THE FORAGING HABITS OF BOTTLENOSE DOLPHINS: INSIGHTS INTO
TEMPORAL, DEMOGRAPHIC, AND INDIVIDUAL VARIATION
By
Samuel Rossman

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Zoology–Doctor of Philosophy
Ecology, Evolutionary Biology and Behavior–Dual Major
2014

ABSTRACT
THE FORAGING HABITS OF BOTTLENOSE DOLPHINS: INSIGHTS INTO
TEMPORAL, DEMOGRAPHIC, AND INDIVIDUAL VARIATION
By
Samuel Rossman
Bottlenose dolphins (Tursiops truncatus) are large apex carnivores that often live
in close proximity to metropolitan areas. As a result, they experience disturbances such
as habitat alteration, competition with humans for food resources, and anthropogenic
injury and mortality. Understanding variation in foraging habits of bottlenose dolphins
over time and within populations reflects how these predators use and respond to
changes in habitat and prey availability. Stable isotope analysis provides powerful
insight into the foraging habitat and tropic level (foraging habits) of top predators.
Carbon isotope values (δ13C) differentiate between seagrass (c.a. 12‰) and
nonseagrass (phytoplankton and mangrove, < -15 ‰) based food webs in coastal
estuaries and nitrogen isotope values (δ15N) increase with trophic level. I evaluated
temporal, demographic and individual variation in the foraging habits of bottlenose
dolphins resident to Sarasota Bay, Florida, using stable isotope analysis.
Chapter one examines bottlenose dolphin response to a series of disturbances.
From 1991 to 2010, bottlenose dolphin showed a significant decrease in δ 13C values
and a significant increase in δ15N values, trends likely related to changes in habitat use
after a 1995 commercial net fishing ban. Across a larger time period, bottlenose dolphin
δ15N values significantly increased from 1944 to 1990, then data became variable. This
is likely related to an initial increase and subsequent decline in anthropogenic nitrogen
loading to Sarasota Bay which contributes 15N enriched nitrogen to the system. These

data suggest that environmental legislation is effective in remediating anthropogenic
nutrient excess in food webs.
The second chapter examines the foraging habits of three demographic groups:
male, female and juvenile bottlenose dolphins. Isotope values of common bottlenose
dolphin prey fish were used to inform a Bayesian mass-balance model to estimate the
abundance of certain groups to the diet of males, females, and juveniles. Low trophic
level, seagrass associated prey are important for all demographic groups. Bayesian
standard ellipses identified the degree of variability in foraging habits. The ellipse center
identifies the mean isotope value for a demographic group and the ellipse size and
shape determined from a covariance matrix indicates the degree of variation in the
demographic group's foraging habits. While males show relatively low variability, the
female standard ellipse indicates large variability in foraging habitat and trophic level.
Female bottlenose dolphin foraging habits are examined further in chapter three,
which investigates individual specialization. Individual specialization results from each
individual using a subset of the resources used by the population. Multiple chronological
isotope samples from the same individual allow the partitioning of isotopic variation
(total niche width, TNW) into a within individual component (WIC, high in generalists)
and a between individual component (BIC, high in individual specialists). Isotope
analysis of sequential growth layers from teeth indicate that BIC constitutes 60% of
TNW for δ15N and 88% of TNW for δ13C, indicative of a large degree of individual
specialization. The incidence of individual specialization likely results from processes
associated with learning, social structure, and intraspecific competition.

ACKNOWLEDGMENTS

While my name alone appears on the author line this work would not have been
possible without the numerous scientists who are my collaborators, mentors, and
friends. I would like to offer a special thanks to:
My adviser, Peggy Ostrom, your confidence in me as a scientist was both humbling and
inspiring and without it this dissertation would not exist. Thank you.
Nélio Barros, whose passion for dolphin biology and conservation planted the seeds
which grew into this dissertation. While he is not here to see the fruits of his efforts I am
honored that I could contribute to his legacy.
Randall Wells, who has been a wonderful mentor demonstrating excellence as a
scientist, leader, and conservationist.
Hasand Gandhi, who was always willing to give of his time and effort and whose
patience with an inexperienced graduate student knew no bounds.
Dick Hill, whose mentorship taught me the power of ideas and that the merit of a scholar
derives from the rigor their philosophy.
Craig Sticker, who taught me that good friends make the best collaborators.
Josh Haslun, Briana Hauff and Kateri Salk who reminded me science is fun.
I would also like to thank the countless others who have given of their time and effort to
make this work possible: Nathaniel Ostrom, Brian Mauer, Elise Zipkin, Elizabeth Berens
McCabe, Aleta Hohn, Anne Wiley, Ramona Beckman, Aaron Barleycorn, Brian Balmer,
Jason Allen, Jenn Yordy, Gretchen Lovewell and everyone involved in Mote Marine
Lab’s Stranding Investigations Program, the volunteers of the Sarasota Dolphin
Research Program.
On a personal note:
A special thanks to my mom and dad, Bruce and Linda Rossman, whose lack of
concern for my future is the greatest vote of confidence I have ever received.
And Heather whose love and support didn’t just help get me through, it gave me a
reason to do it.

iv

A note on authorship:
I have had the fortunate pleasure to collaborate and publish with a large network of
scientists. The first chapter of my dissertation is published in Marine Mammal Science
with coauthors: Nélio B. Barros, Peggy H. Ostrom, Craig A. Stricker, Aleta A. Hohn,
Hasand Gandhi, and Randall S. Wells. Chapter two is also published in Marine Mammal
Science with coauthors: Elizabeth Berens McCabe, Nélio B. Barros, Hasand Gandhi,
Peggy H. Ostrom, Craig A. Stricker, and Randall S. Wells. When chapter three is
submitted for publication coauthors will be Megan Stolen, Nélio B. Barros, Hasand
Gandhi, Peggy H. Ostrom, Craig A. Stricker, and Randall S. Wells.

v

TABLE OF CONTENTS

LIST OF TABLES…………………………………...…………………………………………viii
LIST OF FIGURES……………………………………………………………………………..ix
CHAPTER 1
RETROSPECTIVE ANALYSIS OF BOTTLENOSE DOLPHIN FORAGING: A LEGACY
OF ANTHROPOGENIC ECOSYSTEM DISTURBANCE…………………………………...1
ABSTRACT………………………………………………………………………………1
INTRODUCTION………………………………………………………………………..2
METHODS……………………………………………………………………………….5
Sample Site and Acquisition…………………………………………………...5
Sample Preparation……………………………………………………………..8
Statistical Analysis………………………………………………………………9
RESULTS………………………………………………………………………………10
DISCUSSION…………………………………………………………………………..15
LITERATURE CITED………………………………………………………………….21
CHAPTER 2
FORAGING HABITS IN A GENERALIST PREDATOR: SEX AND AGE INFLUENCE
HABITAT SELECTION AND RESOURCE USE AMONG BOTTLENOSE
DOLPHINS……………………………………………………………………………………..26
ABSTRACT…………………………………………………………………………….26
INTRODUCTION………………………………………………………………………27
METHODS……………………………………………………………………………...30
Sample Site and Acquisition………………………………………………….30
Sample Preparation……………………………………………………………31
Statistical Analysis……………………………………………………………..31
RESULTS………………………………………………………………………………34
DISCUSSION…………………………………………………………………………..40
LITERATURE CITED………………………………………………………………….47
CHAPTER 3
INDIVIDUAL SPECIALIZATION IN THE FORAGING HABITS OF FEMALE
BOTTLENOSE DOLPHINS LIVING IN A TROPHICALLY DIVERSE AND HABITAT
RICH ESTUARY……………………………………………………………………………….52
ABSTRACT..…………………………………………………………………………...52
INTRODUCTION……………………………………………………………………....53
METHODS……………………………………………………………………………...57
Sample Site and Acquisition………………………………………………….57
Sample Preparation……………………………………………………………57
Statistical Analysis…………………………………………………………..…58
RESULTS………………………………………………………………………………62
vi

DISCUSSION…………………………………………………………………………..70
LITERATURE CITED………………………………………………………………….77

vii

LIST OF TABLES
Table 1. Bottlenose dolphin prey species δ13C and δ15N values (means ± standard
deviation), the number of samples used to determine the average isotope value (n)
dietary indicators, SCA and DNA and prey cluster designation. SCA and DNA are the
fractional contribution of each species to the diet of bottlenose dolphins as determined
by stomach content analysis or molecular prey detection (Barros and Wells 1998,
Berens McCabe et al. 2010, Dunshea et al. 2013). Prey cluster refers to the cluster
designation as determined by K means cluster analysis. Asterisk indicates instances
where Dunshea et al. 2013 report genus sp., the genus and species are provided
here……………………………………………………………………………………………...35
Table 2. ANOVA results for competing models where δ15N and δ13C values are a
function of individual and/or age class. Deviance information criterion (DIC) values are
used for model selection. The lowest DIC value identifies the model with the highest
predictive capacity. The R2 value indicates the amount of variation in the dependent
variable explained by the model. Our Bayesian treatment of the data provides a 95% CI
for the R2 denoted by a lower bound (CI: 2.5%) and an upper bound (CI: 97.5%)…….63

viii

LIST OF FIGURES

Figure 1. Distribution of habitats used by bottlenose dolphins resident to Sarasota Bay,
Florida…………………………………………………………………………………………….7
Figure 2. δ13C, δ34S, and δ15N values of bottlenose dolphins and percentage of
observed feedings in seagrass habitat. Percentage of observed feedings in seagrass
habitat are from Barros and Wells (1998). Lines indicate the best-fit trend between
percent of feedings observed in seagrass and isotope values…………………………...11
Figure 3. δ13C, δ34S, and δ15N values of muscle from 1991-2010. Open symbols
indicate calves <2.5 yr of age while closed symbols are individuals >2.5 yr old. Lines
indicate the best-fit trend between year and isotope values………………………………13
Figure 4. δ13C values of collagen isolated from the crown tip dating from 1944 to 2007.
Closed circles represent the time period of increasing nutrient loading (1944–1989) and
the open circles represent the time period of nutrient loading mitigation (1990–
2007)…………………………………………………………………………………………….14
Figure 5. δ15N values of collagen isolated from the crown tip dating from 1944 to 2007.
Closed circles represent the time period of increasing nutrient loading (1944–1989), the
open circles represent the time period of nutrient loading mitigation (1990–2007) and
triangles show the human population size of Sarasota and Manatee Counties
summed……………………………………………………………………………………….14
Figure 6. δ13C and δ15N values of prey clusters and bottlenose dolphin skin from
different demographic groups including males (>6 yr old), females (>6 yr old) and
juveniles (2 to 6 yr old). Prey clusters show mean with error bars depicting the standard
deviation. Prey clusters were adjusted for trophic enrichment (cluster average + TEF).
TEF estimates derive from this study (δ13C: 0.0‰, δ15N: 2.0‰), not the literature…38
Figure 7. SIAR results for bottlenose dolphin demographic groups: juveniles (J; 2-6 yr
old males and females), females, (F; >6 yr old) and males (M; >6 yr old). Horizontal
lines within boxes represent mean dietary contributions of each prey cluster to the
demographic group. Boxes enclose the 50% credible interval and vertical lines with end
caps depicting the 95% credible interval. Cluster 1, contains snook, ladyfish, and spot.
Cluster 2 contains seatrout and crevalle jack. Cluster 3 contains toadfish, seagrass
pinfish, pigfish, red drum, lane snapper, sheepshead, mojarra, and striped mullet.
Cluster 4 contains nonseagrass prey fish, threadfin herring, nonseagrass pinfish, and
scaled sardine.…………………………………………………………………………………39

ix

Figure 8. Carbon and nitrogen isotope values of prey clusters and standard ellipses
controlled for small sample size (SEAc) for female, male and juvenile bottlenose
dolphins. The upper right inset table provides SEAc area (‰2), mean Bayesian standard
ellipse area (SEAb) (‰2), and the lower (2.5%) and upper (97.5%) bounds for the 95%
credible interval around the mean SEAb for males (M; >6 yr old), females (F; >6 yr old),
and juveniles (J; 2 to 6 yr old)………………………………………………………………..40
Figure 9. The frequency of an isotope value over the lifetime of three individuals is
depicted by red, blue and yellow curves in panels A and B. Arrows at the bottom of the
curves reflect the degree of variation in isotope value over the lifetime of an individual
also termed the within individual component (WIC) of the population’s isotope variation.
Arrows between curves reflect the between individual component (BIC) of the isotopic
variation in the population. A) Isotope frequency curves for individual specialists, each
with a narrow range in isotope values. In this case BIC is larger than WIC. B) Isotope
frequency curves for individual generalists, each with a wide range in isotope values. In
this case BIC is much smaller than WIC. C) Microdrilled tracks (green lines) imposed
over a photograph of a bucco-lingual longitudinal section of a dolphin tooth. The table
shows age class for each track and hypothetical δ13C data for two individuals. D)
Histograms for the hypothetical isotope data in panel C. The large BIC compared to
WIC indicates that these dolphins are individual specialists……………………………...55
Figure 10. δ15N values from bottlenose dolphin teeth for five age classes covering the
lifetime of the individual. Age class >15 is adult. Age classes were determined from
bottlenose dolphin tooth ontogeny. Individuals are differentiated by color and Sarasota
Dolphin Research Program identification codes given in the figure legend. The inset
shows average δ15N values for each age class (closed triangles) and associated
credible intervals (bars) from Bayesian ANOVA……………………………………………64
Figure 11. δ13C values from bottlenose dolphin teeth for five age classes covering the
lifetime of the individual. Age class >15 is adult. Age classes were determined from
bottlenose dolphin tooth ontogeny. Individuals are differentiated by color and Sarasota
Dolphin Research Program identification codes given in the figure legend……………65
Figure 12. An individual’s isotope value as and adult plotted against its isotope value as
a calf, A) carbon isotope values and B) nitrogen isotope values. Calf δ13C or δ15N value
is the average of age classes 1-2 yr and 2-4 yr and adult isotope values is age class >
15 yr. Individuals are differentiated by color (same color scheme as earlier figures) and
Sarasota Dolphin Research Program identification codes given in the figure legend.
Regression equations are shown in the bottom right of each panel. For δ 15N the 95%
credible interval for the slope was 0.38-1.06 and for the intercept it was -0.56-4.44. For
δ13C the 95% credible interval for the slope was 0.72-1.14 and for the intercept it was 0.86-1.34. The 95% credible intervals for the slopes are depicted on the graphs as grey
dashed lines. R2 values and their 95% confidence intervals are shown in the lower right
of each graph. The R2 value documents the proportion of the variance in adult isotope
values that can be explained by calf isotope values……………………………………….66

x

Figure 13. Isotope standard ellipses representing variation in δ13C and δ15N values for
each individual over their lifetime. Mean δ13C and δ15N values for each individual define
the center of the standard ellipse and the shape and orientation are determined by a
covariance matrix. Individuals appear in three non-overlapping groups. Compared to the
population mean (-10.5‰ and11.1‰ for δ13C and δ15N respectively) group one is
defined by high δ15N values and low δ13C values, group two has low δ15N values and
high δ13C values, and group three possess low δ15N values and intermediate δ13C
values. Individuals are differentiated by color using the same scheme as in earlier
figures. Sarasota Dolphin Research Program identification codes given in the figure
legend…………………………………………………………………………………………...68
Figure 14. Maps of Sarasota Bay depicting habitat distribution (left) and observational
standard ellipses generated from the latitude and longitude data from sightings (right).
Data for habitat use was obtained from the Southwest Florida Water Management
District (2010). Inset depicts location of Sarasota Bay in relation to the state of Florida.
Colors represent different habitats in Sarasota Bay as defined in the figure legend.
Individuals in the observational standard ellipse maps are differentiated by color (same
color scheme as earlier figures) and Sarasota Dolphin Research Program identification
codes given in the figure legend…………………………………………………………….69

xi

CHAPTER 1
RETROSPECTIVE ANALYSIS OF BOTTLENOSE DOLPHIN FORAGING: A LEGACY
OF ANTHROPOGENIC ECOSYSTEM DISTURBANCE
ABSTRACT
We used stable isotope analysis to investigate the foraging ecology of coastal
bottlenose dolphins (Tursiops truncatus) in relation to a series of anthropogenic
disturbances. We first demonstrated that stable isotopes are a faithful indicator of
habitat use by comparing muscle isotope values to behavioral foraging data from the
same individuals. δ13C values increased, while δ34S and δ15N values decreased, with
the percentage of feeding observations in seagrass habitat. We then utilized stable
isotope values of muscle to assess temporal variation in foraging habitat from 1991 to
2010 and collagen from tooth crown tips to assess the time period 1944 to 2007. From
1991 to 2010 δ13C values of muscle decreased while δ34S values increased, indicating
reduced utilization of seagrass habitat. From 1944 to 1989 δ13C values of the crown tip
declined significantly, likely due to a reduction in the coverage of seagrass habitat, and
δ15N values significantly increased, a trend we attribute to nutrient loading from a rapidly
increasing human population. Our results demonstrate the utility of using marine
mammal foraging habits to retrospectively assess the extent to which anthropogenic
disturbance impacts coastal food webs.

1

INTRODUCTION
Coastal marine ecosystems can be impacted by a multitude of anthropogenic
disturbances including pollution, habitat alteration, overfishing, and climate disruption.
Ecosystem change is often difficult to document because disturbances occur across
multiple temporal scales and the vast majority of ecological studies span only a few
years (Knowlton and Jackson 2008). Because of their high trophic position within the
food web and their typically long lifespan, large marine predators integrate disturbance
across multiple trophic levels and are of particular interest as sentinels of ecosystem
change (Sergio et al. 2008). However, collection of consistent, continuous data over
large spatial scales is expensive, requires extensive effort, and is generally only
available from data associated with commercial fishing. While fisheries catch data have
documented profound global declines in predatory fish populations (Myers and Worm
2003), causal mechanisms are difficult to assess. Declining fish stocks are often
attributed to intense fishing pressure, but other factors contribute to the diversity and
abundance of marine organisms. For example, changes in foraging ecology (trophic
level, foraging location) have been implicated as important regulators of population size
and distribution of marine organisms (Emslie and Patterson 2007, Österblom et al.
2008). Because top predators rely on energy transfers from each trophic step, they are
sensitive to perturbations at any point in the food web. Thus, changes in population
size, distribution, or foraging ecology of top predators are indicators of ecosystem
change.

2

The study of marine predator foraging ecology traditionally involves stomach
content analysis or direct observation of feeding behavior (Barros and Wells 1998).
Stomach content analysis usually requires deceased animals and stomachs are often
empty and may be influenced by differential digestion and/or cause of death. Direct
observation of prey consumption is informative for predators who capture or consume
prey at or near the water’s surface but can be time consuming, expensive and may be
biased toward prey living in shallow water or high in the water column. Stable isotope
analysis of selected tissues from predators is a relatively inexpensive alternative to
traditional diet analysis and has a long-standing history in food web studies (Peterson
and Fry 1987). Carbon isotope values (δ13C) are valuable because they differ among
forms of primary production and are passed up the food web to consumers. Thus, they
distinguish organisms from habitats with isotopically unique primary producers
(Peterson and Fry 1987). Similar to carbon, sulfur isotope values (δ34S) distinguish
consumers who derive nutrition from rooted marine plants vs. phytoplankton (Connolly
et al. 2004). Nitrogen isotopes (δ15N) have been extensively used in food web studies
because of the strong positive correlation with trophic level (Peterson and Fry 1987,
Harrigan et al. 1989, Gould et al. 1997).
Isotope analysis of different tissues aids in the assessment of foraging over
different time scales or through life history milestones because animal tissues turn over
at varying rates. Blood plasma records foraging over a period of days (Hobson 1999)
and muscle integrates over many months for large mammals (Sponheimer et al. 2006).
Protein from tooth dentin remains metabolically inert once formed and, thus, records
foraging at the time of synthesis (Knoff et al. 2008). The teeth of marine mammals, such

3

as bottlenose dolphins (Tursiops truncatus) have annual growth layers (Hohn et al.
1989) in which collagen can be examined to assess isotopic variation of an individual
over its lifetime (Mendes et al. 2006, Newsome et al. 2009). The isotope value of
collagen in well-preserved tooth and bone can provide a historical record of foraging
(Hobson and Sease 1998).
Using bone collagen samples from North Sea harbor porpoises (Phocoena
phocoena), Christensen and Richardson (2008) attributed a decrease in δ 15N to the
declining availability of high trophic level prey from 1848 to present. In a study of dentin
from mid-Atlantic bottlenose dolphin teeth, Walker et al. (1999) found no significant
increase in δ15N from the 1880’s to the 1980’s suggesting little to no change in dolphin
diet over this time period. Unfortunately, information on parameters such as life history,
geographic range, migration, sex, and age are often limited for historical samples or
those from stranded animals of unknown geographic origin; however, any one of these
factors may confound the ability to attribute changes in foraging to disturbance. Data on
any subset of these parameters may offer the possibility to partition variance in isotope
values and control for factors unrelated to ecological disturbance.
Bottlenose dolphins from Sarasota Bay (SB), Florida, have a well established,
long-term, residency within SB (Wells et al. 1987, Wells 2003) and thus provide an ideal
study population for probing long-term effects of anthropogenic disturbance on foraging
ecology. In addition, the Sarasota Dolphin Research Program has collected data on vital
rates and a number of other parameters for a multi-generational resident community of
ca. 160 dolphins since 1970 (Wells 2009, Bowen et al. 2010). These dolphins have
experienced numerous anthropogenic disturbances likely to influence various aspects of

4

their ecology. For example, since about 1950 Sarasota Bay has undergone large-scale
changes through coastal construction that altered seagrass coverage. Furthermore, the
human population surrounding the bay has tripled since 1970 with associated changes
in commercial and recreational fishing pressure, and nutrient inputs (Sarasota Bay
National Estuary Program 2010, U.S. Census Bureau 2010). As predators that consume
a wide variety of prey items across numerous estuarine and coastal marine habitats
(Barros and Wells 1998), bottlenose dolphins are likely influenced by ecosystem
disturbance.
The purpose of this study was to examine the role of anthropogenic disturbance
on the foraging ecology of bottlenose dolphins resident to SB. We first explore the
relationship between isotope values and observational foraging data (percentage of
feedings in seagrass habitat). We then examined muscle tissue from dolphins stranded
between 1991 and 2010 to assess changes in foraging ecology (habitat and trophic
level) during a period including the implementation of a 1995 state-wide commercial net
fishing ban. To extend our retrospective analysis of foraging ecology to six decades
(1944-2007), we examined the isotope values of dentin from the crown tip (external 2-3
mm of the crown) of teeth collected from known SB stranded or live dolphins sampled
during health assessments (Hohn et al. 1989).

METHODS
Sample Site and Acquisition
SB is a complex of shallow bays (< 4m deep, 40 km long) on Florida’s central west
coast, communicating with the Gulf of Mexico through narrow deep passes separating a

5

series of narrow barrier islands. The study area encompasses a diverse array of
habitats including seagrass meadows, mangroves, shorelines and open bay (Fig. 1). All
samples were obtained from a long-term resident community of ca. 160 bottlenose
dolphins. As part of the Sarasota Dolphin Research Program, most dolphins from SB
are identifiable and of known age and sex (Wells 2003, 2009). Some of the dolphins
included in this study were also reported on in Barros and Wells (1998). Thus, we were
able to compare isotope values from this study to the percentage of observed feedings
in seagrass habitat, documented in Barros and Wells (1998), for the same individuals.
Observational data on seagrass foraging and isotope values of muscle tissue are
reported for six individuals, identified by freeze brand number (FB): FB19, FB21, FB31,
FB41, FB57, FB98. Four of the individuals used for observation-isotope comparison
died in 1991, the other two stranded in 1994 and 1996. Thus, all but one individual died
prior to the net ban.

6

Figure 1. Distribution of habitats used by bottlenose dolphins resident to Sarasota
Bay, Florida.

Samples of muscle tissue from SB individuals were obtained from stranded,
deceased dolphins (n = 41) recovered and sampled by Mote Marine Laboratory’s
Stranding Investigations Program during 1991 - 2010. Tooth samples were obtained
from live animals under local anesthesia during brief capture-release health
assessments (n = 27) (Hohn et al. 1989) or from stranded, deceased dolphins (19842009) (n = 40, total for crown tip data set n = 67).

7

Sample Preparation
Muscle tissue was freeze-dried, lipid extracted, and homogenized to a fine powder in a
ball and capsule amalgamator (Crescent Industries). Aliquots (1 mg) were transferred to
tin capsules for stable carbon and nitrogen isotope analysis. Stable sulfur isotopes were
analyzed separately by weighing 6 mg of sample into tin capsules amended with 2 mg
of vanadium pentoxide.
The external 2-3 mm of the crown of the tooth (crown tip) was used for isotope
analysis. This material is formed prior to one year of age during gestation and early
development and is therefore an indicator of maternal diet. Ages used in this study were
obtained from dolphins of known age via observation or quantification of growth layer
groups (GLG) in teeth (Hohn et al. 1989).
The crown was separated using a Dremel tool and demineralized in HCl (1N) for
72 hours leaving behind a collagenous replica of the tooth. The replica was then dried,
and lipid extracted. We were able to extract the oldest material, the crown tip, by
sectioning 2-3 mm of collagen from the apex of the crown. Collagen samples (1 mg)
were transferred to tin cups for δ13C and δ15N analysis. Insufficient mass precluded
sulfur isotope analysis.
Aliquots of muscle and tooth collagen were analyzed for stable carbon and
nitrogen isotopic composition using an elemental analyzer (Eurovector) interfaced to an
Isoprime mass spectrometer (Elementar). Samples were analyzed for sulfur isotopic
composition using an elemental analyzer (Costech Analytical) interfaced to a mass
spectrometer (Thermo-Finnegan DeltaPlus XP). Isotope values are expressed as:

8

δX = {(Rsample / Rstandard ) – 1} x 1,000

Where X represents 13C, 34S, or 15N and R represents the abundance ratios:
34S/32S,

13C/12C,

or 15N/14N respectively. In-house standards used for δ13C and δ15N were

calibrated with respect to international standards VPDB and air, respectively. NBS127
(21.1‰) and IAEA-SO-6 (−34.05‰) were used to normalize δ34S data to V-CDT. Inhouse standard precision is +/- 0.2‰ for δ13C, δ34S and δ15N values.

Statistical Analysis
All data were approximately normal and homoscedastic. To test the relationship
between isotope values of muscle (δ13C, δ34S or δ15N) and the percentage of observed
feedings in seagrass, we used a linear regression model and a random bivariate
resampling for 1,000 permutations. For muscle tissue samples collected from 1991 to
2010 we used a general linear model (GLM) with δ13C, δ34S and δ15N as dependent
variables and age, sex, weaning status, season and year as independent variables.
While most calves are nutritionally weaned at 18 to 20 months (Perrin and Reilly 1984),
calves with ages up to 2.5 years were grouped as pre-weaned to minimize the
contribution of a maternal isotope signal. Thus the variable “weaning/sex” consists of
dolphins divided into classes of male (n = 9), female (n = 25) or pre-weaned calf (n = 8).
For the variable “season”, dolphin isotope data were grouped by seasons based on the
calendar date of stranding: spring (21 April-6 June, n = 8), summer (21 June-20
September, n = 24), fall (21 September-20 December, n = 3) and winter (21 December21 April, n =6). Categorical variables (sex/weaning or season) with a significant effect

9

on isotope value were further examined for differences between groups with Tukey's
post-hoc pair wise comparisons.
For teeth collected from live animals, the date of birth was back-calculated from
the age at date of collection. Isotope values from the crown tip reflect the diet of
bottlenose dolphins prior to age one. We used an analysis of covariance (ANCOVA) in
which δ13C and δ15N values were modeled with a continuous variable, “year,” and a
categorical variable, nutrient loading, separating two time periods that markedly differed
with regard to nutrient loading (prior to and after 1989) (Tomasko et al. 2005). Carbon
isotope values from the crown tip were corrected 0.2‰ per decade to account for
changes in the δ13C of atmospheric CO2 due to the burning of fossil fuels (Suess effect)
(Quay et al. 2003). There should be no influence of sex or age on crown tip isotope
values as the crown tip is formed prior to one year of age and exclusively influenced by
maternal foraging ecology. Statistical tests were conducted in R 2.14 (R Foundation for
Statistical Computing 2011).

RESULTS
Carbon isotope values of muscle indicated a significant positive regression with
percentage of observed foraging in seagrass habitat (Fig. 2A; n = 6, regression with
bivariate resampling, r2 = 0.807, P < 0.001); δ15N and δ34S values were inversely related
to observed foraging in seagrass habitat (Fig. 2B, 2C, n = 6, regression with bivariate
resampling, δ34S: r2=0.836, δ15N: r2 = 0.952, δ34S and δ15N: P < 0.001).

10

Figure 2. δ13C, δ34S, and δ15N values of bottlenose dolphins and percentage of
observed feedings in seagrass habitat. Percentage of observed feedings in seagrass
habitat are from Barros and Wells (1998). Lines indicate the best-fit trend between
percent of feedings observed in seagrass and isotope values.

11

For muscle samples collected from strandings (1991 to 2010), the results of the
GLM model explained a significant portion of the variance for carbon and nitrogen
isotope values but not sulfur (δ13C : n = 41, F5, 35 = 4.17, P = 0.004, δ34S: n=39, F5, 33 =
1.81, P = 0.1383, δ15N: n = 41 F5, 35 = 6.24, P < 0.001,). Age was not a significant factor
for any of the three isotope values (δ13C: t35 = 1.14, P = 0.29, δ34S: t33 = 1.12, P = 0.23,
δ15N: t35 = 1.08, P = 0.29). Weaning/sex was a significant determinant of δ13C (F2, 38 =
6.85, P = 0.002) and δ15N (F2, 38 = 15.05, P < 0.001) but not δ34S (F2, 36 = 0.62, P =
0.544). Pre-weaned calves had higher δ13C values compared to males but did not
significantly differ from females (pre-weaned vs. males, t35 = 2.33, P = 0.026, preweaned vs. females t35 = 1.71, P = 0.095). Pre-weaned calves had significantly higher
nitrogen isotope values compared to both males and females (pre-weaned vs. males, t35
= 2.61, P = 0.013, pre-weaned vs. females, t35 = 4.04, P < 0.001). Males and females
were not significantly different for carbon or nitrogen isotope values (δ13C: t35 = 0.78, P
= 0.710, δ15N: t35 = 1.48, P = 0.310). Season was not significant for any of the three
isotopes (δ13C: F3, 37 = 0.73, P = 0.543, δ34S: F3, 35 = 1.12, P = 0.353, δ15N: F3, 37 = 0.03,
P = 0.993). Carbon isotope values significantly decreased as a function of year, while
sulfur isotope values significantly increased with year (δ13C: Fig. 3A, t35 = 2.32, P =
0.026, δ34S: Fig. 3B, t33 = 2.63, P = 0.013). Nitrogen isotope values showed no
significant trend with year (Fig. 3C, t35 = 0.50, P = 0.62).

12

Figure 3. δ13C, δ34S, and δ15N values of muscle from 1991-2010. Open symbols
indicate calves <2.5 yr of age while closed symbols are individuals >2.5 yr old. Lines
indicate the best-fit trend between year and isotope values.

13

Figure 4. δ13C values of collagen isolated from the crown tip dating from 1944 to 2007.
Closed circles represent the time period of increasing nutrient loading (1944–1989) and
the open circles represent the time period of nutrient loading mitigation (1990–2007).

Figure 5. δ15N values of collagen isolated from the crown tip dating from 1944 to 2007.
Closed circles represent the time period of increasing nutrient loading (1944–1989), the
open circles represent the time period of nutrient loading mitigation (1990–2007) and
triangles show the human population size of Sarasota and Manatee Counties summed.

14

For data from the tooth crown tip spanning the time period 1944-2007 an
ANCOVA for both δ13C and δ15N explained a significant portion of the variance (δ13C :
Fig. 4, n = 67, F2, 64 = 4.97, P = 0.001, δ15N: Fig. 5, n = 67, F2, 64 = 10.68, P < 0.001). For
carbon and nitrogen isotope values both year and nutrient loading had a significant
effect in the ANCOVA model (δ13C: Fig. 4, year: t64 = 3.05, P = 0.003, nutrient loading:
t64 = 2.72, P = 0.008, δ15N: Fig. 5, year: t64 = 4.38, P < 0.001, nutrient loading: t64 = 2.04,
P = 0.045).

DISCUSSION
Our multi-decadal study of bottlenose dolphin foraging ecology documents large
temporal variation in the δ13C, δ15N and δ34S values of individuals resident to SB.
Because our study population emphasizes use of inshore bay, sound, and estuary
habitats over the open waters of the Gulf of Mexico (Wells et al. 1987, Wells 2003),
isotopic differences derive primarily from variation in foraging habits within SB. Isotopic
variation is useful because it can provide insight into the ecology of organisms and
historical changes in environmental conditions (Newsome et al. 2010) particularly in
near-shore environments. Carbon isotope values have long been used to distinguish
seagrass from non-seagrass based food webs (Peterson and Fry 1987, Harrigan et al.
1989). Similarly, δ34S values of consumers frequenting nearshore seagrass habitats are
unique compared to those utilizing non-seagrass habitat (Barros et al. 2010). This is
because sulfur isotopes distinguish rooted and non-rooted phototrophs and these
differences are passed up the food web. The uniquely low δ 34S values of rooted plants,
such as seagrass, derive from the incorporation of

15

34S-depleted

sulfides, common in

reducing sediments (Brunner and Bernasconi 2005, Connolly et al. 2004, Oakes and
Connolly 2004). However, we cannot rule out the possibility that low δ34S values are
influenced by the transfer of sulfur from the terrestrial environment (δ34S value
approximately 0‰) into estuaries where it is taken up by benthic plants and
incorporated into the food web (Trust and Fry 1992). Regardless, individuals with low
sulfur isotope and high 13C values likely derive proportionally more nutrition from
seagrass-based food webs than open water areas within SB (Bottcher et al. 2006).
While δ13C and δ34S values are established indicators of habitat use, few studies
of wild populations have compared the isotope value of highly mobile organisms to their
habitat of origin. This requires an isotope study that is coordinated with tagging or
observational efforts, both of which can be quite expensive and labor intensive. Thus,
we took the unique opportunity to compare our isotope values to previously reported
observational data for six individuals. δ13C values are positively correlated with percent
observations in seagrass and δ34S are negatively correlated with the observational data
suggesting a close association between observational data on habitat use and stable
isotopes. Thus, it appears that dolphins frequenting seagrass habitat also derived their
carbon and sulfur from seagrass based food webs.
Like δ13C and δ34S, δ15N values also vary as a function of percent observations
foraging in seagrass habitat. δ15N values significantly decrease as the percentage of
observed feeding in seagrass increases (Fig. 2). This may result from variation in the
isotope composition of nutrients supplied to the base of the food web (Dillon and
Chanton 2008, Graham et al. 2010). Alternatively, prey trophic level and availability vary
by habitat and may also be important controls on δ15N values. For example, pinfish are

16

a dominant dietary item of SB dolphins, are frequently associated with seagrass, and
often derive a large portion of their diet from seagrass (Luczkovich and Stellwag 1993,
Barros and Wells 1998). Thus, bottlenose dolphins utilizing seagrass habitats would
likely forage at a low trophic level. In contrast, low trophic level fish that are abundant in
open water habitats (e.g., clupeids) are not common dietary items of bottlenose
dolphins (Barros and Wells 1998, Berens McCabe et al. 2010, Gannon et al. 2009).
Thus, there may be a tendency for dolphins from non-seagrass habitats to feed at
higher trophic levels than those that forage in seagrass, consistent with the observed
trend in δ15N.
Muscle isotope data provide foraging information from 1991 to 2010. Among the
variables we tested only weaning status and year had a significant influence on δ13C
and δ34S values. Calves less than 2.5 years old had significantly higher δ13C values
than older individuals. High δ13C values are likely the result of both trophic enrichment
and the use of seagrass habitats as a nursery ground for Sarasota Bay dolphins (Wells
1991, 1993). In addition to weaning and year, several other factors (e.g., predator
distribution or boat traffic) could influence foraging ecology and contribute to the large
variation we observed in our data. Our observational data suggest that SB dolphins
demonstrate some degree of individual foraging specialization that may also contribute
to the wide range of isotope values observed for muscle data from 1991 to 2010.
Despite this, the significant influence of year on δ13C and δ34S values suggests that the
use of non-seagrass habitats by bottlenose dolphins in Sarasota Bay increased from
1991 to 2010 consistent with observations of increased use of open bay habitat by SB
dolphins in recent years (Wells 2003). A state-wide net fishing ban may have been a

17

salient factor contributing to this trend. Prior to 1995 dolphins were likely in competition
with the fishery for some prey species. As a consequence, prior to 1995 dolphins may
have been forced to consume species of no commercial value that derive their carbon
primarily from inshore seagrass habitats (i.e., pinfish) (Berens McCabe et al. 2010).
However, decreased abundance of predator sharks and increased boat traffic in shallow
water may also contribute to the trend in δ13C (Wells 2003). δ15N values showed no
significant change between 1991 and 2010.This suggests that factors other than habitat
use are impacting δ15N values, a possibility that can be explored further with our longterm tooth crown tip data set.
Analyses from tooth crown tips enable us to examine temporal records of
bottlenose dolphin foraging ecology from 1944 to 2007, as teeth have been collected
from known resident dolphins born as early as 1944. Even after the Suess correction,
δ13C values significantly declined from 1944 to 1989; a trend that would be falsely
inflated if the data had not been corrected. The temporal decrease in δ 13C values likely
results from a 25% decrease in SB seagrass cover from 1950 to 1989 (Tomasko et al.
2005).
In contrast to δ13C values, δ15N values of crown tips increase from 1944 to 1989.
This trend likely reflects an increase in wastewater loading to Sarasota Bay (Tomasko
et al. 2005). The 15N-enriched nutrients typical of wastewater are taken up by
phototrophs, are passed to consumers, and result in elevated δ15N values throughout
the food web (Macko and Ostrom 1994, Cole et al. 2005, Schlacher et al. 2005). The
increase in δ15N values from 1944 to 1989 is proportional to the increase in the human
population size of Sarasota and Manatee Counties combined, the two counties

18

encompassing Sarasota Bay (Fig. 5) (U.S. Census Bureau 2010). After 1989, total
nitrogen loading was dramatically reduced due to advances in the treatment of
wastewater. In 1988 the total nitrogen entering SB was estimated at 905,386 kg
compared to 488,981 kg in 1990 and a baseline of 191,419 kg in 1890 (Tomasko et al.
2005). The absence of a temporal trend in δ15N values after 1989 likely reflects a
response of the ecosystem to reduced influence of

15N-enriched

wastewater.

Like the post-1989 tooth data, δ15N values of muscle tissue show no relationship
with time from 1991 to 2010. The observation that only carbon and sulfur isotope data
suggest an increased use of non-seagrass foraging habitats during this period is
enigmatic in that all three isotopes correlate with the percentage of observed foraging in
seagrass habitat (Fig. 2). The influence of habitat use on δ15N values may be
confounded by other variables. For example, the expected increases in δ15N values of
dolphins resulting from a higher propensity to forage in open water habitat may be
obscured by a decrease in nutrient loading and associated with a decline in the
contribution of 15N enriched nutrients to the food web.
In addition to the effects of nutrient loading on nitrogen isotope data, δ13C values
of primary producers may be impacted. For example, phytoplankton blooms stimulated
by nutrient loadings can reduce dissolved CO2 concentrations and increase δ13C of
phototrophs (Popp et al. 1998). However, δ13C values declined from 1944 to 1989.
Thus, we propose an alternative explanation that addresses the temporal trends in all
three isotopes in the context of historical changes to SB.
For much of the last half century, SB experienced a slow, steady decline in
health that is strongly related to nitrogen pollution. Nitrogen loadings in 1989 were 4.7

19

times higher than they were in 1890, a phenomenon linked to increased inputs from
wastewater, storm water, and groundwater (e.g., from septic systems) as human
population and development increased (Tomsko et al. 2005). The excess nitrogen
stimulated algal growth which, in turn, reduced light penetration and caused widespread
declines in seagrass across the bay through ca. 1989 (loss of 25% of total seagrass
coverage1950-1988) (Tomasko et al. 2005). The recovery of seagrass after 1989 is, in
part, a response to the Grizzle-Figg legislation of August 1988 passed by the Florida
state legislature (Section 403.086, Florida Statutes). This legislation placed limits on the
amount of nitrogen in wastewater discharge and promoted wastewater recovery efforts.
From 1988 to 1990 nitrogen loads into SB decreased 46% (Tomasko et al. 2005). Since
then improvements in water treatment technology resulted in further reduction of
nitrogen loading bay-wide. Today SB receives a nitrogen load equivalent to only 5% of
the 1988 value (Sarasota Bay National Estuary Program 2010). Since 1988, an
improvement in water clarity has been accompanied by a 24% increase in seagrass
acreage compared to 1950 values (Sarasota Bay National Estuary Program 2010).
The undulation in nitrogen loadings and associated seagrass coverage are
recorded in the carbon, sulfur, and nitrogen isotope data preserved within various
dolphin tissues. We demonstrate that changes in foraging patterns that correlate with
perturbations to the SB ecosystem have been propagated throughout the food web from
the earliest time we investigated, the 1940’s. Thus, marine mammals are not only
sentinels of marine ecosystem health (Wells et al. 2004, Bossart 2010) but can provide
historical context for past ecological change, providing compelling data related to
human disturbance.

20

LITERATURE CITED

21

LITERATURE CITED

Barros, N. B., P.H. Ostrom, C. A. Stricker and R. S. Wells. 2010. Stable isotopes
differentiate bottlenose dolphins off west-central Florida. Marine Mammal
Science 26:324-336.
Barros, N. B., and R. S. Wells. 1998. Prey and feeding patterns of resident bottlenose
dolphins (Tursiops truncatus) in Sarasota Bay, Florida. Journal of Mammalogy
79:1045-1059.
Berens, McCabe, E. J., D. P. Gannon, N. B. Barros and R. S. Wells. 2010. Prey
selection in a resident common bottlenose dolphin (Tursiops truncatus)
community in Sarasota Bay, Florida. Marine Biology 157:931-942.
Bossart, G. D. 2006. Marine mammals as sentinel species for oceans and human
health. Oceanography 19:134–137.
Bottcher, M. E., H.-J. Brumsack and C.-D. Durselen. 2006. The isotopic composition of
modern seawater sulfate: I. Coastal waters with special regard to the North Sea.
Journal of Marine Systems 67:73–82.
Bowen, W. D., J. D. Baker, D. Siniff, Stirling et al. 2010. Long-term Studies. Pages 283305 in I. Boyd, D. Bowen and S. Iverson eds., Marine mammal ecology and
conservation: A handbook of techniques. Oxford University Press. Oxford, United
Kingdom.
Brunner, B., and S. M. Bernasconi. 2005. A revised isotope fractionation model for
dissimilatory sulfate reduction in sulfate reducing bacteria. Geochimica et
Cosmochimica Acta 69:4759–4771.
Christensen, J., and K. Richardson. 2008. Stable isotope evidence of long-term
changes in the North Sea food web structure. Marine Ecology Progress Series
368, 1–8.
Cole, M. L., K. D. Kroeger, J. W. McClelland and I. Valiela. 2005. Macrophytes as
indicators of land-derived wastewater: Application of a δ15N method in aquatic
systems. Water Resources Research 41:1-9.
Connolly, R. M., M. A. Guest, A. J. Melville and J. M. Oakes. 2004. Sulfur stable
isotopes separate producers in marine food-web analysis. Oecologia 138:161–
167.
Dillon, K. S., and J. P. Chanton. 2008. Nitrogen stable isotopes of macrophytes assess
stormwater nitrogen inputs to an urbanized estuary. Estuaries and Coasts
31:360–370.

22

Emslie, S. D., and W. P. Patterson. 2007. Abrupt recent shift in δ13C and δ15N values in
Adélie penguin eggshell in Antarctica. Proceedings of the National Academy of
Sciences 104:11666-11669.
Gannon, D. P., E. J. Berens McCabe, S. A. Camilleri et al. 2009. Effects of Karenia
brevis harmful algal blooms on nearshore fish communities in southwest Florida.
Marine Ecology Progress Series 378:171–186.
Gould, P., P. H. Ostrom and W. Walker. 1997. Trophic relationships of albatrosses
associated with squid and large-mesh drift-net fisheries in the North Pacific
Ocean. Canadian Journal of Zoology 75:549–562.
Graham, B. S., P. L. Koch, S. D. Newsome, K. W. McMahon and D. Aurioles. 2010.
Using isoscapes to trace the movements and foraging behavior of top predators
in oceanic ecosystems. Pages 299–318 in J. B. West, G. J. Bowen, T. E.
Dawson, and K. P. Tu eds. Isoscapes: Understanding movement, pattern, and
process on earth through isotope mapping. Springer, New York, NY.
Harrigan, P., J. C. Zieman and S. A. Macko. 1989. The base of nutritional support for
the gray snapper (Lutjanus griseus): an evaluation based on a combined
stomach content and stable isotope analysis. Bulletin of Marine Science 44:65–
77.
Hobson, K. A., and J. L. Sease. 1998. Stable isotope analysis of tooth annuli reveal
temporal dietary records: an example using Stellar sea lions. Marine Mammal
Science 14:116-129.
Hobson, K. A. 1999. Stable-Carbon and Nitrogen Isotope Ratios of Songbird Feathers
Grown in Two Terrestrial Biomes: Implications for Evaluating Trophic
Relationships and Breeding Origins. The Condor 101:799-805.
Hohn, A. A., M. D. Scott, R. S. Wells, J. C. Sweeney and A. B. Irvine. 1989. Growth
layers in teeth from known-age, free-ranging bottlenose dolphins. Marine
Mammal Science 5:315-342.
Knoff, A., A. A. Hohn and S. A. Macko. 2008. Ontogenetic diet changes in bottlenose
dolphins (Tursiops truncatus) reflected through stable isotopes. Marine Mammal
Science 24:128-137.
Knowlton, N., and J. B. C. Jackson. 2008. Shifting baselines, local impacts, and global
change on coral reefs. PLoS Biology 6:215-220.
Luczkovich, J. J., and E. J. Stellwag. 1993. Isolation of cellulolytic microbes from the
intestinal tract of the pinfish, Lagodon rhomboides: size-related changes in diet
and microbial abundance. Marine Biology 116:381-388.
Macko, S. A., and N. E. Ostrom. 1994. Pollution studies using stable isotopes. Pages
45-62 in: K. Lajtha and R. H. Michener eds.. Stable isotopes in ecology and
23

environmental science. Blackwell Scientific Publications, Oxford, United
Kingdom.
Mendes, S., J. Newton, R.J. Reid, A.F. Zuur and G.J. Pierce. 2006. Stable carbon and
nitrogen isotope ratio profiling of sperm whale teeth reveals ontogenetic
movements and trophic ecology. Oecologia 151:605-615.
Myers, R. A., and B. Worm. 2003. Rapid worldwide depletion of predatory fish
communities. Nature 423:280-283.
Newsome, S. D., M. A. Etnier, D. H. Monson and M. L. Fogel. 2009. Retrospective
characterization of ontogenetic shifts in killer whale diets via δ 13C and δ15N
analysis of teeth. Marine Ecology Progress Series 374:229–242.
Newsome, S.D., M.T. Clementz and P.L. Koch. 2010. Using stable isotope
biogeochemistry to study marine mammal ecology. Marine Mammal Science,
26:509-572.
Oakes, J. M., and R. M. Connolly. 2004. Causes of sulfur isotope variability in the
seagrass Zostera capricorni. Journal of Experimental Marine Biology and
Ecology 302:153–164.
Österblom, H., O. Olsson, T. Blenckner and R.W. Furness. 2008. Junk-food in marine
ecosystems. Oikos 117:967-977.
Perrin, W. F., and S. B Reilly. 1984. Reproductive parameters of dolphins and small
whales of the family Delphinidae. Pages 97-134. in W.F. Perrin, D.P. DeMaster
and R.L. Brownell, Jr., eds. Report of the International Whaling Commission
Special Issue 6.
Peterson, B. J., and B. Fry. 1987. Stable isotopes in ecosystem studies. Annual
Reviews in Ecology and Systematics 18:293-320.
Popp, B. N., E. A. Laws, R. R. Bidigare, J. E. Dore, K. L. Hanson and S. G. Wakeham.
1998. Effect of phytoplankton cell geometry on carbon isotopic fractionation.
Geochimica et Cosmochimica Acta 62:69–77.
Quay, P., R. Sonnerup, T. Westby, J. Stutsman and A. McNichol. 2003. Changes in the
13C/12C of dissolved inorganic carbon in the ocean as a tracer of anthropogenic
CO2 uptake. Global Biogeochemical Cycles, 17:1004.
Sarasota Bay National Estuary Program. 2010. State of the Bay. Available:
http://sarasotabay.org/about-sbep/accomplishments.
Schlacher, T. A., B. Liddell, T. F. Gaston and M. Schlacher-Hoenlinger. 2005. Fish track
wastewater pollution to estuaries. Oecologia 144:570-584.

24

Sergio F., T. Caro, D. Brown, et al. 2008. Top predators as conservation tools:
ecological rationale, assumptions, and efficacy. Annual Review of Ecology,
Evolution and Systematics 39:1-19.
Sponheimer, M., T. F. Robinson, T. E. Cerling, L. Tegland, B. L. Roeder, L. Ayliffe, M.
D. Dearing and J. R. Ehleringer. 2006. Turnover of stable carbon isotopes in the
muscle, liver, and breath CO2 of alpacas (Lama pacos). Rapid Communications
in Mass Spectrometry 20:1395–1399.
Tomasko, D. A., C. A. Corbett, H. S. Greening and G. E. Raulerson. 2005. Spatial and
temporal variation in seagrass coverage in Southwest Florida: assessing the
relative effects of anthropogenic nutrient load reductions and rainfall in four
contiguous estuaries. Marine Pollution Bulletin 50:797-805.
Trust, B. A., and B. Fry. 1992. Stable sulphur isotopes in plants: a review. Plant, Cell &
Environment 15:1105–1110.U.S. Census Bureau. 2010. Census of Population
and Housing. Available http://quickfacts.census.gov/qfd/index.html.
Walker, J. L., C. W. Potter and S. A. Macko. 1999. The diets of modern and historic
bottlenose dolphin populations reflected through stable isotopes. Marine Mammal
Science 15: 335-350.
Wells, R.S., M.D. Scott and A.B. Irvine. 1987. The social structure of free-ranging
bottlenose dolphins. Current Mammalogy 1:247–305.
Wells, R. S.1991. Bringing up baby. Natural History. August: 56-62.
Wells, R.S. 1993. The marine mammals of Sarasota Bay. Chapter 9, pages. 9.1 - 9.23
in Sarasota Bay: 1992 Framework for action, P. Roat, C. Ciciccolella, H. Smith,
and D. Tomasko, eds. Published by the Sarasota Bay National Estuary Program,
5333 N. Tamiami Trail, Sarasota, FL 34234.
Wells, R.S. 2003. Dolphin social complexity: Lessons from long-term study and life
history. Pages 32-56 in F.B.M. de Waal and P.L. Tyack, eds. Animal social
complexity: Intelligence, culture, and individualized societies. Harvard University
Press, Cambridge, MA.
Wells, R. S., H. L. Rhinehart, L. J. Hansen, et al. 2004. Bottlenose dolphins as marine
ecosystem sentinels: Developing a health monitoring system. EcoHealth 1:246254.
Wells, R. S. 2009. Learning from nature: Bottlenose dolphin care and husbandry. Zoo
Biology 28:1-17.

25

CHAPTER 2
FORAGING HABITS IN A GENERALIST PREDATOR: SEX AND AGE INFLUENCE
HABITAT SELECTION AND RESOURCE USE AMONG BOTTLENOSE DOLPHINS
ABSTRACT
This study examines resource use (diet, habitat use, and trophic level) within and
among demographic groups (males, females, and juveniles) of bottlenose dolphins
(Tursiops truncatus). We analyzed the δ13C and δ15N values of 15 prey species
constituting 84% of the species found in stomach contents. We used these data to
establish a trophic enrichment factor (TEF) to inform dietary analysis using a Bayesian
isotope mixing model. We document a TEF of 0‰ and 2.0‰ for δ13C and δ15N,
respectively. The dietary results showed that all demographic groups relied heavily on
low trophic level seagrass-associated prey. Bayesian standard ellipse areas (SEAb)
were calculated to assess diversity in resource use. The SEAb of females was nearly
four times larger than that of males indicating varied resource use, likely a consequence
of small home ranges and habitat specialization. Juveniles possessed an intermediate
SEAb, generally feeding at a lower trophic level compared to females, potentially an
effect of natal philopatry and immature foraging skills. The small SEAb of males reflects
a high degree of specialization on seagrass associated prey. Patterns in resource use
by the demographic groups are likely linked to differences in the relative importance of
social and ecological factors.

26

INTRODUCTION
The use of particular food resources or habitats by members of a population impacts the
intensity of intraspecific competition, social interactions, and risk of predation or
parasitism (Bolnick et al. 2003). Further, a wealth of recent literature documents that
ecological inequality within a population results in differential use of the total resource
pool (e.g., habitat or prey type) by individuals or assemblages within that population
(Bolnick et al. 2003, Newsome et al. 2009, Torres and Read, 2009). Such intraspecific
variation in resource use can confer both benefit and detriment to the long-term viability
of a population. For example, differential habitat use among assemblages within a
population can result in increased population viability amid the loss of a specific habitat
type (Tyler and Rose 1994). However, loss of a habitat upon which all or most juveniles
are dependent would likely result in population decline, emphasizing the importance of
studying variation in resource use among demographic groups (Dahlgren et al. 2006).
The impact on certain demographic groups, and thus population viability, is of particular
concern for wildlife populations in close proximity to human development because
resources within these communities are more likely to experience disturbance (e.g.,
habitat loss, competition with humans for resources). Marine mammals are of particular
interest as they are large predators that often inhabit metropolitan coastal areas where
habitats have undergone degradation and increased fishing pressure.
The bottlenose dolphins (Tursiops truncatus) of Sarasota Bay (SB), FL, represent
a model system to study patterns in resource use within and between demographic
groups. SB dolphins are year-round, multi-generational residents that have been
intensively studied for nearly five decades (Wells et al. 1987; Wells 2003, 2014).

27

Previous stomach content studies of stranded deceased dolphins from the SB
population (Barros and Wells 1998, Berens McCabe et al. 2010) provide estimates of
diet that reflect recently ingested prey. Additionally, dolphin foraging has been assessed
using molecular prey detection on fecal samples (Dunshea et al. 2013). Unlike stomach
content analysis, molecular prey detection is not limited to salvaged individuals.
However, the time period represented by the molecular data is similarly short. Together
stomach content data and molecular prey detection are excellent indicators of the prey
species commonly consumed by SB bottlenose dolphins. Yet, to date, there is no
information on potential differences in resource use among demographic groups within
the long-term resident community.
Carbon and nitrogen stable isotope analysis of bottlenose dolphin tissues
complements stomach content and molecular techniques by providing information on
habitat use, trophic position, and diet over a long time period (half-life of months for skin
and years for muscle) (Newsome et al. 2010). The foraging habits of a large number of
individuals can be obtained because, in addition to tissues from salvaged animals,
analysis can be conducted on biopsies from extant members of the population. Unique
types of information can be derived from stable isotope data. Carbon isotope values
(δ13C) are indicators of primary production at the base of the food web. δ 13C values
differentiate seagrass from phytoplankton and nonseagrass habitat (mangrove and
open bay) (Peterson and Fry 1987). For Sarasota Bay bottlenose dolphins, high δ13C
values indicate frequent foraging in seagrass habitat (Rossman et al. 2013). Nitrogen
isotope values (δ15N) generally increase by 3‰-4‰ with each step in the food chain,
offering a good indicator of trophic level (Peterson and Fry, 1987); although, recent

28

studies indicate trophic dynamics may be lower for high trophic level organisms (Hussey
et al. 2014). The information provided by stable isotope analysis is particularly valuable
for cryptic foragers such as bottlenose dolphins who capture and consume prey
underwater. Because of the ecological information derived from stable isotope
approaches, mass-balance models using multiple isotopes are well established tools for
determining the relative contribution of various prey sources to a population’s diet
(Harrigan et al. 1987, Phillips et al. 2005).
Recently, dietary assessments using stable isotopes have been advanced via the
development of Bayesian mass balance models. By incorporating variability in prey and
consumer isotope values, Bayesian diet modeling programs, such as Stable Isotope
Analysis in R (SIAR) (Jackson et al. 2011) have a unique advantage over earlier mass
balance models (Harrigan et al. 1989, Phillips et al. 2005). Unlike the earlier models that
report the fractional contribution of prey to the diet as a range, Bayesian models provide
a true probability distribution for each prey item (Parnell et al. 2010).
While SIAR provides an estimate of the average contribution of each prey item to the
diet, the degree of variation within a population or demographic group is also an
important descriptor of foraging ecology. Stable isotope Bayesian ellipses in R (SIBER)
(Jackson et al. 2011) uses δ13C and δ15N values to construct ellipses that represent two
dimensional equivalents of the standard deviation. While these ellipses were originally
intended to compare isotopic variability among species within a community they can
also be used to quantify variability within and between demographic groups and to infer
differences in habitat use and trophic diversity.

29

In this study we assess resource use (diet, habitat use, and trophic level) among
male, female, and juvenile bottlenose dolphins (demographic groups). We first compare
diet among the groups using stable isotope-based Bayesian dietary estimation. We then
compare variation in resource use among the groups via stable isotope Bayesian
standard ellipses. These tools provide important insight into a cryptic forager and the
manner in which the unique ecology of the demographic groups relates to resource use
at the population level.

METHODS
Sample Site and Acquisition

SB is a complex series of shallow bays (<4m deep, 40 km long) on the central west
coast of Florida, communicating with the Gulf of Mexico through narrow, deep passes
separating a series of narrow barrier islands. The study area encompasses a diverse
array of habitats, including seagrass meadows, mangrove fringing forests, humanaltered shorelines, and open bay (Rossman et al. 2013). Fish commonly found in the
diet of bottlenose dolphins were collected for isotopic analysis during fish abundance
surveys 2009-2012 (Table 1) (Barros and Wells 1998, Berens McCabe et al. 2010,
Dunshea et al. 2013). In total, 15 species and 234 individual samples were processed
for isotope analysis. Sampling targeted fish between 100 and 300 mm, the size range
most often found in stomach contents of bottlenose dolphins (Barros and Wells 1998).
Bottlenose dolphin skin samples were obtained from a community of ca. 160 individuals,
mostly of known age and sex that are resident to SB (Wells 2003, 2009). Skin samples
were taken during health assessments when dolphins were briefly captured, biopsied,
30

and released (Wells et al. 2004). In the field, tissues were stored in liquid nitrogen and
retained frozen prior to analysis. Isotope values for bottlenose dolphin muscle tissue
used to determine the trophic enrichment factor (TEF) were taken from Rossman et al.
(2013). Because muscle data derive from stranded, deceased dolphins it is independent
from the skin samples used for diet estimation in SIAR.

Sample Preparation
White muscle tissue from fish and bottlenose dolphin skin were freeze-dried, lipid
extracted, and homogenized to a fine powder in a high-energy ball mill (SPEX
SamplePrep) and a 1 mg aliquot of homogenate was transferred to tin capsules for
stable carbon and nitrogen isotope analysis. Isotope values were determined using an
elemental analyzer (Eurovector) interfaced to an Isoprime mass spectrometer
(Elementar). Isotope values are expressed as:
δX = {(Rsample / Rstandard ) – 1} x 1,000
where X represents 13C or 15N, and R represents the abundance ratios:
15N/14N

13C/12C

or

respectively. In-house standards used for δ13C and δ15N were calibrated with

respect to international scales V-PDB and air respectively. In-house precision was +/0.2‰ for δ13C and δ15N values.

Statistical Analysis
To delineate factors impacting prey fish isotope values, we assessed the influence of
standard length, habitat, season, and location on δ13C and δ15N values of each of five
prey fish species whose sample size was sufficient for general linear modelling (GLM).

31

These species included spot, Leiostomus xanthurus; spotted seatrout, Cynoscion
nebulosus; pigfish, Orthopristis chrysoptera; pinfish, Lagodon rhomboides; and
sheepshead, Archosargus probatocephalus. The variable habitat included three levels:
seagrass, mangrove, and open bay. Tukey’s HSD was used to test for pairwise
differences in mean δ13C or δ15N value between habitats. For the variable “season”, fish
were grouped based on the calendar date of capture: spring (21 April-6 June), summer
(21 June-20 September), fall (21 September-20 December), and winter (21 December21 April). For the variable location, latitude and longitude were combined into a single
variable via principal components analysis (PCA). The first PCA, retained for modeling
efforts as the variable “location”, explained 97% of the variation in latitude and
longitude. All statistical analyses were conducted in R 3.0.1 (R Development Core
Team 2013).
Bottlenose dolphin demographic groups were defined as follows: males (>6 yr of
age), females (>6 yr of age), and juveniles (2 to 6 yr of age). Dolphins younger than 2 yr
of age are not included in health assessments and, thus, not represented in this study.
Nearly all calves leave their mothers around 6 yr of age, thus the age ranges for males
and females (>6 yr of age) only reflect foraging by males and females independent from
their mothers (Wells 1991). The package, Stable Isotope Analysis in R (SIAR) 4.2
(Parnell and Jackson 2013) was run within R 3.0.1 (R Development Core Team 2013)
to provide dietary estimates for each bottlenose dolphin demographic group. SIAR
requires isotope values for consumers, isotope values of prey species, and a TEF.
Because SIAR cannot differentiate prey items with similar isotope values (Phillips et al.
2005), we used a k-means cluster analysis to group prey fish into four clusters. The

32

resulting clusters differed by trophic level and/or source of primary production at the
base of the food web. The TEF value used in this study was based on stomach content
data, molecular prey detection, prey isotope values, and muscle isotope values of
stranded deceased dolphins from SB. First, a weighted average δ13C value of
bottlenose dolphin diet was calculated as follows:

Where δ13Cd is the average carbon isotope value of bottlenose dolphin diet, Fi is the
fractional contribution of diet item i based on stomach content analysis or molecular
prey detection (Barros and Wells 1998, Berens McCabe et al. 2010, Dunshea et al.
2013), and δ13Ci is the average carbon isotope value of diet item i as reported in this
study. To determine the δ13C TEF the isotope value of the bottlenose dolphin diet
(δ13Cd) was subtracted from the average SB dolphin muscle δ13C value (Rossman et al.
2013). The same method was then applied to determine the TEF for δ15N. With only two
TEF values we could not provide an estimate of variance associated with trophic
fractionation which was assumed to be negligible compared to other sources of isotopic
variation. Because the TEF derives from isotope values of bottlenose dolphin muscle
and the dietary estimates used isotope values of skin, the TEF determination was
independent of data used in Bayesian dietary estimation.
Stable isotope Bayesian ellipses in R (SIBER) including standard ellipse areas
were produced using SIAR (Jackson et al. 2011). To test for significant differences we
ran 100,000 Markov-chain Monte Carlo iterations for SEAb and constructed 95%
credible intervals around the mean of each subgroup. The probability that there was a
33

significant difference in SEAb between demographic groups was determined by
calculating the proportion of times the posterior estimate for one demographic group
was smaller than another (Turner et al. 2010). We considered the two SEAb to be
significantly different when no more than 5% of the posterior estimates for one group
were smaller than those of another group (α = 0.05). For graphical representations we
used standard ellipse areas controlled for small sample size (SEAc).

RESULTS
We assessed isotope values of prey species constituting 81% (stomach content
analysis) or 84% (molecular prey detection) of bottlenose dolphin diet (Barros and Wells
1998, Berens McCabe et al. 2010, Dunshea et al. 2013). Species average δ13C values
ranged from -17.5‰ (ladyfish, Elops saurus, and threadfin herring, Opisthonema
oglimum) to -12.9‰ (striped mullet, Mugil cephalus), and average δ15N values ranged
from 7.8‰ (mojarra, Gerreid sp.) to 11.7‰ (ladyfish) (Table 1). Habitat, length, season,
and location did not significantly influence the isotope values of pigfish, spot,
sheepshead, or spotted seatrout. Pinfish δ13C values were significantly related to habitat
and location (F2, 42 = 36.09, P < 0.001, F1, 42 = 5.20, P = 0.028, respectively). δ13C
values decreased from northwest to southeast and from seagrass (mean ± standard
deviation: -13.7‰ ± 1.2‰) to open bay (-15.5‰ ± 1.0‰) to mangrove habitat (-17.1‰ ±
1.0‰) (seagrass vs. open bay: P < 0.001, seagrass vs. mangrove: P < 0.001, mangrove
vs. seagrass: P < 0.001). Length and season did not appear to influence pinfish δ13C
values. Pinfish δ15N values demonstrated a significant relationship with habitat and
length (F2, 42 = 6.43, P =0.004, F1, 42 = 5.80, P = 0.021, respectively) but not location or

34

Table 1. Bottlenose dolphin prey species δ13C and δ15N values (means ± standard
deviation), the number of samples used to determine the average isotope value (n)
dietary indicators, SCA and DNA and prey cluster designation. SCA and DNA are the
fractional contribution of each species to the diet of bottlenose dolphins as determined
by stomach content analysis or molecular prey detection (Barros and Wells 1998,
Berens McCabe et al. 2010, Dunshea et al. 2013). Prey cluster refers to the cluster
designation as determined by K means cluster analysis. Asterisk indicates instances
where Dunshea et al. 2013 report genus sp., the genus and species are provided here.

35

season. δ15N values of pinfish increased with length and were higher in open bay vs.
seagrass habitat (9.3‰ ± 0.5‰ vs. 8.0‰ ± 1.3‰, P = 0.003). δ15N values from
mangrove pinfish (8.6‰ ± 0.7‰) did not significantly differ from those of seagrass or
open bay.
The K-means cluster analysis of prey fish isotope values combined fish species
that shared a common food web base (δ13C values) and trophic level (δ15N values) (Fig.
6). High δ13C values are indicative of seagrass based food webs (e.g., seagrass grazing
pinfish, -13.7‰) and low δ13C values are associated with open bay and mangrove
primary production (e.g., clupeids whose food web is generally phytoplankton based:
threadfin herring, -17.5‰, scaled sardine, Harengula jaguana, -16.2‰). Cluster 1,
contained high trophic level nonseagrass prey and included snook (Centropomus
undecimalis), ladyfish, and spot. Cluster 2 consisted of high trophic level prey
associated with seagrass and consisted of seatrout and crevalle jack (Caranx hippos).
Cluster 3 included low trophic level, seagrass-associated species and consisted of Gulf
toadfish (Opsanus beta), seagrass pinfish, pigfish, red drum (Sciaenops ocellatus), lane
snapper (Lutjanus synagris), sheepshead, mojarra, and striped mullet. Cluster 4
contained low trophic level nonseagrass associated prey and consisted of threadfin
herring, nonseagrass pinfish, and scaled sardine. Because pinfish isotope values
significantly differed between habitats, pinfish were separated by habitat prior to the
cluster analysis. This resulted in cluster 3 containing seagrass pinfish and cluster 4
containing nonseagrass pinfish with low δ13C values (open bay and mangrove). The
mean isotope values and associated standard deviation for all species within a cluster

36

were used to estimate cluster averages and standard deviation, which were
implemented as sources in the Bayesian dietary estimation.
The weighted average δ13C and δ15N of SB bottlenose dolphin diet based on the
relative contribution of prey from stomach content analysis or molecular prey detection
was -14.2‰ and 8.9‰ and -14.5‰ and 8.9‰ respectively (Barros and Wells 1998,
Berens McCabe et al. 2010, Dunshea et al. 2013). Given the average isotope value for
bottlenose dolphin muscle (-14.4‰ and 10.8‰ for δ13C and δ15N, respectively) the TEF
estimate based on stomach content analysis was -0.1‰ for δ13C and 2.0‰ for δ15N.
The TEF estimate based on molecular prey detection was 0.1‰ for δ13C and 2.0‰ for
δ15N. The TEF used in this study was an average of the two values: 0.0‰ for δ 13C and
2.0‰ for δ15N.The mean skin isotope values, reported as mean ± standard deviation,
for the three demographic groups were similar to one another: juveniles (δ13C = -14.9‰
± 0.9‰, δ15N = 10.6‰ ± 0.6‰, n = 14), females (δ13C = -14.6‰ ± 1.3‰, δ15N = 11.2‰
± 0.9‰, n = 14) and males (δ13C = -14.2‰ ± 0.4‰, δ15N = 10.7‰ ± 0.7‰, n = 14) (Fig.
6).

37

Figure 6. δ13C and δ15N values of prey clusters and bottlenose dolphin skin from
different demographic groups including males (>6 yr old), females (>6 yr old) and
juveniles (2 to 6 yr old). Prey clusters show mean with error bars depicting the standard
deviation. Prey clusters were adjusted for trophic enrichment (cluster average + TEF).
TEF estimates derive from this study (δ13C: 0.0‰, δ15N: 2.0‰), not the literature.

The results of the Bayesian mass balance model, given as mean ± standard
deviation, showed that juveniles consumed prey cluster 3 in the highest proportion (0.58
± 0.09) (Fig. 7) followed by cluster 4 (0.31 ± 0.11), cluster 1 (0.06 ± 0.05), and cluster 2
(0.05 ± 0.05). Females consumed cluster 3 in the highest proportion (0.60 ± 0.11),
followed by cluster 4 (0.20 ± 0.12), cluster 2 (0.11 ± 0.08), and cluster 1 (0.09 ± 0.07).
Males consumed prey cluster 3 in the highest proportion (0.71 ± 0.07) followed by
cluster 4 (0.14 ± 0.08), cluster 2 (0.09 ± 0.06), and cluster 1 (0.06 ± 0.05).

38

Figure 7. SIAR results for bottlenose dolphin demographic groups: juveniles (J; 2-6 yr
old males and females), females, (F; >6 yr old) and males (M; >6 yr old). Horizontal
lines within boxes represent mean dietary contributions of each prey cluster to the
demographic group. Boxes enclose the 50% credible interval and vertical lines with end
caps depicting the 95% credible interval. Cluster 1, contains snook, ladyfish, and spot.
Cluster 2 contains seatrout and crevalle jack. Cluster 3 contains toadfish, seagrass
pinfish, pigfish, red drum, lane snapper, sheepshead, mojarra, and striped mullet.
Cluster 4 contains nonseagrass prey fish, threadfin herring, nonseagrass pinfish, and
scaled sardine.

Females had the largest SEAb at 3.59‰2 (95% CI: 2.15-6.00‰2) followed by juveniles
at 1.93‰2 (95% CI: 1.10-3.21‰2), and males at 1.29‰2 (95% CI: 0.77- 2.14‰2) (Fig.
8). The SEAb of males was significantly smaller than that of females (P = 0.003) but not
juveniles (P = 0.144). The SEAb of juveniles was significantly smaller compared to the
SEAb of females (P = 0.046). SEAc was calculated for graphical representations of
standard ellipse area (Fig.8).
39

Figure 8. Carbon and nitrogen isotope values of prey clusters and standard ellipses
controlled for small sample size (SEAc) for female, male and juvenile bottlenose
dolphins. The upper right inset table provides SEAc area (‰2), mean Bayesian
standard ellipse area (SEAb) (‰2), and the lower (2.5%) and upper (97.5%) bounds
for the 95% credible interval around the mean SEAb for males (M; >6 yr old),
females (F; >6 yr old), and juveniles (J; 2 to 6 yr old).

DISCUSSION
Male, female, and juvenile bottlenose dolphins possess unique combinations of
ecological, physiological, and social constraints (Wells 2003, 2014), all of which may
impact how these different demographic groups find and consume prey. We probed
differences in the foraging ecology of these three groups using stable isotope analysis
and Bayesian modeling. Because our analyses included a large number of prey species
(15) we performed cluster analysis to group prey. This resulted in ecologically significant

40

clusters differentiated by source of primary production at the food web base and/or
trophic level.
Our Bayesian mass balance modeling benefited from a TEF estimate that was,
a) based on prey data specific to the SB bottlenose dolphin food web, b) independent of
our skin isotope data used in SIAR and c) derived from two independent estimates of
population diet, one based on stomach content analysis and the other on molecular
prey detection (Barros and Wells 1998, Berens McCabe et al. 2010, Dunshea et al.
2013). In using the average isotope value of muscle to produce a TEF that was later
applied to Bayesian modeling based on skin data, we assumed that the TEF of muscle
and skin were similar. This assumption is consistent with the findings of Borrell et al.
(2012) and Fernández et al. (2011), who documented identical average δ13C and δ15N
values for skin and muscle from cetaceans. For SB bottlenose dolphins, the mean δ13C
and δ15N isotope values of skin and muscle differed by less than 0.1‰.
The average TEF values calculated in this study were 0‰ for δ13C and 2.0‰ for δ15N.
While our δ15N TEF is low compared to the often cited average of 3.4‰ (Post 2002), a
growing body of literature documents widespread variability in TEF across and within
taxa (McCutchan et al. 2003, Lecomte et al. 2011). Our values are in agreement with a
low δ15N TEF expected for a consumers with a high protein diet and who utilize marine
environments (Vanderklift and Posand 2003).
Results from the Bayesian mass balance model indicate that all bottlenose
dolphin demographic groups predominantly depend on low trophic level, seagrassassociated prey fish (prey cluster 3). Among the clusters, prey cluster 3 contains the
largest number of species including both prey fish frequently small (e.g., pinfish, pigfish,

41

and mojarra) and more massive prey items (e.g., mullet). Prey cluster 3 is likely the
most important dolphin diet because seagrass serves as an excellent foraging habitat,
supporting high densities of prey fish and providing safety from predators in SB (e.g.,
bull sharks) (Barros and Wells 1998, Gannon et al. 2009, Mann et al. 2000, McHugh et
al. 2011). Our findings are consistent with those of Barros and Wells (1998)
documenting seagrass as an important habitat for bottlenose dolphins. In addition to
prey cluster 3, juveniles, in particular, appeared to consume low trophic level,
nonseagrass prey (cluster 4) to a larger extent than male or females. This may be
related to physiological and behavioral constraints that are unique to juveniles. The
acquisition of foraging skills by calves is likely a slow process; upon nutritional weaning
(typically in second year of life), calves may not yet possess skills associated with
complex foraging behavior typical of adults. For example, the detection and capture of
prey species associated with structures (e.g., Gulf toadfish) or fast-moving high trophic
level fish may be too difficult for young dolphins (Berens McCabe et al. 2010). The small
gape size of juveniles may work in conjunction with undeveloped foraging skills to
constrain diet to small, easily detectable, low trophic level prey, such as all members of
prey cluster 4 and pinfish, pigfish and mojarra in prey cluster 3. However, the finding
that high trophic level prey were not a large contributor to the diet of any demographic
group is consistent with the observation that high trophic level prey species occur in
only low abundances and may be too energetically costly for consumption at the
population level.
An increasing body of literature suggests that variation in resource use within and
between individuals of a population is an important ecological parameter that has

42

profound implications for conservation biology (Bolnick et al. 2003, Johnson et al. 2009).
To assess variation in the foraging behavior between male, female, and juvenile
bottlenose dolphins, we used SEAb as a quantitative indicator of variation in resource
use (Jackson et al. 2011). Male and female bottlenose dolphins appear widely divergent
in their diversity of resource use. Males possess a significantly small SEAb, nearly one
fourth the size of the female SEAb. The small SEAb size indicates that differences in
foraging habits among males are small. Even though most male bottlenose dolphins
possess home ranges larger than those of females (Wells 2003, Urian et al. 2009) and
have access to numerous habitat types, the position and small size of the male
standard ellipse indicates a predominant reliance on seagrass-associated prey. This
may occur because seagrass habitat has one of the highest densities of prey fish of any
habitat (Gannon et al. 2009) often including numerous large prey species (e.g., mullet),
which generally provide more calories compared to smaller fish. Because gape size can
be a constraint, the larger gape size of males offers an advantage. More importantly,
consuming large prey allows males, relative to females and juveniles, to maximize their
caloric intake while minimizing energy expenditure.
In contrast to the males, the large SEAb of females indicates broader foraging
habits. The large SEAb predominately derives from the wide range of δ13C values, an
indicator of habitat use (i.e., seagrass vs. open water). This suggests that individual
female dolphins consistently utilize a subset of available habitats (habitat
specialization). Observations of SB bottlenose dolphins suggest a high degree of habitat
specialization among females. For example, females occupy smaller home ranges
compared to males (Wells 2003, Urian et al. 2009). For reproductively active females,

43

habitat familiarity, predator avoidance, and proximity to familiar female associates favor
a small home range and place limitations on foraging habitat. Thus, some females in the
SB population may specialize on prey from seagrass habitats while others may
consume prey associated with phytoplankton based food webs. Thus, it appears that
while most males specialize on seagrass-associated prey, females may specialize on a
variety of habitats or prey resources. Furthermore, female bottlenose dolphin habitat
selection appears correlated with trophic level. Females who forage in seagrass also
consume lower tropic level prey compared to females who forage in open water. This
likely results from seagrass associated female dolphins utilizing the abundant low
trophic level prey commonly found in seagrass habitat. While low trophic level species
are found in open water (e.g, clupeids) they do not substantially contribute to the diet of
female bottlenose dolphins as indicated by SIAR results, stomach content analysis and
molecular prey detection (Barros and Wells 1998, Berens McCabe et al. 2010, Dunshea
et al. 2013).
Samples obtained during health assessments are, out of necessity, generally
from dolphins traveling in shallow habitats. This possibly introduces a bias favoring
individuals who frequently utilize seagrass, a common habitat type for shallow water.
However, a bias of this nature would likely impact males and females similarly yet, we
document an SEAb for females significantly larger than that of males.
Juveniles possessed an intermediate SEAb between males and females. Relative to
females, the ellipse for juveniles is similar in spatial orientation but smaller in total area.
The shape of the ellipse for juveniles is contracted in δ13C and lower in δ15N compared
to females. The similarity in shape between the female and juvenile ellipses is likely the

44

result of maternal habitat selection and/or the high degree of philopatry demonstrated
by calves newly independent of their mother (McHugh et al. 2011). The lower δ15N
values suggest that juveniles feed at a lower trophic level than females possibly
resulting from limitations of a small gape size and less developed foraging skills.
Bottlenose dolphins have traditionally been described as opportunistic generalists
(Shane et al. 1986). However, recent studies have demonstrated that bottlenose
dolphins do not indiscriminately capture and consume prey; instead, they specialize on
a subset of available prey, especially soniferous fishes (Berens McCabe et al. 2010),
yet the manner in which dietary variability and specialization is partitioned within dolphin
populations remains uncertain. In this study, we found an ecologically significant
disparity in the diversity of resource use by male and female bottlenose dolphins, with
females accounting for the majority of the variation in foraging habits at the population
level. Differences in resource use among male, female, and juvenile bottlenose dolphins
likely resulted from trade-offs associated with social interactions, predator avoidance,
and energetic needs related to body size, activity levels, and reproductive condition.
This dichotomy in resource use may have impacts for conservation and management.
Because specialization in resource use within a population impacts social interaction
and exposure to habitat degradation, certain subsets of the population may be more
likely to strand as the result of disturbance. This is because individuals who utilize a
particular foraging habitat may be predisposed to stranding or increased mortality if
nutritional stress results from habitat loss or disturbance (Johnson et al. 2009). In
addition, an elevated mortality risk may result from increased exposure to pathogens or
diseased individuals, phenomena that may be promoted by some types of foraging

45

specialization. Consequently, foraging habits may play a critical role in understanding
the ecology of phenomena such as unusual mortality events (Gulland and Hall, 2007).

46

LITERATURE CITED

47

LITERATURE CITED

Barros, N. B., and R. Wells. 1998. Prey and feeding patterns of resident bottlenose
dolphins (Tursiops truncatus) in Sarasota Bay, Florida. Journal of Mammalogy
79:1045-1059.
Berens McCabe, E. J., D. P. Gannon, N. B. Barros and R. S. Wells. 2010. Prey
selection in a resident common bottlenose dolphin (Tursiops truncatus)
community in Sarasota Bay, Florida. Marine Biology 157:931-942.
Bolnick, D. I., R. Svanback, J. A. Fordyce, L. H. Yang, J. M. Davis, C. D. Hulsey and M.
L. Forister. 2003. The ecology of individuals: Incidence and implications of
individual specialization. American Naturalist. 161:1-28.
Borrell, A., N. Abad-Oliva, E. Gómez-Campos, J. Giménez and A. Aguilar. 2012.
Discrimination of stable isotopes in fin whale tissues and application to diet
assessment in cetaceans. Rapid Communications in Mass Spectrometry
26:1596-1602.
Dahlgren, C. P., G. T. Kellison, A. J. Adams, et al. 2006. Marine nurseries and effective
juvenile habitats: concepts and applications. Marine Ecology Progress Series
132:291-295.
Dunshea, G., N. B. Barros, E. J. Berens McCabe, N. J. Gales, M. A. Hindell, S. N.
Jarman and R. S. Wells. 2013. Stranded dolphin stomach contents represent the
free-ranging population's diet. Biology Letters 9.
Fernández, R., S. García-Tiscar, M. B. Santos, A. López, J. A. Martínez-Cedeira, J.
Newton and G. J. Pierce. 2011. Stable isotope analysis in two sympatric
populations of bottlenose dolphins Tursiops truncatus: evidence of resource
partitioning? Marine Biology 158:1043-1055.
Gannon, D. P., E. J. Berens McCabe, S. A. Camilleri, et al. 2009. Effects of Karenia
brevis harmful algal blooms on nearshore fish communities in southwest Florida.
Marine Ecology Progress Series 378:171-186.
Gulland, F. M., and A. J Hall. 2007. Is marine mammal health deteriorating? Trends in
the global reporting of marine mammal disease. EcoHealth 4: 135–150.
Harrigan, P., J. C. Zieman, and S. A. Macko. 1989. The base of nutritional support for
the gray snapper (Lutjanus griseus): an evaluation based on a combined
stomach content and stable isotope analysis. Bulletin of Marine Science 44:65–
77.

48

Hussey, N. E., M. A. MacNeil, B. C. McMeans, et al. 2014. Rescaling the trophic
structure of marine food webs. Ecology Letters 17: 239–250.Jackson, A. L., R.
Inger, A. C. Parnell and S. Bearhop. 2011. Comparing isotopic niche widths
among and within communities: SIBER–Stable Isotope Bayesian Ellipses in R.
Journal of Animal Ecology 80:595-602.
Johnson, C. K., M. T. Tinker, J. A. Estes, et al. 2009. Prey choice and habitat use drive
sea otter pathogen exposure in a resource-limited coastal system. Proceedings
of the National Academy of Sciences 106:2242-2247.
Lecomte, N., Ø. Ahlstrøm, D. Ehrich, E. Fuglei, R. A. Ims and N. G. Yoccoz. 2011.
Intrapopulation Variability Shaping Isotope Discrimination and Turnover:
Experimental Evidence in Arctic Foxes. PLoS ONE 6.
Macleod, C. D., M. B. Santos, A. Lopez and G. J. Pierce. 2006. Relative prey size
consumption in toothed whales: implications for prey selection and level of
specialisation. Marine Ecology Progress Series 326:295-307.
Mann, J., R. C. Connor, L. M. Barre, and M. R. Heithaus. 2000. Female reproductive
success in bottlenose dolphins (Tursiops sp.): life history, habitat, provisioning,
and group-size effects. Behavioral Ecology 11:210–219.
McCutchan, J. H., W. M. Lewis, C. Kendall, and C. C. McGrath. 2003. Variation in
trophic shift for stable isotope ratios of carbon, nitrogen, and sulfur. Oikos 102:
378–390.
McHugh, K. A., J. B. Allen, A. A. Barleycorn, and R. S. Wells. 2011. Natal philopatry,
ranging behavior, and habitat selection of juvenile bottlenose dolphins in
Sarasota Bay, Florida. Journal of Mammalogy 92:1298–1313.
Newsome, S. D., M. T. Tinker, D. H. Monson, et al. 2009. Using stable isotopes to
investigate individual diet specialization in California sea otters (Enhydra lutris
nereis). Ecology 90:961-974.
Newsome, S. D., M. T. Clementz and P. L. Koch. 2010. Using stable isotope
biogeochemistry to study marine mammal ecology. Marine Mammal Science
26:509-572.
Parnell, A. C., R. Inger, S. Bearhop, and A. L. Jackson. 2010. Source partitioning using
stable isotopes: coping with too much variation. PLoS One, 5:e9672.
Parnell, A. C., and A. L. Jackson. 2013. SIAR: Stable Isotope Analysis in R. Available at
cran.r-project.org/web/packages/siar/index.html.
Peterson, B. J., and B. Fry. 1987. Stable Isotopes in Ecosystem Studies. Annual
Reviews in Ecology and Systematics 18:293-320.
49

Phillips, D. L., S. D. Newsome and J. W. Gregg. 2005. Combining sources in stable
isotope mixing models: alternative methods. Oecologia 144:520-527.
Post, D. M. 2002. Using stable isotopes to estimate trophic position: models, methods,
and assumptions. Ecology 83:703–718.
Rossman, S., N. B. Barros, P. H. Ostrom, C. A. Stricker, A. A. Hohn, H. Gandhi and R.
S. Wells. 2013. Retrospective analysis of bottlenose dolphin foraging: A legacy of
anthropogenic ecosystem disturbance. Marine Mammal Science 29:705-718.
Shane, S. H., R. S. Wells, B. Würsig. 1986. Ecology, behavior and social organization of
the bottlenose dolphin: a review. Marine Mammal Science 2:34–63.
R Development Core Team. 2013. R: A Language and Environment for Statistical
Computing. R Foundation for Statistical Computing, Vienna, Austria.
Torres, L. G., and A. J. Read. 2009. Where to catch a fish? The influence of foraging
tactics on the ecology of bottlenose dolphins (Tursiops truncatus) in Florida Bay,
Florida. Marine Mammal Science 25:797-815.
Turner, T. F., M. L. Collyer, and T. J. Krabbenhoft. 2010. A general hypothesis-testing
framework for stable isotope ratios in ecological studies. Ecology, 91:2227–2233.
Tyler, J. A., and K. A. Rose. 1994. Individual variability and spatial heterogeneity in fish
population models. Reviews in Fish Biology and Fisheries 4:91–123.
Urian, K. W., S. Hofmann, R. S. Wells, and A. J. Read. 2009. Fine-scale population
structure of bottlenose dolphins (Tursiops truncatus) in Tampa Bay, Florida.
Marine Mammal Science 25:619–638.
Vanderklift, M. A. and S. Ponsard. 2003. Sources of Variation in Consumer-Diet δ15N
Enrichment: A Meta-Analysis. Oecologia 136:169-182.
Wells, R. S., M. D. Scott and A. B. Irvine. 1987. The social structure of free-ranging
bottlenose dolphins. Current Mammalogy 1:247–305.
Wells, R. S.1991. Bringing up baby. Natural History. August: 56-62.
Wells, R. S. 2003. Dolphin social complexity: Lessons from long-term study and life
history. Pages 32-56 in F. B. M. de Waal and P. L. Tyack, eds. Animal social
complexity: Intelligence, culture, and individualized societies. Harvard University
Press, Cambridge, MA.

50

Wells, R. S., H. L. Rhinehart, L. J. Hansen, et al. 2004. Bottlenose Dolphins as Marine
Ecosystem Sentinels: Developing a Health Monitoring System. EcoHealth 1:246–
254.
Wells, R. S. 2009. Learning from nature: Bottlenose dolphin care and husbandry. Zoo
Biology 28:1-17.
Wells, R.S. 2014. Social structure and life history of common bottlenose dolphins near
Sarasota Bay, Florida: Insights from four decades and five generations. Pp. 149172 in J. Yamagiwa and L. Karczmarski, eds. Primates and cetaceans: Field
research and conservation of complex mammalian societies. Springer, Tokyo,
Japan.

51

CHAPTER 3
INDIVIDUAL SPECIALIZATION IN THE FORAGING HABITS OF FEMALE
BOTTLENOSE DOLPHINS LIVING IN A TROPHICALLY DIVERSE AND HABITATRICH ESTUARY
ABSTRACT
We examined the prevalence of individual specialization in foraging habits (foraging
habitat and trophic level) of female bottlenose dolphins resident to Sarasota Bay, FL by
analyzing time series of stable isotope (δ15N and δ13C) values in sequential growth layer
groups within teeth. The isotope data provide a chronology of foraging habits over the
lifetime of the individual and allowed us to show that female bottlenose dolphins exhibit
a high degree of individual specialization in both foraging habitat and trophic level. The
foraging habits used by adult females are similar to those they used as calves and may
be passed down from mother to calf through social learning. We also characterized the
foraging habits and home range of each individual by constructing standard ellipses
from isotope values and dolphin sightings data (latitude and longitude) respectively.
These data show that Sarasota Bay bottlenose dolphins forage within a subset of the
habitats in which they are observed. Moreover, females with similar observational
standard ellipses often possessed different foraging specializations. Female bottlenose
dolphins may demonstrate individual specialization in foraging habits because it
reduces some of the cost of living in groups, such as competition for prey.

52

INTRODUCTION
Predatory populations often meet their energetic needs by consuming prey from
multiple trophic levels or food webs. Some populations accomplish this when individuals
demonstrate a large degree of foraging diversity by exploiting a wide variety of prey
species and/or habitats (individual generalists). In other populations, individuals use
only a small subset of prey, or confines foraging to specific habitats (individual
specialists). Because individual specialization impacts intraspecific competition, social
interactions, and population viability, the degree to which a population is comprised of
individual specialists vs. individual generalists has ecological consequences (Bolnick et
al. 2003, Matich et al. 2010). Furthermore, as a consequence of foraging specialization,
certain individuals may be predisposed to decreased pathogen exposure, increased
reproductive success, or decreased risk of predation. Thus, understanding patterns of
intraspecific resource use is critical understanding population dynamics (Mann et al.
2000, Darimont et al. 2007, Johnson et al. 2009).
Carbon and nitrogen isotope values (δ13C and δ15N) are well-established tracers
of foraging habitat and trophic level, respectively (Harrigan et al. 1989, Newsome et al.
2010, Rossman et al. 2013). δ13C values often differ among primary producers (e.g.,
low values for open water phytoplankton vs. high values for seagrass) that form the
base of the food web and δ15N values increase with trophic level (Peterson and Fry
1987, Layman et al. 2011). Stable isotope values of metabolically inert tissues that are
formed over time (e.g., growth layers in teeth, scutes, vibrissae) provide a chronological
record of an individual’s habitat and trophic level (Mendes et al. 2006, Newsome 2009a,
2009b, Vander Zanden et al. 2010). Such isotope chronologies can be used to evaluate
53

the degree of foraging diversity within and between individuals and, hence, the degree
of specialization within a population. The degree of foraging diversity within and
between individuals is an indicator of total niche width (TNW). More precisely, TNW is
the variance in the δ13C or δ15N values of the population and is partitioned into a within
individual component (WIC) and a between individual component (BIC) (Bolnick et al.
2003, Newsome et al. 2009b). WIC is large in populations of individual generalists as
each member of the population uses a wide range of habitats and trophic levels (Fig.
9A). In populations of individual specialists, BIC predominates, as individuals are
invariant in their range of isotope values (Fig. 9B) (Bolnick et al. 2003, Newsome et al.
2009).
Identifying patterns of individual specialization may be particularly important for
coastal marine mammals which are subject to habitat loss and compete with humans for
prey resources. This is the case for bottlenose dolphins (Tursiops truncatus) in Sarasota
Bay, Florida (Rossman et al. 2013). This population is particularly amenable to an
investigation of foraging specializations because they are year-round, multigenerational
residents that have been extensively studied for more than 43 years (Wells 2003,
2014a, 2014b). Members of the population are generally of known age and reproductive
history, deceased individuals are the subject of stomach content and stable isotope
analyses, and an extensive observational program has been ongoing since the
inception of the Sarasota Dolphin Research Program in 1970 (Barros and Wells, 1998,
Wells 2003, 2009, 2014a, Berens McCabe et al. 2010, Rossman et al. 2013, 2014).
While Sarasota Bay bottlenose dolphins prefer certain prey types at the population level
(Berens McCabe et al. 2010)
54

Figure 9. The frequency of an isotope value over the lifetime of three individuals is
depicted by red, blue and yellow curves in panels A and B. Arrows at the bottom
of the curves reflect the degree of variation in isotope value over the lifetime of an
individual also termed the within individual component (WIC) of the population’s
isotope variation. Arrows between curves reflect the between individual
component (BIC) of the isotopic variation in the population. (A) Isotope frequency
curves for individual specialists, each with a narrow range in isotope values. In
this case BIC is larger than WIC. (B) Isotope frequency curves for individual
generalists, each with a wide range in isotope values. In this case BIC is much
smaller than WIC. (C) Microdrilled tracks (green lines) imposed over a photograph
of a bucco-lingual longitudinal section of a dolphin tooth. The table shows age
class for each track and hypothetical δ13C data for two individuals. (D) Histograms
for the hypothetical isotope data in panel C. The large BIC compared to WIC
indicates that these dolphins are individual specialists.

55

and differences in foraging habits exist among demographic groups (Rossman et al.
2014), the degree to which they demonstrate individual specialization remains
uncertain. Isotope analysis of sequentially micro-drilled tracks containing one or more
growth layers provides a record of the foraging habits (foraging habitat or trophic level)
of the individual over its lifetime (Fig. 9C). Low variance in isotope values among the
tracks of an individual is consistent with the use of a particular foraging habitat or trophic
level over time. In this case, WIC is small and BIC accounts for the majority TNW and
the data reflect individual specialization (Fig. 9D).
In this study, we assess the degree of individual foraging specialization within a
population of bottlenose dolphins resident to Sarasota Bay by establishing an isotope
chronology from teeth. Because previous work documented that relative to males,
female dolphins exhibited a range in trophic level and higher diversity in foraging
habitats, our investigation focuses on this group (Rossman et al. 2014). Additionally,
examining female bottlenose dolphins allowed us to evaluate the relationship between
foraging habit and reproductive success. To assess if an adult's foraging habit was predetermined by the habits developed as a calf, we modeled individual adult isotope
values as a function of their calf isotope values. Finally, we compared isotope and
sighting data to determine how foraging habits are related to the habitat associated with
the individual’s observed range.

56

METHODS
Sample site and acquisition
Sarasota Bay is a complex series of shallow bays (< 4m deep, 40 km long) on the
central west coast of Florida, communicating with the Gulf of Mexico through narrow,
deep passes separating a series of narrow barrier islands. The study area
encompasses a diverse array of habitats, including seagrass meadows, channels,
mangrove fringing forests, human-altered shorelines, and open bay (Wilkinson 2014).
All bottlenose dolphin tooth samples were obtained from females known to the Sarasota
Dolphin Research Program, recovered as carcasses by Mote Marine Laboratory’s
Stranding Investigations Program (Wells 2014b). Both sighting and reproductive
success data were collected as part of the Sarasota Dolphin Research Program’s long
term monitoring program (Wells 2003, 2009, 2014a).

Sample Preparation
Teeth were cut in buccal-lingual longitudinal sections using a water-cooled diamond
bladed saw (Buehler IsoMet). Discrete tracks of hydroxylapatite containing one or more
growth layer groups (GLGs, Hohn et al. 1989) were extracted with a computer guided
micromill (Merchantek) with a drill bit 1 mm in diameter. Five tracks were milled from
each tooth to produce hydroxylapatite samples weighing between 1.0 and 2.5 mg. The
small size of bottlenose dolphin teeth resulted in some tracks containing more than one
GLG. Tracks reflecting older portions of an individual’s lifetime (closer to the pulp cavity)
contained more GLGs compared to tracks originating near the neonate line. After
milling, teeth were visually inspected under a microscope to determine approximate age

57

classes for each track and validated by a second observer. Age class estimates for
tracks 1-5 were 0-1 yr, 1-2 yr, 2-4 yr, 4-8 yr, and >15 yr (referred to as “adult”),
respectively. Substantial wear on samples from FB 41 and FB 9 prevented the sampling
of age class one. Milled tooth powder was directly deposited into 9x5 mm silver
capsules (Costech). Hydroxylapatite was demineralized with 100 µL of 1 N HCl pipetted
into each capsule to isolate the organic portion of the tooth, dentin, which makes up
approximately 30% by dry weight (Koch 2007). Samples were left at 4oC over night then
dried in a 60oC oven. Isotope values were determined using an elemental analyzer
(Eurovector) interfaced to an Isoprime mass spectrometer (Elementar Americas Inc.).
Isotope values are expressed as:
δX = {(Rsample / Rstandard ) – 1} x 1,000
where X represents 13C or 15N, and R represents the abundance ratios:
15N/14N

13C/12C

or

respectively. In-house standards used for δ13C and δ15N were calibrated with

respect to international scales V-PDB and air respectively. In-house precision was +/0.2‰ for δ13C and δ15N values.
To ensure isotopic reproducibility, three teeth were sectioned longitudinally to
produce two equivalent halves per tooth. The isotope values of corresponding age
classes from each half of the tooth were analyzed and the average difference between
corresponding tracks was estimated from the data.

Statistical Analysis
As described below, we used a Bayesian approach to evaluate our statistical models.
Female bottlenose dolphin isotope values were modeled as a general linear mixed

58

model containing two categorical explanatory variables, individual and age class, in a
two-way ANOVA design. Because we were interested in generalizing the effect of the
variable “individual” to females in the bottlenose dolphin population of Sarasota Bay, it
was modeled as a random effect (Bolker et al. 2009). We also were interested in
estimating the foraging habits associated with specific bottlenose dolphin age classes,
and for this reason, the variable “age class” was incorporated into the model as a fixed
effect (Bolker et al. 2009). We also constructed reduced models to examine the
influence of the variables “age class” or “individual” on isotope values and identified the
most predictive model using deviance information criterion (DIC).
R2 values for the reduced model containing the variable “individual” are
statistically equivalent to BIC/TNW (Newsome et al. 2009). The ratio of BIC/TNW was
used as an indicator of the degree of individual specialization with values near 0
indicating that female bottlenose dolphins are generalists and values approaching 1
identifying a high degree of individual specialization.
We modeled the relationship between adult δ13C and δ15N values and
reproductive success using linear regression. Reproductive success was defined as the
number of calves a female produced surviving to age three divided by the number of
years from becoming reproductively active until death or from first being observed until
death. Our model of the influence of foraging habits on reproductive success is limited
by lack of information on environmental variables (e.g. red tide) and other covariables
(e.g. body condition) that also may be influential to reproductive success. The model
was bivariate because our a priori hypothesis was that foraging habit was the dominant

59

control on reproductive success and an assessment of numerous covariates is beyond
the scope of this study.
Another goal of this study was to determine if particular foraging habits used as a
calf predispose an individual to exploiting the same foraging habits as an adult. To
evaluate this possibility, we modeled adult isotope values as a function of their mean
calf isotope values, using a linear regression. Calf isotope value was determined by
averaging age classes 1-2 yr and 2-4 yr, the time period when a calf is likely to be in
close proximity to its mother (Wells 2003, Wells 2014a).
We examined individual foraging habits and home range by constructing two sets
of standard ellipses using the R package, SIAR (Parnell et al. 2008, Jackson et al.
2011, Rossman et al. 2014). δ13C and δ15N values were used to create standard
ellipses representing foraging habits over the lifetime of the individual (“isotope standard
ellipses”) and latitude and longitude from sightings were used to construct
“observational standard ellipses.” An individual’s mean δ13C and δ15N or mean latitude
and longitude define the center of the standard ellipse and the shape and orientation
are determined by a covariance matrix (Jackson et al. 2011). Observational standard
ellipses were plotted alongside a habitat map of Sarasota Bay to enable a visual
comparison of an individual’s sightings and Sarasota Bay habitat types (Wilkinson
2014).
While the benefits of Bayesian analysis of mixing models within stable isotope
ecology are well known (Parnell et al. 2008), Bayesian interpretation of common
hypothesis testing models such as ANOVA and regression is rare. For this reason, we
suggest scientists unfamiliar with Bayesian model analysis see Kery (2010). All models

60

in this study were assessed using a Bayesian hierarchical approach with the programs
R (R core development team, 2013) and Jags (Plumber, 2003). We ran three Markov
Chain Monte Carlo chains for 32,000 iterations with the first 2,000 discarded as burn in.
Bayesian analysis requires that priors be specified for model parameters describing the
current state of knowledge. We allowed parameter estimates to derive predominately
from the data by using vague, uninformative priors. Model convergence was monitored
via the R-hat statistic (Gelman and Hill, 2007). The R-hat statistic examines the
variance ratio of the MCMC algorithm within and between chains across iterations for
each parameter value. R-hat values near one indicate convergence and values below
1.2 are considered acceptable (Gelman and Hill, 2007). All R-hat values in this study
were below 1.01. We used the approach of Gelman et al. (1996), to calculate Bayesian
p-values which assess the adequacy of our models. We defined a discrepancy value, D,
(the summation of the square of the observed values minus the predicted values
obtained at each iteration of the MCMC algorithm). By simulating data sets from the
posterior distribution and computing the discrepancy measure, Dsim, a reference
distribution is computed for the simulated data set. The Bayesian p-values are the
probability: P(D>Dsim) (Gelman et al. 1996). Extreme values (less than 0.05 or greater
than 0.95) indicate inadequate modeling of the data. All Bayesian p-values in our study
were 0.5-0.6 indicating that our models provided an adequate description of the data.
Lastly, while model selection involving random effects models is difficult, we used DIC,
the most widely used goodness of fit criterion in Bayesian applications (Gelman et al.
2013, Tenan et al. 2014).

61

We express parameter estimates as a mean and credible interval (CI). The latter
represents the range which encompasses a 95% probability of containing the true
parameter value. For example, CI of the slope from a regression that does not contain
zero would indicate a “significant” relationship between the two variables tested. In this
study we also incorporated contrasts between successive age classes into our
Bayesian analysis. Differences between age classes were identified by probabilities. For
example, in a contrast to determining if age class 0-1 is larger than age class 1-2,
probability values greater than 0.95 (P>0.95) indicate age class 0-1 is larger than age
class 1-2 while low probabilities (P<0.5) indicate age class 1-2 is smaller than age class
0-1.

RESULTS
Tooth GLG stable isotope values ranged from 9.0‰ to 14.4‰ (mean ± SD: 11.1‰ ±
1.1‰) for δ15N and from -12.3‰ to -8.6‰ (mean ± SD: -10.5‰ ± 1.0‰) for δ13C across
all age classes. In our comparison of corresponding age classes of tooth halves, the
average differences in δ15N or δ13C was 0.4‰ ± 0.4‰ for three teeth. Given that these
data derive from the analysis of tooth halves; the total error is low relative to the
analytical reproducibility of 0.2‰. Among the ANOVA models for nitrogen isotope
values, DIC indicated the full model with individual and age class provided the best
predictive capacity (Fig. 10). The high R2 associated with individual (0.60) indicates a
large BIC/TNW compared to WIC/TNW (0.40) (Table 2). Of the 40% of the total isotopic
variance accounted for by WIC, 12% (relative to TNW) is explained by the variable “age
class.” Contrasts for δ15N values between subsequent age classes showed 0-1 yr was

62

significantly higher than 1-2 yr (P=1.0), 1-2 yr was significantly higher than 2-4 yr
(P=0.98), and age class >15 was significantly higher than 4-8 (P=0.99). For δ13C, age
class did not account for any of the variance in the data (Table 2). The ANOVA model

Table 2: ANOVA results for competing models where δ15N and δ13C values are a function
of individual and/or age class. Deviance information criterion (DIC) values are used for
model selection. The lowest DIC value identifies the model with the highest predictive
capacity. The R2 value indicates the amount of variation in the dependent variable
explained by the model. Our Bayesian treatment of the data provides a 95% CI for the R2
denoted by a lower bound (CI: 2.5%) and an upper bound (CI: 97.5%).
δ15N
model

DIC

R2

CI: 2.5%

CI: 97.5%

individual + age class

128.90

0.81

0.76

0.84

individual

175.50

0.60

0.52

0.66

age class

220.90

0.11

0.01

0.19

δ13C
model

DIC

R2

CI: 2.5%

CI: 97.5%

individual

71.50

0.88

0.85

0.90

individual + age class

80.00

0.88

0.85

0.90

age class

216.90

-0.07

-0.19

0.00

with only “individual” provided the highest predictive capacity, where the R2 indicates a
large BIC/TNW and small WIC/TNW with respect to foraging habitat (Fig. 11, Table 2).
Reproductive success was not related to adult δ15N or δ13C values (slope CI: 0.38 – 0.10, -0.10 – 0.41, respectively).
The 95% CI for the slopes of the regressions describing adult isotope value as a
function of calf isotope value did not contain zero, indicating a significant relationship
between the two variables (Fig. 12).
63

Figure 10. δ15N values from bottlenose dolphin teeth for five age classes
covering the lifetime of the individual. Age class >15 is adult. Age classes were
determined from bottlenose dolphin tooth ontogeny. Individuals are differentiated
by color and Sarasota Dolphin Research Program identification codes given in
the figure legend. The inset shows average δ15N values for each age class
(closed triangles) and associated credible intervals (bars) from Bayesian
ANOVA.

64

Figure 11. δ13C values from bottlenose dolphin teeth for five age classes
covering the lifetime of the individual. Age class >15 is adult. Age classes
were determined from bottlenose dolphin tooth ontogeny. Individuals are
differentiated by color and Sarasota Dolphin Research Program identification
codes given in the figure legend.

65

Figure 12. An individual’s isotope value as an adult plotted against its
isotope value as a calf, A) carbon isotope values and B) nitrogen
isotope values. Calf δ13C or δ15N value is the average of age classes 12 yr and 2-4 yr and adult isotope values is age class > 15 yr. Individuals
are differentiated by color (same color scheme as earlier figures) and
Sarasota Dolphin Research Program identification codes given in the
figure legend. Regression equations are shown in the bottom right of
each panel. For δ15N the 95% credible interval for the slope was 0.381.06 and for the intercept it was -0.56-4.44. For δ13C the 95% credible
interval for the slope was 0.72-1.14 and for the intercept it was -0.861.34. The 95% credible intervals for the slopes are depicted on the
graphs as grey dashed lines. R2 values and their 95% confidence
intervals are shown in the lower right of each graph. The R2 value
documents the proportion of the variance in adult isotope values that
can be explained by calf isotope values.

66

Individual standard ellipses clustered into three distinct groups. Group 1
contained 7 individuals with high δ15N values and low δ13C values relative to
the average isotope value for the population (Fig. 13). In comparison to the average
isotope values of the population, Group 2 contained 6 individuals with lower δ15N values
and higher δ13C values and Group 3 contained only two individuals who exhibited
intermediate and low δ13C and δ15N values, respectively.
Observational standard ellipses showed that while some individuals used nearly
the entire study area, others were only observed in the northern or southern portion of
Sarasota Bay (Fig. 14). For example, the observational standard ellipses of FB 9 and
FB 57 extend over the northern portion of Sarasota Bay near Palma Sola Bay while FB
195 and Bardot (Group 3) predominately use the southern portion of our study area,
Little Sarasota Bay. Even though northern Sarasota Bay has extensive seagrass
meadows, dolphins using this portion of Sarasota Bay still interfaced with multiple
habitat types.

67

Figure 13. Isotope standard ellipses representing variation in δ13C and δ15N values
for each individual over their lifetime. Mean δ13C and δ15N values for each individual
define the center of the standard ellipse and the shape and orientation are
determined by a covariance matrix. Individuals appear in three non-overlapping
groups. Compared to the population mean (-10.5‰ and11.1‰ for δ13C and δ15N
respectively) Group 1 is defined by high δ15N values and low δ13C values, Group 2
has low δ15N values and high δ13C values, and Group 3 has low δ15N values and
intermediate δ13C values. Individuals are differentiated by color using the same
scheme as in earlier figures. Sarasota Dolphin Research Program identification
codes given in the figure legend.

68

Figure 14. Maps of Sarasota Bay depicting habitat distribution (left) and observational standard ellipses generated
from the latitude and longitude data from sightings (right). Data for habitat use was obtained from Wilkinson (2014).
Inset depicts location of Sarasota Bay in relation to the state of Florida. Colors represent different habitats in Sarasota
Bay as defined in the figure legend. Individuals in the observational standard ellipse maps are differentiated by color
(same color scheme as earlier figures) and Sarasota Dolphin Research Program identification codes given in the
figure legend.

69
69

DISCUSSION
Our data indicate significant variation in trophic level over the lifetime of female
Sarasota Bay bottlenose dolphins but little variation in habitat use. We also show that
variation within the population predominately derives from individual specialization as
evidenced by large BIC values. Despite large differences among individuals, foraging
habits were not correlated with reproductive success. This may suggest that multiple
foraging strategies are successful in providing female bottlenose dolphins adequate
nutrition to reproduce, or that other covariates (e.g. environmental variables) are more
influential on reproductive success.
Ontogenetic changes in trophic level and differences between individual
bottlenose dolphins can be inferred from δ15N values (Fig. 10). Nearly every individual
within the population showed similar trends in δ15N over their lifetime. These include a
decrease in δ15N between age classes 0-1 yr and 2-4 yr and an increase in δ15N
between age class 4-8 yr and >15 yr (adult). The early-in-life decrease is likely related
to weaning. While consuming milk, calves are feeding at one trophic level above their
mothers. For bottlenose dolphins, we expect an increase of 2.0‰ for each trophic level
(Rossman et al. 2014). Thus, the δ15N value of calf tissue formed while exclusively
consuming milk should be 2.0‰ higher than that of their mother. The decline in δ15N
between age classes 0-1 yr and 2-4 yr, 1.4‰, was less than a full trophic level
suggesting that calves derive nutrition from sources other than milk as early as their first
year. This is consistent with observational data indicating that calves supplement their
diet with fish within the first few months of life (Wells, 2003). Bottlenose dolphin δ15N
values reach a minimum by age class 2-4 yr, in agreement with observational findings

70

that calves obtain the vast majority of their nutrition from fish by 2 years of age (Perrin
and Reilly 1984, Wells and Scott, 1999, Wells 2003, Wells 2014a). In addition to the
early decline, we observed an increase in trophic level between age classes 4-8 yr and
>15 yr. This result parallels previous demographic comparisons showing that juvenile
dolphins (less than 6 yrs old) feed at a lower trophic level than adult females (Rossman
et al. 2014). Previous studies of sperm whales, killer whales and Commerson’s dolphins
which documented a similar increase in δ15N values with age, attributed the trend to
many factors including changes in trophic fractionation due to growth rate, a change in
diet resulting from leaving the natal band or offshore migration, or an increase in the
size of prey consumed (Mendes et al. 2006, Newsome et al. 2009, Riccialdelli et al.
2013). We cannot exclude isotope effects resulting from changes in growth rate.
However, the observations over decades that Sarasota Bay females associate with their
natal band and remain coastal residents throughout their life (Wells 2003, Wells 2014a),
indicates that migration does not influence age related trends in δ15N for female
Sarasota Bay bottlenose dolphins. The low δ15N value of age classes 2-4 yrs and 4-8
yrs may result from reliance on pinfish, the lowest trophic level diet item consumed by
Sarsota Bay dolphins (Rossman et al. 2014). The increase in female bottlenose dolphin
δ15N values in the >15 yrs age class likely reflects the ability to consume larger, higher
trophic level fish as the result of larger gape size or development of foraging skills
(Mendes et al. 2006, Riccialdelli et al. 2013, Rossman et al. 2014).
Unlike trophic level, δ13C values suggest that there is no ontogenetic trend in
foraging habitat. However, the large BIC/TNW, indicates that individuals use only a
subset of the foraging habitats available to the population. Individual specialization of

71

bottlenose dolphins in habitat and trophic level, may result from social learning
(Sargeant and Mann 2009, Tinker et al. 2009). Social learning may be facilitated when
there are extended periods of association between mothers and offspring (Wells 2003,
2014a, Sargeant and Mann 2009). Although some may be capable of nutritional
independence after their first year of life, calves generally stay in close proximity to their
mothers until 3 to 6 yrs old (Wells, 2003, 2014a). The long period of mother-calf
association has been hypothesized as critical to the social development of the calf
(Wells et al. 2003, 2014a, Sargeant and Mann 2009). Moreover, the close association
between mother and calf provides opportunity for vertical transmission of individual
foraging specializations (Tinker et al. 2009). Our results showing that the isotope values
of adults are predicted by those of calves support the role of social learning in the
development of calf foraging skills.
To further explore variation in foraging habits among bottlenose dolphins from
Sarasota Bay, we constructed isotope standard ellipses. Standard ellipses describe
variation in foraging habits over a dolphin’s lifetime and, are thus, an indicator of
individual niche width. This construct is particularly appropriate for Sarasota Bay
bottlenose dolphins, where isotope values have been interpreted to reflect trophic level
and foraging habitat is well documented (Rossman et al. 2013).
Standard ellipses of females span a broad range of δ13C and δ15N values (ca.
3‰; Fig. 13). With the exception of two dolphins, individuals cluster into two groups
(groups 1 and 2), with non-overlapping δ13C values and a wide spread in δ15N (ellipse
center range of 2.5‰). This observation parallels our previous study describing prey
isotope values (Rossman et al. 2014). Prey were clustered into four groups

72

characterized by either high or low δ13C values, reflecting seagrass and nonseagrass
habitats (i.e. channel, sand flat, open water, mangrove), respectively. Each prey cluster
was also distinguished by a large range in δ15N values, indicating high trophic diversity
within both seagrass and nonseagrass habitats. Because δ13C values are transferred up
the food web, unique δ13C values of the two bottlenose dolphin ellipse clusters
differentiate individuals that forage in seagrass and nonseagrass habitats. Furthermore,
the large variation in δ15N exhibited by the two main groups suggests that individuals
specialize on prey from different trophic levels.
Relative to clusters 1 and 2, the ellipse center of cluster 3 is intermediate and the
δ15N values are lower. Unlike the other clusters, individuals from cluster 3 are from Little
Sarasota Bay in the southern portion of our study area. While it is difficult to draw
meaningful conclusions from two individuals, their relatively low δ13C and δ15N values
may be associated with increased foraging on low trophic level, nonseagrass prey such
as clupeids (Rossman et al. 2014). Little Sarasota Bay is poorly flushed compared to
Sarasota Bay and may possess different assemblages of prey fish (Sheng and Peene
1992). Thus, the unique isotope values of cluster 3 may reflect the response of
bottlenose dolphins to differences in habitat and prey availability.
To determine if individual foraging specializations result from habitat
heterogeneity, we compared our interpretations of foraging habits from isotope standard
ellipses to observational standard ellipses and habitat distribution maps. Observational
standard ellipses show that the home range of each individual bottlenose dolphin
circumscribes a range of habitats. For example, FB 59 was observed in seagrass,
mangrove and open water. Yet, her isotope standard ellipse indicates that FB 59 tended

73

to forage in nonseagrass habitat. Clusters 1 and 2 differentiate on the basis of foraging
habitat, yet their observational standard ellipses demonstrate largely overlapping habitat
use. This is illustrated by FB 19 and FB 111 who have similar observational standard
ellipses, yet their isotope standard ellipses indicate unique foraging habits. Relative to
FB 111, FB 19 had a greater tendency to forage in seagrass. Thus, two individuals who
shared nearly identical ranging patterns can possess very different foraging habits.
While bottlenose dolphins are observed in many different food webs, they do not
indiscriminately forage but rather specialize on a subset of the habitats used by the
population.
Heterogeneity in habitat type within a confined geographic range is thought to
promote foraging generalization (Miller et al. 2009, Montevecchi et al. 2009, Matich et
al. 2011). Yet, in Sarasota Bay, where seagrass, mangrove and phytoplankton based
food webs are largely overlapping, bottlenose dolphins demonstrate significant
individual specialization. Enhanced foraging efficiency and intraspecific competition
likely promote individual specialization even despite diversity of food webs within
Sarasota Bay. By consistently utilizing a particular foraging strategy, dolphins likely
enhance their efficiency at capturing and consuming prey. Switching to a new habitat
may require a unique set of skills costing time and effort to learn (Tinker et al. 2009).
Furthermore, results from our study support the findings of McHugh et al. (2011) and
indicate that dolphins continue to use their natal foraging habitat after independence
from their mother, a trend that may result from the experience and skills gained specific
to their natal habitat. However, our data set contained two mother-daughter pairs FB 59,
daughter of FB19, and FB 9 daughter of FB 63 possessing dissimilar isotope values.

74

Although sample size is low, in both cases mother and daughter were in different
groups indicating that the development foraging habits may be complex.
In addition to foraging strategy and skill development, the social structure of
female bottlenose dolphins may promote individual specialization. Females live in fluid
social groups (bands) often associating with individuals of a similar reproductive status
(Wells 2003, 2014a). While living in a group may confer many benefits (predator
avoidance, shared rearing of offspring, observational learning opportunities for
offspring), it also potentially results in increased competition for prey resources (Connor
et al. 2000). The observation that females with nearly identical home ranges often have
dramatically different foraging habits may indicate one way dolphins decrease
intraspecific competition within female bands. This partitioning of resources would allow
females the benefits of social living while negating some intraspecific competition.
The high degree of specialization in the female dolphins of Sarasota Bay has
management and conservation implications. We show that individual dolphins are not
ecologically equivalent and each dolphin contributes to the large diversity of foraging
habits observed at the population level. The loss of even a few individuals could result
in a reduction in the foraging diversity for the Sarasota Bay population. The significance
of this relates to the transmission of foraging habits from mother to calf through social
learning whereby the removal of individuals may result in the loss of specific foraging
habits for generations. For instance, the loss of individuals from group 1 (nonseagrass
specialists) coupled with a seagrass die-off may destabilize the Sarasota Bay dolphin
population and jeopardize population viability. Furthermore, because each bottlenose
dolphin may possess unique foraging habits, the prevalence of human induced mortality

75

of bottlenose dolphins is of particular concern (Wells et al. 2008, Wells et al. 2014b).
When a bottlenose dolphin dies as the result of the ingestion of fishing line,
entanglement or other preventable causes it is difficult to evaluate the loss in ecological
diversity and the impact it will have on future generations of bottlenose dolphins, without
knowledge of individual specialization as we have presented herein.

76

LITERATURE CITED

77

LITERATURE CITED

Barros, N. B., and R. Wells. 1998. Prey and feeding patterns of resident bottlenose
dolphins (Tursiops truncatus) in Sarasota Bay, Florida. Journal of Mammalogy
79:1045-1059.
Berens McCabe, E., D. Gannon, N. Barros, and R. Wells. 2010. Prey selection by
resident common bottlenose dolphins (Tursiops truncatus) in Sarasota Bay,
Florida. Marine Biology 157:931-942.
Bolker, B. M., M. E. Brooks, C. J. Clark, S. W. Geange, J. R. Poulsen, M. H. H. Stevens,
and J.-S. S. White. 2009. Generalized linear mixed models: a practical guide for
ecology and evolution. Trends in Ecology & Evolution 24:127-135.
Bolnick, D. I., R. Svanback, J. A. Fordyce, L. H. Yang, J. M. Davis, C. D. Hulsey, and M.
L. Forister. 2003. The ecology of individuals: Incidence and implications of
individual specialization. American Naturalist 161:1-28.
Connor, R. C. 2000. Group living in whales and dolphins. Pages 199-218 in J. Mann, R.
C. Connor, P. L. Tyack, and H. Whitehead, eds. Cetacean societies: field studies
of dolphins and whales. The University of Chicago Press, Chicago, IL.
Darimont, C. T., P. C. Paquet, and T. E. Reimchen. 2007. Stable isotopic niche predicts
fitness of prey in a wolf-deer system. Biological Journal of the Linnean Society
90:125-137.
Gelman, A., and J. Hill. 2006. Data analysis using regression and multilevel/hierarchical
models. Cambridge University Press, Cambridge, UK.
Gelman, A., J. Hwang, and A. Vehtari. 2013. Understanding predictive information
criteria for Bayesian models. Statistics and Computing, 1–20. doi:10.1007/s11222013-9416-2
Gelman, A., X. L. Meng, and H. Stern. 1996. Posterior predictive assessment of model
fitness via realized discrepancies. Statistica Sinica, 6: 733–760.
Harrigan, P., J. C. Zieman and S. A. Macko. 1989. The base of nutritional support for
the gray snapper (Lutjanus griseus): an evaluation based on a combined
stomach content and stable isotope analysis. Bulletin of Marine Science 44:6577.
Hohn, A.A., M.D. Scott, R.S. Wells, J.C. Sweeney and A.B. Irvine. 1989. Growth layers
in teeth from known-age, free-ranging bottlenose dolphins. Marine Mammal
Science 5(4):315-342.
78

Jackson, A. L., R. Inger, A. C. Parnell, and S. Bearhop. 2011. Comparing isotopic niche
widths among and within communities: SIBER–Stable Isotope Bayesian Ellipses
in R. Journal of Animal Ecology 80:595-602.
Johnson, C. K., M. T. Tinker, J. A. Estes, et al. Prey choice and habitat use drive sea
otter pathogen exposure in a resource-limited coastal system. Proceedings of the
National Academy of Sciences 106:2242-2247.
Kery, M. 2010. Introduction to WinBUGS for Ecologists: Bayesian approach to
regression, ANOVA, mixed models and related analyses. Academic Press.
Koch, P.L. 2007. Isotopic study of the biology of modern and fossil vertebrates. Pages
99–154 in R. Michener and K. Lajtha, eds. Stable isotopes in ecology and
environmental science. Blackwell Scientific Publications, Boston, MA.
Layman, C. A., M. S. Araujo, R. Boucek, et al. 2011. Applying stable isotopes to
examine food-web structure: an overview of analytical tools. Biological Reviews
87:545-562.
Mann, J., R. C. Connor, L. M. Barre, and M. R. Heithaus. 2000. Female reproductive
success in bottlenose dolphins (Tursiops sp.): life history, habitat, provisioning,
and group-size effects. Behavioral Ecology 11:210-219.
Mann, J. and B. Sargeant. 2003. Like mother, like calf: the ontogeny of foraging
traditions in wild Indian Ocean bottlenose dolphins (Tursiops sp.). Pages 236-266
in D. Fragaszy and S. Perry, eds. The biology of traditions: models and evidence.
Cambridge University Press, Cambridge, UK.
Matich, P., M. R. Heithaus, and C. A. Layman. 2011. Contrasting patterns of individual
specialization and trophic coupling in two marine apex predators. Journal of
Animal Ecology 80:294-305.
McHugh, K. A., J. B. Allen, A. A. Barleycorn, and R. S. Wells. 2011. Natal philopatry,
ranging behavior, and habitat selection of juvenile bottlenose dolphins in
Sarasota Bay, Florida. Journal of Mammalogy 92:1298-1313.
Mendes, S., J. Newton, R. J. Reid, A. F. Zuur, and G. J. Pierce. 2006. Stable carbon
and nitrogen isotope ratio profiling of sperm whale teeth reveals ontogenetic
movements and trophic ecology. Oecologia 151:605-615.
Miller, A. K., N. J. Karnovsky, and W. Z. Trivelpiece. 2009. Flexible foraging strategies
of gentoo penguins Pygoscelis papua over 5 years in the South Shetland Islands,
Antarctica. Marine Biology 156:2527-2537.

79

Montevecchi, W. A., S. Benvenuti, S. Garthe, G. K. Davoren, and D. Fifield. 2009.
Flexible foraging tactics by a large opportunistic seabird preying on forage-and
large pelagic fishes. Marine Ecology Progress Series 385:295-306.
Newsome, S. D., M. T. Clementz, and P. L. Koch. 2010. Using stable isotope
biogeochemistry to study marine mammal ecology. Marine Mammal Science
26:509-572.
Newsome, S. D., M. A. Etnier, D. H. Monson, and M. L. Fogel. 2009a. Retrospective
characterization of ontogenetic shifts in killer whale diets via δ 13C and δ15N
analysis of teeth. Marine Ecology Progress Series 374:229-242.
Newsome, S. D., M. T. Tinker, D. H. Monson, et al. 2009b. Using stable isotopes to
investigate individual diet specialization in California sea otters (Enhydra lutris
nereis). Ecology 90:961-974.
Parnell, A., R. Inger, S. Bearhop and A. L. Jackson. 2008. SIAR: stable isotope analysis
in R. Available at www.cran.r-project.org/web/packages/siar/index.html.
Perrin, W. F. and S. B. Reilly. 1984. Reproductive parameters of dolphins and small
whales of the family Delphinidae. Report of the International Whaling
Commission, Special: 97-134.
Peterson, B. J., and B. Fry. 1987. Stable Isotopes in Ecosystem Studies. Annual
Reviews in Ecology and Systematics 18:293-320.
Plummer, M. 2003. JAGS: a program for analysis of Bayesian graphical models using
Gibbs Sampling. Proceedings of the Third International Workshop on Distributed
Statistical Computing. R Project for Statistical Computing, Vienna, Austria.
Riccialdelli, L., S. D. Newsome, N. A. Dellabianca, R. Bastida, M. L. Fogel, and R. N. P.
Goodall. 2013. Ontogenetic diet shift in Commerson's dolphin (Cephalorhynchus
commersonii commersonii) off Tierra del Fuego. Polar Biology 36:617-627.
Rossman, S., N. B. Barros, P. H. Ostrom, C. A. Stricker, A. A. Hohn, H. Gandhi, and R.
S. Wells. 2013. Retrospective analysis of bottlenose dolphin foraging: A legacy of
anthropogenic ecosystem disturbance. Marine Mammal Science 29:705-718.
Rossman, S., E. Berens McCabe, N. B. Barros, H. Gandhi, P. H. Ostrom, C. A. Stricker,
and R. S. Wells. 2014. Foraging habits in a generalist predator: Sex and age
influence habitat selection and resource use among bottlenose dolphins
(Tursiops truncatus). Marine Mammal Science.
Sargeant, B. L. and J. Mann. 2009. From social learning to culture: Intrapopulation
variation in bottlenose dolphins. Pages 152-173 in K. N. Laland and B. G. Galef,
eds. The question of animal culture. Harvard University Press, Cambridge, MA.
80

Sheng, Y.P., Peene, S. 1992. Circulation and its effect on water quality. Sarasota Bay
National Estuary Program "1992 Framework for Action".
Spiegelhalter, D. J., N. G. Best, B. P. Carlin, and A. Van Der Linde. 2002. Bayesian
measures of model complexity and fit. Journal of the Royal Statistical Society:
Series B (Statistical Methodology) 64:583-639.
R Development Core Team. 2013. R: A Language and Environment for Statistical
Computing. R Foundation for Statistical Computing, Vienna, Austria.
Tenan, S., R. B. O’hara, I. Hendriks, and G. Tavecchia. 2014. Bayesian model
selection: The steepest mountain to climb. Ecological Modelling 283:62-69.
Tinker, M. T., M. Mangel and J. A. Estes. 2009. Learning to be different: acquired skills,
social learning, frequency dependence, and environmental variation can cause
behaviourally mediated foraging specializations. Evolutionary Ecology Research
11:841-869.
Vander Zanden, H. B., K. A. Bjorndal, and A. B. Bolten. 2013. Temporal consistency
and individual specialization in resource use by green turtles in successive life
stages. Oecologia, 173: 767–777.
Wells, R. S. 2003. Dolphin social complexity: Lessons from long-term study and life
history. Pages 32–56 in F. B. M. de Waal, and P. L. Tyack, eds. Animal social
complexity: Intelligence, culture, and individualized societies. Harvard University
Press, Cambridge, MA.
Wells, R. S. 2009. Learning from nature: bottlenose dolphin care and husbandry. Zoo
Biology 28:1-17.
Wells, R. S. 2014a. Social structure and life history of common bottlenose dolphins near
Sarasota Bay, Florida: Insights from four decades and five generations. Pages
149–172 in J. Yamagiwa and L. Karczmarski, eds. Primates and cetaceans: Field
research and conservation of complex mammalian societies. Springer, Tokyo,
Japan.
Wells, R. S., J. B. Allen, G. Lovewell, J. Gorzelany, R. E. Delynn, D. A. Fauquier, and N.
B. Barros. 2014b. Carcass-recovery rates for resident bottlenose dolphins in
Sarasota Bay, Florida. Marine Mammal Science.
Wells, R. S., and M. D. Scott. 1999. Bottlenose dolphins Tursiops truncatus (Montagu,
1821). Pages 137-182 in S. H. Ridgway and R. Harrison, eds. Handbook of
Marine Mammals, vol 6, The Second Book of Dolphins and Porpoises. San
Diego Academic Press, San Diego, CA.

81

Wilkinson, K. 2014. An analysis of shark bites on resident bottlenose dolphins
(Tursiops truncatus) in Sarasota Bay, Florida and implications for habitat use.
Master's Thesis. University of Florida, Gainesville.

82