Ecological Applications, 9(1), 1999, pp.

288–300
q 1999 by the Ecological Society of America

ESTIMATING SIZE AND ASSESSING TRENDS IN A COASTAL

BOTTLENOSE DOLPHIN POPULATION
BEN WILSON,1,2,3 PHILIP S. HAMMOND,2 AND PAUL M. THOMPSON1
1University of Aberdeen, Department of Zoology, Lighthouse Field Station, Cromarty, Ross-shire IV11 8YJ, UK
2 Sea Mammal Research Unit, Gatty Marine Laboratory, University of Saint Andrews, Saint Andrews, Fife KY16 8LB, UK

Abstract. We used a case study of a coastal bottlenose dolphin population to present

a framework for determining the number of individuals present and assessing the likely
time scale over which trends in abundance may be determined. Such a framework is ap-
propriate for animal species that possess natural markings sufficient for individual recog-
nition, and may be valuable in the development and implementation of management and
monitoring programs for vulnerable populations.
Population abundance was estimated using mark–recapture methods applied to photo-
identification data. This experiment was designed to minimize violation of method as-
sumptions so as to allow use of the most parsimonious model for analysis. The data were
examined critically to investigate mark–recapture assumptions, while analytical methods
and data were selected to minimize and, where necessary, account for violations. The
estimated number of animals with long-lasting marks from left and right side estimates
were 73 6 12 and 80 6 11 individuals, respectively (means 6 1 SE). When divided by the
estimated proportion of such animals in the population (0.57 6 0.043 and 0.61 6 0.035,
respectively) and averaged, weighted by inverse variance, a total population size of 129 6
15 individual animals was estimated (95% CI 5 110–174 animals).
Data on calves observed and carcasses recovered suggest that the population could be
increasing or decreasing at an annual rate of up to 5%. A power analysis, undertaken to
investigate the length of monitoring program required to detect changes in population
abundance at a 90% level of certainty, showed that detection of a trend could only occur
following .8 yr of research effort. Biennial sampling has power similar to that of annual
sampling, but savings in resources are offset by the loss of data on the reproductive histories
of individuals.
Key words: abundance; bottlenose dolphin; cetacean; management; mark–recapture; monitoring;
Moray Firth, Scotland; North Sea; photoidentification; population trends; power analysis; Tursiops
truncatus.

INTRODUCTION to estimate cetacean population density, and therefore

Bottlenose dolphins (Tursiops sp.) are widespread in abundance, in a defined study area via shipboard or
the world’s temperate and tropical oceans, occupying aerial sightings surveys (Hammond 1986a, Buckland
a variety of marine habitats (Leatherwood and Reeves et al. 1993). But for practical reasons, line transect
1983). Many populations inhabit coastal waters where sampling can be difficult to implement for coastal ce-
they are subject to human activities, and as with many taceans because the behavior of the animals is often
other species of small cetacean, there is a demand for related to coastal topography (Wilson et al. 1997 a).
the development and implementation of conservation This coupled with highly variable school sizes (Wells
management and monitoring programs. et al. 1980) can lead to estimates of abundance with
An integral part of any management strategy is the poor precision. Furthermore, if the exact range of a
assessment of the number of individuals in a population population is unknown then line transect methods can-
and any trends in abundance (Taylor and Gerrodette not determine population size.
Mark–recapture methods use data on the number of
1993). Estimating the number of individuals in a pop-
animals marked and their proportion in subsequent
ulation of cetaceans presents practical difficulties be-
samples to estimate population parameters including
cause they live in the sea, are wide-ranging, and spend
abundance (Seber 1982). These methods have begun
much of the time underwater. However, a number of
to be applied to data derived from photographic records
field techniques first developed for terrestrial animals
of naturally marked individuals (Hammond et al.
have been adapted for the study of cetaceans (Ham-
1990a). When the assumptions of the technique are
mond 1987, 1995). Line transect methods can be used
fulfilled, mark–recapture techniques can provide un-
Manuscript received 2 September 1997; revised 24 Feb- biased estimates of population size that are more pre-
ruary 1998; accepted 10 March 1998. cise than those derived from line-transect sampling
3 E-mail: bw9@st-andrews.ac.uk (Fairfield 1990, Calambokidis et al. 1990). Bottlenose
288
February 1999 ESTIMATING DOLPHIN ABUNDANCE AND TRENDS 289

minimum estimate of 62 individuals based on a coor-

dinated land-based watch (Hammond and Thompson
1991). No estimate of the absolute abundance or of
trends in abundance of this, or indeed of any other,
population of bottlenose dolphins in the northeastern
Atlantic is available.
Previous studies of cetacean abundance have been
carried out as a byproduct of general purpose photo-
identification studies (Hammond et al. 1990). As a con-
sequence, they frequently suffer from violations of the
assumptions of standard mark–recapture methods, in-
cluding unequal capture probabilities of individuals
(resulting from poor geographical spread of effort or
preferential photographic effort towards particular in-
dividuals), and mark loss between sampling occasions
(due to indiscriminate use of marks for identification
or inappropriate sampling intervals). These lead, at
best, to a necessity for more complex methods of anal-
FIG. 1. Map showing the location of the Moray Firth, ysis, resulting in a loss of precision. At worst, if such
Scotland, study area. Known distribution of bottlenose dol- violations are ignored (Begon 1983), it leads to results
phins is also shown as heavy shading (regular sightings) and
lighter shading (occasional sightings). and conclusions which may be severely biased.
In this paper, we present the results of an intensive,
3-yr photoidentification study, to estimate the absolute
dolphins are readily photo-identified, and the abun- size of the resident population of bottlenose dolphins
dance of some populations has been estimated in this in the Moray Firth. Our study was designed as a mark–
way (Hansen 1990, Hansen and Defran 1990, Wells recapture experiment. Spatial and temporal distribution
and Scott 1990, Williams et al. 1993). of surveys, the photographic effort at sea, the matching
Bottlenose dolphins have long life spans and low of photographs, and the choice of the most appropriate
reproductive rates (Wells 1991), and extensive periods data sets and models were made to minimize violation
of monitoring may be required to detect a trend in of mark–recapture assumptions. These procedures
population size. Monitoring programs may thus rep- avoided, as much as possible, the need for complicated
resent a considerable investment of time and resources analytical models. In this way, we aimed to minimize
and require careful consideration at the outset. One bias and maximize precision of estimated population
underutilized way to investigate the ability of moni- size. The data were sufficient to investigate whether
toring programs is to make use of statistical power variants of standard analysis models were needed to
calculations (Gerrodette 1987). These can help address account for assumption violations.
practical questions related to approach, sample size, In this paper we examine the available data on num-
length of program, and required resources (Taylor and bers of calves born each year and numbers of dead
Gerrodette 1993, Van Strien et al. 1997). animals of all ages to make a preliminary assessment
In waters around the British Isles and northwest Eu- of trends in abundance. Finally, we use power analyses
rope, it is widely believed that numbers of bottlenose to explore the investment in time and effort required
dolphins have declined in recent decades (Kayes 1985). to detect likely changes in the number of individuals
Particular concern has been expressed over the well- in this population from repeated mark–recapture esti-
being of the resident population in the Moray Firth, mates.
northeast Scotland (578409 N, 38309 W, Fig. 1), which
is the only known population remaining in the North MATERIALS AND METHODS
Sea. A high prevalence of skin lesions (Wilson et al.
Study area
1997b), use of waterways polluted by human sewage
(Curran et al. 1996), and a growing dolphin-watching The Moray Firth is a large triangular embayment of
industry (Janik and Thompson 1996) have led to a de- the North Sea off northeast Scotland which can be di-
mand for information on the possible impacts on the vided into two parts (Fig. 1). The inner Moray Firth is
population. In addition, part of its range has been pro- sheltered from prevailing winds, is influenced by fresh-
posed as a Special Area of Conservation under the water inputs, and features several coastal narrows with
European Union’s Habitats Directive (Scottish Natural associated tidal races. The outer Moray Firth, and wa-
Heritage 1995), requiring the establishment of man- ters farther to the south, more closely resembles the
agement and monitoring programs. open sea with fewer freshwater inputs and a straight
Prior to the work reported here, the best indication rocky coastline.
of the number of dolphins in the Moray Firth was a Sightings of bottlenose dolphins from land-based
290 BEN WILSON ET AL. Ecological Applications
Vol. 9, No. 1

volunteers (Evans 1980) and boat-based surveys for

seabirds (G. P. Mudge et al., unpublished report) center
around the inner Moray Firth and southern outer Moray
Firth, with more occasional sightings occurring in the
northern outer Moray Firth and along the coasts off
Aberdeen and Saint Andrews, farther south (Fig. 1).
Data collection
Photographs of dolphins were collected during two
types of boat-based survey between 1990 and 1993. In
the inner Moray Firth, surveys were conducted along
a predefined 42-km route. Ten surveys were carried out
in 1990, and two per month were conducted between
March 1991 and February 1993. To avoid problems
associated with pseudoreplication these surveys were
not carried out on consecutive days. In addition, 35
surveys were conducted along only part of this route.
Nine surveys were conducted during September 1992,
which did not follow a predefined route but ranged
#250 km outside of the inner Moray Firth in order to
determine whether different individuals were present
in the outer Moray Firth. A detailed description of sur-
vey design and effort is given in Wilson et al. (1997a).
On surveys, dolphins encountered were counted,
their location and activities noted, and photoidentifi-
cation pictures of their dorsal fins and flanks taken. An
autofocus 35-mm camera equipped with 75–300-mm
FIG. 2. Photo-identification pictures of two adult bottle-
zoom lens, databack, and ISO 200 or 400 color-trans- nose dolphins illustrating the types of mark used for identi-
parency film was used throughout the study. Schools fication. F 5 fin nick (a piece of tissue missing from the
were defined as aggregations of dolphins within 100 trailing edge of the dorsal fin); MS 5 major scratches (as
m of one another engaged in similar activities and, if ‘‘Deeper scratches or minor wounds’’ [Lockyer and Morris
1990]); WFF 5 white fin-fringe (a white apigmented region
moving, heading in the same direction (Wells et al. around the edge of the dorsal fin); AL 5 active epidermal
1987). The sampling protocol was to photograph as disease (as Wilson et al. 1997b); HL 5 healed epidermal
many animals as possible, from both left and right disease (as Wilson et al. 1997b); mS 5 minor scratches (as
sides, in each school. Individuals with obvious marks ‘‘Superficial scratches’’ [Lockyer and Morris 1990]); U 5
were not preferentially recorded and back-lit shots unusually wide, tall, or leaning dorsal fin; D 5 deformity of
the normal body contour.
(which give little detail of skin markings) were avoid-
ed.
Identifying individual dolphins from photographs of each individual from each trip were graded for pho-
Developed films were viewed with an eyepiece over tographic quality and filed in the archive. Grade 3 pic-
a light-table, and individuals identified from features tures were well lit, in focus, free from spray, and taken
such as nicks, rakes, deformities, and epidermal disease parallel with the exposed flank of the animal such that
on both their dorsal fins and flanks (Fig. 2). Unlike if any patches of skin with active disease were present
several other studies of bottlenose dolphins, which used (the most subtle marking used for identification) they
back-lit shots and were therefore limited to using only would be visible. Pictures graded 1 or 2 were of lower
fin nicks (Defran et al. 1990, Würsig and Jefferson quality and were not considered further. All dolphins
1990, Williams et al. 1993), we were able to use all of encountered in the Moray Firth (except some neonates)
the types of mark, given above, for identification. possessed sufficient markings so that they could be
Matches with previously identified individuals were identified from a grade 3 photograph.
made by comparing each new photograph, taken of the Individuals were classified into three broad age
animal’s left or right side, with all others in an existing groups based on their appearance. Large, robust ani-
archive of left- or right-side pictures (Wilson 1995). mals with a dark gray or black coloration were defined
As many features as possible were used to confirm as adults. Dolphins of similar length but with paler,
matches and reduce the possibility of false positives often olive-colored skin, and a less massive body form
(Scott et al. 1990a, Würsig and Jefferson 1990). Ani- were classed as subadults. Small dolphins with fetal
mals that could not be matched were given a new iden- folds (Kastelein et al. 1990) or substantially paler skin
tification number. The best left- and right-side picture than subadults or adults were defined as calves.
February 1999 ESTIMATING DOLPHIN ABUNDANCE AND TRENDS 291

TABLE 1. The durability of markings on bottlenose dolphins in the Moray Firth, Scotland.

Duration of mark (days)

Interquartile
Type of mark Min. Max. Median range Sample size
Unusual fin shapes 623† 1318† 1285 295 10
Deformities 1002† 1002† 1
Major scratches 309 1318† 696 415 15
Minor scratches 68 1318† 395.5 216 20
White fin-fringes 380 1318† 1223.5 195 6
Active disease 75 1054† 495 468 19
Healed disease 66 1318† 340 424.5 20
† Marks still present in last available photograph.

ASSESSMENT OF MARK–RECAPTURE ASSUMPTIONS individual, this behavioral response will bias estimates
Mark recognition of abundance. ‘‘Trap shy’’ behavior (lower probability
of capture following marking) will result in overesti-
Mark–recapture analyses assume that a marked an- mation of population size; ‘‘trap happy’’ behavior
imal will be recognized with certainty if recaptured. (higher probability of capture following marking) will
Failure to do so will bias estimates of population size result in underestimation. As photoidentification uses
upward. This assumption can be violated if poor quality existing marks it involves no physical interaction be-
photographs or ambiguous markings are used to iden- tween animal and researcher in relation to the marking
tify individuals. In a study of North Atlantic humpback event and so behavioral responses of this type cannot
whales, Friday (1997) found, using sensitivity analy- occur.
ses, that mark–recapture estimates of abundance were
significantly biased only if the poorest quality photo- Mark loss
graphs were included. In this study, to avoid any prob- Mark–recapture methods assume that marks are not
lem, both poor and intermediate quality photographs lost during the experiment; the loss of marks results in
were rejected. Furthermore, the high quality of the pho- upward bias of estimates of abundance. Because the
tographs used usually allowed a variety of marks to be duration of different kinds of marks on small cetaceans
used to confirm each identification. Therefore, we be- is variable (Würsig and Jefferson 1990), we determined
lieve that violation of the mark recognition assumption how long each type of mark could be expected to be
was extremely unlikely. visible, so that appropriately durable marks could be
used for identifying individuals to include in estimates.
Behavioral responses The types of mark used for identification were de-
Standard mark–recapture methods assume that fined and their longevity measured. To do this, indi-
marked animals have the same probability of being viduals with dorsal fin nicks (a feature believed to be
recaptured as unmarked animals. If the action of cap- permanent [Scott et al. 1990a]) and long photographic
ture changes the future probability of recapture for an histories were chosen. The other marks were then fol-
lowed from their first photographic documentation until
the last time that they were visible and their durations
calculated (Table 1). In addition to nicks, these data
suggested that deformities and unusual fin shapes per-
sisted throughout the study whilst other marks were
observed to fade or completely disappear over a period
of weeks to a year.
In addition, so-called ‘‘discovery curves’’ (Williams
et al. 1993) were plotted for all individuals and for the
subset of individuals possessing those marks deter-
mined to be long lasting (Fig. 3 [source data for the
subset shown in Fig. 4]). These plots support the result
that mark loss would be a problem if analyses were
based on animals identified using all types of marks
over the entire period of the experiment, but that this
would not be a problem if identifications were made
FIG. 3. Discovery curves showing the number of indi- using only long-lasting marks.
viduals identified against the cumulative number of dolphins Geographical closure
encountered during the study. Line A is the curve for all
animals. Line B is for animals with long-lasting markings An estimate of abundance that purports to represent
only (nicks, deformities, and unusual fin shapes). population size has limited value unless that population
292 BEN WILSON ET AL. Ecological Applications
Vol. 9, No. 1

FIG. 4. Sightings of dolphins with long-lasting marks (nicks, deformities, and unusual fin shapes).

can be defined. Standard mark–recapture models are identified over time represents an addition of ;4 new
available for analyzing data collected where, first, the individuals/yr. As on average six neonate calves were
population is assumed closed to births, deaths, immi- observed each year, the slight increase in the discovery
gration, or emigration and, second, for those where curve can readily be explained by recruitment of sur-
these assumptions are relaxed. Population sizes esti- viving calves into the marked population.
mated from these latter, open, population models are The acquisition of dorsal fin nicks and other long-
invariably less precise than those estimated from the lasting marks in small cetaceans is cumulative (Würsig
former, closed, models. and Jefferson 1990) therefore animals with these marks
The discovery curve for individuals with long-last- will tend to be the older individuals, especially adults.
ing marks (shown in Fig. 3) strongly suggests that our If immature bottlenose dolphins range more widely, as
study population of bottlenose dolphins in the Moray is the case for many mammals (Greenwood 1980), be-
Firth was closed to permanent immigration. The slight tween-population mobility of younger and therefore
increase in curve B from 500 to 1500 animals photo- poorly marked individuals could be occurring. We in-
February 1999 ESTIMATING DOLPHIN ABUNDANCE AND TRENDS 293

of capture. This assumption is likely to be violated in

cetacean mark–recapture experiments because of in-
herent differences in the behavior of individuals. In-
dividual preferences for certain areas may affect the
probability of encountering an animal, and individual
differences in surfacing rates or boat avoidance be-
havior may affect the probability of obtaining a usable
photograph (Hammond 1986b). Violation of the as-
sumption of equal probability of capture, known as
heterogeneity of capture probabilities, results in un-
derestimation of population size.
In the Moray Firth, bottlenose dolphins most fre-
quently use two areas: the inner Moray Firth and the
southern part of the outer Moray Firth (Wilson et al.
FIG. 5. Discovery curves for two age classes of animals 1997a). Occasionally, animals extend their range to
without long-lasting markings. Each curve shows the number other parts of the outer Moray Firth and areas further
and point at which previously unrecognized animals joined south (Fig. 1). In addition, individuals show significant
the photo-archive.
preferences for specific parts of the main areas (Wilson
et al. 1997a). In this experiment, sampling covered as
vestigated this by comparing the discovery curves for wide a geographical range as possible but was mostly
adult and subadult animals with temporary markings. within the inner Moray Firth and thus truly equal prob-
If there were a pool of mobile subadults in the popu- abilities of encountering each individual could not have
lation we would expect the discovery curve for sub- been achieved.
adults to be steeper than that for adults. In fact, the Inherent differences in surfacing rates or boat avoid-
terminal rates of discovery for these two groups were ance may also be expected (Payne et al. 1983, White-
almost identical (slope of subadult curve 0.012, adult head 1996). To minimize this problem, as noted above,
curve 0.011, Fig. 5) providing further evidence of geo- data collection protocol dictated that attempts be made
graphical isolation. to photograph every individual in a school and that
Male bottlenose dolphins range more widely (Scott preferentially photographing any particular individual
et al. 1990b) and tend to have heavier body scarring was avoided. Furthermore, no data were derived from
than females (Tolley et al. 1995, Wilson 1995). They visual identifications made at sea.
may, therefore, be overrepresented in the group of in- Despite these attempts to minimize heterogeneity, it
dividuals with long-lasting marks. However, the com- was unlikely to have been completely overcome, and
position of animals with long-lasting marks changed thus we investigated models which allowed relaxation
very slowly (Fig. 3, and curve B) and so suggests that of this assumption.
interchange of males between this population and oth-
ers is unlikely. Furthermore, even in populations of ESTIMATING THE NUMBER OF ANIMALS WITH
bottlenose dolphins with abutting ranges, as in Sarasota LONG-LASTING MARKS
and Tampa Bays in western Florida, measured levels
Model selection
of interchange are small (3%/yr) (Wells et al. 1987).
In contrast, the Moray Firth population is considerably Sufficient data were available to use either open or
more isolated than the populations off Florida; with the closed population models for estimating abundance
nearest other known aggregation being .450 km away but, because of the need to allow for heterogeneity of
(K. Grellier, in press). This adds support to our con- capture probabilities, which is problematic in open
tention that levels of interchange between bottlenose population models, closed models were used. Two
dolphins in the Moray Firth and elsewhere are likely models were selected. One, known as Mt, allows cap-
to be negligible. ture probabilities to vary by time (sampling occasion)
Our data do not permit us to rule out permanent only. Darroch (1958) derived a maximum likelihood
emigration as a possibility but there is no evidence that estimator for abundance for model Mt and an expression
animals from the Moray Firth travel more widely than for its asymptotic variance. The second model, known
indicated in Fig 1. Overall, therefore, we are confident as Mth, allows capture probabilities to vary by time and
that the bottlenose dolphins in the Moray Firth do rep- by individual. Chao et al. (1992) have derived a non-
resent a geographically isolated and hence definable parametric estimator for abundance for model Mth using
population. the idea of sample coverage (defined as the relative
fraction of the total individual capture probabilities of
Heterogeneity of capture probabilities the captured animals) and an expression for its as-
Standard mark–recapture methods assume that, with- ymptotic variance from an expanded Taylor series. We
in a sample, all individuals have the same probability implemented both these models using program CAP-
294 BEN WILSON ET AL. Ecological Applications
Vol. 9, No. 1

TURE (Otis et al. 1978, Rexstad and Burnham 1991) where n is the total number of animals from which u
to estimate the number of animals with long-lasting was estimated.
marks. Program CAPTURE derives confidence inter- Confidence intervals for total population size were
vals under the assumption that the number of individ- calculated by assuming that the error distribution was
uals not captured in the population is lognormally dis- the same as for the estimate of the number of animals
tributed. This has the desirable property that the lower with long-lasting marks, as estimated for the Darroch
bound of the confidence interval cannot be less than and Chao models, i.e., that the lower and upper con-
the number of captured individuals. The upper bound fidence limits were the equivalent number of standard
tends to be larger than would be the case if the abun- errors away from the estimate.
dance estimator were assumed to be normally distrib- Williams et al. (1993) estimated u from the propor-
uted. tion of photographs of animals with long-lasting mark-
ings. We were able to improve on this by estimating u
Data selection from the proportion of individuals encountered. This
Population closure is only a reasonable assumption was because the variety of skin markings made it pos-
when experiments are of relatively short duration. Dur- sible to distinguish all individuals in grade 3 photo-
ing our study, bottlenose dolphins showed a seasonal graphs on any particular day. The number of individ-
distribution in the Moray Firth (Wilson et al. 1997a) uals with and without long-lasting marks was deter-
with the greatest numbers found within the inner Moray mined for each survey and summed for each May–
Firth in summer (May–September). Limiting analysis September period. If probability of capture was inde-
to data from May through September in a single year pendent of whether or not an individual had long-last-
is thus a good approximation to closure. These summer ing marks, our estimates of u for each year should be
data were used to estimate population size for the three unbiased.
years (1990, 1991, 1992) independently. Animals identified as calves were treated in the same
Individuals encountered during this period included way as those without long-lasting marks and were in-
almost all individuals seen in other areas of the Moray cluded in the estimates of total population size through
Firth and should thus also be representative of the the parameter u.
whole population. To explore the effect of including For any set of samples, data were available to cal-
data from a wider area than the inner Moray Firth, culate estimates of abundance from left-side photo-
estimates of population size were also calculated with graphs and right-side photographs. Both the left and
additional data collected between May and September right sides were known for some animals but not all.
of 1992 in the outer Moray Firth. We therefore calculated separate estimates for left- and
The investigation of mark duration (Table 1) showed right-side data and combined them as an inverse vari-
that nicks, unusual fin shapes, deformities, major ance weighted average, assuming independence.
scratches and white fin-fringes were sufficiently long-
ANALYSIS OF STATISTICAL POWER OF MONITORING
lived to be considered permanent marks over these 5-
PROGRAMS
mo periods. Only animals identified from these marks
were used to estimate population size. Animals re- To investigate the power of a series of population
corded as calves were not included because their prob- estimates to detect change, we used Gerrodette’s (1987)
ability of capture is not independent from that of their general inequality model
mothers (Wells and Scott 1990).
r 2 n 3 $ 12 CV 2 (z a /2 1 z b ) 2
Results from the mark–recapture analyses give es-
timates of the number of animals with long-lasting where r is the annual rate of population change (in-
marks. Separate estimates were calculated from the left crease or decrease), n is the number of estimates of
and right side data sets. population size, CV2 is the squared coefficient of vari-
ation of estimated total population size, za/2 is the one-
ESTIMATING TOTAL POPULATION SIZE tailed probability of making a Type I error, and zb is
Total population size was estimated as the probability of making a Type II error. The proba-
bility of making a Type I or II error was set at the 0.10
N̂
N̂total 5 level.
û The effect of investing in different amounts of sam-
where N̂total 5 estimated total population size, N̂ 5 pling effort was explored using a range of values for
mark–recapture estimate of the number of animals with the estimated CV of population size, which encom-
long-lasting marks, and û 5 estimated proportion of passed the lowest and highest values from analyses
animals with long-lasting marks in the population, with conducted to estimate population size.
variance estimated using the delta method as The seasonal distribution of bottlenose dolphins in
the Moray Firth precludes estimation of population size

1 2
ˆ total ) 5 N var(Nˆ) 1 2 û using mark–recapture methods more than once a year.
ˆ 2 1 nû
var(N ˆ total
2
N The effect of varying the frequency of population size
February 1999 ESTIMATING DOLPHIN ABUNDANCE AND TRENDS 295

TABLE 2. The number of bottlenose dolphin (Tursiops trun- dolphins with long-lasting marks is most appropriately
catus) individuals identified from grade 3 (best quality)
pictures on surveys in the Moray Firth between May and estimated from inner 1 outer Moray Firth survey data.
September, 1990–1992. Outer Moray Firth surveys were
only conducted in 1992. Proportion of animals with long-lasting marks
The estimates of the proportion of dolphins with
1990 1991 1992
Moray Firth long-lasting marks varied from 0.56 to 0.68 depending
survey area Left Right Left Right Left Right on the sample data used (Table 4). Data were drawn
Inner 67 78 66 72 80 75 from inner Moray Firth surveys (and inner 1 outer
Outer 0 0 0 0 11 38 Moray Firth surveys in 1992), for the three years and
Inner and outer 67 78 66 72 85 98
for left- and right-side photographs.

Total population size

estimation was investigated using five different sam- The estimates of the number of animals with long-
pling regimes (annual, biennial, triennial, etc.). lasting marks from the inner 1 outer Moray Firth sur-
RESULTS veys in 1992 for left- and right-side photographs are
73 and 80 individuals, respectively, and the estimates
Number of animals with long-lasting marks of the proportion of dolphins with long-lasting marks
Between 66 and 98 individuals were identified, from for these data sets are 0.57 and 0.61 of the dolphins.
grade 3 left- or right-side photographs during each of When combined, using the method outlined above,
the three years in the experiment (Table 2). these give a best estimate of the total number of bot-
Results from the mark–recapture analyses using the tlenose dolphins in the Moray Firth in 1992 of 129 6
Mt and Mth models are presented in Table 3. For each 15 animals (mean 6 1 SE; CV 5 0.12). The 95% con-
model, there are independent estimates for the three fidence interval is 110–174 animals.
years and, within each year, independent estimates for
left and right side photographs. In 1992, estimates are Number of calves and dead animals
presented for inner Moray Firth surveys alone, and in- Between 1988 and 1994, 30 births of identified
ner plus outer Moray Firth surveys combined. In all calves occurred within the Moray Firth dolphin pop-
cases, the estimates from the Mth model were larger, in ulation, two calves disappeared soon after birth, and
some cases much larger, than the equivalent estimates 19 other carcasses, representing all age groups, were
from the Mt model. This indicates that there was het- found stranded (Table 5; R. K. Reid, SAC Veterinary
erogeneity of capture probabilities within our data Services, Scottish Strandings Scheme, personal com-
(Chao et al. 1992). As expected, therefore, we were munication; B. Wilson, University of Aberdeen, un-
unable to ensure equal probability of capture in the published data).
field. The Mth model is thus the most appropriate model The mean number of known births is 6 animals/yr.
for analysis and only these estimates are considered When divided by the estimated population size in 1992
further. this gives a mean minimum annual birth (and early
If there were differences in the use of the inner and survival) rate of 4.6% for the population. The exact
outer Moray Firth by individual animals, the wider cov- birth date of many of these calves is known and so this
erage of the inner 1 outer surveys should provide a value is unlikely to be inflated by the presence of calves
more representative sample. For this reason, and be- .1 yr of age. However, as not every animal in the
cause the larger number of surveys resulted in larger population was photographed in each year (an esti-
sample sizes, we propose that the number of bottlenose mated 8–32% were unphotographed), it is possible that

TABLE 3. Dolphin population data from mark–recapture analyses for the Mt (Darroch 1958) and Mth (Chao et al. 1992)
models.

Model Mt estimates Model Mth estimates

Side of No. of
Year Survey area dolphin surveys N̂ SE (N̂) CV (N̂) 95% CI N̂ SE (N̂) CV (N̂) 95% CI

1990 Inner Left 11 58 5.8 0.100 50–75 100 26.2 0.262 67–178
Right 11 58 3.8 0.066 53–69 76 12.2 0.161 60–112
1991 Inner Left 21 42 1.6 0.038 40–48 45 3.7 0.082 41–58
Right 21 46 1.5 0.033 44–51 49 3.9 0.080 45–63
1992 Inner Left 13 68 9.2 0.135 55–93 84 19.2 0.229 60–142
Right 14 48 3.0 0.063 44–57 59 8.9 0.151 48–86
1992 Inner and Left 16 68 7.8 0.115 57–89 73 12.4 0.170 57–110
Outer Right 17 64 3.8 0.059 59–75 80 11.1 0.139 66–113
Notes: Side 5 the side of the dolphin used to calculate the estimate. N̂ 5 estimate of number of individuals in the population
with long-lasting marks; SE 5 standard error; CV 5 coefficient of variation; 95% CI 5 95% confidence interval derived under
the assumption that the number of individuals in the population not captured is lognormally distributed.
296 BEN WILSON ET AL. Ecological Applications
Vol. 9, No. 1

TABLE 4. Proportion of bottlenose dolphin individuals in the Moray Firth population with long-lasting marks, as estimated
from the different data sets.

Long-lasting marks
Survey
Year Side area Number with Number without û SE (û )
1990 left Inner 75 36 0.68 0.044
right 97 53 0.65 0.039
1991 left Inner 107 78 0.58 0.036
right 121 96 0.56 0.034
1992 left Inner 68 52 0.57 0.045
right 88 58 0.60 0.040
1992 left Inner and 77 57 0.57 0.043
right Outer 116 73 0.61 0.035
Notes: Side 5 the side of the dolphin used to calculate the estimate; û 5 estimated proportion of animals with long-lasting
marks in the population; SE 5 standard error.

some calves were unidentified during their first year of change increases, the importance of precision in pop-
life. Accounting for this increases the estimated annual ulation estimates decreases.
birth rate from 4.6% to ;5–6%. This approximation As the interval between estimates increases, the
encompasses the 5.5% given by Wells and Scott (1990). number of estimates required to detect the trend de-
The mean number of known mortalities is 3 animals/ creases (Table 6). This is because the effective rate of
yr, representing a mean minimum annual mortality rate change increases with interval. This apparent saving in
of 2.3% of the population. This value is a minimum resources is offset, however, by an increase in the time
because we know that not all dolphins that died were taken to detect the trend. Surveys conducted once every
recovered; for example the two lost calves in 1991 and 5 yr may take up to double the time to detect a trend
1992 (Table 5) were never found. It is possible, likely compared with annual surveys. This increased time un-
even, that significant numbers of dolphin carcasses til detection leads to substantial differences in the size
have not been recovered. If only ;one-third of car- of the population at the point of trend detection. For
casses were actually recovered then the annual mor- example, a population of 129 animals that is decreasing
tality rate could be as high as 10% or more. at a rate of 5%/yr would have decreased to 85 animals
Thus, our most optimistic scenario is that the pop- before the trend was detected using annual estimates
ulation is increasing at ;3%/yr but, if significant num- but would have decreased to just 59 individuals using
bers of dead dolphins were not recovered, the popu- estimates every five years.
lation could be declining at a rate of $5%/yr.
DISCUSSION
Effectiveness of monitoring programs Estimates of population size
When the relationship between the rate of population The Moray Firth contains the only resident group of
change and the time required to detect trends for three bottlenose dolphins in the North Sea and one of the
levels of estimate precision are plotted (Fig. 6) the best-known in European waters, but our estimate of
following becomes clear: (i) the length of time required
to detect a trend in population size decreases with in-
creasing rate of population change; (ii) the precision
of the annual estimates of population size has a con-
siderable effect on trend detection; and (iii) as rate of

TABLE 5. Demographic data for the bottlenose dolphin pop-

ulation observed in Moray Firth, Scotland, during 1988–
1994.

Known mortalities
Known births
(observed Recovered Missing
Year neonates) carcasses calves
1988 no data 3 no data
1989 no data 3 no data
1990 5 1 0 FIG. 6. Relationships between the rate of population
1991 6 0 1 change, time until trend detection, and estimate precision for
1992 7 4 1 annual population estimates. Three levels of precision are
1993 2 6 0 given. The highest and lowest (CV 5 0.07 and 0.31) represent
1994 10 2 0 the range encountered in this study, and the intermediate (CV
5 0.12) that of the best estimate. The probability of Type I
Annual mean 6.0 3.0
and II errors was set at 0.10.
February 1999 ESTIMATING DOLPHIN ABUNDANCE AND TRENDS 297

TABLE 6. Observation effort required to detect a statistically significant (at the 0.10 probability level) trend in population
size under different directions of 5%/yr population change. Data variability is specified at CV 5 0.12.
a) Increasing population
Number of Number of survey Effective % change Number of years Total % change
years between episodes required per interval t to detection at detection
estimates (t) ( n) (1.05t 2 1) (t [n 2 1]) (1.05t(n 2 1) 2 1)
1 9 5 8 48
2 6 10.3 10 63
3 4 15.8 9 55
4 4 21.6 12 80
5 3 27.6 10 63

b) Decreasing population
Number of Number of survey Effective % change Number of years Total % change
years between episodes required per interval t to detection at detection
estimates (t) ( n) (0.95t 2 1) (t [n 2 1]) (0.95t(n 2 1) 2 1)
1 9 25 8 234
2 6 29.8 10 240
3 5 214.3 12 246
4 4 218.5 12 246
5 4 222.6 15 254

absolute abundance is the first available for this pop- as possible were used to confirm each identification,
ulation. It confirms that the minimum estimate of 62 so that false matches could be avoided, and each new
dolphins derived by Hammond and Thompson (1991) picture was compared with all previous pictures so that
was precisely that; only about half of the population recaptures would not be missed. Individual dolphins
was counted. Nevertheless, our estimate of 129 indi- were stratified by their mark types so that only those
viduals (95% CI 110–174 individuals) shows that the with long-lasting marks were included in the mark–
population is very small. Many coastal populations of recapture analyses to further reduce mark loss. Finally,
bottlenose dolphins elsewhere in the world appear to the choice of models was based on the assumptions of
be of similar size (Hansen 1990, Wells and Scott 1990, the mark–recapture technique most likely to have been
Williams et al. 1993, Liret et al. 1994). However, the broken during the sampling or photo-analysis stages.
geographic isolation of the Moray Firth population jus- However, it is clear that even with a substantial input
tifies the particular concerns expressed about its vul- of time and resources, a wide range of estimates is
nerability. possible (Table 3), especially so because results were
A key feature of this study was that the mark–re- only calculated for analyses conducted after initial data
capture experiment was designed specifically to enable and model selection. This point cannot be overempha-
population size to be estimated with minimum bias and sized. It is straightforward to calculate mark–recapture
maximum precision within practical constraints. To do
estimates from photoidentification data, but care and
this, the spatial distribution of survey effort was
thought are needed before and after the data are col-
planned so that it provided samples from the main parts
lected to ensure that such estimates are meaningful.
of the population’s known range, allowing the whole
In this study, 60% of photographs taken were of
population to be included in the estimate. Survey effort
quality grade 3. Of the factors that reduced this quality,
was further spread over time, so that samples could be
the most common was low light intensity. By simply
drawn to represent a variety of intervals, allowing later
mark-longevity analyses (rather than sample availabil- choosing less cloudy days to carry out survey work,
ity) to dictate the selection of an appropriate capture the number of rejected photographs could be reduced.
history duration for the analyses. During the surveys However, in temperate areas such as the Moray Firth,
themselves, photographs of as many individuals as pos- suitable weather windows for this work are at a pre-
sible were taken irrespective of their marks so that the mium and already limit the number of surveys possible.
proportion of marked to unmarked individuals could This problem could be side-stepped by using more than
be estimated and so that the impact of individual het- one survey platform simultaneously during periods of
erogeneity could be reduced. Field sightings of rec- the best weather. However, increasing sampling effort
ognizable individuals were not included in later anal- is costly and needs to be balanced against the value of
yses unless, like all other individuals, they were rep- increased precision of population estimates. Alterna-
resented in the photographs. At the photo-analysis tively, improvements could be relatively easily made
stage, all pictures were graded and middle to low stan- by choosing higher speed film or camera lenses more
dard pictures rejected to reduce the probability of efficient at light gathering. The use of equivalent grain,
marks going unrecognized at recapture. As many marks but higher speed black and white, rather than color,
298 BEN WILSON ET AL. Ecological Applications
Vol. 9, No. 1

film could be considered and an investment in high Nevertheless, these approximate values provide a
quality lenses has few drawbacks. useful guide for the development of a management pro-
Bottlenose dolphins, like other odontocetes, are gram for Moray Firth bottlenose dolphins. Changes in
highly social and associations among individuals are population size are likely to be slow (probably ,5%/
not random (Wells et al. 1987). In analyses of mark– yr). If mark–recapture estimates of precision similar to
recapture experiments, ‘‘captures’’ of individuals are that presented here (CV 5 0.12) were made annually,
assumed to be independent events, but for populations it would take .8 yr to detect a significant (at the 10%
that form cohesive groups this assumption will be vi- probability level) trend in population size (Table 6). At
olated. This should not bias estimates of the number such small levels of annual change, the precision of
of individuals in a population, but it may result in a estimated population size has a large effect on the
false sense of precision (an underestimation of vari- length of time needed to detect this change with con-
ance). The extent of this effect will depend on the type fidence (Fig. 6). Clearly, information on population
of social structure, on the fluidity of associations among trends cannot be produced within the space of a few
individuals, and on the proportion of the population years. Provision must be made to ensure the continu-
captured at each sampling event. The influence of social ation of a consistent research effort, ideally as part of
behavior on mark–recapture estimates of populations an overall management strategy.
of social cetaceans is a complex issue that has yet to Monitoring population size through biennial esti-
be addressed. However, studies of the social organi- mates would make little difference to the total time
zation of bottlenose dolphins (Wells et al. 1987) sug- needed to detect the population change (Table 6) and
gest that they live in so-called ‘‘fission–fusion’’ soci- would save survey effort and resources. However, these
eties in which schools often split and join, making savings need to be balanced against the loss of infor-
school membership highly dynamic. Thus in this spe- mation on the birth of calves to known females and on
cies, the impacts of social structuring on mark–recap- calf survival. These data are required for application
ture estimates are likely to be slight. of the birth-interval approach of Barlow and Clapham
Similarly, if school size were to influence the de- (1997) and to obtain a fuller understanding of the rea-
tectability and subsequent photographic capture of in- sons for changes in population size (Taylor and Ger-
dividuals in schools then social structuring within a rodette 1993).
population could lead to biased estimates. Since a fea- If mark–recapture methods continue to be used as a
ture of bottlenose dolphins is their highly dynamic basis for monitoring the population of bottlenose dol-
school membership, such a bias is unlikely to occur in phins in the Moray Firth there are some important con-
studies of this species. However, other cetaceans in siderations. Firstly, consistency in data collection pro-
which the composition and hence size of schools is cedures must be ensured. Hammond (1990b) found that
more stable (such as pilot and killer whales [Amos et the magnitude of increase in mark–recapture popula-
al. 1991, Bigg et al. 1990]) some schools (usually the tion estimates of humpback whales (Megaptera no-
larger) may to be more visible at sea than others and vaeangliae) in the Gulf of Maine was most likely
so underestimation of population abundance could oc- caused in part by a combination of an expansion of the
cur. area sampled and site specificity in the distribution of
individual humpbacks, resulting in a progressive re-
Trends in population size and the effectiveness of duction of the effects of heterogeneity of capture prob-
monitoring programs abilities. Furthermore, Hansen and Defran 1990 found
Although the current trend of the Moray Firth bot- that changes in the range of bottlenose dolphins near
tlenose dolphin population is uncertain, data on the the Californian coast lead to wide discrepancies be-
number of calves observed and the number of carcasses tween estimates. There is evidence of site specificity
recovered provide an indication of the levels of pop- among individual dolphins in the Moray Firth (Wilson
ulation change that might be expected. et al. 1997a) and minor changes in range (B. Wilson,
The figures given above for birth and mortality rates Aberdeen University, unpublished data). It is impor-
are crude and approximate; indeed, these vital rates are tant, therefore, that future survey coverage is sufficient
difficult to determine for cetacean populations (Wells to minimize the effects of this. This necessitates that
and Scott 1990). However, the collection of individual- the logistically more difficult surveys in the outer Mo-
based data via photoidentification does allow for their ray Firth and fringes of the population’s known range
estimation (Barlow 1990, Buckland 1990) and their be continued. Secondly, the precision of each annual
incorporation into models to estimate population estimate should be maintained or preferably increased
growth rates (Barlow and Clapham 1997). Data col- by increasing survey effort or through the efficiency
lection on Moray Firth bottlenose dolphins began in of data collection.
1989 and has continued uninterrupted since then. We However, it is also clear that a population changing
plan to apply the birth-interval approach of Barlow and in size very slowly would require considerable effort
Clapham (1997) to these data in the near future when for this to be established. Furthermore, one can never
our data set is sufficiently long to allow it. be statistically confident that a population is experi-
February 1999 ESTIMATING DOLPHIN ABUNDANCE AND TRENDS 299

encing no change. This leads to a paradox, whereby ulation size for capture–recapture data when capture prob-
stable or healthy populations at carrying capacity may abilities vary by time and individual animal. Biometrics
48:201–216.
require considerable resources in monitoring when they Curran, S., B. Wilson, and P. M. Thompson. 1996. Rec-
may actually need the least conservation effort. Un- ommendations for the sustainable management of the bot-
derstanding the power of survey techniques themselves tlenose dolphin population in the Moray Firth. Scottish
is therefore critical if limited resources for conserva- Natural Heritage Review Number 56. Battleby, UK.
Darroch, J. N. 1958. The multiple recapture census: I. es-
tion are to be targeted at the populations that are ac-
timation of a closed population. Biometrika 45:343–359.
tually in most need. Defran, R. H., G. M. Schultz, and D. W. Weller. 1990. A
Many coastal cetacean species are potentially threat- technique for the photographic identification and catalogu-
ened by anthropogenic activities and require the de- ing of dorsal fins of the bottlenose dolphin (Tursiops trun-
velopment of management strategies. The proposed catus). Reports of the International Whaling Commission.
Special Issue 12:53–56.
framework for estimating population size and assessing Evans, P. G. H. 1980. Cetaceans in British waters. Mammal
the likely time scale over which any trends may be Review 10:2–52.
determined is applicable to those species that possess Fairfield, C. P. 1990. Comparison of abundance estimation
natural markings sufficient for individual recognition. techniques for the western north Atlantic right whale (Eu-
Similarly, as the value of photoidentification is rec- balaena glacialis). Reports of the International Whaling
Commission. Special Issue 12:119–126.
ognized and is increasingly being applied to study other Friday, N. 1997. Evaluating photographic capture–recapture
taxa, from ungulates to fish, this framework may have estimates of abundance of north Atlantic humpback whales.
wider use. Thesis. University of Rhode Island, Providence, Rhode Is-
land, USA.
ACKNOWLEDGMENTS
Gerrodette, T. 1987. A power analysis for detecting trends.
We thank those who, often at short notice, made up the Ecology 68:1364–1372.
boat crews. The International Fund for Animal Welfare and Greenwood, P. J. 1980. Mating systems, philopatry and dis-
Jonathan Gordon gave us the use of the research yacht ‘‘Song persal in birds and mammals. Animal Behavior 28:1140–
of the Whale’’; thanks go to her crew, Richard McLanaghan, 1162.
Oliver Chappell, and Nicky Spurr. We thank Harry Ross and Grellier, K. In press. Bottlenose dolphins in the Sound of
Robert Reid of the Scottish Strandings Scheme for infor- Barra? A pilot study. Hebridean Naturalist.
mation on bottlenose dolphin carcasses. We also thank Martyn Hammond, P. S. 1986a. Line transect sampling of dolphin
Gorman, Paul Racey, and Hal Whitehead for comments on populations. Pages 251–279 in M. M. Bryden and R. J.
this work and Kate Grellier and Sofie Van Parijs for comments Harrison, editors. Research on dolphins. Oxford University
on the manuscript. This study was initiated with a small pro- Press, Oxford, UK.
ject grant from the British Ecological Society and subse- . 1986b. Estimating the size of naturally marked
quently supported by the Association for the Study of Animal whale populations using capture–recapture techniques. Re-
Behavior and the Greenpeace Environmental Trust. ports of the International Whaling Commission. Special
LITERATURE CITED Issue 8:253–282.
. 1987. Techniques for estimating the size of whale
Amos, B., J. Barrett, and G. A. Dover. 1991. Breeding be-
haviour of pilot whales revealed by DNA fingerprinting. populations. Symposia of the Zoological Society of London
Heredity 67:49–55. 58:225–245.
Barlow, J. 1990. A birth-interval model for estimating ce- . 1990a. Capturing whales on film—estimating ce-
tacean reproductive rates from resightings data. Reports of tacean population parameters from individual recognition
the International Whaling Commission. Special Issue 12: data. Mammal Review 20:17–22.
155–160. . 1990b. Heterogeneity in the Gulf of Maine? Esti-
Barlow, J., and P. J. Clapham. 1997. A new birth-interval mating humpback whale population size when capture
approach to estimating demographic parameters of hump- probabilities are not equal. Reports of the International
back whales. Ecology 78:535–546. Whaling Commission. Special Issue 12:135–139.
Begon, M. 1983. Abuses of mathematical techniques in ecol- . 1995. Estimating the abundance of marine mam-
ogy: applications of Jolly’s capture–recapture method. Oi- mals: a North Atlantic perspective. Pages 1–12 in A. S.
kos 40:155–158. Blix, L. Walløe, and Ø. Ulltang. Whales, seals, fish and
Bigg, M. A., P. F. Olesiuk, and G. M. Ellis. 1990. Social man. Developments in Marine Biology 4:1–12.
organization and genealogy of resident killer whales (Or- Hammond, P. S., S. A. Mizroch, and G. F. Donovan, editors.
cinus orca) in the coastal waters of British Columbia and 1990. Individual recognition of cetaceans: use of photo-
Washington State. Reports of the International Whaling identification and other techniques to estimate population
Commission. Special Issue 12:383–405. parameters. Reports of the International Whaling Com-
Buckland, S. T. 1990. Estimation of survival rates from sight- mission. Special Issue 12.
ings of individually identifiable whales. Reports of the In- Hammond, P. S., and P. M. Thompson. 1991. Minimum es-
ternational Whaling Commission. Special Issue 12:149– timate of the number of bottlenose dolphins (Tursiops trun-
153. catus) in the Moray Firth. Biological Conservation 56:79–
Buckland, S. T., D. R. Anderson, K. P. Burnham, and J. L. 88.
Laake. 1993. Distance sampling: estimating abundance of Hansen, L. 1990. California coastal bottlenose dolphins.
biological populations. Chapman and Hall, London, UK. Pages 403–420 in S. Leatherwood and R. R. Reeves, ed-
Calambokidis, J., J. C. Cubbage, G. H. Steiger, K. C. Bal- itors. The bottlenose dolphin. Academic Press, San Diego,
comb, and P. Bloedel. 1990. Population estimates of hump- California, USA.
back whales in the Gulf of the Farallones, California. Re- Hansen, L. J., and R. H. Defran. 1990. A comparison of
ports of the International Whaling Commission. Special photo–identification studies of California coastal bottle-
Issue 12:325–334. nose dolphins. Reports of the International Whaling Com-
Chao, A., S. M. Lee, and S. L. Jeng. 1992. Estimating pop- mission. Special Issue 12:101–104.
300 BEN WILSON ET AL. Ecological Applications
Vol. 9, No. 1

Janik, V. M., and P. M. Thompson. 1996. Changes in the Taylor, B. L., and T. Gerrodette. 1993. The uses of statistical
surfacing patterns of bottlenose dolphin in response to boat power in conservation biology: the vaquita and northern
traffic. Marine Mammal Science 12:597–601. spotted owl. Conservation Biology 7:489–500.
Kastelein, R. A., T. Dokter, and P. Zwart. 1990. The suckling Tolley, K. A., A. J. Read, R. S. Wells, K. W. Urian, M. D.
of a bottlenose dolphin calf (Tursiops truncatus) by a foster Scott, A. B. Irvine, and A. A. Hohn. 1995. Sexual di-
mother, and information on transverse birth bands. Aquatic morphism in wild bottlenose dolphins (Tursiops truncatus)
Mammals 16:134–138. from Sarasota, Florida. Journal of Mammalogy 76:1190–
Kayes, R. 1985. The decline of porpoises and dolphins in 1198.
the southern North Sea; a current status report. Number 14. Van Strien, A. J., R. Van De Pavert, D. Moss, T. J. Yates, C.
Political Ecology Research Group, Oxford, UK. A. M. Van Swaay, and P. Vos. 1997. The statistical power
Leatherwood, S., and R. R. Reeves. 1983. The Sierra Club of two butterfly monitoring schemes to detect trends. Jour-
handbook of whales and dolphins. Sierra Club Books, San nal of Applied Ecology 34:817–828.
Francisco, California, USA. Wells, R. S. 1991. The role of long-term study in under-
Liret, C., P. Allali, P. Creton, C. Guinet, and V. Ridoux. 1994. standing the social structure of a bottlenose dolphin com-
Foraging activity pattern of bottlenose dolphins, Tursiops munity. Pages 199–225 in K. Pryor and K. S. Norris. Dol-
truncatus, around Ile de Sein, Brittany, France, and its re- phin societies: discoveries and puzzles. University of Cal-
lations with some environmental parameters. Pages 188– ifornia Press, Oxford, California, USA.
191 in P. G. H. Evans, editor. European Research on Ce- Wells, R. S., A. B. Irvine, and M. Scott. 1980. The social
taceans 8:188–191. ecology of inshore odontocetes. Pages 263–317 in L. M.
Lockyer, C. H., and R. J. Morris. 1990. Some observations Herman, editor. Cetacean behavior: mechanisms and pro-
on wound healing and persistence of scars in Tursiops trun- cesses. Robert E. Krieger Publishing, Malabar, Florida,
catus. Reports of the International Whaling Commission. USA.
Special Issue 12:113–118. Wells, R. S., and M. D. Scott. 1990. Estimating bottlenose
Otis, D. L., K. P. Burnham, G. C. White, and D. R. Anderson. dolphin population parameters from individual identifica-
1978. Statistical inference from capture data on closed an- tion and capture–recapture techniques. Reports of the In-
imal populations. Wildlife Monographs. ternational Whaling Commission. Special Issue 12:407–
Payne, R., O. Brazier, E. M. Dorsey, J. S. Perkins, V. J. 415.
Rowntree, and A. Titus. 1983. External features in south- Wells, R. S., M. D. Scott, and A. B. Irvine. 1987. The social
ern right whales (Eubalaena australis) and their use in iden- structure of free-ranging bottlenose dolphins. Pages 247–
tifying individuals. Pages 371–445 in R. Payne. Commu- 305 in Current mammalogy. Volume 1. H. Genoways, ed-
nication and behavior of whales. Reports of the American itor. Plenum, New York, New York, USA.
Association for the Advancement of Science Selected Sym- Whitehead, H. 1996. Babysitting, dive synchrony, and in-
posium 76, Westview Press, Boulder, Colorado, USA. dications of alloparental care in sperm whales. Behavioural
Rexstad, E., and K. Burnham. 1991. User’s guide for inter- Ecology and Sociobiology 38:237–244.
active program CAPTURE. Colorado Cooperative Fish and Williams, J. A., S. M. Dawson, and E. Slooten. 1993. The
Wildlife Research Unit, Colorado State University, Fort abundance and distribution of bottlenosed dolphins (Tur-
Collins, Colorado 80523 USA. siops truncatus) in Doubtful Sound, New Zealand. Cana-
Scott, M. D., R. S. Wells, A. B. Irvine, and B. R. Mate. 1990a. dian Journal of Zoology 71:2080–2088.
Tagging and marking studies on small cetaceans. Pages Wilson, B. 1995. The ecology of bottlenose dolphins in the
489–514 in S. Leatherwood and R. R. Reeves, editors. The Moray Firth, Scotland: a population at the northern extreme
bottlenose dolphin. Academic Press, San Diego, California, of the species’ range. Dissertation. University of Aberdeen,
USA. UK.
Scott, M. D., R. S. Wells, and A. R. Irvine. 1990b. A long- Wilson, B., P. M. Thompson, and P. S. Hammond. 1997a.
term study of bottlenose dolphins on the west coast of Habitat use by bottlenose dolphins: seasonal distribution
Florida. Pages 235–244 in S. Leatherwood and R. R. and stratified movement patterns in the Moray Firth, Scot-
Reeves, editors. The bottlenose dolphin. Academic Press, land. Journal of Applied Ecology 34:1365–1374.
San Diego, California, USA. Wilson, B., P. M. Thompson, and P. S. Hammond. 1997b.
Scottish Natural Heritage. 1995. Natura 2000: a guide to the Skin lesions and physical deformities in bottlenose dol-
1992 EC Habitats directive in Scotland’s Marine Environ- phins in the Moray Firth: population prevalence and age–
ment. Scottish Natural Heritage, Perth, UK. sex differences. Ambio 26:243–247.
Seber, G. A. F. 1982. The estimation of animal abundance Würsig, B., and R. A. Jefferson. 1990. Methods of photo-
and related parameters. MacMillan, New York, New York, identification for small cetaceans. Reports of the Interna-
USA. Second edition. tional Whaling Commission. Special Issue 12:42–43.