Scribd
Upload
42 views

Sampling Protocol

The document provides guidelines for assessing coastal and marine ecosystems, outlining a 10-point sampling protocol developed by Green (1979). The protocol includes steps such as clearly stating objectives, using replication and randomization to reduce bias, including controls, conducting pilot studies to refine methods, considering efficiency and stratification, determining appropriate sample and unit sizes, and analyzing and accepting results objectively. The overall aim is to inventory resources, identify threats, and establish metrics to monitor changes over time in a standardized, statistically valid manner.

Uploaded by

baksten
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
42 views

Sampling Protocol

The document provides guidelines for assessing coastal and marine ecosystems, outlining a 10-point sampling protocol developed by Green (1979). The protocol includes steps such as clearly stating objectives, using replication and randomization to reduce bias, including controls, conducting pilot studies to refine methods, considering efficiency and stratification, determining appropriate sample and unit sizes, and analyzing and accepting results objectively. The overall aim is to inventory resources, identify threats, and establish metrics to monitor changes over time in a standardized, statistically valid manner.

Uploaded by

baksten
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 19

BMB TECHNICAL BULLETIN

Guidelines on the Assessment of


Coastal and Marine Ecosystems

Sampling Protocol
why assess?
• Inventory our resources
• Where are they located?
“
• What is their extent and cover?
• What resources do we have?
• Species?
• Distribution?
• Abundances?
• What are the threats to these resources?
• levels?; Intensity?
• Usefulness
• Identify correct approaches to address threats
• Working metric indicators
• Monitoring and evaluation
Green’s (1979) ten-point [field]
Sampling Protocol

“
Green R.H. 1979. Sampling design and statistical methods
for environmental biologists. John Wiley and Sons. N.Y. 257
pp.
1. State your questions clearly and
concisely (objectives of the study)

 Doable “
 Measurable (Condition? Good, fair, poor; uniformly good? Fair
and poor?; What is their variability?)

 It is at this stage that you should seek advice from a statistician


about the design and subsequent analyses you may need

 Avoid collecting data in an unplanned manner and then seek


statistical advice.
2. Replication

• Unit of sample
• set of observations in a transect,
• group of quadrats,
“
• Independent (free from bias or uninfluenced)

• Take replicate samples at all levels of interests (ex. Locations, times).


Be wary of pseudoreplication (Hurlbert 1987).

• Remember that variation or differences among locations (or times)


can only be demonstrated by comparison to variation within locations
(or times)

• The basis of any statistical test of your hypothesis is the ratio of the
variation among locations to that within locations
• For example, you wish to examine species richness and percentage
cover of corals between locations in the region

• Single factor design for each parameter (species richness and


abundance of corals (% cover))

FACTOR LEVELS n
Poblacion, Brgy1, Brgy2, …,
Location 10
Brgy10

Replicates 2

Total samples 20
Estimated
Extent of Baseline Selection of
Randomization
Coastal Assessment monitoring Sites
Habitat Area

Suggested hierarchal sampling strategy


Sampling Design

Factors

Locations: Poblacion Brgy. 10

Replicates: Transect 1 Transect 2 Transect 1 Transect 2


3. Randomization

“
• Taking samples from “representative” or “typical” places is NOT
random sampling.

• Collect samples independently (without bias).

• Most of the serious violations of statistical tests is the non-


independence of residual (error) terms

• Bias from non-independent samples cannot be removed!

• Make sure that all possible samples are given the same
chance of being sampled.
How to randomize
sampling?

• Sample randomly within a


location; use grid on maps and
“ 1 2 3 4 5
choose a grid from a table of
random numbers 0 9 8 7 6
• Similarly, when choosing day of
sampling, use random numbers
to choose the day. M T W Th F S Su
• In field sampling, avoid
systematic sampling, e.g.
regular intervals- this increases 1 2 3 4 5 6 7
systematic errors.
4. Controls

“
• Concept of “control” in an experiment is clear to all of us (i.e. it is
a “site” or a “set-up” where “treatment” is absent and all else is
the same)

• Replicated samples from a control is important to measure


natural variability

• It is quite difficult but not impossible to look for a control in field


studies.

• In our case, the most pristine site or the site with the least number
of environmental threats (e.g. exploitation, pollution, illegal
activities…)
5. Pilot studies

“
• This is an important component of a good sampling design but rarely
given emphasis.

• Pilot studies are preliminary sampling to:


• Refine and fine tune sampling design,
• Test efficiency of methods,
• Provide a range of variation of a parameter (expected variances).

• This is usually the only way to:


• Assess efficiency of sampling method
• Determine the presence of a large-scale spatial variability that would
make stratification desirable
• Determine size of sampling unit to maximize precision of sample
estimates
• Determine number of replicates required per site
6. Efficiency of sampling
methods
“
• Your sampling method is efficient under all relevant conditions
• If your sampling method more efficient in Site A than at Site B,
then your comparison can be biased. Try standardization.

• Efficiency can mean either “accuracy” (closeness of a


measurement or estimate to the true value) or “precision” (degree
of concordance among a number of measurements or estimates
for the same population).

• In broad terms, accuracy can be thought as reliability and


precision as repeatability
Sampling Design

High precision, poor High accuracy, High accuracy, high


accuracy –good poor precision – precision –good
replication but poorly Well randomized replication and well
randomized design but poorly randomized design
replicated design

Replication improves precision and randomization


improves accuracy
6. Stratification

“
• This requires previous information

• If there is a large-scale pattern of variation in a factor of interest,


then there is a need to stratify.

• Break up your areas into homogenous sub-areas or STRATA

• Allocate sampling effort (number of replicates) optimally to the


strata.
8. Size of sampling units/ number
of replicate samples/cost-benefit
analysis
“
• The size of your sample unit must be appropriate and this
depends on the following:

• Size and shape of organism


• Size of the study area
• Spatial distribution of the organism (clustered or random)
• Cost of the sample collection (time, money, number of
personnel, etc.)

• Maximizes the precision of the estimate.


9. Analysis of data

“
• You must subject your data to the most appropriate type of
analysis.

• You need to determine your design and match it with the


appropriate analysis a priori

• There are a wide range of analytical packages available for


ANOVAs, Multivariate tests, Classification and Ordination tests.
10. Accept your results

• Be objective. “
• Avoid entering into a search for a method that will give you a
more palatable answer.
Thanks!
Any questions?

You might also like