces before importing it as space delimited:

Stratum_A 100 Line_1A 10 14

Stratum_A 100 Line_1A 10 8

In Distance, you are guided through the import process by the Import Data
Wizard. This wizard can be started in one of two ways:
• from the last page of the Setup Project Wizard, by choosing the
option Proceed to Import Data Wizard. This is the ideal way
to import data into a new project.
• by selecting the menu item Tools | Import Data Wizard. This is
the best way to add extra data from file into an existing project.
You can also replace your existing data with the imported data -
this is an option at the end of the Import Data Wizard.

User's Guide Distance 6.0 Beta 5 Chapter 5 - Data in Distance • 47

 You can streamline data import by having the columns in your text file
in the same order as the fields in the Distance Data Explorer. That way, you can
tick a box in the Import Data Wizard and have the wizard automatically pair up
columns in the text file with fields in the Distance database. See also
Streamlining import of data from one flat file.

Additional information about the Import Data Wizard is given in the Program
Reference page Import Data Wizard. If you are having problems, check the page
Troubleshooting the Import Data Wizard.

Once you have imported your data, you should always double-check
that the correct number of strata, samples (transects) and observations are present
in the Distance project. For example, if you forget to sort the data correctly (by
stratum and sample label)

Importing large datasets into Distance takes a long time. We

hope to improve the performance of the import routines in future releases.

Overview of More Advanced Import Topics

This section contains a brief mention of some more advanced uses of the Import
Data Wizard. Further details of the wizard are in the Import Data Wizard section
of the Program Reference. Some specific import scenarios are covered in the
next topics after this one.
Importing A Subset of Layers
In the Import Data Wizard, you specify which layers you want to import to, and
where the new data records should be located.
For example, this is useful for importing survey data where there is no stratum.
By default, the stratum layer contains one record. So, in the Import Data
Wizard, you specify that you want to import data into only the Sample and
Observation layers, and that the new records should be added below the first
record in the Stratum layer.
Another example where this is useful is when you already have one text file for
each data layer, rather than one large text file containing all layers joined
together.
Creating New Fields
The Import Data Wizard is capable of importing additional columns of data and
will automatically create new fields in the Distance project file to hold them if
the fields do not exist already. Examples are given later in the Users Guide
where additional columns of data are useful - these data include extra columns in
the Sample data layer (such as year of survey in multi-year surveys) and the
Observation data layer (such as sex or species of animal).
Ignoring Columns
The wizard is also capable of ignoring columns in your data text file, so there is
no need to exclude all unwanted columns from the file prior to import.
Appending New Data
You can choose whether to append the imported data to existing data, or to
overwrite the existing data. This is useful, for example, for adding a new year of
data to an existing project file.
Specifying Destination for New Records

48 • Chapter 5 - Data in Distance User's Guide Distance 6.0 Beta 5

 You can specify where the new records will be added, relative to the parent
layer, using the parent layer’s ID or Label field.
For example, imagine you are adding a new year of survey data to an existing
project file. You have covered new transects, so you want to add new records to
the Sample layer, and you have new sightings for the Observation layer. You
want the new transects to go into the correct place below their parent records in
the Stratum layer.
To do this, you add a column to your text file containing the ID of the stratum
for each new transect. You will also need a column saying which year the
records come from – this should go in the Sample layer. In the Import Data
Wizard you specify that you want to import into the Sample and Observation
layer, and that you want to put new records below the ID field of the parent
Stratum layer. You specify that the new data should be appended, and when you
press the Finish button the new data are added to the old, in the correct stratum.
Importing Geographic Information
Unfortunately, there is currently no user-friendly way to import geographic
information into Distance. The only way, apart from typing it in, is to create a
shapefile in a separate package and then link this shapefile to the database – see
Importing Existing GIS Data, below.

Streamlining import of data from one flat file

If you find yourself importing lots of data that are basically similar, then there
are several steps you can take to make the import process quicker. This topic
describes these steps, using an example of a single text file that contains all the
survey data.
Before reading further, you should be somewhat familiar with the Import Data
Wizard, and should read the Program Reference pages on the Import Data
Wizard.
Let’s consider the following data, stored in a text file:
Stratum A;100;Line 1A;10;14
Stratum A;100;Line 1A;10;8
Stratum A;100;Line 1A;10;22
Stratum A;100;Line 2A;10.3;7
Stratum A;100;Line 2A;10.3;37
Stratum A;100;Line 2A;10.3;13
Stratum B;123;Line 1B;5.7;
Stratum B;123;Line 2B;8.4;27
Stratum B;123;Line 2B;8.4;76
Stratum B;123;Line 2B;8.4;44
Stratum B;123;Line 2B;8.4;7

The data are semicolon delimited, and the columns are: stratum label, stratum
area, transect label, transect length and distance. Normally, to get such data into
a Distance project you would:
4. create a new project, going through the Setup Project Wizard,
choosing the option to Analyze a survey that has been
completed, and filling in the options in the successive screens.
5. Proceed to the Import Data Wizard, and specify the appropriate
source file
6. In the Data File Structure screen of the Import Data Wizard,
manually match up the field names with the columns in the text
file.
7. Finish the Import Data Wizard and import the data.

User's Guide Distance 6.0 Beta 5 Chapter 5 - Data in Distance • 49

 This process can be made more efficient using some combination of the
following tips:
• If you already have a project with the data structure you require,
then in the first page of the Setup Project Wizard, choose the option
to Use an existing Distance project as a template. See
Creating a New Project in Chapter 4.
• Rather than manually match up field names with columns in the
text file, make sure the columns are in the same order as the fields
in the Distance database. Then, in the Data File Structure screen,
tick the option Columns are in the same order as they will
appear in the data sheet.
• Another alternative is to put the layer and field name in the first
row of the text file. Then tick the option First row contains
layer names and field names of each column. For example
(the text below is supposed to be on one line, but may be wrapped
on some screens/formats):
Region*Label;Region*Area;Transect*Label;Transect*Length;Observation
*Distance
Stratum A;100;Line 1A;10;14
Stratum A;100;Line 1A;10;8
. . .

The “*” delimiting layer names and field names in the code
can be replaced by other delimiters. Alternatives are * | _ and .
(.e., a full stop or period). During data import, you can choose
from these alternatives the appropriate one for the text file you
are importng.

Importing one file per data layer

Flat files, such as the one used in the example of the previous topic, are useful
ways to store small datasets. However, for large datasets they are inefficient.
For example, in the previous topic, the stratum label and area for stratum 1 was
repeated six times. Imagine if there were 10,000 observations in stratum 1! A
more efficient way to store and import large datasets is to have each data layer in
a separate file, and to import one layer at a time.
Continuing the example from the previous topic, you would have 3 files:
File 1: stratum.txt
Columns: stratum label, area
Stratum A;100
Stratum B;200

File 2: transect.txt
Columns: stratum label, transect label, transect length
Stratum A;Line 1A;10
Stratum A;Line 2A;10.3
Stratum B;Line 1B;5.7
Stratum B;Line 2B;8.4

File 3: observation.txt
Columns: transect label, distance
Line 1A;8
Line 1A;22
Line 2A;7
Line 2A;37
Line 2A;13
Line 2B;27
Line 2B;76

50 • Chapter 5 - Data in Distance User's Guide Distance 6.0 Beta 5

 Line 2B;44
Line 2B;7

Notice that the transect file contains a column giving the stratum of each
transect, and that the observation file contains a column giving the transect of
each observation. In general, each file has to have a column giving an unique
identifier to the record in the parent layer. You don’t need one for the stratum
file because it’s parent, the global layer, has only one record.
To import these data:
• Create a new distance project, with 4 data layers and appropriate fields
– probably using the Setup Project Wizard.
• Begin by importing the stratum layer. Fire up the Import Data Wizard,
and enter “stratum.txt” in the Data Source page.
• Under Data Destination, both the highest and lowest Destination
data layers are the stratum layer (called Region by default). Under
Location of new records, choose the first option Add all new
records under the first record in the parent data layer.
• Under Data File Format, choose semicolon delimited, and in Data File
Structure, match the columns in “stratum.txt” to the fields in the stratum
layer. Click Next and Finish.
• Use the Data Explorer to check the stratum data were imported
correctly.
• Now import the transect file. This time in Data Destination, the highest
and lowest Destination data layers are the transect layer. Under
Location of new records, choose the second option, Input file
contains a column corresponding to the following field in
the parent data layer, and make sure the label field is selected from
the drop down box.
• In the Data File Structure page, match the columns in “transect.txt” to
those in the Distance database, including the stratum label field.
• Import the data, and check it in the Data Explorer.
• Repeat this process for the observation data file.
Non-unique label fields
If the label field is not unique, then you will have to add an extra column
containing the ID of each record. For example, imagine that the transect labels
are not Line 1A, Line 2A, Line 1B, Line 2B, but instead are Line 1, Line 2, Line
1, Line 2. In this case the observation data file will need to be as follows:
File 3: observation.txt
Columns: transect ID, distance
1;8
1;22
2;7
2;37
2;13
4;27
4;76
4;44
4;7

When the transects are created, they are assigned IDs sequentially, so transect
“Line 1” in stratum A will have ID 1, transect “Line 2” in stratum A will have
ID 2, “Line 1” in stratum B will be ID 3, and “Line 2” in stratum B will be ID 3.
In the above file, because the transect labels are not unique, the transect IDs have
been used instead. The only difference in the Import Data Wizard will come in
the Data Destination step, where under Location of new records, the Field
name will be “ID”, rather than “Label”.

User's Guide Distance 6.0 Beta 5 Chapter 5 - Data in Distance • 51

Geographic (GIS) Data
Distance has a built-in GIS (Geographic Information System), which allows it to
store and manipulate spatially-referenced data. Geographic data for spatially-
reference data layers is stored in a shapefile, a widely used data format invented
by the GIS company ESRI (see Chapter 4 - Distance Projects). Geographic data
is used in the survey design part of Distance, and in future releases will be used
in analyses involving spatial modelling of density.
There are two types of Distance project: geographic projects, which can contain
spatial data, and non-geographic projects, which can not. The project type can
be set when the project is created (see Setup Project Wizard in the Program
Reference). To see whether a project is geographic, look under the Geographic
tab of the Project Properties dialog (choose File | Project Properties…).
Not all data layers in a geographic project have to be spatially-referenced,
although all are by default. This is set when the layer is created. To see whether
a data layer is spatially-referenced, in the Data Explorer, either (i) look in the
data sheet for a column with name “Shape”, or (ii) look under the Geographic
tab of the Layer Properties dialog (choose Data | Data Layer Properties…).
Shapes in a spatially-referenced data layer can be one of three types: points, lines
or polygons. The shape type is set when the layer is created. You can see the
shape type of a layer, either by looking under the “Shape” field in the data sheet,
where the shape type is written for each record, or by looking under the
Geographic tab of the Layer Properties dialog.
Geographic data in a data layer can be stored according to a geo-coordinate
system, and projected for viewing and survey design calculations. This is
covered in the section on Coordinate Systems and Projections.
There are two ways to get geographic data into distance: type it in using the
Shape Properties Dialog (see Shape Properties Dialog), or import it (see
Importing Existing GIS Data).
If you are experiencing strange behaviour or error messages after importing GIS
data into Distance, it could be because of geometry problems in the data. For
more on this see GIS Problems in Chapter 12.

Viewing and Manipulating Geographic Data

Viewing Geographic Data in Maps
Geographic data in Distance can be viewed in maps, which are accessable from
the Map Browser – accessed via the Maps tab of the Project Browser. The
Map Browser allows creation of new maps, and allows you to sort, rename,
delete and preview the maps that have been created. For more information see
Map Browser in the Program Reference.
From the Map Browser, select a map and choose View Map to open the map in
a Map window. The map window allows you to add or remove data layers, pan
and zoom, and view information about features on the map (using “Map Tips”).
In the future it will be possible to customize the way each data layer is presented
– the colour, etc of each shape – but this facility is currently not developed. For
more information, see Map Window in the Program Reference.
Maps are also produced during survey design, when using a design to investigate
probability of coverage and to produce example surveys. These maps are
displayed in the Results pages of the Survey Details and Design Details
windows. For more information, see Chapter 6 - Survey Design in Distance.

52 • Chapter 5 - Data in Distance User's Guide Distance 6.0 Beta 5

 Maps in the results pages of Survey Details and Design Details
cannot be customized, but copies can be saved to the Map Browser for
customization by pressing the Add to Map Browser button .

Viewing and Manipulating Geographic Data in the Shape

Properties Dialog
If a data layer contains geographic information, it will have a field called
“Shape” in the Data Explorer. The contents of this field will depend on the type
of shape – either “Point”, “Line” or “Polygon”. Double-clicking on any of these
will open the Shape Properties Dialog, which lists the vertices of the shape and
allows you to edit them. For more information, see the Shape Properties Dialog
page of the Program Reference. You can also use the Shape Properties Dialog to
import shapes from a text file or spreadsheet – for more information, see
Importing GIS Data via the Windows Clipboard, later in this chapter.

Coordinate Systems and Projections

This section provides a brief introduction to the use of coordinate systems in

Distance. For more information, refer to any good book on cartography.

About coordinate systems

There are two types of coordinate systems: geographic and projected.
Geographic coordinate systems use latitude and longitude coordinates to define
the position of a point, line or polygon on the earth’s three-dimensional surface.
Most geographic data is stored in latitude and longitude, according to a specified
geo-coordinate system. Projected coordinate systems use a mathematical
conversion to transform latitude and longitude coordinates to a two-dimensional
surface. Most maps use a projected coordinate system, and calculations such as
distance and area are performed on projected data.
Geographic coordinate systems
Unfortunately for us, the earth is not a perfect sphere, or even a perfect ellipsoid.
Instead, it is nearly ellipsoidal and in addition is covered in lots of small lumps
and bumps. Geographic coordinate systems approximate the shape of the earth
using a reference sphere or ellipsoid. The accuracy of the approximation
depends upon which coordinate system you use and which part of the earth you
use it for.
Projected coordinate systems
Because the earth is round and maps are flat, getting information from a curved
surface to a flat one involves a mathematical formula called a map projection or
simply a projection
This process of flattening the earth will cause distortions in one or more of the
following spatial properties: distance , area, shape, direction.
No projection can preserve all these properties; as a result, all flat maps are
distorted to some degree. Over a small geographic area the distortions caused by
the map projection are not significant, but the selection of an appropriate map
projection is crucial for larger regions. Fortunately, you can choose from many
different map projections. Each is distinguished by its suitability for representing
a particular portion and amount of the earth’s surface and by its ability to
preserve distance, area, shape, or direction. Some map projections minimize
distortion in one property at the expense of another, while others strive to
balance the overall distortion.

User's Guide Distance 6.0 Beta 5 Chapter 5 - Data in Distance • 53

 Coordinate systems and Distance data
Distance can cope with data stored in either a geographic or projected coordinate
system, or neither (sometimes referred to as “non-earth” – in the Distance
interface we use “[None]”). In Distance, a projected coordinate system is
defined as a geographic coordinate system plus a projection together with any
projection parameters required.
The default coordinate system for geographic data in a project is set in the
Geographic tab of the Project Properties dialog.

Data section of the Geographic tab, Project Properties dialog

You set the coordinate system of a data layer when it is created – by default it is
the same as the default coordinate system, but this is not required. (The only
requirement is that all data layers use the same datum.)
Do you need to worry about the coordinate system of the data?
Most geographic data is stored as latitude and longitude according to some
geographic coordinate system. Latitude and longitude are expressed in angular
units (usually decimal degrees). If you want to work with survey design, you
will likely want to work in linear units (e.g., meters), so you will need to
transform your data. To do this you will need to know the geocoordinate
system.
If your data are already expressed in linear units, then likely they are already
projected. This can happen, for example, because you digitized the study area
using a map. So long as you are happy with the projection, then you can set the
geographic coordinate system to “[None]” and forget about coordinate systems.
Similarly, if your study area is small, and you have measured its boundaries
directly, then no coordinate system is required.

Coordinate systems, maps and calculations in Distance

If the data are stored in a geographic coordinate system then you can project the
data for viewing it on maps and performing calculations. All maps must use the
same projection, which is set in the Geographic tab of the Project properties
dialog.

Maps section of the Geographic tab, Project Properties dialog

If the data are stored projected, then this projection is used for all maps and
calculations, while if the data have no coordinate system then they cannot be
projected.
In survey design, a different projection can be defined for each design, so you
can compare the effect of different projections on the results. (See Chapter 6 -
Survey Design in Distance for more on survey design.)

Projecting data takes time. This can significantly affect the

performance of survey design calculations, such as calculating coverage
probability and creating new surveys. Therefore, if you always use the same
projection for a particular set of data, consider projecting the data using an

54 • Chapter 5 - Data in Distance User's Guide Distance 6.0 Beta 5

 external GIS package and storing the projected data in a shapefile. When you
bring the data into Distance, set it as no coordinate system (you can do this by
setting the coordinate system and projection to [None] in File | Project
Properties | Geographic, and setting the units correctly, and then not copying
over the .prj part of the shapefiles when you import your shapefiles – see
Importing Existing GIS data). Distance will then treat the data as simple x,y
coordinates and will not spend time projecting and re-projecting it.

Which projection?
Over a small study area the projection used will make relatively little difference.
Over larger areas the projection can make a significant difference.
If you need to project your data but are not sure which projection to use,
probably the best option is to cheat and refer to maps of your study area to see
what projection they use (the map will usually give the projected coordinate
system – a literature search will reveal the composite projection, projection
parameters and geocoordinate system).
Alternatively, many cartography books describe the properties of the different
projections, which may help you decide which is appropriate. You will need to
consider the following questions:
• Which spatial properties do you want to preserve? Is it just for
displaying the study area, or for performing calculations such as for
survey design, transect distance, etc.
• Where is the study area? Is your data in a polar region? An
equatorial region?
• What shape is the study area? Is it square? Is it wider in the east–
west direction?
• How big is the study area?
Map projection classifications
Map projections can be generally classified according to what spatial attribute
they preserve.
• Equal Area projections preserve area. Many thematic maps use an
equal area projection. Maps of the United States commonly use the
Albers Equal Area Conic projection.
• Conformal projections preserve shape and are useful for
navigational charts and weather maps. Shape is preserved for small
areas, but the shape of a large area such as a continent will be
significantly distorted. The Lambert Conformal Conic and
Mercator projections are common conformal projections.
• Equidistant projections preserve distances, but no projection can
preserve distances from all points to all other points. Instead,
distance can be held true from one point (or a few points) to all
other points or along all meridians or parallels. If you will be using
your map to find features that are within a certain distance of other
features, you should use an equidistant map projection.
• Azimuthal projections preserve direction from one point to all other
points. This quality can be combined with equal area, conformal,
and equidistant projections, as in the Lambert Equal Area
Azimuthal and the Azimuthal Equidistant projections.
Other projections minimize overall distortion but don’t preserve any of the four
spatial properties of area, shape, distance, and direction. The Robinson
projection, for example, is neither equal area nor conformal but is aesthetically
pleasing and useful for general mapping.

User's Guide Distance 6.0 Beta 5 Chapter 5 - Data in Distance • 55

 GIS Data Format
Geographic data is stored in Distance as ESRI shapefiles – a standard, vector-
based GIS format. There are several ways to get GIS data into Distance (see
Importing Existing GIS Data). If you have a GIS, you can probably simply
export data from there into a shapefile format, and use the instructions in the
above section to get the data into Distance. If not, you will probably need to
enter the geographic coordinates into a text file, database file or the Distance
Shape Properties Dialog by hand.
For point and line data, it is quite straightforward to see which order to enter the
points, or the vertices of the lines. For polygons, especially complex polygons
(for example those containing holes), it is less straightforward. If you are
entering data by hand, it is worth first checking the ESRI Shapefile Technical
Description – available from the ESRI web site and also from the Program
Distance Web Site under “Support, Updates and Extras”. An excerpt of this
information is given below, under GIS Format for Polygons.
For more information about checking a shapefile is valid, see GIS problems in
Chapter 12.

GIS Format for Polygons

The following is an excerpt from the ESRI Shapefile Technical Description. The
full text is available from the Program Distance Web Site under “Support,
Updates and Extras”.
A polygon consists of one or more rings. A ring is a connected sequence of four
or more points that form a closed, non-self-intersecting loop. A polygon may
contain multiple outer rings. The order of vertices or orientation for a ring
indicates which side of the ring is the interior of the polygon. The neighborhood
to the right of an observer walking along the ring in vertex order is the
neighborhood inside the polygon. Vertices of rings defining holes in polygons
are in a counterclockwise direction. Vertices for a single, ringed polygon are,
therefore, always in clockwise order. The rings of a polygon are referred to as its
parts.
Because this specification does not forbid consecutive points with identical
coordinates, shapefile readers must handle such cases. On the other hand, the
degenerate, zero length or zero area parts that might result are not allowed.
[…]
The following are important notes about Polygon shapes.
• The rings are closed (the first and last vertex of a ring MUST be
the same).
• The order of rings in the points array is not significant.
• Polygons stored in a shapefile must be clean. A clean polygon is
one that
1. Has no self-intersections. This means that a segment belonging
to one ring may not intersect a segment belonging to another
ring. The rings of a polygon can touch each other at vertices
but not along segments. Colinear segments are considered
intersecting.
2. Has the inside of the polygon on the "correct" side of the line
that defines it. The neighborhood to the right of an observer
walking along the ring in vertex order is the inside of the
polygon. Vertices for a single, ringed polygon are, therefore,
always in clockwise order. Rings defining holes in these
polygons have a counterclockwise orientation. "Dirty"

56 • Chapter 5 - Data in Distance User's Guide Distance 6.0 Beta 5

 polygons occur when the rings that define holes in the polygon
also go clockwise, which causes overlapping interiors.

Importing Existing GIS Data

There are four ways to get geographic information into Distance:
• Enter the vertices (corners) of each shape by hand using the Shape
Properties Dialog (see Shape Properties Dialog in the Program
Reference). Clearly, this is only useful for a small number of
simple shapes.
• Copy the vertices of each shape from a text file or spreadsheet and
paste them into the Shape Properties Dialog. This option will work
well if you have only a few shapes to import (such as just a few
regions) and already have the vertices in some other file.
• Copy an existing ESRI shapefile into the Distance project’s Data
Folder. This will work well if you already have the geographic
data in a shapefile (or can export your data into this format), and
have access to a GIS package for preparing the shapefile.
• Link and existing ESRI shapefile by editing the Data File. This
requires a separate package for preparing the shapefile, and
Microsoft Access for editing the Data File.
The last three options are covered in more detail in the sections below. Before
reading further, you need to understand how data, and particularly geographic
data, are stored in Distance. Make sure you have read all of Chapter 5 - Data in
Distance up to this point, and also read the section How Distance Stores Data, in
the Inside Distance appendix.

Consider projecting your data before importing it, and then importing
it into Distance without the projection – see the tip in Coordinate Systems, Maps
and Calculations in Distance on page 14 for details.

Importing GIS Data via the Windows Clipboard

The Shape Properties Dialog has a facility to copy and paste the vertices
(corners) of an individual shape to and from the Windows clipboard. You can
use this to transfer GIS data between Distance and other formats such as text
files and spreadsheets. For a step-by-step example of importing GIS data from a
text file into Distance, see the Getting Started chapter Example 3: Using
Distance to Design a Survey.
To copy data from a spreadsheet or text file into Distance:
• Highlight the data in the text file or spreadsheet and copy it to the
Windows clipboard.
• In the Distance project you want to copy to, select the shape you
wish to replace in the Data Explorer and double-click on it to open
the Shape Properties Dialog.
• Choose Paste from Clipboard.
To copy data from Distance to a spreadsheet or text file:
• In Distance, double-click on the shape in the Data Explorer. This
opens the Shape Properties Dialog.
• Choose Copy to Clipboard

User's Guide Distance 6.0 Beta 5 Chapter 5 - Data in Distance • 57

 • In your text file or spreadsheet editor, select the area you want to
paste to and choose Paste.
Text file format
Each vertex should be on a separate line, with the x coordinate first, followed by
a tab, followed by the y-coordinate. For example, a text file containing the
following four lines indicates a square:
0 0
0 100
100 100
100 0

To separate the parts of a multi-part polygon, leave a blank line (or have a line
that contains anything other than the above number-tab-number format). For
example the following indicates two triangles:
0 0
0 100
100 0

100 0
100 100
200 0

Spreadsheet format
Each vertex should be in a separate row, with two columns: the first for the x-
coordinate and the second for the y-coordinate. To separate the parts of a multi-
part polygon, leave a row blank.

Example from an Excel spreadsheet, showing data for two triangles. Both columns are
highlighted, ready to copy to the windows clipboard.

Importing GIS Data by Copying an Existing Shapefile

into the Data Folder
In Distance, each geographic data layer has an associated shapefile. By default,
these shapefiles are located in the Data Folder for that project. For example, in
the Mexico sample project, the data layer “Mex” has an associated shapefile
“Mex.shp” (and other files “Mex.dbf”, “Mex.shx” and “Mex.prj”). In this
method, we import the GIS data by copying our shapefiles into the Data Folder,
and then renaming them so they overwrite shapefiles already attached to a data
layer in Distance. That way, next time Distance opens the project, it will use our
shapefiles rather than the original ones. The advantage of this method is that it
does not require the use of a database package to edit the project’s Data File (see
Importing GIS Data by Linking and Existing Shapefile to the Project Data
Folder, below). The disadvantage is that your shapefile has to be copied into the
Data Folder from its original location – so you then have two copies of your GIS
data to manage.
The following instructions assume that you have software to create and edit
shapefiles (e.g., ESRI ArcView). They guide you through the process of
importing a single shapefile to and linking it to a single Distance data layer. If
you are importing more than one shapefile then you need to repeat the same
process for each shapefile. They start by assuming you have created the
Distance project, and have the shapefile you wish to import. The Distance
project must be geographic.
Prepare the Distance Data Layer

58 • Chapter 5 - Data in Distance User's Guide Distance 6.0 Beta 5

 • In Distance, open the project file.
• If you have not yet created the data layer that will hold the
shapefile, then you should do so now.
• In the Data Explorer, click on the Create New Data Layer
button. The Create New Layer dialog box opens.
• Give the layer an appropriate name, and choose the appropriate
parent layer and layer type. Make sure that the options
Create new tables for layer, Create internal data table
within project file and Create shapefile are all ticked.
• Note down the name of the shapefile that will be created. For
example, if the new data layer is called “Antarctica”, the
shapefile will be “Antarcti.shp”
• Click the OK button. The new layer will be created.
• If you had already created the data layer that will hold the
shapefile, open the Data Layer Properties dialog for that layer
(highlight the layer in the Data Explorer, and click the Data
Layer Properties button). In the Geographic data tab, note
down the name of the shapefile (first line).
• Make sure that the data layer contains the same number of records
as the shapefile. For example, if your shapefile contains 5 shapes
(corresponding, say, to 5 strata), then the data layer inside Distance
needs to have 5 records, with ID 1 to 5.
If the data layer does not contain the same number of records (for
example, because you have just created it and it is empty), then add
the appropriate number of new records. (See the Data Explorer
section of the Program Reference for how to do this, if you do not
know.)
• Close the project in Distance.
Prepare the Shapefile
• In Windows, go to the Data Folder for this project. Locate the
shapefile currently associated with the data layer you are going to
import to – in the above example this was called “Antarcti.shp”.
Delete this file, and all other related files – those with the same
name but ending in “.dbf”, “.shx”, “.prj”, etc.
• Locate the shapefile that you have previously created, and wish to
import. Copy this shapefile (and all associated files) into the Data
Folder. Rename the file (and all associated files) so they have the
same name as the shapefile you just deleted. In the above example,
we would rename our shapefile so it is called “Antarcti.shp”,
“Antarcti.dbf”, etc.
• Open the newly renamed shapefile in your GIS package, and add a
field to the table. The field should be able to contain long integer
numbers – in ArcView this means the field type should be
“Number”, the width 16 and decimal places 0. Name this field
“LinkID”.
• The LinkID field will be used to link records in the shapefile to
records in Distance’s internal data table for this layer. Each
LinkID value must correspond with a value in the ID field of
Distance’s internal table.
• If you only just created the Data Layer in Distance, then it
doesn’t matter what order you number the records in the

User's Guide Distance 6.0 Beta 5 Chapter 5 - Data in Distance • 59

 LinkID field of the shapefile. Start with 1, and go up to the
number of records you have.
• If you created the Data Layer previously in Distance, and
already have information in the layer such as Labels, Areas,
etc., then you should take care to number the records in the
shapefile so that the LinkID field of a shape corresponds to the
ID value of that record inside Distance.
• Once you have added records to the LinkID field, you can close the
shapefile and exit your GIS.
Clean up
• You can now reopen the project inside Distance. The new
shapefile should now be attached. You can confirm this by looking
in the Data Layer Properties, or by creating a new Map.
• If your data are from a specific coordinate system, and you did not
copy across .prj (projection) file with the rest of the shapefile, then
you need to tell Distance about the coordinate system. You do this
in the Geographic tab of the Data Layer Properties.

Importing GIS Data by Linking and Existing Shapefile to

the Project Data Folder

This method is similar to the previous one, but does not require you to copy the
shapefile into the Data Folder, or to rename it. The advantage, therefore, is that
you only have one copy of your shapefile to manage. The disadvantages are:
• it is more complicated, requiring you to edit the Data File using a
database package
• because the shapefile remains outside the Data Folder, it is harder
to move the project onto other machines (for example, Export
Project only copies files in the Data Folder).
The following instructions assume that you have software to (1) create and edit
shapefiles (e.g., ESRI ArcView), (2) edit the Distance Data File DistData.mdb
(e.g., Microsoft Access 97). If you have Access 2000 or later, then you should
read Accessing DistData.mdb using newer versions of Access in the Appendix –
Inside Distance.
Prepare the Distance Data Layer
• Follow the instructions under this section in Method 1:Prepare the
Distance Data Layer.
Prepare the Shapefile
• Open the shapefile you wish to import in your GIS package, and
add a field to the table. The field should be able to contain long
integer numbers – in ArcView this means the field type should be
“Number”, the width 16 and decimal places 0. Name this field
“LinkID”.
• The LinkID field will be used to link records in the shapefile to
records in Distance’s internal data table for this layer. Each
LinkID value must correspond with a value in the ID field of
Distance’s internal table.
• If you haven’t created the Data Layer in Distance yet, then it
doesn’t matter what order you number the records in the

60 • Chapter 5 - Data in Distance User's Guide Distance 6.0 Beta 5

 LinkID field of the shapefile. Start with 1, and go up to the
number of records you have.
• If you have created the Data Layer already in Distance, and
already have information in the layer such as Labels, Areas,
etc., then you should take care to number the records in the
shapefile so that the LinkID field of a shape corresponds to the
ID value of that record inside Distance.
• Once you have added records to the LinkID field, you can close the
shapefile and exit your GIS.
Edit the Distance Data File
• In your database package, open the project’s Data File,
DistData.mdb. This is located in the project’s data folder.
• Open the DataTables table. Locate the record that corresponds to
the shapefile of the data layer in which you are interested. For
example, if you have created a data layer called “Antarctica” then
look for the record with a LayerName of “Antarctica” and
TableName that starts with “geo” (e.g., “geoAntarcti”). The
SourceDatabaseType should be “Geog”.
• For this record, change the SourceTableName to the name of
the shapefile you want to import. Don’t use any suffix. For
example, if the shapefile files are “Boundary.shp”,
“Boundary.shx” and “Boundary.dbf” then you enter
“Boundary”.
• Change the SourceDatabaseName to the path of the folder
containing your shapefile. You must include the full absolute
path – e.g., “D:\data\shapefiles”. If the shapefile is located in
the Data Folder for this project (for example if you copied it
there), then this field should be blank.
• Close the database DistData.mdb
Clean up
• You can now reopen the project inside Distance. The new
shapefile should now be attached. You can confirm this by looking
in the Data Layer Properties, or by creating a new Map.
• If your data are from a specific coordinate system, and you did not
copy across .prj (projection) file with the rest of the shapefile, then
you need to tell Distance about the coordinate system. You do this
in the Geographic tab of the Data Layer Properties.
• If the data imported correctly, you can delete the old shapefile from
the Data Folder.

Advanced Data Topics

Linking to Data From Other Databases

It is possible to directly link to data in external database tables, spreadsheets and

text files, instead of importing them into the distance database. This is an
advanced technique, and should be used only by those confident poking around
inside Microsoft Access databases. The technique involves editing the project
Data File (which is a Microsoft Access database) using Access, or some other

User's Guide Distance 6.0 Beta 5 Chapter 5 - Data in Distance • 61

 database tool (i.e., outside of the Distance interface). Note that the Distance
database engine is quite unforgiving - if you make a small mistake specifying the
location of the data, Distance will generate an error next time you try to open the
project within Distance.
For more information, see Linking to External Data from Distdata.mdb in the
Inside Distance Appendix. The text there refers to a sample project,
LinkingExample.dst, which is located in the sample projects folder (see Sample
Projects).

62 • Chapter 5 - Data in Distance User's Guide Distance 6.0 Beta 5

Chapter 6 - Survey Design in
Distance

Introduction to Survey Design in Distance

This chapter gives some background information about survey design, and an
overview of the interface for survey design in Distance. The methods
implemented here are based on work by Strindberg (2001), and are described in
Strindberg et al. (2004). This latter text, which is Chapter 7 of Buckland et al.
(2004) is recommended reading for anyone using the design engine in Distance.
In addition, Thomas et al. (2007) review these concepts, with particular
application to design of shipboard surveys in complex regions using Distance.
A number of frequently used survey designs are implemented within the
geographic survey design component of the Distance software (see Design
Classes Available in Distance, below). You can use the component either to
evaluate the properties of a design class or to generate an instance from that class
which can act as a survey plan. Simulation is used to calculate design class
properties, such as coverage probability, estimates of distance travelled while on-
and off- effort and between sampling locations. Design classes comprise a
sampler type, e.g. point or line, and a design type associated with the sampler.
The design type has an associated survey design algorithm, which has been
automated to generate the survey designs.
By creating a number of different designs in Distance, you can compare the
properties of the designs using simulation (see previous paragraph), and then
select a suitable design for your study. Most designs are stratified, in which case
the different designs may contain different number of strata or stratum
boundaries.
Within each stratum, sampler points, or lines, might be randomly located.
Alternatively, they may be placed on a regular grid that is randomly
superimposed on the stratum. Sometimes, more complex algorithms are
required. For example, shipboard surveys typically use continuous zigzag
samplers, so that costly ship time is not wasted in travelling from one line to the
next. A number of different zigzag designs are implemented in the software.
When these designs are realized within a convex survey region the sampler line
is continuous. Non-convex survey regions lead to some sampler discontinuity, as
the sampler is clipped against the survey region boundary.
For complex surveys, in which coverage probability is not uniform, the software
permits evaluation of coverage probability by location, using simulation as
mentioned above. Transect survey data are frequently analysed under the
assumption of an equal coverage probability design, as this avoids the necessity
of making assumptions about the distribution of the survey population. A design

User's Guide Distance 6.0 Beta 5 Chapter 6 - Survey Design in Distance • 63

 that leads to uneven coverage probability throughout the survey region can lead
to biased abundance estimates, if the analysis assumes coverage probability is
constant. Unbiased estimates can be calculated from the sample data if an
appropriate estimator, such as the Horvitz-Thompson-like estimator, is used and
the unequal coverage probabilities are taken into account. The properties that
such an algorithm should possess include the following: no part of the survey
region should have zero coverage probability, and coverage probability should
be as nearly constant as possible. The former property affects bias, whereas the
latter affects efficiency. We hope to implement an unequal coverage estimator
in Distance in the future.
One advantage of design automation by means of software is that it enables each
possible design to be compared for efficiency, accuracy, and bias of the
subsequent abundance estimates, using simulation over a population of interest.
Simulation capabilities of this nature are planned for a future version of
Distance.
An introduction to survey design issues for distance sampling is given in Chapter
7 of Buckland et al. (1993, 2001). These issues will be covered in detail in
Strindberg (in prep.) and Buckland et al. (in prep.). A good general text on
survey design is Thompson (1992).

Design Classes Available in Distance

The following design classes are available in Distance. We expect to add more
in future releases. Generally, all except the zigzag designs produce even
probability of coverage in the survey region (or stratum, for stratified surveys),
although care should be taken at the region boundaries (see Concept: Edge
Effects). For more on zig-zag designs, see Concept: Zigzag Sampling Designs.

Sampler Design Class Example Description

Point Simple Random Randomly distributes a fixed number
Sampling of points over the survey region.
Systematic Grid Randomly superimposes a systematic
Sampling point grid of fixed dimensions and
rotation onto the survey region.
Line Parallel Random Randomly distributes a number of
Sampling parallel lines over the survey region.
Systematic Randomly superimposes a systematic
Random set of parallel lines onto the survey
Sampling region.
Systematic Randomly superimposes a systematic
Segmented set of segmented parallel lines onto the
Trackline survey region. A set of parallel
Sampling tracklines is used for this purpose.
Systematic Randomly superimposes a systematic
Segmented Grid set of segmented parallel lines onto the
Sampling survey region. A set of grid points is
used for this purpose.
Equal Angle Superimposes a continuous zigzag
Zigzag sampler of fixed angle on the survey
region.
Equal Spaced Superimposes a continuous zigzag
Zigzag sampler that passes through equally
spaced points on opposite sides of the
survey region boundary.
Adjusted Angle Superimposes a continuous zigzag
Zigzag sampler whose angle is continuously

64 • Chapter 6 - Survey Design in Distance User's Guide Distance 6.0 Beta 5

 adjusted by survey region height.

Survey Design Concepts

Concept: Coverage Probability

The coverage (or inclusion) probability at an arbitrary location within the survey
region is the probability of it falling within the sampled portion of the survey
region. Transect survey data are frequently analysed under the assumption of an
equal coverage probability design. A design with uneven coverage probability
leads to biased abundance estimates if even coverage probability is assumed.
Thus, in some respects the ideal is to attain equal probability of coverage
throughout the survey region, as this simplifies the statistical analysis. However,
if equal coverage probability is not feasible then it is possible to use a sampling
design that gives different, but known coverage probabilities throughout the
survey region; unbiased estimates can be calculated from the sample data if an
appropriate estimator, such as a Horvitz-Thompson estimator, is used and the
unequal coverage probabilities are taken into account. (A generalized Horvitz-
Thompson-like estimator is planned for a future version of Distance.) Even if an
estimator that takes unequal coverage probabilities into account is used, designs
providing nearly even coverage probabilities are preferable. Animals detected in
a region of relatively low coverage probability can contribute substantially to the
abundance estimate, and estimation may be very imprecise.
It is also more difficult to get precise estimates of the coverage probabilities if
these probabilities are small. A high coverage probability leads to more robust
estimation, and improved precision for the coverage probability estimates, but
this requires an increase in effort, which may not be affordable.
Simulation can be used to estimate the coverage probability at locations
throughout the survey region for those designs that give uneven coverage
probability. Even for the more straightforward designs that are frequently
assumed to give even coverage probability throughout the survey region,
simulations can be used to examine the potential problems caused by edge
effects at the survey region boundaries.
For more details about the various options associated with estimating coverage
probability, see the Program Reference section on Coverage Probability
Information on the Design Results Tab.

Concept: Edge Effects

Point or line samplers have an associated radius or width, so parts of the areas
sampled by each point or line sampler may fall outside the survey region. If the
relative area of the sampled region is small relative to that of the survey region
then discarding sampling units that intersect the boundary of the survey region
causes negligible bias. However, if the relative area is large then discarding
sampling units at the edge of the survey region may cause considerable bias in
the estimates.
Minus Sampling
If points or lines are generated exclusively within the survey region we call this
minus sampling. This leads to some under-sampling at the edge and an uneven
coverage probability. Depending on the relative area of the sampled region and
the relative density of the population along the edge of the survey region, this
may or may not lead to significant bias in the estimates. This under-sampling is
due to potential line or point samplers that lie just outside the boundary, whose
sampled area intersects the survey region.

User's Guide Distance 6.0 Beta 5 Chapter 6 - Survey Design in Distance • 65

 Plus Sampling
If a buffer zone is created around the survey region, then samplers can be
generated in the survey region plus the buffer zone, which we call plus sampling.
Plus sampling leads to an even coverage probability, but to some loss in
efficiency (as part of the survey effort falls outside the region of interest).
With either plus or minus sampling, portions of the sampled area associated with
samplers along the edge of the survey region will fall outside the region itself. If
the terrain immediately outside the survey region differs substantially from that
within the region and an abundance estimate for a particular habitat is required,
then you should not include sample data from those areas outside the survey
region. On the other hand, if the buffer zone surrounding the survey region is
such that no population members are found in it (e.g. the survey region is an
island the buffer the surrounding water), then the abundance estimate can be
obtained for the survey region together with the buffer region and the same
abundance estimate will apply to the survey region.

Concept: Zigzag Sampling Designs

Zigzag line transect designs can be more efficient than conventional parallel line
designs, because no time is spent moving from one line to the next (“off effort”).
This type of design is often used in shipboard surveys, where ship time is
extremely costly and survey areas are large, so moving from one line to the next
in a conventional design would be very expensive.
The downside is that zigzag designs zigzag surveys are difficult to generate in
complex survey regions (see Zigzag Sampling in Non-convex Regions).
Further, some zigzag designs do not produce even probability of coverage for
anything but a rectangular survey design.
Here, we give a brief overview of the zigzag designs available in Distance – for
more information see Strindberg (2001), Stringberg et al. (2004) and Strindberg
and Buckland (2004).
The lines in a zigzag design are generated with respect to a design axis, which
is a user-defined line overlaid on the survey region.
Distance offers 3 different zigzag design classes:

• Equal Angle Zigzag. In this design class, the angle of the

lines is fixed with respect to the design axis. This design class
produces even coverage only if the survey region is rectangular,
and then only if the design axis runs parallel to one side of the
rectangle.

• Equal Spaced Zigzag. Here, the lines run through equally

spaced points on opposite sides of the survey region boundary.
This design class also produces uneven coverage probability for all
but rectangular survey designs. However, it gives more even
coverage probabilities than the equal angle design. The coverage
probabilities become more even as you increase the sampling
intensity (i.e. line length), so you have to make a trade-off between
the length of line you can afford to survey and the evenness of the
coverage probability. There is also an issue with how to place the
first and last lines – see First and last line placement in equal
spacing zigzag designs.

• Adjusted Angle Zigzag. In this design class, the angle of the

lines is continuously adjusted, depending on the height of the
survey region. If the survey region is convex, this produces a
design with even probability of coverage in the direction of the
design axis. The downside is that it is difficult to implement on the

66 • Chapter 6 - Survey Design in Distance User's Guide Distance 6.0 Beta 5

 ground – in practice you would approximate the design using small
straight segments.

Zigzag Sampling in Non-convex Regions

Zigzag sampling designs can only be generated in a convex survey region. If any
of the survey strata in the survey layer are non-convex, then you can choose to
generate the design for each stratum in either a convex hull or bounding
rectangle. For non-convex strata, the zigzag line will no longer be continuous.
The amount of discontinuity is generally less using the convex hull, but this may
lead to uneven coverage probabilities. If simulation shows such an effect to be
extreme, then it’s better to use the bounding rectangle. Details of how to set
these options are given in the Program Reference topic, Zigzag Sampling Non-
convex Survey Region Options.
Another way of dealing with non-convex survey regions is by using
stratification. The figure below shows how you can make a non-convex region
convex by defining 3 strata. This is only an option for certain survey regions
whose size and shape permit such stratification.

Stratifying the survey region into 3 strata can eliminate non-convexity or at least reduce
the discontinuity in the sampler

Even if stratification doesn’t let you take care of non-convexity entirely, it may
at least reduce the discontinuity in your sampler. An example of this, for a real
survey in British Colombia, is shown below.

Single realizations of three equal-spacing zigzag designs applied to inshore waters of

southern British Colombia (taken from Fig 3. of Thomas et al. 2007; see that paper for
more details). In the first, the study region is treated as a single stratum. In the second it
is divided into two strata. In the third it is divided into four strata. The designs are laid
out in covex hulls fit to the strata; note that as the number of strata goes up, the strata
become more convex and so the amount of discontinuity (i.e., off-effort time) between
lines decreases.

First and last line placement in equal spacing zigzag

designs
The material in this section is a brief abstract of Strindberg (2001), section 4.6,
which should be consulted for full details on the algorithm used.
The equal spacing zigzag design works by placing a series of equally spaced
lines across the study area, perpendicular to the design axis, and then joining up
alternate waypoints formed by the intersection of these lines with the survey area
boundary. However, for the first and last transect lines, the equally spaced
perpendicular lines do not intersect the survey area boundary. Instead, a line is
drawn from the survey area boundary until it intersects the perpendicular line at
ninety degrees (the dashed line in the figure below), and this is used as the
waypoint.

User's Guide Distance 6.0 Beta 5 Chapter 6 - Survey Design in Distance • 67

 Method of constructing the first and last lines in an equal spacing zigzag design.
Reproduced from Figure 4.6 of Strindberg 2001.

Although this method does not produce exactly equal coverage in the area of the
first and last transects, it usually comes close (see, e.g., Thomas et al. 2007).
Alternative algorithms may be implemented in future, such as the use of an
adjusted angle zigzag sampler for the first and last segments.

Setting Up a New Project for Survey Design

This section outlines the options for creating and setting up a new project ready
for use with the survey design component of Distance. You should read Chapter
4 - Distance Projects and Chapter 5 - Data in Distance before going any further!
Creating the new project
The first step is to create a new Distance project, using File | New Project….
In the first page of the New Project Setup Wizard, choose the option Design a
new survey. Click Next and on the next page click Finish.
Importing or entering geographic data
If you’ve followed the instructions above, Distance has created a new project
which contains a geographic global data layer with one record. In many cases,
your survey designs will include strata. If this is true for you, then create a new
layer below the global layer, and make it of type Stratum (see Chapter 5
Changing the Data Structure). Then add one record to the new layer for each
stratum.
You’re now ready to import or enter your data – see Chapter 5 Getting Data into
Distance, for instructions how to do this. The most common route is to import
the data by copying the appropriate shapefiles into the project folder. If you take
this route, note that you need shapefiles both for your stratum layer(s) and the
global layer. It is easy to form the global polygon in your GIS by joining the
polygons from the stratum shapefile. You need to have a valid shapefile for the
global layer before you can create a coverage probability grid (see below), even
if you are doing the design at the stratum layer.
Creating a coverage layer
The next step is to create a grid of points at which coverage probability will be
assessed for your designs – this is called a coverage layer.
In the Data Explorer, click on the icon for the global layer in the Data Layers
tree ( ), and then click the Create New Data Layer… button. Under “Layer
type”, choose Coverage, and then click Properties… and fill in the required
grid spacing and other properties. Once you have the options set the way you
want, click OK in the Grid Properties dialog, and then OK again in the Create
New Layer dialog. Distance will then create the coverage layer and the points
that go in it.
You can check the number of points from the Data Explorer by clicking on the
icon for the coverage layer and then clicking on the Data Layer Properties
button. You could also create a Map (in the Map tab of the Project Browser)
and add the coverage layer to a map to see how it looks.

68 • Chapter 6 - Survey Design in Distance User's Guide Distance 6.0 Beta 5

 Creating and running designs
You’re now ready to create some designs and run them to calculate coverage
probability or to generate single realizations of the design (“surveys”). For
examples of this process, see Chapter 3 - Getting Started.

User's Guide Distance 6.0 Beta 5 Chapter 6 - Survey Design in Distance • 69

Chapter 7 - Analysis in Distance

Introduction to Analysis in Distance

Distance is designed to promote interactive modeling of distance sampling data.
It is easy to set up and run many different models, possibly using different
subsets of the data, and to compare and catalogue the results.
In this chapter, you will find general information about the way that analyses are
set up and stored in Distance, and the kind of analyses that you can perform.
This chapter is essential reading if you want to get the most out of the program.
More detailed information about each of the analysis windows is in the
Appendix - Program Reference.

Introduction to the Analysis Browser

In Distance, your data modeling is divided into individual analyses. Each
analysis has a name, and a set of inputs associated with it. If you have run the
analysis, it will also have some results.
The main interface for creating, managing and comparing analyses is the
Analysis Browser, which you can access by clicking on the Analyses tab in
the Project Browser. This displays a summary of each analysis, in table form:

Example of the Analysis Browser, from the Ducknest sample project

In this example, there are four analyses. The first analysis, which is highlighted
in blue, has not been run – its status light (left hand column) is grey, and the
results columns (right columns) are blank. The next two analyses have been run,
but the run generated some warnings – the status light is amber. If an analysis
encounters an error during a run, the status light will be red. The last analysis
ran with no errors or warnings – its status light is green.
You may also notice in the example that the toolbar along the top of the
Analysis Browser has a box labelled “Set: All data”. In Distance you can
group your analyses into different Sets. If you were to click on the down arrow
beside “All data”, you would see that there is another set in this project, called
“Truncation at 6 feet”. If you chose that set then another table of analyses would

User's Guide Distance 6.0 Beta 5 Chapter 7 - Analysis in Distance • 71

 replace those currently displayed. Sets provide a convenient way of grouping
related analyses.
The other buttons in the toolbar allow you to create, delete and manage sets, to
create and delete analyses, run analyses, etc.

Introduction to Analysis Details Windows

You can find out more about individual analyses by opening an Analysis
Details window. You do this either by double-clicking on the status light of an
analysis, or by highlighting the analysis you are interested in and clicking the
Show Details button on the Analysis Browser’s toolbar.
Each Analysis Details window contains three tabs (along the right hand side)
• Inputs Tab, where you specify how the analysis is to be done.
• Log Tab, where you view a log of the analysis once it has been
run (useful for pinpointing problems).
• Results Tab, where you can read many detailed pages of results
from the analysis.

The Analysis Details windows for two analyses, open on different Results pages.

In the above example, the results tab of the top analysis (“Analysis 2”) is green
because it ran without generating any errors or warnings. For an analysis that
has not been run, all three tabs are grey, while if an analysis encounters problems
during the run, the Log tab is colored amber (warnings) or red (errors).

Analysis Components

Data Filters, Model Definitions and Survey

Objects
In Distance, each analysis is made up of three components: a Survey, a Data
Filter, and a Model Definition.

• Survey objects tell Distance what kind of survey you

performed (e.g., point transect or line transect), and where the data
are stored in the project (which data layers and fields).
• Data Filters manipulate the survey data before passing it to the
analysis. For example, you could define a Data Filter that discards

72 • Chapter 7 - Analysis in Distance User's Guide Distance 6.0 Beta 5

 all observations with perpendicular distance greater than a given
distance (i.e., performs right truncation), or one that selects only
data from one survey region.
• Model Definitions tell Distance how to analyze the data that
passes through the Data Filter. Model Definition options include
the analysis engine to use, the type of detection function model
(e.g., half-normal with cosine adjustments) and the method of
estimating variance (analytic vs. bootstrap), as well as many other
options.
Each Survey, Data Filter and Model Definition can be attached to one or more
analysis (in fact, they can also be attached to no analyses!). Changing the
properties of a Survey, Data Filter or Model Definition affects all of the analyses
that it is attached to.
Example
For example, in the Ducknest project (Ducknest.dst, in the Sample Projects
folder), one Survey, two Data Filters and four Model Definitions have been
defined.
You can get a list of the Surveys by clicking on the Survey tab of the Project
Browser:

Survey Browser in the Ducknest project.

You can get a list of Data Filters and Model Definitions by clicking on the View
Analysis Components button on the main toolbar. This opens the
Analysis Components Window:

Analysis Components window, showing a list of the two Data Filters (left) and four Model
Definitions (right), in the Ducknest project.

Using the Analysis Browser, we can see that the four analyses in the Analysis
Set “All data”, all use Survey number 1 (called “New Survey”), and Data Filter
number 1 (called “Default data filter”). However, each one uses a different
Model Definition.

Part of the Analysis Browser from the Ducknest project

User's Guide Distance 6.0 Beta 5 Chapter 7 - Analysis in Distance • 73

 If you were to change the Data filter “Default data filter” in some way, then all
four of these analyses would be affected (and any other analyses in other
Analysis Sets that use this Data Filter). For one thing, the three analyses that
have already been run (those with green or orange status lights) would have to be
reset and their results deleted, because these results would be out of date. In
Distance, if you change a Survey, Data Filter or Model Definition that is being
used by analyses that have already been run, Distance automatically issues a
warning message and asks you if you want the analyses to be reset.

Working with Data Filters and Model Definitions

Selecting a Data Filter or Model Definition for your
Analysis
In Distance, you select the Data Filter and Model Definition for an analysis in
the Inputs tab of its Analysis Details window. For example, the following
analysis has the Data Filter called "Default data filter" and the Model Definition
called "Half normal / hermite" selected.

Example of the Analysis Details Inputs tab for an analysis, from the Ducknest sample
project

The analysis is also called “Half normal / hermite” (top of the picture, in the title
bar and beside Name:). This is the name that appears in the Analysis
Browser, so it is always a good idea to give the analysis a name that lets you
distinguish it from other analyses in the Analysis Set.
If you wanted to change, say, the Data Filter for this analysis to “Truncation at 6
feet”, you would click on that Data Filter in the list. In the case of the analysis
shown above, it would not be a good idea to change the Data Filter as the
analysis has already been run and has results associated with it (you can tell this
because the Results tab is green). If you change the selected Data Filter or
Model Definition in an analysis that has already been run, then Distance will
warn you that the results have become out of date and ask whether you want
them deleted.
The best way to do an analysis with a new combination of Data Filter or Model
Definition is to create a new analysis in Distance for this combination. You do
this in the Analysis Browser, by clicking on the New Analysis button .
This automatically creates a new analysis, based on the one that you currently
have selected in the Analysis Browser. You can then open up the Analysis
Details for the new analysis. Because the new analysis has not yet been run,
you are free to choose the combination of Data Filter and Model Definition you
want. If this seems a little confusing, take a few moments to try creating a new
analysis in the Analysis Browser for the Ducknest project.

74 • Chapter 7 - Analysis in Distance User's Guide Distance 6.0 Beta 5

 Creating New Data Filters and Model Definitions
Imagine that in the Ducknest example, above, you have created a new analysis
based on the analysis called “Half-normal / hermite”, above. Instead of selecting
one of the existing Model Definitions for this analysis, you wish to create a new
one. Perhaps this new one will use the same detection function model, but will
use bootstrapping to estimate the variance of the density estimate.
There are two ways to create a new Model Definition (or Data Filter):
• by clicking the New... button on the Inputs tab of the Analysis
Details window of the new analysis. This is described in more
detail below.
• by using the Analysis Components window. This is described
in the section Using the Analysis Components Window, below.
Creating a New Model Definition from the Inputs tab of the Analysis
Details window
To create a new Model Definition, you click on the New... button under
Model definition on the Inputs tab of the Analysis Details window:

Model Definition section of the Inputs tab on the Analysis Details window

Distance creates a new Model Definition, based on the one you currently have
selected, and opens the Model Definition Properties dialog for this Model
Definition:

Example Model Definition Properties dialog

You can then edit the properties to reflect the changes you want. As in our
example, to change the variance option to use boostrapping, you click on the
Variance tab and tick the Select non-parametric bootstrap checkbox.
(More details about the options in this dialog are given in the Model Definition
Properties Dialog section of the Program Reference.) You may also want to
change the name to reflect the change in properties, for example calling the
anlysis “Half-normal / hermite – bootstrap”. You do this by editing the Name:
text box.

User's Guide Distance 6.0 Beta 5 Chapter 7 - Analysis in Distance • 75

 Example Model Definition Properties dialog, with bootstrapping option selected

You can now press the OK button to save the new options and close the Model
Definition Properties dialog. The new Model Definition is automatically
selected in the Analysis Details:

Example of the Analysis Details Inputs tab, with a new Model Definition selected

You can then run the analysis by pressing the Run button, or you can close the
Analysis Details window and run the analysis from the Analysis Browser by
pressing the button on the Analysis Browser's toolbar.
Creating new Data Filters is exactly analogous to the process just described for
Model Definitions. In this case, when you press the Data filter New... button, a
new Data Filter is created based on the one currently selected, and the Data Filter
Properties dialog opens. To find out more about the Data Filter options, see the
Data Filter Properties Dialog section of the Program Reference.

Viewing and Editing Existing Data Filters and Model

Definitions
In some cases, you may want to view or edit the properties of an existing Data
Filter or Model Definition. For example, you may wish to check the options in a
Model Definition before selecting it for an analysis. There are two ways to do
this:
• by clicking the Properties... button on the Inputs tab of the
Analysis Details window. This will open the properties dialog

76 • Chapter 7 - Analysis in Distance User's Guide Distance 6.0 Beta 5

 of the Data Filter or Model Definition that is currently selected for
that analysis.
• by using the Analysis Components window. This is described
in the section Using the Analysis Components Window, below.

If you change the options in a Data Filter or Model Definition that is

being used by any analyses that have been run, Distance will warn you when you
press the OK button that the analyses will be reset.

To find out more about the options available in the Data Filter Properties and
Model Definition Properties dialogs, see the Program Reference sections
Data Filter Properties Dialog and Model Definition Properties Dialog.

Using the Analysis Components Window

The Analysis Components window is designed to be a convenient way of

manipulating Data Filters and Model Definitions. The window shows a list of
all Data Filters or Model Definitions in your project. Using buttons on the
toolbar you can create, delete, view and arrange the listed components.

Analysis Components window, showing a list of the Model Definitions in the Ducknest
sample project

In the Analysis Components toolbar, you click on the first button to get a list
of all the Data Filters, and the second button to get a list of all the Model
Definitions in your projects. The other buttons allow you to copy (i.e. create),
delete, view and arrange (move up and down the list) the component that you
have selected.
You can also work with Data Filters and Model Definitions in the Analysis
Details window. Using the Analysis Components window is most useful
when you have a large number of components in your project, as you can
arrange them into a logical order, delete the ones you are not using, and easily
rename them.

The last column in the table of analysis contents tells you whether that
component is currently being used in any analyses: “Y” means it is being used
and “N” means that it is not. This is useful because when there are many
components (e.g., many Model Definitions if you have been doing a lot of
analyses), it is easy to loose track of which are being used and which are no
longer required. Also, if you double-click on a “Y”, you get a list of the analyses
that use that component.

Example
In the section Creating New data Filtes and Model Definitions we showed how
to create a new Model Definition and associate it with a new Analysis using the

User's Guide Distance 6.0 Beta 5 Chapter 7 - Analysis in Distance • 77

 Inputs tab of the Analysis Details window. Here, we show how to do the
same thing using the Analysis Components window.
In the Ducknest sample project, there are four analyses in the set “All data”.
These analyses use four different Model Definitions, each with a different
detection function model. Imagine you want to add a new analysis, with a new
Model Definition. Your new analysis will be based on the “Half-normal /
hermite” analysis, but will use bootstrapping to estimate the variance of the
density estimate.
You begin by creating a new Model Definition. In the Analysis Components
window, you highlight the Model Definition “Half-normal / Hermite” and click
on the New Item button . A new Model Definition is created, based on the
one you had highlighted:

Analysis Components window, showing the new Model Definition

The new Model Definition has been given the default name “Half-normal /
hermite 1”. You may want to change the name to reflect the options you’re about
to set – double click on the name and type “Half-normal / hermite – bootstrap”.
You can now edit the new Model Definition properties, by double-clicking on
the ID of the new Model Definition, or by clicking the View Item Properties
button . The Model Definition Properties dialog opens. To change the
variance option to use boostrapping, you click on the Variance tab and select
the “non-parametric bootstrap” option. (More details about the options in this
dialog are given in the Model Definition Properties Dialog section of the
Program Reference.) You can now press the OK button to save the new options
and close the dialog.
You have now set up the new Model Definition ready for use. If you want to
perform other types of analysis, you could set up more Model Definitions at this
point.
You now need to create a new Analysis, and attach the new Model Definition to
this new Analysis. In the Analysis Browser (i.e., the Analysis tab of the
Project Browser), click on the New Analysis button . Double-click on the
status button of the new analysis – this opens the Analysis Details window. In
the Model definition section, select your new model definition, and in the
Name: section, type in a suitable name for your new analysis, e.g., “Half-
normal / hermite – bootstrap”. You can now run the analysis.
This approach to setting up new Model Definitions (or Data Filters) is most
useful when you have several to set up at once. You can use the Analysis
Components window to set up your new components, then use the Analysis
Browser to create new Analyses, and associate the new analyses with the new
components. Then, in the Analysis Browser, you can highlight all the new
Analyses, click the run button , and go and have a cup of tea while they all
run! (For more about running analyses, see Running Analyses on page 12).

Working with Surveys during Analysis

Surveys tell Distance what kind of survey you performed (e.g., point transect or
line transect), and where the data are stored in the project (which data layers and
fields). Each analysis is associated with a Survey. Surveys are also used when
designing new field surveys – for more on this see Chapter 6 - Survey Design in
Distance.

78 • Chapter 7 - Analysis in Distance User's Guide Distance 6.0 Beta 5

 In routine analysis of Distance sampling data, you don’t have to worry too much
about Surveys. If, in the Setup Project Wizard, you tell Distance that you want
to analyze a survey that has been completed, Distance will automatically create a
Survey for you, based on what you tell it about your survey in the wizard. This
Survey will be used by default for all the Analyses you create.
You only need to work with Survey during analysis if:
• you want to set up the project manually. In this case, you will need
to create the Survey and set its properties yourself.
• you wish to do analyses involving more than one Survey.
Situations where this may be useful are covered in the next section
(Analysis with Multiple Surveys).
You can get more information about the Survey associated with your analysis by
clicking the Details... button on the Inputs tab of the Analysis Browser.
This opens the Survey Details window. Click on the Properties... button of
the Survey Details window.

Analysis with Multiple Surveys

Normally, for analysis in Distance, you only need one Survey. This Survey tells
Distance what type of survey you performed and where the data from the survey
are stored. However, there are some situations when it is useful to have more
than one Survey in a project. Examples include:
• you have a complicated data structure, for example with two or
more layers of type Sample (e.g., for two or more survey regions or
years). In this case, you will set up one Survey to point to each
sample layer. You could then create one Analysis set for analyses
that use the first layer, and another for analyses that point to the
second.

Having multi-site or multi-year data in separate data layers means

you will not be able to do a single combined analysis of all data. For this reason,
it is often better to put this type of data into a single data layer, with an extra
field indexing the year or site number. You can then use the data selection page
of the Data Filter to select out individual years/sites or combinations of
years/sites for analysis as required.

• you have multiple measurements of the same objects, for example

from multiple observers. One case where this can occur is in field
trials on artificial objects, where multiple observers traverse the
same lines or points. Your data will then have multiple “distance”
fields. You can set up a Survey for each distance field (in the Data
Fields tab of the Survey), and then analyze each observer
separately.
• you have clustered data, but you want to ignore the clustering for
some analyses, and just calculate density of clusters. In this case,
you would create a new survey with the Observations option set to
Single objects.
• you have multi-species data where some species occur in mixed
species groups (see Zero Cluster Sizes in CDS Analysis for more
on this).
If you do want to set up multiple Surveys, you do this from the Survey
Browser, which you access by clicking on the Survey tab of the Project
Browser.

User's Guide Distance 6.0 Beta 5 Chapter 7 - Analysis in Distance • 79

 • To create a new survey, click on the New Survey button.
• To edit the properties of a survey, click on the Show Details
button in the Survey Browser. A Survey Details window
opens. On the Input tab, click on the Properties... button.
Once you have created and set up the Surveys you require, you associate them
with Analyses by selecting the appropriate Survey from the list on the Survey
section of the Analysis Details window.

Example of Analysis Details Inputs tab, showing selection of Survey object

Analysis Engines
In Distance, you have a choice of analysis engines to perform an analysis.
Each analysis engine has different capabilities, and has different inputs and
outputs. Distance has three analysis engines built in: a conventional distance
sampling (CDS) engine, a multiple covariate distance sampling (MCDS) engine,
and a mark recapture distance sampling (MRDS) engine. More engines are
planned for future versions of Distance.
You choose the analysis engine when you are setting up a model definition –
select the appropriate engine from the drop down list at the top of the Model
Definition Properties dialog:

Choosing an analysis engine

A description of the options available for each engine is given in the Model
Definition Properties Dialog section of the Program Reference.

Conventional Distance Sampling (CDS) Engine

This engine provides a design-based analysis of line or point transect data, using
the approach described by Buckland et al. (1993, 2001). Probability of detection
is modeled as a function of observed distances from the line or point, using
robust, semi-parametric methods. One level of stratification is allowed, and
there are various methods for dealing with cluster size bias. Variance can be
estimated empirically, or via a non-parametric bootstrap.
For more information about this engine, see Chapter 8 - Conventional Distance
Sampling Analysis.

80 • Chapter 7 - Analysis in Distance User's Guide Distance 6.0 Beta 5

 Multiple Covariate Distance Sampling (MCDS)
Engine

This engine contains (almost) all the features of the CDS engine, but also allows
additional covariates to be included in the detection function model, in addition
to observed distance. These covariates enter through the scale parameter of the
key function (via a log link function). This means that the covariates are
assumed to influence the scale of the detection function, but not its shape (see
picture, below).

Example estimated detection functions, where cluster size (Sbar) is the covariate. The
basic shape of the function is the same (half-normal), but the effective strip width is wider
at cluster size 750.

For more information about this engine, and when it can be useful, see Chapter 9
- Multiple Covariates Distance Sampling Analysis. This engine was first
introduced in Distance 4.0.

Mark Recapture Distance Sampling (MRDS)

Engine

This engine permits analysis of data collected from two survey platforms, where
the assumption of certain detection detection of objects on the trackline can be
relaxed. CDS and MCDS analyses are also possible, although adjustment terms
are not currently available to modify the shape of the detection functions in this
engine.
For more information about this engine, see Chapter 10 - Mark Recapture
Distance Sampling. This engine was first introduced in Distance 5.0.

Density Surface Modelling (DSM) Engine

This engine allows users to model variation in density in relation to covariates,

using distance sampling (and plot sampling) data.
For more information about this engine, see Chapter 11 - Density Surface
Modelling. This engine was first introduced in Distance 6.0.

Running Analyses
Once you have created an analysis, and set it up by choosing an appropriate
Survey, Data Filter and Model Definition, you are then ready to run it.
There are two ways to run an analysis in Distance:

User's Guide Distance 6.0 Beta 5 Chapter 7 - Analysis in Distance • 81

 • in the Analysis Browser, select the analysis and press the Run
Analysis button (or choose Analyses | Run Analysis)
• in the Inputs tab of the Analysis Details window of that
analysis, press the Run button Analysis - Inputs | Run
Analysis.

You can run more than one analysis at once. Simply highlight all of
the analyses you want to run in the Analysis Browser and then press the Run
Analysis button. This is useful if you are planning on doing a number of long
analyses - simply set them all up, select and run them all at once, and go and
have a cup of tea!

In Distance, when you select more than one analysis to run

at once, one is run straight away, and the others are queued up to run in turn.
You will see the status light of the queued analyses turn to a “q”. This change
was made for two reasons. Firstly, it makes more efficient use of a single-
processor computer to have only one computer-intensive process running at a
time. Secondly, Distance is designed to have numerical engines running that
require exclusive access to the distance database. One example of this is the
survey design engine. Because of this, only one item (design, survey, analysis)
can run at a time.

In Distance, Analyses run in a background process. This means that

you can carry on working with the interface while the analysis is running. The
only time you should go easy on the interface is when an analysis is initializing
or finishing (you can tell when this is happening because the pointer is turned
into an hourglass and a message is printed on the status bar at the top of the
screen). This feature is particularly useful if you are doing analyses such as
bootstraps that take a long time to run.

Windows NT, 2000 and XP are generally better built than Windows
95, 98 and ME. However, one unfortunate consequence of this for us is that
background processes tend to get more CPU time. This means that the Distance
interface may slow down quite noticeably while an analysis is running under
Windows NT/2000/XP, even on well-configured machines. You can give the
interface a boost by changing the Foreground performance setting in the
Performance tab of the Windows System Properties dialog (System icon in
Control Panel) to Maximum.

You can stop an analysis that is running by pressing the Stop button
in the Analysis Details window (this button replaces the Run button while the
analysis is running), or by highlighting the analysis in the Analysis Browser
and pressing the Reset Analysis button. However, on some systems, the
analysis will appear to stop but will carry on running in the background, using up
system resources. For more about this, see Stopping an Analysis, in Chapter 10.

Locking the Data Sheet

A complete analysis of distance sampling data in Distance usually proceeds in
three stages:
• creating and setting up the distance project file

82 • Chapter 7 - Analysis in Distance User's Guide Distance 6.0 Beta 5

 • getting the data into Distance
• analyzing the data
Clearly, it is generally not a good idea to change the data after it has been
analyzed. Because of this, Distance has built-in safeguards to prevent you from
accidentally editing the data once the analysis phase has begun. By default,
whenever you perform an analysis, Distance locks the Data Sheet. This means
that the data cannot be edited or deleted. You can tell that the data have been
locked, as the Lock Data Sheet button in the Data Explorer toolbar becomes
depressed.
In some circumstances you may want to change the data after the analysis phase
has begun. For example, you may discover an error in the data entry as a result
of some exploratory analysis. You may also want to add another year's data to
your project. You may want to add additional columns, in order to perform post-
stratification. If you wish to change the data, it is simple to unlock the data
sheet: click on the depressed button to pop it back out.
If you regularly update the data during the analysis stage then you can disable
the automatic locking feature of Distance, by unchecking the option
Automatically lock data sheet whenever an analysis is run in the
Analysis tab of the Preferences dialog (choose Tools | Preferences from the
main menu bar).

Cleaning the Temp folder

When analyses are run, temporary files are written to the Windows temp folder
(the location of this folder is system-dependent). As analyses finish, these
temporary files are deleteted. If an analysis crashes, then the temporary files
may not be deleted correctly, so over time they may build up in the temporary
folder. To clear all temporary Distance files from this folder, choose Tools |
Clean temp folder ….

R Statistical Software

Distance has a link to the free statistical software R. The Mark Recapture
Distance Sampling (MRDS) analysis engine is implemented as an R library, and
so you must have a working copy of R installed on your computer before you
can use that engine.
You can download and install R after you have installed Distance, and you can
run Distance without having R installed - but you will get an error message if
you try to run the MRDS engine without R.
R is under very active development, and new versions are released quite
frequently. Unfortunately, new versions are sometimes not compatible with
libraries compiled in old versions. We will endeavour to test our libraries with
each new version as it appears, and update it as required. For more information
about the version we are currently supporting, please browse to the Program
Distance Web Site, Support, Updates and Extras page.

To use the MRDS engine, you don’t have to know anything about R
beyond how to install it. However, R is a fully featured, widely-used statistics
package which you may consider using for your other analyses. You can find
out more about R from the R project home page, http://www.r-project.org/.

User's Guide Distance 6.0 Beta 5 Chapter 7 - Analysis in Distance • 83

 Installing and Configuring R
Instructions for installing R are given in the file readme.rtf in the Distance
program folder. You can also access this file from within Distance by choosing
Help | Release notes.
Before installing, you should check which versions of R are currently supported
by Distance. Latest information is on the Support, Updates and Extras page of
the Program Distance Web Site.
Once you have R installed, you should be able to run MRDS analyses straight
away from within Distance. Distance automatically recognizes the latest version
of R you have on your system the first time you run an MRDS analysis or open
the Preferences dialog, and automatically installs the mrds library. Once it has
registered the presence of R, it keeps using the same version until you manually
tell it to update (see Updating the Version of R That Distance Uses). You can
check which version it is using by choosing Tools | Preferences and looking
under the Analysis tab.

If you cannot see the plots that R produces, see Images Produced by R

Updating the Version of R That Distance Uses

If you install a new version of R after running some analyses using an older
version, Distance does not automatically switch to the new version. This is
because Distance might not be compatible with the newer version of R. Before
telling Distance about the new version of R, you should check it is compatible,
by looking on the Support, Updates and Extras page of the Program Distance
Web Site. Having checked, you can make Distance use the new version by
choosing Tools | Preferences, choosing the Analysis tab and updating the
Folder containing R to the folder that contains the new version. Distance will
automatically install any required libraries the first time R is run from within
Distance. Once you have checked the new version of R works, you can delete
the folder containing the old version.

Contents of the R folder

Distance projects that contain analyses run with R (e.g., MRDS analyses) have a
folder within the project data folder called “R”. This folder contains:
• the R object file, .RData. This file holds all the objects created by
R for that project. By default, new objects that are created for an
analysis are deleted at the end of that analysis, so the .RData file is
virtually empty. However, this default behaviour can be changed -
see Analysis Preferences Tab.
• image files generated by R. These files are loaded into the Results
tab when the Analysis Details for an analysis is opened. For more
about these files, see Images Produced by R.

Images produced by R
The images produced by R are stored as files in the R folder (see Contents of the
R folder). They are of the general form
[prefix].[analysis ID].[plot number].[suffix]

for example qq plot 1 for analysis 8 in windows metafile (.wmf) format would be
qq.8.01.wmf.

84 • Chapter 7 - Analysis in Distance User's Guide Distance 6.0 Beta 5

 These files are loaded into the Results tab when the Analysis Details for an
analysis is opened inside Distance.

The image files can be used in producing manuscripts and other

reports of analyses done. Many aspects of the images can be changed - see
below.

Changing the image properties

By default, R produces images in Windows Metafile format (.wmf). This is a
vector format that can be viewed at a variety of sizes without loss of quality.
However, .wmf files do not display properly on older Windows operating
systems, so you may wish to switch to another format. You may also wish to
switch to a more compact format to save disk space.
In addition, many other aspects of the images can be configured, such as the
image size (this only affects the size in the image file; images are automatically
scaled to fill the Results tab of the Analysis Details window), line width, font
size, etc.
To change image properties, choose Tools | Preferences, Analysis tab, and
under R Software click on Image Properties…. For more about the options,
see R Image Properties Dialog in the Program Reference.

User's Guide Distance 6.0 Beta 5 Chapter 7 - Analysis in Distance • 85

Chapter 8 - Conventional
Distance Sampling Analysis

Introduction to CDS Analysis

Conventional distance sampling (CDS) analysis refers to analysis of distance
sampling data using the methods described by Buckland et al. (1993, 2001).
Probability of detection is modeled as a function of observed distances from the
line or point, using robust, semi-parametric methods. The distances can be
recorded exactly, or grouped into non-overlapping intervals (also called “bins”).
Various methods are described for dealing with data where the objects are
clusters, rather than individuals.
The CDS analysis engine in Distance implements these methods. Detection
function, encounter rate and cluster size (where relevant) are estimated
separately, and the results are combined to estimate density or abundance.
Additional factors that influence density estimation (such as violation of the
assumption that all object at the point or line are seen) can be incorporated
through the use of multipliers.
The CDS engine allows one level of stratification – strata may be, for example,
geographic regions. Detection function, encounter rate and/or cluster size can be
calculated either by stratum or globally, and density can be estimated globally,
by stratum and/or by sample (a sample in this context is an individual line or
point). Variance can be estimated analytically, or via a non-parametric
bootstrap. The bootstrap can also be used to produce point and interval
estimates that include model selection uncertainty (see Model averaging in CDS
Analysis).
The mechanics of setting up and running analyses in Distance is outlined in
Chapter 7 - Analysis in Distance. This chapter covers topics that are specific to
the CDS engine, such as guidelines for approaching CDS analysis, how to deal
with grouped (binned) data, stratification and multipliers. (Many of these topics
also apply to the multiple covariates MCDS engine, so we recommend reading
this chapter as well as the next one before embarking on any MCDS analyses.)
This manual is designed to complement the standard text on Distance sampling
(Buckland et al. 1993 or 2001). Users of Distance are referred to that text for a
detailed explanation of conventional distance sampling, and an extensive set of
examples.

The CDS and MCDS engines are implemented as a single

FORTRAN program which is run from within the Distance interface. You can
also run the engine as a stand-alone program – for details see the Appendix -
MCDS Engine Reference.

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 87
 It is also possible to perform a CDS (or MCDS) analysis using the
MRDS engine – see Single Observer Configuration in the MRDS Engine in
Chapter 10 of the Users Guide.

Modelling the Detection Function

A central part of the analysis of distance sampling data is modeling of the
detection function. The CDS engine implements the robust key function + series
expansion (adjustment term) approach outlined by Buckland et al. (1993, 2001).
The candidate key functions offered are Uniform, Half-normal, Hazard-rate and
Negative exponential (this last function is not recommended, except for salvage
analyses; see Buckland et al. 1993, 2001). The candidate series expansions are
Cosine, Simple polynomial and Hermite polynomial. The user is free to choose
any combination of key function and series expansion, and there are a wide
range of options for both automatic and manual selection of the number and
order of series expansion terms that are fit to the data. However it is not
necessary or desirable to try every possible combination! A list of suitable
candidate models is presented in Buckland et al. (1993, 2001), and their selection
illustrated throughout the book.
In Distance, you implement a detection function model by defining a Model
Definition and then selecting the appropriate options from the Detection
Function Model tab. These options are outlined in the Program Reference
section on the Model Definition Properties Dialog.

Setting up a Project for CDS Analysis

For an example of how to set up a project for CDS, see Getting Started Example
1: Using Distance to Analyze Simple Data in Chapter 3. You should also read
Chapter 4 - Distance Projects, Chapter 5 - Data in Distance and Chapter 7 -
Analysis in Distance.

CDS Analysis Guidelines

The following is a condensed version of the guidelines discussed in Section 2.5
of Buckland et al. (2001), with some specific recommendations regarding the
organization of analyses in Distance. We have steered clear of defining specific
“cookbook” procedures for data analysis and, above all, recommend that you do
not unthinkingly use the Distance defaults.
Generally, we recommend that you start by thoroughly exploring your data, by
plotting histograms of recorded distances with the data sub-divided into a large
number of intervals. This can be done in Distance by creating a Model
Definition with an arbitrary model and manually defining a large number of
intervals in the Detection Function, Model Diagnostics page of the Model
Definition Properties dialog. Look for evidence of heaping, evasive
movement, outliers and possible gross errors. Line transect data that are
recorded as perpendicular distance and angle are better examined outside of
Distance, where the distribution of angles can be inspected and problems such as
rounding to zero degrees can be diagnosed.
At this stage, you should not be concerned with estimating density – indeed any
estimates produced are often distracting and misleading. Therefore, it is good
practice to tell Distance not to estimate density, by un-checking the boxes in the
Density row of the section labelled Quantities to estimate and levels of
resolution on the Estimate tab.

88 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 The issue of suitable truncation can be examined by creating Data Filters with
different truncation distances. Similarly, it may prove beneficial to group exact
distance data into intervals - this is also done in the Data Filter. This exploratory
phase is open ended, but you should strive to fully understand the data and
possible violations of the assumptions of distance sampling analyses (Buckland
et al. 2001, Section 2.1). In Distance, it may be worth grouping these
exploratory analyses into a suitably named Analysis Set.
Once the data have been properly prepared, and a decision has been made about
truncation and other Data Filter issues, the model selection phase can begin. We
recommend selecting a small number of sensible candidate models from those
available in Distance, and defining a separate Model Definition for each one.
This way, a separate analysis can be created for each model, and the AIC and
Delta AIC (or AICc and Delta AICc) columns in the Analysis Browser can be
used to sort and compare the analyses. Of course, other criteria should also be
used in selecting among the candidate models, such as goodness of fit (especially
near zero distance). A great deal of useful information about each model is
stored in the Analysis Details, Results tab.
In many cases these analyses will suggest additional explanatory work, so the
process of model selection and exploration is often iterative. Other issues, such
as the appropriate levels for estimating parameters (sample, stratum, global)
must also be considered (see Stratification and Post-stratification in this Chapter
for a discussion of some of these issues).
As the number of analyses defined and run starts to build up, it becomes worth
considering grouping the analyses into different Analysis Sets in the Analysis
Browser. The Analysis Components window can also be used to move
related Data Filters and Model Definitions so that they are positioned adjacent to
one another. The Comments section of the Analysis Details window for
each Analysis can be used to record pertinent information, such as what you
learnt by running the analysis.
At some point, you select a model you believe to be the best for the data set
under consideration. This is the time to consider making bootstrap estimates of
variance (see Model Definition, Variance Tab - CDS and MCDS in the
Program Reference), and beginning to make inferences from the abundance
estimates produced. In many cases there will be perhaps two or three models
that appear to fit the data equally well. Distance allows you to define multiple
models in the Model Definition, Detection Function Models tab - if you
create a Model Definition that includes all of the final models and specify
bootstrap variance estimates, then the estimated variance will account for this
uncertainty in model selection as well as the other sources of variation. This
approach has much to recommend it. (For more on this, see Model Averaging in
CDS Analysis in this Chapter).
The above guidelines give a broad overview of how the analyst might proceed.
These ideas are developed much more fully in Buckland et al. (1993, 2001), and
extensive examples are given to illustrate the approach.

Output from CDS Analyses

The CDS engine produces the following output:
• a summary of results in the Analysis Browser. For general
information about the Analysis Browser, see the section
Introduction to the Analysis Browser in Chapter 7. There are many
results statistics available, and you can select which ones are shown
independently for each analysis set using the Column Manager (see
Column Manager Dialog in the Program Reference). An
explanation of some of the columns is given in the section CDS
Analysis Browser Results.

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 89
 • a detailed listing of results in the Results tab of the Analysis
Details window. These are described in the following section,
CDS Results Details Listing.
• a log of the analysis, highlighting any possible problems, in the
Log tab of the Analysis Details window. For information about
troubleshooting problems, see Chapter 12 - Troubleshooting.
• (optionally) text files, containing the results listing, analysis log,
summary statistics, bootstrap statistics and plot data. For more
about these, see the section on Exporting CDS Results.

CDS Results Details Listing

When an analysis has run, a great deal of information is available in the Results
tab of the Analysis Details window. This information is split into pages, as
follows:
• Estimation options listing. Gives a summary of the analysis
options you chose.
• Detection Fct. A set of detection function pages for each subset
of data used for modeling the detection function. If the detection
function is estimated globally, there are one set of these pages. If
by stratum, there is one set for each stratum.
• Model fitting. Gives details of the models fit, and the final
model selected.
• Parameter estimates. A summary of the parameter
estimates for the final model selected, including correlations
among estimates.
• Plot: Qq-plot. (Not for interval data) Qq-plots are another
graphical method for assessing model fit – for more
information, see CDS Qq-plots in this Chapter.
• K-S GOF Test. (Not for interval data) Kolmogorov-
Smirnov and Cramér-von Mises tests of goodness-of-fit. For
more information, see CDS Goodness of fit tests in this
Chapter.
• Plot: Detection probability. Plot of the detection function,
superimposed on histograms showing the frequency of counts.
(These frequencies are scaled for point transects, see Buckland
et al. 2001)
• Plot: Pdf. Probability density function plot – only for point
transect data.
• Chi-sq GOF test. Table of observed and expected
frequencies in each histogram bin, together with a
χ 2 goodness-of-fit test. This test gives a measure of how well
the model fit the data, based on a comparison of the observed
and expected frequencies of observations within distance bins.

The χ2 test is known to be biased if expected cell counts are small. If

the expected counts are less than 2.0, Distance produces a second table, where
adjacent bins are pooled until the expected counts are greater than 2. This
procedure is rudimentary, and users can probably construct a better test by hand.

90 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 The previous three pages are designed to help you diagnose model
fit. By default, you get three sets of these pages, with the data divided into
equally spaced intervals, and the number of intervals being n0.5, 2/3n0.5 and
3/2n0.5 (where n is the count of objects). Instead of using the defaults, we
recommend you always define your own cutpoints.

These pages are probably the most important of the whole results
output, and you should check them carefully. Check the Model Fitting output to
see if any of the models that were fit did not converge, or hit any of the
constraints. Even if the model affected is not the one that was eventually
selected (if you're using automatic model selection), the selection process can be
affected if there was a problem in the fitting. Check the plot(s) and GOF tables
to look for evidence of lack-of-fit, and possible problems with the data such as
rounding and evasive movement. These issues are mentioned in this Chapter on
the page entitled CDS Analysis Guidelines, and are covered in more detail in the
Distance Book.

If there is a problem in the fitting routine, such as non-convergence, it

may be useful to look at the parameter estimates for each iteration of the fitting
algorithm. To do this, check the option "Report results for each iteration of the
detection function fitting routine" in the Model Definition Misc. Tab, and then
re-run the analysis.

If any of the parameter estimates hit the default upper or lower

bounds, you should consider setting bounds manually. You do this in the
Constraints page of the Detection Function tab in Model Definitions.

• Cluster size. A set of pages for each subset of data used in

estimating expected cluster size (e.g., one for each stratum). If
objects are not in clusters, these pages are omitted.
• Estimates. Gives the estimated expected cluster size,
including the size-bias regression results, if requested (the
default).
• Regression plot. Text-based plot showing size-bias
regression, if requested.
• Density estimates. One per subset of data for which density is
estimated (e.g., one for each stratum)
• Estimation summary. Set of pages containing tables
summarizing the results, and giving variance estimates and
confidence limits.
• Encounter rates
• Detection probability
• Expected cluster size (if objects in clusters)
• Density & Abundance
• Bootstrap summary. (Only if variance by bootstrap option
selected.) A set of pages similar to the estimation summary, but
with bootstrap point estimates, standard errors and confidence
limits. The bootstrap point estimate is the mean of the point
estimates from the bootstrap replicates (useful for model averaging

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 91
 – see Model Averaging in CDS Analysis). Two types of
confidence limits are given. The first use the bootstrap standard
error to generate parametric, log-normal confidence limits. The
second use the percentile method – i.e., for x% confidence intervals
the (x/2)th and (100-x/2)th quantiles of the bootstrap estimates are
given. In general, the latter confidence intervals are considered
more reliable.

The parametric bootstrap confidence intervals on density and

abundance use the same degrees of freedom as the original analysis, rather than
re-calculating the degrees of freedom using formula 3.75 of Buckland et al.
(2001).

For information about how to export the results text or plots into
another program, see Exporting CDS Results from Analysis Details Results.

CDS Qq-plots
Distance gives quantile-quantile (qq)-plots for all analyses that use exact data
(i.e., those where the data is not transformed into intervals in the Data Filter).
Qq plots are useful for diagnosing problems in the data such as rounding to
preferred values and other systematic departures from the fitted model. A major
advantage of these plots over the histograms of the detection function and
probability density function (pdf) is that they do not require the data to be
grouped into intervals. A disadvantage is that they require a little effort to
understand the output.
In statistics, qq-plots are used to compare the distribution of two variables – if
they follow the same distribution, then a plot of the quantiles of the first variable
against the quantiles of the second should follow a straight line.
To compare the fit of a detection function model to the data, a standard method
is to plot the fitted cumulative distribution function (cdf) against the empirical
distribution function (edf). The cdf, F(x), gives the probability of getting a
distance less than or equal to x for a given model. The edf, S(x), gives the
proportion of the data with distances less than x. (Note – this explanation
ignores tied values.) If the data fit the model, then the fitted cdf and edf should
be the same.
To make the qq-plot, the fitted cdf is evaluated for each observation. The data
are then sorted into increasing order i=1, … n and the edf is calculated as (i-
0.5)/n. The following plot shows an example where 58 of the 204 data points are
at 0 distance (yes, this is a real dataset!). The red dots show the data, and the
blue line is where they should lie if the fit of the model was perfect.

Example qq plot, showing a severe ‘spike’ at 0 distance

92 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 The cdf at 0 distance is 0, so the 58 points appear along the bottom left side of
the plot. Clearly in this case, the data do not fit the model. This can be
confirmed in Distance using the Kolmogorov-Smirnov and Cramér-von Mises
goodness of fit tests on the next page of results output.
The following plot shows an example where the fit appears quite good, with
most points close to the line and little systematic departure. The data have
clearly been rounded (e.g., to the nearest meter) as there are several data points
at each level of the cdf. Such rounding should not affect the reliability of the
parameter estimates at all.

For more on qq-plots, see Chapter 11 of Buckland et al. (2004). For options
associated with qq-plots in the CDS engine, see the Program Reference page on
Model Definition Properties, Diagnostics - Detection Function Tab - CDS and
MCDS. For information about how to export qq-plots (and other output) from
Distance into Word processors, spreadsheet and other graphing programs, see
Exporting CDS Results from Analysis Details Results.

CDS Goodness of fit tests

Previous versions of distance allowed users to test the fit of a model using χ 2
goodness-of-fit tests. A disadvantage of this test for ungrouped data is that the
data must first be put into intervals before the test can be performed, and the
selection of cutpoints can have a strong influence on the outcome of the test.
Therefore, we have added three tests of goodness-of-fit that operate directly on
the exact distances observed. These tests are not produced when the data are
analyzed in intervals.
Kolmogorov-Smirnov test
The Kolmogorov-Smirnov (k-s) statistic focusses on the largest difference
between the cumulative distribution function (the cdf, F(x)), and the empirical
distribution function (the edf, S(x) – see the previous topic for more on these
functions).
To calculate the statistic, the fitted cdf is evaluated for each observation, and
these cdfs are arranged in ascending order and indexed i=1, … n. The k-s
statistic is then given by
n (
Dˆ = max Dˆ + , Dˆ −
n n )
where Dˆ n+ ( ) ( )
= max i / n − Fˆ (xi ) and Dˆ n− = max Fˆ (xi ) − (i − 1) / n .

The k-s statistic can be used to test whether there is a significant departure
between the edf and cdf – in other words whether the data fit the model.

The upper tail probabilities (“p-values”) reported by Distance are

calculated by evaluating the following expansion:
∞
⎛ z ⎞
∑
2 2
P⎜⎜ Dn > ⎟⎟ = 2 ( −1) i −1 e −2i z
⎝ n⎠ i =1

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 93
 where z = Dˆ n n and n is the number of observations (Gibbons 1971, page 81).
This should be suitably accurate for practical application for sample sizes of
about 35 or greater.

Cramér-von Mises test with uniform weighting function

Unlike the k-s test, which focuses on the largest difference between the cdf and
edf, Cramér-von Mises (C-vM) family tests focus on the sum of squared
differences between cdf and edf. They are of the form:
∞
Q = n ∫ [F ( x ) − S ( x )] ψ ( x )dF ( x )
2

−∞

where ψ ( x ) is a weighting function that allows us to give different weights to

different parts of the distribution. If we give all observations the same weight
then we obtain the standard C-vM statistic, calculated by:
2
1 n
⎡ i − 0.5 ⎤
W2 = + ∑ ⎢ Fˆ ( xi ) −
12n i =1 ⎣ n ⎥⎦

The tail probabilities for W2 depend on sample size, and a closed

form expression for them is not available. Instead, we used simulation to
estimate critical values for W2 at a range of sample sizes from 5 to 1000 and at
α-levels of 0.001, 0.005, 0.01, 0.025, 0.05, 0.1, 0.15, and 0.2, 0.3, … , 0.9.
These values are stored in Distance, and the program uses linear interpolation to
construct a set of critical values for the observed sample size. It then reports
which of the critical values the observed W2 falls between, using the form: “αl <
p <= αh” where αl and αh are the bounding critical values.

Cramér-von Mises test with cosine weighting function

This is similar to the above test, but with a weighting function that puts more
emphasis on observations closer to 0 distance. The rationale is that these
observations have more influence on the estimated value of f(0) or h(0), so we
expect this test to have more power to detect influential departures from the
fitted function, and be more robust to departures at larger distances that don’t
unduly affect the f(0) or h(0) estimates.
The weighting function we use is cos(π xi/2w) where xi is the perpendicular
distance of observation i and w is the truncation distance. This leads to the
statistic
− 16n
C2 = +
π3
⎡⎛ ˆ 2i − 1 ⎞ ⎛ π ˆ ⎞ ⎤
⎢ ⎜ 2 F ( xi ) − ⎟ sin⎜ F ( xi )⎟ + ⎥
n ⎠ ⎝2
⎢⎝ ⎠ ⎥
n
2
∑
π i =1 ⎢ 4 ⎛π ⎞⎥
⎢ cos⎜ Fˆ ( xi )⎟⎥
⎣ π ⎝2 ⎠⎦
Tail probabilities for C2 are calculated and presented in the same way as for W2,
above.
These tests are also discussed in Chapter 11 of Buckland et al. (2004). For
options associated with goodness-of-fit tests in the CDS engine, see the Program
Reference page on Model Definition Properties, Diagnostics - Detection
Function Tab - CDS and MCDS.

94 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 About CDS Detection Function Formulae
Understanding the Detection Function Model Formulae

On various pages of the results details listing, Distance presents the detection
function model used and the parameters estimated. For example, under
Detection Function, Model Fitting, you may see:
Model
Hazard Rate key, k(y) = 1 - Exp(-(y/A(1))**-A(2))
Simple polynomial adjustments of order(s) : 4
A( 1) bounds = ( .00000 , 1.0000 )
A( 2) bounds = ( 1.0000 , 2.0000 )

and under Detection Function, Parameter Estimates:

Model
Hazard Rate key, k(y) = 1 - Exp(-(y/A(1))**-A(2))
Simple polynomial adjustments of order(s) : 4

Point Standard Percent Coef.

Parameter Estimate Error of Variation
--------- ----------- ----------- --------------
A( 1) 0.4582 0.3373
A( 2) 1.195 1.085
A( 3) -1.954 0.4343
f(0) 1.5652 0.49259 31.47
--------- ----------- ----------- --------------

Detection functions are modelled using the following general form:

g ( y ) ∝ key ( y )[1 + series ( y s )]
where g(y) is the detection function, key(y) is the key function, series(ys) is the
series adjustment, y is distance and ys is the scaled distance y/w where w is the
truncation distance. The function is scaled so that g(0) is 1. In the above
example, the key function is hazard rate and the series adjustment is one
polynomial term of order 4.
The formula for the hazard rate function is given in Buckland et al. (2001) as

(
Hazard rate: 1 − exp − ( y σ )− b )
where σ is the scale parameter and b is the shape parameter. This formula is
given in the output above, and it can be seen that parameter A(1) corresponds to
σ and A(2) to b.
The formula for the polynomial series adjustment is given in Buckland et al.
(2001) as

∑ a j ( y s )2 j
m
Simple polynomial:
j =2

where m is the number of adjustment terms, and aj is the parameter for the
adjustment term of order 2j. In the above output, parameter A(3) corresponds to
the order 4 adjustment term (i.e., where j=2).

In the MCDS engine, ys can also be y σ where σ is the scale

parameter of the key function - see Scaling of distances for adjustment terms in
the MCDS chapter.

Calculating probability of detection

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 95
 To calculate probability of detection at a given distance, y, you need to substitute
the parameter estimates into the formula
key ( y )[1 + series ( y s )]
g(y) =
key (0)[1 + series (0)]
For example, taking the results given above and assuming a truncation distance
w = 10, the probability of detection at y = 3 is

g(y) =
(1 − exp(− (3 0.4582) −1.195
))[1 − 1.954(3 10) ]
4

(1 − exp(− (0 0.4582) −1.195

))[1 − 1.954(0 10) ]
4

=
(1 − 0.8995)[0.9842]
(1 − 0)[1]
= 0.0989
To calculate average probability of detection over the surveyed strip, an easy
approach is to divide the interval (0, w) into a large number of evenly spaced
intervals, evaluate g(y) at each cutpoint and take the mean. (This is almost
equivalent to numerical integration of g(y)/w using the trapeziodal rule.)
Alternatively, use a numerical integration routine (e.g., the function integrate
in R) to integrate g(y) and divide the result by w.

Because g(y) is a nonlinear function of y, the accuracy of

calculated average probability of detection will be much less than the 4 or 5
significant figures Distance reports for the parameter estimates under Detection
Function, Parameter Estimates. For accurate estimates, it is better to use
the estimated parameter values given in the stats file (to 7 significant figures).
To do this, ask Distance to save the results stats file (see Saving CDS Results to
File in this chapter) and then look for the statistics numbered 101, 102, etc. in
module 2 (see MCDS Engine Stats File in the MCDS Engine Reference for more
about the stats file format).

Calculating effective strip width

The estimated effective strip width (line transects) or effective area (point
transects) in CDS analysis is given in the Results Details listing. However, there
are some circumstances when you may wish to calculate it outside of Distance
using the parameter estimates (one example is for MCDS analyses to calculate it
at a given covariate level).
For line transects, effective strip width is given by
w
μ= ∫ g ( y )dy
0

where g(y) is the probability of detection at distance y and w is the truncation

distance. Effective strip with can therefore be calculated by numerical
integration of g(y). Calculating g(y) from parameter estimates output by
Distance is described in previous sections of this chapter.
For point transects, effective area is given by
w
ν = 2π ∫ rg (r )dr
0

where g(r) is the probability of detection at distance r and w is the truncation

distance. Effective area can therefore be calculated by numerical integration of
g(r).

96 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 See the previous topic for a tip on how to get the best accuracy when
calculating statistics such as effective strip width (and average probability of
detection).

Understanding parameter indexing

To set starting values or bounds on parameters, it is important to understand the

order in which they are indexed. This can become quite complicated when there
are multiple strata and/or multiple models.
• Within a stratum/model, the key function parameters come first and
adjustment term parameters next. In the above example,
parameters A(1) and A(2) are key function parameters and A(3) is
the adjustment term parameter. Additional adjustment term
parameters would be A(4), A(5), etc
• When fitting detection function by stratum (or sample), parameters
are indexed sequentially between strata (or samples). For example,
if there were two strata in the above example, then the hazard rate
key function parameters would be A(1) and A(2) in stratum 1 and
A(4) and A(5) in stratum 2, and the adjustment term parameter
would be A(3) in stratum 1 and A(6) in stratum 2. You can see
these parameter indexes in the Detection Fct part of the output.
• When fitting multiple models, the parameters are indexed
separately. So, if in the Model Definition under Detection
Function, Models, you specified both a hazard rate key function
and a half normal key function, then the hazard rate parameters
would be indexed starting with A(1) and A(2) and the half-normal
parameter would be indexed starting with A(1).
• You can use the above rules to determine which parameters refer to
which strata when setting starting values. When setting bounds,
remember that only key function parameters are included in the
ordering.

The best way to be sure which parameter is which is to run the analysis
without starting values and check the Detection Fct part of the output.

• When there are multiple models and multiple strata, the parameters
are not indexed separately when printing results in the Density
Estimates section. Only the model selected in each stratum
contributes to the indexing. For example, imagine Distance is
choosing among hazard rate and half-normal key functions, with no
adjustments, and that there are two strata. In stratum 1 it chooses
half-normal and in stratum 2 it chooses hazard rate. Then in the
Density Estimates output, parameter A(1) corresponds to the
half-normal parameter in stratum 1 and parameters A(2) and A(3)
correspond to the hazard rate parameter in stratum 2.
• This means that in some cases the parameter indexes in the
Density Estimates part of the output can be different from those
in the Detection Fct part. Hopefully the output in each section is
self-explanatory. The important thing to remember is that it is the
parameter indexes in the Detection Fct part that are used for
setting starting values – the output in the Density Estimates part
is for display purposes only.

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 97
 CDS Analysis Browser Results
When an analysis is run, a summary of the results is given in the right-hand pane
of the Analysis Browser. You can select which statistics that are displayed
separately for each analysis set by using the Column Manager (click the
button).
Most of the columns that are available for selection have obvious interpretations.
However a few require some additional explanation or amplification:
• Many columns will appear blank in the Analysis Browser when the
analysis is stratified. For example, the number of parameters,
probability of detection and chi-square p columns will be blank if
detection function is estimated by stratum.
• One exception to the above is the model selection statistics AIC,
AICc, BIC, and LogL (and respective Delta AIC, Delta AICc, …).
When detection function is estimated by stratum, these statistics are
summed across the estimated detection functions, making it easy to
compare models where detection function is estimated separately
by stratum vs those where it is pooled.
• The goodness of fit Chi-square p value is for the last test performed
(if more than one diagnostic test is performed), after automatic
pooling has taken place - see the last “Chi-sq GOF” page of the
Analysis Details Results Tab.
• Both bootstrap and analytic estimates of coefficient of variation
and confidence limits for the abundance and density estimates can
be displayed. The bootstrap estimates use the percentile method
(cf. bootstrap in the Distance Book). Bootstrap estimates obtained
from the bootstrap variance estimate, assuming a lognormal
distribution for the density estimate, are available in the Analysis
Details Results Tab Bootstrap Summary page.
• Bootstrap point estimates of abundance and density are also
available – these are the mean of the estimates from the bootstrap
replicates. They are especially useful if you have run the bootstrap
with multiple key functions as they are then model-averaged point
estimates – see Model Averaging in CDS Analysis for details.

Exporting CDS Results

There are many reasons to want to export the results of analyses to other
programs. For example, you may want to:
• present a summary table of results from the Analysis Browser in a
report
• save part of the detailed results from the Analysis Details window
• copy a detection function graph into a spreadsheet program and
modify the formatting
• export the parameter estimates from Distance as a text file, as part
of a simulation
This section summarizes the various ways of getting results out of Distance.

Exporting CDS Results from the Analysis Browser

To copy the current analysis set to the clipboard, click the Copy to Clipboard
button on the main toolbar, or choose Analyses | Copy to Clipboard. The
column and row separators can be set in the General tab of the Preferences
dialog (Tools | Preferences).

98 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 Non-integer results are only shown in the Analysis Browser to a few
(usually 2) decimal places. However, if you copy and paste them into another
application (e.g., Excel) you can see them to 7 significant figures.

Exporting CDS Results from Analysis Details Results

Exporting Text
To transfer the results text, make sure a page of text is showing and then click on
the Copy to Clipboard button on the main toolbar, or choose the Analysis –
Results menu button Copy Results to Clipboard. Then paste into whatever
application you choose.
To transfer just part of the result text, highlight the text, right click and choose
Copy Selected Text.

Exporting Plots or Plot Data

To transfer a high-resolution plot, make sure the plot is showing and then click
on the Copy to Clipboard button on the main toolbar, or choose the Analysis
– Results menu button Copy Plot to Clipboard. You can only paste the plot
itself into applications that are designed to take picture objects, such as word
processors and spreadsheets. In your application, choose Paste Special, and
then choose the option “Picture” from the list of formats that appears (you can
also choose “Device Independent Bitmap”, but you will end up with a much
larger file).
Alternatively, you can paste the data that was used to generate the plot file into,
say, a spreadsheet and then regenerate the file yourself. This has the advantage
that you can then change the format of the plot. To do this, in your application,
choose Paste, or choose Paste Special and then the “Unformatted Text”
option.

If you wish to use Excel to recreate a plot from the plot data, there is a
simple macro available on the Support page of the Program Distance Web Site to
do this.

If you wish to use R to recreate a plot, the following instructions may

help. (1) Paste the data from the clipboard to a text file. Let’s say the file is
“plot.txt”. (2) Paste the following commands into R:

#this reads in the file just created

forplot<-read.table(file="plot.txt", header=T, sep="\t", dec=".")
#note, depending on your language, dec might be "," rather than "."
#this plots the detection function or pdf (if point transects)
plot(forplot$C1, forplot$C2, type="l", ylim=c(0,max(forplot$C4)),
xlab="Distance", ylab="Detection probability")
#Define labels as you wish
#this adds in the data bars
lines(c(0,0), c(forplot$C3[1], forplot$C4[1]))
lines(forplot$C3, forplot$C4)

The following code can be used to recreate qq-plots:

#this reads in the file just created

forplot<-read.table(file="plot.txt", header=T, sep="\t", dec=".")
#note, depending on your language, dec might be "," rather than "."
#this plots the detection function or pdf (if point transects)
plot(forplot$C1, forplot$C2, type="p", ylim=c(0,max(forplot$C4)),
pch = ".", xlab="Empirical distribution function",
ylab="Fitted cumulative distribution function")
#Define labels as you wish

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 99
 #this adds in the (0,0) (1,1) line
lines(c(0,1), c(0,1))

and the labels should be changed.

Saving CDS results to file

In the Model Definition Properties dialog, there are options to save files
containing summaries of the results of an analysis. Four files can be saved, all of
which are standard ASCII text files:
• Results Details File - see Misc. Tab - CDS and MCDS.
• Results Stats File - see Misc. Tab - CDS and MCDS.
• Bootstrap Stats File - see Variance Tab - CDS and MCDS.
• Plot File - see Detection Function Tab - CDS and MCDS.
These files may be useful in providing an interface between Distance and other
applications - for example you could write a spreadsheet macro to paste the
results stats file and extract information into spreadsheet cells. In addition, the
Bootstrap file is often useful for making diagnoses of problems encountered
while doing bootstrap resampling. The formats of these four files are given in
the MCDS Engine Command Language Appendix section Output from the
MCDS Engine.
Note that the results details file is the same as the text displayed in the Results
tab of the Analysis Details window for an analysis that has been run. You can
easily obtain this text by choosing the menu item Analysis – Results | Copy
Results to Clipboard or pressing the the Copy to Clipboard button on the
main toolbar. Similarly, you can obtain a copy of the plot data by displaying the
plot in Distance and pressing the Copy to Clipboard button. For more on this,
see Exporting CDS Results from Analysis Details Results.

Miscellaneous CDS Analysis Topics

Interval (Binned/Grouped) Data

Summary
If your data were collected in intervals (bins), then enter them as exact distances
and convert them to intervals in the Intervals tab of the Data Filter.
Details
In Distance, each record in the Observation data layer corresponds to one
observation, and this observation must be entered as an exact radial or
perpendicular distance. In reality, however, some surveys collect distance data in
distance intervals (bins, also called grouped data).
To enter these into Distance, enter each observation at the mid-point of the
interval. For example if your intervals span 0-10m, 10-20m and 20-50m then an
observation in the first bin would be entered as 5m, in the second bin would be
15m and in the third bin would be 35m.

If you observed, say 50 objects in the first bin of a transect you won't
want to have to create 50 records by hand and type the distance for each one.
Luckily there is a shortcut - simply create a record for the first object and enter

100 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 the distance (in this case 5m). Then double-click on the ID field to bring up the
multi-record add dialog. Select 49 and press append - this will automatically add
the other 50 records. See the Program Reference page Editing, Adding and
Deleting Records for more details.

When you have entered your data, the first thing you should do is tell Distance to
turn the data into intervals for analysis. Create a new analysis in the Analysis
Browser, and in the Analysis Details window, click on Properties… for the
default Data Filter. In the Intervals tab of the Data Filter click on “Transform
distance data into intervals for analysis” and enter your intervals.
Look at the Data Filter Truncation tab help page to find out about choosing the
level of truncation for your distance data.

Missing Data in CDS Analysis

Missing Distances
The conventional distance sampling engine is not designed to deal explicitly
with missing distance data. With good field methods, it should be very rare that
you detected an object but did not record a distance for it. However, if this does
occur you have two options:
• Discard the observations with missing distances and analyze the
data as usual. So long as the observations were not at zero
distance, this should cause no bias, as you are effectively just
making the detection function steeper by discarding some
observations that were made away from the line or point. A
disadvantage of this approach is that you are discarding data that
could help to estimate encounter rate, so the overall variance may
be higher than the second approach.
• Use a data filter to select only the observations with recorded
distances. Fit a detection function to these data, and record the
estimated probability of detection and SE. Enter these as
multiplers and the estimate density/abundance using the whole
dataset, as described under Multipliers in CDS Analysis. Note an
important assumption here is that the missing distances are missing
at random – for example it will not work if you are less likely to
record the distance for objects farther from the line or point. For
this reason, the first approach is probably safer. Note also that this
approach won’t work if you use stratification – in that case you’ll
need more than one multiplier (one for each stratum), and will have
to calculate the global density estimate by hand.

If you run an analysis with data that includes missing distances, the
CDS engine will issue a warning and exclude the observations with missing
distances from the analysis.

Missing Cluster Sizes

See Missing Cluster Size Data in CDS Analysis.
Missing Survey Effort
Nothing can be done about this – to estimate density you need to know how
much survey effort you expended!
Missing Study Area Size
See Unknown Study Area Size.

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 101
 Clusters of Objects
In many studies, the objects of interest (usually animals) occur in clusters
(schools, flocks, etc.). In this case, each observation represents a cluster, and in
addition to the distance from the transect to the cluster the observer also records
the cluster size.
Distance sampling theory is readily extended to include clustered populations, as
outlined in Buckland et al. 2001. Distance allows you to specify that objects are
in clusters during the New Project Setup Wizard. It then automatically creates a
field for cluster size in the Observation data layer, and specifies that objects are
clusters in the default Survey object.
The key decision in the analysis of clustered data is how to estimate the expected
cluster size at zero distance. Distance offers a number of options for this, as
explained in Section 3.5 of Buckland et al. 2001. In the Data Filter Properties,
under Truncation, there is an option to right-truncate the data for cluster size
estimation independently of the truncation for estimating the detection function.
In addition, in the Model Definition Properties, there is a Cluster Size tab which
gives a number of options for estimating expected cluster size (see the Program
Reference page on the Cluster Size Tab - CDS and MCDS).

Cluster size is usually an integer value, but Distance allows you to

enter non-integer cluster sizes. This may be useful, for example, if cluster size is
estimated independently by several observers – you could then improve accuracy
by using the mean of these estimates.

Missing Cluster Size Data in CDS Analysis

In some cases, you may not know the cluster size of an observation. For
example, some shipboard line transect surveys of cetaceans operate in two
survey modes: “passing mode” where observers guess the cluster size for each
sighting and “closing mode” where they break of the survey and go to an
observation to get a confirmed school size. One protocol would be to go into
closing mode for every 5th sighting, say. You may want to do an analysis on
only the confirmed school sizes, treating the other observations as having
missing cluster sizes.
To tell Distance that a cluster size value is missing, enter a value of -1 for that
cluster size. Observations with a cluster size of -1 are included in estimating the
detection function and encounter rate, but are excluded from estimation of
expected cluster size. If there are any -1 cluster size values in the data, Distance
issues a warning.

In previous versions of Distance, missing cluster sizes were coded as

0. Zero cluster sizes are now no longer treated as missing, but are analyzed (see
next topic).

Zero Cluster Sizes in CDS Analysis

There are some situations where you may observe clusters of size zero. For
example, imagine that you are surveying for a species of parasitic plant that only
occurs in a certain tree. You survey by walking a line transect looking for trees,
and if you find one you approach it and count the parasitic plants in the tree.
One way to enter these data would be by recording the distance to each tree, and
the number of parasitic plants in each tree as the cluster size. You may find no
plants in a tree, in which case you record a cluster size of zero. In this case, you
would not want to use a size bias method to get the expected cluster size, but

102 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 instead would use the mean cluster size (see Cluster Size Tab - CDS and MCDS
in the Program Reference).
An alternative analysis method would be to only include the trees where one
more parasitic plant was seen in the data – cluster size will then always be >1. A
disadvantage of this approach is that there will be less data available to fit the
detection function.
Another example of zero cluster sizes are in multi-species analyses where
species are encountered as mixed groups. You may have one field giving cluster
size for one species and a second field giving cluster size for the second species.
To do the analysis you would then a separate Survey object for each species and
use one survey object for each analysis (see Analysis with Multiple Surveys in
Chapter 7 for more on the use of multiple survey objects).

Stratification and Post-stratification

Stratification is a useful way of handling heterogeneity in the survey data, of

improving precision and reducing bias (see Distance Book, Section 3.8).
Stratification might be carried out by geographic region, environmental
conditions, cluster size, time, animal behaviour, detection cue, observer, or many
other factors.
The CDS engine can analyze data with one level of stratification per analysis.
There are two ways to deal with the stratification in Distance: (1) using the
stratum data layer and the Model Definition stratify option, and (2) using
extra data fields and the Model Definition post-stratification option. These
two methods are discussed in detail below.

Implementing Stratification via the Stratum Data Layer

This is the recommended approach when the strata are geographic. For example,
imagine a line transect study in which the study area has been stratified into two
separate regions: a small area of ideal habitat and a larger area of marginal
habitat. Animal density is expected to be higher in the ideal habitat and so it is
given a higher density of transects than the marginal habitat (see Buckland et al.
2001, Chapter 7).
In Distance, the two regions are entered as two records in the Stratum data layer
(called “Region” in this example):

Data explorer, showing two records in the stratum layer “Region”

To analyze data that has been entered this way, you should click on the option
Use layer type: Stratum in the Model Definition Properties dialog
Estimate tab:

In the lower part of the Estimate tab, you can then select the level of estimation
for density, encounter rate, detection function and cluster size (if the
observations are clusters of individuals). In the following picture, density is
estimated separately for each stratum as well as overall (globally), encounter rate
and cluster size are estimated by stratum, but detection function is estimated

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 103
 globally (i.e., pooled across strata). You can see this from the location of the
tick marks in the boxes.

Let’s assume you have no a prior reason to believe that detection

function is the same across all strata. Let’s also assume that you have plenty of
detections in each stratum. Then you can try an analysis where you allow the
detection function to vary by stratum (by ticking the “Stratum” box for detection
function), and try another where the detection function is fit to data pooled
across strata (i.e., estimated globally). Look at the goodness-of-fit statistics in
the Results tab of the Analysis Details windows, and compare the detection
function histograms. Assuming both stratum and global models produce
reasonable fits, the one with the lowest AIC is to be preferred. You need to use
the same Data Filter for both models, otherwise the AICs may not be
comparable. See the Stratify example sample project for an example. This topic
is discussed in Buckland et al. 2001, section 3.8.

Assuming you wish to estimate density by stratum and globally, you must tell
Distance how to combine the stratum estimates to produce a global estimate. If
your strata are geographic, the following options should be used:

The other options for the global density estimate are discussed in the following
sections.

Implementing Stratification via the Post-stratification

Option
This is the recommended approach to non-geographic stratification. For
example, imagine a shipboard line-transect survey conducted using two different
vessels. The vessels are assigned to transects at random, but it is known that
there are large differences in the effective strip width achieved by the two
vessels: one (“Beagle”) has a high crows-nest from which observers can see long
distances, while the other (“Bounty”) has no such raised platform. Distance is
quite robust to such heterogeneity in detection function (see Buckland et al. 2001
and Buckland et al. in prep), but nevertheless, gains in precision may be made by
modeling the heterogeneity. Therefore, the biologist wishes to allow for
differences in detection function by vessel.
In Distance, the vessel information is entered as an additional field called
“Vessel” in the Sample data layer (called “Line transect” in this example):

Data sheet part of the data explorer, showing the Vessel field in the sample layer “Line
transect”

104 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 (For more information about how to create additional fields in Distance, see the
Program Reference page about the Data Explorer. Additional fields such as this
can also be imported into Distance as with the other survey data - see Chapter 5
of the Users Guide on Data Import. To analyze data where the stratum is entered
as an additional field, click on the Post-stratify, using option in the Model
Definition Properties dialog Estimate tab, and select the appropriate data
layer and data field:

Example from Estimate tab, showing post-stratification by the Vessel field in the sample
layer

In this example, we want to estimate detection function by stratum (i.e., by

Vessel), so we fill in the Levels of Estimation options as follows:

Finally, we estimate the global abundance estimate as the mean of the stratum
estimates, but in this case we weight by survey effort.

When you choose to weight by survey effort, the tick box Strata are
replicates is enabled. In this case we do not want to tick that option. This, and
the other options, are discussed more in the next section, and in the Program
Reference page on the Model Definition Estimate Tab - CDS and MCDS.

The CDS engine only supports one level of stratification. This

means that you cannot use both geographic strata and Post-stratification at the
same time. If your survey design includes geographic stratification and you
create a Model Definition that defines a post-stratum field then the geographic
stratification is ignored.

If you wish to estimate overall density for multi-strata surveys, you can use the
Data Selection option in the Data Filter to set up a separate filter for each
level of your highest stratum. Then, do a separate analysis for each level and
combine the resulting density estimates by hand, remembering to include the
appropriate weightings. Variances can be calculated using the delta method
(Buckland et al. 2001). We hope to include multi-level stratification in a future
release of Distance.

Situations where the Post-stratification option is useful

This section gives various post-stratification scenarios and outlines the way you
would set the data up and the options you would choose under Quantities to
Estimate and Level of Resolution on the Estimate tab of the Model
Definition. More on these options is also given in the Estimate tab page of the
Program Reference.
1. Where there are different types of survey effort.

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 105
 In these situations you create an extra field that indexes the type of effort. You
post-stratify on this field and estimate density as the mean of the post-stratum
estimates, weighted by survey effort:

One example is the scenario described in the previous section, where there is
observer heterogeneity. In this case, you would add an extra field in the sample
data layer for observer (/vessel/survey party/etc.), and then post-stratify by this
field. The global density estimate is given as the mean of the stratum estimates,
weighted by survey effort.
Another example is where the study area is surveyed in multiple time periods,
using a different set of samples (transects) in each time period. Even without
wanting to post-stratify, it would be a good idea to add an extra field that
indicates the time period of each transect in the sample data layer, as this would
enable you to use the Data Selection feature of the Data Filter to pick out only
certain time periods for analysis. Post-stratification becomes useful if you want
a combined estimate of the average density over all periods. This is done by
post-stratifying on the time period field, and asking for a combined estimate of
density that is the mean of the post-stratum estimates, weighted by survey effort.
If observers, methods and conditions were the same at all time points it would be
reasonable to investigate the possibility of pooling the detection function over
the time periods.
A third example is similar to the previous one, in that the study area is surveyed
at multiple time points, but in this case the same set of samples (transects) were
used at each time point. You could set up the project in the same way as for the
previous case, but this would necessitate entering each transect into the sample
data layer once for each time period. Instead, you may consider adding the time
period field to the Observation data layer - this way each transect only has to
appear in the sample data layer once, while you indicate the time period that
each object was observed. To estimate mean density over the whole study, you
post-stratify on the time period column in the observation data layer and estimate
overall density as the mean in the post-strata. One small problem occurs in this
scenario when you come to estimate variance - each stratum in each year will be
treated as independent when in fact they are not. We hope to address this in a
future release of Distance.
Strata as replicates when there are different types of survey effort
In all of the above cases, we chose to weight by survey effort. When you weight
by survey effort, there is an option to treat the strata as replicates. This affects
how the variance of the global density estimate is calculated. Ticking the Strata
are replicates option means that you consider the strata you surveyed to be a
random sample from some larger population of possible strata that could have
been surveyed.
For example, consider the case when we survey at multiple time points – say
multiple days during a year. We may consider the days we surveyed to be a
sample from all those in the year, and we want to make inferences about the
average density of animals during the year. In this case we tick the Strata as
replicates option. Our estimate is then the effort-weighted mean density, and
our variance is calculated from the variation in density between days – i.e.,
treating strata as replicate samples.
On the other hand, let’s imagine that our multiple time points are actually years,
and we only surveyed in two years (we pooled the data within year). We could
consider the two years of data to be a sample from some larger set of possible
years – but this does not seem very useful and in any case a sample of two is not
very large for making inferences about average density over this larger set of
years. Instead, we will make inferences only about the average density over the
two years we sampled. We do not tick the Strata as replicates option. Our

106 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 estimate is still the effort-weighted mean density, but now the variance is
calculated from a weighted average of the stratum variances.
The difference between the two scenarios is known in the statistical literature as
treating the strata as random effects (the first scenario) or fixed effects (the
second scenario). The right one for your study depends on the inferences you
are making – is it to the average density over a larger set of strata from which
you have a random sample (random effect), or is it just to the average density
over the strata in which you sampled (fixed effect).
The variance calculation is given in more detail in the Program Reference page
on the Model Definition Estimate Tab - CDS and MCDS.
2. Where there are different types of object (animal) in the
population.
If the population can be divided into different “sub-populations”, each with
different encounter rates or detection functions, then it may be possible to
increase precision through post-stratification. This is done by creating an extra
field in the Observation data layer that indexes the sub-population type. You
then post-stratify on this field, with the global density estimate as the sum of the
post-stratum estimates:

One example of this would be where male and female animals have very
different detectabilities. For each animal, its sex would be entered as an
additional field (of type "Other") in the Observation data layer. In the Model
definition, you would choose Post-stratification by the sex field.
3. Where there is not enough data to estimate a detection function
for some subsets of the study.
For example, in a multi-species study it is often not possible to estimate f(0)
reliably for the rarer species. In this case, it may be acceptable to estimate the
detection function by pooling over similar species. To do this, you would add a
column to the Observation data layer for species name or ID. Then, define a
Data Filter that uses the Data Selection to include only the species for which you
wish to pool the detection function. In the Model Definition, post-stratify by
species and choose the following Levels of estimation:

You are not interested in obtaining a global estimate of density, as density

pooled over species is not meaningful.
To test your assumption about the detection function being similar among
species, it may be possible to do an analysis with detection function estimated by
species (by “stratum”) and to compare the AICs, although in this scenario there
are often too few observations for the rare species to make reliable estimates of
the detection function.
This situation can also be addressed using Multipliers - see the topic on
Multipliers in CDS Analysis.

Variance Estimation in CDS

The methods of variance estimation in CDS analysis are detailed in Buckland et
al. (2001), section 3.6. The default method is an analytic variance estimate, but
it is also possible to select a nonparametric bootstrap. A summary of the

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 107
 methods used is given here; this can be safely skipped on first reading through
this manual.
Variance options are chosen in the Model Definition, Variance tab – for more
on these options, see Variance Tab - CDS and MCDS in the Program Reference
appendix.
Analytic variance estimation
Density can be estimated globally, by stratum and/or by sample. At the lowest
of the levels requested (e.g., at the stratum level of both globally and by sample
are selected), variance estimates for encounter rate, detection probability and
expected cluster size are combined using the delta method (formula 3.68 of
Buckland et al. 2001) to give the variance of the density estimate at that level.
Lognormal confidence intervals are calculated using formulae 3.71-3.74 except
that t-based limits are calculated using degrees of freedom calculated using the
Satterthwaite method given in formula 3.75.
If there are any multipliers (see Multipliers in CDS Analysis) with non-zero
variance these are included in formula 3.68 as extra terms. If they have a non-
zero degrees of freedom they are also included in the degrees of freedom
calculation of equation 3.75. If degrees of freedom for the multiplier is not
specified, it is assumed zero, and the multiplier is omitted from both the top and
bottom line of equation 3.75.
By default, encounter rate variance is calculated using the empirical between-
sample variation in encounter rate, as detailed in section 3.6.2 (formulae 3.77-
3.82). Alternatively, the user may specify that encounter rate variance follows a
Poisson or overdispersed Poisson distribution (see Variance Tab - CDS and
MCDS). In this case, the encounter rate variance is assigned zero degrees of
freedom and the encounter rate term is omitted from both the top and bottom line
of equation 3.75.

When there is only one line, the default option is to set encounter
rate variance to zero – see Analysis of Data from a Single Transect in CDS.

When density is requested to be estimated at multiple levels (e.g., by stratum and

globally) then the higher density estimates are calculated as a weighted average
of the lower estimates (see Estimate Tab - CDS and MCDS in the Program
Reference for how to specify what weighting to use). In this case, the formulae
in section 3.7.1 are used to calculate variance of the estimate at the higher levels
– details of exactly which formula is used in which circumstance are given in the
Program Reference under Estimate Tab - CDS and MCDS.
Bootstrap variance estimation
The bootstrap is a robust procedure for estimating variance and confidence limits
(among other things). Details of the bootstrap implemented in the CDS engine
are given in Buckland et al. (2001), section 3.6.4, and a description of how to tell
Distance to produce bootstrap estimates is in the Program Reference appendix
under Variance Tab - CDS and MCDS. Note that the bootstrap can also be used
for producing estimates averaged across a number of candidate detection
function models – for more on this see Model Averaging in CDS Analysis in this
chapter.

Multipliers in CDS Analysis

In Distance sampling, there are often situations where the standard methods
produce a density estimate that is only proportional to the true density. For
example, when detection probability on the trackline (g0) is less than one, the

108 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 true density is the Distance density estimate divided by g0. Another example
comes from cue counts, where the density of animals is the density of cues
divided by the cue rate.
In Distance, such factors are dealt with using Multipliers. You enter the
multiplier value in the global data layer and in the Multipliers tab page of the
Model Definition Properties, you tell Distance which multipliers you want to use
in your analyses and whether they divide or multiply the density estimate.
Distance then scales you density estimate appropriately.
Some multipliers are known with certainty. One such multiplier is the “sampling
fraction” – the proportion of each line or point surveyed. Normally this is 1, but
in some cases you may only survey one side of a transect line, so the sampling
fraction is 0.5. Another example is a cue count survey, where the sampling
fraction is the proportion of a full circle that is covered by the observation sector.
A third example is when all points or lines are visited multiple times – then the
sampling fraction is the number of visits. In these cases, you tell Distance to
create a field for the multiplier, enter the appropriate value, and tell Distance to
multiply the Density estimate by the value in this field.

If the sampling fraction is not the same for all lines or points, you
account for this by adjusting the survey effort at the data entry stage. For
example, if all your transects were 10km long, but you visited some 3 times, then
you set the survey effort for these to 30km, and leave the others at 10km. In this
situation you don’t need a sampling fraction multiplier.

Other multipliers are based on estimates from other experiments. For example,
in a cetacean survey, g(0) may be less than one because some animals are below
the surface and so not available for detection. You may have estimated g(0)
based on a separate experiment where you follow a sample of animals and record
the proportion of time they are on the surface. In these cases, your estimate of
the multiplier will have uncertainty associated with it. If you want this
uncertainty to be reflected in the variance of the final density estimate, you do
this in Distance by having additional fields for the multiplier standard error (SE)
and degrees of freedom (df). Note that you can have fields for both SE and df,
or just the SE. In this latter case, Distance assumes the DF for the multiplier is
infinity. (Another way to specify infinite DF is to have a field for DF containing
the value 0.0) Degrees of freedom of the multiplier affect the DF of the density
estimate as the density estimate DF is calculated using the Satterthwaite
approximation (formula 3.75 of Buckland et al. 2001). This in turn affects the
log-normal confidence limits on the density estimate.
In the Setup Project Wizard you are given the chance to create fields in the
Global data layer for a number of common multipliers. You can also add more
fields manually after the project is created using the Append Field button in the
Data Explorer (see Data Explorer in the Program Reference).

In this version of Distance you can only define Multipliers in the

global data layer. This means that you cannot account for factors that vary by
stratum or transect. In future versions of Distance we hope to offer the ability to
define multipliers at any level of the data hierarchy.

If you use the Setup Project Wizard to define your multiplier fields,
then they will appear automatically in the Multiplier tab in Model Definition
Properties for the default CDS analysis. For these fields, Distance also knows
whether they should multiply or divide the density estimate. The default value
for a multiplier created by the wizard is 1.0, with SE 0 and DF 0 (i.e., infinity) –

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 109
 in other words a multiplier that doesn't affect the density estimate at all! It's up
to you to enter appropriate values into the multiplier fields.

It is a good idea to follow Distance's convention for naming your

multiplier, SE and DF fields. If the multiplier field name is, for example, Nest
Prod Rate (for Nest Production Rate in indirect surveys of nests), then the
corresponding SE field should be Nest Prod Rate SE and DF field should be
Nest Prod Rate DF. This makes it much easier to recognize which SE and DF
goes with which multiplier.

If you are estimating variance using the bootstrap, be aware that

the variance due to any multipliers is not included in the bootstrap variance. To
do this, we would have to resample the multiplier value in each bootstrap
iteration, according to some distribution – a possible feature to add to a future
version of Distance…

In fact, the CDS analysis engine only recognizes the multiply

operator - i.e., all multipliers multiply the density estimate. So, the Distance
Interface uses a trick to pass multipliers to the Analysis Engine when the
operator is divide. In this case, to pass the multiplier value, it takes 1 / (input
value). To pass the multiplier SE, it takes SE / (input value2). You can see this
happening if you look at the first page of the Analysis Details Results tab for an
analysis that has been run with multipliers, where the operator is a divider.
About half-way down the first page is a small table listing the multiplier values
and SEs used. If, for example, you used a multiplier with a value of 0.9 and an
SE of 0.5, you'll see that the actual value passed in was 1.1 (i.e., 1 / 0.9) and the
SE was 0.6128 (i.e., 0.5 / (0.92) ).

Additional Uses of Multipliers

This section lists an example of another use for multipliers. If you think of any
more, let us know and we’ll include them in future versions of the help file.
In a multi-species study it is often not possible to estimate detection probability
reliably for the rarer species. In this case, you may want to assume that the
probability of detection for a rare species is the same as that for a similar, more
common one. You can do this using multipliers:
8. Define a Data Filter that uses the Data Selection tab to include only
the more common species. Perform a normal distance analysis and
note down the estimated P (detection probability) and P SE
(standard error) from the "Density Estimates/Global" page of the
Analysis Details Results.
9. In the Data Explorer, create a new multiplier, and multiplier SE and
enter the estimated P and P SE in the cells. (You can do this using
copy and paste - try highlighting the value in the results page and
clicking the right mouse button).
10. Now, define a Data Filter that uses the Data Selection tab to
include only the rare species of interest. Define a Model Definition
that uses a Uniform key function (under Detection Function
Models) and 0 adjustment terms (under Detection Function
Adjustment Terms). Include the new multiplier and multiplier SE
in this Model Definition under the Multipliers tab.
11. When you run the analysis, Distance will fit a uniform distribution
to the detection function with no adjustments and so will estimate

110 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 probability of detection as 1.0. The multiplier you created will
allows you to specify the probability of detection yourself.
Variance of the density estimate is calculated correctly
automatically.

Another approach to this scenario is to estimate a pooled detection

function for both the common and rare species. This is outlined in the previous
page on post-stratification.

Another example of the use of multipliers is given in the Getting Started chapter,
Example 2: More Complex Data Import.
A third example would be when some distances are missing from the dataset,
and you are confident that these are missing at random (i.e., it isn’t just the
farthest away distances that are missing). Then you could fit a detection
function to the data for which you have distances, enter the estimated detection
probability (and associated SE and df) as a multiplier and estimate density using
the whole dataset. For more on this, and a perhaps better approach, see Missing
Data in CDS Analysis.

Model Averaging in CDS Analysis

When using AIC to select among alternative candidate models of the detection
function, it is not unusual to find that more than one model have similar AIC
scores (perhaps differing by AICs of 2 or fewer). When this happens, more
reliable inferences can be obtained by basing the final results on an AIC-
weighted average of these plausible alternative models (Buckland et al. 1997;
Burnham and Anderson 2002). To do this:
• Create a new Model Definition.
• In the Detection Function, Models tab, click on the + button to
create one line for each candidate model. For each one, set the
appropriate key function and adjustment terms. Select the option to
Select among multiple models using AIC
• In the Variance tab, tick on Select non-parametric
bootstrap, and set appropriate options for the level of
bootstrapping and number of bootstraps.
When you run an analysis using this model definition, each bootstrap replicate
will use AIC to choose among the candidate models. The bootstrap point and
interval estimates you get will then be an average over all the replicates, and so
will include uncertainty as to which model is best.

You can only use this approach to include different candidate key
function + adjustment term combinations. There is currently no way within
Distance to do model averaging over, say, global and by-stratum analyses, or
CDS and MCDS models, or different covariates within an MCDS model. This
would have to be done by writing an external bootstrap routine and running the
analysis engine as a stand-alone program (see Running CDS Analyses From
Outside Distance).

For more on the bootstrap options, see Variance Tab - CDS and MCDS in the
Program Reference.

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 111
 Sample Definition in CDS Analysis

Sample definition allows you to specify the data layer to be used in the
estimation of the encounter rate variance.
For example, imagine you have line transect data from a survey where 3 lines are
divided into 4 short segments. The spacing between segments is approximately
the same as the spacing between transects, so a map of the design is as follows:
- - - -
- - - -
- - - -

The data are stored in distance in a Global layer called “Study area”, a Sample
layer, “Transect”, a Sub-sample1 layer, “Segment”, and an Observation layer,
“Sightings”:

Data layers view from the example project

When you come to analyze these data, you must choose whether to treat the 3
transects or the 12 segments as the samples, for estimating variance in encounter
rate. You make this selection in the Sample definition section of the
Estimate tab of the Model Definition Properties:

Selecting the layer type to use as sample

In this case, because the distance between segments is the same as the distance
between transects, it is valid to treat each segment as a separate sample. So you
would choose the layer type “SubSample1” as the sample definition.
If the between-segment spacing was much less than the between-transect
spacing, then you would choose the layer type “Sample” as the sample
definition, and Distance would pool the data from the segments on each transect
for estimating encounter rate variance.

Unknown Study Area Size

If you don't know the size of the study area you sampled, you can obtain density
estimates from Distance, but not abundance estimates.
When entering your data, leave the area of each stratum at its default value of 0.
When you come to analyze the data, Distance will automatically detect that the
area is 0 and will not provide an abundance estimate.

Restricting Inference to Density or Abundance in

the Covered Region in CDS Analysis
Normally in distance sampling, we lay down a series of line or point samplers
within some large study area and use the observations from these samplers to

112 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 make inferences about the density or abundance of animals in the study area.
However, in some (rare) cases, we may just want to make inferences about the
density or abundance of animals in the region covered by the samplers. One
example where we actually survey the whole study area – for example in a
simulation experiment, where we lay out objects such as golf tees or wooden
stakes in a strip and then have observers do a line transect experiment within that
strip. Another example is when we do not lay out the transect lines using a
survey design with an element of randomization in it, but lay the lines
purposively. In this case, we cannot guarantee that our survey lines will be
representative of the study area and so one approach is to restrict inferences only
to density in the region actually covered by the lines. (Another is to use a
model-based approach such as that of Hedley et al. 2004)
Restricting inferences in this way does not affect the density estimate at all.
Abundance is simply density multiplied by the area covered (the sum of the
areas of the samplers). The big difference comes in the variance. Normally in
distance sampling, there are three components that make up the variance of the
density or abundance estimate: variance from estimating the detection
probability, variance from estimating population mean cluster size and variance
from spatial variability in encounter rate between samplers. The third
component is often the largest, and typically it is estimated from the empirical
variation in encounter rate between samplers. When we only wish to make
inferences about density or abundance in the covered region, the only relevant
sources of uncertainty are the first two – the third is no longer relevant.
If we wish to restrict inferences in this way, how can we ensure that Distance
sets the encounter rate variance to zero? The trick is as follows. For the CDS
(and MCDS) engine, in the Model Definition, Variance tab, under Analytic
variance estimate, choose the option Assume distribution is Poisson,
with overdispersion factor and set the factor to 0. Once you’ve run the
analysis, you can check in the results that the encounter rate variance is zero. If
you’re only interested in density, this is all you have to do. If you also want the
correct abundance estimate, you need to set study area (either in the global layer
or stratum layer, if you have strata) to be the sum of the area of the samplers
(either globally or by stratum) – i.e., the total length times twice the truncation
width for line transects or number of points times truncation radius for point
transects.

Analysis of Data from a Single Transect in CDS

In general, it is bad practice to try to estimate density in some large study area
(or stratum) using just a single transect (see, e.g., Buckland et al. 2001 section
7.2). When data come from a single transect, it is not possible to use the default
method to estimate encounter rate variance – that is using the empirical between-
line variation in encounter rate. Instead, the CDS (and MCDS) engine issues a
warning, and assumes that encounter rate variance is zero. The resulting
variance is appropriate if you only wish to make inferences about the density or
abundance of animals in the area actually sampled – possibly a wise choice when
there is only one line. For more about this see the earlier section on Restricting
Inference to Density or Abundance in the Covered Region in CDS Analysis.
An alternative, if you wish to try to make inferences about the whole area from
one line, is to assume the distribution of observations is a Poisson or
overdispersed Poisson (an overdispersion factor of 3 has been suggested in a
related context by Buckland et al. 2001: section 7.2.2). You select these options
in the Variance tab of the Model Definition – see Variance Tab - CDS and
MCDS in the Program Reference appendix.

User's Guide Distance 6.0 Beta 5 Chapter 8 - Conventional Distance Sampling Analysis • 113
 Running CDS Analyses From Outside Distance
The CDS engine is implemented as a stand-alone FORTRAN program,
MCDS.exe. This program is called behind the scenes by Distance when you
press the Run button on the Analysis Details Inputs tab.
Some users may wish to run the engine from outside the Distance interface –
either from the Windows command line or from another program. For example,
you may want to automate the running of analyses for simulations, or you may
want to perform a more complicated bootstrap than Distance allows.
Full documentation for running MCDS.exe is provided in the Appendix - MCDS
Engine Reference.

114 • Chapter 8 - Conventional Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
Chapter 9 - Multiple Covariates
Distance Sampling Analysis

Introduction to MCDS Analysis

This chapter gives an outline of multiple covariates distance sampling (MCDS)

in Distance, together with guidelines for approaching MCDS analyses. The
methods are described in more detail in Marques (2001), Marques and Buckland
(2003) and Marques and Buckland (2004) and Marques et al. (2007 – see
Bibliography). These last two texts are recommended reading to anyone using
these methods. The dataset analyzed in Marques et al. (2007) is included with
Distance as a sample project (see the Sample Projects page of Chapter 3,
Amakihi project). Note that MCDS is an advanced topic that should only be
considered once you are familiar with conventional distance sampling.
The standard method for performing MCDS analysis in Distance is by using the
MCDS Engine, as introduced in the next topic Introducing the MCDS Engine. It
is also possible to perform an MCDS analysis using the MRDS engine – see
Single Observer Configuration in the MRDS Engine in Chapter 10 of the Users
Guide.

Introducing the MCDS Engine

The multiple covariates distance sampling (MCDS) engine extends the key
function + adjustment term approach of the CDS engine, allowing additional
covariates into the scale parameter of the key function (via a log link function).
This means that the covariates are assumed to influence the scale of the detection
function, but not its shape (see figure, below). In other words, we are assuming
the covariates affect the rate at which detectability decreases with distance, but
not the overall shape of the detection curve.

Example estimated detection functions, where cluster size (Sbar) is the covariate. The
basic shape of the function is the same (half-normal), but the effective strip width is wider
at cluster size 750.

User's Guide Distance 6.0 Beta 5 Chapter 9 - Multiple Covariates Distance Sampling Analysis • 115
 Of course, it is possible for covariates to affect both the shape and
scale of the detection function. Such models could be fit in Distance (for factor
covariates) using stratification. There is, however, some evidence that at least
some covariates may only affect the scale: Otto and Pollock (1990) examined the
effect of cluster size and distance on the detection function of graduate students
searching for beer cans. They found that a model where cluster size influenced
only the scale of the detection function fit the data best.

When adjustment terms are included in the model, it is possible for

covariates to influence both the scale and shape of the fitted function – see the
section Scaling of Distances for Adjustment Terms further on in this chapter for
more information about this.

The MCDS engine is implemented as a FORTRAN program which

is run from within the Distance interface. You can also run the engine as a
stand-alone program – for details see the Appendix - MCDS Engine Reference.

Setting up a Project for MCDS Analysis

Before you contemplate an MCDS analysis, you should be familiar with the
setup and analysis of CDS data (see previous chapters).
Covariates are simply entered or imported as extra fields of data, in the
appropriate data layer. For example, covariates such as habitat or observer might
be associated with each point transect, and so be entered as extra fields in the
sample layer, with one record for each point. Alternatively, covariates such as
the sex or species of detected animals might be associated with each observation,
and so be entered as extra fields in the observation layer.
Then, for an example of importing data containing an extra field for species, see
Getting Started Example 2: More Complex Data Import in Chapter 3.

Defining MCDS Models

This section describes how MCDS models are defined. More details are given in
Marques and Buckland (2001, 2004) and Marques et al. (2007).
Like the CDS engine, the MCDS engine uses a key function + series expansion
formulation to model the detection function. The difference is the incorporation
of covariates in addition to distance into the key function. So,
g (x, z ) = key(x, z )[1 + series(x )]

where g (x, z ) is the probability of detecting an object at distance x and covariates

z , key(x, z ) is the key function, and series (x ) is the series expansion.

The covariates are assumed to affect the scale parameter of the key function, σ .
The scale parameter controls the “width” of the detection function. Of the four
key functions available in the CDS engine, the half-normal and hazard-rate are
both available in the MCDS engine; the other two either do not have a scale
parameter (uniform), or provide an implausible shape close to 0 distance
(exponential).

{
Half-normal key function, exp − x 2 2σ (z )2 }

116 • Chapter 9 - Multiple Covariates Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 {
Hazard-rate key function, 1 − exp − [x σ (z )]− b }
The scale parameter is modeled as an exponential function of the covariates:
σ (z ) = exp(β 0 + β1 z1 + β 2 z 2 + Kβ q z q )
where q is the number of covariate parameters. The term inside the brackets is
akin to a linear model – the β ’s are parameters to be estimated, with β 0
corresponding to the intercept. The exponential term prevents the scale
parameter from being negative.

Output in distance is actually given with first parameter, β 0 ,

( )
outside the exponential – i.e.,: σ ( z) = β 0 exp β1 z1 + β 2 z 2 + K + β q z q
This is because in this formulation, β 0 parameter estimates can be directly
compared with estimates of the scale parameter from the CDS engine.

Factor and Non-factor Covariates in MCDS

When setting up MCDS models in Distance, you need to distinguish between
factor covariates, and non-factor covariates. This is because the type of
covariate affects how the model is parameterized. Parameterization in the
MCDS engine is briefly outlined below, but will be covered in more detail in any
standard book on linear models.
Factor covariates classify the data into more than one distinct category or level.
Examples include habitat, sex, observer, etc. For factor covariates, the actual
covariate values are not important, as a separate parameter is defined for each
factor level. The last factor level is incorporated into the intercept. For example,
imagine a habitat covariate with 3 levels: “grassland”, “scrub”, “wood” (factor
covariates are sorted into alphabetical order when run in Distance). The model
for the scale parameter will be:
σ ( z ) = exp(β 0 + β1 z1 + β 2 z 2 )
where the β 0 parameter corresponds to the effect of wood, β1 corresponds to
the additional effect of grassland, above that of wood alone, and β 2 corresponds
to the additional effect of scrub, above that of wood alone. z1 and z 2 are
indicator variables (i.e., they take values 0 or 1 to indicate which habitat each
observation corresponds to).
Non-factor covariates must be numeric, and are treated just as in standard linear
regression. For example, imagine a non-factor covariate “Beaufort” (Beaufort is
a measure of wind speed). The model for the scale parameter will be:
σ ( z ) = exp(β 0 + β1 z1 )
where β 0 is the intercept (i.e., the effect of Beaufort when Beaufort is 0)
and β1 is the slope.

Some covariates could be either factor or non-factor. In the

example above, we could specify “Beaufort” as a factor covariate, with one level
for each Beaufort level. Specifying Beaufort as a non-factor covariate means
that we have to fit fewer parameters, but also assumes an exponential
relationship between the scale parameter and Beaufort level. Which is best
depends on the data – we could use a model selection criterion (such as AIC) to
help choose.

User's Guide Distance 6.0 Beta 5 Chapter 9 - Multiple Covariates Distance Sampling Analysis • 117
 More complex models are constructed along similar lines. For example, a model
containing habitat as a factor covariate and Beaufort as non-factor covariate will
be:
σ ( z ) = exp(β 0 + β1 z1 + β 2 z 2 + β 3 z 3 )
where the β 0 parameter corresponds to the effect of wood and the intercept of
Beaufort, β1 corresponds to the additional effect of grassland, β 2 corresponds to
the additional effect of scrub, and β 3 is the slope of the Beaufort covariate.
Interactions between covariates can be modeled by creating a new field in the
Distance data sheet that contains one covariate multiplied by the other. We hope
to extend the MCDS engine to allow easier modeling of interactions in a future
version.

Estimating the Detection Function at Multiple Levels

In some cases, it is useful to fit the detection function at one level, but use the
fitted model to estimate f(0) (or h(0) for point transects) and probability of
detection at a lower level. For example, imagine you want to estimate density by
stratum, but don’t have enough data to fit a separate detection function for each
stratum. One solution is to fit a global model for the detection function, but
using stratum (or lower) level covariates. You can then use the fitted model to
estimate a separate average f(0) (or h(0)) in each stratum, using the covariates
that apply to that stratum. In this way, both the stratum and global estimates of
density will hopefully be less biased than if you had used the global detection
function estimate.
The same argument applies if you want to estimate density by sample (e.g.,
transect) – for example as input to another analysis such as a spatial or habitat
modelling exercise. One rarely has enough data to fit a separate detection
function to each transect. However, by including transect (and lower) level
covariates in a global or stratum level detection function model, you can then
estimate average f(0) at the transect level using the observations, and their
corresponding covariates, that apply to that transect.
To fit the detection function at one level, but estimate at a lower level, you tick
both levels in the Estimate tab of the Model Definition Properties dialog (see
Estimate Tab - CDS and MCDS in the Program Reference for further details).
Variance estimation
A restriction if you fit detection function at one level and estimate at a lower
level is that the variance of density at the higher level must be estimated using
the bootstrap. This is because Distance estimates density at the higher level by
combining the density estimates from the lower level. The lower level density
estimates were calculated using the same detection function model, and so are
not independent, so the appropriate formula for variance at the higher level is
complex – which is why we use the bootstrap instead.
Another issue is determining appropriate degrees of freedom for estimating
confidence limits at the lower level. Currently, we divide the degrees of freedom
from the higher level according to the number of observations for each estimate
at the lower level. For example, consider a global detection function model that
has 96 degrees of freedom, used to estimate f(0) at the stratum level for two
strata, one with 76 and and the other with 24 observations. The degrees of
freedom for the stratum level estimates will be 96*(76/100)=72.96 and
96*(24/100)=23.04 respectively.

Cluster Size as a Covariate

When objects occur in clusters, the detection function often shows “size-bias” –
that is large clusters are more likely to be seen further from the line than small
clusters. In conventional distance sampling, there are several methods for

118 • Chapter 9 - Multiple Covariates Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 dealing with this (Buckland et al. 2001), including predicting expected cluster
size at zero distance using a regression of cluster size against distance (or against
probability of detection). The expected cluster size is then used when converting
the estimated density of clusters into density of individuals. In Distance, these
options are accessible in the Cluster Size tab of the Model Definition
Properties window.
In MCDS, there is another alternative: cluster size can be included in the
detection function model as a covariate. In this case, the size bias will be
allowed for in the detection function model. Density of individuals can be
obtained directly from the Horvitz-Thompson-like estimator used in the MCDS
engine (Marques and Buckland 2004), so the options in the Cluster Size tab
become obsolete (indeed, this tab is greyed out when you select cluster size as a
covariate).
When cluster size is a covariate, several options also change in the Model
Definition Properties window Estimate tab. The changes are outlined
below; see also the Estimate Tab - CDS and MCDS page of the Program
Reference.
Stratification and post-stratification
A restriction when you select cluster size as a covariate is that stratification and
post-stratification is no longer possible – see Stratification and Post-stratification
in MCDS for more on this).
Variance estimation
A second restriction when you select cluster size as a covariate is that the
variance of the estimate of density of individuals is not estimated analytically,
but is instead obtained using the bootstrap. We also use the bootstrap to obtain a
variance for the estimated expected cluster size. Analytic formulae for these
variances are given by Marques and Buckland (2004) but are not currently
implemented in the software.

Scaling of Distances for Adjustment Terms

In the key function + adjustment term formulation of Buckland et al. (2001,
Section 2.4), the detection function is defined using a parametric key function
which is then “adjusted” using a series expansion to give a flexible model. The
series expansion (or adjustment terms) are one of three forms:

Series expansion Form1

Cosine m
∑ a j cos( jπy s )
j =2

Simple polynomial m
∑ a j y s2 j
j =2

Hermite polynomial m
∑a j H 2 j (ys )
j =2

(1Note that when a uniform key function is used in CDS, the summation is from j=1 to m)

Here, ys is the perpendicular distance, standardized to avoid numerical problems.

In Buckland et al. (2001), the standardization ys=y/w is assumed, where w is the
right truncation distance. However, in the case of the multiple covariate engine,
this means that the adjustment of the detection function is independent of the
covariate values, so the shape of the detection function will change at different
values of the covariate. This may be appropriate in some cases, for example

User's Guide Distance 6.0 Beta 5 Chapter 9 - Multiple Covariates Distance Sampling Analysis • 119
 when search effort was conducted in such a way that the probability of detection
is consistently higher at a given distance than the model without adjustments
would predict. Nevertheless, to enable models with a truly constant shape to be
fit in the presence of adjustment terms, Distance also offers the ability to scale
the distance by σ, the scale parameter of the half-normal or hazard rate key
functions (see Defining MCDS Models on page 3 of this chapter for more on the
scale parameter). Since σ is a function of the covariates, this means that the
scaling will be different at each covariate level, and the shape of the function
will be preserved.
This option is set in the Model Definition, under the Detection function |
Adjustment terms tab. For more information, see the Program Reference
pages on the Model Definition Properties Dialog.

MCDS Analysis Guidelines

This section gives specific advice on doing analyses with the MCDS engine. For
more general guidelines for approaching analyses in Distance, see Chapter 7 -
Analysis in Distance. See also Marques and Buckland (2004) and Marques et al.
(2007).

Choosing Covariates to Include in the Model

When specifying one or more covariates to be included in the model for the
detection function, care must be taken to ensure that they do not violate any
inherent assumptions of the method. Of primary importance are the following:
Covariates should be independent of (perpendicular or radial)
distance.
A fundamental assumption of the theory behind the MCDS engine is that the
distribution of distances is independent of the distribution of the covariates. If
this assumption is violated, results from the MCDS engine are no longer valid or
reliable. Imagine you carried out line transect surveys while walking along a
path, and recorded vegetation type for each detected object as either low shrubs
or tall shrubs. Imagine also that the vegetation near or around the path consisted
primarily of low shrubs, whereas farther away from the path there were mostly
tall shrubs. Clearly in this example, vegetation type will depend on perpendicular
distance, as low shrubs will only occur at smaller distances, whereas taller
shrubs will only occur at greater distances. Hence the inclusion of vegetation
type as a covariate in this case will not be appropriate.
Avoid using covariates that are strongly correlated.
The use of highly correlated covariates may result in poor estimates of the
detection function, or highly correlated parameter estimates, or both. If you have
two covariates that are strongly correlated, and would like to include them in the
model, a better approach would be to include one covariate at a time, fitting the
model separately to each of them, and selecting the covariate which gives the
best model fit.
Covariates should only affect the scale parameter.
It is possible to have factor covariates with levels that exhibit different shapes for
the detection function – for example, the distribution of distances for one of the
levels may be almost uniform, whereas for the other(s) it may follow a half-
normal or hazard-rate shape. In such cases you will almost invariably end up
with a very poor model fit. The use of graphic tools during exploratory data
analysis (e.g. examination of histograms of the distances stratified by factor
levels), prior to the actual modeling of the data, is useful for identifying this type
of problem.

120 • Chapter 9 - Multiple Covariates Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 Specifying the Model
Fitting the detection function with multiple covariates is significantly harder,
computationally, than in the case where there is only one covariate. This has
several consequences:
• the analysis engine takes longer to run
• the algorithm will fail to converge more often
Because of this, it is important to be careful and thoughtful when setting up and
running MCDS analyses. Here are some recommendations:
• Rather than including numerous covariates at once, start by
including one covariate at a time, and selecting the covariate that
gives the best model fit/lowest AIC/AICc/BIC. You can then carry
out forward stepwise selection by adding one additional covariate
at a time, while retaining the one(s) already selected, until there is
no decrease in the AIC/AICc/BIC value (depending on the criteria
you are using).
• Bear in mind that factor covariates are usually harder to fit than
non-factor covariates, especially as the number of factor levels
increases (see Factor and Non-factor Covariates in MCDS earlier in
this Chapter for information on factor covariates). If you encounter
problems while trying to fit a factor covariate, try reducing
(condensing) its number of levels, if possible.
• Avoid using the feature that allows automatic selection of
adjustment terms – at least to start with. Instead, start by using a
model with no adjustments, and if this converges try one with one
adjustment. Gradually work up to more adjustments if required.

The default for the MCDS engine’s Model Definition Properties

is for no adjustment terms. This is different from the default for the CDS engine.

One advantage of automatic forward or sequential selection is that the

parameter estimates from the previous fit are used as starting values for the next
model. So, for example, the parameter estimates for the fit with 0 adjustment
terms would be used as starting values when trying 1 adjustment term. This
helps the algorithm to converge. So, it is sometimes helpful to use automated
selection, but to set the maximum number of adjustment terms to a low value,
such as 2, so that too many terms are not tried. Alternatively, you can set
starting values manually (see below).

• Use the option in Model Definition Properties, Misc., Report

results for each iteration of detection function fitting
routine. This will give you some clues about whether the
parameter estimates have converged.

This option is on by default in the MCDS engine.

• Another check for convergence is to compare the fitted likelihood

with that from a CDS analysis with the same key function and
adjustment terms. Assuming that the CDS analysis converged, the
MCDS analysis should always have a likelihood that is as high, or
higher. (This is because the CDS analysis contains a subset of the
covariates in the MCDS analysis, so it must fit the data as well or
worse.)

User's Guide Distance 6.0 Beta 5 Chapter 9 - Multiple Covariates Distance Sampling Analysis • 121
 • Convergence is often very sensitive to the starting values used.
You can set starting values manually using Model Definition
Properties, Detection Function, Adjustment Terms,
Manually select starting values.

Truncation for MCDS Analyses

Because MCDS methods are based on a Horvitz-Thompson like estimator of
abundance, in which the inclusion probabilities are estimated (Marques and
Buckland 2001, 2004), abundance estimates can be sensitive to errors in the
estimated probabilities. The sensitivity is greater for smaller estimated
probabilities. As a rough guide, we recommend that the method not be used if
more than say 5% of the estimated probabilities of detection of observed objects,
given that they were within the strip and with given covariate values, are less
than 0.2, or if any are less than 0.1. If these conditions are violated, the
truncation distance can be reduced. This causes some loss of precision relative to
standard distance sampling without covariates.

For MCDS analyses, a summary of the proportion of estimated

detection probabilities in bands 0.0-0.1 through to 0.9-1.0 is given at the bottom
of the Detection Fct | Parameter Estimates page of results.

The probabilities discussed here are a function of the observed

covariate values, but not of distance from the line or point; distance has been
integrated out, to give more robust estimation of abundance (Marques and
Buckland 2001, 2004).

Output from MCDS Analyses

The MCDS engine produces very similar output to the CDS engine (see the
section Output from CDS Analyses in Chapter 8). Differences are in the Results
Details listing, as outlined below.

MCDS Results Details Listing

MCDS output differs from CDS output only in the detection function pages
(those that start with Detection Fct). The following detection function pages
are output:
• Model fitting. Similar to CDS output, except that the results of
each iteration of the fitting engine are reported by default. The
model formulae and parameter indexing (A(1), A(2), etc) are also
slighly more complicated, as there are now covariate parameters,
but this should be self-explanatory. (For more about model
formulae and parameter indexing, see About CDS Detection
Function Formulae in Chapter 8.)

• Parameter estimates. Similar to CDS output, except

for the addition of a table summarizing the distribution of the
estimated probability of detection, given the covariate values. This
is useful for diagnosing possible problems with the estimates due to
low estimated detection probabilities – see Truncation for MCDS
Analyses in this Chapter.
• Plot: Qq-plot. Similar to CDS output. For more about CDS Qq-
plots, see Chapter 8.

122 • Chapter 9 - Multiple Covariates Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 • K-S GOF Test. Similar to CDS output. For more about these
CDS Goodness of Fit Tests, see Chapter 8.
• Plot: Detection probability. These plots can be used in the
same way as for the CDS engine – for comparing the estimated
function with the histograms of counts (or scaled counts for point
transects). However, because covariates affect the detection
function, there is no one single detection function to display.
Instead, the plotted functions are the average detection function,
conditional on the observed covariates. The histograms show
observed frequencies at given distances, pooled over all covariates.
For more information, see Chi-square GOF tests and related plots
in this Chapter.
• Plot: Pdf. (Point transects only) See above.
• Chi-sq GOF test. See above.
• FCx. When there are factor covariates, you get a set of diagnostic
plots for each factor combination, using only the data for that
combination. For example, if there are two factor covariates, one
with 3 levels and the other with 4, there will be 3x4=12 possible
factor combinations (although some may not occur in the dataset,
so will not be shown).
• Plot: Det Prb – the detection probability plot for the given
factor combination. If there are any non-factor covariates,
then the plotted function will be the average detection
probability, conditional on the observed non-factor covariates,
and the histograms will show observed frequencies pooled
over the non-factor covariates.
• Plot: Pdf. (Point transects only) See above.
• Plot: Examp Det Funcs. If there are any non-factor
covariates, then the above plots show the detection function (or
pdf) averaged over the observed covariate levels. Depending
on the observed covariates, the shape of these functions can be
quite different from the detection function given fixed
covariate values. So, for each non-factor covariate, the MCDS
engine outputs a plot showing 3 example detection functions.
These are evaluated at the 25th, 50th and 75th quartiles of the
covariate. If there is more than one non-factor covariate, the
values of the other covariates are fixed at the 50th percentile.

Currently the example detection functions are evaluated at the

observed quartiles of non-factor covariates. An alternative would be to estimate
the population quartiles, using a Horvitz-Thompson-like estimator. We may
implement this in a future version of Distance.

• Plot: Examp Det Funcs. If there are no factor-covariates, then

there are no factor combination plots to display. For each non-
factor covariate, the engine outputs a plot showing 3 example
detection functions, as described above.

For information about how to export the results text or plots into
another program, see Exporting MCDS Results.

User's Guide Distance 6.0 Beta 5 Chapter 9 - Multiple Covariates Distance Sampling Analysis • 123
 Chi-square GOF Tests and Related Plots
In the CDS engine, the detection function depends on distance alone, and this
function is displayed in the detection probability plots. By contrast, in the
MCDS engine, probability of detection depends on other covariates, so there are
many possible detection functions, depending on the covariate levels. Hence
detection probability plots show the average detection function, conditional on
the observed covariates. Similarly, for point transects, the probability density
function (pdf) shown is the average pdf conditional on the observed covariates.
For example, the average conditional pdf at distance j is calculated as:

f ( j, z i )
n
1
f ( j | z) =
n ∑ w
i =1
∫ f (x, z i )dx
x =0

where f (x, z ) is the joint density function, n is the number of observations, and
w is the truncation distance. (Assuming no left truncation, otherwise x=0 in the
integral is replaced by x=left truncation distance.)
Expected frequencies for the chi-square GOF test are calculated similarly. For
the jth bin, with cutpoints c j1 to c j 2 , expected frequency is:
c j2
n ∫ f (x, z i )dx
( ) ∑
Eˆ n j =
c j1
w
i =1
∫ f (x, z )dx
0
i

As with the CDS engine, the observed frequencies correspond to the total
number of observations which fall within each bin.

Exporting MCDS Results

The methods of exporting results from MCDS analyses to other programs are the
same as those for CDS analyses, as documented in Exporting CDS Results in
Chapter 8.

In Exporting CDS Results from Analysis Details Results we show how

to reproduce detection function, pdf and qq-plots in R. MCDS results contain
one addition type of plot: the example detection functions produced with non-
factor covariates. These can be reproduced using the following code:

#this reads in the file just created

forplot<-read.table(file="plot.txt", header=T, sep="\t", dec=".")
#note, depending on your language, dec might be "," rather than "."
#this plots the detection function or pdf (if point transects)
plot(forplot$C1, forplot$C2, type="l", ylim=c(0,1),
xlab="Distance", ylab="Detection probability")
#Define labels as you wish
#this adds in other two lines
lines(forplot$C3, forplot$C4, lty=2)
lines(forplot$C5, forplot$C6, lty=3)

Miscellaneous MCDS Analysis Topics

Missing Data in MCDS Analysis

For information about how to deal with missing distance and cluster size data,
see Missing Data in CDS Analysis in Chapter 8.

124 • Chapter 9 - Multiple Covariates Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
 If you have missing covariate data for some observations, then this can be treated
in a similar way to missing distances (see above).

If you run an analysis with data that includes missing covariates, the
MCDS engine will issue a warning and exclude each observation with a missing
covariate value.

Stratification and Post-stratification in MCDS

In most cases, the stratification and post-stratification options for MCDS are the
same as those for CDS analyses – see Stratification and Post-stratification in
Chapter 8 on CDS. An exception is when cluster size is a covariate in the
detection function (see Cluster Size as a Covariate). In this case, Distance
currently does not allow stratification or post-stratification. This is not an
inherent limitation of the theory, rather a limitation of the current analysis engine
software.

Running MCDS Analyses from Outside Distance

The MCDS engine is implemented as a stand-alone FORTRAN program,
MCDS.exe. This program is called behind the scenes by Distance when you
press the Run button on the Analysis Details Inputs tab.
Some users may wish to run the engine from outside the Distance interface –
either from the Windows command line or from another program. For example,
you may want to automate the running of analyses for simulations, or you may
want to perform a more complicated bootstrap than Distance allows.
Full documentation for running MCDS.exe is provided in the Appendix - MCDS
Engine Reference.

Analysis of Double Observer Data with the

MCDS Engine

Double observer data comes from surveys where two (semi-) independent
observer teams perform a distance sampling survey and duplicate detections are
identified. The method for setting up double observer data in Distance is
outlined in the topic Setting up a Project for MRDS Analysis in Chapter 10 of
the Users Guide.
There are two ways to achieve an MCDS (or CDS) analysis of double observer
data in Distance:
• Analyze the data using the MRDS engine, with Detection
Function | Method in the model definition set equal to ds –
single observer. Using this approach, duplicate observations are
automatically removed. For more information, see Single Observer
Configuration in the MRDS Engine in Chapter 10 of the Users
Guide.
• Analyze the data using the MCDS engine. Using this approach, it
is necessary to use the data filter to specify either observer 1 or
observer 2 data or both should be used.
Here, we focus on the second option.
To analyze double-observer data using the MCDS engine, you set up an analysis
where the analysis engine option in the associated model definition is MCDS.

User's Guide Distance 6.0 Beta 5 Chapter 9 - Multiple Covariates Distance Sampling Analysis • 125
 However, it is also necessary to set up a data filter specifically to achieve the
desired analysis. This is because double-observer data is entered into distance
with two records for each detected object in the observation layer. So, an
analysis that ignores this will have two records for each object, and so more data
than there should be. Double observer data has two fields that can be used in
conjunction with the data selection options in the Data Filter to achieve the
desired analysis. The two fields are observer (which is 1 for the first observer
and 2 for the second) and detected (which is 1 if the animal was detected by that
observer and 0 if not; note that the actual field names may be different from
this). For more details on these, see Setting up a Project for MRDS Analysis in
Chapter 10 of the Users Guide.
Three types of analysis can be envisaged:
12. An analysis of only objects detected by observer 1. To achieve
this, set up a Data Filter with data selection at the Observation
layer, and the data selection criterion
(observer=1) and (detected=1)
Note that “observer” and “detected” are the default field names for
these fields, but the actual field names in your project may be
different if you set up the double-observer project manually (rather
than via the Setup Project Wizard).
13. An analysis of only objects detected by observer 2. To achieve
this, you want the data selection criterion
(observer=2) and (detected=1)

14. An analysis of all objects detected, regardless of which observer

detected them. To achieve this, you want the data selection
criterion
(observer=1)
or you could equally use
(observer=2)
This assumes that the distance and other covariate data are the
same for both observers, and also that there are no records in the
dataset where detected=0 for both observer 1 and observer 2 (i.e.,
no-one saw it!). If both of these criteria are met, then choosing
either observer=1 or observer=2 will result in half of the data being
discarded, so that the number of observations going into the MCDS
engine will be the number of unique objects detected, which is
what you want.

126 • Chapter 9 - Multiple Covariates Distance Sampling Analysis User's Guide Distance 6.0 Beta 5
Chapter 10 - Mark Recapture
Distance Sampling

Introduction to Mark Recapture Distance Sampling

This entire chapter is for advanced users only!

Mark Recapture Distance Sampling (MRDS) refers to the analysis of double-
observer distance sampling data, as described by Laake and Borchers (2004).
Double-observer methods allow estimation of the probability of detection at zero
distance, g(0), in contrast to conventional methods, where this probability is
assumed to be 1.
The MRDS engine in Distance implements these methods, and this chapter
describes how to use the MRDS engine. We do not describe the methods at all –
instead we assume the reader is familiar with Laake and Borchers (2004). Please
note that double-observer methods are an advanced technique that should only
be considered once you are familiar with conventional distance sampling.
The MRDS engine currently has the following capabilities. We expect to extend
these in future versions.
• Estimation for independent observer and trial configurations (see
Introduction to MRDS Models; removal configuration to follow in
a future version)
• Point and full independence models (in the forms of Models 1 and
3 from Table 6.1 of Laake and Borchers 2004)
• Estimation for single observer CDS and MCDS surveys (see Single
Observer Configuration in the MRDS Engine; support for mixed
single and double platform surveys may follow).
• Line transect surveys (point transects to follow)
• Perpendicular distance (radial distance and angle to follow)
• Data where the object of interest is an individual or cluster (see
Clusters of Objects in MRDS)
• Right truncation of distances, and left truncation for single observer
configuration (left truncation for double observer configuration to
follow)
• Exact and interval data (manual specification of cutpoints only;
more complex interval data, where intervals vary by observation to

User's Guide Distance 6.0 Beta 5 Chapter 10 - Mark Recapture Distance Sampling • 127
 follow, and even more complex non-continuous intervals are
planned). Note that you can’t do left or right truncation when the
data are in intervals.
• Up to one level of geographic stratification (see Stratification and
Post-stratification in MRDS; extension to multiple levels of
stratification and other types of stratification may follow)
• Three analytic methods of variance estimation (see Variance
Estimation in MRDS; bootstrap may follow)
• Estimation of density/abundance using a detection function fitted in
a previous analysis – allows different subsets of the data to be used
for encounter rate and detection function fitting (see Using a
Previously Fitted Detection Function to Estimate Density in
MRDS).
The MRDS engine is implemented as a library in the free statistical software R.
When you run an MRDS analysis from Distance, Distance creates a sequence of
R commands, calls the R software, waits for the results and then reads them back
in. Therefore, before you can use the MRDS engine, you must first ensure that
you have R correctly installed and configured. For more on this, see R
Statistical Software in Chapter 7 of the Users Guide.
To analyze double-observer data in Distance, you then need to set up the project
appropriately and include data in the correct format – see Setting up a Project for
MRDS Analysis. You must next create one or more model definitions that
specifies to use the MRDS analysis engine, and associate these model definitions
with analyses, which you can then run. For more about the basics of setting up
analyses, see Chapter 7 - Analysis in Distance. More details of the various
models available in the MRDS engine are given in Defining MRDS Models, and
a detailed description of the options available in the Model Definition Properties
pages for this engine is given in the Program Reference pages Model Definition
Properties - MRDS.

If you are new to Distance, we strongly recommend you familiarize

yourself with the CDS analysis engine, for example by working through Chapter
3 - Getting Started before trying to analyze MRDS data.

The Golftees sample project is an example of how to set up a double-

observer study and specify analyses – see Sample Projects.

In this chapter we also provide some analysis guidelines, give a list of the output
the engine can produce and cover various miscellaneous topics such as how to
deal with interval data, stratification, etc.

If you are familiar with the R software, you can run the MRDS
engine directly from within R, bypassing the Distance interface altogether. For
more information, see Running the MRDS Analysis Engine from Outside
Distance.

There is an extensive online help that accompanies the MRDS R

library. This contains more details about the methods and options available, and
will be of use to users familiar with R. To open these help pages from within
Distance, choose Help | Online manuals | MRDS Engine R Help (html).
The main functions to look at are ddf() and dht().

128 • Chapter 10 - Mark Recapture Distance Sampling User's Guide Distance 6.0 Beta 5
Setting up a Project for MRDS Analysis
The easiest way to set up a new project for an MRDS analysis is using the Setup
Project Wizard.
• In Step 1, under I want to:, select Analyze a survey that has
been completed.
• In Step 3, under Observer configuration, select Double
observer.
• Follow through the rest of the wizard as usual.
Distance then creates the appropriate data fields for double observer data, and
you can then import your data using the Import Data Wizard. For more about
the data requirements, see Setting up your Data for MRDS Analysis.
Alternatively, you can create the appropriate fields by hand, and manually create
a new survey object with the appropriate observer configuration and data files.
For more about survey objects, see Working with Surveys During Analysis in
Chapter 7.

Setting up Your Data for MRDS Analysis

When you use the select a double observer configuration in the Setup Project
Wizard, Distance creates three extra data fields in the Observation layer (in
addition to the appropriate distance and cluster size fields). These are:
• object – this must contain a unique integer for each object
detected by either observer. There should be two records in the
observation layer for each object seen – one for observer 1 and the
other for observer 2.
• observer – tells Distance which observer the record refers to:
either 1 or 2.
• detected – tells Distance whether the object was detected by that
observer or not: 1 for detected, 0 for not detected.
The fields do not have to have those names – if you set up the project manually,
you can call them what you like. However, in the Data Fields tab of the survey
object for double observer configurations, you have to tell Distance which fields
play the object, observer and detected roles.
An example of double observer data, from the Golftees sample project, is shown
below:

Part of the Observation layer from the Golftees double observer example project

If you open the Golftees sample project you will notice that:
• The object number goes up sequentially in this example (1, 2, 3,
…) – in general the object number should be unique but it doesn’t
need to be sequential.

User's Guide Distance 6.0 Beta 5 Chapter 10 - Mark Recapture Distance Sampling • 129
 • In this example the two records for each object come one after the
other (e.g., lines 1 and 2 are Observer 1 and Observer 2 for object
1) – in general they don’t have to so long as there are two lines for
each object – one with Observer 1 and one with Observer 2. For
example, you might like to structure your data with all records for
observer 1 first and then all records for observer 2
• The detected field indicates whether an observer saw the object or
not. For example, object 1 was seen by observer 1 but not by
observer 2.
• In this example, the distance field (Perp distance) contains the same
distance for both observers, regardless of whether the observer saw
the object or not. In general you should always put the same
distance for both observers – a version that can deal with
measurement error and so allow different distances is planned.
• There are some additional covariates in the observation layer in this
example: sex, exposure and Cluster size. In general, covariates can
be placed in any of the data layers – although the rules for referring
to the covariates differ between the layers – for more on this, see
Defining MRDS Models.
• In this example, there is only one transect, and so there are no
transects on which no objects were seen. In general there may well
be transects with no objects. On these transects you should not
enter any observations (just as with the CDS engine – see the
Example Data Sheet picture on the Data Fields page in Chapter 5
for an example, and see Introduction to Data Import for how to
import such data).

Defining MRDS Models

Introduction to MRDS Models

Specifying a detection function in MRDS requires specifying the form of up to
two functions (Laake and Borchers, 2004, section 6.3.2.3). The first is the
unconditional detection function g . ( y, z ) – the probability of one or more
observer detecting the object, given it’s distance and covariate values. The
second is the conditional detection function p j|3− j ( y , z ) – the probability of
observer j detecting the object, given that the other observer (observer 3-j) has
detected it and also given its distance and covariate values. We refer to the
former as the distance sampling or DS model and the second as the mark
recapture or MR model. Which of these needs to be specified depends on the
fitting method chosen, as follows:
Fitting Configuration Independence DS model MR model
method
ds single1 - yes no
io independent point yes yes
io.fi independent full no yes
trial trial point yes yes
trial.fi trial full no yes
2
removal removal point yes yes
2
removal.fi removal full no yes

130 • Chapter 10 - Mark Recapture Distance Sampling User's Guide Distance 6.0 Beta 5
 1
For more on this, see the topic Single Observer Configuration in the MRDS Engine later
in this chapter.
2
These methods are not implemented in the current version of Distance

The fitting method is chosen on the Detection Function | Method page of the
Model Definition Properties.

DS and MR Models
The form of the two DS and MR models are different. The implementation here
corresponds with models 1 and 3 of Table 6.1 from Laake and Borchers (2004) –
see also that reference for more information.
DS Model
The DS model is of the same form as the models used in the MCDS engine,
except that currently only the key function part of the model is implemented,
with no adjustment terms. The two key functions implemented are half-normal
and hazard rate:

{
Half-normal key function, exp − x 2 2σ (z )2 }
Hazard-rate key function, 1 − exp − [x σ (z )] { −b
}
The scale parameter, σ (z ) , is modeled as an exponential function of the
covariates:
σ ( z ) = exp (β 0 + β 1 z 1 + β 2 z 2 + K β q z q )
The key function and covariates to use are specified on the Detection
Function | DS Model page of the Model Definition Properties. One can also
choose no additional covariates, which corresponds with a CDS model. Note
that distance (x in the above formulae) is automatically a part of this model.

If you’re comparing DS model results with those obtained from the

MCDS engine, you should note that in the MCDS engine, the scale parameter is
( )
modeled as σ (z ) = β 0 + exp β1 z1 + β 2 z 2 + Kβ q z q . Therefore to compare the
two values of β 0 , you need to exponentiate the β 0 from the MRDS Engine’s
DS model.

MR Model
The MR model is currently implemented as a logistic model – i.e. it is of the
form:

p j|3− j ( y , z ) =
(
exp β 0 + β1 z1 + β 2 z 2 + Kβ q z q )
(
1 + exp β 0 + β1 z1 + β 2 z 2 + Kβ q z q )
The covariates to use are specified on the Detection Function | MR Model
page of the Model Definition Properties. Note that these covariates need not be
the same as those chosen for the DS model, if there is a DS model with the
chosen fitting method. Note also that unlike for DS models, distance is not
automatically a part of the model – if you want to include distance as a covariate
in an MR model you must explicitly include it in the model formula.

Specifying DS and MR Model Formulae

For both DS and MR models, one must specify formulae that tell Distance which
covariates to include in the model (the general form of the two models are given
in the previous section, DS and MR Models). These formulae consist of a series

User's Guide Distance 6.0 Beta 5 Chapter 10 - Mark Recapture Distance Sampling • 131
 of terms joined by operators such as “+”. The terms represent covariates and the
operators tell Distance how the covariates relate to one another.
For example, the MR formula
distance + sex + exposure

means include the data from the fields distance, sex and exposure as covariates.
To understand how to specify formulae, we need to understand (1) how to
translate a field in the Distance database into a covariate to specify and (2) what
are the possible operators and how do they work. We also need to understand
the difference between factor and non-factor covariates and how to specify
which is which. These are covered in the following sections.

To see some examples of model formulae, have a look at the Golftees

example project.

Translating Distance Fields into DS and MR Covariates

Unfortunately, there is not a 1:1 correspondence between the field names in the
Distance project database and the covariate names you can use to specify DS and
MR model formulae. This is due to some limitations of the R statistical
language that the MRDS engine is implemented in, and also in the way the
MRDS engine is written.
Field Translation Made Simple
The easiest way to work out how to write your formulae is to run a simple
MRDS analysis in Distance, perhaps with a trial.fi fitting method and “1” as the
MR model (i.e., and intercept-only model). Then, look in the log tab for
something like the following, which lists the field name and translated covariate
name for all fields in the database:
The following fields will be written to the detection function data
file, and can be used in detection function model formulae. Note
that you should use the new names, not the original field names in
formulae, and that formulae names are case sensitive.
Format: [layer name].[field name] AS new name
[Observation].[Perp distance] AS distance
[Observation].[Cluster size] AS size
[Observation].[object] AS object
[Observation].[observer] AS observer
[Observation].[detected] AS detected
[Observation].[sex] AS sex
[Observation].[exposure] AS exposure
[Line transect].[Label] AS label
[Line transect].[Line length] AS line.length
[Region].[Label] AS stratum.label
[Region].[Area] AS area
[Study area].[Label] AS global.label

For example, if you then wanted to specify a formula with the label field from
the region layer (i.e., [Region].[Label]) and observer from the Observation layer
(i.e., [Observation].[observer]) as covariates, you would write the formula as:
stratum.label + observer

Field Translation in Detail

If you want to try to predict what translations Distance will do, here are the rules
it applies:
• You can specify covariates using fields in any data layer. However, if
the field name occurs in more than one data layer, and so is not unique,
then Distance has to use some method to determine which layer you are
referring to. So it does the following. It starts at the lowest layer and
stores the field names. If it encounters the same name in a higher layer

132 • Chapter 10 - Mark Recapture Distance Sampling User's Guide Distance 6.0 Beta 5
 it renames it by adding the layer type and a dot in front of the field
name. For example, by default, there is a field called “Label” in the
sample, stratum and global layers. The sample layer is the lowest, so
the formula name for this is label (note it is lower case as all formula
names are changed to lower case – see below). The stratum and global
layers are higher, so the formula names for the Label fields in those
layers are stratum.label and global.label.
• To comply with R object naming rules, and make life simpler:
• spaces and anything else that isn’t a letter or number are
replaced by dots – e.g., “type of habitat” becomes
type.of.habitat.

• all letters become lower case – e.g., “Type of Habitat”

becomes type.of.habitat.
• fields that don’t start with a letter get an “X” appended - e.g.,
“1 covariate” becomes X1.covariate.
• Fields with certain roles are always translated to a fixed name, as
follows.
Role Translated to
Perpendicular distance distance
Cluster size size
Object object
Observer observer
Detected detected

As an example, if you have a field “Cluster size” which is defined as

the having the role of cluster size in the survey definition, then the name
you should specify in MR or DS formulae is size.

DS and MR Model Formulae Operators

• + means the two covariates on either side of the operator are both
in the model as main effects – e.g., a + b means include terms for
covariate a and covariate b
• : means the interaction of the two covariates on either side of the
operator – e.g., a:b means the interaction of a and b
• * means the two main effects plus all interactions – e.g., a*b is the
same as a + b + a:b. More than two covariates can be included –
e.g., a*b*c is the same as a + b + c + a:b + a:c + b:c + a:b:c
• ^ indicates crossing to a specified degree – e.g., (a+b+c)^2 is the
same as (a+b+c)*(a+b+c) which in turn expands to a formula
containing the main effects for a, b and c together with their
second-order interactions
• %in% indicates that the terms on its left are nested within those on
the right. For example a + b %in% a expands to the formula a + a:b
• - removes the specified terms, so that (a+b+c)^2 -a:b is identical to
a + b + c + b:c + a:c. It can also used to remove the intercept term:
x - 1 is a line through the origin. (A model with no intercept can be
also specified as y ~ x + 0 or 0 + x)

User's Guide Distance 6.0 Beta 5 Chapter 10 - Mark Recapture Distance Sampling • 133
 More About DS and MR Model Formulae
• An intercept term is included in formulae by default. To remove it
you can use “- 1” - for example sex - 1, while to specify an
intercept-only formula, you use 1 alone.

As an example of the use of the use of an intercept-only formula,

one way to specify a Peterson model is using a trial configuration where there is
full independence and an intercept-only MR model.

• While formulae usually involve just variable and factor names,

they can also involve arithmetic expressions. The formula sex +
log(size) is quite legal. Take care if you use arithmetic
expressions that there isn’t a field with the same name (e.g., a field
in the observation layer called “log”)!

Factor and Non-factor Covariates in MRDS

For an explanation of the difference between factor and non-factor covariates,
see Factor and Non-factor Covariates in MCDS. In the MRDS engine, all
covariates are assumed to be non-factor, unless specified otherwise. To specify a
covariate as a factor, include it as part of a comma-delimited list in the
Detection Function | Factors page of the Model Definition.
For example, if the MR model is given as distance * sex * exposure, and the
Factors page specifies the following: sex, exposure, then the MRDS engine
will assume that distance is a non-factor covariate and that sex and exposure are
factors.

You can specify covariates as factors even if they are not included in a
model in the current model definition. It saves time to list all the factor
covariates in your first model definition, as this list will then be copied to all
subsequent model definitions that you define, saving you the bother of having to
type the factor list for each model definition. In the above example, we could
have specified observer, sex, exposure as the factor list – observer is not in
the current model, but if we subsequently define a model definition based on this
one and include observer as a covariate we won’t have to remember to include it
in the list of factors as it will already be there.

MRDS Analysis Guidelines

These methods are relatively new, so we are only beginning to gain experience
on effective analysis strategies. Some preliminary guidelines follow – please let
us know if you can suggest some further guidelines or amendements.
• Start with some CDS and MCDS analyses (i.e. using the CDS/MCDS
engines) to get a ball park on the detection function shapes, etc. – see
Analysis of Double Observer Data with the MCDS Engine in Chapter 9
for some tips on doing this. You can also perform CDS and MCDS
analyses using the MRDS engine – see Single Observer Configuration
in the MRDS Engine later in this chapter.
• Your first MRDS analysis could be a Peterson or other simple model –
helps work out what covariate names to use (see Translating Distance

134 • Chapter 10 - Mark Recapture Distance Sampling User's Guide Distance 6.0 Beta 5
 fields into DS and MR covariates), and should also fit without
problems.
• Build up covariates slowly. You may need to specify starting values
(although this option isn’t available currently in the interface) look at
the iteration history (Detection Function | Control page of Model
Definition), etc. to work around any convergence problems.
• If you experience problems, check Problems with the MRDS Engine in
the Troubleshooting chapter, and also check the Program Distance Web
Site for the latest list of known problems.
• This is a new analysis engine – you can expect some teething problems.
Contact the program authors if you can’t resolve them (see Sending
Suggestions and Reporting Problems).

Output from MRDS Analyses

The MRDS engine produces the following output:
• a summary of results in the Analysis Browser. For general
information about the Analysis Browser, see the section
Introduction to the Analysis Browser in Chapter 7. There are many
results statistics available, and you can select which ones are shown
independently for each analysis set using the Column Manager (see
Column Manager Dialog in the Program Reference). An
explanation of some of the columns is given in the section MRDS
Analysis Browser Results.
• a detailed listing of results in the Results tab of the Analysis
Details window. These are described in the following section,
MRDS Results Details Listing.
• a log of the analysis, highlighting any possible problems, in the
Log tab of the Analysis Details window. For information about
troubleshooting problems, see Chapter 12 - Troubleshooting.
• (optionally) plots from the results details can be imported into other
programs. For more about this, see the section on Exporting
MRDS Results.

MRDS Results Details Listing

When an analysis has run, a great deal of information is available in the Results
tab of the Analysis Details window. This information is split into pages, as
follows:
• Detection Fct/Summary. Gives a summary of the results of the
detection function modelling.

You can get more information about the fitting process using the
showit control setting – for details, see the section on Fine-tuning an MRDS
Analysis.

• Detection Fct. A set of pages about the detection function model

fited.
• Plot: Qq-plot. A quantile-quantile plot, showing the
goodness-of-fit of the fitted model (exact data only). For
information about what qq-plots are, see CDS Qq-plots in
Chapter 8.

User's Guide Distance 6.0 Beta 5 Chapter 10 - Mark Recapture Distance Sampling • 135
 • Goodness-of-fit. Chi-squared goodness of fit tests for the
DS and/or MR models (depending on the fitting method), and
Kolmogorov-Smirnov and Cramér-von Mises tests (for exact
data). For more about these latter tests, see CDS Goodness of
Fit Tests in Chapter 8.
• Plot: Detection Probability. Plots of the fitted detection
functions, superimposed on histograms showing the frequency
of counts. The estimated probability of detection of each
observation (given its covariate values and distance) is also
shown. The number of plots depends on the fitting method.

The plots are stored as image files in the R directory, and so can easily
be imported into other programs - see Exporting MRDS Results.

• Density Estimates and associated quantities.

• Standard output consists of 3 tables: (i) a summary containing
area, covered area, survey effort, number of observations and
encounter rate; (ii) abundance and associated CV and
confidence limits; (iii) density and associated CV and
confidence limits. All are reported by stratum and pooled
across strata. If objects are in clusters there are 3 tables for
estimates by cluster and 3 more for estimates by individual,
followed by a further table of estimated expected cluster sizes.
• Further output is available by choosing the Extended output
option in the Misc tab of the Model Definition (see Misc -
MRDS in the Program Reference). This consists of: (i)
variance-covariance matrix for the abundance estimates due to
estimating the detection function parameters; (ii) variance-
covariance matrix for the abundance estimates due to selection
of samples; (iii) resulting correlation matrix of the
abundance/density estimates; (iv) estimates by sample of
sample effort, covered area, number of observations, estimated
abundance and density.

MRDS Analysis Browser Results

Currently, a relatively limited number of statistics are available in the Analysis
Browser: number of parameters, AIC, log likelihood, density and abundance
with associated CVs and confidence limits. We will be expanding the list of
outputs in future versions.

Exporting MRDS Results

The methods of exporting results from the Analysis Browser and results pages of
the Analysis Details to other programs are the same as those for CDS analyses,
as documented in Exporting CDS Results in Chapter 8. For example you can
copy the results details text by choosing Analysis Results | Copy Results
to Clipboard, and you can copy plots by choosing Analysis Results | Copy
Plot to Clipboard. Note though that in the case of the plot, the underlying
data are not copied as well (unlike for CDS plots) – just the plot picture.
There is another way to get hold of the plots produced by MRDS analyses: to
directly access the image files produced when each analysis is run. Each plot
that is displayed in the Results Details has a corresponding image in the project’s
R Folder. For more details, see Images Produced by R in Chapter 7 of the Users
Guide.

136 • Chapter 10 - Mark Recapture Distance Sampling User's Guide Distance 6.0 Beta 5
Miscellaneous MRDS Analysis Topics

Interval Data in MRDS

Not implemented in the current version of this engine.

Clusters of Objects in MRDS

If the objects detected are clusters of individuals, you tell Distance this in the
same way as for CDS analyses, in the Setup Project Wizard (see Survey
Methods Wizard Page in the Program Reference).
Unlike CDS analysis, the Model Definition does not offer any regression
methods for dealing with size bias. If you suspect size bias is a potential
problem, the appropriate way to deal with it in an MRDS analysis is to include
cluster size (or some transformation of cluster size) as a covariate in the
detection function model(s).

The cluster size field is one of the fields with a fixed name in
detection function formulae in MRDS (see Translating Distance Fields into DS
and MR Covariates) – in formulae you should use the name size regardless of
the actual field name.

Stratification and Post-stratification in MRDS

At the moment, the MRDS engine only accommodates one level of stratification,
and this stratification is assumed to be geographic (i.e., the global density
estimate is calculated as the mean of the stratum estimates, weighted by stratum
area). More options are planned for future versions.

Variance Estimation in MRDS

The MRDS engine currently offers three analytic methods for estimating
variance of the density estimate. These are:
• Innes et al. (2002) - based on the empirical variance in
estimated density between samples. This is the default method,
and the one that should generally be used. The formula has terms for
variation in density between samples given the estimated detection
function parameters, and for uncertainty due to estimating the detection
function parameters, as follows (from Marques and Buckland et al.
2004, formula 3.27):
⎧ ⎛ Nˆ csk Nˆ cs ⎞
2 ⎫
⎪ l ⎜ − ⎟ ⎪
K k⎜ l L ⎟⎠
⎛ A ⎞ ⎪⎪
() ⎪⎪
2
∂Nˆ cs ∂Nˆ cs −1
( )
r r
⎝
∑ ∑∑
k
vâr Nˆ s =⎜ ⎟ ⎨L + H jm θˆ ⎬
K −1 ˆ ˆ
⎝ 2 wL ⎠ ⎪ k =1 j =1 m =1 ∂θ j ∂θ m ⎪
⎪ ⎪
⎪⎩ ⎪⎭

where N̂ s is the estimated abundance of clusters (or individuals if

cluster size is always 1) in the study region (of size A), N̂ cs is the
estimated abundance of clusters in the covered region (of size 2wL,
where w is truncation distance and L is the total line length), N̂ csk is the
estimated abundance of clusters on the kth line (k=1…K), lk is the

User's Guide Distance 6.0 Beta 5 Chapter 10 - Mark Recapture Distance Sampling • 137
 length of the kth line, θˆ is a vector of length r containing the parameter
estimates of the detection function model, and H −1 θˆ is the jmth jm ()
element of the inverse of the Hessian matrix for θˆ . (See also formula
3.35 for the variance of the number of individuals.) More details are
given in Innes et al. (2002) and Marques and Buckland (2004).
• Buckland et al. (2001) - based on the delta method, using the
empirical variance in encounter rate between samples. This
method is based on the conventional distance sampling variance
estimator of Buckland et al. (2001, Sections 3.6.1 and 3.6.2), extended
to allow probability of detection to vary among individuals. The
method assumes independence between the estimates of detection
function parameters, encounter rate (and mean cluster size for variance
of the estimate of abundance of individuals) – an assumption not made
by the Innes et al. estimator. For that reason, the Innes et al. estimator
is preferred. The formula can be written:
⎧ ⎛n n ⎞
2 ⎫
⎪ K k⎜
l ⎜ sk − s ⎟⎟ ⎪
()
2
( ) ⎛ A ⎞ ⎪ Nˆ cs L ∂Nˆ cs ∂Nˆ cs −1 ˆ ⎪
r r
l L
vâr Nˆ s =⎜ ⎟ ⎨
⎝ 2 wL ⎠ ⎪ (n s w )2
∑k =1
⎝ k
K −1
⎠ +
∑∑ ˆ
j =1 m =1 ∂θ j ∂θ m
ˆ
H jm θ ⎬
⎪
⎪ ⎪
⎩ ⎭

where ns is the number of clusters seen, and nsk is the number of clusters
seen on line k.
• Binomial variance of detection process. This variance estimator
should only be used when the entire study area is sampled (as happens
sometimes, for example in simulation experiments). It only contains
the term for uncertainty due to estimating the detection function
parameters – i.e., it assumes no variance comes from scaling up the
estimated density on the surveyed area to the whole study area. The
formula is:
2 ⎧ ns
∂Nˆ cs ∂Nˆ cs −1 ˆ ⎫⎪
( ) ()
r r
⎛ A ⎞ ⎪
vâr Nˆ s = ⎜ ⎟
⎝ 2 wL ⎠
⎨ ∑
⎪⎩ i =1
fˆ (0 | z i )2 − Nˆ cs + ∑∑ ∂θˆ
j =1 m =1
ˆ
j ∂θ m
H jm θ ⎬
⎪⎭

where fˆ (0 | z i ) is the estimated pdf of observed distances given the

covariate values zi, evaluated at zero distance.
The appropriate option can be selected on the Variance tab of the Model
Definition Properties Dialog (see Variance - MRDS in the Program Reference).

Multipliers in MRDS Analysis

Multipliers are currently not implemented in the MRDS engine. Two main uses
for multipliers are dealt with using other methods in this engine – the case when
g(0)<1 is dealt with via the double-observer data, and the case where we wish to
fit a detection function to some data and apply it to a subset is dealt with
explicitly in the section Using a Previously Fitted Detection Function to Estimate
Density in MRDS.

Model Averaging in MRDS Analysis

This is currently not implemented in the MRDS engine. For more on model
averaging, see Model Averaging in CDS Analysis in Chapter 8 of the Users
Guide.

138 • Chapter 10 - Mark Recapture Distance Sampling User's Guide Distance 6.0 Beta 5
 Sample Definition in MRDS Analysis
This is implemented in exactly the same way as for CDS and MCDS analyses -
see Sample Definition in CDS Analysis in Chapter 8 of the Users Guide for
details.

Using a Previously Fitted Detection Function to

Estimate Density in MRDS
Using the MRDS engine, you can fit the detection function to your survey data,
and use this fitted detection function to estimate detection probability for subset
of the data. This can then be combined with encounter rate information for that
subset to estimate density. There are several reasons why you might want to do
this. For example:
• In a multi-species study, you may want to explore fitting a
combined detection function to a group of species. You can use
model selection criteria such as AIC to decide whether to fit
individual models by species, or a combined model with or without
species (or guild) level covariates. Once the model selection is
done, you are usually interested in obtaining estimates of density
by species, so you would apply the global density function (if that
is what was chosen) to subsets of the data for each species.
• You may have two surveys of the same species at different times
but over the same range of conditions, where one has too few
sightings to estimate the detection function but the other has
enough. You can fit the detection function to the larger dataset (or
a combined dataset) and then, given the covariate values from the
smaller one, predict detection probabilities in the smaller dataset.
• You may want to fit a detection function to a relatively common
species, and then apply this function to a species for which there
are only a few observations (assuming you collected the same
covariate information for both species).
• Rather than species, other components of the population could be
the target (e.g., sex, etc).
How is this kind of analysis done in the MRDS engine?
You start with the the data for which you want to model the detection function –
you can select this data from the whole dataset using the Data Selection tab of
the Data Filter. Then define the model definitions you want and run the
corresponding analyses. Decide which analysis you want to use to provide a
detection function for the new subset of data.

For these model definitions, you probably don’t want to estimate

density, just the detection function. You can achieve this by un-checking the
option in the Estimate tab of the Model Definition to Estimate density /
abundance.

Recall that the MRDS engine is actually implemented as a language in the

statistical software R, and that to run an MRDS analysis, Distance writes out
command and data files, then runs R in batch mode, waits for it to finish and
then harvests the results from output files. By default, the R objects generated
by the run are not saved by R, but are deleted. In this case, we want to overwrite
that behaviour and save the fitted detection function object so that it can be used
in a future analysis. We do this by selecting Tools | Preferences | Analysis
and then under R software, uncheck the option Remove the new objects that
are created with each run. Now run the analysis that you are going to use to

User's Guide Distance 6.0 Beta 5 Chapter 10 - Mark Recapture Distance Sampling • 139
 provide the detection function fit again, and the R objects will be saved (in the R
object file .RData in the R Folder – see Contents of the R Folder in Chapter 7 for
more on this file).
Now we’ll apply this detection function to the new subset of data. Define a new
data filter that selects the new subset of data. Note that it must be a subset of the
data used to fit the detection function. Define a new Model Definition, and in
the Estimate tab, check the option to Estimate density / abundance and
also the option to Use the fitted detection function from previous
MRDS analysis. Select the ID of the analysis you want to use. For example,
if you want to use analysis 2 (called “FI - MR dist” in this case), the lower half
of the Estimate tab will look like this:

Applying the fitted detection function from one analysis to another analysis

When you run the new analysis, the probability of detection for each object in
the new analysis is estimated using the fitted detection function from analysis 2
(in this case).

Restricting Inference to Density or Abundance in

the Covered Region in MRDS Analysis
For more on the circumstances when this may be appropriate, see Restricting
Inference to Density or Abundance in the Covered Region in CDS Analysis in
the CDS Chapter. It is achieved in MRDS by selecting the Binomial variance
of detection process option in the variance tab – see the section on Variance
Estimation in MRDS earlier in this chapter.

Running the MRDS Analysis Engine from

Outside Distance
The MRDS engine is implemented as a library in the free statistical software R.
When you run an MRDS analysis from Distance, Distance creates files
containing a set of R commands and the appropriate input data, calls the R
software, waits for the results and then reads them back in. For more about R,
see R Statistical Software in Chapter 7 of the Users Guide.
Some users may wish to run the engine from outside the Distance interface –
either from within the R GUI interface or from another program. For example,
you may want to automate the running of analyses for simulations or
bootstrapping.
To see the format of the input and data files produced by Distance, try running
an MRDS analysis in debug mode. To set debug mode on, choose Tools |
Preferences, Analysis tab, and tick Debug Mode. When you run analyses
in debug mode, the input and data files are created, but the analysis is not run.
The Log tab displays the location of the files – these are created in a directory
with a name “dst” followed by up to 4 numbers, located within the Windows
temporary directory. The file “in.r” contains the commands. Data are located in
the files “ddf.dat.r” (used in the detection function modelling), “region.dat.r”,
“sample.dat.r” and “obs.dat.r” (used in estimating density given a fitted detection

140 • Chapter 10 - Mark Recapture Distance Sampling User's Guide Distance 6.0 Beta 5
 function). You can use these files as templates for creating your own command
and data files.
To run the analysis from within the R GUI (Graphical User Interface), you can
cut and paste the commands from the file in.r. To run the analysis from another
program, you can call R in batch mode – this is achieved by calling the program
RCmd.exe, which is located within the /bin subdirectory of your R installation.
For more details, see the R for Windows FAQ (in R, type help.start() and
when a browser window opens, click on the FAQ for Windows port). For an
example of its use, see the Log tab of any MRDS analysis you have run that was
not in debug mode – you should see a line of the form:
Starting engine with the following command:
C:\PROGRA~1\R\rw1091\bin\Rcmd.exe BATCH C:\temp\dst90474\in.r
C:\temp\dst90474\log.r

Users familiar with R may wish to work inside the R GUI. The MRDS engine is
contained in the library mrds. To load the library from within R GUI, type
library (mrds)

All the functions in the mrds library are documented – the main functions are
ddf() (fits the detection function) and dht() (estimates abundance using the
Hortvitz-Thompson-like estimator). You can open a copy of the help files from
within Distance by choosing Help | Online Manuals | MRDS R Engine
Help (html).

The use of these libraries in operating systems other than Windows

is not supported (but may well work – let us know!)

Installing an Updated Version of the MRDS

Engine
The MRDS engine is implemented as a library (called “mrds”) in the statistical
software R. From time to time, we may issue updated versions of the library, for
example in response to reported problems. These will come as an archive file
mrds.zip. To install the new version,
• find the old version of mrds.zip in the Distance program directory.
• copy the new mrds.zip over the top of it (you may want to rename
the old version first, as a backup)
• Choose Tools | Preferences | Analysis, and tick the option to
Re-install analysis engine library on next run.
• Open a project containing analyses that use the MRDS engine (for
example the Golftees sample project).
• Run one of these analyses.
• Check in the Log tab of the Analysis Details. You should find the
line:
package 'mrds' successfully unpacked and MD5 sums
checked

The next topic describes how to check which version of the library is being used.

Checking Which Version of the MRDS Engine is

Being Used
The MRDS engine is implemented as a library (called “mrds”) in the statistical
software R. From time to time, we may issue updated versions of the library, for
example in response to reported problems. Therefore, before downloading a

User's Guide Distance 6.0 Beta 5 Chapter 10 - Mark Recapture Distance Sampling • 141
 new version or reporting a problem you may want to check which version of the
library is currently in use. To do this, re-run an analysis that uses the MRDS
Engine (such as one from the Golftees sample project) and look in the Log tab
for the line
> library(mrds)

After it, you should see a line which looks something like the following:
This is mrds 1.2.7
Built: R 2.3.1; i386-pc-mingw32; 2006-08-09 17:33:03; windows

If you are reporting a problem you should quote both the build number (in the
above case 1.2.7) and the build date and time (2006-08-09 17:33:03).
The previous topic describes how to update to a newer version of the MRDS
Engine, if one is available.

When reporting results, you may want to cite the exact version (i.e.,
build number) of the library that used in the analysis. This is stored in the Log
tab, as outlined above.

Fine-tuning an MRDS Analysis

In most cases, the default options for estimation of detection function parameters
in an MRDS analysis work well. However, there are times when you need to
tweak the analysis, for example by setting starting values or bounds on
parameters. You may also want to get more information on the fitting process,
such as parameter estimates at each iteration of the optimization. This kind of
fine-tuning is specified in the Detection function | Control page of the
model definition.
To enter options on this page, type them in as a comma-delimited list. Any
noninteger numbers should have a decimal point separating the integer and
fractional parts – e.g., 38.98. You can also use engineering notation – e.g.,
3.898E1. For some options (e.g., starting values), you need to specify a vector
of numbers. To do this, write them out as a comma-delimited list prefixed by c(
and suffixed by ) – e.g., c(4.7, 0.1, 0.2). An example of a control list is:
showit=T, doeachint=T, lowerbounds=c(0,0,0)

The options you specify are exported to R without change and

turned into a list object, which is why the above format is required.

The options vary slightly depending on which detection function method is being
fit (ds, io, etc). More details are in the mrds R library online help, under the
appropriate ddf function (ddf.ds, ddf.io, etc). However, in general, the options
are as follows.
• showit – F (false, the default) or T (true); if true gives output at each
iteration of the fit
• doeachint – F (false, the default) or T (true); if true uses numerical
integration rather than an interpolation method during fitting
• estimate – T (true, the default) or F (false); if false fits the detection
function model but doesn’t estimate predicted probabilities.
• refit – T (true, the default) or F (false); if true, the algorithm tries
multiple optimizations at different starting values if it doesn’t converge

142 • Chapter 10 - Mark Recapture Distance Sampling User's Guide Distance 6.0 Beta 5
 • nrefits – integer number – controls the number of refitting attempts
• initial – a vector if initial values for the parameters
• lowerbounds – a vector of lower bounds for the parameters
• upperbounds – a vector of upper bounds for the parameters
• limit – T (true, the default) or F (false); if true then restricts analysis
to observations with detected=1 (this option is ignored if fitting method
= ds and there is no detected field)

Single Observer Configuration in the MRDS

Engine
You can analyze single observer data using the MRDS analysis engine if you set
the detection function method to ds. To do this, choose ds - single observer
from the drop-down list on the Detection function | Method page of the
Model Definition. Then specify the detection function model you want on the
Detection function | DS Model page.
You can expect results to differ slightly from those you can obtain using the
CDS and MCDS engines, as the optimization algorithm is different. Also note
that the default variance estimation method in the MRDS engine is different (see
Variance Estimation in MRDS).
There is currently no great advantage to fitting single observer data in the MRDS
engine, as the CDS and MCDS engines offer more features – however this may
change in the future as more features are added to the MRDS engine.

You can also fit CDS and MCDS models (i.e., models that assume all
animals on the trackline are detected) to double observer data by choosing the ds
detection function method. When you run such an analysis, Distance will pool
the data from the two observers, so that the data are the total number of unique
detections.

User's Guide Distance 6.0 Beta 5 Chapter 10 - Mark Recapture Distance Sampling • 143
Chapter 11 – Density Surface
Modelling

Introduction to Density Surface Modelling

This entire chapter is for advanced users only!

Density surface modelling (DSM) refers to the analysis of distance sampling
data, for the purposes of predicting the spatial arrangement of animals in the
study region as described by Hedley and Buckland (2004).
The DSM engine in Distance implements the “count” method described in that
paper, and this chapter describes how to use the DSM engine. We do not
describe the methods in detail – instead we assume the reader is familiar with
Hedley and Buckland (2004). Please note this is an advanced technique that
should only be considered once you are familiar with conventional distance
sampling.
The DSM engine currently has the following capabilities. We expect to extend
these in future versions.
• Line transect surveys (strip and point transects forthcoming)
• Perpendicular distance to objects (distance to object and sighting
angle not accommodated)
• Data where the object of interest is an individual or cluster (see
Clusters of Objects in MRDS)
• Response variable is estimated abundance within a segment
(estimated density and number counted in a segment in future
implementation)
• Up to one level of geographic stratification
• Variance estimation via moving block bootstrap; with uncertainty
derived from the estimation of detectability included in an ad hoc
manner
• Estimation of density/abundance using a detection function fitted in
a previous analysis – allows different subsets of the data to be used
for encounter rate and detection function fitting (see Using a
Previously Fitted Detection Function to Estimate Density in
MRDS).

User's Guide Distance 6.0 Beta 5 Chapter 11 – Density Surface Modelling • 145
 As with the MRDS engine, the DSM engine is implemented as a library in the
free statistical software R. When you run a DSM analysis from Distance,
Distance creates a sequence of R commands, calls the R software, waits for the
results and then reads them back in. Therefore, before you can use the DSM
engine, you must first ensure that you have R correctly installed and configured.
For more on this, see R Statistical Software in Chapter 7 of the Users Guide.
To produce a density surface model in Distance, you then need to set up the
project appropriately and include data in the correct format – see Setting up a
Project for DSM Analysis. You must next create one or more model definitions
using the MRDS analysis engine, and associate these model definitions with
analyses to derive detection probabilities for each objected detected. For more
about the basics of setting up analyses, see Chapter 7 - Analysis in Distance.
More details of the various models available in the MRDS engine are given in
Defining MRDS Models, and a detailed description of the options available in
the Model Definition Properties pages for this engine is given in the Program
Reference pages Model Definition Properties - MRDS. After deriving detection
probabilities, then a density surface model can be fitted. In addition, you must
also create a prediction grid that contains values of the covariates used for spatial
prediction at a grid of locations throughout the study region, not only along the
surveyed line transects. This prediction grid must be geo-referenced, and read
by Distance in a particular fashion to take advantage of the spatial data contained
therein. See Producing a prediction grid in GIS for further details.

If you are new to Distance, we strongly recommend you familiarize

yourself with the CDS analysis engine, for example by working through Chapter
3 - Getting Started before trying to analyze DSM data.

The Dolphin sample project is an example of how to set up a double-

observer study and specify analyses – see Sample Projects.

In this chapter we also provide some analysis guidelines, give a list of the output
the engine can produce and cover various miscellaneous topics.

If you are familiar with the R software, you can run the DSM engine
directly from within R, bypassing the Distance interface altogether. For more
information, see Running the DSM Analysis Engine from Outside Distance.

There is an online help that accompanies the DSM R library. This

contains more details about the methods and options available, and will be of use
to users familiar with R. To open these help pages from within Distance, choose
Help | Online manuals | DSM Engine R Help (html). The main function
to examine is dsm.fit().

Setting up a Project for DSM Analysis

The easiest way to set up a new project for an DSM analysis is using the Setup
Project Wizard.
• In Step 1, under I want to: select Analyze a survey that has been
completed.
• Be sure to tick the box indicating the Project will contain
geographic information at the bottom of the Step 1 screen

146 • Chapter 11 – Density Surface Modelling User's Guide Distance 6.0 Beta 5
 • In Step 3, under Observer configuration, select Double observer.
But see also Single Observer Configuration in the MRDS Engine.
• Follow through the rest of the wizard as usual.
Distance then creates the appropriate data fields for double observer data, and
you can then import your data using the Import Data Wizard. For more about
the data requirements, see Setting up your Data for MRDS Analysis.
Alternatively, you can create the appropriate fields by hand, and manually create
a new survey object with the appropriate observer configuration and data files.
For more about survey objects, see Working with Surveys During Analysis in
Chapter 7.

Setting up Your Data for DSM Analysis

It must be said that preparing your data is non-trivial. If you have not previously
constructed data for a Distance project, you will be operating at a bit of a
disadvantage. You will be well-served to closely study Chapter 5 of the
Distance Users Guide (particularly the sections on Importing one file per data
layer and Importing existing GIS data).

Segmenting your line transect data

There is no best way to accomplish this; some of the criteria for good
segmentation will be dependent upon your dataset. Nevertheless, your transects
will need to be divided into a number of "sub-transects" we will call segments.
As a rule we can use to move forward, you will wish to have segments that are
approximately equal in length to twice the truncation distance w, e.g., if you
conducted a survey of ants and were able to detect them to a distance of 1 meter
off the transect line, then you should divide your ant walking transects into
segments roughly 2 meters in length. In this manner the resulting areas in which
your survey has been divided is roughly square in shape.

Line transect after segmentation with individual segments of length li and truncation
distance w.

Layer-specific data files

In conjunction with segmentation of your data, you must think about the
hierarchal structure of your Distance project. The project we will be describing
possesses this structure:
Study region
Transects
Segments
Observations

We will not be concerned with strata that may arise in some distance sampling
applications. The sampling of our study region (otherwise know as the 'global
layer' in Distance) consists of transects. Each transect is composed of segments

User's Guide Distance 6.0 Beta 5 Chapter 11 – Density Surface Modelling • 147
 (thanks to the segmentation effort we performed previously). Within each
segment we may have detections of the objects we are studying.
As you know from your previous work with Distance projects, each layer of a
Distance project contains multiple fields. You will want to populate the
Observation layer with data that you think may be influential in modelling the
detectability of objects (e.g., observer, cluster size, etc.). This is exactly the type
of modelling you have done with previous versions of the Distance software.
You will also wish to populate the segment layer with data that you will wish to
use in your spatial modelling (e.g., latitude, longitude, soil depth, prey biomass,
etc.). Note these data are specific to the segment, so you will wish to think
carefully about how to integrate data that is defined at a point so that it will be
relevant at the spatial scale of a segment.

Constructing layer-specific files

If you surveyed your study area with sufficient effort, you have 20-40 transects
(as suggested by Buckland et al. (2001)). Now that you have segmented each
transect such that there are perhaps 10-20 segments per transect, the result will
be 200-800 segments. It goes without saying it will be important to track the
transect and segment within transect where each detection took place.
So, imagine a resulting dataset:.
Area T-A S-1 O-a
O-b
S-2
S-3 O-c
T-B S-4 O-d
S-5 O-e
S-6
S-7 O-f
O-g
T-C S-8
S-9 O-h
O-i
T-D S-10 O-j
S-11 O-k
O-l
S-12
T-E S-13
S-14 O-m
S-15 O-n
O-o

This represents 5 transects within the study area (labeled A through E), with
Transect A comprised of 3 segments, Transect B comprised of 4 segments, C
comprised of 2 segments, D having 3 segments, and E also having 3 segments).
By coincidence there were also 15 observations (2 in segment 1, none in
segment 2, 1 each in segments 3, 4 and 5, none in segment 6, 2 in segment 7,
none in 8, 2 in 9, 1 in 10, 2 in 11, none in either 12 or 13, one in 14 and finally 2
in 15). Note, this depiction omits the covariates at either the segment or
observation layer. If there are a large number of covariates included for
consideration in modelling either the detection function or the response surface,
the amount of data to bring into program Distance could be considerable.

148 • Chapter 11 – Density Surface Modelling User's Guide Distance 6.0 Beta 5
 We advocate the use of importing data by layers, which in this case might
consist of 3 files (one for each layer ''transect.txt'', ''segment.txt'', and
''observation.txt''). ''transect.txt'' would contain the transect label and its length,
''segment.txt'' would contain the label of the transect to which each segment
belonged, along with segment-specific data such as segment length, latitude, and
longitude mentioned earlier. Finally ''observation.txt'' would contain the
identifier of the segment in which the detection took place, plus items such as the
perpendicular distance of the detection, cluster size, and other data associated
with detectability.
The observation layer file will need to contain three other fields beyond those
already mentioned. If you are working with double platform (MRDS) designs,
you will already include them in your data. If you are unfamiliar with double
observer designs, and the letters MRDS mean nothing to you, then consult the
section of this users guide regarding the MRDS engine. Setting up your Data for
MRDS Analysis. There are two fields that will take the value '1' for each
detection. They will be called 'Observer' and 'Detected' when they are imported.
Finally the last field in this layer will be called 'Object' and it will contain a
unique number for each object detected. However see also Single Observer
Configuration in the MRDS Engine regarding elimination of these fields in
favour of a simpler approach.

Producing a prediction grid in GIS

However, we are finished with the easy part of preparing our data for analysis
with the spatial modelling capacity of Distance. The next step involves creation
of an ESRI shapefile. You may have expected this step, because if you were
going to perform spatial modelling it is only logical there would need to be a
GIS in your future. So, take the plunge.
You will construct a prediction grid, because you will be predicting with
covariates across your study region. Define some part of the universe that you
call your study region (this requires considerable thought). Use your GIS system
to produce a shape file of type point. These points should be thought of as the
centers of cells into which you (courtesy of your GIS) have subdivided your
study region.

Try not to construct a prediction grid with tens of thousands of cells;

this will be agonizingly slow to import and analyze (when you are performing
your final analysis with a dense prediction grid, be prepared to wait for
considerable lengths of time for import, and particularly for bootstrap variance
computations). Try not to exceed a couple thousand cells in your prediction
grid.

For each of these cells you will need to populate an attribute table with all of the
covariates you specified at the segment layer of your Distance project and that
are included in the density surface model you wish to use for prediction. So,
assuming you are using ArcGIS as your GIS software, perform the following:
• In ArcGIS, create a shape file of type point; containing covariates
of interest; clipped to the extent of the study region,
• Compute the area of the cells (boundary cell sizes can be ignored
for sufficiently dense prediction grid spacings relative to the size of
the study region),
• Open attribute table of this object, and create a new field called
“LinkID” of type "Number" and width 16.
• use the “Calculate Values” tool to fill that attribute field with a
formula equivalent to the value of the FID field plus 1. This is

User's Guide Distance 6.0 Beta 5 Chapter 11 – Density Surface Modelling • 149
 accessed by highlighting the newly-created field and pressing the
right mouse button
o Having made this modification the .dbf file you have worked
with will also be modified (without explicit saving by you)
• Export the attribute table to ASCII. This file will be imported into
the Distance project via the data import wizard in due course.

Importing prediction grid into Distance

This prediction grid exercise has resulted in the creation of a host of files. These
will need to be mated with Distance to make the prediction grid available to the
analytical tools in Distance. The steps involved in this process require
considerable care. A refresher (or perhaps a first inspection) of the section of the
Distance Users Guide Geographic (GIS) Data and Importing GIS Data by
Copying an Existing Shapefile into the Data Folder would be helpful.
• A new layer must be created in your Distance project. Before
creating a new layer, remember to ask Distance to either turn off
the coordinate system for the global (study area layer), or change
the default coordinate system in the Project settings (by selecting
Data |Preferences from the menu).
o You will need to have a shape file associated with the study
area layer before creating the coverage layer in the following
step. This is because the boundaries of the study region need
to be known before Distance can try to drop a coverage layer
atop the study area. If you wish to manufacture study area
boundaries that represent a simple shape (such as a rectangle),
this can be simply done by double clicking on the shape field
(polygon) of the study area layer, and entering coordinates of
four vertices.
• Create the new layer, making it a child of the Study Area layer (not
a child of any other layers extant in the project). This layer will be
of type coverage.
o This layer will need to have exactly the same number of
records as the prediction grid you created with your GIS. If
you have a small number of cells in the prediction grid, you
can add records to the layer after you've specified spacing of
cells so as to get approximately the correct number of records.
With a large number of records in your prediction grid, you
will wish to iterate the cell spacing such that you get close to
the correct number of records, and then add or subtract records
to arrive at the correct number of cells in your prediction grid.
o It seems a bit of a shame, now that you've gone to all this work
building this layer, but the layer created in this process will be
overwritten by the GIS shapefile constructed earlier. It is
critical that you note (write it down) the name of the layer
containing the coverage grid you have constructed.
• Import the attribute data (exported to an ASCII file when you were
working with ArcGIS). The attributes are only those fields
containing covariates you may use to predict abundance throughout
your study area. You will want to import data into the coverage
layer of your Distance project.
• Go into the Data Folder for the Distance project you are
constructing. Find the shape file (''.shp''), and other files with the
same prefix (but different extensions) for this coverage grid layer.
Delete these.

150 • Chapter 11 – Density Surface Modelling User's Guide Distance 6.0 Beta 5
 • Copy the shapefile (and companion files) manufactured by your
GIS work with the prediction grid into the Distance project data
folder, renaming them identically with the name of the coverage
grid data layer you have just manufactured.
• In this fashion, you have swapped the GIS-generated shapefile for
the artificially manufactured coverage grid layer.
• Check to see that in the prediction grid layer, you have values in
the shape (type point) fields: these are the coordinates of your
prediction grid cells. Also make sure you have values in all rows
for the attribute fields you have imported. As an extra check of the
integrity of the prediction grid layer, you can ask Distance to map
the prediction grid cells, by using the mapping function of
Distance.

Defining DSM Models

Introduction to DSM Models

The concept underlying density surface modelling is that a response variable is
modeled as a function of predictor variables. However, the response variable for
our purposes can be a count, an estimated density, or an estimated abundance
within each transect segment. For purposes of this users guide, we will focus
upon estimated abundance as our response variable; with the estimated
abundance having been adjusted by detection probability to account for
individuals not detected during the survey.
To proceed, we will specify a response variable, a link function, and a
distribution for error terms. Different link functions and error distributions are
appropriate for different response variables. The remainder of this chapter will
discuss the modelling of segment abundance, but for the sake of completeness
we will depict some common combinations of response, link, and error
distribution.
Area of Possible
Response Link
segment Weighting error
variable function
(offset) models

Abundance Covered Quasi-poisson

None Log
n̂i ai = 2wli Negative
binomial

Count Effective Quasi-poisson

None Log
ni aˆi = 2 μ̂li Negative
binomial
Density
Covered area Normal
nˆ
Dˆ i = i None
ai = 2wli Identity
Gamma
ai

Specifying DSM Model Formulae

One must specify formulae that tell Distance which covariates to include in the
model for fitting the density surface model to the predictor covariates. These
covariates are attributes of the transect segment (or higher spatial scale). Aside:
it does not make sense to include predictors at the observation level in models

User's Guide Distance 6.0 Beta 5 Chapter 11 – Density Surface Modelling • 151
 that predict response at the segment level because there may be multiple
observations within a given segment.
Formulae consist of a series of terms joined by operators such as “+”. The terms
represent covariates and the operators tell Distance how the covariates relate to
one another.
For example, the formula
latitude + longitude + depth
means include the data from the fields latitude, longitude and depth as
covariates.
The names of predictor covariates are possible transformations of the field
names in the Distance project database. See the section Translating Distance
Fields into DS and MR Covariates. The concepts of factor covariates (Factor
and Non-factor Covariates in MRDS) also apply to the use of covariates for
density surface modelling. Operators for formulae are the same as for MRDS
modelling (); however there is the additional smoother operator s, that can
operate either in a univariate (s(depth)) or bivariate (s(latitude, longitude))
manner. The smoother operator will fit a nonlinear spline. See Wood (2006) for
details on fitting of GAMs to data.

To see some examples of model formulae, have a look at the Dolphins

example project.

DSM Analysis Guidelines

These methods are relatively new, so we are only beginning to gain experience
on effective analysis strategies. Some preliminary guidelines follow – please let
us know if you can suggest some further guidelines or amendments.
Producing point and interval estimates of abundance in a study region using the
DSM engine requires four steps
• Modelling the detection function,
• Modelling the estimated segment-specific abundance as a function
of covariates,
• Extrapolation, courtesy of the model, from the covered region of
the study region to the unsurveyed portion of the study region
(given measures of the predictive covariates throughout the study
region). This is termed the prediction step, and
• Producing variance estimates (and confidence limits) using
parametric bootstrapping techniques.
We will not discuss the modelling of the detection function, as that is
discussed elsewhere (Modelling the Detection Function). The remaining
steps however, are unique to the density surface modelling, and receive
attention herein.

Density surface model

The modelling of the response variable (abundance or count or density) proceeds
by specification of the analysis that contains the estimated detection probability

152 • Chapter 11 – Density Surface Modelling User's Guide Distance 6.0 Beta 5
 for each object (individuals or cluster). The level at which the segments are
represented in the Distance project is also required.
Specification of the form the relationship between the response and the
covariates include the following:
• Generalized linear model or generalized additive model
framework,
• Link function,
• Offset term (if necessary),
• Form of the error distribution, and
• Weighting factor for observations (if necessary);
along with the model formula that uses the operators described in Specifying
DSM Model Formulae.
Modelling involves fitting a series of models to the data, and employing model
selection to compare them. For GAMs, the model selection diagnostic metric is
the generalized cross-validation (GCV) score. Models with small GCV scores
tend to do best at predicting response when presented with a new set of
covariates, and this is our objective when modelling density surfaces.
The number of knots in a GAM smooth (governing the wiggliness of the
predicted relationship) is estimated from the data, with the default for a
univariate smooth being 10 knots, and 30 knots for a bivariate smooth. The
number of knots can be fixed by the user by stating s(latitude, k=5,
fx=TRUE) as part of the model formula, where k is the number of knots.

Prediction of abundance to unsurveyed areas

Having modeled the relationship between the response variable and the predictor
covariates within the covered region, we rely upon the model to predict the
response in unsurveyed portions of the study region (hence the concept of these
estimates being model-based). We use a prediction grid that partitions the study
area into cells, with all relevant predictor covariates being measured within all
cells. The estimated response in each of the cells can be aggregated (summed
for abundance and count responses, averaged for density responses) to produce a
study region estimate (or prediction to any other spatially-delineated region
within the study region).
To perform this estimation, the necessary ingredients are
• Identification of the analysis that produced the preferred density
surface model preferred for model-based estimates of the response
variable.
• Name of the spatial layer in the Distance project that contains the
prediction grid and the covariate predictors employed in the chosen
density surface model for each cell (Importing prediction grid into
Distance).
• Size of each cell in the prediction grid (this may be constant for all
cells in the study area, or may be cell-specific). If the value is
constant, the scalar value can be specified in the Distance form
when the analysis is performed. Otherwise, a field in the prediction
grid layer is populated with the size of each cell, and the name of
that field is specified at the time of analysis (Producing a prediction
grid in GIS).
• The regions over which the aggregation is to take place. Note you
must specify a layer that contains polygons (because the edges of
the polygon are necessary to determine prediction cells

User's Guide Distance 6.0 Beta 5 Chapter 11 – Density Surface Modelling • 153
 inside/outside of the region). The polygons may be either part of
the survey in the Distance project that gave rise to the sightings
(such as strata), or some other polygon contained within the
Distance project (an area of special interest such as a biological
reserve, within the study region). The default is aggregation over
only the study region.

Variance estimation using parametric bootstrap

Because density surface modelling consists of using estimated abundance or
counts as the response variable, density surface modelling constitutes a form of
meta-analysis (analyzing the results of analyses). As such, there are two
components to uncertainty in the resulting estimates of abundance in the study
region.
We can approximate the combination of these two forms of uncertainty by using
the delta method approximation.

{cv( Nˆ overall )}2 = {cv( pˆ )}2 + {cv( Nˆ DSM )}2

The precision of the detection probabilities ( p̂ ) are provided by the engine that
fitted the detection function. The uncertainty of the abundance estimate
produced by the density surface model is determined by applying a parametric
bootstrap using something called a moving block that shuffles the residuals from
the fitted density surface model among segments.
This particular application of the parametric bootstrap is capable, in our
experience, of producing rogue replicates that are orders of magnitude larger
than the median replicate. This has the consequence of producing quite large
estimates of variance for the density surface modelling component of estimation.
We have instituted a trimming algorithm attributed to Tukey (1977). Values that
lie more than coefficient*interquartile range below the first quartile, or the same
amount greater than the third quartile are considered outliers, and are not
incorporated in the calculation of the quantile confidence interval for N̂ DSM ;
likewise these outliers are not included in the computation of the
cv( Nˆ DSM ) used in the delta method approximation to produce the measure of
precision that incorporates uncertainty both in the density surface model as well
as the detection function model.
This is a new analysis engine – you can expect some teething problems. Contact
the program authors if you can’t resolve them (see Sending Suggestions and
Reporting Problems).

Output from DSM Analyses

The DSM engine produces the following output:
• a summary of results in the Analysis Browser. For general
information about the Analysis Browser, see the section
Introduction to the Analysis Browser in Chapter 7. There are many
results statistics available, and you can select which ones are shown
independently for each analysis set using the Column Manager (see
Column Manager Dialog in the Program Reference).
• a detailed listing of results in the Results tab of the Analysis
Details window. These are described in the following section,
MRDS Results Details Listing.

154 • Chapter 11 – Density Surface Modelling User's Guide Distance 6.0 Beta 5
 • a log of the analysis, highlighting any possible problems, in the
Log tab of the Analysis Details window. For information about
troubleshooting problems, see Chapter 12 - Troubleshooting.
• (optionally) plots from the results details can be imported into other
programs.

DSM Results Details Listing

When an analysis has run, a great deal of information is available in the Results
tab of the Analysis Details window. A separate result will be produced for the
fitting, prediction, and variance estimation steps of a density surface
analysis.

DSM fitting step

Results from this step are split into the following pages:
Response surface/summary
• A summary table showing the result of the fitting of the GAM (or
perhaps GLM) showing estimated degrees of freedom for each
term along with an overall percent of deviance explained by the
model and generalized cross-validation score.
Response surface/Plot: gam-check
• Plots of various diagnostics (Q-Q goodness of fit plot, histogram of
residuals, residuals against linear predictor, and response against
fitted values)
Response surface/Plot: residual versus fitted
• A detailed (larger) plot of the residuals against the fitted values
(rather than against the linear predictor so that the fitted values are
on a scale comparable with the original per-segment estimated
abundance). This is a way of examining residuals that we find
most useful because it is depicted on a biologically relevant scale.
Response surface/plot: diagnostic plot
• As many of these plots are produced as there are smoothed
predictor covariates in the model (non-smoothed predictors are not
plotted). The response as a function of the predictor is shown,
along with a confidence interval and rug plot showing values of the
predictor variable for which there were observations.

DSM prediction step

Results from this step are split into the following pages:
Response surface/prediction: Estimation for entire study area
• Produces a single value, i.e., response aggregated over the entire
study area
Response surface/prediction: aggregated sums
• Table of point estimates for each of the requested polygons within
the study region for which an aggregated response was requested

DSM variance estimation step

Results from this step are split into the following pages:
Response surface/variance: Bootstrapped measure of precision
• Merely indicates version of the MRDS engine being used

User's Guide Distance 6.0 Beta 5 Chapter 11 – Density Surface Modelling • 155
 Bootstrap variance computations: outlier removal
• Describes result of outlier removal (percent of replicates meeting
the outlier criterion and list of those replicates values removed)
Response surface/Variance plot: bootstrap distribution
• Histogram showing distribution of bootstrap replicates after outlier
removal
Bootstrap confidence interval for abundance within study area
• Percentile confidence limits incorporating only uncertainty
associated with the density surface modelling,

• cv( Nˆ DSM ) and point estimate of N̂ DSM and standard error

derived from trimmed bootstrap distribution

• cv( Nˆ overall ) from which a confidence interval around N̂ overall is

calculated by back calculating a standard error from the CV, and
applying a log-based interval around the point estimate
Bootstrap confidence interval for the sub-units specified
• Percentile confidence intervals from the bootstraps for each of the
sub-regions requested.

DSM Analysis Browser Results

Currently, a relatively limited number of statistics are available in the Analysis
Browser, for the density surface modelling step of the analysis. Statistics
available for inspection via the Analysis Browser are: proportion of deviance
explained, generalized cross-validation score, generalized cross-validation scale,
and number of segments included in the analysis (including cells with response
of zero). We will be expanding the list of outputs in future versions. There are
no results available to the Analysis Browser for the prediction or variance
estimation steps.

Exporting DSM Results

The methods of exporting results from the Analysis Browser and results pages of
the Analysis Details to other programs are the same as those for CDS analyses,
as documented in Exporting CDS Results in Chapter 8. For example you can
copy the results details text by choosing Analysis Results | Copy Results
to Clipboard, and you can copy plots by choosing Analysis Results | Copy
Plot to Clipboard. Note though that in the case of the plot, the underlying
data are not copied as well (unlike for CDS plots) – just the plot picture.
There is another way to get hold of the plots produced by DSM analyses: to
directly access the image files produced when each analysis is run. Each plot
that is displayed in the Results Details has a corresponding image in the project’s
R Folder. For more details, see Images Produced by R in Chapter 7 of the Users
Guide.

Miscellaneous DSM Analysis Topics

Clusters of Objects in DSM

If the objects detected are clusters of individuals, you tell Distance this in the
same way as for CDS analyses, in the Setup Project Wizard (see Survey
Methods Wizard Page in the Program Reference).

156 • Chapter 11 – Density Surface Modelling User's Guide Distance 6.0 Beta 5
 Unlike CDS analysis, the Model Definition does not offer any regression
methods for dealing with size bias. If you suspect size bias is a potential
problem, the appropriate way to deal with it in an MRDS analysis is to include
cluster size (or some transformation of cluster size) as a covariate in the
detection function model(s).

The cluster size field is one of the fields with a fixed name in
detection function formulae in DSM (see Translating Distance Fields into DS
and MR Covariates) – in formulae you should use the name size regardless of
the actual field name.

Stratification and Post-stratification in DSM

At the moment, the MRDS engine only accommodates one level of stratification,
and this stratification is assumed to be geographic. There is no allowance for
weighting responses among strata in any fashion.

Running the DSM Analysis Engine from Outside

Distance
The DSM engine is implemented as a library in the free statistical software R.
When you run a DSM analysis from Distance, Distance creates files containing a
set of R commands and the appropriate input data, calls the R software, waits for
the results and then reads them back in. For more about R, see R Statistical
Software in Chapter 7 of the Users Guide.
Some users may wish to run the engine from outside the Distance interface –
either from within the R GUI interface or from another program. For example,
you may want to automate the running of analyses for simulations or
bootstrapping.
To see the format of the input and data files produced by Distance, try running a
DSM analysis in debug mode. To set debug mode on, choose Tools |
Preferences, Analysis tab, and tick Debug Mode. When you run analyses
in debug mode, the input and data files are created, but the analysis is not run.
The Log tab displays the location of the files – these are created in a directory
with a name “dst” followed by up to 4 numbers, located within the Windows
temporary directory. The file “in.r” contains the commands. Data are located in
the files “ddf.dat.r” (used in the detection function modelling), “region.dat.r”,
“sample.dat.r” and “obs.dat.r” (used in estimating density given a fitted detection
function). You can use these files as templates for creating your own command
and data files.
To run the analysis from within the R GUI (Graphical User Interface), you can
cut and paste the commands from the file in.r. To run the analysis from another
program, you can call R in batch mode – this is achieved by calling the program
RCmd.exe, which is located within the /bin subdirectory of your R installation.
For more details, see the R for Windows FAQ (in R, type help.start() and
when a browser window opens, click on the FAQ for Windows port). For an
example of its use, see the Log tab of any DSM analysis you have run that was
not in debug mode – you should see a line of the form:
Starting engine with the following command:
C:\PROGRA~1\R\rw1091\bin\Rcmd.exe BATCH C:\temp\dst90474\in.r
C:\temp\dst90474\log.r

Users familiar with R may wish to work inside the R GUI. The DSM engine is
contained in the library DSM. To load the library from within R GUI, type
library(dsm)

User's Guide Distance 6.0 Beta 5 Chapter 11 – Density Surface Modelling • 157
 All the functions in the dsm library are documented – the main functions are
dsm.fit() (fits the density surface model) and dsm.predict() (produces the
estimated response across the prediction grid). You can open a copy of the help
files from within Distance by choosing Help | Online Manuals | DSM
Engine R Help (html).

The use of these libraries in operating systems other than Windows

is not supported (but may well work – let us know!)

Installing an Updated Version of the DSM Engine

The DSM engine is implemented as a library (called “DSM”) in the statistical
software R. From time to time, we may issue updated versions of the library, for
example in response to reported problems. These will come as an archive file
dsm.zip. To install the new version,
• Find the old version of dsm.zip in the Distance program directory.
• Copy the new dsm.zip over the top of it (you may want to rename
the old version first, as a backup)
• Choose Tools | Preferences | Analysis, and tick the option to
Re-install analysis engine library on next run.
• Open a project containing analyses that use the DSM engine (for
example the dolphins sample project).
• Run one of these analyses.
• Check in the Log tab of the Analysis Details. You should find the
line:
package 'dsm' successfully unpacked and MD5 sums checked

The next topic describes how to check which version of the library is being used.

Checking Which Version of the DSM Engine is

Being Used
The DSM engine is implemented as a library (called “DSM”) in the statistical
software R. From time to time, we may issue updated versions of the library, for
example in response to reported problems. Therefore, before downloading a
new version or reporting a problem you may want to check which version of the
library is currently in use. To do this, re-run an analysis that uses the DSM
Engine (such as one from the dolphins sample project) and look in the Log tab
for the line
> library(dsm)

After it, you should see a line which looks something like the following:
This is dsm 1.0
Built: R 2.5.1; ; 2007-07-20 14:41:38; windows

If you are reporting a problem you should quote both the build number (in the
above case 1.0) and the build date and time (2007-07-20 14:41:38).
The previous topic describes how to update to a newer version of the DSM
Engine, if one is available.

When reporting results, you may want to cite the exact version (i.e.,
build number) of the library that used in the analysis. This is stored in the Log
tab, as outlined above.

158 • Chapter 11 – Density Surface Modelling User's Guide Distance 6.0 Beta 5
 Fine-tuning a DSM Analysis

In most cases, the default options for estimation of detection function parameters
in a DSM analysis work adequately. If you are a seasoned veteran in the use of
mgcv() for fitting density surface models, you may wish to access some of the
inner levers and knobs. This kind of fine-tuning is specified in the Detection
function | Control page of the model definition.
To enter options on this page, type them in as a comma-delimited list. Any
noninteger numbers should have a decimal point separating the integer and
fractional parts – e.g., 38.98. You can also use engineering notation – e.g.,
3.898E1. For some options (e.g., starting values), you need to specify a vector
of numbers. To do this, write them out as a comma-delimited list prefixed by c(
and suffixed by ) – e.g., c(4.7, 0.1, 0.2). Consult the R documentation for
detailed instructions regarding the use of gam.control() features.

The options you specify are exported to R without change as

arguments to gam.control(), which is why the above format is required.

User's Guide Distance 6.0 Beta 5 Chapter 11 – Density Surface Modelling • 159
Chapter 12 - Troubleshooting

Known Problems
A list of known problems at time of release is in the file ReadMe.rtf, in the
Distance program directory.
For a more up-to-date list, see the Support page of the Distance web site (you
can access the web site from Distance by choosing Help | Distance on the
web | Distance home page…).

Internal Errors in the Interface

Internal errors occur when Distance encounters a situation it has not been
programmed to expect. In response, Distance shows a message box that gives a
technical description of the problem, as well as the place in the code that the
problem occurred.
In some cases the cause of the problem is obvious. For example, if you use a
database application to delete part of a distance project file and then open the file
in Distance, internal errors will result. In most cases, however, an internal error
indicates that you have found a program bug. If you suspect that the error is
indeed caused by a bug, you should contact the program authors, detailing the
exact circumstances under which the problem occurred and the exact text
generated by the internal error message box. See Staying in Touch in Chapter 1
of the Users Guide for more information.

Recovering from Internal Errors

If you press the OK button on the internal error message box, Distance will try
to continue as if the error had not occurred. In some cases it is successful, but in
most cases further internal errors will be generated. If you are able, you should
try to close the project file and exit Distance. If Distance will not allow you to
exit, you should manually end the program using the Windows Task Manager.
To do this, press the Ctrl, Alt and Delete keys at the same time, highlight the
Distance program and press End Task. (You may have to press End Task again
if a Wait message appears.) Now follow the instructions on the page Recovering
from Unexpected Program Exit further on in this chapter.

User's Guide Distance 6.0 Beta 5 Chapter 12 - Troubleshooting • 161

Problems with the Analysis Engines

Errors and Warnings in the CDS and MCDS

Analysis Engines
Some analyses may results in warnings or errors being generated by the CDS
and MCDS analysis engine. The log tab will then be amber (warnings) or red
(errors), and you should look in the analysis log to find the appropriate warning
or error message. A complete list of these messages is given in the Appendix on
the MCDS Engine Command Reference, under MCDS Engine Error and
Warning Messages.

Internal Errors in the CDS and MCDS Analysis Engines

Occasionally, while running an analysis, the CDS and MCDS analysis engine
encounters a situation that it was not programmed to deal with and either shuts
down with an Internal Error message, or, more rarely, crashes with a Fortran run-
time error. Neither of these will cause the Distance interface to crash – instead
the analysis will be given error status, and any errors will be reported in the Log
tab. If any results were generated they will be in the Results tab, but results stats
are not presented in the Analysis Browser.
A list of internal errors is given in the Appendix on the MCDS Engine
Command Reference, under MCDS Engine Error and Warning Messages. More
about the output from a program crash is in that appendix under MCDS Engine
Command Line Output. Note that output from a CDS program crash will only
be saved if you ticked the option Capture command line output from CDS
and MCDS engines in WinNT in the Analysis Preferences Tab of the
Preferences Dialog.

Run-time errors are rather more common in the MCDS engine. If you
are using Windows NT/2000/XP, then Distance saves a copy of the command-
line output to the Log file. This can be useful in diagnosing the source of the
problem, so you should make a note of what is recorded there when discussing
the problem with the program authors.

If you are using Windows 98/ME, you can still access the command-line output,
by running the analysis in debug mode within the interface, and then running the
analysis using the analysis engine from the Windows command line. For more
information, see Running the MCDS Engine in the Appendix on the MCDS
Engine Command Reference.

Problems with the MRDS Engine

Please report any problems not listed below to the program authors.

Before reporting any problems, please note down which version of

the mrds library in R you are using, and the build date, and include this
information with your problem report. To find this information, see the page on
Checking Which Version of the MRDS Engine is Being Used in Chapter 10.

Problem Solution
Plots cannot be viewed or are poor quality Change image format or image properties
in Results tab of Analysis Details. - see Images Produced by R.

162 • Chapter 12 - Troubleshooting User's Guide Distance 6.0 Beta 5

 Analysis did not produce any output, or Look in the Log tab at the R commands
ran with unexpected errors used and any error messages produced by
R (see Analysis Details Log Tab).
If there seems to be a problem with the
data (e.g., a data selection query problem,
it is worth running again with Tools |
Preferences | Analysis | Echo data
to log turned on.
If you are an advanced user familiar with
R, you could run the analysis again in
Debug mode (see Running the MRDS
Analysis Engine from Outside Distance),
and then cut and paste the commands
from the input file line by line. Once the
problem has been encountered, you may
be able to use your knowledge of R to
determine the appropriate fix. If this is
caused by a bug in the program, please
contract the program authors.
Detection function estimates not produced You may need to manually specify
due to problem in the fitting algorithm starting values and bounds on parameters,
(e.g., lack of convergence) or maximum iterations – see Fine-tuning
an MRDS Analysis for details.

Stopping an Analysis
On occasion, you will want to stop an analysis that is running. For example, you
may have set off a long bootstrap analysis by mistake. Or perhaps the analysis
engine has locked up.
• To stop an analysis from the Analysis Browser, click on the Reset
Analysis button .
• To stop the analysis from the Input tab of the Analysis Details
window, click on the Stop button (which replaces the Run button
when an analysis is running).

On some systems, although the analysis appears to have

stopped, it actually continues running in the background. You can see this in
Windows NT, 2000 and XP if you open the Windows Task Manager (from the
Windows Start menu, choose Run and type in taskmgr, or click Ctrl-Alt-Del
and choose the Task Manager option). If you look in the Processes tab of
the task manager, for CDS and MCDS analyses you may see the process
MCDS.exe still running after you clicked the Stop button in Distance. This is
the CDS and MCDS analysis engine. To kill the process, highlight it in the Task
Manager and click End Process. This may also happen when running the
MRDS engine – in which case you will see Rterm.exe still running. You can
end this process in a similar manner.

GIS Problems
If you are experiencing strange behaviour with a project that contains geographic
data, it could be because the GIS data is invalid in some way. Symptoms include
maps that are blank or for which Full Extent button in the Map Window
doesn’t seem to work properly, error messages when generating a grid layer or a
design.

User's Guide Distance 6.0 Beta 5 Chapter 12 - Troubleshooting • 163

 In these circumstances, the first option should be to check that the global and
stratum data layers are valid (for examples of various possible geometry
problems, see
http://support.esri.com/index.cfm?fa=knowledgebase.techarticles.articleShow&d
=26920).
For simple shapes, this is as simple as opening the Shape Properties Dialog for
each polygon and checking the vertices are all there and in the appropriate order
(for more on this, see GIS Data Format in Chapter 5). For more complex shapes
(and for simple shapes as a double-check), it is worth running one of ESRI’s
shapefile checking tools. For example, there is a cleanshapefile macro available
for Arc version 8 at
http://arcobjectsonline.esri.com/default.asp?URL=/arcobjectsonline/samples/utili
ties/cleanshapefile/cleanshapefiles.htm, while for version 9 there is a “Check
Geometry” tool available in ArcTools, under Data Management Tools | Features
(see
http://support.esri.com/index.cfm?fa=knowledgebase.techarticles.articleShow&d
=27429).
A second possible problem is an inappropriate projection. For more on
projections, see Coordinate Systems and Projections in Chapter 5.

Recovering from Unexpected Program Exit

Distance saves the changes you make to your project straight to the project files.
This means that should Distance quit unexpectedly, for example due to a power
cut or a fatal program error, all of your work is most likely to be safe. However,
under some rare circumstances, your project can become corrupted. This may
happen, for example, if Distance is part-way through making an update when the
problem occurs.

Fixing a corrupted project

In Distance, the Project File and Data File are actually database files. If these
files become corrupted, for example due to power failure while performing a
write operation, it may be possible to fix them using the database repair
functionality built into the compact database function. To do this, ensure all
projects are closed in Distance, then select the menu item Tools | Compact
Project. You will be prompted for the name of a project to compact. Select the
damaged project and press Compact.
For more about compacting projects, see Compacting a Project in Chapter 4 of
the Users Guide.

164 • Chapter 12 - Troubleshooting User's Guide Distance 6.0 Beta 5

Appendix - Program Reference

Introduction to Program Reference

This part of the documentation is designed to be referred to whenever you want
to know how to use a specific part of the Distance interface. Each section covers
a different window in Distance.
For a more in-depth treatment of major concepts involved in using Distance, see
the Users Guide.

Setup Project Wizard

The Setup Project Wizard guides you through the process of creating a new
project. It is started by choosing File | New Project… from the main Distance
menu.
At any stage during the setup process you can click on the Next button (or press
Alt+N) to move on to the next step, or the Back button (Alt+B) to return to a
previous step and amend previously selected options. Press Cancel (or Alt+C)
to cancel the wizard and creation of the new project.
The first page prompts you for the type of project you want to create. You
choice here will dictate the content and number of the next wizard pages. The
choices are:
15. Analyze a survey that has been completed.
Choose this option if you have performed a standard distance
sampling survey, and want to use Distance to analyze your data. In
the following pages the wizard will ask you for information about
your survey. It will use this information to set up one survey
object, and a simple data structure containing four data layers of
type:Global, Stratum, Sample and Observation. If you want to set
up a more complex data structure, you should choose option 5,
below.
16. Design a new survey.
Choose this option if you want to design a new survey, and
Distance will create a global data layer, containing one record.
You can then enter the co-ordinates of your study area using the
Data Explorer. A more extensive design setup wizard is planned
for a future version of Distance.
17. Use an existing Distance project as a template.
Choose this option if you want to use an existing project as the
basis for your new project. Distance will copy the project settings,
data structure, survey objects, data filters and model definitions

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 165

 from the project you select. The survey data and analysis results
are not copied. You can then import your survey data and run the
analyses. An example where this option is useful is where you
have a set of standard analyses that you want to perform on several
different datasets.
18. Import a project or command file created in a previous
version of Distance
Choose this option to import survey information and data from a
Distance 2.2-3.0 command file, or Distance 3.5 project file, or to
import all the information (including designs, surveys, analyses and
results) from a Distance 4 project file.
19. Exit the wizard and set up the project file manually.
Choose this option if you want to set up the project file by hand.
Click Finish, to be taken straight to the Project Browser, from
where you can create data layers and fields, survey objects, etc as
required.
For more information about these options, see Creating a New Project in Chapter
4 of the Users Guide.
There is also a check box where you can specify whether the project will contain
geographic (GIS) information. Your choice of option, above, dictates whether
this check box is accessable – for example if you choose option 2 (design a new
survey), then the project must be geographic.

After you have finished the wizard, and the project has been created,
you can turn a non-geographic project into a geographic one by choosing File |
Project Properties … and ticking the check box “Project can contain
geographic information” on the Geographic tab.

Setup for Analyzing a Survey

This part of the Setup Project Wizard is for when you have performed a standard
distance sampling survey, and want to use Distance to analyze your data.
In the following pages, Distance will ask you for information about your survey.
It will use this information to set up one survey object, and a simple data
structure containing four data layers of type Global, Stratum, Sample and
Observation.
If you want to set up a more complex data structure, you should go back to Step
1 of the Setup Project Wizard and choose the option to set up the project
manually.

Setup for Analyzing a Survey Introductory Wizard Page

This introductory screen gives some information about the part of the Setup
Project Wizard for when you want to analyze a survey already completed.
To skip this screen in the future, tick Don’t show this introductory screen
again.

Survey Methods Wizard Page

In this screen the Setup Project Wizard guides you through the second step of
project creation, which involves providing information about your survey
methods.
The following information is required:

166 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 • The type of survey, which can be either a line transect, point
transect or cue count survey.
• The observer configuration, which can be single or double
observer configuration. Single observer is the norm for
conventional distance sampling. Double observer configuration is
used to estimate g(0) when it is expected to be < 1 – see Chapter 10
- Mark Recapture Distance Sampling for more on this.
• The distance measurements – that is the type of distances that
were measured in the field. For line transects this can be
perpendicular distances or radial distances together with the angle
of the object relative to the trackline. For point transects and cue
counts only radial distances are measured.
• The type of observations – this is whether recorded
observations were of single individuals or clusters of individuals.

For those used to Distance 4 and earlier, the sampling

fraction option is now on the Multipliers Wizard Page..

Measurement Units Wizard Page

At this stage in the project setup you need to specify the measurement units for
your data, from the drop-down list. These units are used when setting up the
data structure.

You can change the measurement units later in the Data Explorer –
double-click on the 4th header row of the field you want to change and select
from the list of units.

You can measure data in one set of units, and output analysis results in
another set. The units of measurement are specified here and in the data sheet.
The units of analysis are specified in the Units tab of the Data Filter. See Data
Filter Units in the Program Reference for more information.

Multipliers Wizard Page

At this stage in the project setup you need to specify the multipliers for your
analyses. Multipliers are constants that are used to scale the density estimate.
For more detail, see Multipliers in CDS Analysis in Chapter 8 of the Users
Guide, and look up “Multipliers” in the index of the Distance book.
In this setup project wizard, you can add multipliers to account for the following
situations:
• When the sampling fraction (proportion of the lines or points that
were surveyed) is not 1. For example, imagine that only one side
of the transect line was observed. In this case the sampling fraction
will be 0.5. Another example is cue counting, where the sampling
fraction is the proportion of a full circle that is covered by the
observation sector. A third example is when all points or lines are
visited multiple times – then the sampling fraction is the number of
visits.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 167

 If the sampling fraction is not the same for all lines or points, you
account for this by adjusting the survey effort at the data entry stage. For
example, if you only sampled on one side of the line on 2 out of 20 transects, and
all transects were 10km long, then you should enter a transect length of 5 km for
those two transects. In this situation you set the overall sampling fraction to 1.

• Surveys where g(0) is less than 1.

• Cue count surveys (this box will automatically be checked if you
selected cue counts in the Survey Methods screen).
• Indirect surveys - that is surveys where the animal of interest is not
surveyed directly, but instead some object that is produced by the
animal is surveyed. Examples include nest counts and dung
counts. In this situation, two multipliers are added: one for the
object production rate and another for the mean time to object
disappearance. (Cue counts are a special case of indirect counts,
where the decay time is instantaneous.)
• Other - this gives you the ability to define a generic multiplier at
this stage.
Ticking the boxes associated with these options causes Distance to create a field
for the multiplier in the Global data layer. You enter the multiplier value at the
same time as your other data.
Most multiplier values are not known with certainty, but instead are estimates
with associated standard error (SE) and degrees of freedom (DF). As well as
creating a field for you to enter the multiplier value, Distance also creates fields
to enter the SE and DF. An exception is the sampling fraction, which should be
known. For the ‘Other’ multiplier, you can choose whether Distance should
create the SE and DF fields or not.

You can also define multipliers after the project has been created,
using the Append Field button in the Data Explorer. This gives you the
ability to define as many multipliers as you like. However, if you add the
multipliers in the Setup Project Wizard, Distance will automatically include
them in the default Model Definition. These issues are discussed in more detail
in the Users Guide page on Multipliers in CDS Analysis.

Finished Setup for Analyzing a Survey Wizard Page

This is the final stage of the setup project process prior to the creation of the new
project database.
You are required to choose one of the following destination options:
• Data Entry Wizard: Go here if you want to enter your data from
the keyboard and want to be guided through the process.
• Data Import Wizard: Go here if you want to import your data
from a text file.
• Return to Distance: Choose this option if you want to enter your
data using the Data Explorer (a non-guided interface that is
similar to the Data Entry Wizard but is more flexible). You can
still start up either the Data Entry Wizard or the Data Import
Wizard by selecting them from the Tools menu.
Save your setup settings for the next time you setup a new project by ticking
Save current settings as default.

168 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Setup for Designing Surveys
This part of the Setup Project Wizard is for when you want to design a new
survey. Distance creates a global data layer, containing one record. You can
then enter the co-ordinates of your study area using the Data Explorer. A more
extensive design setup wizard is planned for a future version of Distance.

Setup for Designing Surveys Wizard Page

This window gives some information about the Setup Project Wizard option to
setup a project for designing surveys. For more information about survey
design, see Chapter 6 - Survey Design in Distance of the Users Guide.

Use Another Project as Template

This part of the Setup Project Wizard is for when you want to use an existing
project as the basis for your new project.

Use Another Project as Template Wizard Page

Distance can use any Distance project as a template for the new project.
Distance will copy the project settings, data structure, survey objects, data filters
and model definitions from the project you select. The survey data and analysis
results are not copied.
An example where this option is useful is where you have a set of standard
analyses that you want to perform on several different datasets. For more
information, see the Using an Existing Project as a Template in Chapter 4 of the
Users Guide.
In this window, you select the project to import information from. You are then
taken to the final page, where you choose whether to import the new data or type
it in by hand.

Finished Use Another Project As Template Wizard Page

This is the final stage of the setup project process prior to the creation of the new
project database.
You are required to choose one of the following destination options:
• Data Import Wizard: Go here if you want to import the new data
from a text file.
• Return to Distance: Choose this option if you want to enter the
data using the Data Explorer (a non-guided interface that is
similar to the Data Entry Wizard but is more flexible). You can
still start up either the Data Entry Wizard or the Data Import
Wizard by selecting them from the Tools menu.
Save your setup settings for the next time you setup a new project by ticking
Save current settings as default.

Import from Previous Version of Distance

This part of the Setup Project Wizard is for when you want to import survey
information and data from previous version of Distance.

Import from Previous Version of Distance Wizard Page

Distance can import options and data from the following previous versions:
• Distance 2.0 to 3.0.
Under Files of type, specify “Distance 2.0 to 3.0 command files”, and

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 169

 select the command file you wish to import.
Distance will use information from the OPTIONS section of the
command file to create a Survey object containing information about
the type of survey (line, point, etc), type of object (line or point), and
type of distance measurements (radial or perpendicular). Distance will
also create the appropriate data structure and import the survey data.
Distance will not read the ESTIMATE section, so you will have to set
up the analysis specifications again yourself.

If you are having difficulties importing a Distance 2.0 – 3.0 command

file, the first thing to check is that it runs okay in the previous version of
Distance. You can download Distance version 2.2 from the Distance web site if
you do not have a copy.

• Distance 3.5.
Under Files of type, specify “Distance 3.5 project files”, and select
the project file you wish to import.
Distance imports the project settings, and uses them to create a Survey
object. It will also create the appropriate data structure and import the
survey data. Distance will not import the Data Filters, Model
Definitions or Analyses – you will have to recreate these again
manually.
• Distance 4
Under Files of type, specify “Distance 4 project files”, and select the
project file you wish to import. Distance will import all of the
information from the old project, including project settings, data, maps,
designs, surveys and analyses.

Data Entry Wizard

The Data Entry Wizard guides you through the process of data entry. It has
similar features to the Data Explorer (the Data tab of the Project Browser)
but is more suitable for beginners as it guides you through the process of
entering data, giving on-screen advice via a text box at the top of the window.
For new projects, you are taken to the Data Entry Wizard automatically from
the end of the Setup Project Wizard if you choose the option Proceed to Data
Entry Wizard. You can also access the Data Entry Wizard from the main
Distance toolbar, by choosing Tools | Data Entry Wizard.

You can only access the Data Entry Wizard if you have a simple
data structure, with a single global, stratum, sample and observation layer. In
other cases, you must manipulate your data using the Data Explorer.

The introductory screen in the wizard provides you with an overview of the 4
data layers (Global, Stratum, Sample and Observation) in your project. If you
want to find out more about the way that Distance stores survey data, you should
read the Users Guide Chapter 5 - Data in Distance.
As with the other wizards in Distance, you navigate through the wizard by
pressing the Next and Back buttons on the navigation bar (or pressing Alt-N
and Alt-B).
If you do not want to see the Data Entry Wizard introductory screen again
then tick the box Don’t show this introductory screen again.

170 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Global Layer Wizard Page
At the Global step of the Data Entry Wizard you should enter values for any
multipliers you selected in the Setup Project Wizard, along with their standard
errors. If you did not define any multipliers then you can move straight to the
next screen.
The Data Entry Wizard's screen is split into three parts. The upper section
contains the help text to assist you at each stage of the data entry process. This
section can be resized at any stage to make more space for your data. On the
lower left there is a hierarchical view of the data layers in your project – this is
the Data Layers Viewer. (The Data Layers Viewer is visible, but disabled
in the Data Entry Wizard - but it's fully functional in the Data Explorer!) On
the lower right there is a spreadsheet-like window, called the Data Sheet.
Have a look at the Data Explorer help page to find out more about the Data
Layers Viewer and entering data in the Data Sheet.
If you have not read Chapter 5 - Data in Distance in the Users Guide, you should
probably do so before proceeding much further.

Stratum Layer Wizard Page

At the Stratum step of the Data Entry Wizard you should enter information
about your survey strata. If your survey did not include stratification, you should
enter the total survey area in the “Area” field, and then move to the next screen.
Note that if you do not know the area of your strata, or the total study area, you
should enter 0. See Unknown Study Area Size in Chapter 8 of the Users Guide
for more details.
If you have not read Chapter 5 - Data in Distance in the Users Guide, you should
probably do so before proceeding any further. To find out more about
stratification in Distance, see Stratification and Post-stratification in Chapter 8 of
the Users Guide (although this is quite advanced!).
Have a look at the Data Explorer help page to find out more about the Data
Layers Viewer and entering data in the Data Sheet.

Sample Layer Wizard Page

At the Sample step of the Data Entry Wizard you should enter information
about your samples (transects or points).
If you want to find out about Data Field Types and other new Distance jargon,
you should read Chapter 5 - Data in Distance of the Users Guide.
If you want to find out more about the Data Layers Viewer and entering data
in the Data Sheet, have a look at the Data Explorer help pages.

Observation Layer Wizard Page

At the Observation step of the Data Entry Wizard you should enter
information about your observations.
Have a look at the Data Explorer help page to find out more about the Data
Layers Viewer and entering data in the Data Sheet.
If you have not read Chapter 5 - Data in Distance in the Users Guide, you should
probably do so before proceeding any further.
To find out more about stratification in Distance, see Stratification and Post-
stratification in Chapter 8 of the Users Guide (although this is quite advanced!).

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 171

 The Cluster Size column is of field type decimal, which means that
non-integer cluster sizes can be entered. This could occur if, for example, cluster
size is estimated by more than one observer and the mean is taken.

Finished Data Entry Wizard Page

To find out more about the data structure that you have created, see Chapter 5 -
Data in Distance in the Users Guide.
To find out more about the Data Explorer, see the Program Reference page on
the Data Explorer.

Import Data Wizard

This wizard guides you through the process of importing data into a Distance
project file. The wizard can be started in one of two ways:
• from the last page of the Setup Project Wizard, by choosing the
option Proceed to Data Import Wizard. This is the ideal way
to import data into a new project.
• by selecting the menu item Tools | Data Import Wizard. This is
the best way to add extra data from file into an existing project.
You can also replace your existing data with the imported data -
this is an option at the end of the Data Import Wizard.
General information about data import is given in the Users Guide pages on Data
Import in Distance. Before you try importing data it is essential to have a good
understanding of how Distance handles your survey data - such as would be
gained by reading Chapter 5 - Data in Distance in the Users Guide.
The data import wizard has six screens. The first is introductory. The second
asks you for the data source – the text file to import data from. The third allows
you to specify the destination of the data – which data layers to put the data into,
and how to assign rows in the text file to records in the Distance database. The
fourth screen asks you to specify the delimiter used in the text file, and the fifth
asks you to match up columns in the text file with fields in the database. The
last screen allows you to check if the number of columns and rows is correct,
and displays a log of any errors that occur during the import process.

After importing data, it is generally a good idea to carefully check it in

the Data Explorer, even if the import seemed to go smoothly.

Importing large datasets into Distance takes a long time. We

hope to improve the performance of the import routines in future releases.
Meanwhile, if you have a very large dataset, consider importing it into Distance
3.5 (which was much quicker) and then importing the Distance 3.5 project into
the latest version of Distance.

Data Source Wizard Page

At this step of the Import Data Wizard, you select the text file to import data
from. For more information about the types of text file and required format, see
the Users Guide pages on Data Import in Distance.

172 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Only files with the following extensions are allowed: txt, csv, tab,
asc, htm, html. This is because of a security restriction in the database engine
used by Distance. For more information, see the Microsoft knowledge base
articles Q247861 and Q239105 (go to www.microsoft.com and click on support,
then search).

Data Destination Wizard Page

At this step of the Import Data Wizard, you tell Distance where to store the
imported data.
Destination data layers
Here, you specify which data layers the new data will be stored in. You choose
the Lowest data layer and Highest data layer, and Distance shows you a
list of the data layers where records will be added, and the parent of the highest
data layer.
Normally, if you are importing data from a flat file containing all the survey
data, the lowest data layer is an Observation layer, and the highest a Stratum
layer. However there are other scenarios where you will want to make different
selections. For example:
• when you are importing the data one layer at a time – see Importing
one file per data layer in Chapter 5 of the Users Guide
• when there are no strata in your study, and you have already
created a single stratum record
• when you have a complex data structure, with more than the
default 4 data layers for analysis (Global, Stratum, Sample,
Observation)
Location of new records
• Add all new records under the first record in the parent
data layer. An example of when you would choose this option is
when the highest data layers is a stratum layer (as the parent data
layer is the global layer, which only contains one record).
• Input file contains a column corresponding to the
following field in the parent data layer. Choose this option
when you want rows from the input file to be assigned to specific
records in the parent data layer. For example, imagine you are
importing a text file into an observation data layer, containing one
row for each sighting. Clearly, you want each observation to be
assigned to the correct transect. Your data file contains a column
for the sighting distance, and also a column with the transect ID.
You choose this option, and under Field name, select ID.
Another example of the use of this option is given in the Users
Guide page Importing one file per data layer.
Creation of new records in lowest data layer
• Create one new record for each line of the import file. If
your lowest data layer is an observation layer, this is usually the
option you want to select. That way, if two successive records are
the same (for example, two sightings at 0 distance in transect 1),
then two records will be created for them in the data file.
• Create new records only when the line differs from the
previous line. This option is useful when you have multiple
rows that are the same in your input text file. For example,

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 173

 imagine you are importing a file containing just the transect data,
so that the lowest and highest data layer are both the sample layer.
The semicolon-delimited text file has columns for stratum label,
transect label and transect length:
Stratum A;Line 1A;10
Stratum A;Line 1A;10
Stratum A;Line 2A;10.3
Stratum A;Line 2A;10.3
Stratum B;Line 1B;5.7
Stratum B;Line 1B;5.7
Stratum B;Line 2B;8.4
Stratum B;Line 2B;8.4

there are 8 rows of data, but each transect is repeated twice.

Choosing this option ensures that only 4 records are created in the
sample data layer.

Data File Format Wizard Page

At this step of the Import Data Wizard, you specify the delimiter used to separate
columns of the text file you are importing data from.
Options
Delimiter. The delimiter is the character used to separate columns of data. For
more about the delimiters and data format, see Data Import in Chapter 5 of the
Users Guide. Here, you choose the appropriate delimiter for your text file.
Ignore rows. Tick the box to allow Distance to ignore the first row of your
text file. This is useful if, for example, you have put column names, or some
other reminder, on the first row.

Sometimes, when coming to this page, you get a “Problem reading

data file” message, and then no data is displayed in the bottom part of the wizard
page. This is often caused by the wrong delimiter being selected by default –
selecting the correct delimiter should fix the problem. If setting the delimiter
doesn’t solve the problem, then see Troubleshooting the Import Data Wizard.

If you commonly use this delimiter then on the Finished page, choose
the option to Save current settings as default, and your delimiter will
become the default.

Data File Structure Wizard Page

This is the screen where you tell Distance which column in the import text file
corresponds to which field in the Distance database.
There are two ways to achieve this: manually, by clicking on the first and second
row of grey boxes (Layer name: and Field name:); or using one of the
shortcuts.

Manually assigning columns to fields

You manually assign columns in the import text file to fields in the Distance
database by clicking on the first and second row of grey boxes. In the first row
you specify the Data Layer Name, and in the second, the Data Field Name.
For each column, click on the first row and choose from the drop-down list of
data layer names. Press Enter to confirm the selection. Then click on the
second row and choose from the list of available data field names. Once you

174 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 have assigned a field to one column it disappears from the list of available fields.
(When you have assigned all of the available fields in that data layer to a column
then further fields are created for you automatically, because Distance assumes
you want to import additional fields of data - see Importing additional columns
not yet in the Distance database, below.)
When you are using the drop down lists, pressing Enter confirms your selection
and pressing Esc cancels the selection.
If you assign a field to a column by accident, simply click on the first row for
that field, select “[Ignore]” and press Enter. This will clear the Data Field
entry.
Ignoring a column in the import text file
By default Distance ignores all columns of text with the word “[Ignore]” in the
data layer name row. By choosing Ignore, you can skip extra columns of text
that you do not want to import.
Importing additional columns not yet in the Distance database
Imagine that you wish to import an additional column of data that identifies the
sex of each animal observed into the observation layer. (You are going to use
this field for post-stratification, so that you can see if detectability varies by
species.) You have not already created a field for sex in the Data Explorer, so it
is not on the list of available fields that drops down when you click on the
second row. To specify a new field, you simply click on the row and type in a
name for the field. In this case, because there are no remaining fields in the
Observation layer, Distance automatically supplies the default field name “New
Field”; however you may prefer to call the field “Sex”. Distance automatically
chooses Data Type “Text” (it will be stored as Text inside Distance) – choose
from the drop-down list to assign another field type. It’s important to get the
field type right here as once the field is created, you won’t be able to change it.

Using shortcuts to assign columns to fields

The following two checkboxes can be used to save time compared with
manually assigning columns to fields in the Distance database.
Columns are in the same order as they will appear in the data sheet
If the columns in the text file correspond exactly to the order of the fields in the
Data Explorer, then ticking this box automatically assigns the columns to the
correct fields.

Before opening the Import Data Wizard, you can move fields in the
Data Explorer by dragging and dropping them. You will also need to create any
new fields before starting the Wizard.

First row contains layer names and field names of each column
In many database and spreadsheet packages, you can specify the contents of the
first row when exporting data. To use this shortcut, you will need the first row
to contain both the layer name and field name for each column, separated by
some delimiter. For example, first row of the column corresponding to the field
“Area” in the data layer “Region” would be “Region*Area”, assuming that “*” is
the delimiter used. Possible delimiters are: * | _ - and . (i.e., a full stop or
period).

This option is most useful as part of streamlining data import from a

database or other programmable application. Combined with the option to setup
a new project using another project as template (see Using an Existing Project as

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 175

 a Template in Chapter 4 of the Users GuideGuide,), it streamlines the application
of a prespecified, standard set of analyses by relatively inexperienced users to a
new set of data (such as might be required in a regulatory framework).

Finished Import Data Wizard

This is the last screen of the Import Data Wizard. Click Finish to import that
data according to the options you have chosen. If there are problems during the
import, a log of the errors, showing the row and column where they occurred, is
displayed. If the problem cannot be readily corrected, see Troubleshooting the
Import Data Wizard.
Options
Existing data: here you can choose to overwrite the existing data in the
project, or append the new data to the current records.
Save current settings as default Tick this box to save the selections you
have made in previous screens as the default for next time you run the Import
Data Wizard.

Troubleshooting the Import Data Wizard

Only files with the following extensions are allowed: txt, csv, tab, asc, htm, html.
This is because of a security restriction in the database engine used by Distance.
For more information, see the Microsoft knowledge base articles Q247861 and
Q239105 (go to www.microsoft.com and click on support, then search).
This page contains some hints that may help you if you're having problems
importing data into Distance. For more general information about importing data
in Distance, see Data Import in Chapter 5 of the Users Guide. For an overview
of the Import Data Wizard, see theImport Data Wizard page of the Program
Reference.
Problems between Step 3 (Data Destination) and Step 4 (Data File
Format)
When you press the Next button, to move from Step 3 to 4, Distance tries to
load the file you have selected and parse it using the default delimiter. Any
problems that occur at this stage result in a “Problem reading data file” message
being displayed.
Some of the possible messages generated by the database (last line of message
box) are explained in more detail below:
• Error 3440: “An attempt was made to import or link an empty text
file.” This message typically indicates that the wrong delimiter is
selected – selecting the correct delimiter should fix the problem.
This message usually indicates that the file you chose to import
was not a text file. You should check that it is a valid ASCII text
file (for example by opening it in Notepad), and that it is correctly
formatted for import (see Data Import in Chapter 5 of the Users
Guide).
• Error 3047: “Record is too large.” This message often indicates
that Distance failed to recognize any end-of-line symbols in the
file. Distance therefore assumes that the file is one line containing
a large number of fields (columns) - too many to import. The
Distance import engine expects the end of each line to be denoted
by a Carriage return (Cr - ASCII 13) + Line feed (Lf - ASCII 10)
combination. This is the standard used by almost all windows
software packages. However, we have come across some cases in
which certain versions of some packages can put either Cr+Cr+Lf

176 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 or just Lf at the end of each line. Lf alone is also the standard end-
of-line symbol on Unix machines. If your file is not of the correct
format, there are some possible remedies:
o Many software packages have different options for exporting
text files - for example Text Only, MS DOS Text, Text with
Line Breaks, etc. Try playing around with these options.
o Try opening the file in a different word-processor or text
editing package and saving it as text. From our experience, we
particularly recommend the shareware software TextPad
(www.textpad.com), which has options to save files into PC
format (File format, under Save As…), which automatically
makes the line end in Cr+Lf.
o One user reported that the problem disappeared when they
moved to a different machine, despite all the software settings
apparently being identical.
o If the file comes from a Unix machine, FTP it to the windows
machine as an ASCII file (not as a binary) - this will ensure
that the Lf 's are replaced with CrLf (or use TextPad, above).
If none of these remedies work for you, or you are having ongoing
problems, please contact the Distance Development Team.
• Error 3051: “The Microsoft Jet Engine cannot open the file ' '. It is
already opened exclusively by another user, or you need
permission to view its data.” This message often occurs because
the file is still open in the software package that you used to export
it. Close the file in the other package and try again.
Problems after pressing Finish
If there are problems during the import operation, a box containing a list of error
messages appears on the Finished page. These messages give the line number
that the problem occurred on, and so are often helpful in pinpointing the cause of
the problem.
Please let us know if you encounter a problem you cannot solve, or manage to
troubleshoot a particularly tricky import problem. We can add your experience
to the above list!

Project Properties Dialog

The project properties dialog box can be accessed from the main Distance menu
by selecting File | Project properties…. It contains information about the
current project. There are two tabs, General and Geographic.

General Project Properties Tab

The General tab of the Project Properties dialog displays information about the
project file and associated data folder, including file sizes and locations. There
is also a comments box in which you can enter some remarks about your project.

You can easily cut and paste from the comments box - right click on
the box to see a pop-up list of options.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 177

 Geographic Project Properties Tab
The geographic tab of the Project Properties dialog allows you to view and
change the geographic settings of the project. To find out more about
geographic data in Distance, see Geographic (GIS) Data in Chapter 5 of the
Users Guide.
Options
Project can contain geographic information.
If this box is checked, the project is a geographic one, and the data layers can be
spatially referenced. Once a project is geographic, you cannot un-check the box.
However, if a project is not geographic, you can convert it to a geographic
project by checking the box.
Default coordinate systems
Use the options here to define the default coordinate systems for geographic data
in the project, and for maps and geographic calculations. To find out more about
coordinate systems, see Coordinate Systems and Projections in Chapter 5 of the
Users Guide.
Geographic data. These settings will be applied to all new data layers created
in the project. It is best to set these options before creating any data layers, as
then all data layers will have the same coordinate system. Although this isn’t
strictly necessary (the only requirement is that they all have the same datum), it
will make calculations quicker.
Maps and geographic calculations. These settings will be applied to all
maps displayed in the project. In addition, they will be used as the default for
survey designs. If the geographic data’s geographic coordinate system is
“[None]”, or if the geographic data is already projected, then the projection is
automatically set. In other cases, you can specify a map projection and
projection parameters. Map units are the unit of distance used when displaying
maps.
Conversion between coordinate systems
Densification is the process of adding vertices to lines when projecting them
from one coordinate system to another. Densification tolerance is the maximum
distance allowed before adding a new vertex. Distances are defined in terms of
the geographic data units (usually latitude and longitude).
A densification tolerance of 0 (the default) means that no extra vertices will be
added. Above 0, the higher the value, the longer the distance between new
vertices and therefore the fewer the new vertices. Smaller values (above 0) mean
more vertices are added – which takes more computer time but increases the
accuracy of the projection.
The default densification tolerance can be set in the Preferences dialog – this
default is applied to all projects. Because the optimal densification tolerance
depends on the scale of the data and the accuracy required, it can be over-ridden
here on a project-by-project basis if required.

Preferences Dialog
The Preferences dialog lets you set options that relate to the behaviour of
Distance across all projects on this computer.

General Preferences Tab

Projects

178 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 • Default project folder: the default location used when creating
and opening projects. Click Browse… to choose a new location.
• Always create a backup copy when opening projects.
For more information about backups, see Saving and Backing up a
Project in Chapter 4 of the Users Guide.
• Check projects are on the local hard drive before
opening them. By default, when you open a project Distance
checks that it is located on a local hard drive, as opposed to a
removable drive, a network drive or a CD. If it is not on a local
hard drive, it issues a warning. This is usually sensible, as
performance is significantly degraded for projects not located
locally. However, if you are sure you want to be able to access
projects not stored on the local drive, then un-check this option.
• Store results in compressed format in project databases.
By default, the information shown in the Results tab of the Details
windows (e.g., Analysis Details) are compressed in the distance
project database to make the project smaller. If this box is not
checked, the results are stored as plain text. You will normally
only want to un-check this if you are directed to do so by the
program authors (for example, in response to a problem with the
current compression routine that can occur in some versions of
Windows such as Chinese, Japanese and Korean language
versions).
Interface
• Show Tip of the Day at startup. If checked, the Tips window
loads each time Distance is started. You can also open the Tips
window by selecting Help | Tip of the Day from the main menu.
• Save current position and size of all open windows as
default for new projects. Use this option to customize the look
of new projects in Distance.
• Default tab for runs with warning status. In some types of
runs, warnings are generated routinely – for example in MCDS
runs with cluster size as a covariate. By default the Log tab opens
when opening Analysis Details windows with Warning status, but
you can use this option to bypass the Log tab and go straight to
results.
• Results and Log font size. Use this option to set the default
font size. You can also alter the font size by right-clicking in both
the Results and Log windows and choosing the appropriate options.
Copy to clipboard
• Column separator. This option sets the separator used between
columns when copying tables (e.g., Browser contents, data sheet,
etc.) to the clipboard.
• Row separator. This option sets the separator used at the end of
lines when copying tables to the clipboard. Most packages expect
the end of line to be denoted by carriage return (Cr– ASCII 10) +
line feed (Lf – ASCII 13). However, if you are experiencing
trouble pasting tables into a package, try experimenting with other
separators.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 179

 Geographic Preferences Tab
The geographic tab of the Preferences dialog allows you to view and change the
default geographic settings. To find out more about geographic data in Distance,
see Geographic (GIS) Data in Chapter 5 of the Users Guide.
Default coordinate systems for new projects
Use the options here to define the default coordinate systems for new geographic
projects. To change the default coordinate system for a particular project, open
the project and then go to the Geographic tab of the Project Properties dialog
(File | Project Properties…).
To find out more about coordinate systems, see Coordinate Systems and
Projections in Chapter 5 of the Users Guide.
• Geographic data. These settings are applied to new data layers.
• Maps and geographic calculations. These settings applied to all
maps. In addition, they are used as the default for survey designs. If
the geographic data’s geographic coordinate system is “[None]”, or if
the geographic data is already projected, then the projection is
automatically set. In other cases, you can specify a map projection and
projection parameters. Map units are the units of distance used when
displaying maps.
Conversion between coordinate systems
Densification is the process of adding vertices to lines when projecting them
from one coordinate system to another. Densification tolerance is the maximum
distance allowed before adding a new vertex. Distances are defined in terms of
the geographic data units (usually latitude and longitude).
A densification tolerance of 0 (the default) means that no extra vertices will be
added. Above 0, the higher the value, the longer the distance between new
vertices and therefore the fewer the new vertices. Smaller values (above 0) mean
more vertices are added – which takes more computer time but increases the
accuracy of the projection.
The default densification tolerance is set here, but can be over-ridden on a
project-by-project basis in the Project Properties dialog. This is because the
optimal densification tolerance depends on the scale of the data and the accuracy
required, which will vary between projects.
Maps
• Scale factor for copying and exporting maps. This setting
determines whether the map image is enlarged or shrunk when copying
a map to the clipboard. The default value is 1.0, which means the map
is copied at the size it is displayed. Choose a value larger than 1 to
enlarge the map and get a better quality image. Chose a value smaller
than 1 to shrink the map and get a smaller image.

Survey Design Preferences Tab

Design engine
• Echo commands to log. When this box is checked, the
commands issued to the design engine are written to the log tab of
the design or survey details window. This is helpful for the
developers for debugging purposes, but ordinarily should be
unchecked to save project file space.
• Time stamp log entries. When this box is checked, each log
file entry is accompanied by the time it occurred. This can be
useful for recording how long different operations took.

180 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Design details window - Coverage probability map.
• Symbol size. Allows you to change the size of the dots that
make up the grid points. This is largely useful for producing nice
looking maps to output.
• Number of classes. Determines the number of classes the
coverage probability is split into.
• Maximum and minimum values of classes.
• Setting the range of the classes to range from 0 – 1 can be
useful when you are comparing between maps, as they are then
all drawn on the same scale – however you will likely need
lots of classes as most points will have similar coverage
probabilities in many designs.
• Setting the range of the classes between the minimum and
maximum coverage for each design gives you the best chance
to spot patterns in coverage probability for each design, but
makes it harder to compare designs visually, as each will have
a different scale.
Browser windows
• Default columns for new browser sets. Clicking the Design
Browser… or Survey Browser… buttons opens the Column
Manager dialog, and allows you to select default columns for new
Design Browser or Survey browser sets.

Analysis Preferences Tab

Analysis Engines
• Automatically lock data sheet whenever an analysis is
run. Normally, you don’t want to change the data after starting
with your analyses. Checking this box prevents you from
inadvertently changing the data, because the data sheet is locked
automatically after the first analysis is run. For more information,
see Locking the Data Sheet, inChapter 7 of the Users Guide.
• Echo commands to log. When this box is checked, the
commands issued to the design engine are written to the log tab of
the design or survey details window. This is helpful for the
developers for debugging purposes, but ordinarily should be
unchecked to save project file space.
• Time stamp log entries. When this box is checked, each log
file entry is accompanied by the time it occurred. This can be
useful for recording how long different operations took.
• Capture command line output from CDS and MCDS
engines in WinNT. When this box is checked, any output
written to the command line when running MCDS.exe is captured
and loaded into the log window. Generally, this only occurs if the
engine crashes - saving this output can be useful for helping the
program authors to diagnose the source of the crash. The option is
selected by default, and only works in operating systems based on
Windows NT (i.e. NT4, 2000, XP and subsequent OSs). For more
about the output from program crashes, see MCDS Engine
Command Line Output in the MCDS engine reference appendix.
• Debug mode. When this box is checked, running an analysis
causes the temporary command files to be generated and placed in
the Windows temp folder, but the analysis engine is not run. The

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 181

 name and location of the command files is written to the Log tab,
and the status set to Warnings (amber). This option is useful when
debugging the analysis engine, or for making template command
files.
R Software
For more details about the link between the R statistical software and Distance,
see R Statistical Software in Chapter 7 of the Users Guide.
• Folder containing R. This contains the path to the R software.
If you have R installed on your machine, this path is automatically
read from the Windows registry the first time the Preferences
dialog is opened, or the first time an MRDS analysis is run. You
may want to update it when you install a new version of R – see
Updating the Version of R that Distance Uses.
• Properties of images generated by R. Allows you to change
various aspects of the image formatting and image file type.
Clicking Image Properties… opens the R Image Properties Dialog.
• Remove new objects that are created with each run.
When selected (the default) any new objects created during an
analysis run are removed from the R objects file (called “.RData”,
this is stored in the R folder, below the project data folder). This
keeps the R object file as compact as possible. However, under
some circumstances you may want to keep the R objects – for
example when they will be re-used in a subsequent analysis. For
more details, see Using a Previously Fitted Detection Function to
Estimate Density in MRDS in Chapter 10 of the Users Guide.
• Re-install analysis engine library on next run. When
selected, running an analysis that uses R (either the MRDS or DSM
engines) causes the appropriate R library to be re-installed from its
archive in the Distance program directory (file mrds.zip or dsm.zip)
into the current version of R before the analysis is started. This
option is only selected if the library has become corrupted
(unlikely), or if an updated version of the archive has been placed
in the Distance program directory (e.g., the program authors have
sent you a patched version of mrds.zip or dsm.zip). If selected, the
option is automatically de-selected after the next run of R once the
library has been re-installed.
Analysis Components window
• Opening Data Filter and Model Definitions. Both the Data
Filter and Model Definition properties dialogs have lots of tabs. In
many cases, you want to access the same tab repeatedly. To save
time, you have the option here to go directly to the tab that was
used last time you opened the dialog. Alternatively the dialog can
open on the first tab, which was the default in previous versions of
Distance.
Analysis Browser window
• Default columns for new browser sets. Clicking the
Analysis Browser… button opens the Column Manager dialog,
and allows you to select default columns for new Analysis Browser
sets.

182 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

Project Browser
The Project Browser is the main interface to Distance projects. It opens
automatically when a project is first opened, and is closed when the project is
closed. Its’ 6 tabs allow you to access the following parts of the interface:
• Data. The Data Explorer is the main interface to the data in your
project. For more information about data, see Chapter 5 - Data in
Distance in the Users Guide. For more about the Data Explorer,
see Data Explorer in the Program Reference.
• Maps. The Map Browser allows you to create and manage maps
of the geographic data in your project. If the project is not
geographic, this tab is disabled. For more about geographic data,
see Geographic (GIS) Data – Chapter 5 in the Users Guide. For
more about the Map Browser, see Map Browser in the Program
Reference.
• Designs. The Design Browser is for creating and managing
survey designs. For more information about survey design, see
Chapter 6 - Survey Design in Distance in the Users Guide. For
more about the Design Browser, see Design Browser in the
Program Reference.
• Surveys. The Survey Browser is for creating and managing
survey objects. Surveys are generated by designs (see Chapter 6 -
Survey Design in Distance in the Users Guide), and are also used
as part of the analysis specification (see Analysis Components –
Chapter 7 in the Users Guide). For more about the Survey
Browser, see Survey Browser in the Program Reference.
• Analyses. The Analysis Browser is for creating and managing
analyses. For more about setting up and running analyses, see
Chapter 7 - Analysis in Distance in the Users Guide. For more
about the Analysis Browser, see Analysis Browser in the Program
Reference.
• Simulations. The Simulation Browser will allow access to
the simulation capabilities of Distance – it is disabled at the
moment but should be implemented in a future version of the
software.

Data Explorer
The Data Explorer is the main interface for viewing and manipulating data in
Distance. You access it via the Data tab of the Project Browser.
An alternative data interface is the Data Entry Wizard, which can be accessed
from the Setup Project Wizard or by choosing Tools | Data Entry Wizard.
The Data Entry Wizard has a more restricted interface, and can only be used
with a simple data structure. For more information, see Data Entry Wizard in
the Program Reference.
You cannot use the Data Explorer effectively until you understand the way that
survey data is stored in Distance – make sure you’ve read the Chapter 5 - Data in
Distance in the Users Guide before you continue!
The Data Explorer is split into three sections, the Toolbar, the Data Layers
Viewer and the Data Sheet.
The Toolbar functions are summarized below. The other sections of the
Explorer are discussed on the next pages.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 183

 The Toolbar
Buttons that Change the Way the Data Sheet Looks
• Compact View. Shows a compact view of the data. This makes
the Data Sheet display only the ID and Label fields for the layers
that are not currently selected. Compare the following picture with
that for expanded view, below.

• Expanded view. Shows an expanded view of the data, with all

fields from all rows showing. In the example below we can see
that there are actually g0 and g0 SE fields in the Global data layer,
which were hidden in Compact view. Compact and Expanded
view are mutually exclusive.

• Show Field Types. Shows or hides the Data Type and

Modeling Type of each field. All of the examples above have
shown the Data Type and Modeling Types hidden. The following
example is with them shown. The first header row is the Data
Layer Name, the second is the Field Name, the third is the Field
Type, the fourth is the units and the fifth is the Source Database
Type:

The Data Sheet Lock/Unlock button

• Locks or unlocks the Data. When the data are locked, you
cannot change or delete any cells. See Locking the Data Sheet in
Chapter 7 of the Users Guide for more information.
Buttons for Managing Data Layers
• Create New Data Layer. Opens the Creates a new layer dialog,
allowing you to create a new data layer.
• Delete Current Data Layer. Deletes the currently selected data
layer.

If you delete a data layer, all data in the layer and in all child
layers are lost. There is no undo button!

• Data Layer Properties. Opens the Data Layer Properties dialog,

allowing you to view information about the layer.
Buttons for Adding and Deleting Fields.
• Insert Field Before Current. Inserts a new field before the
current field.
• Append Field After Current. Appends a new field after the
current field.
• Delete Current Field. Deletes the current field.

184 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Proceed with caution, as deleting a field can prevent a Distance
analysis from running properly!

Buttons for Editing the Data Sheet

For more details about the functions of these buttons, consult the pages on
Editing, Adding and Deleting Records in the Program Reference.
• Insert New Record Before Current. Inserts a new record before
the current record.
• Append Record After Current. Appends a new record after the
current record. Alternatively, you can press Ctrl+Enter or Ctrl-
Insert.
• Delete Current Record. Deletes the current record.
Alternatively, you can press Ctrl+Delete.

When you delete a row from a higher data layer then the
corresponding rows from lower data layers also get deleted!

If you want to know more about the Data Explorer, proceed to next page on the
Data Layers Viewer.

Data Layers Viewer

(For an overview of the data explorer, see the Program Reference page Data
Explorer.)

Example of the Data Layer Viewer. In this case the project has 4 data layers.

The Data Layer Viewer appears on the left in the Data Explorer. It presents a
hierarchical view of the data layers in your project (see Chapter 5 - Data in
Distance in the Users Guide for a discussion of the data layers). The icons by
the data layer names indicate the Data Layer Type (see List of Data Layer Types
in Chapter 5 of the Users Guide for a complete list).
Clicking on a data layer in the viewer shows the data for that layer in the Data
Sheet, as well as data for all higher layers. When the Data Explorer first opens,
only the Global Layer is displayed:

Part of the Data Explorer from the Stratify example project when first opened

If you click on the stratum data layer icon (in this case the stratum layer is called
“Region”), the stratum data appears beside the global data:

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 185

 Part of the Data Explorer from the Stratify example project after the Stratum icon in the
Data Layer Viewer has been clicked

You can compress the fields so that only the Label and ID column from the
Global layer appears by clicking on the button on the toolbar (see Toolbar,
above):

Same view as above, but with Compact View button clicked

Similarly, if you click on the sample data layer icon (in this case the sample data
layer is called “Line transect”), the sample data appears beside the stratum data.
Because the Compact View button is enabled, all fields in the Stratum data layer
(Region) except the label and ID fields disappear from view:

Part of the Data Explorer from the Stratify example project after the Sample icon in the
Data Layer Viewer has been clicked

Why go to all the trouble of selecting different data layers - why not just click on
the observation layer and open up the whole data sheet at once? When all the
data layers are open, the Data Sheet can become cluttered. It is often simpler, for
example, if you want to see how many strata there are, to only open the strata
data layer, and leave the sample and observation layer hidden.

186 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 When focus is on the Data Layer Viewer, you can use the arrow keys
to move up and down the data layers, rather than clicking.

If you want to learn more about the Data Explorer, you should read the next
page, which is about the Data Sheet.

Data Sheet
(For an overview of the data explorer, see the Program Reference page Data
Explorer.)
The Data Sheet is intended to be as intuitive to use as any spreadsheet grid.
However, the hierarchical nature of the data (observations within transects
within strata within global) imposes some restrictions. In the following three
pages, we describe how to perform the common tasks associated with
manipulating survey data in the Data Sheet:
• Navigating Around the Data Sheet
• Editing, Adding, and Deleting Records
• Editing, Adding and Deleting Fields

You can easily copy your data from the data sheet to a spreadsheet or
database package. Click on the Data Explorer to give it focus and then click on
the Copy to Clipboard button on the main toolbar, or choose the Data
Explorer menu item Copy Data to Clipboard. In your spreadsheet or
database, choose Paste. You can use this Copy to Clipboard facility, together
with the Data Import Wizard to provide a crude Import/Export facility for your
data.

Sometimes you may run into problems pasting the data into your target package.
This is usually caused by the symbol used by Distance to signify an end-of-row
– you can change this in the General tab of the Preferences window.

Navigating the Data Sheet

(For an overview of the data explorer, see the Program Reference page Data
Explorer.)
You can move around the Data Sheet by clicking on the grid and/or using the
Up, Down, Left, Right Arrows as well as the Home, End, Page Down , Page Up
and Tab cursor keys. Holding down Ctrl and a cursor key simultaneously, allows
you to move in larger steps in that direction.

You can also move around by clicking on the grid and then moving
the mouse while holding the left mouse button down. In this mode, if you move
past the end of the visible grid, it scrolls for you.

Setting Data Sheet Column Widths

The data sheet columns widths are set automatically when the grid is first shown,
based on the contents of the column. For all columns except ID, the width can be
changed by clicking and dragging the grid lines in the column headers. Distance
will then remember the column width you have set.
To find out more about the Data Explorer, consult the next page about Editing,
Adding and Deleting Records.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 187

 Editing, Adding and Deleting Records
(For an overview of the data explorer, see the Program Reference page Data
Explorer.)
Editing Records
The data grid has a distinct edit mode. The edit mode is entered by a mouse click
or a key press. To replace the current contents of a cell, simply begin typing. To
edit the contents, double click on the cell. The Left and Right Arrow keys are
used within cells to move between individual characters and digits, and Home
and End takes you to the beginning and end. Ctrl+Left and Right takes you to
the beginning and end of the next word in the cell, while Shift+Left and Right
selects part of the cell. Use Delete and Backspace keys as normal within the cell.
To exit edit mode, click on another part of the cell, use the Up or Down Arrow
keys to get out, use the Left and Right keys to go past the contents of the cell, or
use Page Up and Page Down.
Pressing Esc leaves edit mode without entering the data, returning the cell to its
original state. Note you can also cut and paste into cells using the usual
Windows keys, or right clicking the mouse button in edit mode. To delete the
contents of a cell, press Delete (you will not be allowed to do this if the cell is
required to contain a value).
You cannot edit the ID field, nor can you edit rows without an ID field (see
Adding Records, below). Shape fields (in geographic projects only) are also
special – if you double-click on a record in a shape field, the Shape Properties
dialog appears.

You can tell whether you can edit a cell because the Focus rectangle
(dashed box that shows you which cell is currently selected) turns from light
dashes to heavy dashes when you can edit the cell.

When you enter data into a field some data validation takes place to
check that the data is of the correct type. This prevents you, for example, from
entering text into a decimal or integer field, or entering decimal points into an
integer field. Most cells must contain some value and cannot be left blank.

When you are in edit mode, the cell behaves just like any other text
box. For example, you can right-click to bring up a pop-up menu containing
useful commands such as Cut and Paste. You can also use the usual keyboard
shortcuts (Ctrl-Insert, Shift-Insert and Shift-Del for Copy, Paste and Cut).

Shift+Enter takes you out of edit mode and on to the first field of the
next record for that layer. Its designed to help with data entry - try it!

Adding Records
The Data Sheet is a representation of the underlying database, not a simple
spreadsheet. Therefore, you cannot edit a cell in the Data Sheet unless there is a
corresponding record in the database. For example, consider the following data
sheet:

188 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 The focus is on the observation cell that is below the record with ID 3 (you can
tell this because there is a light focus rectangle on that cell). However, this cell
has no ID in the ID field. This means that there is no record for this cell in the
database. So, before you can enter the distance for this observation, you must
add a record. You do this by clicking the Insert Record or Append Record
buttons, or by typing the keyboard shortcut for Append Record, which is
Ctrl+Enter. Insert Record puts a record before the current one, while Append
Record puts one after the current one. In this case, since there is no record in the
cell, they both have the same effect:

A record has been inserted with ID of 4. Notice that the focus rectangle is now
heavy too, indicating that the value in the cell is editable. We enter the value
0.53 and hit Enter:

Now we wish to append another record, so we type Ctrl+Enter (the keyboard

shortcut for Append record):

Notice that the value of the record in the cell that previously had focus has been
copied into the new record. We can now enter the next value for distance, 1.98
and hit Enter:

We could continue this process until all the observations have been entered.

If no objects were seen on a transect, then we do not add any records

in the Observation data layer for that transect. For example, if there had been 3
objects seen on line 1, none on line 2 and 4 on line 3 then the completed data
sheet would look like this:

Adding Multiple Records at Once

Often it is more convenient to add more than one record at once. For example, in
the above scenario we may know that there were 10 objects sighted on Line 2.
We may wish to add all 10 records at once, and then type in the values for the
distances. We can do this by double clicking on the ID field to bring up a

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 189

 special window called the multi-record add dialog. This window allows you to
add up to 99 rows at once.
Starting from the situation

we double click on the ID field, bringing up the multi-record add dialog:

In the text field, we type in 10: , and then click on either of the two
buttons to the right of the up and down arrows (again, because there are no
records associated with the cell, it doesn't matter which we press). Voilla!:

We are now ready to enter the 10 values.

The multi-record add facility is particularly useful when you have

interval (binned) data. Say you saw 50 objects in the first bin, which is between
0 and 10 meters. Insert a single record for the first object, and type in a distance
of 5 meters (mid-way between the cutpoints). Then double-click on the ID field
to bring up the multi-record add dialog. Enter 49 and press append. You one
record gets copied into 49 new records, so now you have the 50 you want. See
Interval (Binned/Grouped) Data in Chapter 8 of the Users Guide page on Interval
Data for more details about how to enter and analyze interval data in distance.

This example used the Observation data layer, but of course it could have been
done with any other layer.
Deleting Records
To delete a record, highlight the record in the Data Sheet and press the Delete
Current Record button. You can only delete one record at once.

If you delete a record from the sample data layer, all of the
observations associated with that sample will be deleted. Similarly, if you delete
a record from the transect data layer, all of the samples and observations will be
deleted. Remember, there is no undo button, so you may want to consider
Backing up first!

To find out more about the Data Explorer, consult the last page in this chapter,
Editing, Adding and Deleting Fields.

190 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Editing, Adding and Deleting Fields
In many cases, you won't need to change the fields that Distance provides by
default. However, there are a number of circumstances in which you may want
to edit, add or delete fields – some of these are outlined in the relevant sections.
Bear in mind if you are setting up the project to analyze data that it is usually
best to make any changes to the fields before you start the analysis phase.
Editing Field Names
Distance provides default names for all the fields, but you may wish to change
them. To edit a field name double-click on it in Data Sheet. You can also edit
the Data Layer Names by double-clicking on them. If you edit the field names
after you have created any Data Filters or Model Definitions, the results can be
unpredictable!
Editing Units
To change the units of a field, double click on the 4th row of the data sheet
header for the field you want to change, and choose from the list of field types.
n/a means “not applicable” – i.e., no units are required (e.g., the cluster size field
doesn’t require any units).
Moving Fields
You can change the order in which the fields are presented in the data sheet by
using the mouse to drag the field name to a new location.
For example, the default order of fields in the observation layer for studies where
objects are clusters is Perp Distance and then Cluster size. To swap the two
fields, click on the Perp Distance column header with the left mouse button, hold
the mouse down, move the mouse over the Cluster size column and release the
mouse. As you drag the mouse you should see the cursor change from an arrow
the drag pointer:

Adding Fields
There are many reasons for adding fields to the Distance database, beyond those
provided by default. You may want to add extra multiplier fields in the global
layer (see Multipliers in CDS Analysis in Chapter 8 of the Users Guide). You
may want to add a field that will be used for post-stratification (see Chapter 8 of
the Users Guide:Stratification and Post-stratification). You may want to add a
column that defines a subset of the data that you will use to select data in a Data
Filter, such as different species or years of data (see the Data Selection Tab of
the Data Filter entry in the Program Reference). You may want to add fields for
covariates in MCDS analyses, or you may be setting up a complicated data
structure by hand.
To add a field to the data sheet, click on any cell in the appropriate data layer to
give that layer focus. The click on either the Insert Field or Append Field
button. Insert Field puts the new field where the current field is, and moves the
current field to the right. Append Field places the new field after the current
one.
After clicking the button, a small window will appear prompting you for the new
field's name, Field Type and Units.
New fields are automatically filled with default values in the Data Sheet.
Deleting Fields

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 191

 You may want to delete a field if, for example, you have defined a multiplier that
you don't need any more, and is not being used by any Model Definitions.
Another example is during survey design if you have deleted a design and want
to remove the coverage probability field from the coverage layer. To delete a
field, click on any cell in that field and click the Delete Field button . Distance
will issue a standard warning, and will delete the field if you press OK.

Map Browser
The Map Browser allows you to create, sort, rename, delete and preview maps of
the geographic data in your Distance project. It is accessed via the Maps tab of
the Project Browser. The Map Browser is only accessible if the project is
geographic. For more information about geographic data in Distance, see
Geographic (GIS) Data in Chapter 5 of the Users Guide.
The Map Browser comprises a table showing a list of the maps that have been
created, and optionally a preview pane, which gives a preview of the map that is
currently selected. You can change the size of the preview pane by dragging the
bar between the map table and preview pane.

For maps containing many shapes, the preview pane can take a while
to draw. In these cases, it’s better to hide the preview pane (click the button).

Because of a limitation in the GIS engine Distance uses, the preview

pane is blanked out (together with all other map windows) if any geographic data
is changed.

To create a new map, click the button, or choose Maps | New Map. To
rename the map, double click on the name in the map table, and type the new
name. To view the map, so you can add layers, etc, select the map in the map
table and click , or choose Maps | View Map.
Toolbar

• Show Preview Pane. Opens the preview pane on the right

hand side of the Map Browser.
• Hide Preview Pane. Closes the preview pane.
• Refresh Preview. Causes the map in the preview pane to be
refreshed, if it has become out of date (for example because a data
layer has been deleted).
• New Map. Create a new map.
• Delete Map. Delete the currently selected map.
• View Map. Open the Map window for the currently selected
map.
• Up. Move the currently selected map up in the map table.
• Down. Move the currently selected map down in the map
table.

192 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

Design Browser
The Design Browser is the launching pad for survey design in Distance. From
here you can create and run designs, arrange and sort them and view summaries
of the results.
Before starting the survey design process, you should read Chapter 6 - Survey
Design in Distance of the Users Guide.
Overview
The Design Browser is laid out like a spreadsheet, with one row for each design,
and columns that give you useful information about the designs. The window is
split into two panes. On the left there are columns of summary information about
the design inputs: the status , unique ID number, design name, time created
and time run. On the right are columns summarizing the results (which will be
blank if the design has not been run yet) the actual columns showing can be
customized using the Column Manager (press the button).

You can resize the panes by dragging the bar that divides them.

If you hold your mouse over a column header for a few moments, a
small window pops up giving you an explanation of that column. This also
works if you hold your mouse over a survey, data filter or model definition
number: a window pops up giving you the name that corresponds with the
number.

Designs can be grouped into Design Sets. A Design Set is a group of related
designs – you are free to create, delete and rename sets, and choose which
designs to group together. The current set name is listed after the word “Set:” on
the design browser toolbar, and you can access a list of sets by clicking on the
down arrow beside the current set name. You can create, delete and move sets
using the buttons to the right of the current set name.

Transferring results to another application, such as a word processor is

easy. Press the Copy to Clipboard button on the main toolbar or choose the
Designs | Copy Set to Clipboard. This copies the contents of the current
Design Set. In your word processor or spreadsheet, choose the Paste button.

Toolbar, and Designs menu

Set:
• Set Name. Gives the name of the current design set. Click on the
name to edit it. Click on the drop-down arrow to get a list of other
sets, from where you can click on another set to display its’
contents.
• New Set. Creates a new Design Set.
• Delete Set. Deletes the current Design Set and all designs in it.
• Arrange Sets. Opens the Arrange Sets dialog, from where you
can change the order that sets appear in the drop-down list of sets.
Design:
• New Design. Creates a new design.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 193

 The new design is based on the one that is currently selected in the
Design Browser.

• Delete Design. Deletes the selected designs.

• Design Details. Opens Design Details windows for the selected
designs
• Run Design. Runs the selected designs. For examples, see
Example 3 - Automated Generation of New Surveys and Example
3 - Design Statistics in Chapter 3 of the Users Guide.
• Reset Design. Resets the selected designs. For designs that have
been run, this deletes their Log and Results and returns the status to
Grey (not run). For designs that are currently running, this cancels
the run.

Sometimes it is useful to cancel an design while it is running - for

example you may have set a very long coverage probability simulation running
by mistake!

• Move Design. Moves the selected designs to another set. You

are prompted for a set to move the designs to.
• Arrange Columns. This opens the Column Manager dialog for
the current Design Set – see Column Manager Dialog in the
Program Reference for more details.
• Copy Set to Clipboard (menu only). Copies the current set to
the clipboard, from where it can be pasted into word processors,
spreadsheets, etc.
• Preferences (menu only). Opens the Preferences dialog on the
Survey Design page.

For the New Design, Delete Design and Design Details buttons, you
can work with more than one design at once by highlighting multiple designs in
the browser. To highlight more than one design, either:
(i) Hold the Ctrl key down and click on each design to highlight them.
(ii) Hold the Shift key and click on two non-adjacent designs to select all designs
in between them.
(iii) Hold your mouse button down and move it over the designs you want to
highlight, if they are adjacent.
(iv) Hold the Shift key and use the up or down keys to extend the current
highlighting.

A shortcut way of opening the Design Details for a design is to double

click on the status button in the right hand pane of the browser.

Sorting designs
To sort your design by any column, just click on the column header. One click
sorts the column in ascending order, while another clicks sort in descending
order. A little red arrow tells you which column is currently being used as the
sort column, and whether it is an ascending or descending sort.

194 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

Survey Browser
The Survey Browser is the interface for managing survey objects in Distance.
From here you can create and run surveys, arrange and sort them and view
summaries of the results. Survey objects are used for two purposes: they are
created from designs (as described in Chapter 6 - Survey Design in Distance in
the Users Guide), and they are a component of an analysis (as described in
Chapter 7 - Analysis in Distance of the Users Guide).
Overview
The Survey Browser is laid out like a spreadsheet, with one row for each survey,
and columns that give you useful information about the surveys. The window is
split into two panes. On the left there are columns of summary information about
the survey inputs: the status , unique ID number, designID, survey name, time
created and time run. On the right are columns summarizing the results (which
will be blank if the survey has not been run yet) the actual columns showing can
be customized using the Column Manager (press the button).
Surveys created from a design have a design number, and their status will be run
(green), warnings (amber), or errors (red). Surveys used for analysis are usually
not created from a design, and have no design number and status not run (grey).
The right-hand results pane is therefore only useful for surveys created from a
design.

You can resize the panes by dragging the bar that divides them.

If you hold your mouse over a column header for a few moments, a
small window pops up giving you an explanation of that column. This also
works if you hold your mouse over a survey, data filter or model definition
number: a window pops up giving you the name that corresponds with the
number.

Surveys can be grouped into Survey Sets. A Survey Set is a group of related
surveys – you are free to create, delete and rename sets, and choose which
surveys to group together. The current set name is listed after the word “Set:” on
the survey browser toolbar, and you can access a list of sets by clicking on the
down arrow beside the current set name. You can create, delete and move sets
using the buttons to the right of the current set name.

Transferring results to another application, such as a word processor is

easy. Press the Copy to Clipboard button on the main toolbar or choose the
Surveys | Copy Set to Clipboard. This copies the contents of the current
Survey Set. In your word processor or spreadsheet, choose the Paste button.

Toolbar, and Surveys menu

Set:
• Set name. Gives the name of the current survey set. Click on the
name to edit it. Click on the drop-down arrow to get a list of other
sets, from where you can click on another set to display its’
contents.
• New set. Creates a new Survey Set.
• Delete set. Deletes the current Survey Set and all surveys in it.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 195

 • Arrange sets. Opens the Arrange Sets dialog, from where you
can change the order that sets appear in the drop-down list of sets.
Survey:
• New Survey. Creates a new survey.

The new survey is based on the one that is currently selected in the
Survey Browser.

• Delete Survey. Deletes the selected surveys.

• Survey Details. Opens Survey Details windows for the selected
surveys
• Run Survey. Runs the selected surveys. Only surveys associated
with a design can be run – see Chapter 6 - Survey Design in
Distance in the Users Guide for more information.
• Reset Survey. Resets the selected surveys. For surveys that have
been run, this deletes their Log and Results and returns the status to
Grey (not run). For surveys that are currently running, this cancels
the run.
• Move Survey. Moves the selected surveys to another set. You
are prompted for the set to move the surveys to.
• Arrange Columns. This opens the Column Manager dialog for
the current Survey Set – see Column Manager Dialog in the
Program Reference for more details.
• Copy Set to Clipboard (menu only). Copies the current set to
the clipboard, from where it can be pasted into word processors,
spreadsheets, etc.
• Preferences (menu only). Opens the Preferences dialog on the
Survey Design page.

For the New Survey, Delete Survey and Survey Details buttons, you
can work with more than one survey at once by highlighting multiple surveys in
the browser. To highlight more than one survey, either:
(i) Hold the Ctrl key down and click on each survey to highlight them.
(ii) Hold the Shift key and click on two non-adjacent surveys to select all
surveys in between them.
(iii) Hold your mouse button down and move it over the surveys you want to
highlight, if they are adjacent.
(iv) Hold the Shift key and use the up or down keys to extend the current
highlighting.

A shortcut way of opening the Survey Details for a survey is to double

click on the status button in the right hand pane of the browser.

Sorting surveys
To sort your survey by any column, just click on the column header. One click
sorts the column in ascending order, while another clicks sort in descending
order. A little red arrow tells you which column is currently being used as the
sort column, and whether it is an ascending or descending sort.

196 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

Analysis Browser
The Analysis Browser is the launching pad for data analysis in Distance. From
here you can create and run analyses, arrange and sort them and view summaries
of the results.
Before starting your analyses, you should read Chapter 7 - Analysis in Distance
in the Users Guide.
Overview
The Analysis Browser is laid out like a spreadsheet, with one row for each
analysis, and columns that give you useful information about the analyses. The
window is split into two panes. On the left there are columns of summary
information about the analysis inputs: the status , unique ID number, survey
number, data filter number, model definition number, analysis name, time
created and time run. On the right are columns summarizing the results (which
will be blank if the analysis has not been run yet) the actual columns showing
can be customized using the Column Manager (press the button). Some
additional notes on these columns are given under CDS Analysis Browser
Results in Chapter 8 of the Users Guide.

You can resize the panes by dragging the bar that divides them.

If you hold your mouse over a column header for a few moments, a
small window pops up giving you an explanation of that column. This also
works if you hold your mouse over a survey, data filter or model definition
number: a window pops up giving you the name that corresponds with the
number.

Analyses can be grouped into Analysis Sets. An Analysis Set is a group of

related analyses – you are free to create, delete and rename sets, and choose
which analyses to group together. The current set name is listed after the word
“Set:” on the analysis browser toolbar, and you can access a list of sets by
clicking on the down arrow beside the current set name. You can create, delete
and move sets using the buttons to the right of the current set name.

Transferring results to another application, such as a word processor is

easy. Press the Copy to Clipboard button on the main toolbar or choose the
Analyses | Copy Set to Clipboard. This copies the contents of the current
Analysis Set. In your word processor or spreadsheet, choose the Paste button.

Toolbar, and Analyses menu

Set:
• Set name. Gives the name of the current analysis set. Click on
the name to edit it. Click on the drop-down arrow to get a list of
other sets, from where you can click on another set to display its’
contents.
• New set. Creates a new Analysis Set.
• Delete set. Deletes the current Analysis Set and all analyses in it.
• Arrange sets. Opens the Arrange Sets dialog, from where you
can change the order that sets appear in the drop-down list of sets.
Analysis:

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 197

 • New Analysis. Creates a new analysis.

The new analysis is based on the one that is currently selected in the
Analysis Browser (Half-normal / hermite in the picture above - when more than
one are selected, look for the dashed focus rectangle around the one that has
focus)

• Delete Analysis. Deletes the selected analyses.

• Analysis Details. Opens Analysis Details windows for the
selected analyses
• Run Analysis. Runs the selected analyses. See the Running
Analyses page in Chapter 7 of the Users guide for more
information and tips.
• Reset Analysis. Resets the selected analyses. For analyses that
have been run, this deletes their Log and Results and returns the
status to Grey (not run). For analyses that are currently running,
this cancels the run.

Sometimes it is useful to cancel an analysis while it is running - for

example you may have set a large bootstrap analysis running by mistake!

On some systems, clicking on the Reset button while an analysis

is running appears to stop the analysis, but it carries on running in the
background, eating up system resources. For more about this, see Stopping an
Analysis in Chapter 10 of the Users Guide.

• Move Analysis. Moves the selected analyses to another set. You

are prompted for the set to move the analyses to.
• Arrange Columns. This opens the Column Manager dialog for
the current Analysis Set – see Column Manager for more details.
• Copy Set to Clipboard (menu only). Copies the current set to
the clipboard, from where it can be pasted into word processors,
spreadsheets, etc.
• Preferences (menu only). Opens the Preferences dialog on the
Analysis page.

For the New Analysis, Delete Analysis and Analysis Details buttons,
you can work with more than one analysis at once by highlighting multiple
analyses in the browser. To highlight more than one analysis, either:
(i) Hold the Ctrl key down and click on each analysis to highlight them.
(ii) Hold the Shift key and click on two non-adjacent analyses to select all
analyses in between them.
(iii) Hold your mouse button down and move it over the analyses you want to
highlight, if they are adjacent.
(iv) Hold the Shift key and use the up or down keys to extend the current
highlighting.

A shortcut way of opening the Analysis Details for an analysis is to

double click on the status button in the right hand pane of the browser.

198 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Sorting analyses
To sort your analysis by any column, just click on the column header. One click
sorts the column in ascending order, while another clicks sort in descending
order. A little red arrow tells you which column is currently being used as the
sort column, and whether it is an ascending or descending sort.

Some columns, such as AIC and Delta AIC are special. When you
click on these columns, your analyses are sorted by Data Filter first, and then
AIC or Delta AIC column second. Why? AIC stands for Akaike's Information
Criterion - an index of the relative fit of competing statistical models. The lower
the AIC, the more parsimonious the model (other things being equal – look for
AIC in the Distance Book). Delta AIC is the difference in AIC between the
model with the lowest AIC and the current model. However, it only makes sense
to compare AIC values, calculate Delta AIC and rank the analyses based on
models fit to the same data. Sorting by the data filter column first ensures that
this happens.

AIC is not the only model selection criterion that can be used. Distance also
provides the following columns: AICc (“corrected” AIC), BIC (Bayes
Information Criterion) and LogL (the log-likelihood). For more information
about these criteria, and model selection in general, see Burnham and Anderson
(2002).

Map Window
Map windows provide a view of the geographic data layers in a project. You can
customize the map by choosing which data layers to include, and by panning and
zooming around the map area. Any changes you make to a map are saved when
you close the map.
You create maps in the Map Browser (see Program Reference, Map Browser for
details). From there, click on the View Map button to open the Map window.
You can have more than one map open at a time.
The map window is split into two panes. On the left is a pane containing map
tools (currently just the layer control), and on the right is the map itself.

Several features of the map window are not yet implemented. These
include: the Info, Find and Spatial select map tools; the ability to customize the
properties of each data layer, such as its colour, and add legends. We expect to
implement these features in future releases.

Layer control
The layer control displays a list of the data layers currently shown on the map,
with a legend showing the symbol used to display shapes on that layer. There is
a tick box where you can turn off display of that layer. You can change the
ordering of layers by clicking on a layer and dragging it above or below another
layer.
Map tips
Map tips is a popup window that appears when you hover over a map feature,
giving information about the feature. To enable map tips, click on the Map
Tips button on the toolbar. A second row of tools opens on the toolbar,
prompting you for the Map Tip Layer and Field. Select from the list of layers
and fields, and then position the cursor over a feature on the map. The value of
the selected field in that position will appear. For example, if you select a

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 199

 sample layer “Line transect” and the “Label” field, then position the cursor over
a particular transect, the map tip will display the label (name) of that transect.

The use of map tips can significantly slow down the map display, so
turn them off when you’re finished!

Toolbar and Map menu

Tools in left-hand pane
• Layer control. Depressing this button displays a list of the layers
in the map.
• Info. Not currently implemented.
• Find. Not currently implemented.
• Spatial Select. Not currently implemented.
About the map
• Map Properties. Opens the Map Properties dialog.
• Refresh Map. Refreshes the map – useful if some features have
been edited so the map is out of date.
Manipulating layers
• Add Layer. Opens the Add Layer dialog, prompting you to choose the
layer to add.
• Remove Layer. Removes the currently selected layer from the map.
• Layer Properties. Not currently implemented. This will allow you
to customize the look of the data layer – for example the colour of the
symbols.
• Remove All Layers. Removes all layers from the map.
Pan/zoom tools
• Full Extent. Zooms out to show the entire map area.
• Zoom In. When you click this button the cursor change to a
“zoom in” symbol. Select on the map the area to zoom in to. To
return the cursor to normal, click on the button again.
• Zoom Out. When you click this button the cursor change to a
“zoom out” symbol. Click on the map to zoom out. To return the
cursor to normal, click on the button again.
• Pan. When you click this button the cursor change to a “pan”
symbol (a hand). Click on the map, and drag to move the display
to another area. To return the cursor to normal, click on the button
again.
Others
• Map Tips (toolbar only). select this button to turn on the display
of map tips (see above for details).
• Preferences… (menu only). Opens the Preferences dialog at the
Geographic tab.
• Close (menu only). Closes the map.

200 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

Design Details Window
This window is the main interface to each individual design. From here you can
change the inputs, run the design to estimate probability of coverage or generate
surveys, view the log file and the results pages. For more background about
survey design, see Chapter 6 - Survey Design in Distance of the Users Guide.
To open a Design Details window, select the design in the Design Browser and
click the Design Details button [picture here]. You can open up more than one
design details window at once by selecting multiple designs before clicking the
button. This could be useful if, for example, you want to view the results from
two designs side-by-side. You can resize the design details window to fit your
requirements by dragging the edges of the window (although there is a fixed
minimum size). The last window you close sets the size for the next one to
open.

A shortcut method of opening an Design Details window from the

Design Details is to double click on an design’s status button.

The Design Details window is divided into three tabs: Inputs, Log and Results.
The tab that Distance first displays when you open an Design Details window
depends on the status of the design. For designs that have not been run (grey
status light in Design Browser), it opens the Inputs window. For designs that ran
with warnings or errors (amber or red), it opens the Log tab. For designs than
ran OK (green) it opens the Results.

Design Details Inputs Tab

Use the inputs tab to set up and run your design. It is divided into three sections:
Design, Type of Design, and Comments.
Design
This section gives you some information about the design, such as the name,
time created and time last run (i.e., time coverage probability estimated).
To change the name of the design, simply type a new name into the box labeled
Name. Once you have typed the new name, hit Enter or click somewhere else on
the window to apply the name to the design.
To run a design, click the Run button. This opens the Run Design dialog, where
you are given the choice of either estimating coverage probability or creating a
new survey. You cannot run a design until you have: (i) created a coverage
probability grid in the project; (ii) selected the type of design; (iii) set the
properties for the type of design. For more about the process, see Chapter 6 -
Survey Design in Distance of the Users Guide.
Type of Design
To select the type of design, first choose the type of sampler (Line or point), and
then the class of design. A list of all design types is given in Chapter 6 of the
Users guide entitled Design Classes Available in Distance. A description and
picture is given of the chosen class.
Before you run a design, you need to set the design properties, which you do by
clicking on the Properties… button. This takes you to the Design Properties
dialog.
Comments
The comments section is there for you to type some comments to yourself about
the current design. For example, you might want to remind yourself of why you

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 201

 chose to use these input parameters. The same section appears in the Results tab,
so you can make comments about your results too.

You can give yourself more room by resizing the comments section.
Put your mouse just above the Comments section header and dragging the
section up and down. You may want to increase the height of the whole Design
Details window (by dragging on its border) before you do this.

Design Details Log Tab

The Log tab is pretty-much identical for Design and Survey details, and both of
them are covered here. This tab allows you to check any warnings or errors that
occurred when you ran a design or generated a survey. Some messages in the log
just report on the general progress of the design or survey run, thus if a warning
or error does occur there is some indication of where in the sequence of the run
this took place. Warning messages are coloured amber, and error messages red.
Warnings may indicate some problem with effort allocation, the definition of the
stratum, or the sampler properties. Some warnings may also occur due to
problems with the GIS component or due to problems with some geometric
calculations for certain survey regions. This may possibly lead to problems
calculating the design properties (which may then be invalid or very
approximate) or generating a design (only part of a survey plan may be created).
For example, a warning appears if no effort is allocated to any of the survey
strata. If an incorrect stratum definition leads to the stratum having zero surface
area, this is reported as a warning. If the sampler width or radius is large relative
to the size of the survey region, then some coverage probability grid points will
be covered by more than one sampler during a survey simulation. This may lead
to coverage probability values greater than one. A warning letting you know that
the coverage probabilities were constrained to fall within the [0,1] range will be
displayed in this case.
Errors appear when an event occurs that completely thwarts the attempt to
calculate design properties or generate an instance of a design. Such events occur
when:
• An invalid description of the design is given.
• The project database cannot be accessed.
• The coverage probability grid layer cannot be located or the
associated database table is missing.
• The coverage probability field name is invalid or cannot be
validated.
• The temporary coverage probability grid layer cannot be created
(the temporary layer is produced by projecting the original layer to
the design coordinate system .
• The stratum layer cannot be located, it contains no strata, or the
strata are incorrectly defined.
• The sample layer cannot be created or set up.
• The convex regions (required for the generation of some designs)
data layer cannot be created or set up.
• The design axis coordinates used to orientate some designs are
incorrectly defined.
• Problems with the GIS component or with some geometric
calculations for certain survey regions lead to invalid designs.

202 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 • Too little effort has been allocated to a survey region to allow a
particular design to be generated.

Design Details Results Tab

Header Information on the Design/Survey Results Tab
The Results tab for both designs and surveys displays some header information
describing the design. Firstly, the design and sampler class are displayed. For
surveys the name of the sampler layer is also shown. Some general properties to
do with effort allocation are listed. The results show whether the coverage
probabilities for a design run are simulation calculated or assumed even, and also
which coverage grid is used, and the field where the coverage probability values
are stored. For coverage probabilities estimated by simulation, the number of
simulations is shown. A description of the stratum layers coordinate system type
as well as that of the design is given. If the design coordinate system is
projected, then the type of projection and its associated units are also given. The
seed value used to initialize the random number generator (RNG) is also shown.

Coverage Probability Information on the Design Results

Tab
If you selected the option that assumes the coverage probabilities to be even ,
approximate estimates of coverage probability are calculated analytically. For
each stratum in the stratum layer the proportion of the sampled area relative to
the survey region surface area is displayed. This proportion does not take
sampler overlap into account, or that parts of the sampler may fall outside the
study area along the region boundary.
If you opted to estimate the coverage probabilities by simulation, the design is
created for the number of repetitions you specified. For each stratum in the
stratum layer the minimum, maximum, mean, and standard deviation of the
number of times each point in the coverage point grid is hit by the point or line
sampler is shown. This is followed by the minimum, maximum, mean, and
standard deviation of the coverage probability at each point in the grid. If the
sampling intensity is very low the coverage probability values may be very
small. If any of the coverage probability statistics are less than 0.001 then “<
0.001” is displayed instead of the value. It is always possible to retrieve the
coverage probability values from the field in the coverage layer in which they
are stored. You should aim for a design whose coverage probabilities are as even
as possible, and the minimum number of hits for any design should be greater
than zero.
To calculate coverage probabilities by simulation for designs using lines, each
segment making up the sampler line is enclosed in a rectangle whose width is the
same as that of the sampler, Similarly, for point designs, each point sampler is
enclosed in a circle whose radius is the same as that associated with the point.

Design Property Information on the Design Results Tab

For each stratum in the survey layer the following general properties are
displayed on the design Results tab:
• The number of point or line samplers that were specified on the
effort allocation page and are expected to be generated.
• The samplers’ radius or half-width.
• The expected sampler area coverage, which is the surface area
covered by the samplers (generally less than will be realized in a
survey).

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 203

 • The surface area of the stratum and the expected proportion of the
stratum covered by the samplers (based on the expected sampler
area coverage).

Simple Random Point Sampling - Results Tab

The Results tab for both designs and surveys displays some header information
for all survey designs.

Survey Plan Results

For each stratum in the survey layer the following are displayed:
• The number of point samplers that were specified on the effort
allocation page and are expected to be generated.
• The actual number of point samplers generated and the associated
sampler radius
The expected sampler area coverage, which is the surface area covered by the
sampler points. Each point sampler line is enclosed in a circle whose radius is
the same as that associated with the point. The area intersection of the circle is
used in calculating the realized sampler area coverage. As the circles may fall
partly outside the survey region the realized sampler area coverage is generally
less the expected value. The potential overlap between the uniformly distributed
point samplers is not taken into account when calculating the realized sampler
area coverage.
The surface area of the stratum and the proportion of the stratum covered by the
samplers.
Design Class Results
The design Results tab displays some general design properties and coverage
probability information for all survey designs.

Systematic Point Grid Sampling - Results Tab

The Results tab for both designs and surveys displays some header information
for all survey designs. For each stratum in the survey layer the following are
displayed:
Survey Plan Results
For each stratum in the survey layer the following are displayed:
• The approximated number of point samplers displayed on the effort
allocation page. This may differ from the actual number generated,
as the points are generated according to the spacing specified for
the systematic regular grid of sampler points.
• The actual number of point samplers generated and the associated
sampler radius.
• The spacing between the systematic grid of points in the vertical
and horizontal direction.
• The angle of the systematic point grid with respect to the x-axis,
measured in an anti-clockwise direction from the positive x-axis.
• The expected sampler area coverage, which is the surface area
covered by the sampler points. Each point sampler line is enclosed
in a circle whose radius is the same as that associated with the
point. The area intersection of the circle is used in calculating the
realized sampler area coverage. As the circles may fall partly
outside the survey region the realized sampler area coverage is
generally less the expected value. The potential overlap between

204 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 the systematically distributed point samplers is not taken into
account when calculating the realized sampler area coverage.
• The surface area of the stratum and the proportion of the stratum
covered by the samplers.
Design Class Results
The design Results tab displays some general design properties and coverage
probability information for all survey designs.

Parallel Random Line Sampling - Results Tab

The Results tab for both designs and surveys displays some header information
for all survey designs. For each stratum in the survey layer the following are
displayed:
Survey Plan Results
For each stratum in the survey layer the following are displayed:
• The number of line samplers that were specified on the effort
allocation page and are expected to be generated. This number may
differ from the number actually generated if the effort allocation is
determined by line length rather than the number of lines. In this
case the number displayed on the effort allocation page is an
approximation.
• The actual number of line samplers generated and the associated
sampler half-width.
• The total aggregated sampler length.
• The angle of the sampler lines with respect to the x-axis, measured
in an anti-clockwise direction from the positive x-axis.
• The total length of the trackline, including distance spent off-effort
moving from the end of one sampler to the beginning of the next
one. Total cyclic trackline length includes the extra distance
required to return from the end of the last sampler to the beginning
of the first sampler.
• The expected sampler area coverage, which is the surface area
covered by the sampler lines. Each sampler line is enclosed in a
rectangle whose width is the same as that of the sampler. The area
intersection of the rectangle is used in calculating the realized
sampler area coverage. As the rectangles may fall partly outside the
survey region the realized sampler area coverage is generally less
the expected value. The potential overlap between the uniformly
distributed line samplers is not taken into account when calculating
the realized sampler area coverage.
• The surface area of the stratum and the proportion of the stratum
covered by the samplers.
Design Class Results
The design Results tab displays some general design properties and coverage
probability information for all survey designs.

Systematic Random Line Sampling - Results Tab

The Results tab for both designs and surveys displays some header information
for all survey designs.
Survey Plan Results
For each stratum in the survey layer the following are displayed:

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 205

 • The approximated number of line samplers displayed on the effort
allocation page. This may differ from the actual number generated,
as the lines are generated according to the spacing specified for the
systematic sampler lines.
• The actual number of line samplers generated and the associated
sampler half-width.
• The total estimated and realized aggregated sampler length.
• The spacing between the systematic line samplers.
• The angle of the sampler lines with respect to the x-axis, measured
in an anti-clockwise direction from the positive x-axis.
• The total length of the trackline, including distance spent off-effort
moving from the end of one sampler to the beginning of the next
one. Total cyclic trackline length includes the extra distance
required to return from the end of the last sampler to the beginning
of the first sampler.
• The expected sampler area coverage, which is the surface area
covered by the sampler lines. Each sampler line is enclosed in a
rectangle whose width is the same as that of the sampler. The area
intersection of the rectangle is used in calculating the realized
sampler area coverage. As the rectangles may fall partly outside the
survey region the realized sampler area coverage is generally less
the expected value. The potential overlap between the uniformly
distributed line samplers is not taken into account when calculating
the realized sampler area coverage.
• The surface area of the stratum and the proportion of the stratum
covered by the samplers.
Design Class Results
For each stratum in the stratum layer the minimum, maximum, mean, and
standard deviation of the on-effort and total trackline length is displayed. The
total trackline length is an approximation and does not assume that the observer
will return to the start of the first sampler line, but rather that the survey is over
after the last sampler line has been traversed. The distance covered to get to the
first line sampler and back from the last line sampler is not considered. The
design Results tab also displays some general design properties and coverage
probability information for all survey designs.

Systematic Segmented Line Sampling - Results Tab

The Results tab for both designs and surveys displays some header information
for all survey designs. For each stratum in the survey layer the following are
displayed:
Survey Plan Results
For each stratum in the survey layer the following are displayed:
• The approximated number of line segment samplers displayed on
the effort allocation page. This may differ from the actual number
generated, as the line segments are generated according to the
spacing specified for the systematic segment samplers.
• The actual number of line segment samplers generated and the
associated sampler half-width.
• The total estimated and realized aggregated sampler length.
• The length of the line segments.
• The spacing between the systematic line segment samplers.

206 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 • The spacing between the tracklines along which the segments run.
• The angle of the tracklines with respect to the x-axis, measured in
an anti-clockwise direction from the positive x-axis.
• The total length of the trackline, including distance spent off-effort
moving from the end of one sampler to the beginning of the next
one. Total cyclic trackline length includes the extra distance
required to return from the end of the last sampler to the beginning
of the first sampler.
• The expected sampler area coverage, which is the surface area
covered by the sampler segments. Each sampler segment is
enclosed in a rectangle whose width is the same as that of the
sampler. The area intersection of the rectangle is used in
calculating the realized sampler area coverage. As the rectangles
may fall partly outside the survey region the realized sampler area
coverage is generally less the expected value. The potential overlap
between the uniformly distributed sampler segments is not taken
into account when calculating the realized sampler area coverage.
• The surface area of the stratum and the proportion of the stratum
covered by the samplers.
Design Class Results
The design Results tab displays some general design properties and coverage
probability information for all survey designs.

Systematic Segmented Grid Line Sampling - Results Tab

The Results tab is the same as for the Systematic Segmented Line
Sampling design, except that total trackline length is not calculated.

Equal Angle Zigzag - Results Tab

The Results tab for both designs and surveys displays some header information
for all survey designs. For each stratum in the survey layer the following are
displayed:
Survey Plan Results
For each stratum in the survey layer the following are displayed:
• The approximated line length displayed on the effort allocation
page. This may differ from the actual length of the zigzag sampler
generated, as the sampler is generated according to the angle
specified for the equal angle zigzag.
• The actual length of the zigzag sampler.
• The number of zigzag segments generated (each determined by a
change of zigzag direction) and the associated sampler half-width.
• The constant angle of the equal angle zigzag.
• The angle of the design axis, used to orientate the zigzag, with
respect to the x-axis, measured in an anti-clockwise direction from
the positive x-axis.
• The expected sampler area coverage, which is the surface area
covered by the sampler lines. Each segment making up the zigzag
sampler line is enclosed in a rectangle whose width is the same as
that of the sampler. The area intersection of the rectangle is used in
calculating the realized sampler area coverage. As the rectangles
may fall partly outside the survey region the realized sampler area
coverage is generally less the expected value. The potential overlap

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 207

 between the rectangles is not taken into account when calculating
the realized sampler area coverage.
• The surface area of the stratum and the proportion of the stratum
covered by the samplers.
Design Class Results
The design Results tab displays some general design properties and coverage
probability information for all survey designs.

Equal Spaced Zigzag - Results Tab

The Results tab for both designs and surveys displays some header information
for all survey designs. For each stratum in the survey layer the following are
displayed:
Survey Plan Results
For each stratum in the survey layer the following are displayed:
• The approximated line length displayed on the effort allocation
page. This may differ from the actual length of the zigzag sampler
generated, as the sampler is generated according to the spacing
specified for the equal spaced zigzag.
• The actual length of the zigzag sampler.
• The number of zigzag segments generated (each determined by a
change of zigzag direction) and the associated sampler half-width.
• The angle of the design axis, used to orientate the zigzag, with
respect to the x-axis, measured in an anti-clockwise direction from
the positive x-axis.
• The expected sampler area coverage, which is the surface area
covered by the sampler lines. Each segment making up the zigzag
sampler line is enclosed in a rectangle whose width is the same as
that of the sampler. The area intersection of the rectangle is used in
calculating the realized sampler area coverage. As the rectangles
may fall partly outside the survey region the realized sampler area
coverage is generally less the expected value. The potential overlap
between the rectangles is not taken into account when calculating
the realized sampler area coverage.
• The surface area of the stratum and the proportion of the stratum
covered by the samplers.
Design Class Results
The design Results tab displays some general design properties and coverage
probability information for all survey designs.

Adjusted Angle Zigzag - Results Tab

The Results tab for both designs and surveys displays some header information
for all survey designs. For each stratum in the survey layer the following are
displayed:
Survey Plan Results
For each stratum in the survey layer the following are displayed:
• The actual length of the zigzag sampler. The sampler is generated
according to the length specified for the adjusted angle zigzag.
• The number of zigzag segments generated (each determined by a
change of zigzag direction) and the associated sampler half-width.

208 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 • The angle of the design axis, used to orientate the zigzag, with
respect to the x-axis, measured in an anti-clockwise direction from
the positive x-axis.
The expected sampler area coverage, which is the surface area covered by the
sampler lines. Each segment making up the zigzag sampler line is enclosed in a
rectangle whose width is the same as that of the sampler. The area intersection of
the rectangle is used in calculating the realized sampler area coverage. As the
rectangles may fall partly outside the survey region the realized sampler area
coverage is generally less the expected value. The potential overlap between the
rectangles is not taken into account when calculating the realized sampler area
coverage.
The surface area of the stratum and the proportion of the stratum covered by the
samplers.
Design Class Results
The design Results tab displays some general design properties and coverage
probability information for all survey designs.

Survey Details Window

Survey Details Inputs Tab

Survey objects are used for two purposes: they are created from designs (as
described in Chapter 6 - Survey Design in Distance of the Users Guide), and
they are a component of an analysis (as described in Chapter 7 - Analysis in
Distance of the Users Guide). If you are using the survey as part of a survey
design exercise, then you may want to Run the survey to generate an example
sample data layer. If you are using the survey for data analysis, you will
probably be most interested in setting or viewing the survey properties, via the
Properties… button.
The Survey Details Inputs tab is divided into four sections: Survey, Survey
Methods and Data, Design, and Comments.
Survey
This section gives you some information about the survey, such as the name,
time created and time of the last run.
To change the name of the survey, simply type a new name into the box labeled
Name. Once you have typed the new name, hit Enter or click somewhere else on
the window to apply the name to the survey.
To run a survey, click the Run button. Note that you can only run a survey that
is based on a design -- for more about the process, see Chapter 6 - Survey Design
in Distance in the Users guide.
Survey Methods and Data
This section gives some outline information about the survey. To view more
detailed information, and to edit the properties, click on the Properties…
button. This will take you to the Survey Properties dialog.
Design
If your survey is based on a design, the design set and name will be displayed
here. You can choose from the list if you want to change the design, although
bear in mind that if the survey has been run (i.e., an example sample layer
generated from the design), then the results will no longer apply, and the survey
status will be reset to “not run”.
Comments

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 209

 The comments section is there for you to type some comments to yourself about
the survey. For example, you might want to remind yourself of why you chose to
use these input parameters. The same section appears in the Results tab, so you
can make comments about your results too.

You can give yourself more room by resizing the comments section.
Put your mouse just above the Comments section header and dragging the
section up and down. You may want to increase the height of the whole Survey
Details window (by dragging on its border) before you do this.

Survey Details Log Tab

See the Log tab of the Design Details window for more information.

Survey Details Results Tab

See the Results tab of the Design details window for more information.

Analysis Details Window

Analysis Details Inputs Tab

You use the Analysis Details Inputs tab to set up your analysis. It is divided into
five sections: Analysis, Survey, Data Filter, Model Definition and Comments.
Before you begin setting up and running analyses, you should read Chapter 7 -
Analysis in Distance of the Users Guide, and the appropriate chapter after that
relating to the analysis engine you want to use.

Analysis section of Analysis Details Inputs Tab

This section of the Analysis Details Inputs tab gives some information about
your analysis, such as the name of the analysis, and the time it was created. You
can also change the analysis name and run the analysis from here.
To change the name of the analysis, simply type a new name into the box labeled
Name. Once you have typed the new name, hit Enter or click somewhere else on
the window to apply the name to the analysis.
To run an analysis, click the Run button. See the Users Guide page in Chapter 7
on Running Analyses for more information and tips. Once the analysis has
finished, it will automatically take you to the Results tab if it ran okay, or the
Log tab if there were errors or warnings. While the analysis is running, the Run
button changes to a Stop button. Press this to abort the analysis. (Pressing the
Stop button has the same effect as pressing the Reset Analysis button for a
running analysis in the Analysis Browser).

Survey section of Analysis Details Inputs Tab

In this section of the Analysis Details Inputs tab, you specify which Survey to
use for the current analysis. For more about the use of surveys in data analysis,
see Working with Surveys during Analysis in Chapter 7 of the Users Guide.
To find out more about the survey currently selected, you can click on the
Details… button to open the Survey Details window for that survey.

210 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Data filter section of Analysis Details Inputs Tab
In this section of the Analysis Details Inputs tab, you specify which Data Filter
to use for the current analysis. For more background about Data Filters, see the
Users Guide pages on Working with Data Filters and Model Definitions.

It is often easier to manipulate data filters (e.g., create new ones,

delete them, rename them, etc. in the Analysis Components window – see the
Analysis Components Window page in Chapter 7 of the Users Guide for more
information).

The central window lists the Data Filters that are available for you to choose
from. The one selected for the current analysis is highlighted on the list.
Choosing a different Data Filter for your analysis
If you want to choose another data filter for this analysis, click on the data filter
you want. If you have results already for your analysis, Distance issues a
warning that they will be deleted. This is because your results were generated
with the old Data Filter and so will not correspond to your new choice. If you
want to do a new analysis but keep your old results, then you need to go back to
the Analysis Browser, click on the new analysis button and select the new data
filter in Analysis Details window for the new analysis.
Making a new Data Filter
To make a new Data Filter press the New… button. A new filter will be created
and appended to the current list. The new data filter is based on the data filter
you have highlighted in the central window when you press the new button. The
Data Filter Properties window is then opened up, so you can edit this new filter.

Scenario: Imagine you have run an analysis, and now want to try
another analysis, but with just one part of the Data Filter changed (say a different
truncation distance). Highlight the analysis you just ran in the Analysis Browser
and click the New Analysis button. A new analysis is created, based on your
current one. Now click the Show Details button to open an Analysis Details
window. The old Data Filter will already be highlighted, so click the New…
button to make a new Data Filter based on the old one. Make the changes in the
Data Filter Properties and press OK to return to the Analysis Details window.
Then click Run to run the analysis. Easy, eh!

Editing the Data Filter

Click the Properties… button. The Data Filter Properties window appears.
Make any change you want in the Data Filter Properties and then press OK to
return. Distance will warn you if you the data filter is associated with any
analyses that have already been run.

Click Properties… if you just want to view the properties for this
Data Filter, rather than edit them, and then press Cancel in the Data Filter
Properties window to return without saving any changes.

Double clicking on the ID of a Data Filter in the central window is a

shortcut way of opening the Data Filter Properties for that filter.

Renaming a Data Filter

Click on the data filter name, and start typing.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 211

 Deleting a Data Filter
To do this, go to the Analysis Components window.

Model definition section of Analysis Details Inputs Tab

In this section of the Analysis Details Inputs tab, you specify which Model
Definition to use for the current analysis. For more background about Model
Definitions, see - Working with Data Filters and Model Definitions in Chapter 7
of the Users Guide.

It is often easier to manipulate Model Definitions (e.g., create new

ones, delete them, rename them, etc. in the Analysis Components window – see
the Analysis Components Window in Chapter 7 of the Users Guide for more
information).

The central window lists the Model Definitions that are available for you to
choose from. The one selected for the current analysis is highlighted on the list.
Choosing a different Model Definition for your analysis
If you want to choose another Model Definition for this analysis, click on the
Model Definition you want. If you have results already for your analysis,
Distance issues a warning that they will be deleted. This is because your results
were generated with the old Model Definition and so will not correspond to your
new choice. If you want to do a new analysis but keep your old results, then you
need to go back to the Analysis Browser, click on the new analysis button and
select the new Model Definition in Analysis Details window for the new
analysis.
Making a new Model Definition
To make a new Model Definition press the New… button. A new filter will be
created and appended to the current list. The new Model Definition is based on
the Model Definition you have highlighted in the central window when you
press the new button. The Model Definition Properties window is then opened
up, so you can edit this new filter.
Editing the Model Definition
Click the Properties… button. The Model Definition Properties window
appears. Make any change you want in the Model Definition Properties and then
press OK to return. Distance will warn you if you the Model Definition is
associated with any analyses that have already been run.

Click Properties… if you just want to view the properties for this
Model Definition, rather than edit them, and then press Cancel in the Model
Definition Properties window to return without saving any changes.

Double clicking on the ID of a Model Definition in the central

window is a shortcut way of opening the Model Definition Properties for that
filter.

Renaming a Model Definition

Click on the Model Definition name, and start typing.
Deleting a Model Definition
To do this, go to the Analysis Components window.

212 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Comments section of Analysis Details Inputs Tab
The comments section is there for you to type some comments to yourself about
the current analysis. For example, you might want to remind yourself of why you
chose to use these input parameters. The same section appears in the Results tab,
so you can make comments about your results too.

You can give yourself more room by resizing the comments section.
Put your mouse just above the Comments section header and dragging the
section up and down. You may want to increase the height of the whole Analysis
Details window (by dragging on its border) before you do this.

Analysis Details Log Tab

This tab of the Analysis Details window allows you to check any warnings or
errors that occurred when you ran an analysis.
If you run an analysis and there are warnings, the Log tab is colored amber. If
there are errors, this tab is colored red. In either case, when you open the
Analysis Details window, this tab will display on top by default.

You can change the default tab for analyses that ran with warnings to
the Results tab. The option is under the Analysis tab of the Preferences
window. This option is useful if you are running analyses that regularly generate
warnings, but where you want to disregard the warning message.

The Log tab is split into two sections. The top contains the analysis log – a list of
the commands that Distance used to do your analysis, together with the message
that Distance sent back while executing the commands. The bottom section gives
a summary of any warning and error messages. You can resize the bottom
section by clicking just above it and dragging.

To quickly go to the part of the analysis Log that contains a warning

or error message, click on that message in the bottom section of the Log tab.

To save time, Distance only indexes and colors the first 30

warnings and 30 errors.

One common reason for analyses returning an error status is that the
data selection criteria in the Data Filter have been entered incorrectly. To check
what data are being sent to an analysis, tick the “Echo data to log” option in the
Analysis tab of the Preferences dialog.

To copy the log file to the clipboard, click the Copy to Clipboard
button on the main toolbar, or choose the Analysis - Log menu option Copy
Log to Clipboard.

You can change the font size of the Log text in the General tab of the
Preference dialog.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 213

 Analysis Details Log Tab - CDS and MCDS
Users of Distance 3.0 and earlier will recognize this as the old Log file. To help
you find the important messages, warnings are colored Amber and errors Red.
Unfortunately, many of the messages are rather cryptic, and people who did not
use Distance 3.0 and earlier will not be able to understand the command syntax.
The good news is that we expect warnings and errors to occur rather rarely, at
least for conventional distance sampling analyses, because the graphical user
interface in the new versions of Distance do not allow you to make many of the
mistakes that Distance 3.0 and earlier allowed.
As a check, you may want to see a log of the data that was output to the Distance
analysis engine. To have the data echoed in all future analyses, tick the check-
box in Analysis of the Preferences dialog (under the menu item Tools |
Preferences).
Errors and warnings during bootstrapping
Distance treats errors and warnings that are generated during bootstrap
estimation of variance differently from normal errors. These errors and warnings
are labeled as “Bootstrap Error:” and “Bootstrap Warning:”, and they are all
colored amber. If an error occurs during the bootstrap, the status is not set to
error (red), but to warning (amber) - this is because the error only affects the
variance estimation, not the whole analysis.
Internal errors
On occasion, the analysis engine's internal algorithms may fail. In this case it
usually generates an “Internal Error” in the Log, and the analysis status is set to
error (red). In extreme cases, the analysis engine may crash. For more
information about this, see the Troubleshooting – Chapter 12 of the Users Guide,
particularly the page on Internal Errors in the CDS and MCDS Analysis Engines.

Analysis Details Log Tab - MRDS

As for CDS and MCDS analyses, the analysis log starts with a list of the data
selection queries.
It next details how it has translated the fields names in the Distance project
database into names that can be used in detection function formulae. This is a
very useful part of the log to refer to when building new model formulae, as
incorrect covariate names are a common source of problems in the analysis. For
more on this, see Translating Distance Fields into DS and MR Covariates in
Chapter 10 of the Users Guide.
The log then gives the command used to start the R statistical software in batch
mode, followed by the R commands used and any response from the R software.
This section is very useful for diagnosing any problems, although the error
messages from R are sometimes quite crypic.
The log finishes by reporting the run status returned by R.

Analysis Details Log Tab – DSM

Analysis Details Results Tab

This tab on the Analysis Details Window allows you to view the results of your
analysis in detail.
The Results tab is divided into two sections: Results pages, and Comments.

214 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 At the top of the results pages, there is a drop-down list of all the pages that are
available. You can navigate through the pages by choosing from this list or
using the < Back and Next > buttons. Detailed information about the contents
of the results pages is given under Output from CDS Analyses (Chapter 8) and
Output from MCDS Analyses (Chapter 9) in the Users Guide.

You can increase or decrease the font size of an individual results page
by right clicking and choosing the appropriate button. This is particularly useful
for the text-based cluster size regression plots, which don’t fit easily on a page.
You can choose Set current font as default to make the font size the default
for all results and log pages.

For information about transferring your results to another application,

see Exporting CDS Results in Chapter 8 of the Users Guide, Exporting MCDS
Results in Chapter 9, or Exporting MRDS Results in Chapter 10.

Design Properties Dialog

This dialog window allows you to view and edit the properties for a survey
design class. For more information about design classes and survey plans, see
the Chapter 6 - Survey Design in Distance of the Users Guide.
The dialog is divided into four tab pages:
• General Properties
• Effort Allocation
• Sampler
• Coverage Probability
In addition there are three buttons at the bottom of the page:
• Defaults - resets all options on all tab pages to the Distance
defaults
• OK - saves any changes and closes the dialog
• Cancel - closes the dialog without saving changes
The General Properties and Coverage Probability tabs are identical for all
designs, while there is a different tab for each Sampler type (point and line).
The Effort Allocation tab displays a different set of properties for each design
class. The set of possible sampler and design class combinations is as follows:

Sampler Design Class

Point Simple Random Sampling
Systematic Grid Sampling
Line Parallel Random Sampling
Systematic Random Sampling
Systematic Segmented Trackline
Sampling
Systematic Segmented Grid Sampling
Equal Angle Zigzag
Equal Spaced Zigzag

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 215

 Adjusted Angle Zigzag

The statistics from any design class run include the minimum, maximum, mean,
and standard deviation of the coverage probability. For a survey plan the
statistics from the run include the number of points or lines, the maximum
possible area coverage, the realized area coverage, and the mean realized
proportion of survey area covered. Each statistic is the sum over all strata. If the
design is based on a line sampler then the statistic for the mean realized sampler
line length (mean of all strata) is also generated.

General Design Properties Tab

The General Properties tab lets you choose the stratum layer, coordinate
system and random number generator (RNG) seed value. This tab is the same for
all design classes.
The Stratum Layer
Choose the stratum layer from the drop-down list of available stratum layers.
The global, stratum and substratum type layers are displayed in the list. The
survey design is generated within each region stored in the chosen stratum layer.
A description of the stratum layer’s coordinate system is given. The stratum
layer can be either geo-referenced or not. If it is not geo-referenced, the distance
measurement units of its non-earth coordinate system will be displayed. The
coordinates of a geo-referenced stratum layer are stored in degrees latitude and
longitude and the type of geographic coordinate system is shown. If the stratum
layer coordinates are projected, then the description and units of the map
projection are also displayed.
The Design Coordinate System
The survey designs are generated within the stratum layer regions whose
coordinates correspond to the design coordinate system selected. If the selected
stratum layer is stored within a non-earth coordinate system, then the box “Same
coordinate system as stratum” is checked, and the Non-earth referenced radio
button is selected by default. The reason for this is that the facilities to transform
the non-earth coordinates of the stratum layer into a geographic or projected
coordinate system are not available. This type of geo-referencing can be
achieved in most commercially available GIS packages. For a non-earth stratum
layer, the survey design takes place in the same non-earth coordinate system in
which the original stratum layer coordinates are stored. If the stratum layer
coordinates are stored as degrees latitude and longitude or if they are projected,
then a checked “Same coordinate system as stratum” box leads to the designs
being generated in the same geographic or projected coordinate system as the
stratum layer. Unchecking the box lets you choose a design coordinate system
that is different from that of the stratum layer.
If your survey takes place over a small geographic area, the selected geo-
coordinate system and map projection are not overly important, and you may
well store your stratum layer coordinates within a non-earth coordinate system.
However, an appropriate selection is crucial for larger survey regions. The geo-
coordinate system and projection chosen will make a difference to the results.
Representing the Earth’s surface in two dimensions causes distortion in the
shape, area, distance, or direction of the data. Different projections cause
different types of distortions. Some projections are designed to minimize the
distortion of one or two of the data’s characteristics. For instance, a projection
could maintain the area of a feature but alter its shape.
Equal-area projections preserve the surface area of displayed features, but this is
achieved by distorting the shape, angle, and scale properties. As abundance
estimation requires an accurate value for the surface area of the study survey
region, an appropriately chosen equal-area projection should be used to calculate

216 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 the areas. Equidistant projections preserve the distances between certain points,
which is an important consideration when calculating the length of line samplers
or the distance between sampling locations. True-direction or azimuthal
projections maintain the directions or azimuths of all points on the map
correctly, while conformal projections preserve local shape, both of which play
an important role in navigation. For a more detailed description of the various
types of geo-coordinate systems and projections, as well as some guidelines for
selecting an appropriate combination of these, see the section on Coordinate
Systems and Distance Data (Chapter 5) and Coordinate Systems, Maps and
Calculations in Distance (Chapter 5) of the Users Guide.
Generally, the coordinates of a geo-referenced survey region will be stored as
decimal degrees latitude and longitude and a projection will be chosen as the
design coordinate system. Selecting the Geographic coordinate system radio
button will seldom lead to reasonable results. In the rare instance where the
design is generated within a geo-coordinate system, it will be the same as that of
the stratum layer. Clicking the third radio button will let you select options for
the projected design coordinate system. The geo-coordinate system on which the
projection is based will be the same as that of the stratum layer. Select the
projection and the distance measurement units associated with it. If the stratum is
not projected (the most common case), then the design's projection is set to that
of the project by default (unless this is [None], in which case the first in the list
of projections is set as the default).
The Random Number Generator (RNG)
Pseudo-random numbers are used when running the designs, to provide a
“random” starting point for each survey. The numbers are produced by a
random number generator (RNG). The RNG uses a “seed” to start the sequence
of pseudo-random numbers. By clicking on the “from system clock” radio button
the value of the seed is taken from the computer’s system clock. By selecting the
other radio button the seed value can be set to a fixed value (which must be an
odd whole number greater than 2 million). If you set the seed to a fixed value,
then each run will produce the same results – useful if you want to generate
exactly the same survey again in the future. If you use the system clock, then the
actual value used is recorded in the results, so you can generate the same results
again if you want.

Coverage Probability Design Properties Tab

The Coverage Probability tab lets you choose between an analytic (assume
even) or simulation based estimation of the coverage probabilities, at what
resolution these estimates should be made by selecting the coverage grid, and
where the results should be stored. For more details, see the Users Guide section
on Concept: Coverage Probability in Chapter 6.
If you select the first radio button the coverage probabilities will be estimated
analytically, and assumed even. The estimates of coverage probability that are
calculated analytically are only approximate. A simplistic formula that gives the
proportion of the sampled area relative to the survey region surface area is used.
It does not take into account sampler overlap, or the fact that parts of the sampler
may fall outside the study area along the region boundary. This option will most
likely be chosen if you already know about the coverage probability achieved by
a given design class, and are interested in its other properties – analytic estimates
are much faster to calculate than simulated estimates.
To estimate the coverage probabilities by simulation, select the second radio
button. You then need to set the number of repetitions of the survey design that
should be used to obtain the estimates. For the purpose of estimating coverage
probabilities, point or line transects have an associated radius or half-width,
respectively. The precision of the coverage probability estimates obtained is
highly dependent on the grid point spacing relative to the sampler width or

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 217

 radius, and on the average proportion of the survey region sampled by each
design in the design class. Thus it is important to select an appropriate grid
spacing before starting the simulations. A spacing that is too coarse provides a
very noisy estimate of coverage probability because for most realizations of this
design, a large number of samplers remain undetected by any grid point. A very
fine point grid is computationally intensive (i.e. unduly slow). Good grid spacing
ensures that a sufficient number of grid points are hit by the sampler, relative to
its surface area. A basic rule of thumb is that the number of grid points falling
within a single sampler - for any particular realization of the design - should
never be less than one at a bare minimum.
Increasing the number of simulations can counteract the shortfall caused by a
coarse grid. It is more effective, however, to increase the resolution of the grid.
To obtain a sufficiently precise estimate of coverage probability, while not
sacrificing computational efficiency, a trade-off between grid spacing and the
total number of simulations must be made. We suggest the following
approximate formula. If you require a variance no greater than v in your
coverage probability estimates, then the total number of simulations needed is
approximated by A2v / a(A-a), where a is the total surface area covered by the
samplers and A that of the survey region. This approximation assumes equal
coverage probabilities. If the coverage probability is not even, then the number
of simulations must be increased to achieve the same precision.
The intrinsic stochasticity in the estimation process means that the coverage
probability estimates will be variable for even as well as uneven probability
designs. The level of variability decreases as the number of simulations
increases, but given the coverage probability results and a case where uncertainty
exists about whether the design provides even cover or not, one can test for even
coverage using an index-of-dispersion or a classical χ 2 -goodness-of-fit test.
Note that points falling in adjacent strata that are hit by portions of the sampler
lying outside the design stratum are disregarded. This is justified by the
assumption that potential observations at the points in question would not be
recorded from the sampler over the stratum boundary.
Results Coverage Grid
The regular grid of points contained in the coverage layers is used for estimating
coverage probability for our survey designs. You select the coverage grid layer
in which the coverage probabilities are stored from the previously created drop-
down list of grid layers. The coverage probability estimates at each point in the
coverage probability grid are stored in the field whose name you specify in the
text box.

Sampler Design Properties Tab

Sampler tab for point sampler designs
The Sampler tab for design classes based on point samplers allows you to
specify the radius associated with each point sampler. This radius is required to
estimate the coverage probabilities. We suggest you use the value of your
truncation distance.
Select the point sampler radius units from the drop-down list. If the design
coordinate system is non-earth or projected, these are linear distance
measurement units. If the design takes place in a geo-coordinate system, these
are angular units. It is best to choose the same units that are used in the design
coordinate system and for effort allocation, as then Distance won’t have to
convert between different units. Conversion inevitably leads to some loss in
precision (although this loss is usually very small).
Check the Same properties for all strata box if you want the point samplers
to have the same radius in all survey strata. The box will be checked and

218 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 disabled if there is only a single stratum in the selected stratum layer. Enter a
single positive value for the radius in the Radius column of the grid table.
Unchecking the box will expand the grid table. Each row in the table will
correspond to a stratum in the layer, which allows you to enter a different radius
value for each stratum (if, for example, you have a different truncation distance
in different strata). Each stratum’s label, or ID value if the strata are not labelled,
are shown in the Stratum column of the table.

Sampler tab for line sampler designs

The Sampler tab for design classes based on line samplers allows you to
specify the half-width associated with each line sampler. The width is required to
estimate the coverage probabilities. We suggest you use the value of your
truncation distance.
Select the line sampler half-width units from the drop-down list. If the design
coordinate system is non-earth or projected, these are linear distance
measurement units. If the design takes place in a geo-coordinate system, these
are angular units. It is best to choose the same units that are used in the design
coordinate system and for effort allocation, as then Distance won’t have to
convert between different units. Conversion inevitably leads to some loss in
precision (although this loss is usually very small).
Check the Same properties for all strata box if you want the line samplers
to have the same half-width in all survey strata. The box will be checked and
disabled if there is only a single stratum in the selected stratum layer. Enter a
single positive value for the half-width in the Width column of the grid table.
Unchecking the box will expand the grid table. Each row in the table will
correspond to a stratum in the layer, which allows you to enter a different half-
width value for each stratum (if, for example, you have a different truncation
distance in different strata). Each stratum’s label, or ID value if the strata are not
labelled, are shown in the Stratum column of the table.

Effort Allocation Design Properties Tab

The Effort Allocation property pages let you define the amount of effort you
want to apportion to each stratum in the survey layer.

For survey layers containing multiple strata, you can allocate zero
effort to some of the strata. For those strata with zero effort, the design
properties will not be calculated during a design run, and no design will be
generated during a survey run. The sum of the effort over all strata should,
however, be greater than zero.

Simple Random Sampling - Effort Allocation Properties

Edge Sampling
The Edge Sampling options provide different methods for dealing with point
samplers falling along the boundary of the survey region. For more information,
see the section on Concept: Edge Effects in Chapter 6 of the Users Guide.
Allocation by stratum
Each row in the grid table corresponds to a stratum in the layer, which allows
you to allocate effort for each stratum in the survey layer. Each stratum’s ID and
label (if this field exists) are shown in the Id and Label column of the table,
respectively.
You can select the Absolute values radio button and enter the number of point
samplers you want in the Effort column of the grid table. Otherwise, if you

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 219

 select the other radio button you can enter a point sampler total in the text box
and specify a percentage from that total in the Effort% column of the grid table.
The percentages over all the strata do not have to sum to 100. Under the second
option the Integer Totals box will also be enabled. By checking this box any
effort percentage that leads to a non-integer number will be rounded to an
integer. Point samplers are always generated from integer totals anyway. Check
the Same effort for all strata box if you want the same number or percentage
of point samplers in all survey strata, otherwise you can allocate different effort
values for each stratum. The box will be checked and disabled if there is only a
single stratum in the selected stratum layer. If the box is checked when the
stratum layer contains multiple strata then a single row is displayed in the table,
and the effort values entered in this row are used for all strata. The Total points
text box displays the aggregated total of sampler points over all survey strata.
Each point sampler is stored as a sampling unit when you create a survey plan.

Systematic Grid Sampling - Effort Allocation Properties

Edge Sampling
The Edge Sampling options provide different methods for dealing with point
samplers falling along the boundary of the survey region. For more information,
see the section on Concept: Edge Effects in Chapter 6 of the Users Guide..
Allocation by stratum
Select the between grid point spacing units from the drop-down list. If the design
coordinate system is non-earth or projected these are linear distance
measurement units. Otherwise, if the design takes place in a geo-coordinate
system these are angular units. By selecting the same units that are used in the
design coordinate system or for the sampler radius, imprecision introduced
during unit conversions can be avoided.
Each row in the grid table corresponds to a stratum in the layer, which allows
you to allocate effort for each stratum in the survey layer. Each stratum’s ID and
label (if this field exists) are shown in the Id and Label column of the table,
respectively.
You can select the Absolute values radio button and enter the number of point
samplers you want in the Effort column of the grid table. The second radio
button lets you enter a point sampler total in the text box and specify a
percentage from that total in the Effort% column of the grid table. The
percentages over all the strata do not have to sum to 100. By selecting the
Systematic point grid spacing you can enter the regular spacing between
grid points. If the Square grid box is checked the point grid is square and you
enter the length of each square’s side under the Side column in the table. With
this box unchecked you can enter different values horizontal and vertical grid
spacing values in the Width and Height columns, respectively. Under the
second and third effort allocation option the Integer Totals box will also be
enabled. By checking this box any effort percentages or grid spacing that leads to
a non-integer number will be rounded to an integer. Point samplers are always
generated from integer totals anyway. Enter the angle of the systematic point
grid with respect to the x-axis - measured in an anti-clockwise direction from the
positive x-axis – in the table’s Angle column. The angle should be greater or
equal to zero and less than ninety degrees.
When the Update effort in real time box is checked calculations to estimate
the missing information are performed. So, if you enter an absolute number of
points the software tries to estimate the square spacing required for a systematic
grid with that number of points. In general a systematic spacing can be found
that gives an estimated number of points near to – rather than exactly equal to –
the absolute number you have specified. This is where the Effort Tolerance
text box comes in. This allows the calculated spacing to give an estimated
number that lies within the effort tolerance percentage number of points either

220 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 side of the actual number you have specified. If you specify an effort tolerance
that is really narrow, then the software may not be able to find a spacing. If the
tolerance is too wide, the algorithm may stop before a potentially better spacing
is found. So, try starting off with a fairly narrow tolerance, say one (1%), and if
you keep getting error messages make it wider. If you enter a grid spacing, the
number of points resulting from this spacing will be estimated. The points are
generated according to the spacing specified or estimated for the systematic
regular grid of sampler points, so the number of point samplers generated in a
run of the design may differ from the absolute number specified or the
approximate number calculated on this page – you should therefore use the
estimates on this page only as a guide. If you change the distance units then the
grid spacing for each stratum is updated as are the estimated number of points
for that new spacing. Alternatively, if your computer is slow or you want to enter
all your values and then do the calculations just uncheck the Update effort in
real time box, and press the Update Effort button when you are ready.
Check the Same effort for all strata box if you want either the same grid
spacing, or same number or percentage of point samplers in all survey strata,
otherwise you can allocate different values for each stratum. The box will be
checked and disabled if there is only a single stratum in the selected stratum
layer. The Total points text box displays the estimated aggregated total of
sampler points over all survey strata.
Each point sampler is stored as a sampling unit when you create a survey plan.
Future versions of Distance may allow you to store this design in two sample
layers – points along lines, allowing you to choose the appropriate level of
analysis in the analysis engine.

Parallel Random Line Sampling - Effort Allocation

Properties
Edge Sampling
The Edge Sampling options provide different methods for dealing with line
samplers falling along the boundary of the survey region. For more information,
see the section on Concept: Edge Effects in Chapter 6 of the Users Guide.
Effort determined by
Select the first radio button if you want to determine effort by Sampler
number (i.e., number of lines). With this option the number of survey lines you
specify in the Samplers column of the table will be generated. If you select the
second Sampler length option and specify the length value in the Length
column of the table, then sampler lines will be generated until their aggregated
length exceeds the length specified.
Allocation by stratum
Select the line length units from the drop-down list. If the design coordinate
system is non-earth or projected these are linear distance measurement units.
Otherwise, if the design takes place in a geo-coordinate system these are angular
units. By selecting the same units that are used in the design coordinate system
or for the sampler width, imprecision introduced during unit conversions can be
avoided.
Each row in the grid table corresponds to a stratum in the layer, which allows
you to allocate effort for each stratum in the survey layer. Each stratum’s ID and
label (if this field exists) are shown in the Id and Label column of the table,
respectively.
You can select the Absolute values radio button and – depending on which
Effort determined by option you chose – enter either the number or
aggregated length of line samplers you want in the Samplers or Length
column of the grid table, respectively. The second radio button lets you enter a
number or aggregated length of line total in the text box, and then specify a

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 221

 percentage from that total in the Effort% column of the grid table. The
percentages over all the strata do not have to sum to 100. Under the second effort
allocation option the Integer Totals box will also be enabled. By checking this
box any effort percentages that lead to a non-integer line number will be rounded
to an integer. If effort is determined by line number, then these samplers are
always generated from integer totals anyway. Enter the angle of the parallel line
samplers with respect to the x-axis - measured in an anti-clockwise direction
from the positive x-axis – in the table’s Angle column. The angle should be
greater or equal to zero and less than 180 degrees.
When the Update effort in real time box is checked calculations to estimate
the missing information are performed. So, if effort is determined by Sampler
number and you enter an absolute number of points or a percentage value, the
software tries to estimate the line length of sampler line that would be generated.
This is only an approximation, which is dependent on the shape of your survey
region. In the current version of the software the approximation may also grow
worse as the angle of the sampler lines departs from 90 degrees. If effort is
determined by Sampler length, then as you enter an absolute line length or a
percentage value, the software tries to estimate the number of sampler lines that
would be generated. If you change the distance units then the line length for each
stratum is updated, as are the estimated number of lines for that new length.
Alternatively, if your computer is slow or you want to enter all your values and
then do the calculations just uncheck the Update effort in real time box, and
press the Update Effort button when you are ready.
Check the Same effort for all strata box if you want either the same line
length, the same number of lines, or percentage of sampler number or length, in
all survey strata. Otherwise you can allocate different values for each stratum.
The box will be checked and disabled if there is only a single stratum in the
selected stratum layer. The Total lines and length text boxes display the exact or
estimated aggregated totals of sampler lines or line length over all survey strata,
respectively.
Each line sampler is stored as a sampling unit when you create a survey plan. A
single line sampler may be made up of one or more parts, depending on whether
the shape of the survey region causes a split in the line.

Systematic Random Line Sampling - Effort Allocation

Properties
Edge Sampling
The Edge Sampling options provide different methods for dealing with line
samplers falling along the boundary of the survey region. For more information,
see the section on Concept: Edge Effects in Chapter 6 of the Users Guide.
Allocation by stratum
Select the line length and spacing units from the drop-down list. If the design
coordinate system is non-earth or projected these are linear distance
measurement units. Otherwise, if the design takes place in a geo-coordinate
system these are angular units. By selecting the same units that are used in the
design coordinate system or for the sampler width, imprecision introduced
during unit conversions can be avoided.
Each row in the grid table corresponds to a stratum in the layer, which allows
you to allocate effort for each stratum in the survey layer. Each stratum’s ID and
label (if this field exists) are shown in the Id and Label column of the table,
respectively.
You can select the Absolute values radio button, and depending on whether
you choose lines or line length from the drop-down list, you then enter either
the number or aggregated length of line samplers you want in the Samplers or
Length column of the grid table, respectively. The second radio button lets you

222 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 enter a number or aggregated length of line total – depending on the choice of
lines or line length from the drop-down list - in the text box, and then specify
a percentage from that total in the Effort% column of the grid table. The
percentages over all the strata do not have to sum to 100. Under the second effort
allocation option the Integer Totals box will also be enabled. By checking this
box any effort percentages that lead to a non-integer line number will be rounded
to an integer. The systematic lines are generated according to the spacing
specified or estimated from the number of lines specified, so the number of line
samplers generated may differ from the absolute number specified or the
approximate number calculated. The Integer Totals box is disabled when the
third effort allocation radio button is selected, because the estimated number of
sampler lines is always an integer anyway. Enter the angle of the parallel line
samplers with respect to the x-axis - measured in an anti-clockwise direction
from the positive x-axis – in the table’s Angle column. The angle should be
greater or equal to zero and less than 180 degrees.
When the Update effort in real time box is checked calculations to estimate
the missing information are performed. So, if effort is determined by lines and
you enter an absolute number of lines or a percentage value, the software tries to
estimate the systematic inter-line spacing and the line length of sampler line that
would be generated. This is only an approximation, which is dependent on the
shape of your survey region. In the current version of the software the
approximation may also grow worse as the angle of the sampler lines departs
from 90 degrees. If effort is determined by line length, then as you enter an
absolute line length or a percentage value, the software tries to estimate the
systematic inter-line spacing and number of sampler lines that would be
generated. If effort is determined by Systematic line spacing, then as you
enter the spacing value the software tries to estimate the number of sampler lines
that would be generated, their aggregated length, and the associated percentage
value. If you change the distance units then either the line length or inter-line
spacing – depending on the selected effort allocation option - for each stratum is
updated, as are the values in the remaining columns. Alternatively, if your
computer is slow or you want to enter all your values and then do the
calculations just uncheck the Update effort in real time box, and press the
Update Effort button when you are ready.
Check the Same effort for all strata box if you want either the same line
length, the same number of lines, or percentage of sampler number or length, in
all survey strata. Otherwise you can allocate different values for each stratum.
The box will be checked and disabled if there is only a single stratum in the
selected stratum layer. The Total lines and length text boxes display the
approximate and exact aggregated totals of sampler lines and line length over all
survey strata, respectively.
Each line sampler is stored as a sampling unit when you create a survey plan. A
single line sampler may be made up of one or more parts, depending on whether
the shape of the survey region causes a split in the line.

Systematic Segmented Line Sampling - Effort Allocation

Properties
Edge Sampling
The Edge Sampling options provide different methods for dealing with line
samplers falling along the boundary of the survey region. For more information,
see the section on Concept: Edge Effects in Chapter 6 of the Users Guide.
Non-convex survey regions
The segmented line sampling design is generated by systematically spacing
segments along tracklines. The tracklines can be generated within the survey
region or its minimum bounding rectangle. If the design is generated within
irregular survey regions that have narrow sub-regions this can lead to uneven or,

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 223

 in the extreme case, some zero coverage probabilities, if complete segments are
used (see below). If simulation shows such an effect then check the Use a
minimum bounding rectangle box to counteract it. However, for irregular
survey regions checking this box may lead to survey designs with a less
systematic spatial spread throughout the region. If the designs are generated
within the minimum bounding rectangle of the survey region, then sampler
segments will sometimes necessarily be less than complete, because they are
clipped against the original survey region. For this reason the Allow split
sampler segments box is disabled when the Use a minimum bounding
rectangle box is checked.
For designs generated within the survey region itself, portions of segments may
also lie outside the survey region, where they intersect the boundary. Check the
Allow split sampler segments box if you want to allow boundary segments
to be broken in two. If this box remains unchecked you will always get complete
segments, but the cost of this is some irregularity in the inter-segment spacing. If
split segments are disallowed, then the design ensures you get complete
segments by moving boundary segments along or between the tracklines. If more
than half of the length of a boundary segment falls on its original trackline, then
the segment is moved ‘inwards’ until it completely falls within the survey
region. Otherwise, it is removed from its original trackline and placed
completely on the next trackline in the sequence.
Allocation by stratum
Select the line length and spacing units from the drop-down list. If the design
coordinate system is non-earth or projected these are linear distance
measurement units. Otherwise, if the design takes place in a geo-coordinate
system these are angular units. By selecting the same units that are used in the
design coordinate system or for the sampler width, imprecision introduced
during unit conversions can be avoided.
Each row in the grid table corresponds to a stratum in the layer, which allows
you to allocate effort for each stratum in the survey layer. Each stratum’s ID and
label (if this field exists) are shown in the Id and Label column of the table,
respectively.
If you select the Absolute values radio button, then you choose sampler
segments or total length from the drop-down list. The sampler segments
option lets you then enter the number of segment samplers you want, in the
Samplers column of the grid table. The second option lets you enter the
aggregated segment length in the Length column. The second radio button lets
you enter a total segment number or aggregated segment length – depending on
the choice of sampler segments or total length from the drop-down list - in
the text box, and then specify a percentage from that total in the Effort%
column of the grid table. The percentages over all the strata do not have to sum
to 100. Under the second effort allocation option the Integer Totals box will
only be enabled if sampler segments option is showing. By checking this
box any effort percentages that lead to a non-integer segment number will be
rounded to an integer. Choosing the Systematic line spacing effort
allocation option lets you specify the inter-segment spacing in the Spacing
column of the table. If the Same spacing between segments and lines
box is checked then the tracklines along which the segments run are spaced at
the same distance as the segments. By un-checking this box you can specify a
different inter-trackline spacing in the Trackline column of the table. Enter the
length of each sampler segment in the table’s Segment column. The systematic
line segments are generated according to the inter-segment and inter-trackline
spacing specified, or estimated from either the specified number of segments or
their specified total length. Thus, the number of segment samplers generated
may differ from the absolute number specified or the approximate number
calculated. The Integer Totals box is disabled when the third effort allocation
radio button is selected, because the estimated number of samplers is always an
integer anyway. Enter the angle of the parallel tracklines with respect to the x-

224 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 axis - measured in an anti-clockwise direction from the positive x-axis – in the
table’s Angle column. If you wish use the same angle for each realization, then
enter a value greater or equal to zero and less than 180 degrees. Alternatively, if
you enter a value of -1 then Distance will choose a random angle for each
realization.
When the Update effort in real time box is checked calculations to estimate
the missing information are performed. So, if effort is determined by sampler
segments and you enter an absolute number of segments or a percentage value,
the software tries to estimate the systematic inter-segment spacing (the inter-
trackline spacing is the same) and the total length of sampler segments that
would be generated. You most enter a value in the Segment column for the
calculations can proceed. The result of the calculations is only an approximation,
which is dependent on the shape of your survey region. In the current version of
the software the approximation may also grow worse if the angle of the sampler
lines is not 90 degrees. If effort is determined by segment length, then as you
enter a total segment length or a percentage value, the software tries to estimate
the systematic inter-segments (and trackline) spacing and number of sampler
segments that would be generated. Again only if the length of a single segment
has been entered. If effort is determined by Systematic line spacing, then as
you enter the spacing value(s) the software tries to estimate the number of
sampler segments that would be generated, their aggregated length, and the
associated percentage value. If you change the distance units then either the
segment length or inter-segment (and trackline) spacing – depending on the
selected effort allocation option - for each stratum is updated, as are the values in
the remaining columns. Alternatively, if your computer is slow or you want to
enter all your values and then do the calculations just uncheck the Update
effort in real time box, and press the Update Effort button when you are
ready.
Check the Same effort for all strata box if you want either the same total
segment length, the same number of segments, or percentage of sampler number
or length, in all survey strata. Otherwise you can allocate different values for
each stratum. The box will be checked and disabled if there is only a single
stratum in the selected stratum layer. The Total lines and length text boxes
display the approximate and exact aggregated totals of sampler segment lines
and segment line length over all survey strata, respectively.
Each segment sampler is stored as a sampling unit when you run this design
(split segments are stored as separate sampler, which may lead to some
inappropriately small samplers. This will be dealt with in future versions of the
software). This is fine if spacing between lines and samplers is similar (or in the
unlikely case that spacing between samplers is greater then between lines). But
not if spacing between samplers is small relative to lines. Future versions of
Distance will store this design in two sample layers – segments within lines,
allowing you to choose the appropriate level of analysis in the analysis engine.

Systematic Segmented Grid Line Sampling - Effort

Allocation Properties
The options for this design are the same as for the systematic segmented line
sampling design.

Equal Angle Zigzag - Effort Allocation Properties

Non-convex survey region approximated by a
The Non-convex survey region options provide different methods for dealing
with non-convex survey regions (see Zigzag Sampling Non-convex Survey
Region Options in the Program Reference).
Effort determined by

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 225

 Select the first radio button if you want to determine effort by Sampler angle.
With this option the equal zigzag sampler will be generated with the constant
angle you specify in the Angle column of the table. The constant angle should
be greater than zero and less than ninety degrees. If you select the second
Sampler length option and specify the zigzag’s length value in the Length
column of the table, then the equal angle corresponding to the length specified
will be calculated. The result of the calculations is only an approximation, which
is dependent on the shape of your survey region. In the current version of the
software the approximation may also grow worse if the angle of the sampler
lines is not 90 degrees. The length of the zigzag generated at the calculated angle
will thus vary to a lesser or greater degree from the length specified.
Design Axis
The design axis options provide different methods for specifying the orientation
of the design axis for zigzag samplers (see Zigzag Sampling Design Axis
Options in the Program Reference).
Allocation by stratum
Select the line length units from the drop-down list. If the design coordinate
system is non-earth or projected these are linear distance measurement units.
Otherwise, if the design takes place in a geo-coordinate system these are angular
units. By selecting the same units that are used in the design coordinate system
or for the sampler width, imprecision introduced during unit conversions can be
avoided.
Each row in the grid table corresponds to a stratum in the layer, which allows
you to allocate effort for each stratum in the survey layer. Each stratum’s ID and
label (if this field exists) are shown in the Id and Label column of the table,
respectively.
You can select the Absolute values radio button and – depending on which
Effort determined by option you chose - enter either the sampler angle or
length in the Angle or Length column of the grid table, respectively. The
second radio button is only enabled when effort is determined by length, and lets
you enter a length of line in the text box. You can then specify a percentage from
that total in the Effort% column of the grid table. The percentages over all the
strata do not have to sum to 100.
When the Update effort in real time box is checked calculations to estimate
the missing information are performed. So, if effort is determined by Sampler
angle and you enter an angle value in degrees, the software tries to estimate the
line length of the zigzag sampler that would be generated. Similarly, if effort is
determined by Sampler length, then as you enter an absolute line length or a
percentage value, the software tries to estimate the constant angle of the zigzag.
If you change the distance units then the line length for each stratum is updated,
as are the angle estimates for that new length. Alternatively, if your computer is
slow or you want to enter all your values and then do the calculations just
uncheck the Update effort in real time box, and press the Update Effort
button when you are ready.
Check the Same effort for all strata box if you want either the same line
length or zigzag angle, in all survey strata. Otherwise you can allocate different
values for each stratum. The box will be checked and disabled if there is only a
single stratum in the selected stratum layer. The Total length text box displays
the aggregated totals of sampler line length over all survey strata.
The zigzag sampler is made up of line segments (each determined by a change of
zigzag direction). These segments are stored as sampling units when you create a
survey plan.

Equal Spaced Zigzag - Effort Allocation Properties

Non-convex survey region approximated by a

226 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 The Non-convex survey region options provide different methods for dealing
with non-convex survey regions (see Zigzag Sampling Non-convex Survey
Region Options in the Program Reference).
Effort determined by
Select the first radio button if you want to determine effort by Sampler
spacing. With this option the equal zigzag sampler will be generated at the
spacing you specify in the Spacing column of the table. The equal spaced
zigzag passes through equally spaced points on opposite sides of the survey
region boundary, and this value determines the spacing of those points. If you
select the second Sampler length option and specify the zigzag’s length value
in the Length column of the table, then the equal spacing corresponding to the
length specified will be calculated. The result of the calculations is only an
approximation, which is dependent on the shape of your survey region. In the
current version of the software the approximation may also grow worse if the
angle of the sampler lines is not 90 degrees. The length of the zigzag generated
at the calculated spacing will thus vary to a lesser or greater degree from the
length specified.
Design Axis
The design axis options provide different methods for specifying the orientation
of the design axis for zigzag samplers (see Zigzag Sampling Design Axis
Options in the Program Reference).
Allocation by stratum
Select the line length units from the drop-down list. If the design coordinate
system is non-earth or projected these are linear distance measurement units.
Otherwise, if the design takes place in a geo-coordinate system these are angular
units. By selecting the same units that are used in the design coordinate system
or for the sampler width, imprecision introduced during unit conversions can be
avoided.
Each row in the grid table corresponds to a stratum in the layer, which allows
you to allocate effort for each stratum in the survey layer. Each stratum’s ID and
label (if this field exists) are shown in the Id and Stratum column of the table,
respectively.
You can select the Absolute values radio button and – depending on which
Effort determined by option you chose - enter either the sampler spacing or
length in the Spacing or Length column of the grid table, respectively. The
second radio button is only enabled when effort is determined by length, and lets
you enter a length of line in the text box. You can then specify a percentage from
that total in the Effort% column of the grid table. The percentages over all the
strata do not have to sum to 100.
When the Update effort in real time box is checked calculations to estimate
the missing information are performed. So, if effort is determined by Sampler
spacing and you enter a spacing value, the software tries to estimate the line
length of the zigzag sampler that would be generated. Similarly, if effort is
determined by Sampler length, then as you enter an absolute line length or a
percentage value, the software tries to estimate the constant spacing of the
zigzag. If you change the distance units then the spacing or line length –
depending on how effort is determined - for each stratum is updated, as are
length or spacing estimates corresponding to the new spacing or length,
respectively. Alternatively, if your computer is slow or you want to enter all your
values and then do the calculations just uncheck the Update effort in real
time box, and press the Update Effort button when you are ready.
Check the Same effort for all strata box if you want either the same line
length or zigzag spacing, in all survey strata. Otherwise you can allocate
different values for each stratum. The box will be checked and disabled if there

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 227

 is only a single stratum in the selected stratum layer. The Total length text box
displays the aggregated totals of sampler line length over all survey strata.
The zigzag sampler is made up of line segments (each determined by a change of
zigzag direction). These segments are stored as sampling units when you create a
survey plan.

Adjusted Angle Zigzag - Effort Allocation Properties

Non-convex survey region approximated by a
The Non-convex survey region options provide different methods for dealing
with non-convex survey regions (see Zigzag Sampling Non-convex Survey
Region Options in the Program Reference)
Effort determined by
Select the first radio button if you want to determine effort by Coverage
probability. With this option the adjusted angle zigzag sampler will be
generated approximately at the coverage probability you specify in the Cov
Prob column of the table. Given this value and the line sampler width 4) the
length of the zigzag can be determined. Thus, if you do not want the default
sampler width, you need to change this before the length calculation takes place.
If you select the second Sampler length option and specify the zigzag’s length
value in the Length column of the table, then given the sampler width, the
coverage probability corresponding to the length specified will be calculated.
Design Axis
The design axis options provide different methods for specifying the orientation
of the design axis for zigzag samplers (see Zigzag Sampling Design Axis
Options in the Program Reference).
Allocation by stratum
Select the line length units from the drop-down list. If the design coordinate
system is non-earth or projected these are linear distance measurement units.
Otherwise, if the design takes place in a geo-coordinate system these are angular
units. By selecting the same units that are used in the design coordinate system
or for the sampler width, imprecision introduced during unit conversions can be
avoided.
Each row in the grid table corresponds to a stratum in the layer, which allows
you to allocate effort for each stratum in the survey layer. Each stratum’s ID and
label (if this field exists) are shown in the Id and Label column of the table,
respectively.
You can select the Absolute values radio button and – depending on which
Effort determined by option you chose - enter either the sampler coverage
probability or length in the Cov Prob or Length column of the grid table,
respectively. The second radio button is only enabled when effort is determined
by length, and lets you enter a length of line in the text box. You can then
specify a percentage from that total in the Effort% column of the grid table. The
percentages over all the strata do not have to sum to 100.
When the Update effort in real time box is checked calculations to estimate
the missing information are performed. So, if effort is determined by Coverage
probability and you enter a value, the software tries to estimate the line length
of the zigzag sampler that would be generated using this value and the sampler
width. Similarly, if effort is determined by Sampler length, then as you enter
an absolute line length or a percentage value, the software tries to estimate the
coverage probability of the zigzag, also using the sampler width. If you change
the distance units then the line length for each stratum is updated, as are
coverage probability estimates corresponding to the new length, respectively.
Alternatively, if your computer is slow or you want to enter all your values and

228 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 then do the calculations just uncheck the Update effort in real time box, and
press the Update Effort button when you are ready.
Check the Same effort for all strata box if you want either the same line
length or zigzag coverage probability, in all survey strata. Otherwise you can
allocate different values for each stratum. The box will be checked and disabled
if there is only a single stratum in the selected stratum layer. The Total length
text box displays the aggregated totals of sampler line length over all survey
strata.
The zigzag sampler is made up of line segments (each determined by a change of
zigzag direction). These segments are stored as sampling units when you create a
survey plan.

Zigzag Sampling Design Axis Options

For zigzag sampling designs, the option you select in the Design Axis section
of the Effort Allocation page determines the angle of the design axis. Zigzag
samplers are orientated with respect to this axis. If you select the Runs at an
angle to the x-axis option, the angle of the design axis is defined with respect
to the x-axis - measured in an anti-clockwise direction from the positive x-axis –
and can be entered in the “DA Angle” table column on that page. The angle
should be greater than or equal to zero and less than 180 degrees. Selecting
Determined by a start and end location lets you define the start and end
locations for the design axis in each stratum. Do this by entering the coordinate
values in the StartX, StartY, EndX, and EndY table columns. For this third
way of defining the design axis, if the stratum layer is geo-referenced, then the
Defined as geographic coordinates box is enabled. Check the box if you
want to enter the coordinates in degrees longitude (x-coordinate) and latitude (y-
coordinate).

Zigzag Sampling Non-convex Survey Region Options

This section describes the options on the Non-convex survey region
approximated by a part of the Effort allocation tab for zigzag sampling
design classes. For important background information, see Zigzag Sampling
Non-convex Survey Region Options in the Program Reference.
Zigzag sampling designs can only be generated in a convex survey region. If any
of the survey strata in the survey layer are non-convex then you can choose to
generate the design for each stratum in either a Convex hull or Bounding
rectangle, by selecting the appropriate radio button. For non-convex strata the
zigzag sampler will no longer be continuous. The amount of discontinuity can
generally be reduced by selecting the Convex hull option, but this may lead to
uneven coverage probabilities. If simulation shows such an effect to be extreme,
then select the Bounding rectangle option instead.
If you want to store the convex regions within which the designs are generated
for some or all of the strata, then check the box and enter a valid name for the
new data layer. This new data layer will only be created during survey runs
rather than design runs. The convex layer will appear beneath the survey stratum
you selected for the design in the data layer hierarchy. If a layer of the same
name already exists there, it will be overwritten.

In the current version of the software when you choose the

Bounding rectangle option to deal with non-convex regions, the bounding
rectangle’s width runs parallel to the design axis. This gives you a chance to
choose a design axis that minimizes discontinuity in the zigzag sampler.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 229

Survey Properties Dialog
The Survey Properties dialog allows you to define the survey methods and where
the survey data are located. It is used mostly when setting up survey objects as
part of data analysis in Distance – see Working with Surveys during Analysis in
Chapter 7 of the Users Guide for more information.

A standard survey object is automatically created by the Setup Project

Wizard, if you tell the wizard that you want to setup the project ready for
analysis, so it is often not necessary to edit the survey properties, unless you are
doing complicated analyses, or require multiple survey objects.

The dialog is accessed by clicking the Properties… button on the Survey

Details Input tab. It is composed of three tabs:
• Survey methods
• Data layers
• Data fields
In addition there are three buttons at the bottom of the page:
• Defaults resets all options on all tab pages to the Distance
defaults
• Ok saves any changes and closes the dialog
• Cancel closes the dialog without saving changes
Before saving any changes, Distance checks to see if the Survey is used by any
analyses. If it is, and these analyses have results associated with them, Distance
will show the Confirm Change dialog.

Survey Methods Survey Properties Tab

See Survey Properties Dialog for an overview.
Options
• Type of survey – can be either a Line Transect, Point Transect or
Cue Count survey.
• Observer configuration – can be either single or double
observer. For more on double observer surveys, see Chapter 10 -
Mark Recapture Distance Sampling.
• Distance measurements – that is the type of distances that
were measured in the field. For line transects this can be
perpendicular distances or radial distances together with the angle
of the object relative to the trackline. For point transects and cue
counts only radial distances are measured.

If the distances were collected in intervals (bins), rather than as

exact distances, you should read the Users Guide section on Interval
(Binned/Grouped) Data in Chapter 8.

• Observations – this is whether recorded observations were of

single individuals or clusters of individuals. (See also Clusters of
Objects in Chapter 8 of the Users Guide)

230 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 • Sampling fraction – this option is only here for
backward compatibility with previous versions of Distance where
sampling fraction was specified as a survey property. We now
recommend you specify sampling fraction using a multiplier – see
Multipliers in CDS Analysis in Chapter 8 of the Users Guide for
details. The appropriate multiplier field can be created during the
Setup Project Wizard, or you can create it manually.

Data Layers Survey Properties Tab

See Survey Properties Dialog in the Program Reference for an overview of the
Survey Properties dialog.
Here, you specify which data layers relate to this survey. If the survey has been
completed, choose the observation layer that contains the survey data. If the
survey is planned, but not completed, choose the lowest sample or subsample
layer that contains the planned survey effort.
When you choose the lowest data layer, the parent layers are displayed in a table.
You must then go on to define the Data fields in the next tab.

Data Fields Survey Properties Tab

See Survey Properties Dialog in the Program Reference for an overview of the
Survey Properties dialog.
Here, you specify the role of the fields in your survey. You need to fill in the
table to tell Distance which fields correspond to the following roles for this
survey:
• Area – the area of each stratum. If this is set to [None] then
density can be estimated, but not abundance.
• Effort – the line length for line transects, or number of times each
point was visited for point transects. (You can set this to [None]
for point transects.)
• Perp distance – the perpendicular distance of each observation,
if applicable
• Radial distance – the radial distance of each observation, if
applicable
• Angle – the angle of each observation, if applicable
• Cluster size – the field containing cluster size, if applicable
Depending on the type of survey, some of these will be greyed out in the table –
for example if the survey is a point transect then the Perp distance and Angle
fields will be greyed out, as they are not applicable.

If you used Distance 3.5, then this tab plays a similar role to the
Modelling Types in Distance 3.5. The current setup is more flexible, however,
because you can define multiple surveys within a project.

Data Filter Properties Dialog

The Data Filter Properties dialog allows you to view and edit the properties for a
data filter. For more information about data filters, and how they are used in
analyses, see the pages in Chapter 7 of the Users Guide on Analysis
Components.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 231

 The dialog is divided into four tab pages:
• Data Selection
• Intervals
• Truncation
• Units
In addition there are three buttons at the bottom of the page:
• Defaults resets all options on all tab pages to the Distance
defaults
• Ok saves any changes and closes the dialog
• Cancel closes the dialog without saving changes
Before saving any changes, Distance checks to see if the Data Filter is used by
any other analyses. If it is, and these analyses have results associated with them,
Distance will show the Confirm Change dialog.
You can change the name of the Data Filter by typing a new name into the Name
text box. This name is saved when you press the Ok button.

Data Selection Tab

See Data Filter Properties Dialog of the Program reference for an overview of
the Data Filter Properties dialog.
This tab page allows you to select a subset of your data for analysis. This feature,
when combined with the ability to add extra fields to the dataset, is extremely
powerful because it effectively lets you keep different subsets of your study data
(species, years, etc) in the same Distance Project. You could, for example,
define a separate Data Filter to select each subset separately, but in your analyses
use the same set of Model Definitions. Other possible examples of the use of
data selection is given in Chapter 8 of the Users Guide within the sections
entitled: Stratification and Post Stratification and Multipliers in CDS Analysis.
To activate the data selection criteria options, click on the + button and choose
the data layer type that you want to define criteria for.

You choose from data layer types, rather than layer names at this
stage because it is only when running the analysis that Distance combines your
data filter with a survey object to find out which data layers are to be used in that
run.

You then type the selection criteria in the space to the right of the layer type.

If you need more while editing a edit or view a long selection criterion,
click on the line you want to see more of and press SHIFT-F1 (i.e., the shift and
F1 keys) to open the Data Selection Zoom Dialog.

Selection criteria format

It is important to get the selection criterion exactly right, otherwise the analysis
will not work. The format is:
FieldName Operator Value

• FieldName is the name of the field that the criterion applies to. If
the field contains any spaces or punctuation then put it inside
square brackets – e.g., Cluster size becomes [Cluster size]

232 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 • Operator is the type of logical operator to perform – valid
operators include =, <>, >, >=, <=, IN, LIKE, BETWEEN
• Valueis the values the operator applies to – for example 20 (a
number) or ‘1’ (a text value – put it in quotes)

If you’re a database geek, you’ll recognize the format as it is the

same as that in a SQL WHERE statement.

For example, in the Stratify example project, if we wanted to select only

observations with cluster size 1, we would define a criterion on the observation
data layer:
[Cluster size] = 1

The IN operator is a useful way of selecting on multiple values, e.g., to

select cluster sizes 1, 2 and 3, you could say:
[Cluster size] IN (1,2,3)
Another useful operator is BETWEEN, as in
[Cluster size] BETWEEN 1 AND 3

Selection on multiple data layers

You can define criteria on different layers – for example to select cluster size 1
but only from the Ideal habitat stratum, you would define a second criterion (by
clicking the + button again), and this time choose the stratum layer type. The
second criterion would be
Label = ‘Ideal habitat’

Complex selection criteria

You can join selection criteria on the same layer using up to 40 logical operators
AND, OR and NOT, together with brackets if necessary. Examples of more complex
criteria on some fictional project might be:
[Cluster size] >= 3 AND Beaufort IN (0,1,2,3)
Species = ‘SOSP’ AND (Observer = ‘LT’ OR Observer = ‘KBV’)

You can also manipulate the data using simple functions, such as:
• string functions LEFT, RIGHT, MID
• numerical functions INT, ROUND
For example:
LEFT (Observer,1) = ‘L’
INT(Distance)=0

Unfortunately, we haven’t been able to find a complete reference for which

functions are and are not allowed – we will update this section when we do!

Intervals Tab
See Data Filter Properties Dialog of the Program reference for an overview of
the Data Filter Properties dialog.
On this page, you specify whether you want your distance data analyzed as
interval data (as opposed to exact measurements).
You would use this option under two circumstances:
• Your data were collected in intervals. In this case you would set the
intervals here in the default Data Filter and leave them the same for

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 233

 all analyses. You should read the Users Guide section on Interval
(Binned/Grouped) Data in Chapter 8, for more information on
dealing with interval data within Distance.
• Your data were collected as exact measurements, but you wish to
analyze them as interval data.

Do not use this option if you want to analyze the data as exact,
but want to specify intervals for the goodness-of-fit tests. You do this in the
Diagnostics page, under the Detection Function tab the Model Definition
Properties.

To specify intervals for the analysis:

• Click on the "Transform distance data into intervals for analysis"
check box.
• Choose the number of intervals.
• If your intervals are evenly spaced, you can choose the Automatic
equal intervals option. You then only need to type in the lowest and
highest cutpoints. Distance will fill in the others automatically.
• If your intervals are not evenly spaced, choose the Manual and type
in the interval cutpoints.

When you select interval data on this tab page, your truncation
options change on the Truncation tab page. By default, the data are truncated at
the upper and lower cutpoints you have selected. See the Truncation Tab page of
the Program Reference for more details.

Selecting the Automatic equal intervals option will give you less
flexibility in choosing truncation distances in the Truncation tab. So, even if you
have evenly spaced cutpoints, it is often better to use the Automatic equal
intervals option to speed entering the cutpoints (this way you only have to type
in the lowest and highest cutpoints), but to select the Manual option before going
on to the Truncation tab page.

The Automatic equal intervals option is essential when you have a

large number of intervals. This is because Distance stores Manual cutpoints in a
list, and this list can not be more than 180 characters long. (If you try and enter
Manual cutpoints and the list becomes too long, Distance will give you a
warning message.) For Automatic cutpoints, Distance only has to store the
highest and lowest cutpoint, and number of intervals - so the list is much shorter.
The maximum number of cutpoints you are allowed in either case is 30.

What happens on the cutpoint boundaries? Observations that are

exactly on the lowest cutpoint boundary are included in the lowest interval.
Thereafter, observations that are exactly on a boundary are put in the lower
interval band.

234 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Truncation Tab
See Data Filter Properties Dialog of the Program reference for an overview of
the Data Filter Properties dialog.
The Truncation tab is where you choose the level of truncation for your distance
data.
Exact distances
This is how the tab looks if you have exact distances i.e., you did not choose to
transform your data into intervals on the Intervals tab:

Data Filter Truncation tab when using exact distances

When analyzing distance data, it is common practice to use Right Truncation to

remove observations a long way from the track line. In Distance, you can right
truncate by discarding a certain percentage of the observations with the largest
distances or you can choose a fixed distance for the truncation. We recommend
experimenting with running analyses at variety of truncation distances and
examining the resulting fit in the Detection Probability Plot and Chi-square
Goodness of Fit pages of the Analysis Details Results.

In Distance 3.5 and earlier, when truncating a percentage of data in

analyses where the detection function was stratified, the truncation point was
calculated separately for each stratum. It is now calculated using all the data,
and the same truncation point is applied in all strata. This is more consistent
with the other truncation options.

Left truncation (i.e., discarding observations at less than a given distance) is less
common. In Distance you can choose a fixed distance for the left truncation.

Left truncation is usually used in CDS when detection on the line is

not certain, but detection probability is higher some distance away from the line.
Examples include the case where animals close to the line are freeze or hide so
are likely to be missed, or aerial surveys where it is hard to see right under the
plane. When you use Distance to left truncate, only data beyond the left
truncation distance is used, so the detection function is fit only to these data and
is extrapolated back to distance zero. The estimated detection probability at the
left truncation distance is often less than 1. An alternative analysis method is
appropriate when you are willing to assume that detection probability is 1 at the
left truncation distance. In this case, you can simply subtract the left truncation
distance from the observed distances before importing the data into Distance –
effectively moving zero distance out to your left truncation line. In this case,
there is no need to specify left truncation within distance – you’ve already done
it before importing the data.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 235

 If your observations are clusters, rather than individual animals, you may want to
choose a different truncation distance for estimating cluster size. This is
especially true if you are running an analysis using the Mean of observed
clusters option in the Cluster size tab of the Model Definition Properties. If
your observations are not clusters of objects, the cluster size truncation options
will be greyed out (see picture).

The cluster size options are only relevant for CDS analyses, and
MCDS analyses where cluster size is not a covariate.

Interval data - Manual intervals

When you have specified manual Intervals on the Intervals tab page, the
Truncation tab will look something like this:

Data Filter Truncation Tab when data have been transformed into intervals in the
Intervals tab

By default, Distance will right truncate all observations that fall beyond your
upper interval cutpoint and left truncate all observations that fall within your
lower interval cutpoint. If you want to truncate further, you can right or left
truncate at any of the interval cutpoints by selecting from the drop-down lists.
For cluster sizes, Distance allows you to choose either the same truncation as
above, or to choose from one of your cutpoints.
Interval data - Automatic intervals
If you have specified automatic intervals in the Intervals tab page, then the right
and left truncation are set to the upper and lower cutpoints that you specified and
cannot be changed.

Units Tab
See Data Filter Properties Dialog of the Program reference for an overview of
the Data Filter Properties dialog.
Using the Units tab, you can convert between that the data are stored in (as
specified in the Data Explorer), and the units for reporting analysis results. You
can, if you want, report results in one unit of area (say) in one analysis and a
different set of units in another.

Model Definition Properties Dialog

The Model Definition Properties dialog allows you to view and edit the
properties for a model definition. For more information about model definitions

236 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 and how they are used in analyses, see the pages in Chapter 7 of theUsers Guide
on Analysis Components.
At the top of the dialog, you choose the analysis engine – either CDS
(conventional distance sampling – see Users Guide, Chapter 8 - Conventional
Distance Sampling Analysis), MCDS (multiple covariates distance sampling –
see Users Guide, Chapter 9 - Multiple Covariates Distance Sampling Analysis)
or MRDS (mark-recapture distance sampling – see Users Guide, Chapter 10 -
Mark Recapture Distance Sampling).
The contents of the dialog change depending on the analysis engine chosen. The
CDS and MCDS analysis engines have very similar options, so are grouped
together in the pages that follow. The MRDS engine has somewhat different
options so is dealt with separately below.
There are three buttons at the bottom of the dialog window:
• Defaults resets all options on all tab pages to the Distance
defaults
• Ok saves any changes and closes the dialog
• Cancel closes the dialog without saving changes
Before saving any changes, Distance checks to see if the Model Definition is
used by any other analyses. If it is, and these analyses have results associated
with them, Distance will show the Confirm Change dialog.
You can change the name of the Model Definition by typing a new name into the
Name text box. This name is saved when you press the Ok button.

Model Definition Properties - CDS and MCDS

Both the CDS and MCDS analysis engine have the same Model Definition
properties tabs:
• Estimate
• Detection Function
• Cluster Size
• Multipliers
• Variance
• Misc.
In the following pages, we discuss the contents of these tabs, highlighting any
differences between the engines.

Estimate Tab - CDS and MCDS

See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
On the Estimate page you define the stratum and sample layer to use in the
analysis, and tell Distance which quantities to estimate. The various options are
outlined here, and are also discussed in Chapter 8 of the Users Guide in the
section on Stratification and Post-stratification.
Stratum definition
Here, you specify the level of stratification to use in the analysis. For details, see
Stratification and Post-stratification in Chapter 8 of the Users Guide.

You specify layer types rather than layer names at this stage because
it is only when the analysis is run that the survey object is used to select the data

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 237

 layers for the analysis. If you select a field for post-stratification that is not in
the layers used in the analysis, an error will result.

Sample definition
Here, you specify which sample or sub-sample layer to use as the sample, for
determining encounter rate variation and for bootstrapping. For more
information see Chapter 8 of the Users Guide, the section on Sample Definition
in CDS Analysis.
Quantities to estimate and level of resolution
These options define which quantities you wish to estimate, and at what level. If
you have selected No stratification in the Stratum definition section then the
Stratum column will be greyed out. Also, if your observations are individual
objects, not clusters, then the Cluster Size row will be greyed out. Lastly, if
you are doing an MCDS analysis, and have cluster size as a covariate, the
options here will look different (see below).
If you wish to estimate density, you should first check the boxes at the levels for
which you want density estimates. In most cases you will not have enough
observations in each sample to estimate density by sample, but this is not always
true. The lowest level of density dictates the level of estimation for encounter
rate, and the lowest level for estimation of the detection function parameters and
cluster size. After you have selected the level to estimate density, you can then
select one level to estimate the detection function and cluster size (if applicable).

In exploratory analyses, you may not wish to estimate density. For

example, you may only be interested in investigating the fit of various detection
functions at the global level, and not be interested in estimating density until you
have found a satisfactory fit. In this case, uncheck all of the density, encounter
rate and cluster size boxes, like this:

Notice that the restrictions on the level of resolution of estimates are removed
when you are not estimating density.

If you are estimating density by stratum and also globally, you need to tell
Distance how to combine the stratum estimates together to make the global
estimate. For geographic strata, use the default settings:

If your strata are not spatial strata, for example time periods or different sections
of the population, then you should consider using the other options here;
however we recommend entering non-spatial data as columns in the appropriate
data layer and then using the Post-stratification feature to do the analysis
(Chapter 8 of the Users Guide in the section on Stratification and Post-
stratification has more information and further details of the scenarios when each
of the following options are appropriate).
For reference, the four possible options are:
• Global density estimate is Mean of stratum estimates, weighted by
Stratum area.

238 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 This option is appropriate in the usual case where the strata are
geographic strata. Here density is calculated as a weighted mean of
the stratum estimates, weighted by the area of each stratum.
Abundance is calculated as density multiplied by the sum of the
stratum areas. Variance is a weighted mean of the stratum
variances, although the exact formula depends upon which
components (detection function, encounter rate, cluster size) were
estimated by stratum and which globally - see Buckland et al. 2001
section 3.7.1.
• Global density estimate is Mean of stratum estimates, weighted by
Total effort in stratum.
This option is appropriate when the strata are effort related, such as
different observers or time periods. Here density is calculated as a
weighted mean of the stratum estimates, weighted by the sum of
the survey effort in each stratum. Abundance is calculated as
density multiplied by the area of the first stratum (in almost all
cases the strata represent the same area – for example when the
strata are different observers surveying the same survey region). In
this case the Strata are replicates tick box is enabled. The
formula for variance depends if this box is ticked
• Strata are replicates ticked.
Here, strata are seen as a random sample from a larger
population of possible strata. An example would be if the
strata represent survey days chosen at random (or
systematically) over a year, and the inference is about the
average density of animals over the whole year. Variance is
calculated using the variation in density between strata
(weighted by effort); see equations 3.84 - 3.87 in Buckland et
al. 2001, treating stratum as a sample.
• Strata are replicates not ticked.
Here inferences are restricted only to the strata surveyed. An
example would be if the strata represent two survey vessels
that surveyed the same area and we wish to make inferences
about mean density in the area over the two vessels. Here
variance is calculated as an effort-weighted average of the
stratum variances, using the methods outlined in section 3.7.1
of Buckland et al. 2001, substituting total line length for the
area weighting terms (Av and A).
• Global density estimate is Sum of stratum estimates.
This option is appropriate when the strata represent different
components of the population, such as male and female animals,
and we want a combined estimate of overall density. The stratum
density estimates are summed across strata, abundance is the
density multiplied by the area of the first stratum (since all strata
should have the same area), and the variance is calculated as in
Section 3.7.1 of Buckland et al. 2001, dropping the area weighting
terms (Av and A).
MCDS analyses

For MCDS analyses, it is possible to fit the detection function at one level and
estimate at a lower level. For example you can fit a global detection function
model, but estimate average f(0) and probability of detection separately for each
stratum. For details of why this may be useful, see Estimating the detection
function at multiple levels in Chapter 9 of the Users Guide.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 239

 To implement this, you simply check the Levels to estimate boxes for detection
function at both the level you want to fit the model (the higher level) and the
level you want to estimate average f(0) (the lower level).
For example, imagine you want to estimate density by stratum, but don’t have
enough data to fit a separate detection function for each stratum. One solution is
to fit a global model for the detection function, but using stratum (or lower) level
covariates. You can then use the fitted model to estimate a separate average f(0)
(or h(0)) in each stratum, using the covariates that apply to that stratum. To
implement this, you define the appropriate covariates in the Covariates page,
and set up the Quantities to estimate on the Estimate page as follows:

Estimating detection function at multiple levels. Note that the detection function boxes
are checked both at the global and stratum levels.

MCDS analyses with cluster size as a covariate

When cluster size is a covariate, the options on this page change:
• Stratum definition. The engine currently does not allow for
stratification when cluster size is a covariate, so the stratification
options are disabled.
• Sample definition. Because variance is not estimated
analytically, this section is now only used when bootstrapping to
estimate variance.
• Quantities to estimate. Variance of encounter rate is no longer
estimated, so this line is removed, and cluster size is now for output
only.
For more information see Cluster size as a covariate in Chapter 9 of the Users
Guide.

Detection Function Tab - CDS and MCDS

The Detection Function tab is divided into five pages:
• Models - Detection Function Tab - CDS and MCDS
• Adjustment terms - Detection Function Tab - CDS and MCDS
• Covariates - Detection Function Tab - MCDS (only in the MCDS
engine)
• Constraints - Detection Function Tab - CDS and MCDS
• Diagnostics - Detection Function Tab - CDS and MCDS

Models - Detection Function Tab - CDS and MCDS

See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
The Models page is the first that comes under the Detection Function tab. It is
probably the most important of all the pages in the Model Definition because it
is where you tell Distance what model to use to estimate the detection function.
In almost all cases you will have one detection function model. Select from the
drop down list to choose the Key function (Hazard-rate, Half-normal, Uniform
or Exponential for CDS analyses, Hazard-rate or Half-normal for MCDS

240 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 analyses) and Series expansion (Cosine, Simple polynomial or Hermite
polynomial).

To choose a key function with no series expansion, you will need to

use manual adjustment in the Adjustment terms page and select number of
adjustment parameters to be 0.

The only situation where we recommend you select more than one detection
function model is where you are using bootstrapping to incorporate model
selection uncertainty in your variance estimate (see Model Averaging in CDS
Analysis, in Chapter 8 of the Users Guide). In this case, use the + button to add
more detection function models, and select from the options under Selection
among multiple models using. You can choose either AIC, AICc or BIC.

Users of Distance 3.0 and earlier will be used to regularly selecting

more than one model in the ESTIMATE section of the command file, and using
AIC to select among them. In the current software, we recommend you assign
each model to a separate analysis, and then compare the models in the Analysis
Browser. This way you have access to Analysis Details for each of the models
for example you can look at the Detection Probability Plots in the Results section
of the Analysis details.

Adjustment Terms - Detection Function Tab - CDS and

MCDS
See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
This Adjustment terms page is the second that comes under the Detection
Function tab. This is where you tell Distance what methods to use to fit the
series expansion terms (Adjustment terms) to the data in the detection function
modeling. You can also specify starting values for the estimation procedure, for
both the key and adjustment function parameters, and the method for scaling
distances when calculating the adjustment terms..
Selection of adjustment terms
Automated selection
Click the Automated selection option for automated selection of adjustment
terms. Distance then uses data-based criteria to choose the number and order of
adjustment terms required for the analysis.
The selection methods are
• All This option examines all possible combinations of a limited
number of adjustment terms. If z is the maximum number of
parameters and k is the number of parameters in the key function,
then there are 2 to the power of z-k combinations of the adjustment
terms. Each model is fitted to the data and the model with the
smallest value according to the selection criterion is selected.
• Sequential selection This option considers a subset of models
with different combinations of adjustment terms. A sequence of
models is considered, with one adjustment term being added at
each step of the sequence. The sequence of models can be
represented as
M1 key function with no adjustment terms
M2 key function with 1 adjustment term
M3 key function with 2 adjustment terms

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 241

 …
Mv key function with v-1 adjustment terms
The selection criterion (stopping rule) is either based on likelihood
ratio test or minimizing AIC, AICc or BIC. Model Mt is chosen if
there is no model in the sequence Mt+1, ..., Mt+k which provides a
significantly better fit as determined by the stopping rule. The
value in the Look-ahead field determines the length(k) of the
sequence of models that is examined before choosing model Mt.
Adjustment terms are added sequentially based on the order of the
term. For polynomial adjustment functions, the order of the
adjustment term is the exponent of the polynomial. Terms are
added in the following sequence: (xt, xt+2,...). For cosine
adjustments, cosine terms are added in the following sequence:
(cos(tpixs), cos((t+1)pixs),...). The beginning value, t, is determined
by the shape of the key function. xs, the scaled perpendicular
distance, is defined at the bottom of this page.
• Forward selection Forward selection only differs from
sequential selection in the choice of which adjustment term is
included at each step in the sequence. Forward selection adds one
term at a time, but not necessarily in sequential order. For each
model in the sequence, each term not already in the model is added
and the adjustment term which increases the likelihood the most is
chosen as the term to add. With forward selection it is possible to
select models that cannot be selected with sequential selection. For
example, the following model might be chosen with forward
selection:
f(x) = (1/w)(1+ a1cos(pixs)+ a2 cos(3pixs))
However, with sequential selection, the adjustment term cos(3pixs),
could not be added without first adding the adjustment term
cos(2pixs).
Manual Selection
Choose this option if you want to specify the number and optionally the order of
the adjustment terms yourself. The number of the models on in the tables on this
page correspond to the models chosen on the previous Models page (normally
just 1).
When you type in the number of adjustment parameters, the appropriate number
of cells under “Order of adjustment parameters” become editable. Leave these
cells blank if you want to specify the number of adjustment parameters but not
the order.

To specify a detection function model with just a key function and no

adjustment parameters, set the number of adjustment parameters to be 0.

You can also select starting values for the key function and adjustment terms. To
do this, check the box Manually select starting values, and enter the
number of parameters for each model.
Calculate the number of parameters by summing the number of key function
parameters, the adjustment terms and any covariate parameters. If the detection
function is fit by stratum or sample, sum the number of parameters in each
stratum to get the total. For more information about model parameterization, see
About CDS Detection Function Formulae in Chapter 8 of the Users Guide.

In newer versions of Distance, you specify the starting values

separately for each stratum. In versions 3.5 and earlier, the same starting values
were used in each stratum.

242 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 If you are not sure how many parameters there are in an analysis, run it
first without specifying starting values, and then look in the Parameter Estimates
page of the Analysis Details Results to see how many parameters were used.
This is particularly useful for MCDS analyses, as the number of covariate levels
in non-factor covariates may vary among strata, depending on which covariate
levels occur in each stratum.

Scaling of distances
As explained in Chapter 9, this option is mainly of interest when using the
MCDS engine (although it can be used for the CDS engine too). For all of the
series expansion terms, the scaled distance is used in place of actual distance in
calculating the expansion term values mainly for numerical reasons. There are
two options: scale by w, the truncation distance, or by σ, the scale parameter of
the key function (this does not apply to the uniform key function, which has no
scale parameter). For the CDS engine one will generally want to scale by w, but
for the MCDS engine one may scale by either – see Chapter 9 - Multiple
Covariates Distance Sampling Analysis in the Users Guide.

Covariates - Detection Function Tab - MCDS

See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
The Covariates page is the third under the Detection Function tab. This is where
you specify the covariates to add to the model. This page only appears for the
MCDS analysis engine.
Click on the + button to add a row to the table.
In the first column you select the layer type containing the covariate from the
drop-down list.

This column gives layer types rather than layer names because at
this stage Distance doesn’t know which Survey you’re going to use with this
Model Definition, so it doesn’t know which layer names it can use. This way
you can pair the same Model Definition with many different surveys.

In the second column you select the field name of the covariate.
Tick the box in the third column if the covariate is a factor. Factor, or class
covariates have a finite number of distinct levels. The value of each level is not
significant – for example the factor levels could be text fields “Porpoise”,
“Whale”, “Seal”, or they could be numeric fields 1, 2, 3. If the box in this
column is not ticked, the covariate is assumed to be a non-factor covariate. In
this case, the field must be numeric, otherwise an error will occur when you try
to run the analysis engine. For more about this, see Factor and Non-factor
covariates in MCDS in the Users Guide.
Tick the box in the last column if the covariate is the cluster size field.
Distance needs to know whether any of the fields in your analysis are the cluster
size field because it treats this field a special way in the analysis. For more
information see Cluster size as a covariate in Chapter 9 of the Users Guide.
Some general advice about selecting covariates to include is given in the CDS
Analysis Guidelines section of Chapter 8 - Conventional Distance Sampling
Analysis in the Users Guide.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 243

 Constraints - Detection Function Tab - CDS and MCDS
See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
The Constraints page comes under the Detection Function tab. This is where you
specify constraints on the fitting procedure. In most cases, it is sufficient to use
the default settings. Other situations are discussed below
Constraints on the shape of the fitted function
This option allows you to choose the level of constraints on the shape of the
fitted detection function. The estimators are constrained by default to be strictly
monotonically non-increasing – i.e., the detection curve is either flat or
decreasing as distance increases from 0 to the truncation distance. In some
instances, depending on the tail of the distribution, this can cause a poor fit at the
origin (distance = 0), or can cause lack of convergence.
In these cases there are two options: (1) use the data filter to truncate the
observations in the tail, or (2) relax the constraints. The weak constraint option
allows the curve to go up and down as it fits the data but will not let the curve
dip down at the origin. In some instances this will achieve a better fit at the
origin, which is the point of interest. With no constraints, the curve can take any
form, except it must remain non-negative.
Monotonicity is achieved by constraining the function at a fixed set of points. In
some circumstances it is possible that the curve can be non-monotone between
the fixed points. Typically, this results from trying to over-fit the data with too
many adjustments with a long-tailed distribution. In this case, use a Data Filter
to truncate the data, or constrain the number of adjustment terms in the
Adjustment Terms tab.

You cannot apply constraints in the MCDS engine. This is because

it uses a different fitting algorithm, one for which we have not implemented
constraints on the maximization routine.

Bounds on Key Function Parameters

With some datasets, it may be necessary to bound the parameter estimates to
achieve convergence. In these situations, you can use this table to specify upper
and lower bounds on the key function parameters (one parameter for half-normal
and negative exponential key functions, two for the hazard rate function, plus
any covariate parameters). One common circumstance where this is required is
to impose a lower bound of 1.0 on the second hazard rate parameter.

In newer versions of Distance, you specify bounds separately for

each stratum. In versions 3.5 and earlier, the same starting values were used in
each stratum. To calculate the number of key function parameters, sum across
strata (including any covariates). For more information about model
parameterization, see About CDS Detection Function Formulae in Chapter 8 of
the Users Guide.

If you are not sure how many key function parameters there are in an
analysis, run it first without specifying starting values, and then look in the
Parameter Estimates page of the Analysis Details Results to see how many
parameters were used. This is particularly useful for MCDS analyses, as the
number of covariate levels in non-factor covariates may vary among strata,
depending on which covariate levels occur in each stratum.

244 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Diagnostics - Detection Function Tab - CDS and MCDS
See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
This page is the last under the Detection Function tab. On this page you specify
the diagnostic output that you want Distance to produce, including the intervals
for detection probability plots and goodness-of-fit tests, whether to output qq-
plots and associated statistics for non-interval data, and the filename if you want
Distance to save the plot file.
Chi-sq goodness-of-fit tests and histogram intervals

This option is only relevant if you are analyzing the data as exact
distances. If you to analyze your data as intervals (by selecting Intervals in the
Data Filter) then a single goodness-of-fit test is performed using those intervals.
Any options you set under Intervals on this tab page are ignored.

Automatic selection of intervals

By default, when the data are analyzed as ungrouped, three sets of intervals are
constructed with equally spaced cutpoints and the number of intervals being m,
2/3m and 3/2m, where m is the square root of the number of observations.
Manual selection of intervals
Instead of using the automatically generated intervals, we recommend that you
routinely select your own. To do this, choose Manual selection of intervals,
and enter the number of intervals in the text box provided. You then have two
choices to specify the interval cutpoints:
• Automatic equal intervals. If you choose this option, and you
leave the first and last cutpoint values in the Cutpoints table as 0,
then Distance will generate goodness-of-fit tables with evenly
spaced cutpoints between the left truncation distance and the right
truncation distance. Alternatively, you can specify the first and last
cutpoints in the Cutpoints table, and Distance will generate
evenly spaced cutpoints between these limits.
• Manual. This option allows you to specify uneven cutpoints – for
example you may want smaller intervals close to zero distance (to
look for evidence of a shoulder) or perhaps cutpoints that isolate
favoured distances (to look for evidence of rounding). Enter the
cutpoint values you want in the Cutpoints table. You will
normally set the first cutpoint to the left truncation distance, and
the last one to the right truncation distance.

A good way to get to know your data when you begin analyzing it is
to define a large number of intervals (say 15-20), fit any arbitary model, and then
examine the output histogram for evidence of evasive movement, heaping,
outliers, etc. (you can ignore the model fit for now). See CDS Analysis
Guidelines in Chapter 8 of the Users Guide for more on this.

The Interval cutpoints options can make it easier for you to enter
manual intervals. For example, you can enter the left and right truncation points
in the first and last cutponts rows, and then click on Automatic equal
intervals to have the intermediate cutpoints set. Then go back to Manual and
customize the cutponts to your requirements.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 245

 K-S test goodness-of-fit test and qq plots

This option is only relevant if you are analyzing the data as exact
distances.

Qq plots are a graphical technique for assessing the adequacy of the fit, and the
associated test statistics (Kolmogorov-Smirnov and Cramér-von Mises) test
goodness-of-fit for exact data. For more about these outputs, see CDS Qq-plots
and CDS Goodness of fit tests in Chapter 8 of the Users Guide.
Since these outputs can take a while to generate for large datasets, there is an
option here to turn them off. Qq plots have one plotted point per observation, so
for large datasets it is better to plot only a subset of points. By default, the
maximum number of points to plot is 500, but that can be changed here.
Entering 0 under Maximum num points in qq plots means that all points
are plotted, regardless of how many there are.
Plot file
If higher quality graphical output is required, Distance can save the histogram
data to a file that can then be imported into any graphics or statistics package.
Check the Create file of histogram data option and choose the file name using
the Browse button.
The output format of the file is described in the Users Guide page Saving CDS
results to file in Chapter 8.

You can copy and paste the high quality plots produced by Distance
straight into most word processor and spreadsheet packages. In addition, you
can easily paste the plot data into a spreadsheet and re-create the plot that way.
See the Analysis Details Results Tab help page.

Cluster Size Tab - CDS and MCDS

See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
The cluster size page is used to modify the way the cluster sizes are used in the
estimate of density.
By default, distance uses a regression of observed cluster size against distance to
estimate the average population cluster size (as the expected value at distance of
0). This approach is intended to reduce bias if there is a tendency for smaller
clusters to be missed more than large clusters at large distances from the track
line. The exact regression method is chosen by the option in the Size-bias
regression method box.
A second alternative if cluster size bias is not anticipated to be a problem is to
use the mean of the observed clusters as an estimate of the average cluster size in
the population. Truncation becomes particularly important if you are using this
option (see note below).
Thirdly, you can apply a statistical hypothesis test to the regression of cluster
size on distance and only apply the size bias method if this regression is
statistically significant.
Fourthly, in MCDS analyses, you can use cluster size as a covariate in the
detection function. Doing this makes the options on this page un-necessary, and
the Cluster size tab is therefore disabled. For more information see the section
in Chapter 9 of the Users Guide entitled: Cluster size as a covariate.

246 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Another important cluster size option is the truncation of distances
for cluster size calculations. This is set in the Truncation tab of the Data Filter
Properties window.

Multipliers Tab - CDS and MCDS

See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
In this page you tell Distance which multipliers to use to scale the density
estimate. Multipliers are discussed in more detail in Chapter 8 of the Users
Guide under Multipliers in CDS Analysis.
To add a new multiplier to the list, press the + button. The column “Layer type
containing multiplier” is set to the Global data layer and cannot be changed in
this version of Distance. You should choose the multiplier you want from the
list of multiplier fields that pops up when you click on the "Fields containing
multiplier value" column. If this multiplier has a standard error associated with
it, choose the appropriate Multiplier SE field from the list that pops up when you
click on the “Field containing multiplier SE” column. Note that you don't have
to select an SE if your multiplier value is known without error. Similarly, if you
know the degrees of freedom (DF) associated with the multiplier, click on the
“Field containing multiplier DF” column. You don't have to select a DF; if you
do not select one then Distance assumes the DF for this multiplier is infinity.
(Another way to tell Distance that DF is infinity is to select a field from the
global layer that has a value of 0.0.) Lastly, you must tell Distance what
operator to use: i.e., whether to divide or multiply the density estimate by the
multiplier value to obtain the final estimate.
To remove the most recently added multiplier press the - button.

If the multiplier represents cue rate in a cue count analysis, tick the
“Cue rate” box

You can only add multipliers if you have already created the
appropriate fields in the Data Explorer.

If you used the Setup Project Wizard to define your multiplier fields,
then they will appear automatically in the Multiplier tab in Model Definition
Properties. For these fields, Distance also knows whether the operator is * or /
(i.e. whether to multiply or divide the density estimate).

Variance Tab - CDS and MCDS

See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
In the Variance page you specify the methods of calculating the variance of the
density estimate. For the analytic variance estimate, you can choose the method
for calculating the encounter rate variance component. You can also tell Distance
to calculate a bootstrap variance estimate, and specify the exact methods to use.
Analytic Variance Estimate

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 247

 The analytic variance estimate is usually calculated automatically. It
is not calculated in MCDS analyses when cluster size is a covariate or when the
detection function is fit at one level and estimated at a lower level.

The encounter rate variance can be calculated in three ways (see Buckland et al.
2001 section 3.6.2, and look in the book index under Poisson variance of n)
• Estimate variance empirically. This is usually the best option the
variance is calculated from the variance in observations between
samples. However this is unreliable when there are few samples.
• Assume distribution of observations is Poisson.
• Assume distribution is Poisson, with overdispersion factor b.
Setting b to 1 is equivalent to the previous option..
Bootstrap Variance Estimate
Check on the box Select non-parametric bootstrap to tell Distance to
estimate the variance from bootstrap resamples of the data. When you run an
analysis with this option checked in the Model Definition, the bootstrap results
are given at the end of the Results tab on the Analysis Details. Distance reports
bootstrap confidence limits using two methods: firstly using the bootstrap
estimate of variance but assuming that the distribution of the density estimate is
log-normal; secondly using the bootstrap percentile method (i.e., gives the
appropriate quantiles of the actual bootstrap estimates). Distance also reports the
mean of the bootstrap point estimates.
Each bootstrap resample is made up by sampling with replacement an equal
number of units at the level you specify. For example, if you specify to resample
samples (see below), then each bootstrap resample is made up of the same
number of samples (line or point transects) as your original sample, chosen
randomly with replacement from the original sample. Note that for line
transects, this means that the survey effort (total line length) will differ in each
resample. Note also that each of your original samples has an equal probability
of appearing in the resample (an alternative, which we do not implement, would
be to have probability proportional to line length).

You can add a column for the bootstrap CV and confidence limits in
the Analysis Browser using the Analysis Browser Column Manager.

The options under Levels of Resampling change, depending on whether you

have stratification turned on or not in the Estimate tab.
With No stratification you can resample samples and/or resample
observations within samples. In this case you will normally want to select
Resample samples alone. The resampling observations option is included for
completeness but its routine use is not recommended, and can only be expected
to produce reasonable results if the number of observations per sample is
reasonable (e.g.., > 15).

If your survey includes stratification, but you want to resample across

strata then choose No stratification in the Estimate tab, and then choose
Resample samples in the Bootstraps Levels of sampling box.

With Stratification or Post-stratification enabled in the Estimate tab, you

can resample by strata, samples or observations within samples. Normally, you
will want to resample samples within strata. Resampling strata in addition could
be useful if density is estimated by stratum or if sampling was stratified a priori.

248 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 If you are estimating variance using the bootstrap, be aware that
the variance due to any multipliers is not included in the bootstrap variance. For
more on multipliers, see Multipliers in CDS Analysis in the Users Guide.

Bootstrap Statistics File

Distance can save a file of summary statistics for each bootstrap iteration. This
can be useful if you are having problems with the bootstrapping option. Check
the Create file of statistics option and choose the file name using the Browse
button.
For more information about the format of this file, see Saving CDS results to file
in Chapter 8 of the Users Guide.
One reason to use bootstrapping is for multi-model inference – see Model
Averaging in CDS Analysis in Chapter 8 of the Users Guide.

Misc. Tab - CDS and MCDS

See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
This page contains miscellaneous Model Definition options regarding the
presentation of results, saving of output to files, and options for smearing
distance data before analysis.
Presentation of Results
Use the confidence intervals option to select the percentile of the two sided
confidence intervals presented in the Results section of the Analysis Details
window.
The option to report results for each iteration of the detection function fitting
routine gives you more information about the fitting process, in the Detection
Function pages in the Results section of the Analysis Details window. Use this
option when there are problems with the fitting algorithm and you want more
information about what has gone wrong. It is checked by default for MCDS
analyses.
Results Files
These options allow you to save the results to files when the analysis is run.
Two files can be saved:
• The results details file is identical to the text that is produced in the
Results tab of the Analysis Details window.
• The results stats file is a compact output of summary statistics that
Distance uses internally to extract data for the Analysis Browser
table.
These files may be useful if you wish to import the results into another package -
for example you could write a spreadsheet macro to parse the files and extract
information into spreadsheet cells. For more information about these files, and
their format specification see the Users Guide page Saving CDS results to file in
Chapter 8.
Smearing
Smearing is an ad-hoc method for dealing with measurement error in line
transect surveys (see Buckland et al. 2001 p 269-271).
The smearing option in Distance is only enabled if the survey is a line transect
with distances collected as radial distance and angle. In addition, you should
only select this option of you are going to pair the Model Definition with a Data
Filter in which you have selected automatic Intervals. (In the Intervals tab of

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 249

 the Data Filter, tick Transform distance data into intervals for
analysis, and also select Automatic equal intervals.)
The smearing angle (phi) is the angle sector around the angle measurement. The
proportion of distance parameter (s) is the proportional sector of distance to use
as the basis for smearing. If an observation is measured at angle a and radial
distance r, it is smeared uniformly in the sector defined by the angle range (a-phi,
a+phi) and distance range (r*(1-s), r*(1+s)). The proportion of the sector
contained in each perpendicular distance interval is summed as an observation
frequency and these non-integer frequencies (grouped data) are analyzed to
estimate detection probability.

Smearing is not supported in the MCDS engine

Model Definition Properties - MRDS

Estimate Tab - MRDS
See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
On the Estimate page you define the stratum and sample layer to use in the
analysis, and tell Distance which quantities to estimate.
Stratum definition
Here, you specify the level of stratification to use in the analysis. For details, see
Stratification and Post-stratification in MRDS in Chapter 10 of the Users Guide.

You specify layer types rather than layer names at this stage because
it is only when the analysis is run that the survey object is used to select the data
layers for the analysis. If you select a field for post-stratification that is not in
the layers used in the analysis, an error will result.

Sample definition
Here, you specify which sample or sub-sample layer to use as the sample, for
determining encounter rate variation - see Sample Definition in MRDS Analysis
in Chapter 10 of the Users Guide for details.
Quantities to estimate
These options determine which quantities to estimate, and with what data.
The first option determines whether the engine should Estimate density /
abundance or not. By default, this option is checked, but there are two
circumstances under which you might want to uncheck it:
• In exploratory analysis, you might want to focus on fitting
detection functions and leave the estimation of density until you’ve
selected the detection function to use. This also saves computer
time, since estimating density can be time consuming for larger
datasets.
• You may wish to use different subsets of the data for estimating
detection function and for estimating density given a detection –
see below.
The second set of options determines how to obtain the Detection function
parameters:

250 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 • The default option (Estimate detection function) is to fit the
detection function using the data in this analysis. This makes sense
most of the time.
• The second option is to Use fitted detection function from
previous MRDS analysis. If you select this option, you must select
the ID of the analysis containing the detection function you wish to use.
For more about this option, and when you might want to use it, see
Using a Previously Fitted Detection Function to Estimate Density in
MRDS in Chapter 10 of the Users Guide (which needs finishing).

For the second option to work, you must have run the analysis
containing the target detection function after un-checking the option in Tools |
Preferences | Analysis | R Software to Remove the new objects that
are created with each run.

When you choose the option to use the fitted detection function from
a previous analysis, you cannot choose any options under the Detection
function tab in the current Model Definition – since you are not fitting a
detection function in this analysis.

Detection Function Tab - MRDS

The Detection Function tab is divided into five pages:
• Method - Detection Function Tab - MRDS
• DS Model - Detection Function Tab - MRDS
• MR Model - Detection Function Tab - MRDS
• Factors - Detection Function Tab - MRDS
• Control - Detection Function Tab - MRDS
• Diagnostics - Detection Function Tab - MRDS

Method - Detection Function Tab - MRDS

See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
On this page, you specify the Fitting method – see Introduction to MRDS
Models in Chapter 10 of the Users Guide for more information about the
methods available. Note that the choice of fitting method affects the options
available in the subsequent pages, particularly the DS and MR model pages.

DS Model - Detection Function Tab - MRDS

See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
The DS model is the probability of one or more observer detecting the object,
given it’s distance and covariate values. On this page you specify the form of
this model - the key function and any covariates that affect the scale parameter.
For more about the form of the DS model, see Defining MRDS Models in
Chapter 10 of the Users Guide. That section also includes a detailed description
of how to specify the formula. Note that if you want covariates in the formula to
be factor covariates, you need to specify them as factors – see Factors -
Detection Function Tab - MRDS.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 251

 MR Model - Detection Function Tab - MRDS
See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
The MR model is the conditional detection function p j|3− j ( y , z ) – the
probability of observer j detecting the object, given that the other observer
(observer 3-j) has detected it and also given its distance and covariate values..
On this page you specify the form of this model. You specify the Class of
model (currently only GLM), the link function (currently only logit), and the
formula for the linear (or additive for GAM) predictor. For more about the
form of the MR model, see Defining MRDS Models in Chapter 10 of the Users
Guide. That section also includes a detailed description of how to specify the
formula. Note that if you want covariates in the formula to be factor covariates,
you need to specify them as factors – see Factors - Detection Function Tab -
MRDS.

Factors - Detection Function Tab - MRDS

See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
On this page, you list the covariates in the DS and MR models that should be
considered as factor covariates. Use a comma-delimited list, and use the same
covariate names that appear in the DS and MR model formulae. For more on
factor covariates, see Factor and Non-factor Covariates in MRDS in Chapter 10
of the Users Guide.

Control - Detection Function Tab - MRDS

See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
In this page, you can set various options to control the way the maximization
routine performs in fitting the detection function models. For more details, see
the section on Fine-tuning an MRDS Analysis in Chapter 10 of the Users Guide.

Diagnostics - Detection Function Tab - MRDS

See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
This page is the last in the MRDS Detection function tab. Here you can choose
whether to plot the detection function histograms and perform the goodness of fit
tests. These are done by default, but you may want to un-check the options to
save time in an analysis, or save disk space in the case of the plots.
We anticipate that further options for setting Chi-square GOF cutpoints will be
available in future versions of this engine.

Variance - MRDS
See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
In the Variance page you specify the methods of calculating the variance of the
density and abundance estimates. The options are:
• Innes et al. (2002) – Based on the empirical variance of estimated
density between samples (the default and preferred option).
• Buckland et al. (2001) – Based on the delta method, using the
empirical variance in encounter rate between samples
• Binomial variance of detection process – Only realistic if the entire
study area was sampled.

252 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Misc - MRDS
See Model Definition Properties Dialog in the Program Reference for an
overview of the Model Definition Properties dialog.
In the Misc page you specify various miscellaneous options.
Presentation of results from density estimation
• Standard output. Gives default output options.
• Extended output. Gives extra output – see MRDS Results
Details Listing in Chapter 10 of the Users Guide for details.

Model Definition Properties – DSM

Analysis Components Window

The Analysis Components window is designed to be a convenient way of
manipulating Data Filters and Model Definitions. The window shows a list of
all Data Filters or Model Definitions in your project. Using buttons on the
toolbar you can create, delete, view and arrange the listed components.

Analysis Components window, showing a list of the Model Definitions in the Ducknest
sample project

The last column of the in the table of analysis contents tells you
whether that component is currently being used in any analyses: “Y” means it is
being used and “N” means that it is not. This is useful because when there are
many components (e.g., many Model Definitions if you have been doing a lot of
analyses), it is easy to loose track of which are being used and which are no
longer required. Also, if you double-click on a “Y”, you get a list of the analyses
that use that component.

Toolbar
• List Data Filters. When this button is selected, the Analysis
Components window shows a list of the all Data Filters in the
project.
• List Model Definitions. When this button is selected, the
Analysis Components window shows a list of all the Model
Definitions in the project.
• New Item. Create a new Data Filter or Model Definition,
based on the one currently selected
• Delete Item. Delete the selected Data Filter or Model
Definition

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 253

 • View Item Properties. Show properties dialog for the
selected Data Filter or Model Definition
• Up. Move the selected Data Filter or Model Definition up one
place. This also
• Down. Move the selected Data Filter or Model Definition
down one place.
For more information, see Analysis Components, in Chapter 7 of the Users
Guide.

Other Windows

About Distance Dialog

This window reports information about the program such as the version number,
program sponsors, authors and program files. The information is divided into a
number of tabs:
• About. This shows the program version and release number. Use
this when citing the program (click on the Citation… button for a
suggested wording for the citation).
• Sponsors. Lists the program sponsors.
• Authors. Lists the program authors.
• Use Agreement. Contains a copy of the program use agreement.
Acceptance of this agreement is a condition of program use.
• Program Files. Lists the files in the Distance program folder and
gives information such as the version number of each file.
In addition, there are the following buttons at the bottom of the window:
• Citation… Click on this to obtain a suggested wording for citing
Distance.
• System Info… Click on this to open a system tool that allows
you to retrieve detailed information about the computer running
Distance.
• OK. Click on this to close the About Distance dialog.

Export Project Dialog

The export project dialog allows you to export projects to another location on
your computer. For more information about why you would want to do this, and
some useful tips, see the Users Guide section on Exporting, Transporting and
Archiving Projects (Chapter 4).
Options
File name: Enter the name of the project or zip file to save to
Save as type: You can export either as a Distance project (i.e., .dst project file
and associated .dat data folder) or as a single .zip archive file.
Exclude the following parts: By default, the entire contents of the project
are exported, but you can optionally exclude the following parts:
• Data layer contents. Checking this option causes the data to be
excluded from the exported project, but the data structure (i.e., the
data layer names and types, the data fields, etc) is retained. You

254 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 typically do this to save space if you are exporting a project to
make a template for future project setup – see the Users Guide
(Chapter 4) on Using an Existing Project as a Template.
• Design and survey results. Checking this option causes all
designs and surveys in the project to be reset – i.e., their status is
set to Not Run and any results are deleted. However, the
specifications (the stuff on the Inputs tab of the Details page) are
retained. You typically do this if you are making a template, or if
you want to save space when transporting a file, and the results will
be easy to re-create.
• Analysis results. Checking this option causes all results in the
project to be reset – i.e., their status is set to Not Run and any
results are deleted. However, the specifications – including all
Data Filters and Model Definitions – are left untouched. You
typically do this for the same reasons as you exclude the design and
survey results.

If you are exporting to a zip archive on a removable disk (e.g., a

floppy disk), and the archive is too big to fit on one disk, Distance will
automatically span the archive across multiple disks. You will be prompted to
insert one disk at a time, until the operation is complete. When opening a
spanned archive from disk, you should insert the last disk first.

Projection Parameters Dialog

The Projection Parameters dialog is displayed by clicking the Parameters…
button beside the projection lists in the Preferences or Project Properties dialogs.
It allows you to set parameters associated with a particular projection.
Projection parameters go with a projection and geographic coordinate system to
make a projected coordinate system.
To set a parameter, tick the Set column and enter the appropriate value in the
Value column.

Create New Layer Dialog

The Create New Layer dialog is accessed from the Data Explorer by clicking on
the Create New Data Layer button [picture here], or from the menu Data |
Create Data Layer….
You are prompted to enter the Layer name, Parent layer name, and Layer
type. The list of Layer types depends on the Parent data layer – for example a
stratum data layer must have a global layer as parent.
For layers of type Coverage, the Properties… button allows you to access the
Grid Properties dialog.

Grid Properties Dialog

The Grid Properties dialog is accessed by clicking the Properties… button of
the Create New Layer dialog, when you are creating a new coverage data layer.
The settings you choose here are used to generate a grid of points used to assess
probability of coverage for survey designs. For more information, see Concept:
Coverage Probability in Chapter 6 of the Users Guide.
Projection for grid calculations. This is set to the default projection (see
Project Properties), or [None] if the coverage layer doesn’t have a coordinate

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 255

 system. For more about coordinate systems and projections, see Coordinate
Systems and Projections in Chapter 5 of the Users Guide.
Distance between grid points. Set this to the distance you want between
grid points in the coverage layer. For more about selecting an appropriate
distance, see Concept: Coverage Probability in the Design Properties section of
the Program Reference.
Units of distance. If the calculations are to be done projected, or there is a
non-earth geographic coordinate system, then you can choose from linear units
(e.g., meters, miles, etc). Otherwise, they are angular (e.g., degrees of latitude or
longitude).

Insert or Append Field Dialog

This dialog allows you to insert or append a field into the Distance database. It
is accessed by clicking on the Insert New Field Before Current or Append
New Field After Current buttons or menu items in the Data Explorer.
For more information about creating fields, see Data Explorer – Editing, Adding
and Deleting Fields in the Program Reference.
Options
Field name: Enter the name of the new field here
Field type: Choose an appropriate field type for the field – note that this cannot
be changed once the field has been created.
Units: Choose units for the new field, if appropriate. The units can be changed
later in the Data Explorer (see Data Explorer – Editing, Adding and Deleting
Fields in the Program Reference).

Data Layer Properties Dialog

This dialog gives you information about a particular data layer. It is accessed by
selecting a data layer in the Data Explorer and clicking on the button, or
choosing the menu item Data | Data Layer Properties…
There are two tabs:
• General – gives information about the Layer name, Layer type
and Parent layer. The Description box allows you to enter
comments about the data layer. For coverage layers, this box
contains information about the options used to generate the
coverage.
• Geographic data – gives information about the geographic
information associated with the layer, if any. For more information
about how GIS data are stored, see Geographic (GIS) Data in
Chapter 7 of the Users Guide.
• Shapefile is the name of the ESRI shapefile containing the
geographic data.
• Shape type gives the type of shape (point, line or polygon)
• Folder gives the Windows folder containing the shapefile –
by default this is the project’s Data Folder – but shapefiles can
be contained in any folder (see Importing Existing GIS Data in
Chapter 5 of the Users Guide)
• Coordinate system gives details of the coordinate system
of the shapefile. Click on Change coordinate system …
to go to the New Coordinate System dialog, where you can
specify a different coordinate system for the shapefile data.

256 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Shape Properties Dialog
The shape properties dialog allows you to view and edit the vertices (corners) of
a shape attached to a record in a data layer, as well as copy the vertices to the
Windows clipboard and paste them from the clipboard. It is accessed from the
Data Explorer, by double-clicking on a record in the “Shape” field.
There are three types of shape:
• Point – has only one vertex
• Line – a series of points, joined up. A multi-part line is a broken
line, e.g.: ─── ──. The break is introduced using a Separator in
the list of vertices.
• Polygon – a solid shape, made up of a set of vertices. Multi-part
polygons can be created using separators in the list of vertices.
The dialog contains a table listing the x and y coordinates of the vertices making
up the selected shape. Coordinates are in units defined by the coordinate system
of the data layer (for more about coordinate systems, see Coordinate Systems
and Projections in Chapter 5 of the Users Guide).
You can edit the coordinates by typing into the table entries. You can add
vertices by clicking on the Insert vertex and Append vertex buttons. You
can delete a vertex by highlighting it and clicking on the Delete button. You
can delete all the vertices by clicking the Delete All Vertices button.
To create multi-part lines or polygons, highlight the last vertex in the line or
polygon and click the Append Separator button. Then click Append point
to create the first point of the new line or polygon.
You can copy the vertices to the Windows clipboard by clicking Copy to
Clipboard, and then paste this into another file such as a text file or
spreadsheet. Each vertex is copied to the clipboard as two numbers separated by
a tab character. Separators between multi-part lines or polygons are copied a
line containing just a tab character.
You can also paste the vertices of a shape from the Windows clipboard into the
table by clicking Paste from Clipboard. This provides a useful mechanism
for importing shape information from other packages such as text files or
spreadsheets. For more details, see Importing GIS Data via the Windows
clipboard in Chapter 5 of the Users Guide.
Once you have finished editing the point, click OK to save the edits and return to
the Data Explorer. Alternatively, click Cancel to cancel any changes and
return.

Take care when editing shapes, as once you have pressed OK

there is no undo button!

New Coordinate System Dialog

This dialog lets you change the coordinate system of a data layer. It is accessed
from the Geographic tab of the Data Layer Properties dialog, by clicking
on Change coordinate system….
For more information about coordinate systems in Distance, see Coordinate
Systems and Projections in Chapter 5 of the Users Guide)

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 257

 Column Manager Dialog
The Column Manager allows you to add, delete and rearrange the summary
columns of results that appear in the Design, Survey and Analysis Browser.
Each set can have different results columns. For example, in analysis, you could
have a set for exploratory data analyses, with columns such as number of
parameter, Delta AIC and Chi-sq p, and another set for final results, with
columns such as N and CV of N. (For more information about some of the
columns in the Analysis Browser, see CDS Analysis Browser Results in Chapter
8 of the Users Guide).
The Column Manager consists of two tables - the left table lists the columns that
are already included and the right table lists those that are available to be
included. Each column shows the column name, as it will appear in the browser,
and an explanation of the meaning of the column.
To include a column, click on the column name in the available table and press
the < button. Double clicking on the column name has the same effect.
To exclude a column, click on the column name in the selected table and press
the > buttons. Double clicking on the column name has the same effect.

You can select more than one analysis at once in either table, by
holding the Ctrl or Shift keys while you click, or by pressing Ctrl A or Ctrl / to
select all the analyses.

To rearrange the ordering of the columns in the selected table, use the ↑ and ↓
buttons.
To reset the columns to their state when the Column Manager was opened, press
the Reset button.
To reset to columns to their default arrangement, press the Default button. You
can edit the default arrangement in the Preferences dialog (choose File |
Preferences… on the main menu).
To leave the Column Manager without saving the changes, press the Cancel
button.
To save the changes and exit the Column manager, press OK.

Arrange Sets Dialog

This dialog lets you change the order that Design, Survey and Analysis sets
appear in the drop-down list in the Design, Survey and Analysis browsers. You
access the dialog by clicking on the Arrange Sets button on the browser
toolbar.
For more about sets, look under the appropriate browser (Design, Survey or
Analysis) in the Project Browser section of the Program Reference.

Map Properties Dialog

The Map Properties dialog displays information about a map. It is accessed by
clicking the Map Properties button on the Map toolbar. Information displayed
includes the map’s background colour, rotation angle and coordinate system.

Add Map Layer Dialog

The Add Map Layer dialog allows you to choose which layer to add to the map
from the drop-down list. It is accessed by clicking on the Add Layer button in
a Map.

258 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

 Run Design Dialog
The Run Design dialog opens when you run a design. Here, you choose whether
to estimate probability of coverage for your design, or create a new survey based
on the design.

Confirm Change Dialog

This dialog is displayed:
• when you change the properties of a design that is being used by
one or more survey objects that have been generated from the
design
• when you change the properties of a survey, data filter or model
definition that is being used by an analysis that has been run and
has results.
In it displays a list of surveys or analyses that will be affected by the change, and
asks you to confirm that the log and results for these should be deleted and their
status reset to “not run”.
For more information about why the status needs to be reset, see Analysis
Components in Chapter 7 of the Users Guide.

R Image Properties Dialog

This dialog is displayed by clicking the Image Properties… button in the
Analysis tab of the Preferences Dialog. It allows you to change the properties
of the images produced by the MRDS analysis engine. If you are familiar with
the package R, you will recognize many of these options as relating to the par
function in R.
Image file format
Images are saved to files in a folder “R” within the project data folder - for more
on this see Images Produced by R in the Users Guide. Use these options to alter
the format of this file. Some possible reasons for changing the options are:
• you want to use the image files for inclusion in another document
and would like them to be a specific format or size
• you want to make your project file smaller, by specifying smaller
images, or a format where the images are compressed (jpg)
• your operating system is having trouble displaying images in the
Distance interface in the default format (wmf), so you want to
switch to another
The options are:
• Format of image file – three formats are currently available:
• wmf – Windows metafile is a vector-based format, and gives
the best display quality in the Distance interface, where the
graphic gets resized according to the size of the Analysis
Details window. It is therefore the default. Wmf files are,
however, relatively large (compared with jpeg). There may
also be problems displaying wmf files on older operating
systems (e.g., Windows 98, Windows NT), in which case one
of the other formats will have to be used.
• jpeg – A compressed pixel-based format. Produces relatively
small images, but they tend to look poor when rescaled.

User's Guide Distance 6.0 Beta 5 Appendix - Program Reference • 259

 • bmp – A non-compressed pixel-based format. Produces large
images that tend to look poor when rescaled, but is a format
that can be viewed in any operating system.
• Character point size – default point size of plotted text,
interpreted at 72 dpi, so one point is approximately one pixel
• wmf
• width – the width of the plot in inches
• height – the height of the plot in inches
• jpeg
• width – the width of the plot in pixels for jpeg and bmp
formats
• height – the height of the plot in pixels for jpeg and bmp
formats
• quality – the the “quality” of the jpeg image, as a percentage.
Smaller values will give more compression but also more
degradation of the image.
Graphics parameters
These options allow you to change the look of the plot. They correspond with
options available in the par statement in R. More could be added in future – just
ask!
• Line width (lwd) – The line width, a positive number, defaulting
to 2. We use a default of 2 as it looks better in the wmf plots.
• Line type (lty) – The type of line to draw. An integer: 0=blank,
1=solid (the default), 2=dashed, 3=dotted, 4=dotdash, 5=longdash,
6=twodash.
• Point character (pch) – An integer specifying the symbol to use
– try some numbers between 1 and 20 to see what it does. The
default is 1, a filled circle.
• Symbol scaling (cex) – A numerical value giving the amount by
which plotting text and symbols should be scaled relative to the
default. Default is 1.

Data Selection Zoom Dialog

This dialog is displayed by pressing SHIFT-F1 (i.e., the shift and F1 keys
simultaneously) while editing a data selection criterion in the Data Selection Tab
of the Data Filter Properties Dialog. Here, you have more space to view and edit
long, complex selection criteria.
For more about the format of data selection queries, see the page on the Data
Selection Tab in the Program Reference.

260 • Appendix - Program Reference User's Guide Distance 6.0 Beta 5

Appendix - Inside Distance

Introduction to Inside Distance Appendix

This chapter contains information about how Distance works from the inside. It
is intended for advanced users who want to push Distance to its limits.
The information here is preliminary, and will be expanded in future releases.

Manually editing the distance database files can result in the

project no longer working in Distance. The following information is provided in
the hope that it will be useful, but we cannot be held responsible for any
problems that occur as a result of using it.

Distance components

This section will contain a description of the various components that go to make
up Distance. You can get a list of the component files that make up Distance by
selecting the Help | About Distance… from the main menu, and clicking on the
Program Files tab.

The Distance 6 Database API

All of the functionality used to manipulate Distance 6 projects is packaged into

an ActiveX DLL, D6DbEng.dll. This DLL can be used to create and delete
projects, change project and default properties, create, delete and rename data
layers, etc. An overview class diagram for the API (Application Programming
Interface) is given below, and detailed documentation is available from the
program authors on request.

The database API is primarily intended for internal use by those

working on the Distance project. There is no guarantee that the API will remain
the same in subsequent releases, although we will endeavor not to break the
current interface.

User's Guide Distance 6.0 Beta 5 Appendix - Inside Distance • 261

 Overview of D4DbEng.dll public classes. Where there are items in [], these are the base
class names.

Data File Reference

How Distance Stores Data

Distance can use data from three sources:
• internal data, which is stored in tables in the Data File,
DistData.mdb
• geographic data, stored in ESRI shapefiles
• external data, stored in external database tables, spreadsheet or text
files
Information about the location of the data is stored in three linked tables in
DistData.mdb. This file is created by the Microsoft Jet 3.51 engine, which is the
same engine used in Microsoft Access 97. If you want to get at the internal
workings of this file, your best bet is to use Access 97. You can use Access
2000 or later, but things are a little more complex (see Accessing DistData.mdb
using newer versions of Access.)
The three tables in DistData.mdb are:
• DataLayers – contains one record for each data layer in the
project
• DataTables – contains one record for each database table that
make up the data layers. For example, by default a geographic data
layer is made up of two tables: an internal table with the same
name as the data layer, and the .dbf table from the shapefile.
• DataFields – contains one record for each data field in the data
layers of the project.
More details on each of these tables are given in the following topics.

262 • Appendix - Inside Distance User's Guide Distance 6.0 Beta 5

 DataLayers table in DistData.mdb
This table contains one record for each data layer in the project. It is used to tell
Distance how many layers are in the project and what their relationship is to one
another - this is used, for example, to construct the tree view in the Data Layers
Viewer.
The DataLayers table has the following fields:
Field Name Field Primar Description
Type y Key?
LayerName Text Y Name of data layer
LayerType Long N Enumeration (see
Enumerations in
DistData.mdb)
ParentLayerName Text N Name of parent data layer
Description Memo N Place for user to enter
comments about the layer (can
edit in Description box of the
Data Layer Properties Dialog).

Records in the DataLayers table are subject to the following restrictions:

• The LayerName must be unique
• There are some restrictions on characters that can be used for layer
names (e.g., no special characters such as “[”, “]”, etc.
• There can only be one record with LayerType = 1 (Global)
• All layers, except the global layer, must have a ParentLayerName.
• In general, the layer type of the parent layer should be lower than
that of its children (e.g., parent LayerType = 1 (Global); child
LayerType = 2 (Stratum)).

DataTables table in DistData.mdb

This table contains one record for each database table that makes up each data
layer in the project. For example, by default a geographic data layer is made up
of two tables: an internal table with the same name as the data layer, and the .dbf
table from the shapefile. Fields from tables making up the data layer are joined
together to make up the records for the data layer that you see in the Data
Explorer.
The DataTables table has the following fields:
Field Name Field Primar Description
Type y Key?
TableName Text Y Unique name for table. Can
be either an internal table, or a
linked table. This name is
assigned by Distance when the
table is added to the project –
if you’re adding tables
manually you can make a
unique name up.
LayerName Text N Name of data layer
SourceDatabaseT Text N Either (1) “Int” for internal
ype tables (i.e., tables that are in
DistData.mdb); (2) “Geog” for
the .dbf part of a shapefiles; or
(3) a MS Jet IISAM database
type specifier (e.g., “Text”)

User's Guide Distance 6.0 Beta 5 Appendix - Inside Distance • 263

 (see Note 1).
SourceDatabaseN Text N Either (1) blank if
ame SourceDatabaseType is “Int”;
(2) blank if the table is in the
Data Folder; or (3) the
absolute path to the database
(see Note 2).
SourceTableName Text N The actual table name (see
Note 3).
ShapeType Long N Enumeration (see
Enumerations in
DistData.mdb)
PrimaryTable Boolea N If true, the table should
n contain ID and ParentID
fields. If false, the table
should contain LinkID field.
Only one primary table per
data layer.
Note 1: MS Jet IISAM specification is for this to be blank for native Jet databases, but
this property should contain “Jet” if the table is a native Jet table.

Note 2: for single table databases, this is the folder name, for multiple table databases, its
the file name

Note 3: For external files: for single table databases, this is the file name without
extension, for multiple its the table name in the file.

Records in the DataTables table are subject to the following rules:

• The TableName must be unique, and there are some restrictions on
the characters that can be used.
• There must be at least one table with SourceDatabaseType=“Int”
(i.e., internal) in each data layer
• One table in each layer must be the primary table
(PrimaryTable=T) and this table must be of
SourceDatabaseType=“Int”.
• There can be at most one geographic table
(SourceDatabaseType=“Geog”). The geographic table must have a
nonzero ShapeType.
Linking to external tables using MS Jet IISAM is examined further in Linking to
External Data from DistData.mdb.

DataFields table in DistData.mdb

This table contains one record for each field in each data layer. It is used to
determine which fields to display in the Data Explorer, and also to denote special
fields such as ID fields.
The DataFields table has the following fields:
Field Name Field Primar Description
Type y Key?
LayerName Text Y Name of data layer
FieldName Text Y Name of data field (must be
the same as the name of the
field in the source database
table)
TableName Text N TableName in DataTable

264 • Appendix - Inside Distance User's Guide Distance 6.0 Beta 5

 FieldType Long N Enumeration (see
Enumerations in
DistData.mdb)
Units Text N Units of measurement1, or
blank.
OrdinalPosition Long N Used in Data Explorer to tell
what order to display the
fields
ColumnWidth Long N Used in Data Explorer to
remember user-specified
column widths (usually zero)
Tag Text N Currently unused
Formula Memo N Currently unused
1 For a list of allowable strings, open DistIni.mdb in the Distance program directory,
look in the ProjectSettings table under section UnitsAbbr. All the values under Key are
allowable units.

Records in the DataFields table should abide by these rules:

• Records can have the same field name (e.g., “ID”), but the same
field name cannot be used twice within a layer (i.e., you can’t have
the same field in two Data Tables that make up the same layer)
• Not all fields that physically exist in a Data Table need to have
records here – if a field is omitted then it will not be available to
Distance (e.g., it will not appear in the data sheet)
• In general, the FieldType must match the type of the field in the
database (e.g., you’ll get an error if a field that is a text field in the
database is given FieldType = 1 (Integer)).
• Primary tables must have one field of FieldType = 10 (ID). All
other tables must have one field of FieldType = 13 (LinkID).
These fields are used to join together records from each table. In
general, the ID and LinkID values in each table should go from 1 to
the number of records in the table.
• One of the tables in each layer must contain a field of FieldType =
11 (ParentID). This field is not needed in each table in a layer –
only in one of them. The only exception is tables in the global
layer: none of these need a ParentID field (since there is no parent
layer). The ParentID field tells Distance which record in the parent
data layer to put the child record in.
• If there is a geographic table in the layer, there needs to be a record
with FieldName = “Shape”, and FieldType = 14 (Shape).
• The OrdinalPosition should be unique for fields within a layer –
these dictate the order that the fields appear in the Data Explorer.
ParentID and LinkID fields are not shown in the DataExplorer, so
their ordinal position doesn’t matter.

Unfortunately, at present, you cannot include attribute fields from

the geographic table – the Shape field is the only one you can include. This is
due to a bug in the GIS component from ESRI and we hope to resolve this at
some point in the future. Until then, this means there is no way to link to
attribute data in the shapefiiles.

User's Guide Distance 6.0 Beta 5 Appendix - Inside Distance • 265

 Enumerations in DistData.mdb
Field Name Enumeration Meaning
LayerType 1 Global
10 Stratum
11 SubStratum1
12 SubStratum2
13 SubStratum3
14 SubStratum4
15 SubStratum5
20 Sample
21 SubSample1
22 SubSample2
23 SubSample3
24 SubSample4
25 SubSample5
30 Observation
40 Coverage
99 Other
FieldType 1 Integer
2 Decimal
3 Text
10 ID
11 ParentID
12 Label
13 LinkID
14 Shape
99 Other (not used)
ShapeType 0 No shape
21 Point
22 Line
23 Polygon
24 Multipoint (not used)
25 Rectangle (not used)
26 Ellipse (not used)

Accessing DistData.mdb using newer versions of

Access

The first time you open a DistData.mdb file in versions of Access after Access
97 (e.g., Access 2000, 2002, etc.), it asks you if you want to convert the file to
the new format, or open as is.
If you choose to open as is, you will get a message saying that you cannot
change the database structure. This means that you cannot add new fields or
new tables to the database, but you can edit records or add new records. In many
cases (for example, see Importing Existing GIS Data in Chapter 5 of the Users
Guide) this is fine. In general we recommend this as the easier option.

266 • Appendix - Inside Distance User's Guide Distance 6.0 Beta 5

 If you choose to convert the database, you will be prompted for a new filename
to save the converted database to (e.g., DistData.new.mdb). Once the conversion
has taken place, the new file is opened. You are then will be free to make any
changes you want to the database structure, such as adding new tables, fields,
etc. However, you then need to convert the database back to the old format
before it can be opened in Distance.
To do this:
• Rename the original DistData.mdb file (e.g., to DistData.bak.mdb),
and keep it as a backup in case something goes wrong
• In Access 2000 or 2002, on the Tools menu, point to Database
Utilities, click Convert Database, and then click To Prior
Access Database Version.
• In the Convert Database Into dialog box, type DistData.mdb in
the File name box, and then click Save
• Close the database in Access 2000/2002
• Open the project file in Distance, and check that the changes you
made have registered correctly
• Assuming everything is fine, you can delete the Access 2000/2002
database (e.g., DistData.new.mdb) and the old backup file
(DistData.bak.mdb)
As you can see, this is quite a hassle! There are other issues to consider in
making the conversion – for more information, look in the Access 2000 online
help under “Convert an Access 2000 database to Access 97” and “About
converting an Access 2000 database to Access 97” (Access 2002 and later have
similar topics).

Visual Basic 5 professional came with source code for a

demonstration database access program called VisData, which allows access to
Microsoft Jet 3.51 files. It isn’t nearly as easy to use as Access 97, but is better
than nothing. If it would be useful, we could probably post a modified version
of the program on the Distance web site. We’d have to check the licensing
situation first though. If you think this would be useful, please let us know.

Linking to External Data from Distdata.mdb

This is an advanced technique, suitable only for those

comfortable poking around inside Access databases. Make sure you have read
and understood the previous topics on How Distance Stores Data before
proceeding!

You can, in theory, use Distance to link to data in tabular text files, databases
and spreadsheets – although you cannot do this directly from the Distance GUI.
Instead, you do it by directly editing the Data File, DistData.mdb, using Access.
Briefly: you add an entry for the table you want to link in the DataTables table in
DistData.mdb, and then add entries for the fields you want to link to in
DataFields. An example is provided in the LinkingExample sample project.

Supported External Data Sources

You can link to any type of data for which there is a Microsoft Jet 3.51 IISAM
(Installable Indexed Sequential Access Method) driver. By default, Distance
supplies you with:

User's Guide Distance 6.0 Beta 5 Appendix - Inside Distance • 267

 • a native Jet driver (for Microsoft Access 97 and earlier .mdb
databases)
• a text driver (for text files in tabular formats)
• a Microsoft Excel driver (for versions 3.0-8.0).
For more on the last two, see the topic Working with the Microsoft Jet IISAM
Text File Driver and Working with the Microsoft Jet IISAM Excel File Driver.
There are also drivers available for the following formats:
• Microsoft FoxPro databases, versions 2.0, 2.5, 2.6, 3.0, and DBC
(database containers).
• dBASE databases, versions III®, dBASE IV®, and dBASE 5.0.
• Paradox databases, versions 3.x, 4.x, and 5.x.
• Lotus spreadsheets, versions WKS, WK1, WK3, and WK4.
• Tabular data in Hypertext Markup Language (HTML) files.
These are discussed briefly in Working with Other Microsoft IISAM drivers.

The technology used by the Distance database engine (Jet 3.51) has
been replaced by Microsoft by newer technology, so it is unlikely they will issue
IISAM drivers to link to newer versions of the above software. Given the
overhead that would be required, it is also unlikely that we will be updating the
Distance database engine to use newer technology any time soon. Many newer
programs can, however, work with files in the older formats – for example,
newer versions of Excel can easily save files as Excel 97-2002 (Excel 8.0) and
work with them in that format.

Outline of Linking to External Data

Probably the easiest way to see how to link to external data is to examine the
LinkedExample sample project. This project has two data layers:
• a global layer which links to a table in an Access 97 database
(LinkedData.mdb)
• a stratum layer which links to another table in LinkedData.mdb and
also to a tab-delimited text file, and to a worksheet in an Excel 8.0
file (LinkedData.xls).
Both layers also have a geographic data layer. Examine the DataTables and
DataFields tables in DistData.mdb to see how this was done.
Imagine that we wanted to add a sample-level data layer called “Line transect”
and link to a table “Transect” containing transect information in
LinkedData.mdb. The following outlines how we might do this, from within
DistData.mdb. For simplicity, we’ll assume that the new layer is not going to be
geographic.
20. Create a new record in DataLayers for the new layer. LayerName
is “Line transect”, LayerType is 20 (Sample) and
ParentLayerName is “Region”.
21. Create a primary data table for the new layer. Create a new table
called “Line transect” in DistData.mdb. Give it fields “ID” and
“ParentID” (both of type Long). (Note – I’m assuming the
ParentID field isn’t in the external “Transect” table.) Let’s imagine
that there are 10 records in our external Transect table – so we need

268 • Appendix - Inside Distance User's Guide Distance 6.0 Beta 5

 10 records in this table, with ID from 1-10 and appropriate
ParentID values to put the transects in the correct stratum.
22. Create a record for the primary data table in the DataTables table.
The TableName, LayerName and SourceTableName are all “Line
transect”, the SourceDatabaseType is “Int”, and PrimaryTable is
True.
23. Create records for the fields in this table in DataFields. You need a
record for the ID field, which will have FieldType=10 (ID) and for
the ParentID field, which will have FieldType = 11 (ParentID)
Note that all of these steps could have been more easily performed by
opening the project in Distance, creating a new data layer called “Line
transect” of type Sample, and then creating 10 new records in the layer.
24. Now we want to add a new record to the DataTables table for the
external table. The TableName can be anything, but for
consistency with what’s already in the project let’s use
“lnkTransect”, the LayerName is “Line transect”,
SourceDatabaseType is “Jet”, SourceDatabaseName is
“LinkedData” and SourceTableName is “Transect”.
25. Now we must add records to the DataFields table for each field we
wish to link to in the Transect table. We must include as a
minimum a field of type 13 (LinkID), which tells Distance which
records in Transect link to which records in the internal “Line
transect” table. We can include any other fields we like, so long as
they are in the “Transect” table.
26. Hopefully now the new table will be linked – we can open the
project in Distance to check.

If you run into problems linking files of a specific format, and have
tried everything you can think of, try looking at the settings in
HKEY_LOCAL_MACHINE\SOFTWARE\
Microsoft\Jet\3.5\Engines or \ISAM_Formats to see if they might be the cause of
the problem.

Working With the Microsoft Jet IISAM Text File Driver

The following information is adapted from the Microsoft Jet 3.5 documentation,
and will hopefully be of use in setting up text files for linking to the Distanace
database. An example of linking to a text file is given by the LinkingExample
sample project.
You can use the Microsoft Jet Text IISAM to link and open character-delimited
and fixed-length text files. Commas, tabs, or user-defined delimiters are valid in
the source file.
When specifying connection information for text files, use the following
specifications in the DataTables table:
SourceDatabaseType: Text
SourceDatabaseName: The full path to the directory containing the text file
you intend to access. If you do not specify a path, Distance will look in the
project data folder.
SourceTableName: The name of the text file, including the extension. If you
don't specify an extension, the default .txt extension is used.
Microsoft Jet recognizes null values in character-delimited files by the presence
of two consecutive delimiting characters. Microsoft Jet recognizes null values in
fixed-length files by the absence of data (spaces) in the data column.

User's Guide Distance 6.0 Beta 5 Appendix - Inside Distance • 269

 Microsoft Jet determines the format of the text file by reading the file directly or
by using a schema information file. The schema information file, which is
always named Schema.ini and always kept in the same directory as the text data
source, provides the IISAM with information about the general format of the
file, the column name and data type information, and a number of other data
characteristics. A Schema.ini file is always required for accessing fixed-length
data; you should use a Schema.ini file when your text table contains DateTime,
Currency, or Decimal data or any time you want more control over the handling
of the data in the table.

Microsoft Jet doesn't support multiuser access to text files. When

you open a text file through Microsoft Jet, you have exclusive access to the file.

Depending on the registry settings on your computer, you may only

have read-only access to linked text files.

The following table lists the few limitations to the size of text tables and objects.
Item Maximum size per text file
Field 255
Field name 64 characters
Field width 32,766 characters
Record size 65,000 bytes

Understanding Schema.ini files

Schema.ini files provide schema information about the records in one or more
text files in the same directory as the schema file. Each Schema.ini entry
specifies one of five characteristics of the table:
• The text file name.
• The file format.
• The field names, widths, and types.
• The character set.
• Special data type conversions.
The following sections discuss these characteristics.
Specifying the file name
The first entry in Schema.ini is always the name of the text source file enclosed
in square brackets. The following example illustrates the entry for the file
Sample.txt:
[Sample.txt]

You can specify settings for more than one file in the same Schema.ini file.
Specifying the file format
The Format option in Schema.ini specifies the format of the text file. The Text
IISAM can read the format automatically from most character-delimited files.
You can use any single character as a delimiter in the file except the double
quotation mark ("). The Format setting in Schema.ini overrides the setting in the
Windows Registry on a file-by-file basis. The following table lists the valid
values for the Format option.
Format specifier Table format
TabDelimited Fields in the file are delimited by

270 • Appendix - Inside Distance User's Guide Distance 6.0 Beta 5

 tabs.
CSVDelimited Fields in the file are delimited by
commas (comma-separated
values).
Delimited(*) Fields in the file are delimited by
asterisks. You can substitute any
character for the asterisk except
the double quotation mark.
FixedLength Fields in the file are of a fixed-
length.

For example, to specify a comma-delimited format, you would add the following
line to Schema.ini:
Format=CSVDelimited

Specifying the fields

You can specify field names in a character-delimited text file in two ways:
• Include the field names in the first row of the table and set
ColNameHeader to True.
• Specify each column by number and designate the column name
and data type.
You must specify each column by number and designate the column name, data
type, and width for fixed-length files.

The ColNameHeader setting in Schema.ini overrides the

FirstRowHasNames setting in the Windows Registry on a file-by-file basis.

You can also instruct Microsoft Jet to determine the data types of the fields. Use
the MaxScanRows option to indicate how many rows Microsoft Jet should scan
when determining the column types. If you set MaxScanRows to 0, Microsoft Jet
scans the entire file. The MaxScanRows setting in Schema.ini overrides the
setting in the Windows Registry on a file-by-file basis.
The following entry indicates that Microsoft Jet should use the data in the first
row of the table to determine field names and should examine the entire file to
determine the data types used:
ColNameHeader=True
MaxScanRows=0

The next entry designates fields in a table by using the column number (Coln)
option, which is optional for character-delimited files and required for fixed-
length files. The example shows the Schema.ini entries for two fields, a 10-
character CustomerNumber text field and a 30-character CustomerName text
field:
Col1=CustomerNumber Text Width 10
Col2=CustomerName Text Width 30

The syntax of Coln is:

Coln=ColumnName type [Width #]

The following table describes each part of the Coln entry.

Parameter Description
ColumnName The text name of the column. If
the column name contains
embedded spaces, you must
enclose it in double quotation
marks.

User's Guide Distance 6.0 Beta 5 Appendix - Inside Distance • 271

 Type Data types are:Microsoft Jet data
types: Bit, Byte, Short, Long,
Currency,Single, Double,
DateTime, Text, MemoODBC
data types, Char (same as Text),
Float (same as Double), Integer
(same as Short), LongChar (same
as Memo), Date date format
Width The literal string value Width.
Indicates that the following
number designates the width of the
column (optional for character-
delimited files, required for fixed-
length files).
# The integer value that designates
the width of the column (required
if Width is specified).

Selecting a Character Set

You can select from two character sets: ANSI and OEM. The following example
shows the Schema.ini entry for an OEM character set. The CharacterSet setting
in Schema.ini overrides the setting in the Windows Registry on a file-by-file
basis. The following example shows the Schema.ini entry that sets the character
set to ANSI:
CharacterSet=ANSI

Specifying Data Type Formats and Conversions

The Schema.ini file contains a number of options that you can use to specify
how data is converted or displayed when read by Microsoft Jet. The following
table lists each of these options.
Option Description
DateTimeFormat Can be set to a format string
indicating dates and times. You
should specify this entry if all
date/time fields in the
import/export are handled with the
same format. All of the Microsoft
Jet formats except A.M. and P.M.
are supported. In the absence of a
format string, the Windows
Control Panel short date picture
and time options are used.
DecimalSymbol Can be set to any single character
that is used to separate the integer
from the fractional part of a
number.
NumberDigits Indicates the number of decimal
digits in the fractional portion of a
number.
NumberLeadingZeros Specifies whether a decimal value
less than 1 and greater than –1
should contain leading zeros; this
value can either be False (no
leading zeros) or True.
CurrencySymbol Indicates the currency symbol to
be used for currency values in the
text file. Examples include the

272 • Appendix - Inside Distance User's Guide Distance 6.0 Beta 5

 dollar sign ($) and Dm.
CurrencyPosFormat Can be set to any of the following
values:· Currency symbol prefix
with no separation ($1); Currency
symbol suffix with no separation
(1$); Currency symbol prefix with
one character separation ($ 1);
Currency symbol suffix with one
character separation (1 $).
CurrencyDigits Specifies the number of digits used
for the fractional part of a currency
amount.
CurrencyNegFormat Can be one of the following
values:· ($1);·–$1; $–1; $1–; (1$);
–1$; 1–$; 1$–; –1 $; –$ 1; 1 $–; $
1–; $ –1; 1– $; ($ 1); (1 $). This
example shows the dollar sign, but
you should replace it with the
appropriate CurrencySymbol value
in the actual program.
CurrencyThousandSymbol Indicates the single-character
symbol to be used for separating
currency values in the text file by
thousands.
CurrencyDecimalSymbol Can be set to any single character
that is used to separate the whole
from the fractional part of a
currency amount.

Note: If you omit an entry, the default value in the Windows Control Panel is
used.

Examples of Schema.ini files

Schema.ini contains the specifics of a text data source: how the text file is
formatted, how it's read at import time, and the default export format for files.
The following example shows the layout for a fixed-width file, Filename.txt:
[Filename.txt]
ColNameHeader=False
Format=FixedLength
MaxScanRows=25
CharacterSet=OEM
Col1=columnname Char Width 24
Col2=columnname2 Date Width 9
Col3=columnname7 Float Width 10
Col4=columnname8 Integer Width 10
Col5=columnname9 LongChar Width 10

Similarly, the format for a delimited file is specified as:

[Delimit.txt]
ColNameHeader=True
Format=Delimited(!)
MaxScanRows=0
CharacterSet=OEM
Col1=username Text
Col2=dateofbirth DateTime

Note that both of these format sections can be in the same .ini file.
Another example of a Schema.ini file is in the Data Folder of the
LinkingExample project.

User's Guide Distance 6.0 Beta 5 Appendix - Inside Distance • 273

 Working With the Microsoft Jet IISAM Excel File Driver
The following information is adapted from the Microsoft Jet 3.5 documentation,
and will hopefully be of use in setting up Excel spreadsheet files for linking to
the Distance database. An example of linking to an Excel 8.0 file is given by the
LinkingExample sample project.
The Microsoft Jet IISAMs support the following single-sheet worksheet and
multiple-sheet workbook versions of Microsoft Excel: Excel 3.0 and Excel 4.0
for single-sheet worksheets, and Excel 5.0 (for Microsoft Excel 5.0 and 7.0) and
Excel 8.0 for multiple-sheet workbooks. There are a few operations that you
can't perform on Microsoft Excel worksheets or workbooks through the
Microsoft Excel IISAM:
• You can't delete rows from Microsoft Excel worksheets or
workbooks.
• You can clear data from individual cells in a worksheet, but you
can't modify or clear cells that contain formulas.
• You can't create indexes on Microsoft Excel worksheets or
workbooks.
• You can't read encrypted data through the Microsoft Excel IISAM.
You can't use the PWD parameter in the connection string to open
an encrypted worksheet or workbook, even if you supply the
correct password. You must decrypt all Microsoft Excel
worksheets or workbooks through the Microsoft Excel user
interface if you plan to link or open them in your Microsoft Jet
database.
When specifying connection information for Excel files, use the following
specifications in the DataTables table:
SourceDatabaseType: One of: Excel 3.0, Excel 4.0, Excel 5.0, Excel 8.0
SourceDatabaseName: see Use (1) below
SourceTableName: see Use (2) below

To access In this Use this syntax

this object version of
Microsoft
Excel
Entire sheet 3.0 and 4.0 Use (1) to specify the fully qualified
in a network or directory path to the
worksheet worksheet file (no path needed if in data
file folder); use (2) to specify the sheet as
filename#xls, where filename is the name
of the worksheet.
Entire 5.0, 7.0, and Use (1) to specify the fully qualified
worksheet in 8.0 network or directory path to the workbook
a workbook file (if not in data folder), including the
file workbook file name; use (2) to specify the
sheet as sheetname$, where sheetname is
the name of the worksheet. Important
You must follow the worksheet name
with a dollar sign ($).
Named 3.0, 4.0, 5.0, Use (1) to specify the fully qualified
range of 7.0, and 8.0 network or directory path to the
cells in a worksheet or workbook file (if not in data
worksheet folder), including the worksheet or
or workbook workbook file name; use (2) to specify the
file named range as NamedRange, where

274 • Appendix - Inside Distance User's Guide Distance 6.0 Beta 5

 NamedRange is the name you assigned to
the range in Microsoft Excel.Important
You must name the range in Microsoft
Excel before attempting to open or link it.
Unnamed 3.0 and 4.0 Use (1) to specify the fully qualified
range of network or directory path to the
cells in a worksheet file (if not in data folder),
worksheet including the worksheet file name; use (2)
file to specify the range as A1:Z256. Replace
A1:Z256 with the range of cells you want
to access.
Unnamed 5.0 and 7.0 Use (1) to specify the fully qualified
range of network or directory path to the workbook
cells in a file (if not in data folder), including the
single workbook file name; use (2) to specify the
worksheet in sheet you want to link or open as
a workbook sheetname$ and the range as A1:Z256.
file For example, to access cells A1 through
Z256 in worksheet SheetName, you
would use the following in (2)
SheetName$A1:Z256.

The HDR parameter

By default, the first row of the worksheet or selected text contains the field
name. To suppress this, add ;HDR=No to the SourceDatabaseType (you don’t
need to specify the default ;HDR=Yes, unless the default has been changed –
this is governed by the value of FirstRowHasNames in the following registry
key: \HKEY_LOCAL_MACHINE\SOFTWARE\
Microsoft\Jet\3.5\Engines\Excel)

Working with Other Microsoft IISAM drivers

Microsoft provides drivers for other data sources, in addition to text and native
Jet 3.51 (see Supported External Data Sources for a list).
These drivers are not supplied with Distance. Probably the easiest way to obtain
them is to go to the Support page of the Distance web site, and find the link to
download Microsoft DAO setup package. As part of the install, it will prompt to
ask which IISAM drivers you wish to install. You can also download them from
Microsoft.
A very brief outline of what entries are required in the DataTables table is given
below. More documentation can be supplied on request to the program authors;
however it is not anticipated that this feature of the program will see much use!

Data Source SourceDatab SourceDatab SourceTable

aseType aseName Name
Microsoft Jet Jet Path to the Name of the
database table
dBASE One of: Path to the Name of the
dBASE III database table
dBASE IV
dBASE 5.0
FoxPro One of: Path to the The name of the
FoxPro 2.0 database (For table. (Use the
FoxPro 2.5 Microsoft .dbf file name
FoxPro 2.6 FoxPro DBC, without the
FoxPro 3.0 the path must extension, or
FoxPro DBC include the the complete
name of the file name with
.dbc file.) the extension,

User's Guide Distance 6.0 Beta 5 Appendix - Inside Distance • 275

 but substitute a
number sign [#]
for the dot [.]
that precedes
the file name
extension; for
Microsoft
FoxPro DBC,
use the table
name in the
DBC.)
Paradox One of: Path to the Name of the
Paradox 3.x database table.
Paradox 4.x
Paradox 5.x
Lotus One of: Path to the file filename#wks
Lotus WKS1 (single sheet
Lotus WK12 file) or
Lotus WK32 sheetname:
Lotus WK41 (multi sheet
file) or
sheetname:A1..
Z256 (range of
cells)
HTML HTML Import URL of page title of table
containing table caption, or
Table1, Table2,
etc if no caption
1
Read-only access
2
Read and insert access (can’t modify existing rows)

Valid Names
Valid Field Names
Field names must meet the following criteria to be valid:
• For internal fields, the name must be 64 letters long or less
• For shapefile fields, the name must be 10 letters or less long with no
spaces
• The only permitted characters are letters (A-Z or a-z), numbers (0-9),
spaces or underscores.
• Field names must be unique within a data layer (i.e., the same name is not
allowed in 2 tables, except for the ID and LinkID fields)
• The name must not appear on the list of reserved field names below (not
case sensitive).

Reserved field names

Reserved by Distance
Shape FeatureID None
Reserved by R
NULL NA TRUE
FALSE GLOBAL.ENV Inf
NaN function while
if repeat for
in else next

276 • Appendix - Inside Distance User's Guide Distance 6.0 Beta 5

 break ...
Reserved by the Jet database engine
ADDALL Alphanumeric ALTER
And ANY AS
ASC AUTOINCREMENT Avg
Between BINARY BIT
BOOLEAN BYBYTE CHAR
CHARACTER COLUMN CONSTRAINT
Count COUNTER CREATE
CURRENCY DATABASE DATE
DATETIME DELETE DESC
DISALLOW DISTINCT DISTINCTROW
DOUBLE DROP Eqv
EXISTS FLOAT FLOAT8
FLOAT4 FOREIGN FROM
GENERAL GROUP GUID
HAVING IEEEDOUBLE
IGNORE Imp In
ININDEX INNER INSERT
INT INTEGER INTEGER4
INTEGER1 INTEGER2 INTO
Is JOIN KEY
LEFT Level Like
LOGICAL LOGICAL1 LONG
LONGBINARY LONGTEXT Max
MEMO Min Mod
MONEY Not NULL
NUMBER NUMERIC OLEOBJECT
ON OPTION Or
ORDER Outer OWNERACCESS
PARAMETERS PERCENT PIVOT
PRIMARY PROCEDURE REAL
REFERENCES RIGHT SELECT
SET SHORT SINGLE
SMALLINT SOME StDevStDevP
STRING Sum TABLE
TableID TEXT TIME
TIMETIMESTAMP TOP TRANSFORM
UNION UNIQUE UPDATE
VALUE VALUES Var
VARBINARY VARCHAR VarP
WHERE WITH Xor
YESNO

User's Guide Distance 6.0 Beta 5 Appendix - Inside Distance • 277

Miscellaneous topics

Random number generation

Different algorithms are used to generate pseudo-random numbers in the
different components of Distance, as follows:
• The Distance interface uses the random number class in D4Util.dll.
This is based on Knuth’s subtractive method – see algorithm ran3
in Press et al. 1992. The generator is seeded from the system
clock. Random numbers are used, for example, in generating
probability of coverage grid points.
• The design engine also uses the random number class from
D4Util.dll for generating designs. The generator is either seeded
from the system clock or from a user-specified value.
• The CDS and MCDS analysis engines use the Compaq Visual
Fortran function random_number, for bootstrap resampling. This
uses two congruential generators (see L’Ecuyer 1988 or the Fortran
manual for more details). The seed is either set from the system
clock, or can be specified by the user in the Model Definition
Properties, under Variance.
• The R routines in Distance use R’s default random number
generator – type help(.Random.Seed) in R for more
information.

278 • Appendix - Inside Distance User's Guide Distance 6.0 Beta 5

Appendix - MCDS Engine
Reference

Introduction to MCDS Engine Reference

The CDS and MCDS engines are implemented as a stand-alone FORTRAN
program, MCDS.exe. This program is called behind the scenes by Distance
when you press the Run button on the Analysis Details Inputs tab. Some users
may wish to run the engine from outside the Distance interface – either from the
Windows command line or from another program. For example, you may want
to automate the running of analyses for simulations, or you may want to run a
complicated bootstrap not available in Distance. Here, we provide outline
documentation for running the CDS and MCDS analysis engine as a stand-alone
program.
For more information about the various options available in the engine, see the
Users Guide Chapter 8 - Conventional Distance Sampling Analysis and Chapter
9 - Multiple Covariates Distance Sampling Analysis.

Since the CDS and MCDS analysis engines are both implemented in
MCDS.exe, we refer to both as “the MCDS engine” in what follows.

There have been several messages on the distance-sampling email list

providing tips on how to use the MCDS engine (and previous versions of the
engine) from outside of Distance. Have a look in the online archives.

Some history
In historic versions of Distance (1.0 - 3.0), the program was driven by a simple
command language, which defined the survey design, data, and analysis
methods. Distance could be run in batch mode by passing in the filenames of
input and output files via the DOS command line. It could also be run
interactively, entering the commands at a prompt.
Distance 3.5 and later added a graphic user interface for defining the inputs. The
program that does the actual work of analysis was called an “analysis engine”,
and was called D35Engine.exe in Distance 3.5, D4.exe in Distance 4, and now
MCDS.exe. This program is run from the Distance graphical interface in batch
mode. The exact way that Distance communicates with the MCDS analysis

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 279
 engine is outlined in another Appendix – see Introduction to Inside Distance
Appendix.
The data format and command language used to run MCDS.exe are therefore
very similar to those used to run the old versions of Distance (the major
differences are outlined in a subsection, below). The last complete
documentation for the command language is the Distance 2.2 users manual,
which is available for download from the support page of the Program Distance
web site. Many new features have been added since Distance 2.2 (for example
multiple covariates and flat data file input), but some features are also no longer
supported. These include: interactive mode (batch mode only is now supported)
and hierarchical data input (flat files only). For a full list, see the section
Changes in MCDS Engine Since Distance 2.2.

Running the MCDS engine

The command to run the MCDS engine is
MCDS Parameter1, Parameter2

where
Parameter1 is either a 0 or a 1. 0 is for run mode - i.e. run the analysis. 1 is for
import mode, which is used to implement part of the Project Import feature in
Distance and is not described further here.
Parameter2 is the filename of the input command file – see MCDS Command
Language, below for details of the contents of this file.
The program returns a number to the command line, indicating the status of the
run, as well as up to 6 files of output – see Output From the MCDS Engine.
Example 1:
Assume that we have a command window open in the Distance program
directory (usually C:\Program Files\Distance5), and that we have a file
TestInput.txt in that directory. Then we type:
MCDS 0, TestInput.txt

Example 2:
Assume we have a command window open in some arbitrary directory (e.g.,
C:\). Assume that we have an input command file C:\Temp\Input File.txt
that we want to run. Assume that the MCDS.exe program is in the Distance
program directory C:\Program Files\Distance5. Because both the input file
and the Distance program directory have spaces in them, we need to enclose the
program and file names in quotes:
"C:\Program Files\Distance5\MCDS" 0, "C:\Temp\Input File.txt"

The space between MCDS and parameter 1, and the space and
comma between parameter 1 and parameter 2 are critical. For example, in
Example 1, above,
MCDS 0,TestInput.txt
will not work (it will return the value 4 - file error) because there is no space
between parameter 1 and parameter 2 (see MCDS engine command line output
for more about the numbers returned to the command line).

You can copy the file MCDS.exe to another folder and run it from
there if you want to (e.g., C:\temp). You could also add the Distance program
folder to your windows path (in Windows XP it’s under Control Panel | System |

280 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 Advanced | Environment variables) so you don’t then need to give the full path
when calling it from the command line.

An example of how to run the MCDS analysis engine from a program

written in another language (in this case Visual Basic) is given on the Support
page of the Program Distance Web Site. See also the archives of the Distance-
sampling Email List for some messages discussing how to do this (try searching
for "stand alone").

MCDS Command Language

The MCDS analysis engine is driven by a relatively simple command language.
When you run the MCDS engine from the command line (see Running the
MCDS Engine), you pass in the name of a file containing these commands. The
commands are then interpreted by the engine, which performs the analysis you
have asked for.
The command file is divided into 4 sections:
• The Header section, where the location of the output files is
specified. This is always 6 lines long.
• The Options section, where general program options are set. This
begins with the OPTIONS command and ends with an END command.
• The Data section, where the location and format of the data file is
specified. This begins with the DATA command and ends with an
END command.

• The Estimate section, where estimation options are set. This

begins with the ESTIMATE command and ends with an END
command.
The format of each section is described in the following pages, and an example
command file is given below. You can see many other examples by looking in
the log tab of CDS or MCDS analyses that have been run.
The language interpreter is case insensitive. All commands (apart from the files
in the header) end with a semicolon. Each command is usually given a new line,
but this is not necessary. Commands and options can in theory be shortened so
that they are the minimum length necessary to make them uniquely
distinguishable – but this is not recommended as it leads to incomprehensible
command files! The order of the commands in the Options, Data and Estimate
sections should not matter.

An easy way to generate a template command file for a particular

analysis is to set up that analysis using the Distance graphical interface, and then
run the analysis in Debug mode. In this mode, the Distance interface
generates a command file and data file, and stores them in the Windows
temporary folder, but does not run the analysis. For more about Debug mode,
see the Program Reference page on the Analysis Preferences Tab.

C:\Temp\dst111.tmp
C:\Temp\dst110.tmp
C:\Temp\dst112.tmp
C:\Temp\dst113.tmp
None
None
Options;
Type=Line;

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 281
 Length /Measure='Mile';
Distance=Perp /Measure='Foot';
Area /Units='Square mile';
Object=Single;
SF=1.0;
Selection=Sequential;
Lookahead=1;
Maxterms=5;
Confidence=95;
Print=Selection;
End;
Data /Structure=Flat;
Fields=STR_LABEL, STR_AREA, SMP_LABEL, SMP_EFFORT, the MRDS engine;
Infile=C:\Temp\dst10D.tmp /NoEcho;
End;
Estimate;
Distance /Intervals=0,1,2,3,4,5,6,7,8 /Width=8 /Left=0;
Density=All;
Encounter=All;
Detection=All;
Size=All;
Estimator /Key=HN /Adjust=CO /Criterion=AIC;
Monotone=Strict;
Pick=AIC;
GOF;
Cluster /Bias=GXLOG;
VarN=Empirical;
End;
Example command file

Header Section
This section is required in all command files. Here, you specify the names of the
output files that Distance will generate. If the files do not exist, they will be
created. If they exist, they will be overwritten. The section is 6 lines long, and
each line corresponds with the following file:
• Output file
• Log file
• Stats file
• Plot file
• Bootstrap file

• Bootstrap progress file

If you are not using the bootstrap to estimate variance, you can specify “None”
in the bootstrap and bootstrap progress file lines.
For information about the contents of these files once an analysis has run (or
while the analysis is running in the case of the bootstrap progress file), see the
section Output From the MCDS Engine.
Example:
C:\temp\dst6FA1.tmp
C:\temp\dst6FA0.tmp
C:\temp\dst6FA2.tmp
C:\temp\dst6FA3.tmp
None
None

If you do not include a path for the files (e.g., just dst6FA1.tmp in the
above, for the first file), it is created and written into the current working
directory (the directory you called the program from).

282 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 In previous versions of Distance, the CDS and MCDS engine
required 5 header lines, and not six (because there was no bootstrap progress
file). Also, the bootstrap file came before the plot file. So, if you have any code
for calling previous versions, you’ll need to update it to call the new version.

Options Section
Various options can be set to control program operation. Once an option value
has been set, it retains its value until you change it or exit the program. The data
options define the characteristics of the data collected and how they are to be
entered. The model fitting options define values to be used in fitting a
probability density function to the distance data, some of which can be
overridden in the estimation procedure. Print options control the amount and
format of program output and bootstrap options control the number of bootstrap
samples and the random number seed used to generate a bootstrap sequence.
This section should always begin with the command OPTIONS and end with the
END command.
Below are the valid commands in the options section by category. Each option
and its possible values are individually described in the following sections in
alphabetical order.
Miscellaneous
DEFAULT command Options reset to default
END command Ends options section
LIST command List option values

Output
PRINT command Controls amount of output
QQPOINTS command Max number of points in qq plot
TITLE command Value of output title

Data Options
AREA command Set area quantities
CUERATE command Set cue rate
DISTANCE command Set distance quantities
LENGTH command Set length quantities
OBJECT command SINGLE or CLUSTER
SF command Sampling fraction
TYPE command POINT, LINE or CUE

Model Fitting
LOOKAHEAD command Max for sequential fit
MAXTERMS command Max # model parameters
PVALUE command Significance level (α-level)
SELECTION command Term selection mode

Model Fitting
BOOTSTRAPS command # of bootstrap samples
SEED command Random number seed

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 283
 AREA Command
Syntax:
AREA /CONVERT=value /UNITS='label' ;
Description:
This command defines the area unit for expressing density (D). The switches
are:
/UNITS='label' - a label for the unit of area of the density estimate. The single
quotes are only required to retain lowercase. Only the first 15 characters are
used.
/CONVERT=value - value specifies a conversion factor which is used to
convert the estimated density to new units for area. It is needed for atypical
units.
If the MRDS engine recognizes the measurement unit for DISTANCE (and
LENGTH for line transects) and if it recognizes the Area UNITS label, it will
calculate the appropriate conversion factor. However, if one or more of the
UNITS is not recognized, you will need to specify the conversion value with the
CONVERT switch. The Area units recognized by the program are those listed
under the DISTANCE command and HECTARES (HEC) and ACRES (ACR).
For example, the unit can be entered as Squared Meters or Metres Squared
because the MRDS engine recognizes the unit based on the character string
MET. See the the MRDS engine command below for a definition of recognized
units
Default: AREA /UNITS=HECTARES;
Examples:
Distances are measured in feet but analyzed in meters, length is measured in
miles and density is estimated as numbers per square kilometer. The MRDS
engine will do necessary unit conversions because all unit labels are recognized.
DISTANCE /MEASURE=FEET /UNITS='Meters';
LENGTH/UNITS='Miles';
AREA /UNITS='Sq. kilometers';

BOOTSTRAPS Command
Syntax:
BOOTSTRAPS=value ;
Description:
“Value” is the number of bootstrap samples which should be generated. For a
reasonable variance estimate, this number should be at least 100. We
recommend setting BOOTSTRAPS=999 or 1000 to construct a bootstrap
confidence interval.
Default: BOOTSTRAPS=1000;

CUERATE Command
Syntax:
CUERATE = value1 /SE=value2 /DF=value3;
Description:
For cue counting, “value1” is the average rate at which animals issue visual or
auditory detection cues. The rate should be given in the same units of time as
the values given for sampling effort in the data. For example, if effort is
measured in hours then the cue rate should be number of cues per hour. The cue

284 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 rate must be a positive number (>0). Optionally a standard error for the cue rate
can be given with “value2”, and the degrees of freedom can be given with
“value3” (a DF of 0.0 is interpreted as infinite degrees of freedom). The
standard error and df is accounted for in the estimated standard error of the
density and abundance estimates. This option is only used if TYPE=CUE is
specified.
Default: CUERATE=1 /SE=0 /DF=0;
Example:
An estimate of the cue rate is 12 per hour with a standard error of 2 per hour and
93 degrees of freedom. The sample effort for this cue counting example be
specified in hours sampled.
CUERATE=12 /SE=2 /DF=93;

DEFAULT Command
Syntax:
DEFAULT ;
Description:
This command resets all of the options to their default values. Remember that an
option remains in effect until it is changed or the MRDS engine is terminated.
The default values for each of the options are:

PVALUE = 0.15 PRINT = SELECT TYPE = LINE

SELECT = SEQUENTIAL SQUEEZE = OFF OBJECT = SINGLE
MAXTERMS = 5 BOOTSTRAPS = 1000 DISTANCE = PERP / EXACT / UNITS = METERS
ITERATIONS = 100 SEED = 0 LENGTH / UNITS = KILOMETERS
LOOKAHEAD = 1 SF = 1 AREA / UNITS = HECTARES
CUERATE = 1 / SE = 0 / DF = 0

DISTANCE Command
Syntax:
/ NCLASS = nclass
/ WIDTH = width
PERP / UNITS =' label '
/ CONVERT = value
DISTANCE = / RTRUNCATE = t ;
/ MEASURE =' label '
RADIAL / INTERVALS = c0 , c1, ..., cu
/ LEFT = left
/ EXACT

Synonyms: RIGHT=WIDTH
Description:
This command describes numerous features about the distance data and defines
the default values for estimation. The format of the data entry within the Data
section is determined by the values set with this command. Whereas, the
DISTANCE command in the Estimate section only determines how the distance
data are analyzed.
For line transect data (TYPE=LINE), this command defines whether the data
will be entered as either perpendicular distances or as radial distance and angle
measurements.
• PERP - perpendicular distance was measured for a line transect
• RADIAL - radial distance and angle were measured in line
transects

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 285
 For TYPE=POINT (which includes trapping webs) or CUE, the MRDS
engine=RADIAL is assumed and only radial distances are expected.
Distances can be entered as ungrouped or grouped. Ungrouped implies an exact
distance is entered for each observation in the data. Grouped means a set of
distance intervals is given and the frequency of observations in each interval is
entered. Ungrouped distances are indicated by the switch /EXACT and grouped
data is indicated by the /INTERVALS switch which also specifies the distance
intervals (c0-c1,c1-c2,c2-c3,...). The value c0 specifies the left-most distance
and cu the right-most distance for grouped data. Typically, c0=0 and cu=w.
Intervals can also be specified by using the /NCLASS and /WIDTH and
optionally the /LEFT switch. These switches will create 'nclass' equal width
distance intervals between the values of 'left' and 'width' (i.e., each interval is of
length (width-left)/nclass). For ungrouped data, it is also possible to specify left
and right truncation with the /LEFT and /WIDTH switches. Any values outside
of these bounds are excluded from the analysis. Right truncation as a percentage
of the observations can also be specified for both grouped and ungrouped data
with /RTRUNCATE switch. The value of t must be between 0 and 1. In the
analysis, no more than t*100% of the data is truncated from the right. For
ungrouped data, the width is set at the distance which represents the (1-t)*100%
quantile. For grouped data, intervals are truncated from the right as long as no
more than t*100% of the data is truncated. If t=0 and the data are ungrouped
data, the width is set to the largest distance measurement and if the data are
grouped, the width is set to the endpoint for the right-most interval with a
non-zero frequency. For ungrouped data, if both the /WIDTH and
/RTRUNCATE switch are specified, the RTRUNCATE value specifies the value
of width.
The DISTANCE command is also used to define the measurement unit for
distances:
/MEASURE = 'label' - a label for the units in which distance was measured.
Single quotes are only required to retain lowercase. Only the first 15 characters
are used.
/UNITS='label' - a label for the units for distance after conversion, if any.
Single quotes are only required to retain lowercase. Only the first 15 characters
are used.
/CONVERT=value - value specifies a conversion factor which is used to
convert the input distances for atypical units.
MEASURE and UNITS switches are used to convert from the unit in which the
data are recorded and entered (MEASURE) to the unit for analysis (UNITS). It
is not necessary to convert distances to different units for analysis as long as it is
a unit that is recognized by the MRDS engine (see list below). It is only
provided as a convenience and it is probably easier to leave measurements in
their original units. If you do convert units, take note that values such as f(0),
h(0), effective strip width (ESW) and effective detection radius (EDR) are
expressed in the converted units. Thus, the point estimate and standard errors
will change by the conversion factor from the measured to analysis units. If you
are not converting distance units, you can specify the units with either switch
(/MEASURE or /UNITS). The most common measurement units are recognized
by the MRDS engine and there is no need to enter a conversion value
(/CONVERT= value).
The following are the recognized measurement unit labels:
• CENTIMETERS
• METERS
• KILOMETERS
• MILES

286 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 • INCHES
• FEET
• YARDS
• NAUTICAL MILES
Each label is recognized by its first 3 characters which allows variations in
spelling. For example, if you enter METRES, it will use METRES as the label
and will recognize it based on MET. Values are given in uppercase but can be
entered in upper or lowercase. If the MRDS engine recognizes the /UNITS and
/MEASURE labels and you specify the /CONVERT= switch it will display a
warning message that you are overriding the conversion value. Values for
/WIDTH, /LEFT, and /INTERVALS should be given in original measurement
units and not in converted units.
Default:
DIST=PERP /UNITS='Meters' /MEASURE='Meters' /EXACT /LEFT=0
/RTRUNCATE=0;
Examples:
Perpendicular distance measured in intervals of 2 feet to a distance of 10 feet and
converted to metres (meters) for analysis. The grouped data are entered as the
frequency of observations in each of the 5 distance intervals (see the Data
section). Notice that WIDTH is specified in the original measurement units of
feet and not in meters.
DIST=PERP /MEASURE='Feet' /UNITS='Metres' /WIDTH=10
/NCLASS=5 ;

LENGTH Command
Syntax:
LENGTH /CONVERT=value /UNITS='label' /MEASURE='label' ;
Description:
This command sets the measurement unit for line length and any desired
conversion to different units for analysis. It is not necessary to convert line
length, but may be desirable depending on the original units.
/MEASURE='label' - a label for the units in which line length was measured.
Single quotes are only required to retain lowercase. Only the first 15 characters
are used.
/UNITS='label' - a label for the units for length after conversion, if any. Single
quotes are only required to retain lowercase. Only the first 15 characters are
used.
/CONVERT=value - value specifies a conversion factor which is used to
convert length measured in atypical units.
See further explanation under the DISTANCE command for the /MEASURE,
/UNITS and /CONVERT switches. The LENGTH command is used for line
transects only.
Default:
LENGTH /UNITS=KILOMETERS /MEASURE=KILOMETERS;
Example:
Length is entered in miles but converted to kilometers for display and analysis.
LENGTH /UNITS='Kilometers' /MEASURE='Miles' ;

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 287
 LIST Command
Syntax:
LIST;
Description:
Lists current values of the program options and the program limits to the screen.

LOOKAHEAD Command
Syntax:
LOOKAHEAD=value ;
Description:
For term selection modes SEQUENTIAL and FORWARD (see SELECTION
command), “value” specifies the number of adjustment terms which should be
added to improve the fit, before the added terms are considered to be non-
significant. For example, if LOOKAHEAD=2 and a model with 2 adjustment
terms does not significantly improve the fit over a model with 1 term, a model
with 3 adjustment terms is fitted. If the 3-term model is an improvement over a
1-term model, the algorithm will continue with the 3-term model as the new base
model. If it is not an improvement, the 1-term model would be chosen. If
LOOKAHEAD=1 (the default), in the above example, the 3-term model would
not have been examined because upon finding the 2-term model was not an
improvement, the 1-term model would have been used.
Default: LOOKAHEAD=1;

MAXTERMS Command
Syntax:
MAXTERMS=value ;
Description:
“Value” is the maximum number of model parameters. The maximum number
of adjustment terms (defined as m) that may be added is MAXTERMS minus the
number of parameters in the chosen key function (defined as k). MAXTERMS
must be less than or equal to 5. This option is only useful to limit the number of
model combinations with the term selection mode that considers also possible
combinations of adjustment terms (SELECTION=ALL). Use the NAP switch
on the Estimator command to specify an exact number of adjustment terms to be
used. The maximum number of adjustment terms is also limited by the number
of observations for ungrouped data or number of distance intervals for grouped
data.
Default: MAXTERMS=5;

OBJECT Command
Syntax:
SINGLE
OBJECT = ;
CLUSTER

Description:
This option defines whether objects are detected individually (SINGLE) or as
clusters (CLUSTER).
SINGLE - Object always detected as a single animal or other entity (e.g., duck
nest)
CLUSTER - Object detected as a cluster (e.g., herd, flock,pod of whales)

288 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 Default: OBJECT=SINGLE;

PRINT Command
Syntax:
SELECTION
RESULTS
PRINT = ;
ALL
SUMMARY

Description:
This option sets the default level of printing in the output. The various settings
are hierarchical and more control over the amount of results can be obtained with
the PRINT command in the Estimate section.
ALL - print fitting iterations, model selection results and estimation results
SELECTION - print model selection results and estimation results
RESULTS - print estimation results only
SUMMARY (NONE) - only summary tables are printed
Note: if you choose RESULTS or SUMMARY, warnings are not given about the
algorithm having difficulties fitting a particular model or constraining the fit to
achieve monotonicity.
Default: PRINT=SELECTION;

PVALUE Command
Syntax:
PVALUE=α;
Description:
α is the significance level of likelihood ratio tests to determine significance of
adding adjustment terms and is the default value for the significance test for size-
bias regression of cluster sizes
Default: PVALUE=0.15;

QQPOINTS Command
Syntax:
PVALUE=value;
Description:
Maximum number of points to print in qq-plots. When there are a large number
of data points, plotting all the points can take quite a while and result in a very
large plot file. The default, 0, means no maximum – i.e., plot every point.
Default: QQPOINTS=0;

SEED Command
Syntax:
SEED=value ;
Description:
SEED specifies the random number seed for generating a sequence of random
numbers for bootstrap samples. “Value” should be a large odd number
preferably greater than 2,000,000. If you use the same seed, the same sequence

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 289
 of random numbers will be generated. You can use SEED=0; (the default)
which will use a value from the computer's clock to generate a seed.
Default: SEED=0;

SELECTION Command
Syntax:
SEQUENTIAL
FORWARD
SELECTION = ;
ALL
SPECIFY

Description:
This command specifies the default mode for adjustment term selection in the
ESTIMATE procedure for fitting the detection function. The /SELECT switch
of the ESTIMATOR command overrides the default value. See the Estimate
section for a description of adjustment term and model selection.
SEQUENTIAL - add adjustments sequentially (e.g., for simple polynomial, in
increasing order of the exponent).
FORWARD - equivalent to forward selection in regression; select the
adjustment term which produces the largest increase in the maximum of the
likelihood function
ALL - fit all combinations of adjustment terms with the key function and use
the model with smallest Akaike Information Criterion (AIC) value.
SPECIFY - user-specified number of adjustment terms and possibly order of the
adjustments.
Default: SELECTION=SEQUENTIAL;

SF Command
Syntax:
SF=c ;
Description:
SF defines the value of the sampling fraction which is typically 1. However, if
only one side of a transect line is observed c=0.5, or if some fraction of the circle
surrounding a point transect is searched, c is the fraction searched (e.g.,c=0.5 if a
semi-circle is observed). For cue counting, c is the proportion of a full circle that
is covered by the observation sector. For a sector of 90o (45o either side of the
line) with cue counting, c = 0.25
Note that SF can now be specified using the MULTIPLIER command, with
SE=0, and this is the way that Distance does it.
Default: SF=1;

TITLE Command
Syntax:
TITLE='yourtitle' ;
Description:
This command sets a value for the title which is printed at the top of each page.
Yourtitle should contain no more than 50 characters. Excess characters are not
used. There is only 1 title line. Re-specifying the title will replace the previous
value.

290 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 TYPE Command
Syntax:
POINT
TYPE = LINE ;
CUE

Description:
This option defines the type of sampling, which determines what types of data
can be entered and how data are analyzed.
POINT - point transect data
LINE - line transect data
CUE - cue counting data
Trapping webs should be treated as point transects.
Default: TYPE=LINE;

Data section
In this section, you specify the file containing data, and which column in this file
corresponds with which field. For more about the format of the data file, see see
the section describing the MCDS Engine Required Data Format.
The data section should always begin with the statement DATA
/STRUCTURE=FLAT and end with the statement END. Note that historical
versions of this engine used a hierarchical data format, but that is no longer
supported, so the /STRUCTURE=FLAT switch is now mandatory.
The commands that are valid in the Data section are listed in alphabetical order
below, and described in the following sections.
Data section commands
END command Ends data section
FACTOR command Specifies that a field is a factor covariate
FIELDS command List of fields in the data file
INFILE command Gives filename of data file
SIZEC command Specifies that a field is the cluster size
covariate

FACTOR command
Syntax:
FACTOR /NAME='fieldname' /LEVELS=value /LABELS= 'label1',
'label2', … ;
Description:
This command defines a field in the data file as a factor covariate in MCDS
analyses. For more about factor covariates, see the section on factor and non-
factor covariates in MCDS in Chapter 9 of the Users Guide. Covariates for the
detection function are specified in the Estimator command.
There should be one FACTOR command for each factor field in the data file. If
there are no factor fields, this command will not be present.
/NAME='fieldname' – the name of the field (must be one of the names in the
FIELDS command)
/LEVELS=value – the number of levels in the factor covariate

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 291
 /LABELS= 'label1', 'label2', … - a comma-delimited list giving the value of
each level of the factor
Default: no FACTOR command
Examples:
The data file contains a column for obsererver, which is specified in the FIELDS
command as “Observer”. Observer is to be used as a factor covariate in the
detection function, and each observation can take one of three possible values
“Peter”, “Paul” and “Mary”.
FACTOR /NAME=Observer /LEVELS=3 /LABELS=Peter, Paul, Mary;

FIELDS command
Syntax:
FIELDS= fieldname1, fieldname2, fieldname3, …
Description
This command gives a list of the fields occuring in the data file, reading the
columns of the data file from left to right. The following fieldnames are required:
• SMP_LABEL – sample label
• SMP_EFFORT – sample effort (line length/number of points)
• DISTANCE – perpendicular or radial distances (depending on the
TYPE and DISTANCE commands)
If the TYPE = LINE and DISTANCE = RADIAL then an additional required
field is
• ANGLE – angle of radial distances, in degrees
If OBJECT = CLUSTER then another required field is
• SIZE – the cluster size
Two additional fields with fixed names may be specified:
• STR_LABEL – stratum label
• STR_AREA – stratum area – if areas are ommitted then density but
not abundance is calculated
If covariates are specified in the ESTIMATOR command then these should be
included in the data file, and their names listed in the FIELDS command. In
addition to being listed in the FIELDS command, factor covariates should be
declared as such using a FACTOR command.
Default: No default
Examples:
Standard line transect data, with a column for stratum label, area, transect label,
line length and perpendicular distance:
Fields=STR_LABEL, STR_AREA, SMP_LABEL, SMP_EFFORT,
DISTANCE
Line transect data, with radial distance and angle, objects as clusters, and an
additional field for an Observer covariate:
Fields=STR_LABEL, STR_AREA, SMP_LABEL, SMP_EFFORT,
DISTANCE, ANGLE, SIZE, Observer

INFILE Command
Syntax:

292 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 / ECHO
INFILE = filename ;
/ NOECHO

Description:
This command specifies the data file name. Filename should either give the full
absolute path to the file, or just the filename if the file is in the current directory.
The ECHO and NOECHO switches control whether the data are ECHOed to the
LOG file. Once you are certain that the data are free of errors, using /NOECHO
will reduce the amount of output to the LOG file.
Example:
Infile=C:\temp\dst7035.tmp /NoEcho;

SIZEC command
Syntax:
SIZEC=fieldname;
Description
Specifies that the field “fieldname” is the cluster size field, when cluster size is a
covariate in the detection function.

Estimate section
The following are valid commands in the estimate section:
Estimate section commands
BOOTSTRAP command bootstrap variance/confidence intervals
CLUSTER command estimation of expected cluster size
DENSITY command resolution of density estimation
DETECTION command resolution of detection probability estimation
DISTANCE command analysis treatment of distances
(Estimate section)
ENCOUNTER command resolution of encounter rate estimation
END command initiates estimation
ESTIMATOR command model for g(x)
G0 command estimate of g(0) and its standard error
GOF command intervals for goodness of fit test/display
MONOTONE command monotonicity constraints on g(x)
MULTIPLIER command multipliers in the detection function
PICK command method of model choice
PRINT command (Estimate detailed control of output
section)
SIZE command resolution of expected cluster size estimation
VARN command variance estimation of n

The commands are described below in alphabetical order. You will use these
commands to define:
• which quantities you want to estimate and at what level of
resolution (DENSITY, DETECTION, ENCOUNTER, SIZE ),
• how distance and cluster size are treated in the analysis and which
models are used for estimation (DISTANCE, CLUSTERS,
ESTIMATOR, MONOTONE, PICK, GOF, G0),

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 293
 • how variances are estimated ( VARN, VARF, BOOTSTRAP), and
• how much output should be generated ( PRINT ).
Density and abundance estimates are comprised of the following components:
• detection probability,
• encounter rate,
• expected cluster size (if the detected objects are clusters).
It is possible to restrict estimation to one or more of these components without
estimating density; however, all components must be estimated to obtain an
estimate of density. You will use the commands DENSITY, DETECTION,
ENCOUNTER, and SIZE to define which components will be estimated. If you
do not use any of these commands, each component and density is estimated, by
default. Likewise, if you use the DENSITY command, density and all of its
components are estimated. If you use any or all of the DETECTION,
ENCOUNTER, and SIZE commands and not the DENSITY command, only the
specified components are estimated. For example,
ESTIMATE;
ENCOUNTER ALL;
END;
will only estimate encounter rate.
Estimates of density and its components can be made at different levels of the
sampling hierarchy (Sample < Stratum < All). The DENSITY, DETECTION,
ENCOUNTER, and SIZE commands are used to specify the level at which each
quantity is estimated. Different levels can be used for the various quantities;
although, some combinations are incompatible. An error message is given, if the
levels are incompatible. The lowest level of resolution specified for DENSITY
is the default level for each of its components, if they are unspecified. For
example,
ESTIMATE;
ESTIMATOR /KEY=UNIFORM;
DENSITY BY STRATUM;
END;
will estimate density and each of its components for each stratum defined in the
data. The lowest level for density must coincide with a level assigned to
encounter rate. The level of any component cannot be lower then the lowest
level specified for density. For example, the following is not valid:
ESTIMATE;
ESTIMATOR /KEY=UNIFORM;
DENSITY BY STRATUM;
DETECTION BY SAMPLE;
END;
If a size-bias regression estimate of expected cluster size is computed, the level
for SIZE must be no greater than the level for DETECTION. This feature is
most useful for estimating density by stratum when too few observations exist in
each stratum to estimate f(0) (or h(0)). A solution is to assume f(0) is the same
for all strata, which is illustrated in the following example:
ESTIMATE;
ESTIMATOR /KEY=UNIFORM;
DENSITY BY STRATUM;
DETECTION ALL;
END;
All of the observations are pooled to estimate a common value for f(0), which is
used in each stratum density estimate.

294 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 When there are covariates in the detection function, it is possible to fit the
detection function at one level, and then estimate probability of detection at a
lower level. For more on this, see Chapter 9 of the Users Guide Estimating the
detection function at multiple levels. An example of this, with a global detection
function fit with a habitat covariate which is specific to each stratum, and then
the detection function estimated by stratum is:
ESTIMATE;
ESTIMATOR /KEY=UNIFORM /COVARIATES=Habitat;
DENSITY BY STRATUM;
DETECTION ALL;
DETECTION BY STRATUM
END;
Possibly the most confusing aspect of estimation with the MRDS engine will be
the specification of models for detection probability and model selection. A
model is specified with the ESTIMATOR command which defines a type of key
function and adjustment function. The adjustment function is actually a series of
terms which are added to the key function to adjust the fitted function to the
data. Model selection includes 1) selecting how many and which adjustment
terms are included in the model (term selection) and 2) selecting a “best” model
(estimator) from the specified set of competing models.
The default method of selecting terms (term selection mode) is defined by the
SELECTION command in the Options section. Its value can be overridden with
the /SELECT switch of the ESTIMATOR command. Related options include
LOOKAHEAD and MAXTERMS. There are 4 types of term selection modes
described below: 1) SEQUENTIAL, 2) FORWARD, 3) ALL, and 4) SPECIFY.
The maximum number of adjustment terms that can be included in the model is
limited by the value of (MAXTERMS - number of parameters in the key
function) and less frequently by either the number of observations, for
ungrouped data, or the number of distance intervals for grouped distance data.
The MRDS engine will issue a warning message if the number of parameters is
being limited by the amount of data.
Term selection mode SPECIFY implies the user will specify which adjustment
terms are included in the model. Typically, this is used to specify that a key
function without adjustment terms is to be fitted to the data, as in the following
example:
ESTIMATE;
ESTIMATOR /KEY=HNOR /NAP=0 /SELECT=SPECIFY;
END;
It is not necessary to include the /SELECT switch, but it will prevent the MRDS
engine from issuing a warning message that you are specifying the model. It is
also possible to specify any combination of terms and give starting values for
their coefficients. the MRDS engine does not select the terms to include in the
model but does estimate the parameters to fit the model to the data. For
example,
ESTIMATE;
ESTIMATOR /KEY=UNIFORM /NAP=2
/SELECT=SPECIFY /ORDER=1,3;
END;
specifies the model as the following 2-term cosine series for which the
parameters a1 and a2 are estimated:
1⎛ ⎛ πx ⎞ ⎛ 3πx ⎞ ⎞
f ( x) = ⎜⎜1 + a1 cos⎜ ⎟ + a 2 cos⎜ ⎟ ⎟⎟
w⎝ ⎝w⎠ ⎝ w ⎠⎠
ALL, as its name implies, examines all possible combinations of a limited
number of adjustment terms. If z is the maximum number of parameters

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 295
 (MAXTERMS=z) and k is the number of parameters in the key function, then
there are 2z-k combinations of the adjustment terms. Each model is fitted to the
data and the model with the smallest value of the Akaike's Information Criterion
(AIC) is selected.
SEQUENTIAL and FORWARD both consider a subset of models with different
combinations of adjustment terms. For each of these term selection modes, a
sequence of models is considered. An adjustment term is added at each step of
the sequence. The sequence of models can be represented as:
M1- key function with no adjustment terms
M2- key function with 1 adjustment term
M3- key function with 2 adjustment terms
:
:
Mv- key function with v-1 adjustment terms
A stopping rule (CRITERION) is either based on a likelihood ratio test or
minimizing AIC. Model Mt is chosen if there is no model in the sequence
Mt+1,...,Mt+l which provides a significantly better fit as determined by the
specified CRITERION. The LOOKAHEAD option determines the length (l) of
the sequence of models that is examined before choosing model Mt.
SEQUENTIAL and FORWARD only differ in their choice of which adjustment
term is included at each step in the sequence. SEQUENTIAL term selection
adds the terms sequentially based on the order of the term. For polynomial
adjustment functions, the order of the adjustment term is the exponent of the
t t +2
polynomial. Terms are added in the following sequence: ( x , x ,... . For
cosine adjustments, cosine terms are added in the following
sequence: cos(tπx w), cos((t + 1)πx w),... . The beginning value, t, is determined
by the shape of the key function. FORWARD selection adds 1 term at a time but
not necessarily in sequential order. For each model in the sequence, each term
not already in the model is added and the adjustment term which increases the
likelihood the most is chosen as the term to add. For example, to find model
M2, z-k models are fitted to the data, each with a single adjustment term of a
2 4 6 8 10
different order (e.g., x , x , x , x , or x ). The term which maximizes the
likelihood is selected for model M2. Model M3 would then consider adding
another term not included in M2. With FORWARD selection it is possible to
select models that cannot be selected with the SEQUENTIAL mode. For
example, the following model might be chosen with FORWARD selection:
1⎛ ⎛ πx ⎞ ⎛ 3πx ⎞ ⎞
f ( x) = ⎜1 + a1 cos⎜ ⎟ + a 2 cos⎜ ⎟ ⎟⎟
w ⎜⎝ w
⎝ ⎠ ⎝ w ⎠⎠

⎛ 3πx ⎞
However, with SEQUENTIAL selection, the adjustment term cos⎜ ⎟ , could
⎝ w ⎠
⎛ 2πx ⎞
not be added without first adding the adjustment term cos⎜ ⎟.
⎝ w ⎠
The additional level to model fitting is to choose between the competing models
(ESTIMATORs). This model selection step is determined by the PICK
command. It has 2 values: NONE and AIC. If you assign the value NONE, the
MRDS engine will not choose between the different models and will report the
estimates for each model. However, if you accept the default value, AIC the
MRDS engine will only compute estimates based on the model which has the
smallest AIC value.

BOOTSTRAP Command
Syntax:

296 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 / STRATUM
BOOTSTRAP / SAMPLES / OBS ;
/ INSTRATUM

Description:
The BOOTSTRAP command initiates a non parametric bootstrap of the density
estimation procedure. The number of bootstraps performed is determined by the
BOOTSTRAP command in OPTIONS. The basic re-sampling unit of the
bootstrap is a SAMPLE; however, if strata are replicates they can also be re
sampled with the /STRATUM switch. If both are specified, re-sampling occurs
at both levels (see example below). The switch /INSTRATUM can be set to
restrict the re-sampling of samples or observations within stratum It would be
used if density is estimated by stratum or sampling was stratified apriori.
The switch /OBS can be set to re-sample distances. Using BOOTSTRAP/OBS;
will provide a non-parametric bootstrap of f(0) or h(0) and, if the population is
clustered, E(s). However, the variances and confidence intervals are conditional
on the sample size and do not include the variance of the encounter rate. The
/OBS switch has been included for completeness but its routine use is not
recommended. Reasonable confidence intervals for density could only be
obtained by adding a variance component for the encounter rate. It is also
possible to include the /OBS switch with /SAMPLES, however, this is not
recommended unless the number of observations per sample is reasonable ( >
15).
By default, issuing the BOOTSTRAP command without switches is equivalent
to BOOTSTRAP/SAMPLES/INSTRATUM;. We recommend the default or
dropping the /INSTRATUM if sampling across strata is appropriate. The use of
/STRATUM is only appropriate if the strata represent an additional level of
sampling (e.g., independent observers (stratum) traversing an independent set of
line transects (sample)).
Each bootstrap resample is made up by sampling with replacement an equal
number of units at the level you specify. For example, if you specify to resample
samples within strata (the default), then each bootstrap resample is made up of
the same number of samples (line or point transects) as your original sample,
chosen randomly with replacement from the original sample (within each
stratum). Note that for line transects, this means that the survey effort (total line
length) will differ in each resample. Note also that each of your original samples
has an equal probability of appearing in the resample (an alternative, which we
do not implement, would be to have probability proportional to line length).
The bootstrap summary is given at the end of the output. The point estimate is
the mean of the bootstrap point estimates. Two sets of confidence intervals are
given: 1) log based confidence intervals based on a bootstrap standard error
estimate, and 2) 2.5% and 97.5% quantiles of the bootstrap estimates (i.e.,
percentile confidence intervals).
Summary results from each iteration are stored in the Bootstrap file – see MCDS
Engine Bootstrap File for details.
Example:
Re sample strata and samples within each stratum.
BOOTSTRAP /STRATUM /SAMPLES;

CLUSTER Command
Syntax:

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 297
 CLUSTER / WIDTH = value / TEST = α

/ MEAN
X
XLOG
/ BIAS =
GX
GXLOG

Description:
The CLUSTER command, like the the DISTANCE command, is used to modify
the way the cluster sizes are used in the estimate of density. By default, the
WIDTH is chosen to match the truncation value set by the the MRDS engine
command and the MRDS engine computes a size bias regression estimate
(/BIAS=GXLOG) by regressing the loge(s) (natural logarithm specified as log()
in the output) against (x), where x is the distance at which the cluster was
observed.
The WIDTH switch specifies that only cluster sizes for observations within a
distance less than WIDTH are used in the calculation of the expected cluster
size. This treatment of the data can only be accomplished if the distances and
cluster sizes are both entered as ungrouped.
The MEAN switch specifies that the expected cluster size is to be estimated as
the average (mean) cluster size. Likewise, the BIAS switch specifies that
expected cluster size is to be estimated by a size bias regression defined by the
value of the switch.
Value Meaning
X Regress cluster size against distance x
XLOG Regress loge(s) against distance x
GX Regress cluster size (s) against (x)

The TEST switch specifies the value of the significance level to test whether the
regression was significant. If it is non significant, the average cluster size is
used in the estimate of density. The default value for the significance level is set
by PVALUE in OPTIONS. If the TEST switch is not specified, the size bias
regression estimate will be used regardless of the test value.
Examples:
Estimate the expected cluster size from the loge(s) vs. (x) regression, but use the
average cluster size if the correlation is non significant as determined by the -
level set with PVALUE (default=0.15).
CLUSTER /BIAS=GXLOG /TEST;

DENSITY Command
Syntax:
NONE
DENSITY by SAMPLE /DESIGN = ;
REPLICATE

or

298 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 NONE
/DESIGN =
STRATA
REPLICATE
DENSITY by STRATUM
EFFORT
/WEIGHT = AREA ;
NONE

or
DENSITY by ALL;

Description:
These commands define the levels at which density estimates are made and how
these estimates are weighted. If the DENSITY by ALL; command is used or if
none of the commands (DENSITY, ENCOUNTER, DETECTION, SIZE ) are
given, all of the data are used to make one overall estimate of density.
If the DENSITY BY SAMPLE command is given, density is estimated for each
sample. The DESIGN value defines how the estimates should be treated to
create a pooled estimate. If DESIGN=REPLICATE (default), each sample is
treated as an independent replicate from the stratum or the entire area. In this
case, the estimates are weighted by effort (e.g., line length) to get a stratum
density estimate (if DENSITY by STRATUM is also specified) or a pooled
overall density estimate (see eqns. 3.84-3.87 in Buckland et al. (2001)). If
DESIGN=NONE, the sample estimates are not pooled.
If DENSITY BY STRATUM is specified, an estimate is made for each stratum.
A stratum estimate is a pooled estimate of the sample estimates within the
stratum, if DENSITY by SAMPLE; is specified, or it is an estimate based on the
data within the stratum. An overall (pooled) estimate of density is made unless
DESIGN=NONE is specified. If DESIGN=REPLICATE, the stratum estimates
are treated as replicates to create a pooled estimate and variance weighted by
effort (eqns. 3.84-3.87 in Buckland et al. (1993), treating stratum as a sample).
If DESIGN=STRATA, the pooled estimate is a weighted sum of the estimates
and the variance is a weighted sum of the stratum variances (Section 3.7.1 in
Buckland et al. (2001)). Weighting is defined by the WEIGHT switch. If
WEIGHT=NONE, the densities are summed, which is only useful if the
population is stratified as by sex or age. If WEIGHT=AREA, the densities are
weighted by area (which is the same as adding abundance estimates) and if
WEIGHT=EFFORT, the densities are weighted by effort.
Prior to version 2.1, a combined estimate of abundance (N) was created by
multiplying the combined density estimate by the sum of the areas specified on
each of the STRATUM commands. This produces obviously erroneous results
when DENSITY by STRATUM/DESIGN=REPLICATE; is used and the area
size is repeated on each STRATUM. To avoid this problem in the following two
situations the combined abundance estimate uses the area from the first stratum:
1) DENSITY by STRATUM/DESIGN=REPLICATE;
2) DENSITY by STRATUM/DESIGN=STRATA/WEIGHT=NONE;
In all other cases, the area is totalled from all of the strata.
Default:
For DENSITY by SAMPLE: /DESIGN=REPLICATE
For DENSITY by STRATUM: /DESIGN=STRATA /WEIGHT=AREA
Examples:
An estimate is needed for each stratum and it will be weighted by stratum area:
DENSITY BY STRATUM;

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 299
 An estimate is needed for each stratum and there are enough observations in
each sample to get an estimate from each. The strata represent different
platforms surveying the same area so the strata are treated as replicates.
DENSITY BY SAMPLE;
DENSITY BY STRATUM /DESIGN=REPLICATE;

DETECTION Command
Syntax:
SAMPLE
DETECTION by STRATUM
ALL

Description:
This command explicitly specifies that detection probability (and its functionals
f(0), h(0)) should be estimated and the resolution at which the estimate(s) should
be made (by SAMPLE, by STRATUM, or ALL data).
When there are covariates in the detection function, it is possible to fit the
detection function at one level, and then estimate probability of detection at a
lower level. For more on this, see Chapter 9 of the Users Guide Estimating the
detection function at multiple levels.
Examples:
Density is estimated by stratum but the estimates are based on an estimate of f(0)
for all the data.
DENSITY by STRATUM;
DETECTION ALL;
Estimate detection by stratum choosing between 2 models but do not estimate
any other parameters. Different models may be selected for each stratum.
ESTIMATE;
DETECTION BY STRATUM;
ESTIMATOR/KEY=HAZ;
ESTIMATOR/KEY=UNIF;
END;
Fit detection function globally using a habitat covariate, but estimate by stratum
(the level at which habitat is defined):
ESTIMATE;
ESTIMATOR /KEY=UNIFORM /COVARIATES=Habitat;
DENSITY BY STRATUM;
DETECTION ALL;
DETECTION BY STRATUM
END;

DISTANCE command (Estimate section)

Syntax:
DISTANCE /WIDTH=value /INTERVALS=c0, c1, … , cu /LEFT=l
/RTRUNCATE=t /NCLASS=nclass /SMEAR=angle,pdist;
Description:
The DISTANCE command is used to specify the way the distances should be
treated in the analysis. It can be used to specify left and right truncation of the
distance, to group the distances into intervals prior to analysis, and to “smear”
radial distance / angle measurements into perpendicular distance intervals.

300 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 Right truncation is specified with either the WIDTH or RTRUNCATE switch. If
WIDTH=w, only distances less than or equal to w, are used in the analysis. If
RTRUNCATE=t, the right truncation distance is set to use (1-t)*100% of the
data. If the data are ungrouped the (1-t)*100 percentile is used as the truncation
distance. If the distances are grouped or being analyzed as such, the truncation
distance is set to the uth interval end point where u is the smallest value such that
no more than t*100% of the distances are truncated. The value of t=0 trims the
intervals to the right most interval with a non-zero frequency. If both the
WIDTH and RTRUNCATE are specified, the value of RTRUNCATE defines
the truncation unless the WIDTH is used with NCLASS (see below).
Left truncation is accomplished with the LEFT switch which works in an
analogous fashion to WIDTH. If LEFT=l, only distances greater than or equal to
l are used in the analysis. If LEFT is not specified, it is assumed to be 0.
The INTERVALS command is used to specify u distance intervals for analyzing
data in a grouped manner when the data were entered ungrouped. The value c0
is the left most value and so it can be used for left truncation. If there is no left
truncation, specify c0=0. The values c1 ,c2 ,..., cu are the right end points for the
u intervals. The value cu is the right-most point and is used as the WIDTH
which defines the right truncation point. If all of the distances are less than or
equal to cu, the MCDS engine will not truncate data on the right unless
RTRUNCATE is set. Perpendicular distance intervals can also be created for
analysis with the NCLASS and WIDTH commands. NCLASS intervals of equal
length are created between “Left” and “Width”, if both NCLASS and WIDTH
are given.
The SMEAR switch is used only if TYPE=LINE and radial distance/angle
measurements were entered (DISTANCE=RADIAL). “Angle” defines the angle
sector around the angle measurement and “Pdist” defines the proportional sector
of distance to use as the basis for the smearing (see pg: 269-271 of Buckland et
al. 2001).
If an observation is measured at angle “a” and radial distance “r”, it is smeared
uniformly in the sector defined by the angle range (a-angle,a+angle) and distance
range (r*(1-pdist),r*(1+pdist)).
The NCLASS and WIDTH switches must also be given to define a set of equal
perpendicular distance intervals. The proportion of the sector contained in each
perpendicular distance interval is summed as an observation frequency and these
non-integer frequencies (“grouped data”) are analyzed to estimate detection
probability.
Note: Distances specified by WIDTH, LEFT, and INTERVALS should be in the
same units used for the entered data, even if the distance units are converted in
the analysis.
Examples:
Truncate the distances at 100 feet, hence only use those less than or equal to 100
feet in the analysis. This value would be used even if the distances were
converted to meters for analysis. The conversion is applied to the input width of
100 feet.
DIST /WIDTH=100;
The distance data were entered ungrouped but they were actually collected in
these intervals; alternatively, to mediate the effects of heaping, these intervals
were chosen to analyze the data.
DIST /INT=0,10,20,30,40,50,60,70,80,90,100;
The above example could also be entered as:
DIST/NCLASS=10/WIDTH=100;

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 301
 ENCOUNTER Command
Syntax:
SAMPLE
ENCOUNTER by STRATUM
ALL

Description:
This command explicitly specifies that encounter rate should be estimated and
the resolution at which the estimate(s) should be made (by SAMPLE, by
STRATUM, or ALL data). This command is only necessary if density is not
being estimated.
Examples:
A user wishes to explore the variability in encounter rate by listing the encounter
rate for each sample. The variance of the encounter rate for each sample is
assumed to be Poisson because the sample is a single entity.
ESTIMATE;
ENCOUNTER by SAMPLE;
END;
A user wishes to explore the variability in encounter rate by listing the encounter
rate for each stratum. The variance of the encounter rate for each stratum is
computed empirically for each stratum with more than one sample; otherwise, it
is assumed to be Poisson.
ESTIMATE;
ENCOUNTER by STRATUM;
END;
A user wishes to only see the average encounter rate and an estimate of its
variance. The variance of the encounter rate is computed empirically if there is
more than one sample; otherwise, it is assumed to be Poisson.
ESTIMATE;
ENCOUNTER ALL;
END;

ESTIMATOR Command
Syntax:

302 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 UNIFORM
COSINE
HNORMAL
/KEY = /ADJUST = POLY
NEXPON
HERMITE
HAZARD
SPECIFY
SEQUENTIAL
/SELECT = /ORDER = O(1), O(2),...,0(nap)
FOREWARD
ESTIMATOR ALL
/NAP = nap /START = A(1), A(2),..., A(nkp + nap)
AIC
AICC
/CRITERION = /COVARIATES = cov1, cov2, ...
BIC
LR
/LOWER = val1, val2,...
/UPPER = val1, val2,...
W
/ ADJSTD =
SIGMA
Description:
The ESTIMATOR command specifies the type of model for detection
probability (g(x)) to estimate f(0) or h(0). The KEY switch specifies the key
function to be used and the ADJUST switch specifies the type of adjustment
function. The SELECT switch specifies the type of adjustment term selection
which overrides the default value specified by the SELECTION command in the
OPTIONS procedure (see the discussion on adjustment term in the introduction
to the Estimate section).
If SELECT=SPECIFY is chosen, you can specify the number of adjustment
parameters, the order of the adjustment term and starting values for the
parameters. The number of adjustment parameters is set with the NAP (Number
of Adjustment Parameters). NAP must be less than or equal to MAXTERMS -
'nkp' (number of key parameters). The orders of the adjustment term(s) are
specified with the /ORDER switch. Starting values (/START) for the key and
adjustment parameters can be given if the optimization algorithm suggests there
are problems in finding the maximum of the likelihood function. The first 'nkp'
starting values in the list should be the values for the key parameters and the
remaining are for the 'nap' adjustment parameters. One reason for using the
SELECT and NAP switches is to specify that only the key function should be
fitted to the data. An example is given below.
CRITERION specifies the manner in which the number of adjustment terms is
chosen for SELECT=FORWARD and SEQUENTIAL. LR specifies that a
likelihood ratio test be performed using the PVALUE specified in OPTIONS.
AIC specifies using the Akaike's Information Criterion for adjustment term
selection.
COVARIATES gives a list of covariates that enter the scale parameter of the
detection function – see the Users Guide Chapter 9 section Introduction to
MCDS Analysis for more on this. Note that the covariates must be declared in
the list of FIELDS in the Data section, and that factor covariates need to be
declared as such using the FACTOR command.
The distances passed into the adjustment term formulae are scaled. ADJSTD
determines how they are scaled – either by W (the truncation width) or SIGMA
(the evaluated value of the scale parameter for this covariate). For a discussion
of the difference, see Scaling of Distances for Adjustment Terms in Chapter 9 of
the Users Guide.
LOWER and UPPER enable you to set bounds on the key function parameters
(note that you cannot currently set constraints on adjustment term parameters).

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 303
 If either are missing, default bounds are used (these are reported in the results).
A value of -99 indicates use the default bound for that parameter. See below for
an example.
Multiple ESTIMATORs can be specified within an ESTIMATE procedure and
the “best” model is selected or estimates are given for each model (see PICK
command). Note that if covariates are used, the same covariates must be
declared in all ESTIMATOR commands – automatic selection among covariates
is not currently supported.
The only portion of the command required is the command name ESTIMATOR
because all of the switches have default values.
Default Values:
KEY = HNORMAL
ADJUST = COSINE
SELECT = SEQUENTIAL (or value set in Options section)
CRITERION = LR (except if SELECT=ALL)
Examples:
Use the following to fit a model with a half-normal key function (by default) and
Hermite polynomials for adjustment functions. DISTANCE fits all possible
combinations of adjustment terms and uses AIC to choose the best set of
adjustment terms:
ESTIMATOR /ADJ=HERM /SEL=ALL;
Use the following to fit a model that uses the uniform key function with simple
polynomial adjustment functions:
ESTIMATOR /KEY=UNIFORM /ADJ=POLY;
Use the following to fit a model that uses the hazard key without adjustments:
ESTIMATOR /KEY=HAZARD /SELECT=SPECIFY /NAP=0;
Use the following to fit a 2-term cosine series with terms of order 1 and 3 and
specify the parameter starting values (note: nkp=0 for a uniform key):
ESTIMATOR /KEY=UNIF /SELECT=SPECIFY /NAP=2 /ORDER=1,3
/START=0.3,0.05;
Use the following to fit a hazard key with one polynomial adjustment, and a
specified lower bound on the second parameter of the key function but the
default lower bound of the first parameter. Imagine in this case that there are
two strata and we want a lower bound of 2 on the 2nd parameter in the first
stratum, and a lower bound of 2.5 on the 2nd parameter in the second stratum.
Not a realistic example perhaps, but illustrates that you have to specify
constraints separately for each stratum, as these are treated as separate key
function parameters to be estimated. Also illustrates that you ignore the
adjustment term parameters when setting bounds.
ESTIMATOR /KEY=HAZARD /ADJ=POLY /SELECT=SPECIFY /NAP=1
/LOWER=-99,2.0,-99,2.5;

G0 Command
Syntax:
G0=value /SE=value /DF=value;
Description:
This command assigns a value to g(0) which is assumed to be 1 unless a value is
assigned with this command. The SE and DF switches are used to specify a
standard error for the estimate so that estimation uncertainty of g(0) can be
incorporated into the analytical variance of density. G0 is just a special case of a

304 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 multiplier, so see the MULTIPLIER command for details of its use and the
options.
Default:
G0=1.0/SE=0.0;
Example:
G0=0.85/SE=0.12;

GOF Command
Syntax:
GOF /INTERVALS = c0, c1, c2, … , cu;
or
GOF /NCLASS=nclass;
or
GOF;
Description:
This command is used to specify the distance intervals for plotting a scaled
version of the histogram of distances against the function g(x) (and f(x) ) and for
the chi-square goodness of fit test .
If the data are entered and analyzed ungrouped (/EXACT), the first 2 forms can
be used to define the intervals which are used for plotting the data and for the
chi-square goodness of fit test.The first form specifies the intervals exactly and
the second form provides a shortcut approach of specifying “nclass” equal
intervals (Note: the syntax from previous versions of GOF=c0,c1...; will also
work). You can enter up to 3 of these commands to specify different sets of
intervals. If you do not specify this command and the data are analyzed as
ungrouped, 3 sets of intervals are constructed with equally spaced cutpoints and
the number of intervals being the n0.5 and 2/3 n0.5 and 3/2 n0.5.
If the data are entered grouped or entered ungrouped and analyzed as grouped
(DISTANCE/INTERVALS= used in ESTIMATE) then only the third form can
be used to specify that the GOF statistics should be generated. It is not possible
to specify goodness of fit intervals other than those used to analyze the data
Examples:
Data are ungrouped and 2 different sets of intervals are specified.
GOF /NCLASS=5;
GOF=0,5,10,20,30,40,50;
Data are grouped but GOF statistics are desired.
GOF;

MONOTONE Command
Syntax:
MONOTONE = WEAK or STRICT or NONE
Description:
The estimators are constrained by default to be strictly monotonically non-
increasing (i.e., MONOTONE=STRICT; the detection curve is either flat or
decreasing as distance increases from 0 to w). In some instances, depending on
the tail of the distribution this can cause a poor fit at the origin(x=0). Two
options exist: 1) truncate the observations in the tail, or 2) use the command
MONOTONE=WEAK; or MONOTONE=NONE;. MONTONE=WEAK; will

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 305
 only enforce a weak monotonicity constraint (i.e., f(0) >= f(x) for all distances
x). This will allow the curve to go up and down as it fits the data but it will not
let the curve dip down at the origin. In some instances this will allow the
estimator to achieve a better fit at the origin which is the point of interest.
Setting MONTONE=NONE; will allow the curve to take any possible form
except that it must remain non-negative.
Monotonicity is achieved by constraining the function at a fixed set of points. In
some circumstances it is possible that the curve can be non-monotone between
the fixed points. Typically, this results from trying to over-fit the data with too
many adjustments with a long-tailed distribution. Truncate the data rather than
attempting to over-fit.
Note that MONOTONE = NONE is the only allowed option when there are
covariates in the detection function.
Default:
MONOTONE = STRICT; (no covariates)
MONOTONE = NONE; (covariates)

MULTIPLIER Command
Syntax:
MULTIPLIER = value1 /LABEL='name' /SE=value2 /DF=value3;
Description:
This command specifies a multiplier for the density and/or abundance estimate.
Some uses of multipliers are discussed in the Users Guide section Multipliers in
CDS Analysis.
Density/abundance is multiplied by the value of value1. The analytic variance
estimate takes into account the additional variance due to the multiplier, as
specified by value2, by adding an additional term ot the delta method formula
(equation 3.70 in Buckland et al. 2001). The degrees of freedom for confidence
limits are affected if a non-zero value is specified for value3, because an extra
term is added to the Satterthwaite formula (equation 3.75 in Buckland et al.
2001).
/LABEL='name' – “name” is the name given to the multiplier in the output file
/SE=value2 – value2 is the standard error of the multiplier – use 0 if the
multiplier value is known with certainty
/DF=value3 – value3 is the degrees of freedom associated with the multiplier –
use 0 for infinite degrees of freedom.
Note that if you want a multiplier to divide the density estimate, simply specify
value1 as the inverse of the multiplier value. Value2 (the SE) is then the
multiplier SE divided by the square of the multiplier value.
There is no maximum to the number of multiplier commands within the Estimate
section.

PICK Command
Syntax:
PICK = AIC or AICC or BIC
Description:
If more than one ESTIMATOR command is given a choice must be made as to
which model will be used for the final estimate. The command PICK=AIC;
instructs the program to choose the model that minimizes Akaike's Information
Criterion, AICC minimizes the small sample corrected version of AIC, and BIC
minimizes the Bayesian Information Criterion. If no command is given

306 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 PICK=AIC; is assumed. (Note: the option PICK=NONE, which told the program
not to choose a model and to present the results of each, is no longer supported.)
If the BOOTSTRAP; command is given, the bootstrap is performed and the
estimator is chosen for each analysis. Thus, even though a single estimator is
chosen for the point estimate, different estimators can be chosen for each
bootstrap and the standard errors and interval estimates incorporate the
uncertainty of the model selection process.
Default:
PICK=AIC;

PRINT Command (Estimate section)

Syntax:
PRINT /YES=option list /NO=option list;
Definition:
This command can further expand or limit the output from the estimate
procedure beyond what is defined by the PRINT command in the Options
section. The PRINT command in the Options section allows hierarchical control
of the output and defines the default values for this print command and thus
retains its functionality. However, this command can be used to define whether
each component is printed to the output file by specifying it either in the /YES or
/NO list. The following are the values of the option list:
Options list
All Used in place of listing all options
Estimate Density estiamte table
Explain Explanation of estimation options
Fxest Function parameter estimates
Fxfit Function fitting/model selection
Fxiterations Iterations of MLE
Fxplot Function histogram/plots
Fxtest Goodness of fit tests
Qqplot QQ plots and associated statistics
Sbarest Estimates of E(S)
Sbarplot Size-bias regression plot

Below are listed the default values of the print options as defined by the value set
by PRINT=in OPTIONS. (Y=Yes and N=No)
OPTIONS PRINT=
command ESTI FXE FXP FXT SBA SBA FXFI FXIT QQP
value MAT ST LOT EST RES RPL T LOT
E T OT
SUMMARY N N N N N N N N N
RESULTS Y Y Y Y Y Y N N Y
SELECT Y Y Y Y Y Y Y N Y
ALL Y Y Y Y Y Y Y Y Y

By default, /YES=EXPLAIN is set to provide a printed explanation of the

estimation options and models chosen (Note: in Versions prior to 1.20, the
default was NO and it was set to YES with the more limited PRINT OPTIONS
command).

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 307
 SIZE Command
Syntax:
SAMPLE
SIZE by STRATUM
ALL

Details:
This command explicitly specifies that expected cluster size should be estimated
and the resolution at which the estimate(s) should be made (by SAMPLE, by
STRATUM, or ALL data). This command is only necessary if density is not
being estimated or to specify a level of resolution different from density. The
level of resolution for estimating cluster size must be less than or equal to the
level for estimating detection probability, if a size bias regression estimate is
computed.
Example:
A user wishes to examine detection probability and expected cluster size but not
density at this point:
ESTIMATE;
ESTIMATOR/KEY=UNIF;
DETECTION ALL;
SIZE ALL;
END;

VARN Command
Syntax:
VARN = POISSON or b or EMPIRICAL
Description:
This command specifies the type of variance estimation technique for encounter
rate. The value POISSON specifies that the distribution of n (number of
observations) is Poisson . EMPIRICAL specifies that the variance should be
calculated empirically from the replicate SAMPLEs (section 3.6.2 of Buckland
et al. 2001). If only one SAMPLE is defined in the data, the POISSON
assumption is used unless a value b is specified. If a value b is specified it is
used as a multiplier such that var(n)= bn (e.g., Buckland et al. 2001, section
8.4.1). The Poisson assumption is equivalent to specifying b=1. The default for
VARN is EMPIRICAL unless there is only one SAMPLE, in which case, the
default is POISSON.
Default:
VARN=EMPIRICAL

MCDS Engine Required Data Format

The distance sampling data should be stored in a data file. The location of this
file is specified with the INFILE command in the Data section. The supported
7format for the data file is a flat file – i.e., a file containing columns that
correspond with input fields and one row for each observation (and also transects
without observations). Historical versions of this analysis engine used a
hierarchical data format, but this is no longer supported. Columns should be
separated by tab characters, and the FIELDS command should be used to tell the
MRDS engine which column is which. Other commands may be required – see
the Data section for details.

308 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 An example data file is given below:
Stratum A 100 Line 1A 10 14 F
Stratum A 100 Line 1A 10 8 M
Stratum A 100 Line 1A 10 22 M
Stratum A 100 Line 2A 10.3 7 F
Stratum A 100 Line 2A 10.3 37 F
Stratum A 100 Line 2A 10.3 13 F
Stratum B 123 Line 1B 5.7
Stratum B 123 Line 2B 8.4 27 M
Stratum B 123 Line 2B 8.4 76 F
Stratum B 123 Line 2B 8.4 44 M
Stratum B 123 Line 2B 8.4 7 M
Example data file

The corresponding FIELDS command for these data is:

FIELDS=STR_LABEL, STR_AREA, SMP_LABEL, SMP_EFFORT, DISTANCE, Sex;

This tells Distance that the first column is the stratum label, the second is the
stratum area, the third is the sample label, the forth the sample effort, the fifth is
the distances and the last is a column called “Sex”. This last column will be
used as a factor covariate, so the DATA section also needs the command
FACTOR=Sex /LEVELS=2 /LABELS=’F’,’M’;

Notice that for Line 1B there is nothing in the distance column – this is because
no animals were seen on that line.

An easy way to generate an example data file, to get a feel for the
required format, is to set up an analysis using the Distance graphical interface,
and then run the analysis in Debug mode. In this mode, the Distance interface
generates a command file and data file, and stores them in the Windows
temporary folder, but does not run the analysis. For more about Debug mode,
see the Program Reference page on the Analysis Preferences Tab. This strategy
will also enable you to see what commands are required in the Data section for a
particular data file.

Output From the MCDS Engine

Output from the MCDS engine takes two forms:
• MCDS engine command line output. A number returned to the
command line when the run finishes, giving the status of the run.
Occasionally, other output, such as FORTRAN error or warning
messages, may appear there as well.
• Up to 6 results files, as specified in the header section of the
command file. These files are:
• MCDS engine output file
• MCDS engine log file
• MCDS engine stats file
• MCDS engine bootstrap file
• MCDS engine plot file
• MCDS engine bootstrap progress file
The format of each of these outputs is given in the following sections.

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 309
 MCDS Engine Command Line Output
Run status
When the MCDS engine is run from the command line, it returns a number when
the run finishes. This number gives the status of the run, as follows:
• 1 means the analysis ran OK
• 2 means it ran with warnings (see log file for details)
• 3 means it ran with errors (see log file for details)
• 4 means it ran with file errors (e.g., could not find the specified
command file)
• some other number. A major error occurred (see below).
These numbers are also returned if the engine is run from another program as an
independent process, and so can be used by the program to diagnose whether the
run was OK.
A number returned to the command line when the run finishes, giving the status
of the run. Occasionally, other output, such as FORTRAN error or warning
messages, may appear there as well.
FORTRAN Debugging output
Occasionally, some other text is written to the standard output, which is usually
the command line, by the FORTRAN run time library used to run mcds.exe. A
mild example is that the number of underflow errors are written out, e.g.,:
forrtl: error (74): floating underflow
forrtl: error (74): floating underflow
forrtl: info (300): 30 floating underflow traps
Floating point underflow occurs when a number is calculated that is smaller than
the smallest number the computer can store and the number is instead stored as
zero. This rarely causes problems in practice – although it is worth double-
checking your results.
A more extreme example is if there is a program crash, debugging information is
written out. If this happens, a copy of the Distance project or command file
should be sent to the program authors. In the example below, the program
crashed with a “floating invalid” error on line 398 of the routine SBREG, which
was called from line 205 of CMOD, etc.
forrtl: error (65): floating invalid
Image PC Routine Line Source
MCDS.exe 0040DA68 SBREG 398 Cmod.for
MCDS.exe 0040C294 CMOD 205 Cmod.for
MCDS.exe 0040B2F3 ESMOD 29 Cmod.for
MCDS.exe 0044763C ESTPARM 487 Estmte.for
MCDS.exe 00446471 ESTMTE 291 Estmte.for
MCDS.exe 00445201 ESTPROC 88 Estmte.for
MCDS.exe 004136EA CNTRL 113 Control.for
MCDS.exe 004381CF DISTANCE 263 Distance.for
MCDS.exe 004F2459 Unknown Unknown Unknown
MCDS.exe 004C8BE3 Unknown Unknown Unknown
kernel32.dll 7C816FD7 Unknown Unknown Unknown

MCDS Engine Output File

This file is used in the Distance interface to fill the Results tab of the Analysis
Details Window. It is divided into a number of pages, and each page title is
delimited by tab characters:
[Tab]Page title[Tab]
This file is designed for human reading, not machine processing (except that
pages can easily be separated by searching for two tabs with text in between).
Most important statistics are output in a compact, easily parsed format in the

310 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 MCDS Engine Stats File, which should be the first port of call for extracting
results by machine.

MCDS Engine Log File

This file is used in the Distance interface to fill the Log tab of the Analysis
Details Window. It contains a copy of the input commands, together with output
from the analysis engine about any errors or warnings encountered while
processing the commands. This file is the first place to look if the analysis
engine returns a warning or error value (2 or 3) after a run.
A comprehensive list of the warning and error messages is given in the section
MCDS Engine Error and Warning Messages.

MCDS Engine Stats File

This file contains a compact output of summary statistics. The Distance
interface uses this file to extract data for the Analysis Browser table. A set of
records is output for each model defined in the Detection Function Models.
Each record is given a new line. The record structure is as follows:

Stratum stratum number (or 0 if the estimate is for a sample or all

data).
Sample sample number (or 0 if the estimate is for a stratum or all
data).
Estimator number of the estimator (in the order given in the Estimate
procedure)
Module number of the parameter module (see below)
Statistic number of the statistic within the parameter module (see
below)
Value estimate value
Cv coefficient of variation of estimate or 0.0
Lcl lower confidence limit or 0.0
Ucl upper confidence limit or 0.0
Df degrees of freedom for interval or 0

The modules and statistics within each module are listed in below in the order in
which they are summarized in the output. The FORTRAN format for each
record is:
FORMAT(2(1X,I5),2(1X,I1),1X,I3,5(1X,G14.7))

This is different from the format for previous versions of Distance

Each field is separated by a space, so the records can be read into a spreadsheet
or other program as space delimited or as fixed-width format. The record for a
module/statistic type is only output if it is relevant and it was computed in the
analysis.
The following table defines the module and statistic codes used:
Module Statistic/Parameter Estimate
1 – encounter rate 1 – number of observations (n)
2 – number of samples (k)
3 – effort (L or K or T)
4 – encounter rate (n/L or n/K or n/T)

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 311
 5 – left truncation distance
6 – right truncation distance (w)
2 – detection probability 1 – total number of parameters (m)
2 – AIC value
3 – chi-square test probability
4 – f(0) or h(0) 1
5 – probability of detection (Pw) 1
6 – effective strip width (ESW) or effective
detection radius (EDR) 1
7 – AICc
8 – BIC
9 – Log likelihood
10 – Kolmogorov-Smirnov test probability
11 – Cramér-von Mises (uniform weighting)
test probability
12 – Cramér-von Mises (cosine weighting)
test probability
13 – key function type 2
14 – adjustment series type 3
15 – number of key function parameters
(NKP)
16 – number of adjustment term parameters
(NAP)
17 – number of covariate parameters (NCP)
101 … (100+m) – estimated value of each
parameter 5
3 – cluster size 1 – average cluster size 1
2 – size-bias regression correlation (r)
3 – p-value for correlation significance (r-p)
4 – estimate of expected cluster size
corrected for size bias 1
4 – density/abudance 1 – density of clusters (or animal density if
non-clustered) 1
2 – density of animals 1
3 – number of animals, if survey area is
specified 1
4 – bootstrap density of clusters 1, 4
5 – bootstrap density of animals 1, 4
6 – bootstrap number of animals 1, 4
1
Values for CV, LCL, UCL and DF are included for these statistics.
2
Key function types are: 1 = uniform, 2 = half-normal, 3 = negative exponential, 4 =
hazard rate
3
Adjustment series types are: 1 = simple polynomial, 2 = Hermite polynomial, 3 = cosine
4
Bootstrap CV calculated as bootstrap SE / bootstrap point estimate; df field here is the
number of bootstraps
5
Statistic 101 corresponds with the parameter identified as A(1) in the results, 102 with
A(2), etc.

MCDS Engine Plot File

This file contains the data used to construct the high resolution qq and histogram
plots in the Distance interface. The output format is:

312 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 Title line - plot 1 (up to 80 char)
Sub-title line (up to 60 char)
x-label (up to 30 char)
y-label (up to 30 char)
# of data rows (r)
x1,y11,y21,y31,y41
x2,y21,y22,y31,y41
.
.
.
xr,yr1,yr2,y31,y41
Title line - plot 2 (up to 80 char)
Sub-title line (up to 60 char)
x-label (up to 30 char)
y-label (up to 30 char)
..
etc

The number of columns of data (yr1, yr2, etc) depends on the plot:
• For qq-plots there are 4 columns: 1 and 2 are the x and y
coordinates of the data points (i.e. the edf and the fitted cdf); 3 and
4 are the x and y coordinates of the line that runs from (0,0) to
(1,1). If you are using this file to recreate the plot in another
package, you could easily ignore columns 3 and 4 and replace them
with a (0,0) (1,1) line.
• For plots containing the data histograms and accompanying pdf or
detection function plots there are 4 columns: 1 and 2 give the x and
y coordinates which, when joined up, give the fitted detection
function or pdf and 3 and 4 give a set of x and y coordinates which,
when joined up, produce the data histograms
• For the MCDS example detection function plots, which contain 3
detection functions, there are 6 columns: 1 and 2 give the x and y
coordinates for the first detection function, 3 and 4 give this for the
second detection function and 5 and 6 give the coordinates of the
third detection function.
You can see an example of these kind of data being used to produce a plot in a
tip under Exporting CDS Results from Analysis Details Results in Chapter 8
(although the data there come from copying the plot to the clipboard rather than
directly from the plot file).

MCDS Engine Bootstrap File

This file has the same format as the MCDS Engine Stats File, but contains one
set of stats records for each bootstrap. These records are appended one after
another as the bootstrap progresses.

MCDS Engine Bootstrap Progress File

This file is intended to be read periodically during a bootstrap analysis. It is

only created if a bootstrap is requested, and if the file name is not “None”. If the
file exists already, it is overwritten.
The file is empty until the bootstrapping starts. Then, it contains a 3-digit
integer (between 0-100) which indicates the percentage of the way through the
bootstrap we are.
An example of the use of this file is the Distance interface, which reads it every
second during a bootstrap analysis and uses the contents to add a “Progress x%”
message after the “Running analysis x” on the status line of the main toolbar.

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 313
MCDS Engine Limitations
The following limitations apply to the MCDS engine:
Limitation Maximum number
Observations 100 000
Samples (transects) 50 000
Strata 1 000
Cutpoints in GOF and interval data 25
Detection function models 5
Adjustment terms per model 5
Covariates 10
Levels, for factor covariates 200

In addition, constraints cannot be set on the detection function if there are

covariates in addition to Distance (see next section).

MCDS Engine Fitting Algorithms

CDS Analyses
For analyses with distance as the only covariate in the detection function (CDS
analyses), the detection function is fit using the constrained maximum likelihood
method outlined in section 3.3.5 of Buckland et al. (2001). The fitting algorithm
used is subroutine DNCONG, from the IMSL FORTRAN 90 Mathematics and
Statistics Library version 3.0, written by Visual Numerics Ltd. This subroutine
uses a successive quadratic programming algorithm to minimize the negative
log-likelihood subject to monotonicity constraints on the detection function.
MCDS Analyses
For fitting the detection function when there are one or more covariate, the
MCDS engine uses an iterative maximum likelihood routine, based on the
Newton-Raphson algorithm. It alternates between fitting the key and covariate
parameters conditional on the current estimates for the adjustment parameters,
and fitting the adjustment parameters conditional on the current estimates for the
key and covariate parameters. When close to convergence, it switches to
maximizing all parameters at once.
This algorithm cannot cope with constraints, and therefore no monotonicity
constraints can be enforced with the MCDS engine.
In some cases, the user specifies the MCDS engine but there are no covariates.
For example, imagine that the detection function is to be fit by stratum, and that
there is one factor covariate with two levels. In one stratum, only one level of
the factor occurs in the data. In this stratum, no covariate parameters. When this
occurs, the CDS algorithm is used to fit the data, but still with no monotonicity
constraints.

This means that if you select the MCDS engine in the Distance
interface, but do not specify any covariates, you will get identical output to
selecting the CDS engine with no monotonicity constraints.

314 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
MCDS Engine Error and Warning Messages
This section gives a comprehensive list of the warning and error messages that
can be generated by the MCDS engine, and can occur in the output or log file.
Explanations for some of the messages are given – please contact us if you need
an explanation for a message we don’t explain here, or get a message that is not
documented (it’s possible we missed some!).
The standard format for an error or warning message is
** [Bootstrap] level: message **
where level can be either “Warning”, “Error” or “Internal Error”, message is the
text of the message, and the word Bootstrap appears if the problem occurred
while running a bootstrap replicate dataset.

Some error and warning messages have been discussed on the

distance-sampling email list, so it’s worth searching the list archives to for more
information, including possible remedies.

MCDS Engine Warning Messages

These are used to indicate a mild problem with the analysis (such as small
sample size, possible lack of convergence, etc.), or to alert the user to an unusual
aspect of the analysis that they should be aware of. Warnings do not result in an
analysis terminating, but should be taken seriously as the quality of the results
may be compromised.
Many of these messages will not be seen by users of the Distance interface, as
they relate to mutually incompatible combinations of commands that are already
screened for by the interface.

ID Warning Message Explanation

1 Angle not valid for TYPE=POINT Mutually incompatible
commands
2 SIZE not valid for OBJECT=SINGLE Mutually incompatible
commands
3 The previous stratum had no samples so it will be
ignored.
4 An estimator was not chosen. The default
estimator will be used.
5 Cannot estimate encounter rate variance
empirically when estimating encounter rate by
sample. Sample level encounter rate variances
will be estimated assuming distribution of
observations is Poisson.
6 Degrees of freedom less than 1.0 for estimating
confidence limits on density of individuals.
Confidence limits not calculated.
7 Degrees of freedom less than 1.0 for estimating
confidence limits on F0. Confidence limits not
calculated.
8 Degrees of freedom less than 1.0 for estimating
confidence limits on density of objects.
Confidence limits not calculated.
9 Warning: Estimated correlation between

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 315
 parameters > +/- 1.0.
10 Estimation routine failed to converge due to
negative area estimates on iteration [iteration
number]. Using results from previous iteration.
11 Estimation routine failed to converge due to
singular information matrix on iteration [iteration
number]. Using results from previous iteration.
12 Estimation routine failed to converge.
13 FIELD will be ignored in the data
14 INTERVALS switch ignored because NCLASS Mutually incompatible
specified. commands
15 Missing item in list two adjacent commas [value
here]. Skipping to next item.
16 Negative variance estimate for f0. Invalid
variance. Results may not be reliable.
17 One or more cluster sizes are 0. These See Zero Cluster Sizes in
observations will be used in cluster size CDS Analysis in the Users
estimation. If you intended to code these values Guide.
as missing, please enter them as -1.0
18 One or more cluster sizes is coded as -1. Distance See Missing Cluster Size
assumes -1 to mean a cluster of undetermined Data in CDS Analysis in
size. These observations are used for estimating the Users Guide.
detection probability and encounter rate, but not
cluster size.
19 Parameter [parameter number] is at a lower
bound.
20 Parameter [parameter number] is at an upper
bound
21 Parameters are being constrained to maintain a
positive pdf
22 Parameters are being constrained to obtain
monotonicity.
23 Previously read samples were not assigned a
stratum, so all strata will be ignored.
24 SIZEC is an invalid command when Mutually incompatible
OBJECT=SINGLE, and so was ignored. commands
25 Some of the estimates of f0 are negative. Results
are not reliable.
26 Some parameters are very highly correlated.
27 The /BIAS switch is not allowed when cluster Mutually incompatible
size is a covariate, and so it has been ignored. commands
28 The cluster size covariate is a factor and so it is
assumed that factor levels correspond to cluster
sizes.
29 The estimated analytic variance for f0 gives a CV
greater than 10000%, and hence results may not
be reliable. To avoid numerical problems, the CV
for f0 was assigned a value of 9999.99%.
30 The estimated area is negative and only a single
iteration of the estimation routine has been
carried out.
31 The number of lower bounds for the parameters
does not match the number of parameters in the
model. The default bound for parameter A
[parameter number] is being used instead.
32 The number of starting values for the parameters

316 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 does not match the number of parameters in the
model. The default starting value for parameter A
[parameter number] is being used instead.
33 The number of upper bounds for the parameters
does not match the number of parameters in the
model. The default bound for parameter A
[parameter number] is being used instead.
34 The starting value for parameter A [parameter
number] must be > 0. Using the default starting
value.
35 There are less than 10 data points per estimated
parameter. Results may not be reliable.
36 There is only one level for factor covariate
[covariate]. A minimum of two levels is required
for estimation; hence this covariate will be
omitted.
37 There is only one level for factor covariate
[covariate]. A minimum of two levels is required
for estimation; hence this covariate will be
omitted from estimates for stratum [stratum]
sample [sample].
38 There is only one level for factor covariate
[covaraite]. A minimum of two levels is required
for estimation; hence this covariate will be
omitted from estimates for stratum [stratum]
39 When cluster size is a covariate, variance of the See Cluster Size as a
cluster size, density of individuals, and Covariate in the Users
abundance estimates can only be obtained via the Guide.
bootstrap. You have not specified the bootstrap
variance option, so these variance estimates will
not be produced.
40 When cluster size is a covariate encounter rates See Cluster Size as a
are not computed. Covariate in the Users
Guide.
41 When cluster size is a covariate no stratification
is allowed.
42 When covariates are being used, a number of
intervals > 20 for GOF tests may cause the
program to terminate with an error.
43 The number of intervals - 1 = [number] which is
less than the number of key parameters [number].
No fit possible.
44 The number of intervals - 1 equals the number of
key parameters [number] so no adjustments can
be made.
45 The number of observations - 1 equals the
number of key parameters [number], so no
adjustments can be made.
46 There are no cluster size observations selected.
Cannot estimate expected cluster size.
47 Negative variance for expected cluster size. No
size bias adjustment. Average cluster size used
instead.
48 Convergence failure.
49 Estimated cluster size greater than exp14.
Average cluster size used instead.
50 SEED cannot be a negative number. It has been
set to 0

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 317
 51 Negative variance estimate for parameter. Invalid
variance.
52 Number of cluster size measurements =
[number]. This is not sufficient for size-bias
regression. Average cluster size used instead.
53 Number of cluster size measurements = [number]
This is not sufficient to estimate a mean and
variance
54 Number of observations is small. Do not expect
reasonable results.
55 Size bias adjustment has increased expected
cluster size.
56 The number of adjustment parameters allowed
has been reduced to [number] because of limited
number of observations.
57 Too few observations to calculate AICc. AICc set
to 0.
58 Too few observations. An estimate of f0 cannot
be computed. f0 set to 1/width.
59 Two models have the same [model selection
statistic]. Choosing one of them at random.
60 Zero observations. An estimate of f0 cannot be
computed. f0 set to 0.
61 Angle not valid for DIST=PERP Mutually incompatible
commands
62 Area = 0 for stratum
63 BOOTSTRAPS may not exceed 5000. Set to
5000
64 BOOTSTRAPS should be at least 100
65 Cannot specify CONVERT without MEASURE Mutually incompatible
and UNITS',/, CONVERT value will not be used. commands
66 CONVERT value will overide previous value','
specified.
67 INTERVALS ignored because NCLASS was set. Mutually incompatible
commands
68 Warning: Invalid or missing covariate.
69 NCLASS and INTERVALS both set. Mutually incompatible
INTERVALS',' ignored. commands
70 No observations in stratum [stratum] so
estimating f0 using global average f0|z. Results
are therefore not reliable.
71 SEED should be an odd number greater than
2000000
72 Warning: Seed will be set with value from clock Refers to the random
number seed. This
warning does not occur
when calling MCDS from
the interface as SEED=0 is
specified, which means set
from clock. This warning
only occurs if SEED is not
specified, and is intended
to remind the user the seed
has come from the clock.
73 SMEAR switch only valid for ungrouped Mutually incompatible
dist/angle measurements. commands

318 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 74 SMP_EFFORT not in data. Assumed to be 1 for
each point
75 Specified width [width] does not match an When truncating data in
interval value. It has been set to [new width] intervals.
76 There is only one level for factor covariate
[covariate]. A minimum of two levels is required
for estimation; hence this covariatewill be
omitted from estimates for sample [sample]
77 Warning: TITLE value not found.
78 Too many sets of GOF, only [number] allowed. Currently, [number]=3
79 User is overriding a conversion factor available in Obsolete – conversion
the program. factors between units now
always specified by
Distance interface, so this
warning is suppressed.
80 For goodness-of-fit interval set [number].
Number of goodness-of-fit intervals reduced.
81 For goodness-of-fit interval set [number]. Interval
end-point modified to match width.
82 For goodness-of-fit interval set [number].
Goodness-of-fit intervals specify testing subset of
the data.
83 For goodness-of-fit interval set [number].
Specified intervals are inconsistent with width.
84 For goodness-of-fit interval set [number]. Interval
begin-point modified to match left truncation.
85 For goodness-of-fit interval set [number].
Specified intervals are inconsistent with left
truncation value.
86 Exact distance values, rather than distance
intervals have been used in size bias regression
calculations.
87 No monotocity constraints are allowed when Mutually incompatible
covariates are present. commands

MCDS Engine Error Messages

These are used to indicate a severe problem with the analysis – enough to make
the results invalid. These are sometimes usually enough to terminate the entire
analysis, but more usually result in a part of the analysis being terminated (for
example a row of data is dropped or a detection function estimate is not used).
Many of these messages will not be seen by users of the Distance interface, as
they relate to mutually incompatible combinations of commands that are already
screened for by the interface.
ID Error Message Explanation
1 Total of cluster frequencies = [number] Exceeds
maximum number of observations – [number]
Procedure terminated.
2 Number of adjustment parameters NAP is greater Too many adjustment
than MAXFIT. terms specified.
3 Number of parameter bounds exceeds [number] More parameter bounds
than parameters.
4 Ambiguous value for command [command]
5 Invalid value. Use either Yes, No, On, Off, True,

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 319
 False.
6 Invalid value for command [command]
7 [Command] is an invalid command for Obsolete.
HIERARCHY structure.
8 A maximum of [number] levels is allowed for See MCDS Engine
each factor covariate. Limitations.
9 A maximum of 10 covariates may be specified.
10 ANGLE needed in data
11 Area under fx or gx is zero.
12 At most one group contains observations.
13 Bootstrap will not be done because', observations
are not being re-sampled and density estimated by
sample
14 Cannot scale distances by sigma when using the
uniform key function: change /ADJSTD option to
W.
15 Cannot use multiple DETECTION commands for
CDS analysis or when cluster size is a covariate.
16 Cluster size frequency < 0 [value of cluster freq]
17 CONFIDENCE must be between 1 and 99
18 Covariate specified in the ESTIMATE command
but not in the DATA command: [covariate],
19 CUERATE must be a positive number.
20 Dataset has been cleared.No data has been stored.
21 Density for each sample is unnecessary when
detection and expected cluster size are estimated
at higher levels
22 Detection probability must be estimated for size
bias calculations
23 Distance frequency < 0 [value of distance freq]
24 DISTANCE needed in data
25 Due to errors, this ESTIMATOR command will
be ignored.
26 Due to errors, this GOF command will be
ignored.
27 Error reading in values.
28 Exceeded array size [size] - for entering data.
29 Exceeded maximum number of cluster size Same as number of
observations = [number] distance observations,
below.
30 Exceeded maximum number of distance', See MCDS Engine
observations = [number] Limitations.
31 FIELDS have not been set. Data cannot be read.
32 Filename on INFILE command was not found.
33 FLAT data structure invalid for grouped data
34 Incompatible resolution levels for estimation of
[one component of estimation] and [another]
35 Incorrect number of data values.
36 Interval values for cluster sizes are out of order. Obsolete
37 Interval values for distances are out of order. Obsolete
38 Intervals for clusters cannot be specified because Obsolete

320 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 the data were entered in intervals.
39 Intervals for distance cannot be specified because Obsolete
the', data were entered in intervals.
40 Invalid cluster size < 0 [value]
41 Invalid command encountered – [command]
42 Invalid distance <0 [value]
43 Invalid filename or file could not be found
[filename]
44 Invalid initial values for parameters
45 Invalid option for CUERATE command.
46 Invalid or missing angle.
47 Invalid or missing cluster size.
48 Invalid or missing distance.
49 Invalid or missing value for [variable]
50 Invalid or missing value for sample effort =
[sample name] Sample will be ignored.
51 Invalid radial distance <0 [value]
52 Invalid smearing angle = [value] It must be > 0 &
<90
53 Invalid smearing distance. It must be > 0.
54 Invalid value for adjustment ORDER = [value]
55 Invalid value for NCLASS. It must be between 2 See MCDS Engine
and [max number of classes] Limitations.
56 Invalid value for sighting angle <0 OR >360 =
[value]
57 ITERATIONS must be > or = to 25.
58 Left truncation value cannot be negative.
59 LENGTH command is invalid for point transect Mutually incompatible
data. commands
60 Maximum number of samples [number] See MCDS Engine
exceeded. Limitations.
61 Maximum number of strata [number] exceeded. See MCDS Engine
Limitations.
62 Mismatched number of observations for multiple
measurements.
63 Missing sample label. Further data will be
ignored.
64 More cluster size frequencies were given than
intervals
65 More distance frequencies were given than
intervals.
66 More than 10 multipliers were specified. Excess
will be ignored.
67 Multiplier value must be > 0
68 NCLASS and WIDTH setting needed for
SMEAR switch.
69 NCLASS must be > 1 and <= [max value] See MCDS Engine
Limitations.
70 NCLASS set without a WIDTH value, both must Mutually incompatible
be set. commands
71 Negative variance estimate for f0. Invalid

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 321
 variance.
72 No data available to be analyzed.
73 Not a valid option with grouped data – [option] Mutually incompatible
commands
74 Number of adjustment parameters NAP is greater
than maximum possible for this model = [max.
number possible for this model]
75 Number of starting values exceeds [number of
parameters]
76 One or more estimated CDF is less than zero.
Value has been set to zero.
77 Only a single covariate may be specified as the
cluster size covariate.
78 Re-sampling unit Strata, Sample, Obs not set.
Bootstrap will not be attempted.
79 SE for multiplier must be non-negative number.
80 SF must be between 0.0 and 1.0
81 SMP_LABEL needed in data
82 Specified variance for N must be >= 0.Value
entered was [value]
83 Standard Error for CUERATE must be a positive
number.
84 Strata can not be re-sampled if Density by Strata
85 The [command] command is only valid when
there are covariates.
86 The FIELDS command must be specified before
the [command] command.
87 The number of adjustments specified by ORDER Mutually incompatible
[value] does not match the number specified by commands
NAP [value]
88 The number of adjustments specified by ORDER
exceeds the maximum of [maximum]
89 The number of covariates which are factors
cannot exceed the total number of covariates to
be included.
90 The significance level for the test must be in the
interval 0-1.
91 The specified factor covariate must match one of
the covariates given in the FIELDS', command.
92 The total number of cluster sizes does not match
the total number of distances.
93 The truncation proportion must be >=0 and <1.
94 The width must be a positive value.
95 There must be 2 parameters for smearing.
'Smear=angle,dist’
96 There were [value] covariates specified in the',
ESTIMATE command, but only [value] specified
in the DATA command. The latter must be >= the
former.
97 Unexpected end-of-file encountered.
98 Unknown units for distance measurement. Need
to set conversion factor or correct input.
99 Unknown units for distance/area conversion.
Need to set area conversion factor or correct

322 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 input.
100 Unknown units for distance/length/area
conversion. Need to set conversion factor or
correct input.
101 Valid values for LOOKAHEAD are 1 – [max
lookahead]
102 Values for lower parameter bounds must be
smaller than values for upper parameter bounds!
103 When cluster size is a covariate no stratification
is allowed.
104 When covariates are present, only the half-normal
or hazard-rate keys may be specified.
105 When using the FACTOR command, the switches
/LEVELS, and /LABELS must be specified.
106 WIDTH must be given with NCLASS value.
107 You have requested more estimators than the
maximum of [maximum estimators]
108 You must reset FIELDS or OPTIONS
109 PVALUE must be between 0.0 and 1.0
110 EPS must be BETWEEN 0.1 and 1.0E-8.
111 Exceeded maximum array storage
112 Maximum number of observations – [max obs]
exceeded. Procedure terminated.
113 Negative variance estimate for f0. Invalid
variance.
114 Negative variance estimate for parameter. Invalid
variance.
115 Negative variance estimate for part of f0|Z.
Invalid variance.
116 Bad bootstrap sample.

MCDS Engine Internal Error Messages

These occur when the MCDS engine encounters a situation that should not occur
under normal circumstances – for example a negative estimated effective strip
width. The analysis is aborted and no useful results are produced. If you
encounter such an error, please contact the program authors, sending a copy of
your command and data files, or Distance project.
ID Internal Error Message Explanation
21 invalid ordering sign.
22 CN = [value] INC = [value]
23 ier – [list of values]
24a G0 = 0, MU = [value]
24b MU = 0, G0 = [value]
24c WIDTH=0.0
25 fx not found [list of values]
26 Normalizing factor MU2=0.
27 AREA divisor for plot=0.
28 N equal 0 for plot.
29 Plot interval length for x = 0.

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 323
 30a Problems with incomplete gamma fct (GSER)
30b Problems with incomplete gamma fct (GSF)
30c Problems with incomplete gamma fct (GAMMP)
x or a <=0
31 Invalid degrees of freedom or chi-square value.
32 variance of cluster sizes is <= 0
33 Normalizing factor MU1=0.
34 Area under PDF=0 for at least one observation.
35 Confidence limit ZVAL is NaN. Confidence
limits set to 0. DF= [value]
36 F0NOBS= [value]
37 N= [value]
38 WIDTH has been erroneously set to zero
39a Standard error is 0 for one of the parameters.
39b Standard error is 0 for parameter [parameter]
39c Cannot scale distances by sigma for uniform key
function
40 DERIVS - PJ = 0 value
41 DERIVS - MU = 0
42 PROB - MU = 0
43 Invalid N in inverse routine = [value]
44 WIDTH is zero.
45 Attempt to divide by zero in inversion.
46 compute pdist error [values]
47 compute pdist [values]
50 Mismatched strata – [values]
51 Mismatched samples – [values]
53 Mismatched modules – [values]
54 Mismatched stats – [values]
61 F0 = 0.0
62 N= [value]
70 CLEVMULT=0
71 COVLEVELSI=0
91 Problems with settings for estimation routine.
102 Could not evaluate area under CDF.

Changes in MCDS Engine Since Distance 2.2

• Interactive mode not supported – use only in batch mode
• Flat file data entry added, and the old nested style is no longer
supported. The DATA statement should always have the option
/STRUCTURE = FLAT
• Changes in commands:
• New commands in DATA section: FIELDS, FACTOR and
SIZEC.

324 • Appendix - MCDS Engine Reference User's Guide Distance 6.0 Beta 5
 • New switch /COVARIATES in ESTIMATOR command.
Note that covariates must be the same in all ESTIMATORS in
the same run.
• ASSIGN commands not supported – assign output files
through the first 6 lines of the command file – see Header
Section.
• HELP commands no longer supported
• SQUEEZE command no longer supported
• New MULTIPLIER command
• DF added to the CUERATE command (which is, after all, just
another multiplier).
• Not recommended to use the SF command – use
MULTIPLIER instead
• Model fitting commands EPSILON and ITERATIONS
removed
• LIST command in Data section no longer supported
• In GOF /SAS and /SPLUS switches no longer supported
• VARF command no longer supported
• PICK=NONE no longer supported.
• When bootstrapping, point estimate is mean of the bootstrap
replicate point estimates.
• Changes in output format:
• Output file pages are no longer separated by page break
characters
• Page titles in the output file are surrounded with tab characters
to enable them to be easily recognized by a regular expression
parser
• Format of the stats and bootstrap stats file changed (each line
is longer) – see MCDS Engine Stats File.
• Extra output in the stats file – e.g., parameter estimates
• Bootstrap progress file added to give a way to allow the user to
find out how far the bootstrap has progressed.

User's Guide Distance 6.0 Beta 5 Appendix - MCDS Engine Reference • 325
Appendix – HT estimation of
density when probability of
coverage is unequal

Review of the Horvitz-Thompson-like estimator

This appendix, and the associated analysis engine is new in Distance

6.0 Beta 5. This text will eventually be moved to become a chapter in the Users
Manual.

This entire chapter is for advanced users only!

The theory behind this estimator of animal abundance is described in Strindberg
et al. (2004) and much of the terminology associated with this method can be
found in the Glossary of Terms at the end of this users guide.
The DHT engine currently has the following capabilities. We expect to extend
these in future versions.
• Line transect surveys (has not been tested for strip or point
transects)
• Coverage probability for each detected object is determined by the
survey design engine in Distance
• Estimation of density/abundance using a detection function fitted in
a previous analysis – allows different subsets of the data to be used
for encounter rate and detection function fitting (see Using a
Previously Fitted Detection Function to Estimate Density in
MRDS).
The basic form of the Horvitz-Thompson-like estimator employed by the DHT
engine is
n
s
Nˆ = ∑ i
ˆ i qˆi
i =1 p

where si is the size of the ith cluster, p̂i is the estimated probability of detection
of the ith cluster, and q̂i is the estimated coverage probability of the location
where the ith cluster was located. This formulation is premised on the basis of
estimating abundance of individuals; if interest was instead focused upon the

User's Guide Distance 6.0 Beta 5Appendix – HT estimation of density when probability of coverage is unequal • 327
 estimation of abundance of clusters in the population, a “1” would be substituted
for the numerator.

The job of the DHT engine is to bring together these three pieces of information
to create an estimate of abundance. In its simplest incarnation, the cluster size is
observed by the researcher, coverage probability is estimated from the survey
design (by the survey design engine in Distance prior to the conduct of the
survey), and finally the detection probability is estimated by fitting a detection
function to the observed distances.
As with the MRDS and the DSM engine, the DHT engine is implemented as a
library in the free statistical software R. When you run a DHT analysis from
Distance, Distance creates a sequence of R commands, calls the R software,
waits for the results and then reads them back in. Therefore, before you can use
the DHT engine, you must first ensure that you have R correctly installed and
configured. For more on this, see R Statistical Software in Chapter 7 of the
Users Guide.
To produce an abundance estimate based upon unequal coverage probability in
Distance, you then need to set up the project appropriately and include data in
the correct format – see Setting up a Project for DSM Analysis. You must next
create one or more model definitions using the MRDS analysis engine, and
associate these model definitions with analyses to derive detection probabilities
for each objected detected. For more about the basics of setting up analyses, see
Chapter 7 - Analysis in Distance. More details of the various models available
in the MRDS engine are given in Defining MRDS Models, and a detailed
description of the options available in the Model Definition Properties pages for
this engine is given in the Program Reference pages Model Definition Properties
- MRDS. After deriving detection probabilities, coverage probabilities
associated with each detected object is derived from the coverage grid
constructed by the survey design engine.

If you are new to Distance, we strongly recommend you familiarize

yourself with the CDS analysis engine, for example by working through Chapter
3 - Getting Started before trying to analyze DHT data.

The Williams multi-species survey sample project is an example of

how to set up a double-observer study and specify analyses – see Sample
Projects.

In this chapter we also provide some analysis guidelines, give a list of the output
the engine can produce and cover various miscellaneous topics.

If you are familiar with the R software, you can run the DHT engine
directly from within R, bypassing the Distance interface altogether. For more
information, see Running the DSM Analysis Engine from Outside Distance.

Setting up a Project for DHT Analysis

The premise of this documentation is that users have employed the survey design
in Distance (see Chapter 6 - Survey Design in Distance) to lay out a survey and
then that survey has been carried out in the field. In the process of designing the
survey, the user generated a coverage grid that is stored as a layer within the
Distance project. Having performed the survey, information is entered into the

328 • Appendix – HT estimation of density when probability of coverage is unequalUser's Guide Distance 6.0 Beta 5
 Observation layer of the Distance project in the usual fashion, with the additional
data requirement that the location of each detection (x and y coordinates) is also
recorded in the Observation layer.
The easiest way to set up a new project for a DHT analysis is using the Setup
Project Wizard.
• In Step 1, under I want to: select Analyze a survey that has been
completed.
• Be sure to tick the box indicating the Project will contain
geographic information at the bottom of the Step 1 screen
• In Step 3, under Observer configuration, select Double observer.
But see also Single Observer Configuration in the MRDS Engine.
• Follow through the rest of the wizard as usual.
Distance then creates the appropriate data fields for double observer data, and
you can then import your data using the Import Data Wizard.
Alternatively, you can create the appropriate fields by hand, and manually create
a new survey object with the appropriate observer configuration and data files.
For more about survey objects, see Working with Surveys During Analysis in
Chapter 7.

Output from DHT Analyses

The DHT engine produces the following output:
• a summary of results in the Analysis Browser. For general
information about the Analysis Browser, see the section
Introduction to the Analysis Browser in Chapter 7. There are many
results statistics available, and you can select which ones are shown
independently for each analysis set using the Column Manager (see
Column Manager Dialog in the Program Reference).
• a detailed listing of results in the Results tab of the Analysis
Details window. These are described in the following section,
MRDS Results Details Listing.
• a log of the analysis, highlighting any possible problems, in the
Log tab of the Analysis Details window. For information about
troubleshooting problems, see Chapter 12 - Troubleshooting.

DHT Results Details Listing

When an analysis has run, information is available in the Results tab of the
Analysis Details window. The results at present consist simply of a point
estimate of abundance in the surveyed region.

Exporting DHT Results

The methods of exporting results from the Analysis Browser and results pages of
the Analysis Details to other programs are the same as those for CDS analyses,
as documented in Exporting CDS Results in Chapter 8. For example you can
copy the results details text by choosing Analysis Results | Copy Results
to Clipboard.

User's Guide Distance 6.0 Beta 5Appendix – HT estimation of density when probability of coverage is unequal • 329
Miscellaneous DHT Analysis Topics

Clusters of Objects in DHT

If a cluster size field exists in the Observation layer, it becomes the numerator of
the estimator shown at the beginning of this appendix.
If the objects detected are clusters of individuals, you tell Distance this in the
same way as for CDS analyses, in the Setup Project Wizard (see Survey
Methods Wizard Page in the Program Reference).
Unlike CDS analysis, the Model Definition does not offer any regression
methods for dealing with size bias. If you suspect size bias is a potential
problem, the appropriate way to deal with it in an MRDS analysis is to include
cluster size (or some transformation of cluster size) as a covariate in the
detection function model(s).

The cluster size field is one of the fields with a fixed name in
detection function formulae in DHT (see Translating Distance Fields into DS
and MR Covariates) – in formulae you should use the name size regardless of
the actual field name.

Stratification and Post-stratification in DHT

At the moment, the MRDS engine only accommodates one level of stratification,
and this stratification is assumed to be geographic. There is no allowance for
weighting responses among strata in any fashion.

Running the DHT Analysis Engine from Outside

Distance
The DHT engine is not yet implemented as a library in the free statistical
software R. During this development phase, the DHT engine exists as a series of
R scripts; these are not readily disseminated to the user community. In future,
when DHT matures, we will distribute the code as an R library, and at that time
the comments below will be applicable.
When you run a DHT analysis from Distance, Distance creates files containing a
set of R commands and the appropriate input data, calls the R software, waits for
the results and then reads them back in. For more about R, see R Statistical
Software in Chapter 7 of the Users Guide.
Some users may wish to run the engine from outside the Distance interface –
either from within the R GUI interface or from another program. For example,
you may want to automate the running of analyses for simulations or
bootstrapping.
To see the format of the input and data files produced by Distance, try running a
DSM analysis in debug mode. To set debug mode on, choose Tools |
Preferences, Analysis tab, and tick Debug Mode. When you run analyses
in debug mode, the input and data files are created, but the analysis is not run.
The Log tab displays the location of the files – these are created in a directory
with a name “dst” followed by up to 4 numbers, located within the Windows
temporary directory. The file “in.r” contains the commands. Data are located in
the files “ddf.dat.r” (used in the detection function modelling), “region.dat.r”,
“sample.dat.r” and “obs.dat.r” (used in estimating density given a fitted detection
function). You can use these files as templates for creating your own command
and data files.

330 • Appendix – HT estimation of density when probability of coverage is unequalUser's Guide Distance 6.0 Beta 5
 To run the analysis from within the R GUI (Graphical User Interface), you can
cut and paste the commands from the file in.r. To run the analysis from another
program, you can call R in batch mode – this is achieved by calling the program
RCmd.exe, which is located within the /bin subdirectory of your R installation.
For more details, see the R for Windows FAQ (in R, type help.start() and
when a browser window opens, click on the FAQ for Windows port). For an
example of its use, see the Log tab of any DSM analysis you have run that was
not in debug mode – you should see a line of the form:
Starting engine with the following command:
C:\PROGRA~1\R\rw1091\bin\Rcmd.exe BATCH C:\temp\dst90474\in.r
C:\temp\dst90474\log.r

Users familiar with R may wish to work inside the R GUI. The DHT engine will
be contained in the library DHT. To load the library from within R GUI, type
library(dht)
All the functions in the dsm library will be documented. You will be able to
open a copy of the help files from within Distance by choosing Help | Online
Manuals | DHT Engine R Help (html).

The use of these libraries in operating systems other than Windows

is not supported (but may well work – let us know!)

Checking Which Version of the DHT Engine is

Being Used
The DHT engine is implemented as a library (called “DHT”) in the statistical
software R. From time to time, we may issue updated versions of the library, for
example in response to reported problems. Therefore, before downloading a
new version or reporting a problem you may want to check which version of the
library is currently in use. To do this, re-run an analysis that uses the DHT
Engine (such as one from the Williams sample project) and look in the Log tab
for the line
> library(dht)

After it, you should see a line which looks something like the following:
The previous topic describes how to update to a newer version of the DHT
Engine, if one is available.

When reporting results, you may want to cite the exact version (i.e.,
build number) of the library that used in the analysis. This is stored in the Log
tab, as outlined above.

User's Guide Distance 6.0 Beta 5Appendix – HT estimation of density when probability of coverage is unequal • 331
Bibliography

This section contains a list of references cited in the Users Guide. Much more
complete lists of works related to distance sampling are in Buckland et al. (2001,
2004).
• Borchers, D.L., S.T. Buckland and W. Zucchini. 2002. Estimating
Animal Abundance: Closed Populations. Springer Verlag.
• Borchers, D.L., S.T. Buckland, P.W. Goedhart, E.D. Clark and S.L
Hedley. 1998a. Horvitz-Thompson estimators for double-platform
line transect surveys. Biometrics 54: 1221-37.
• Borchers, D.L., W. Zucchini and R.M. Fewster. 1998b. Mark-
recapture models for line transect surveys. Biometrics 54: 1207-
1220.
• Buckland, S.T., D.R. Anderson, K.P. Burnham and J.L. Laake.
1993. Distance Sampling: Estimating Abundance of Biological
Populations. Chapman and Hall, London, reprinted 1999 by
RUWPA, University of St. Andrews, Scotland.
• Buckland, S.T., D.R. Anderson, K.P. Burnham, J.L. Laake, D.L.
Borchers and L. Thomas. 2001. Introduction to Distance Sampling.
Oxford University Press, London.
• Buckland, S.T., D.R. Anderson, K.P. Burnham, J.L. Laake, D.L.
Borchers and L. Thomas. (eds.) 2004. Advanced Distance
Sampling. Oxford University Press, London.
• Buckland, S.T., K.P. Burnham and N.H. Augustin. 1997. Model
selection: an integral part of inference. Biometrics 53: 603-618.
• Burnham, K. P., and D. R. Anderson. 2002. Model Selection and
Multimodel Inference: A Practical Information-Theoretic
Approach. 2nd edition Springer-Verlag, New York.
• Gibbons, J.D. 1971. Nonparametric Statistical Inference. McGraw-
Hill, New York.
• Hedley, S.L., S.T. Buckland and D.L. Borchers. 2004. Spatial
distance sampling models. Pages 48-70 in Buckland, S.T., D.R.
Anderson, K.P. Burnham, J.L. Laake, D.L. Borchers and L.
Thomas. (eds.) 2004. Advanced Distance Sampling. Oxford
University Press, London.
• Horvitz, D.G., and D.J. Thompson. 1952. A generalization of
sampling without replacement from a finite universe. Journal of
the American Statistical Association 47: 663-685.
• Innes, S., Heide-Jørgensen, M.P., Laake, J.L., Laidre, K.L.,
Cleator, H.J., Richard, P. and Stewart, R.E.A. (2002) Surveys of

User's Guide Distance 6.0 Beta 5 Bibliography • 333

 belugas and narwals in the Canadian High Arctic in 1996.
NAMMCO Scientific Publications 4: 169-190.
• Laake, J.L. and D.L. Borchers. 2004. Methods for incomplete
detection at distance zero. Pages 108-189 in Buckland, S.T., D.R.
Anderson, K.P. Burnham, J.L. Laake, D.L. Borchers and L.
Thomas. (eds.) 2004. Advanced Distance Sampling. Oxford
University Press, London.
• Marques, F.F.C. 2001. Estimating wildlife distribution and
abundance from line transect surveys conducted from platforms of
opportunity. PhD Dissertation, University of St Andrews, Scotland.
• Marques, F.F.C. and S.T. Buckland. 2003. Incorporating covariates
into standard line transect analysis. Biometrics 59: 924-935.
• Marques , F.F.C. and S.T. Buckland and D.L. Borchers. 2004.
Covariate models for the detection function. In Buckland, S.T.,
D.R. Anderson, K.P. Burnham, J.L. Laake, D.L. Borchers and L.
Thomas. (eds.) In prep. Advanced Distance Sampling. Oxford
University Press, London.
• Marques, T. A., Thomas, L., Fancy, S. G. , and S. T. Buckland.
2007. Improving estimates of bird density using multiple covariate
distance sampling. The Auk. 124:1229-1243.
• Otto, M.C. and K.H. Pollock. 1990. Size bias in line transect
sampling: a field test. Biometrics 46: 239-45.
• Strindberg, S. 2001. Optimized Automated Survey Design in
Wildlife Population Assessment. PhD Dissertation, University of St
Andrews, Scotland.
• Strindberg, S. and S.T. Buckland. 2004. Zigzag Survey Designs in
Line Transect Sampling. Journal of Agricultural, Biological &
Environmental Statistics 9: 443 - 461.
• Strindberg , S., S.T. Buckland. and L. Thomas. 2004. Survey
design and Geographic Information Systems. In Buckland, S.T.,
D.R. Anderson, K.P. Burnham, J.L. Laake, D.L. Borchers and L.
Thomas. (eds.) Advanced Distance Sampling. Oxford University
Press, London.
• Thomas, L., R. Williams and D. Sandilands. 2007. Designing line
transect surveys for complex survey regions. Journal of Cetacean
Research and Management, in press.
• Thompson, S.K. Sampling. (2nd edition) 2002. John Wiley and
Sons, New York.

334 • Bibliography User's Guide Distance 6.0 Beta 5

Glossary of Terms

adjusted angle zigzag

Survey design class that superimposes a continuous zigzag sampler whose angle
is continuously adjusted by survey region height.

AIC
Akaike's Information Criterion (AIC) is used in model selection and puts this
process into a function minimization framework. It is based on the Kullback-
Leibler "distance" between two distributions.
For more about AIC and model selection, see Burnham and Anderson (2002).

AICc
Version of AIC corrected for small sample size.
For more information, see Burnham and Anderson (2002).

analysis engine
A component within Distance that runs analyses and produces results. Different
analysis engines have different capabilities. Currently, there are three analysis
engines, one for conventional distance sampling (CDS), one for multiple
covariate distance sampling (MCDS), and one for mark-recapture distance
sampling (MRDS).

API
Abbreviation for Application Programming Interface – an interface that allows a
piece of software to be instructed to perform tasks from within a separate
software package.

CDS
See conventional distance sampling

checkbox
A box in the graphical user interface of Distance that you click on to select. A
selected checkbox displays a tick. Example:

User's Guide Distance 6.0 Beta 5 Glossary of Terms • 335

 conventional distance sampling
A subset of distance sampling methods, where probability of detection on the
point or line is assumed to be 1, and the only covariate in the detection function
is distance. For more details, see Buckland et al. (1993, 2001).

covariate
A variable that you can use to model the detection function. Perpendicular or
radial distance is always used as a covariate, but in the Multiple Covariate
Distance Sampling (MCDS) engine, you can include other covariates, such as
cluster size, sex, platform of observation, habitat, etc.

coverage probability
The coverage (or inclusion) probability of at an arbitrary location within the
survey region is the probability of it falling within the sampled portion of the
survey region.

data file
This file, always called DistData.mdb, contains information about how the data
is stored, and may contain some or all of the data itself. It is stored in the data
folder.

data folder
A folder (directory) containing survey data and related information about the
survey effort, study area boundaries, etc. Data folder names always have the
same beginning as the associated project file, but end in .dat - e.g., Ducknest.dat.
The data folder contains the data file, and one or more other files.

detection function
A function, denoted g(x), that described the probability of detecting an object
(individual or cluster) given that it is at distance x from the transect line or point.
In Distance, the detection function is modeled using the key + series adjustment
framework described in Buckland et al. (1993, 2001).

densification
A line that is straight in one coordinate system will not necessarily be straight
when viewed in a different system. For example, the equator is not a straight
line on many maps. So, when projecting from one coordinate system to another,
a straight line must be broken into a series of smaller straight lines so that it stays
in approximately the same place in the projected coordinate system. This
process of adding vertices to a line when projecting it is called densification.

design axis
User-defined line, superimposed on the survey region, that is used to orient the
samplers in zigzag designs.

dialog
A type of window in the graphical user interface of Distance. Dialog windows
are modal – that is you cannot access any other windows in Distance until you

336 • Glossary of Terms User's Guide Distance 6.0 Beta 5

 close the dialog. Only one dialog can be open at once. Examples include the
properties windows (e.g., Model Definition Properties), and the Open Project
dialog.

distance project
Where all of the information about one study area is stored. A project is made
up of a project file (which ends in .dst) and a data folder (ends in .dat).

distance sampling
A group of related survey methods for estimating the density and/or abundance
of wildlife populations.

double observer
A survey protocol where two (semi-) independent observer teams perform a
distance sampling survey, and duplicate detections are identified. Under this
protocol, more advanced analysis methods (Mark Recapture Distance Sampling)
can be used where it is possible to relax the assumption of standard methods that
all animals at zero distance are seen.
For more information, see Laake and Borchers (2004). For more about how to
set up a double observer dataset in Distance, see the Users Guide chapter on
Mark Recapture Distance Sampling.

equal angle zigzag

Survey design class that superimposes a continuous zigzag sampler of fixed
angle on the survey region.

equal spaced zigzag

Survey design class that superimposes a continuous zigzag sampler that passes
through equally spaced points on opposite sides of the survey region boundary.

external data files

Files that contain information about the survey data in a project, other than the
main data file DistData.mdb.

factor
Name given to a covariate that is divided into distinct classes. Examples include
sex (male / female), observer, etc.

f(0)
The value of the probability density function of observed distances, evaluated at
0 distance.

geographic coordinates
A measurement of a location on the earth’s surface expressed in degrees of
latitude and longitude.

User's Guide Distance 6.0 Beta 5 Glossary of Terms • 337

 geographic coordinate system
A reference system used to locate points expressed in degrees of latitude and
longitude on the earth’s surface. Defined by a spheroid of reference, a datum,
one or more standard parallels, a central meridian, and possible shifts in the x-
and y-directions to locate x,y positions of point, line, and area features

GIS
Geographic Information System - a piece of software that can work with
geographic data.

Horvitz-Thompson
Unbiased estimator of abundance, given by
n
1
Nˆ surv = ∑p
i =1 i

where N surv is the number of animals in the surveyed area (i.e., the strip or
circle actually surveyed), n is the number of animals seen, and pi is the
probability of observing the ith animal, given that it is in the surveyed region.
Given this estimate, assuming equal probability of coverage, an estimate of the
population abundance N is given by
A
Nˆ = N surv
a
where A is the area of the survey region, and a is the surveyed area.
If coverage probability is not equal, population abundance can be estimated by
n
1
Nˆ = ∑pq
i =1 i i

where qi is the probability that surveyed area covers the ith animal, given its
location. qi is dictated by the survey design.
For more information, see Horvitz and Thompson 1952, Borchers et al. 2002,
Thompson 2002, Buckland et al. In prep

Horvitz-Thompson-like
Term used to describe a Horvitz-Thompson estimator, where the probability of
observing the animal is estimated, rather than known:
n
1
Nˆ surv = ∑ pˆ
i =1 i

This estimator is biased, although the bias is usually not large if the pis are not
small.
See the entry for Horvitz-Thompson estimator for notation, and generalization to
the case where coverage probability is not equal.
For more information, see Horvitz and Thompson 1952, Borchers et al. 2002,
Thompson 2002, Buckland et al. In prep

inclusion probability
see Coverage Probabiltiy

338 • Glossary of Terms User's Guide Distance 6.0 Beta 5

 mark recapture distance sampling
A type of distance sampling used when probability of detection on the transect
line or point is less than 1. Multiple observers survey independently (or semi-
independently), and duplicate detections are recorded. For more details, Chapter
6 of the Advanced Distance Sampling book.

MCDS
see multiple covariate distance sampling

MRDS
See mark recapture distance sampling

multiple covariate distance sampling

Analysis of distance sampling data where covariates in addition to distance are
used to model the detection function. For more details, see Chapter 3 of the
Advanced Distance Sampling Book.

multiplier
A quantity you can use when you know your estimates are proportional to the
true abundance or density. If you know the constant of proportionality, you can
use a multiplier to get unbiased estimates. An example would be if you know
that g(0) is less than 1, but you have an independent estimate of g(0). You can
then use the multiplier (and, if you have it, the multiplier SE and DF) to correct
your estimates.

parallel random sampling

Survey design class that randomly distributes a number of parallel lines over the
survey region.

probability of detection
The probability of recording an object (individual or cluster) in the surveyed
area.

project file
A file containing the project settings, survey designs, analysis settings and
results. Project files always end in .dst – e.g., Ducknest.dst. Double-clicking a
project file opens it in Distance.

projected coordinates
A measurement of locations on the earth’s surface expressed in a two-
dimensional system that locates features based on their distance from an origin
(0,0) along two axes, a horizontal x-axis representing east–west and a vertical y-
axis representing north–south. A map projection transforms latitude and
longitude to x,y coordinates in a projected coordinate system.

User's Guide Distance 6.0 Beta 5 Glossary of Terms • 339

 projected coordinate system
A reference system used to measure horizontal and vertical distances on a map.
In Distance, a projected coordinate system is defined by a map projection (which
may include projection parameters such as shifts in the x or y direction)
combined with a geographic coordinate system.

radio button
A round button that you click on to select. Radio buttons usually occur in a
group, of which only one can be selected at once. An example is the constraints
group in the Conventional Distance Sampling Model Definition properties
window:

R folder
A folder (directory) containing the R object file (.RData) and image files
generated by the R statistical software package. It is located within a project’s
data folder, and is created automatically the first time an analysis is run that uses
R.

R software
According to the R web site (http://www.r-project.org), R is a language and
environment for statistical computing and graphics. In the context of Distance,
the mark-recapture distance sampling (MRDS) analysis engine is implemented
as an R library. A working copy of R is therefore required before this engine can
be run.

set
A collection of Analyses, Designs or Surveys that are displayed on the same
browser page. You usually group items together that share some properties – for
example, you could have two different Analyses Sets, one where you use
truncation and one where you do not.

shapefile
A shapefile is a standard format for storing geographic information, invented by
the GIS company ESRI. Each shapefile is actually 3 separate files: an .shp file, a
.shx file and a .dbf file. (In addition, there may be other files such as .prj files.)
Shapefiles are used to store geographic information in Distance.

simple random sampling

Survey design class that randomly distributes a fixed number of points over the
survey region.

single observer
The standard survey protocol where a single team of observers perform a
distance sampling survey. Under this protocol, it is necessary to assume that all
animals at zero distance are detected. cf. double observer methods.

340 • Glossary of Terms User's Guide Distance 6.0 Beta 5

 systematic grid sampling
Survey design class that randomly superimposes a systematic point grid of fixed
dimensions and rotation onto the survey region.

systematic random sampling

Survey design class that randomly superimposes a systematic set of parallel lines
onto the survey region.

systematic segmented grid sampling

Survey design class that randomly superimposes a systematic set of segmented
parallel lines onto the survey region. Segments are placed using a grid of points.

systematic segmented trackline sampling

Survey design class that randomly superimposes a systematic set of segmented
parallel lines onto the survey region. Segments are evenly spaced along a
systematic series of parallel tracklines.

toolbar
The collection of buttons and menus at the top of a window in Distance. An
example is the Analysis Components toolbar:

trackline
The transect line.

User's Guide Distance 6.0 Beta 5 Glossary of Terms • 341

 Analysis in Distance
Analyis Details window 210
Analysis Browser 197
Analysis components 72
Analysis engines 80

Index Introduction 71
Preferences 181
Analysis Results
Output from MCDS Analyses 122
Authors 6

B
Backing up projects 36
Bibliography 333
Binned Data
A CDS 100
Books
About Distance 5
Bibliography 333
About Distance dialog 254
Distance sampling reference books 3
About the Users Guide 1
Bootstrap
Acknowledgements 6
Overview in CDS 107
Adjustment terms
Setting options in CDS and MCDS analysis 247
Specifying in CDS and MCDS analysis 241
Bootstrap file
Algorithms
MCDS engine file format 313
MCDS engine 314
Bootstrap progress file
Analyis Results
MCDS engine file format 313
Output from CDS analyses 89
Bounds on parameters
Output from DSM analyses 154, 329
Setting in CDS and MCDS 244
Output from MRDS analyses 135
Specifying in DSM 159
Analysis
Specifying in MRDS 142
Stopping 163
Analysis Browser 71, 197
CDS Results 98 C
DSM Results 156
Calculating Probability of Detection 95
Exporting CDS Results 98
CDS See Conventional Distance Sampling
MRDS Results 136
Citation for Distance 3
Analysis Components 72, 253
Cleaning the Windows Temp folder 83
Analysis Components Window 77, 253
Cluster Size tab
Analysis Details 72
CDS and MCDS 246
CDS Results 90
Clustered populations
DSM Results 155, 329
In CDS 102
Exporting Results 99
Clusters
MCDS results 122
About 102
MRDS Results 135
CDS 102
Analysis Details window 210
Data Filter options 233
Analysis Details Windows 72
Missing values 102
Analysis Engines
Model Definition options 246
About 80
Zero cluster size 102
CDS Output 89
Column Manager dialog 258
DSM Output 154, 329
Compacting a project 39
MCDS output 122
Components that make up the Distance software 261
MRDS Output 135
Constraints
Running DSM from outside Distance 157, 330
Setting in CDS and MCDS 244
Running MCDS from the command line 279
Control
Running MRDS from outside Distance 140
Specifiying control parameters in MRDS Model
Analysis Guidelines
Definition 252
CDS 88
Conventional Distance Sampling 87
DSM 152
About 87
MCDS 120
Analysis guidelines 88
MRDS 134

User's Guide Distance 6.0 Beta 5 Index • 343

 Limitations of engine 314 Data Selection 232
MCDS engine reference 279 Data structure
Output from CDS analyses 89 About 41
Running from command line 279 Changing 45
Setting up a CDS Project 88 Database API 261
Conventions used in this documentation 2 Density Surface Modelling
Coordinate system, geographic 53 About model formulae 151
Copy Analysis guidelines 152
Data to other Programs 187 Checking the version number 158, 331
Corrupted project Fine-tuning an MRDS analysis 159
Fixing 164 Installing an updated version of the engine 158
Covariates Introduction 145, 327
Density Surface Modelling 145, 327 Output from DSM analyses 154, 329
Factor vs. non-factor in MCDS 117 Running from outside Distance 157, 330
Factor vs. non-factor in MRDS 134 Setting up a DSM Project 146, 329
Mark Recapture Distance Sampling 127 Design Browser 193
Multiple Covariates Distance Sampling 115 Design Classes 64
Specifying in MCDS analysis 243 Design Details window 201
Coverage Probability 65 Design Properties dialog 215
Cramér-von Mises Test 93 Detection function
Creating a new project 34 Constraints
Setting in CDS and MCDS 244
D Specifiying in CDS and MCDS analyses 240
Specifiying in MRDS analyses 251
Data Detection Function
For CDS analysis 88 About Formulae 95
For DSM analysis 146, 329 Multiple detection functions 111
For MCDS analysis 116 Detection Function tab
For MRDS analysis 129 CDS and MCDS 240
Data Entry 46 MRDS 251
Data Entry Wizard 170 Development Team 6
Data Explorer 183 Diagnostic output
Data Fields 43 Specifying in CDS and MCDS analysis 245
Data file reference 262 Specifying in MRDS analysis 252
Data Filter Distance
About analysis components 72 Data file reference 262
Interface 231 Distance
Data Filter Properties dialog 231 About 5
Data Format Authors 6
GIS data 56 Citation 3
Data Import Components that make up the software 261
Advanced 48 Data structure 41
Getting started example 1 13 Database API 261
Getting started example 2 19 History 7
GIS data 57 Inner workings 261
Import Data Wizard 172 New features 8
Importing one file per data layer 50 Sponsors 5
Introduction 46 Use Agreement 5
Linking to data from other databases 61 Web site 4
Non geographical data 46 Distance projects 33
Streamlining import of data from one flat file 49 Distance sampling
Troubleshooting 176 Reference books 3
Data in Distance 41 Distance-sampling email list 3
Data structure 41 Double observer configuration
Getting data into Distance 45 Analysis using CDS and MCDS engine 125
How Distance stores data internally 262 Analysis using MRDS engine 127
Data Layer Properties 256 Project setup 166
Data Layers Double observer methods 127
About 41 DS model
List of layer types 42 About, in MRDS Engine 131

344 • Index User's Guide Distance 6.0 Beta 5

DS Model Geographic projection 53
Specifying in Model Definition 251 Getting Started
DS model formulae Analysis 1 13
About in MRDS Engine 131 Analysis 2 19
DSM See Density Surface Modelling, See Density Objective 13
Surface Modelling Sample Projects 31
Survey Design 21
E GIS data
Format 56
Edge Effects 65 GIS data
Email list 3 Preferences 180
Equal Spacing Zigzag Design GIS Data
First and Last Line Placement 67 Troubleshooting 163
Errors 161 GIS Data
In CDS and MCDS engine 162 About 52
Internal errors in CDS and MCDS analysis engines Copy and paste from Clipboard 57
162 Importing 57
Internal errors in the interface 161 Viewing and Manipulating 52
Known problems 161 Goodness of fit
Estimate tab Chi-square in MCDS 124
CDS and MCDS 237 Cremér-von Mises test 93
MRDS 250 Kolmogorov-Smirnov test 93
Estimating density from a subset of the data in Specifying in CDS and MCDS analysis 245
MRDS 139 Specifying in MRDS analysis 252
Estimating the Detection Function at Multiple Levels Grouped Data
in MCDS 118 CDS 100
Example
Mexico survey design 26
H
More Complex Data Import 19
Sample Projects 31 History of Distance 7
St Andrews Bay survey design 21
Survey Design 21 I
Using Distance to Analyse Simple Data 13
Exporting Data 187 Images produced by R 84
Exporting Projects 38 Import data
Non geographical data 46
Import Data
F
Covariate data example 19
Factor and non-factor covariates Getting started example 1 13
In MCDS 117 Getting started example 2 19
In MRDS 134 Import Data Wizard 172
Factor covariates About 172
Specifiying in MRDS Model Definition 252 Troubleshooting 176
Field names Importing exisisting GIS Data 57
Valid names 276 Importing from previous versions of Distance 39
Fine-tuning a DSM analysis 159 Inferences just on Covered Region
Fine-tuning an MRDS analysis 142 CDS and MCDS 112
Formulae MRDS 140
About in DSM Engine 151 Inside Distance 261
About in MRDS Engine 131 Internal errors 161
About, in CDS Engine 95 Interval Data
CDS 100
G Intervals 233
Iterations
g(0)<1 Showing and controlling maximum number in
CDS and MCDS via Multipliers 108 MRDS 142
MRDS 127
Geographic coordinate system 53
K
Geographic Data 52
Geographic data in Distance 52 Key function

User's Guide Distance 6.0 Beta 5 Index • 345

 Specifying in CDS and MCDS analysis 240 MR model
Known problems 161 About, in MRDS Engine 131
Kolmogorov-Smirnov test 93 MR Model
Specifying in Model Definition 252
L MR model formulae
About in MRDS Engine 131
Limitations MRDS See Mark Recapture Distance Sampling
MCDS engine 314 Multi-model inference 111
Linking to external data 61 Multiple Covariate Distance Sampling
Locking the data sheet 82 Setting up an MCDS Project 116
Log file Multiple Covariates Distance Sampling 115
MCDS engine file format 311 Analysis guidelines 120
Defining MCDS models 116
M Estimating the detection function at multiple levels
118
Manual selection Factor vs. non-factor covariates 117
Of adjustment terms in CDS and MCDS analysis Introduction 115
241 Limitations of engine 314
Map Browser 192 MCDS engine reference 279
Map Properties 258 Output from MCDS analyses 122
Map window 199 Scaling of distances for adjustment terms 119
Maps Multipliers
Map Browser 192 CDS and MCDS 108
Map window 199 MRDS 138
Mark Recapture Distance Sampling Multipliers tab
Analysis guidelines 134 CDS and MCDS 247
Checking the version number 141 Multi-species Study
Defining MRDS models 130 Estimating detection function for rarer species 108
Factor vs. non-factor covariates 134
Fine-tuning an MRDS analysis 142
N
Installing an updated version of the engine 141
Introduction 127 New Features 8
Output from MRDS analyses 135 In Distance 3.5 10
Running from outside Distance 140 In Distance 4.0 8
Setting up an MRDS Project 129 In Distance 4.1 8
Troubleshooting problems 162 In Distance 5.0 8
Using a previously fitted detection function 139 In Future Versions 12
Maximum iterations New Project
Setting in MRDS analysis 142 Creating a New Project 34
Maximum Observations, Samples etc See Limitations Setup Project Wizard 165
MCDS See Multiple Covariates Distance Sampling Non-convex Regions
MCDS engine Zigzag Sampling in 67
Changes 324
MCDS engine fitting algorithms 314
O
MCDS engine reference 279
Misc. tab Opening a project 36
CDS and MCDS 249 Output file
Missing cluster sizes MCDS engine file format 310
CDS 102
Missing covariate values in MCDS 124 P
Missing data
CDS 101 Parametric Indexing 97
MCDS 124 Plot file
Missing cluster sizes in CDS 102 MCDS engine file format 312
Model Averaging Plots
CDS and MCDS 111 Exporting CDS plots 99
MRDS 138 Exporting CDS plots to R 99
Model Definition Exporting MCDS plots 124
Interface 236 Exporting MCDS plots to R 124
Model Definition Properties dialog 236 In CDS Results 90

346 • Index User's Guide Distance 6.0 Beta 5

 In MCDS Results 122 Preferences 181
Produced by R 84 R Image Properties dialog 259, 260
Qq-plots 92 Supported versions 83
Saving to file in CDS and MCDS analysis 245 Updating the Version that Distance Uses 84
Post-stratification R Folder 84
CDS 103 R statistical software 83
MCDS 125 Random number generation algorithms 278
MRDS 137 Recovering from unexpected program exit 164
Preferences 178 References 333
Probability of Coverage 65 Reporting problems 4
Probability of Detection Reserved field names 276
Calculating 95 Results
Problems CDS analysis 89
CDS and MCDS engines 162 DSM analysis 154, 329
GIS 163 Exporting CDS output 98
Known problems 161 MCDS analysis 122
MRDS engine 162 MRDS analysis 135
Recovering from unexpected program exit 164 Running Analyses 81
Reporting 4 Running DSM engine outside the Distance interface
Troubleshooting 161 157, 330
With the Analysis Engines 162 Running MCDS engine outside the Distance
Program Reference 165 interface 279
Project Running MRDS engine outside the Distance
About 33 interface 140
Archiving 38 Running the MCDS engine as a stand-alone
Backing up 36 program 279
Compacting 39 Saving CDS and MCDS results to file 100
Creating 34
Editing 39 S
Exporting 38
Importing 39 Sample definition
Opening 36 In CDS Analysis 112
Sample Projects 31 In MRDS Analysis 139
Saving 36 Sample Projects 31
Template 35 Saving a project 36
Transporting 38 Saving CDS Analysis Results
Viewing 39 Cutting and pasting 99
Project Browser 183 Saving to file 100
Project Properties 177 Saving MCDS Analysis Results
Projection Parameters dialog 255 Cutting and pasting 124
Projection, geographic 53 Scaling of distances for adjustment terms in MCDS
119
Setup Project Wizard 165
Q
Single observer configuration
Qq-plots Analysis using MRDS engine 143
CDS 92 Project setup 166
Qq-Plots Single transect
Specifying in CDS and MCDS analysis 245 CDS and MCDS 113
Smearing 249
R Sponsors 5
Starting values
R 83 Specifying in CDS and MCDS analysis 241
About the link between R and Distance 83 Specifying in DSM analysis 159
Density Surface Modelling 145, 327 Specifying in MRDS analysis 142
dsm library 157, 330 Stats file
Folder 84 MCDS engine file format 311
Images, about 84 Stopping an analysis 163
Installing and Configuring 84 Stratification
Mark Recapture Distance Sampling 127 CDS 103
mrds library 140 DSM 157, 330

User's Guide Distance 6.0 Beta 5 Index • 347

 MCDS 125 Z
MRDS 137
Suggestions Zigzag Sampling Designs
Sending suggestions 4 First and Last Line Placement 67
Survey Browser 195 Introduction 66
Survey Design in Distance Non-convex regions 67
Design Browser 193 zip files
Design Details window 201 Exporting projects as 38
Design Properties dialog 215
Introduction 63
Preferences 180
Setting up a new project 68
Survey Browser 195
Survey Details window 209
Survey Details window 209
Survey Properties dialog 230
Surveys
Working with surveys during analysis 78

T
Temp folder
Cleaning 83
Template, using a project as 35
Troubleshooting 161
Truncation 235

U
Units 236
Unknown Study Area Size 112
Use Agreement 5
Using a previously fitted detection function to
estimate density in MRDS 139

V
Valid field names 276
Variance Estimation
CDS 107
MRDS 137
Variance tab
CDS and MCDS 247
MRDS 252, 253
Version
Distance 254
MRDS Engine 141, 158, 331

W
Warnings
In CDS and MCDS engine 162
Web site 4
Welcome to the Users Guide 1
What is Distance? 5
Which geographic projection? 55
Wizards
Data Entry Wizard 170
Setup Project Wizard 165

348 • Index User's Guide Distance 6.0 Beta 5