Hydrobiologia (2011) 678:1–15

DOI 10.1007/s10750-011-0828-8

REVIEW PAPER

Biotelemetry of crustacean decapods: sampling design,

statistical analysis, and interpretation of data
Edlin Guerra-Castro • Carlos Carmona-Suárez •

Jesús E. Conde

Received: 15 February 2011 / Revised: 27 May 2011 / Accepted: 15 July 2011 / Published online: 7 August 2011
Ó Springer Science+Business Media B.V. 2011

Abstract The implementation of ultrasonic telem- (parametric and nonparametric). Examples are given
etry as a technique to evaluate ecological traits of based on data previously published.
marine organisms has allowed biologists to obtain
information that is hard or impossible to achieve by Keywords Telemetry  Movement analysis 
traditional sampling methods; which in turn improves Sampling design  Crabs  Lobsters
the quantification, in their natural habitats, of activity
patterns, movements, migrations, habitat utilization
and zonation, among other biological and physiolog- History of biotelemetry in crustacean decapods
ical variables. Despite its transcendence, this tech-
nique has been scantly applied in crustacean decapod The application of ultrasonic telemetry in crustacean
studies, mainly due to the high costs involved decapods begins, as far as we know, during the 70’s,
(resources, time, and personnel) and the problem of with Homarus americanus Milne-Edwards (Clifton
how to analyze quantitatively telemetry data to test et al., 1970; Monan & Thorne, 1973; Stasko &
hypotheses. Thus, the aim of this article is to Pincock, 1977). Along this decade, at least seven
evaluate, discuss, and review the statistical methods papers related to movement and migrations of
employed in this discipline and make some recom- lobsters and other crab species were published
mendations concerning the sampling effort and (Table 1). At the end of the 80’s, the interest for
efficiency, as well as to suggest appropriate ways to this technique was retaken by Wolcott & Hines
analyze data, specifically with frequentist statistics (1989) to evaluate muscle activity in blue crabs
(Callinectes spp.). These authors developed special-
ized underwater transmitters that allowed to detect
Handling editor: Koen Martens biological processes, such as ecdysis, mandible
movements, cheliped extension, and other muscular
E. Guerra-Castro (&)  C. Carmona-Suárez  J. E. Conde activities. Afterward, ultrasonic telemetry allowed to
Laboratorio de Ecologı́a y Genética de Poblaciones,
broaden the knowledge on blue crabs; the information
Centro de Ecologı́a, Instituto Venezolano de
Investigaciones Cientı́ficas, Caracas, Venezuela thus obtained could not be acquired by means of
e-mail: [email protected]; [email protected] laboratory experiments, diving, or other conventional
sampling techniques (Wolcott & Hines, 1990; Wol-
E. Guerra-Castro
cott, 1995). In fact, understanding migration, feeding
Escuela de Ciencias Aplicadas del Mar, Universidad de
Oriente, Núcleo Nueva Esparta, Boca del Rı́o, Isla de patterns, agonistic behavior, habitat use, and mating
Margarita, Venezuela patterns of Callinectes sapidus Rathbun, 1896 from

123
2 Hydrobiologia (2011) 678:1–15

the Chesapeake Bay, was only possible by biotelem- technique has been employed mostly on commercial
etry (Clark et al., 1999a, b; Kendall et al., 2001). species. If it is assumed that biotelemetry allows to
Another species which attracted considerable atten- obtain biological and ecological information that is
tion during the 90’s was Maja squinado (Herbst, impossible to acquire by others means, it must be
1788), whose movement patterns, migrations, and recognized that there is a big gap of knowledge
habitat use were poorly known until telemetry was regarding habitat use, activity patterns, migration,
used (see González-Gurriarán & Freire, 1994; Freire movement abilities, and other biological aspects of
& González-Gurriarán, 1998; González-Gurriarán noncommercial decapods. The number of telemetry
et al., 2002). studies on crabs is low as compared to other marine
Recently, biotelemetry in crustaceans has been organisms, like fish or turtles. In fact, since 1970, the
complemented with geographic information systems average number of papers published yearly was 1.05,
(with software such as the Animal Movement Analyst peaking at 1991 with five. After 40 years, only about
Extension (AMAE) to ArcView) to study medium 40 papers have been published on peer-reviewed
and wide scale movements of species such as C. journals; while more than 500 indexed papers of fish
sapidus, Callinectes ornatus Ordway, 1863, Arenaeus telemetry have been published during the same
cribrarius (Lamarck, 1818), and Cancer magister period. The rate of crustacean telemetry publications
Dana, 1852 (see Carr et al., 2004; Holsman et al., has been growing in the last two decades, but the
2006; Guerra-Castro et al., 2007). On the other hand, increment rate is considerable lower than in fishes,
small-scale movements have been approached with whose publication rate has a consistent exponential
the use of automatic positioning systems (mainly behavior (Fig. 1).
using VRAP technology, from VEMCO Radio The reasons that have made biotelemetry unat-
Acoustic Positioning, Vemco Ltd. Halifax, Canada), tractive for carcinologists have not been examined,
specifically in studies on H. americanus and Para- but in any case they should be analyzed and
lithodes camtschaticus (Tilesius, 1815) (see Golet pondered. The main reasons that could explain the
et al., 2006; Jorgensen et al., 2007; Scopel et al., low number of crustacean biotelemetrical studies are,
2009; Watson et al., 2009). The implementation of first, the high costs involved in telemetry projects;
these two systems (GIS and VRAP) has allowed i.e., the receptor system, transmitters (also known as
substantial improvements in the knowledge of eco- pingers), and the resources necessary (time and
logical aspects of these species. funds) to maintain trained personnel in the field.
Callinectes sapidus, H. americanus, P. camtsch- The second aspect is the lack of specific analytical
aticus, M. squinado, C. magister, and C. pagurus methodology to test hypotheses. In general, neither
Linnaeus, 1758 are the species that have attracted the consistent experimental designs, nor statistical pro-
highest efforts in telemetry studies (Table 1). In spite cedures have been followed in crustacean telemetry
of these works, and the comprehensive reviews of studies. Few works properly justify the criteria to fix
Wolcott (1995) and Freire & González-Gurriarán the number of instrumented animals and the moni-
(1998), the potential of this technique in crustacean toring time lapse. Both aspects are discouraging
decapod ecology has not been developed. The enough to deem biotelemetry as an easily accessible
reviews mentioned above describe the advantages technique to obtain biological and ecological infor-
and applicability of telemetry to obtain crustacean mation. Another issue to be considered is the size of
information that is not accessible with conventional the animal and the size of transmitters available. At
experiments or field sampling; however, they do not present, the smallest pinger has a minimum weight in
deal with statistical features of biotelemetry. water of 0.5 g and 16 mm in length; limiting the use
of telemetry in small species or juvenile decapods
which weight lesser than 30 g.
Limitations in the use of biotelemetry However, nowadays analytical limitations can be
in crustacean decapods overcome, and projects can be planned statistically
considering cost and sampling size. In this work, we
Articles in indexed journals dealing with decapod will deal with the following questions: how many
telemetry are listed in Table 1. It is evident that this animals should be instrumented to obtain biological

123
Hydrobiologia (2011) 678:1–15 3

Table 1 Articles related to telemetry of crustacean decapod (extended from Freire & González-Gurriarán, 1998)
Family Species Author Telemetered variable Information obtained

Cancridae Cancer magister Smith & Jamieson (1991) L M

Holsman et al. (2006) L M, HU
Cancer pagurus Bottoms & Marlow (1979) H PH
Hall et al. (1991) L M
Skaaja et al. (1998) L M
Portunidae Callinectes sapidus Wolcott & Hines (1989) L, MA F
Nye (1989) L, MA F
Wolcott & Hines (1990) L, EC M, HU
Shirley & Wolcott (1991) L, EC M, HU
Hines et al. (1995) L HU
Clark et al. (1999a, b) L, C M, A
Carr et al. (2004) L M
Callinectes ornatus Guerra-Castro et al. (2007) L M, HU, AS
Arenaeus cribrarius Guerra-Castro et al. (2007) L M, HU, AS
Scylla serrata Hill (1978) L M
Majidae Chionoecetes opilio Maynard & Webber (1987) L, BP M, E, HU
Maynard (1991) L, BP M, E, HU
Maja squinado González-Gurriarán & Freire (1994) L, T M, MI, HU
Hines et al. (1995) L, T M, MI, HU
González-Gurriarán et al. (2002) L, T M, MI, HU
Anomurans Paralithodes camtschaticus Monan & Thorne (1973) L M, MI, HU
Stone et al. (1992, 1993) L M, MI, HU
Jorgensen et al. (2007) L M, HU
Nephropidae Homarus americanus Lund & Lockwood (1970) L M
Maynard & Conan (1984) L M
O’Dor & Webber (1991) L, T M
Collins & Jensen (1992) L M
Watson et al. (1999) L M, HU
Tremblay et al. (2003) L M, HU
Golet et al. (2006) L M
Scopel et al. (2009) L M, HU
Watson et al. (2009) L M, HU
Homarus gammarus Chapman et al. (1975) L M
van der Meeren (1997) L M, HU
Nephrops norvegicus Bjordal et al. (1993) L M
Newland & Chapman (1993) L M
Palinuridae Panulirus argus Clifton et al. (1970) L M, HU
Herrnkind & McLean (1971) L M
Palinurus elephas Giacalone et al. (2006) L M, HU
A agonistic activity, AS attractive spots, BP branchial pressure, C Chelar spread, E energetics, EC ecdysis, F foraging, H heart rate,
HU habitat use, L location, M movements, MA muscle activity, MI migration, PH physiology, T temperature

and statistically representative results?, how the could be analyzed statistically? Therefore, consider-
sample size should be coupled with resources, time ing the advantages of telemetry as a technique to
and personnel available?, and how the data gathered attain biological data in real time and in the natural

123
4 Hydrobiologia (2011) 678:1–15

56 individuals, a frequency distribution of 100 values of

52 whichever variable can be made up. This is not the
48 case in telemetry studies, because various measure-
44 ments from the same variable can be obtained from
40 the same animal. This problem causes nonindepen-
Number of Articles

36 dence of observations and could produce seriously

32 skewed results (Anderson, 1982; Swihart & Slade,
28 1985).
24 Technically, with ultrasonic telemetry, the
20 researcher is able to detect sounds that indicate, at
16 least, the proximity of a tagged animal. This implies
12 that, with enough number of locations, it is possible
8 to calculate distances between tracked organisms,
4 their position along time, distances from a particular
0 habitat, or distances to the coastline, among others.
1970 1975 1980 1985 1990 1995 2000 2005 2010
All these are measurements of variables related to
Year
habitat use and, as all kind of measurements in
Fig. 1 Number of telemetry studies in fish and decapod biological sampling, there is intrinsic natural vari-
crustaceans during the last 35 years. Filled circles indicate the ability and also sampling errors. Consequently, if
number of papers about crustacean telemetry and the discon-
tinued line represents the linear trend. Open circles indicate the
these distances were measured during the life span of
number of paper in telemetry of fishes and the continued line the crustacean, the real frequency distributions of
represents the exponential behavior each variable for this individual, without sampling
error and with the exact natural variability detected
environment under unfavorable circumstances, this could be obtained. Obviously, this is impossible; but
article focuses on some recommendations about we should be aware that the longer the monitoring or
sampling and analysis of telemetry data. sampling time lapse, the more precise or closer will
be the statistical estimators from this hypothetical
frequency distribution for this individual. On the
The problem of determining the number contrary, it is important to take into account that, no
of individuals and the monitoring time lapse matter how large is the monitoring lapse, the
information obtained is about only one individual
One of the most important objections posed to and, thus, this data is not representative to draw
telemetry manuscripts is the lack of representative- statistical conclusions about the species or group
ness of sample size. Statistically speaking, the only evaluated. This means that the monitoring time lapse
really representative samples are those that have the and the number of animals to tag should be consid-
same frequency distribution as that of the populations ered in telemetry projects. In both situations, basic
being studied (Underwood, 1997). To achieve this, it sampling techniques become important and for that
is necessary to have an adequate sample size and traditional textbooks as Cochran (1977), Green
unbiased measurements (Sokal & Rohlf, 1995). (1979), and Krebs (1999) can be used.
Sample size is a very important issue to consider in To illustrate the importance of the monitoring time
most ecological studies. It is fundamental to distin- lapse and the number of individuals necessary for a
guish between the biological population and the telemetry crustacean project, several analyses using
statistical population of the variable in which we are previously published data will be presented (from
interested (Green, 1979; Underwood, 1997). In Guerra-Castro et al., 2007). In that article, several
conventional crustacean studies, the statistical popu- variables related to activity patterns and zonations of
lations and the biological populations are easily two portunid species (Callinectes ornatus and Are-
distinguishable. From each individual usually is naeus cribrarius) were evaluated. Basically, dis-
extracted only one value (each individual represents tances traveled every 12 h, home range, distances
a sampling unit). Therefore, with a sample of 100 from the shore line, depth and sediment type were

123
Hydrobiologia (2011) 678:1–15 5

estimated. Eleven crabs of each species were instru- However, the problem is important when the vari-
mented with ultrasonic transmitters, and were mon- ability stems from sampling noise or low sampling
itored every 12 h along 7–10 days. For demonstrative intensity (Underwood, 1997). Therefore, if monitor-
purposes, only crabs tracked during 8 days were ing time is not adequate, the estimations could be
selected, and the variable chosen was the distance under- or overestimated in relation to the real value of
traveled, because of its great variability in our study, the parameters that are being studied. Under these
and also because is one of the most measured circumstances, erroneous conclusions can be
variables in crustacean telemetry papers. achieved, and statistical inferences are not sound
The first demonstration deals with how the (Cohen, 1988; Underwood, 1990). These facts bring
precision is improved when the monitoring time on the carcinologist to decide: (1) for how long an
lapse is increased. In Fig. 2, the evolution of the individual should be tracked in order to guarantee an
averaged distance traveled by all the individuals and acceptable precision, considering the costs implied
the standard error is shown (with observation inter- (time in the field, personnel, fatigue, etc.)? and (2)
vals of 12 h). It is evident how the standard errors how many individuals are needed for a sample to be
(the bars) diminish with the increase of the monitor- representative?
ing time lapse and how these get closer until they To answer the first question, an approach using the
approach the mean. The decrease of the standard precision equation commonly used as applied to data
error is a clear indicator of precision improvement after Guerra-Castro et al. (2007) is presented.
(Sokal & Rohlf, 1995); a fact that should be strongly Precision can be obtained from:
considered to understand the wide variability reported
sx
for crustacean movements (e.g., González-Gurriarán P¼ ;
et al., 2002; Guerra-Castro et al., 2007). Certainly, x
high variability is not a problem from a biological or where P is the precision of the sample mean, sx the
a statistical perspective if this variability is real. standard error, and x the sample mean (Krebs, 1999).
In Table 2, the means of the distances traveled every
170 Arenaeus cribrarius 12 h of eight individuals of each species are shown.
The information is presented for 2, 4, 6, 7, and 8 days
or (m)

150
of monitoring. The best precisions were obtained
ean ± Standard erro

130
with 6 days for Arenaeus cribrarius, and 7 and
110 8 days for Callinectes ornatus. Nevertheless, in the
90 original work, most of the individuals were moni-
70 tored for more than 7 days (up to 10 days). This
implied an extra effort and sampling costs that did not
Me

50
improve the precision of the estimators (Table 2).
30
This simple analysis shows that the monitoring time
170 Callinectes ornatus lapse is an important issue to consider in telemetry
Mean ± Standard error (m)

150
projects. All this information can be obtained from a
pilot project or occasionally from data available in
130
articles. In any case, it is also important to consider
110
that making a decision regarding monitoring time is
90 highly dependent on the objectives of the study and
70 the biological questions targeted.
50 Once the adequate time lapse is established, it
must be decided how to catch individuals to guaran-
30
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 tee an unbiased sampling. Statistically speaking, the
Number of measurements every 12 hours during eight days only way to avoid sampling bias is through random
sampling, because it is generally considered to be
Fig. 2 Behavior of average distance traveled and their
standard error with increasing number of localizations in a more likely to lead to independency of data (Under-
timeline of 8 days wood, 1997). Nevertheless, the real experimental

123
6 Hydrobiologia (2011) 678:1–15

Table 2 Means of distances

Crabs Monitoring time lapses (days)
(m) traveled every 12 h
considering different 2 4 6 7 8
monitoring time lapses
A. cribrarius
1 40 25 33 31 29
2 85 48 57 71 70
3 68 43 35 20 20
4 88 59 53 49 46
5 93 119 90 108 103
6 65 44 47 46 47
7 273 154 90 100 100
8 50 31 22 24 29
Mean 95 65 53 56 56
Variance 5,502 2,120 639 1,137 1,039
Precision 0.26 0.23 0.16 0.20 0.19
C. ornatus
1 138 110 94 91 120
2 100 74 62 57 54
3 212 163 127 115 125
4 73 95 93 106 98
5 50 74 66 59 95
6 63 53 45 71 64
7 170 135 101 94 86
8 98 68 56 50 52
Mean 113 97 81 80 87
Variance 3,163 1,401 762 595 795
Data after Guerra-Castro et al. Precision 0.18 0.14 0.12 0.11 0.11
(2007)

units are the crustaceans; thus, it is practically answers to most of these questions are unknown, they
impossible to manage a design so that every member should be kept in mind when setting the pots.
of the population should have an equal and indepen- On the other hand, the number of crabs (samples)
dent chance of being in the sample. Consequently, a should be calculated to obtain representativeness.
way must be found to catch individuals as best as Table 3 shows the statistics of distances traveled by
possible, to give an equal chance to each member of Arenaeus cribrarius, Callinectes ornatus and other
the population of showing up in the sample, and to four decapod species. This table also includes the
achieve this some knowledge on several aspects of sample size needed as estimated using the equation:
population biology of the species is required. For  2
s
example, some crab pots lead to size exclusion; this n  Z2 2 ;
should be evaded if our question includes individuals d
of all sizes within the population. The sites where where n represents the minimum number of crabs,
pots are deployed is another issue to analyze; for Z is the value obtained from a standardized normal
example, if individuals from different sex, size, curve (no matter the original distribution of data
ontogenic stage, etc. use the habitat under a specific because, in general, all distributions of means always
gradient or zonation (latitudinal, depths, distance tend to normality), s2 represents the sample variance,
from shoreline or the surf, etc.), pots must be used in and d is the maximum tolerable error between the
every possible zone to provide equal chance to all sample mean and the population mean (Krebs, 1999).
groups for appearing in the sample. Although the For A. cribrarius and C. ornatus we found that

123
Hydrobiologia (2011) 678:1–15 7

estimated errors, with 95% of confidence, were 14

d = 10 m and 15 m, respectively. At this stage, both precisions
DSS for

were acceptable, considering the locomotion capacity

-12
-15

-60
-430
-202
-144
of displacement that these crabs show. Three exam-
ples of size estimation with desired absolute errors of
Table 3 Accuracy of mean and sample size for desired errors of 20 and 10 m with a confidence of 95% in five telemetrical studies in decapod crustaceans

20, 15, and 10 m will be considered. In the first case,

ASS actual sample size, MTL monitoring time lapse, ESS estimated sample size, d actual error, DSS difference between actual and estimated sample size
the maximum number of crabs needed is six per
d = 10 m
ESS for

species. If this precision is acceptable (under a

443
234
154
23
26

67
biological perspective), Guerra-Castro et al. (2007)
used five extra crabs per species. This implied
unnecessary use of transmitters (10 pingers) and 38
extra days of field work. On the second scenario,
d = 15 m
DSS for

d = 15 m, the sample size used was adequate for

1
0

-72
-58

-23
-184

both species. But in the third case, if the precision

required is 10 m, then 26 individuals of A. cribrarius
and 23 individuals of C. ornatus would have been
d = 15 m

needed to perform an adequate analysis. A similar

ESS for

analysis was applied to the other four species in

10
11

68

30
197
104

Table 3. In all the studies, a substantial number of

individuals were tracked (7–32); nonetheless, if 20 m
is also a desirable error for these species, then more
d = 20 m

crabs should have been instrumented. Likewise, for a

DSS for

desirable error of 15 and 10 m, the number of

-98
-27
-28

-10
5
5

individuals needed is extremely high. Again, the

desirable error should be fixed by the researcher
considering the question posed, the locomotion
d = 20 m

capacity of the species under evaluation, and evi-

ESS for

dently, the capacity to meet the costs that this

6
6

59
38

17
111

precision would have. These aspects could be cleared

up with a pilot project or bibliographic revision;
14
15
58
27
39

31
d

especially the identification of the level of variability

ASS

7
11
11
13
32
10

associated with the species of interest or a close

species.
Standard error

Another way to estimate the number of individuals

should be applied if the purpose of the research is to
6.60
6.50
29.80
13.81
20.00

15.83

compare means. For example, Guerra-Castro et al.

(m)

(2007) compared the movement rates of A. cribrarius

and C. ornatus, Holsman et al. (2006) evaluated the
Mean

González-Gurriarán & Freire (1994)

circadian activity of C. magister, and Golet et al.

75
57

90

51
225
523
(m)

(2006) also studied this topic in H. americanus. In

cases like these, the determination of sampling size is
Daily
Daily

Daily
MTL

12 h

Guerra-Castro et al. (2007)

2h

6h

a little more complicated. The above criterion of

Holsman et al. (2006)

precision is not the best choice, because statistical

Homarus americanusc

Scopel et al. (2009)

Stone et al. (1992)
Arenaeus cribrariusa
Callinectes ornatusa

differences do not necessary mean biological differ-

b
Cancer magister

camtschaticusd

ences (Underwood, 1997; Krebs, 1999). What the

Maja squinadoe

researcher should consider is: (1) the minimum

Paralithodes

difference regarded as biologically important, and

(2) to establish the amount of error Type I (or a, the
probability of rejecting a true null hypothesis) and
b

d
a

123
8 Hydrobiologia (2011) 678:1–15

Type II (or b, the probability of failing to reject a of the project, logistics, personnel, equipment char-
false null hypothesis) that can be tolerated. Once both acteristics and adequate environmental conditions for
criteria are chosen, the standardized difference D is field work. In published crustacean telemetry studies,
obtained using the following formula: the interval time has been very variable, from minutes
d to days, and even months (Table 4). For example,
D¼ ; Golet et al. (2006) monitored 44 lobsters (Homarus
s
americanus) every 5 min for 9 days. The main
where d is the minimum difference desired to detect objective was to determine circadian activity patterns
and s is the standard deviation of the samples that can in a natural environment and thus to evaluate
be obtained from a pilot project or from previous movements at very short periods of time. On the
studies (Krebs, 1999). Using this information, the other hand, Stone et al. (1992) evaluated the seasonal
sample size needed to detect differences of means can migratory patterns of Paralithodes camtschiticus in
be checked in Krebs (1999, p. 238). Another option is Auke Bay (Alaska) with larger time intervals. They
to use software that allows to calculate sample size, monitored 20 individuals once a week for 1 year.
such as G-Power (Faul et al., 2007). Similarly, Holsman et al. (2006) evaluated intertidal
As stated above, to compare the movement rates of migrations and habitat use by Cancer magister in
A. cribrarius and C. ornatus, it was necessary to Willipa Bay (Washington, USA), monitoring 26
decide the minimum difference to be detected. This crabs each 6 h for 7 days. The authors justified this
difference is 40 m; mainly because it is higher than interval time because it allowed to avoid autocorre-
the precision obtainable for each species with our lation problems between locations (a factor that
instruments. Setting a = 0.05, b = 0.05, and always should be taken in account), since it was
D = 1.66, the sample size needed is 10 individuals assumed that each individual had the chance to move
per species. This sample size is substantially higher to wherever possible nearby habitat. In short, the
than the calculated using the precision criterion. important point is that intervals between observations
Nonetheless, it is valid to increase the error Type II should be associated with the biological question.
from 0.05 to 0.10 or 0.20; doing this, the species
sample size would be eight and six individuals per
species, respectively (although the probability of The problem of analyzing and interpreting
failing to reject a false null hypothesis was incre- telemetry data
mented). The important point here is to apply a valid
criterion to pick d, a, and b. Biological considerations Another challenging aspect in the development of
must be appropriate to fix d, taking into account that telemetric projects is the statistical analysis of data,
larger sampling sizes are needed in order to detect especially on those studies aimed at hypothesis
smaller differences. In the second case, a is usually testing. In most works dealing with crustaceans,
fixed at 0.05 by scientific convention (Sokal & Rohlf, parametric statistics have been used, but rarely the
1995), and, b is commonly ignored in ecological data obtained meet the assumptions of parametric
studies, which is an erroneous practice (Underwood, tests (i.e., normality of data, equality of variances,
1997; Krebs, 1999). Anyway, a key point is to and independence of measurements). Some works
understand how to operate the relationship between have faced these problems applying log-transforma-
Error Type I and Type II, and the consequences tion of data, nonparametric tests or generalized linear
involved in fixing a (i.e., 0.10, 0.05, 0.01; 0.001) and mixed models (e.g., Golet et al., 2006; Holsman et al.,
b (i.e., 0.10, 0.05, 0.01; 0.001) in the estimation of 2006; Guerra-Castro et al., 2007). However, it is
sample size and in the interpretation of results in recognized that these alternatives have strong limita-
hypothesis testing (Krebs, 1999). tions. First of all, logarithmic transformations change
The time interval between individual observations the scale of data; this means that statistical differ-
is another aspect to consider in planning a crustacean ences (or not) are conveyed on a nonintuitive scale,
telemetry project. As in monitoring time and sample making it difficult to draw biological interpretations
size, establishing an adequate interval depends on (Quinn & Keough, 2002). More important, logarith-
various factors, such as the objectives and hypothesis mic transformations do not prevent errors Type I in

123
Hydrobiologia (2011) 678:1–15 9

Table 4 Number of crabs and monitoring time lapses used in general telemetry studies of crustacean decapods
Species Author Number of individuals Interval time Monitoring time lapse

Cancer magister Holsman et al. (2006) 13 6h 6–7 days

Callinectes sapidus Wolcott & Hines (1990) 11 &12 h 3–17 days
Clark et al. (1999a) 16 1h 1 day
Carr et al. (2004) 8 1–20 min 37 h
Callinectes ornatus Guerra-Castro et al. (2007) 11 12 h 5–10 days
Arenaeus cribrarius Guerra-Castro et al. (2007) 11 12 h 5–10 days
Maja squinado González-Gurriarán & Freire (1994) 8 0.4–3.1 days 23–56 days
5.0–11.0 days
3.0–6.0 days
González-Gurriarán et al. (2002) 17 3.0–15.0 days 190 days
Paralithodes Stone et al. (1992) 20 Weekly 1 year
camtschaticus Jorgensen et al. (2007) 7 5 min 35–70 h
Homarus americanus Watson et al. (1999) 26 3.0 days 1 year
Golet et al. (2006) 44 1–5 min 5 days
Scopel et al. (2009) 38 2–5 min 5 days
Watson et al. (2009) 25 2 min 5 days

an F test, when variances are heterogeneous (Bude- nonetheless, most of the studies report highly right
scu & Appelbaum, 1981). Also, conventional non- skewed distributions (e.g., Jorgensen et al., 2007:
parametric statistics (e.g., Kruskal–Wallis Test) have Fig. 1) and conventional statistics cannot be applied.
low power (high rate or error Type II) when the Although never or rarely used in crustacean telemetry
number of observations is low (Sokal & Rohlf, 1995). studies, randomization or permuted tests are highly
Furthermore, these techniques are restricted to one recommended when data do not fit normality (Quinn
factor, obstructing the detections of interactions when & Keough, 2002). One of the most powerful tests
two or more factors are included in the design (e.g., developed until now is PERMANOVA (Permuta-
species-sex, circadian activity-sex, circadian activity- tional ANOVA or MANOVA) (Anderson, 2001a, b).
ontogeny, etc.). Similarly, generalized linear model- This is a routine that uses permutations for testing the
ing cannot handle complex experimental designs response of one or more variables to one or more
involving mixtures of nested and orthogonal factors factors in an analysis of variance. In telemetry
(Underwood, 1997); although some suggestions have studies, the basic design should be, at least, of two
been made to overcome this limitation (Breslow & nested factors. The first is referred to individuals,
Clayton, 1993). Taking these into account, we each one with its measurements, and each individual
suggest to carefully consider the following properties nested to the levels of the factor of interest (species,
of the data in order to choose the most appropriate sex, age, stage, etc.). Another alternative to solve the
test. problem of distribution is to work with the means of
each individual as replicates, as we did previously to
Distribution of data estimate sampling size. This is possible because the
Central Limit Theorem states that every sample of
Theoretically, the assumption of normality is required means tends to a normal distribution even if original
to compare the F ratio with the Fisher F distribution data have a skewed distribution (Sokal & Rohlf,
(Sokal & Rohlf, 1995; Underwood, 1997); however, 1995). The limitation of this procedure is that the
even when data do not have a normal distribution but variability within individuals is missed, but this is a
a symmetric one, the F distribution can be used cost to be paid if conventional parametric statistics
(Underwood, 1997). In crustacean movements, are going to be used.

123
10 Hydrobiologia (2011) 678:1–15

Dispersion of data in telemetry studies, in spite of its potential to analyze

this kind of data.
Equality of variances is an assumption of the F test
that allows the estimation of within treatments mean Independence of measurements
squares (Sokal & Rohlf, 1995; Underwood, 1997).
The consequence of the application of an F test with This is one of the most critical assumptions in the
unequal variances is the increase of error Type I in analysis of variance, but paradoxically is the most
hypothesis testing for populations means (Sokal & violated in ecological studies (Hurlbert, 1984; Under-
Rohlf, 1995; Underwood, 1997). For this reason, it wood, 1997; McGuinness, 2002). The core of this
has been recommended to scrutinize the sample assumption in ANOVA is that it allows to remove an
variation before proceeding with an analysis of annoying term in the sum of square estimation. That
variance. Nevertheless, common tests to check for is, if individual replicates do not influence other
equality of variances (e.g., Bartlett’s and Cochran’s replicates (within or among populations), the mea-
test) are more sensitive to nonnormality than surements are uncorrelated and the term is assumed
ANOVA itself, and have a tendency to fail in to be zero (Underwood, 1997). The consequences of
detecting heterogeneity of variances when it is critic nonindependence within treatments are: (1) incre-
to ANOVA, but detect heterogeneity when ANOVA ment of error Type I when replicates are positively
is robust to it (McGuinness, 2002). One good option correlated (spurious difference detected) or (2)
is the Levene’s test; which has been found to be quite increment of Type II error when replicates are
robust to departures from normality (Anderson, 2006) negatively correlated (real differences not detected).
and the most robust and useful of 56 possible tests for Among treatments, positive correlations produce high
homogeneity in extensive simulations (Conover et al., rates of Type II error, while negative correlations
1981). It is also recommended to employ graphical increase error Type I (Underwood, 1997).
methods (residual plots, normal plots, variance vs. In animal movement studies, nonindependence of
means plots) to check the distribution of observations data is a serious problem (Dunn & Gipson, 1977;
and variances, as well as to check for outliers or Swihart & Slade, 1985). First, within individuals,
suspicious values in groups with high variability. The successive locations could be affected by previous
most important point is that ANOVA is robust to ones, generating underestimations of measurements.
error Type I under inequality of variances if sample Second, animals are evaluated simultaneously; this
sizes are sufficiently large and balanced; this is why a means that if animals share the same area, the
balanced design should be used whenever possible. behavior of one could perfectly affect the behavior of
For a detailed discussion of heterogeneity in ANOVA others, creating conditions for nonindependency
see McGuinness (2002). among experimental units. Briefly, telemetry projects
Nonetheless, there are circumstances where vari- have two problems related with dependence: within
ances and sample size are very different, as reported and among individuals.
by Golet et al. (2006), Stone et al. (1992) and Guerra- The solutions to overcome dependence between
Castro et al. (2007), who had to employ generalized localizations within individuals in telemetry studies
linear fixed-effect models and the Kruskal–Wallis test have been discussed by Dunn & Gipson (1977) and
to deal with these problems. An alternative that has Swihart & Slade (1985). The most important recom-
not been used is the application of the F* test from mendation they made is that the tracking plan should
Brown & Forsythe (1974b). This test has been be designed in a way that dependency between
recommended as a parametric alternative when data localizations is brought to a minimum. To do this,
have skewed distributions, different variances and Swihart & Slade (1985) recommend an analytical
unbalanced designs, having a great control of error procedure to calculate the adequate interval time that
Type I under such circumstances (Clinch & Kesel- can avoid potential dependence of measurements. The
man, 1982; Mehrotra, 1997). This test also allows background of this idea is that sufficient time between
two-way orthogonal analysis of variances and multi- localizations gives the individuals the possibility to
ple comparisons (Brown & Forsythe, 1974a). It is utilize all potential habitats in their home range. In
important to highlight that this test has not been used crustacean telemetry studies, this methodology has

123
Hydrobiologia (2011) 678:1–15 11

been only applied by Holsman et al. (2006) in Cancer Example of analysis with existing data:
magister, finding that 6 h is enough time to obtain movements of Callinectes ornatus and Arenaeus
independence of localizations in movements analysis. cribrarius
Moreover, Swihart & Slade (1985) also suggest that
correlation analysis should be made prior to any Taking into account the recommendations of sam-
statistical procedure to identify if the measurements pling analyses presented above with data published
could be considered independent. To detect autocor- by Guerra-Castro et al. (2007) (i.e., mean differences
relations within individuals, two methods have been of 40 m with a = 0.05 and b = 0.20), the best design
suggested: (1) von Neumann temporal autocorrela- was to track six animals per species for 6 days. Since
tions for univariate cases (speed, distance traveled, the original sample size was 11 individuals per
etc.) and (2) Schoener temporal autocorrelations for species and up to 10 days of tracking, randomly we
geographic coordinates (X, Y) (Swihart & Slade, chose six of them and the first 6 days of tracking. It is
1985). On the contrary, no methods have been important to emphasize that this is a demonstrative
suggested to identify correlations among individuals. exercise; the biological analysis of movements of
Dependence between simultaneously tracked indi- these species has already been discussed in Guerra-
viduals is, as far as we know, a problem that has not Castro et al. (2007).
been approached, or even mentioned, in telemetry To test for differences in the movements of these
studies. The main reason to consider this as a problem species, the distribution and dispersion of the data
is that the behavior of one individual could affect the should first be evaluated. It can be safely assumed that
behavior of others, causing positive or negative measurements within individuals are independent
dependence (Underwood, 1997). Besides, tracked because the interval of time between estimates was
individuals can be attracted by local food patches, 12 h. Similarly, independency between individuals
producing spurious correlations between their move- will be assumed since very few animals were tracked
ments. Attraction points should be incorporated in simultaneously. However, after the analysis, tests for
home range analysis because they have an important correlations (nonindependence) and the existence of
effect when more than two individuals are tracked attraction points that could produce erroneous conclu-
simultaneously (Don & Rennolls, 1983). For C. sions will be applied. The distances traveled every
ornatus, this situation was observed; usually some of 12 h have a right skewed distribution (Shapiro–Wilk
the crabs in the study area moved predominantly to test, P \ 0.05; g1 = 2.09) and heterogeneity of vari-
the zones where fishermen anchor their boats and ances between species was detected (Levene’s test,
continuously cast large amounts of fish remains (e.g., P \ 0.05). This implies that data cannot be analyzed
viscera, fins, and heads) overboard (Guerra-Castro with parametric tests as the t test or F test, but could be
et al., 2007). These situations presume that even analyzed with F* test, PERMANOVA, or with non-
correlations among individuals do not indicate parametric tests. Another choice is to obtain means for
dependence if attraction points exist. In brief, the each individual and use these values as replicates for
important issue is to design a tracking plan in which each species. Using this approach, the variability
not all the animals will be simultaneously tracked. If within individuals is not considered, but the assump-
geographic coordinates (X, Y) are considered vari- tion of normality (Shapiro–Wilk test, P [ 0.05;
ables, the potential dependence among individuals g1 = 0.49) and equality of variances (Levene’s test,
could be detected using canonical correlations P [ 0.05) are satisfied and regular parametric tests
(Green, 1978) or nonparametric correlations such as could be applied to find out for differences in the rate of
the Mantel test (Fortin & Gurevitch, 1993), and movements among species. The tendencies in the
observing in a GIS the sequential movements of values of P for the nine tests are shown in Table 5. All
individuals that were simultaneously tracked (Hooge tests indicate the same result: there is enough evidence
et al., 2001). In conclusion, if independency of to reject the null hypothesis of no difference in the
localizations within and among individuals has been movement rate between these two species. These
reached, the data could be analyzed with analysis of results indicate that, in this case, all tests have an
variance; but if not, qualitative analysis using adequate power to detect differences in the movement
geographic information systems is the alternative. rates of C. ornatus and A. cribrarius. Nonetheless, we

123
12 Hydrobiologia (2011) 678:1–15

Table 5 Comparison of P values tendencies in nine tests (parametric and nonparametric) using data of Callinectes ornatus and
Arenaeus cribrarius
Tests Statistical assumptions Complied P

t test (with means as replicates) ND–HV Yes–yes \0.05

Kruskal–Wallis (with means as replicates) ND No \0.05
Mann–Whitney U test (with means as replicates) HV No \0.05
t test (pooling data) ND–HV No–no \0.05
Kruskal–Wallis (pooling data) – – \0.05
Mann–Whitney U test (pooling data) HV No \0.05
ANOVA nested ND–HV No–no
Species \0.05
Individuals (species) [0.05
PERMANOVA nested HV No \0.05
Species \0.05
Individuals (species) [0.05
ANOVA (Brown & Forsythe 1974a, b) – –
Species \0.05
Individuals (species) [0.05
ND normal distribution, HV homogeneity of variances

strongly recommend: (1) F* test, because it does not Conclusions and recommendations
have any restrictions and preserves the variability
within individuals; (2) PERMANOVA, since it pre- These analyses point out that projects in crustacean
serves variability within individuals and develops null biotelemetry can be statistically planned. The key
hypothesis using permutations; and (3) t test using elements that should be considered are the monitor-
means, mainly because all assumptions are satisfied; ing time lapse, the number of animals to be tracked,
although variability within individuals is missed. and the interval time between individual observa-
Since the three nested ANOVA indicate significant tions. The first is essential to assess the individual
differences in movement rates between species, but variability. This knowledge is particularly important
do not among individuals within species, the possible to calculate the necessary amount of time and
statistical erroneous decisions or false outcomes resources that should be invested. Likewise, the
could be Type I between species (negative correla- estimation of the number of individuals to be tracked,
tions) and Type II among individuals (positive or the sample size, is crucial to reach the represen-
correlations). The next step is to identify those tativeness of the sample and to avoid sub or
specimens from different species that where tracked oversampling. This estimation can be obtained using
simultaneously and look for negative correlations two approaches: the precision criterion or the min-
between them, and recognize individuals from the imum difference that should be detected. The third
same species and test for positive correlations among element is particularly important for the problem of
them. At the end, two groups with four crabs and a dependency among observations; but even here, the
third group with two animals were simultaneously biological question could help in defining it. Pilot
tracked. A pairwise Mantel test was applied in each projects, or information obtained from other studies,
group (Table 6), finding no significant correlations are essential to calculate and define these three
among simultaneously tracked animals in neither elements, and are particularly important to make
group. These results imply that the independence appropriate estimations of the resources needed to
assumption within and among groups was reached, reach the objectives of the project.
and possible false outcomes in the ANOVA could be In this work it was determined that, in biotelem-
discarded. etry of crustacean, at least three sources of variation

123
Hydrobiologia (2011) 678:1–15 13

Table 6 Spatial correlation analysis (using X and Y coordinates) among individuals of Callinectes ornatus and Arenaeus cribrarius
tracked simultaneously
Groups Individuals C. ornatus 1 C. ornatus 5 A. cribrarius 1 A. cribrarius 5

I C. ornatus 1
C. ornatus 5 -0.009
A. cribrarius 1 0.538 -0.060
A. cribrarius 5 -0.048 0.303 0.172
II C. ornatus 6 C. ornatus 7 C. ornatus 8 C. ornatus 9
C. ornatus 6
C. ornatus 7 0.219
C. ornatus 8 0.020 0.040
C. ornatus 9 0.062 0.214 0.146
II A. cribrarius 9
A. cribrarius 6 -0.116
Groups are indicated with I, II, III. None of these correlations was statistically significant

exist: within each individual, among individuals from Anderson, M. J., 2001b. Permutation tests for univariate or
the same group, and between groups. This kind of multivariate analysis of variance and regression. Canadian
Journal of Fisheries and Aquatic Sciences 58: 629–636.
data should be analyzed using nested ANOVA, Anderson, M. J., 2006. Distance-based tests for homogeneity of
particularly those ANOVA that are free from the multivariate dispersions. Biometrics 62: 245–253.
assumption of normality (e.g., F*, PERMANOVA). Bjordal, A., A. Engas, A. V. Soldal & T. J. Ovredal, 1993. A
However, there are alternatives. Geographic Infor- new radio link telemetry positioning system. Experiences
from tracking of fish and crustaceans. ICES, Fish Capture
mation Systems (e.g., the Animal Movement Analyst Committee Report B 23: 1–6.
Extension (AMAE) to ArcView, or open software as Bottoms, A. & J. Marlow, 1979. A new ultrasonic tag for the
QuantumGIS) could be very useful to visualize telemetry of physiological functions from aquatic animals.
patterns of movements, and even to produce some Marine Biology 50: 127–130.
Breslow, N. E. & D. G. Clayton, 1993. Approximate inference
specific analysis of movements and home range. in generalized linear mixed models. Journal of the
However, this powerful tool cannot test hypothesis American Statistical Association 88: 9–25.
about biological patterns, a matter fully developed in Brown, M. B. & A. B. Forsythe, 1974a. The ANOVA and
frequentist statistics. In any case, we hope that this multiple comparisons for data with heterogeneous vari-
ances. Biometrics 30: 719–724.
review could broaden the knowledge on sampling Brown, M. B. & A. B. Forsythe, 1974b. The small sample
design and statistical analysis of telemetry in crusta- behavior of some statistics which test the equality of
cean decapods, as well as help in the design and several means. Technometrics 16: 129–132.
analysis of telemetric projects. Budescu, D. V. & M. I. Appelbaum, 1981. Stabilizing trans-
formations and the power of the F-test. Journal of Edu-
cational Statistics 6: 55–74.
Acknowledgments This work is part of the project entitled Carr, S. D., R. A. Tankersley, J. L. Hench, R. B. Forward & R.
‘‘Ecologı́a comunitaria y genética poblacional de jaibas A. Luettich, 2004. Movement patterns and trajectories of
Callinectes ornatus and Arenaeus cribrarius,’’ number 466 ovigerous blue crabs Callinectes sapidus during the
from the Instituto Venezolano de Investigaciones Cientı́ficas, spawning migration. Estuarine Coastal and Shelf Science
IVIC. 60: 567–579.
Chapman, C. J., A. D. Johnstone & A. L. Rice, 1975. The
behaviour and ecology of the Norway lobster, Nephrops
norvegicus (L.). In Barnes, H. (ed.), Proceedings of the
References 9th European Marine Biology Symposium. Aberdeen
University Press, Aberdeen: 59–74.
Anderson, D. J., 1982. The home range: a new nonparametric Clark, M. E., T. G. Wolcott, D. L. Wolcott & A. H. Hines,
estimation technique. Ecology 63: 103–112. 1999a. Foraging and agonistic activity co-occur in free-
Anderson, M. J., 2001a. A new method for non-parametric ranging blue crabs (Callinectes sapidus): observation of
multivariate analysis of variance. Austral Ecology 26: animals by ultrasonic telemetry. Journal of Experimental
32–46. Marine Biology and Ecology 233: 143–160.

123
14 Hydrobiologia (2011) 678:1–15

Clark, M. E., T. G. Wolcott, D. L. Wolcott & A. H. Hines, Green, R. H., 1979. Sampling Design and Statistical Methods
1999b. Intraspecific interference among foraging blue for Environmental Biologists. Wiley, New York.
crabs Callinectes sapidus: interactive effects of predator Guerra-Castro, E., C. Carmona-Suárez & J. E. Conde, 2007.
density and prey patch distribution. Marine Ecology Activity patterns and zonation of the swimming crabs
Progress Series 178: 69–78. Arenaeus cribrarius and Callinectes ornatus. Journal of
Clifton, H. E., V. W. Mahnken, J. C. van Derwalker & R. Crustacean Biology 27: 49–58.
A. Waller, 1970. Tektite 1, man-in-the-sea project: marine Hall, S. J., D. J. Basford, M. R. Robertson, D. G. Rafaelli & I.
science program. Science 168: 659–663. Tuck, 1991. Patterns of recolonisation and the importance
Clinch, J. J. & H. J. Keselman, 1982. Parametric alternatives to of pit-digging by the crab Cancer pagurus in a subtidal
the analysis of variance. Journal of Educational Statistics sand habitat. Marine Ecology Progress Series 72: 93–102.
7: 207–214. Herrnkind, W. F. & R. McLean, 1971. Field studies of homing,
Cochran, W. G., 1977. Sampling Techniques. Wiley, New mass emigration, and orientation in the spiny lobster,
York. Panulirus argus. Annals of the New York Academy of
Cohen, J., 1988. Statistical Power Analysis for the Behavioral Sciences 188: 359–376.
Sciences. Erlbaum, Hillsdale, NJ. Hill, B. J., 1978. Activity, track and speed of movement of the
Collins, K. J. & A. C. Jensen, 1992. Acoustic tagging of lobsters crab Scylla serrata in an estuary. Marine Biology 47:
on the Poole Bay artificial reef. In Swift, I. G. P. S. M. (ed.), 135–141.
Wildlife Telemetry Remote Monitoring and Tracking of Hines, A. H., T. G. Wolcott, E. González-Gurriarán, J. Gon-
Animals. Ellis Horwood, Chichester: 354–358. zález-Escalante & J. Freire, 1995. Movement patterns and
Conover, W. J., M. E. Johnson & M. M. Johnson, 1981. A migrations in crabs: telemetry of juvenile and adult
comparative study of tests for homogeneity of variances, behaviour in Callinectes sapidus and Maja squinado.
with applications to the outer continental shelf bidding Journal of the Marine Biological Association of the Uni-
data. Technometrics 23: 351–361. ted Kingdom 75: 27–42.
Don, B. A. C. & K. Rennolls, 1983. A home range model Holsman, K. K., P. Sean McDonald & D. A. Armstrong, 2006.
incorporating biological attraction points. Journal of Intertidal migration and habitat use by subadult Dunge-
Animal Ecology 52: 69–81. ness crab Cancer magister in a NE Pacific estuary. Marine
Dunn, J. E. & P. S. Gipson, 1977. Analysis of radio telemetry Ecology Progress Series 308: 183–195.
data in studies of home range. Biometrics 33: 85–101. Hooge, P. N., W. M. Eichenlaub & E. K. Solomon, 2001.
Faul, F., E. Erdfelder, A.-G. Lang & A. Buchner, 2007. Using SIG to analyze animal movements in the marine
G*Power 3: a flexible statistical power analysis program environment. In Kruse, G. H., N. Bez, A. Booth, M.
for the social, behavioral, and biomedical sciences. W. Dorn, S. Hills, R. N. Lipcius, D. Pelletier, C. Roy, S.
Behavior Research Methods 39: 175–191. J. Smith & D. Witherell (eds), Spatial Processes and
Fortin, M.-J. & J. Gurevitch, 1993. Mantel tests: spatial Management of Marine Populations. University of Alaska
structure in field experiments. In Scheiner, S. M. & J. Sea Grant, Alaska.
Gurevitch (eds), Design and Analysis of Ecological Hurlbert, S. H., 1984. Pseudoreplication and the design of
Experiments. Chapman and Hall, New York: 342–359. ecological field experiments. Ecological Monographs 54:
Freire, J. & E. González-Gurriarán, 1998. New approaches to 187–211.
the behavioural ecology of decapod crustaceans using Jorgensen, T., S. Lokkeborg, A. Fernö & M. Hufthammer,
telemetry and electronic tags. Hydrobiologia 371(372): 2007. Walking speed and area utilization of red king crab
123–132. (Paralithodes camtschaticus) introduced to the Barents
Giacalone, V. M., G. D’Anna, C. Pipitone & F. Badalamenti, Sea coastal ecosystem. Hydrobiologia 582: 17–24.
2006. Movements and residence time of the spiny lob- Kendall, M. S., D. L. Wolcott, T. G. Wolcott & A. H. Hines,
sters, Palinurus elephas released in a marine protected 2001. Reproductive potential of individual male blue
area: an investigation by ultrasonic telemetry. Journal of crabs, Callinectes sapidus, in a fished population: deple-
the Marine Biological Association of the United Kingdom tion and recovery of sperm number and seminal fluid.
86: 1101–1106. Canadian Journal of Fisheries and Aquatic Sciences 58:
Golet, W. J., D. A. Scopel, A. B. Cooper & W. H. Watson III, 1168–1177.
2006. Daily patterns of locomotion expressed by Ameri- Krebs, C. J., 1999. Ecological Methodology. Addison-Wesley
can lobster (Homarus americanus) in their natural habitat. Longman, Menlo Park.
Journal of Crustacean Biology 26: 610–620. Lund, W. A. & R. C. Lockwood, 1970. Sonic tag for large
González-Gurriarán, E. & J. Freire, 1994. Movement pattern decapod crustaceans. Journal of the Fisheries Research
and habitat utilization in the spider crab Maja squinado Board of Canada 27: 1147–1151.
(Herbst) (Decapoda, Majidae) measured by ultrasonic Maynard, D. M., 1991. Biophysical Ecology of Snow Crab
telemetry. Journal of Experimental Marine Biology and Chionoecetes opilio; Relating Respiration Rates and
Ecology 184: 269–291. Walking Speeds from the Laboratory to Fields Observa-
González-Gurriarán, E., J. Freire & C. Bernárdez, 2002. tions. Dalhousie University, Halifax, NS.
Migratory patterns of female spider crabs Maja squinado Maynard, D. R. & G. Y. Conan, 1984. Preliminary monitoring
detected using electronic tags and telemetry. Journal of of shortterm movements for lobsters (Homarus americ-
Crustacean Biology 22: 91–97. anus) in the Biddeford River, Prince Edward Island, using
Green, P. E., 1978. Analyzing Multivariate Data. The Dryden ultrasonic telemetry. Canadian Atlantic Fisheries Scien-
Press, Illinois. tific Advisory.

123
Hydrobiologia (2011) 678:1–15 15

Maynard, D. R. & D. M. Webber, 1987. Monitoring the Journal of the Fisheries Research Board of Canada 34:
movements of snow crab (Chionoecetes opilio) with 1261–1285.
ultrasonic telemetry. In Proceedings Oceans’87. The Stone, R. P., C. E. O’Clair & T. C. Shirley, 1992. Seasonal
Ocean—An International Workplace: 962–966. migration and distribution of female red king crabs in a
McGuinness, K. A., 2002. Of rowing boats, ocean liners and southeast Alaskan estuary. Journal of Crustacean Biology
test of the ANOVA homogeneity of variance assumption. 12: 546–560.
Austral Ecology 27: 681–688. Stone, R. P., C. E. O’Clair & T. C. Shirley, 1993. Aggregating
Mehrotra, D. V., 1997. Improving the Brown-Forsythe solution behavior of ovigerous female red king crab, Paralithodes
to the generalized Behrens-Fisher problem. Communica- camtschaticus, in Auke Bay, Alaska. Canadian Journal of
tions in Statistics—Simulation and Computation 26: Fisheries and Aquatic Sciences 50: 750–758.
1139–1145. Swihart, R. K. & N. A. Slade, 1985. Testing for independence
Monan, G. E. & D. L. Thorne, 1973. Sonic tags attached to of observations in animal movements. Ecology 66:
Alaska king crab. Marine Fisheries Review 35: 18–21. 1176–1184.
Newland, P. L. & C. J. Chapman, 1993. Locomotory behaviour Tremblay, M. J., Y. Andrade & R. O’Dor, 2003. Small-scale
and swimming performance of the Norway lobster, movements of lobsters (Homarus americanus): an appli-
Nephrops norvegicus, in the presence of an acoustic tag. cation of radio-acoustic positioning and telemetry (RAPT)
Marine Biology 115: 33–37. with an analysis of system resolution. Canadian Technical
Nye, L. A., 1989. Variation in Feeding Behavior of Blue Crabs Report of Fisheries and Aquatic Sciences No. 2439: IV ?
(Callinectes sapidus Rathbun) Measured by Ultrasonic 19 p.
Telemetry. Department of Marine, Earth and Atmospheric Underwood, A. J., 1990. Experiments in ecology and man-
Sciences, University of North Carolina, Raleigh. agement: their logics, functions and interpretations. Aus-
O’Dor, R. K. & D. M. Webber, 1991. Ultrasonic telemetry. The tralian Journal of Ecology 15: 365–389.
Lobster Newsletter 4: 1–2. Underwood, A. J., 1997. Experiments in Ecology: Their Log-
Quinn, G. P. & M. J. Keough, 2002. Experimental Design and ical Design and Interpretation Using Analysis of Vari-
Data Analysis for Biologists. Cambridge University Press, ance. Cambridge University Press, Cambridge.
New York. van der Meeren, G. I., 1997. Preliminary acoustic tracking of
Scopel, D. A., W. J. Golet & W. H. Watson III, 2009. Home native and transplanted European lobsters (Homarus
range dynamics of the American lobster, Homarus gammarus) in an open sea lagoon. Marine and Freshwater
americanus. Marine and Freshwater Behaviour and Research 48: 915–921.
Physiology 42: 63–80. Watson, W. H., A. Vetrovs & W. H. Howell, 1999. Lobster
Shirley, M. A. & T. G. Wolcott, 1991. A telemetric study of movements in an estuary. Marine Biology 134: 65–75.
microhabitat selection by premolt and molting blue crabs, Watson, W. H. III, W. J. Golet, D. A. Scopel & S. Jury, 2009.
Callinectes sapidus (Rathbun), within a subestuary of the Use of ultrasonic telemetry to determine the area of bait
Pamlico River, North Carolina. Marine Behavior and influence and trapping area of American lobster, Homarus
Physiology 19: 133–148. americanus, traps. New Zealand Journal of Marine and
Skaaja, K., A. Fernö, S. Løkkeborg & E. K. Haugland, 1998. Freshwater Research 43: 411–418.
Basic movement pattern and chemo-oriented search Wolcott, T. G., 1995. New options in physiological and
towards baited pots in edible crab (Cancer pagurus L.). behavioural ecology through multichannel telemetry.
Hydrobiologia 371(372): 143–153. Journal of Experimental Marine Biology and Ecology
Smith, B. D. & G. S. Jamieson, 1991. Movement, spatial dis- 193: 257–275.
tribution, and mortality of male and female Dungeness Wolcott, T. G. & A. H. Hines, 1989. Ultrasonic biotelemetry of
crab Cancer magister near Tofino, British Columbia. muscle activity from free-ranging marine animals: a new
Fisheries Bulletin 89: 137–148. method for studying foraging by blue crabs (Callinectes
Sokal, R. R. & F. J. Rohlf, 1995. Biometry: The Principles and sapidus). The Biological Bulletin 176: 50–56.
Practice of Statistics in Biological Research. W. H. Wolcott, T. G. & A. H. Hines, 1990. Ultrasonic telemetry of
Freeman and Company, New York. small-scale movements and microhabitat selection by
Stasko, A. B. & D. G. Pincock, 1977. Review of underwater molting blue crabs (Callinectes sapidus). Bulletin of
biotelemetry, with emphasis on ultrasonic techniques. Marine Science 46: 83–94.

123