Get Complete Project Material File(s) Now! »

Ecological and diet indicators

Stomach content examination can inform on the diet of an individual at the species level as well as on the characteristics (length and mass) of individual prey using allometric relationships based on hard tissues like otoliths. Stomach contents indicate the diet of an animal over the last few days. One major limitation of this technique is the digestion rate that can vary for different prey species. In particular, some species can be overestimated because of the persistence of their hard pieces in the stomachs as they are difficult to digest (Santos et al. 2001a). In addition, analyses can only be performed on dead animals.
The stomach contents of stranded animals have been used to study the diet of numerous cetacean species (e.g. Santos et al. 1999; Spitz et al. 2006). There is some uncertainty if stomach contents of stranded animals are representative of the diet of alive wild individuals. The physical condition of individuals may indeed affect their foraging capacities and some classes such as young or old individuals may eventually be over-represented in strandings. However, in Florida, results on prey species composition obtained using stomach content analyses on stranded animals and using molecular identification of prey in feces and gastric samples of free-ranging dolphins were highly consistent (Dunshea et al. 2013). In this dissertation, stomach contents are used, in complement with stable isotope analyses, to understand the foraging ecology of bottlenose dolphins in Chapter 6.
In the environment, natural elements can be found in different isotopic forms. Isotopes of any given chemical element have different number of neutrons, thus their atomic mass is different. Therefore, in biogeochemical reactions, the heavy isotopes accumulate in substrates as they react slower than light isotopes while products are depleted in heavy isotopes (Figure 2.3). This process, called the isotopic fractionation, controls isotope distribution (ratio of heavy to light isotopes) in the environment (Fry 2006).
In ecology, stable isotope analyses are indirect tools to study foraging ecology. There are based on the principle “you are what you eat”, that is the biochemical composition of the tissue of a consumer is linked to the one of its prey (Kelly 2000). δ13C (13C/12C) and δ34S (34S/32S) vary according to primary producers. In the marine environment, δ13C and δ34S indicate consumer foraging habitats such as inshore vs offshore or pelagic vs benthic habitats. δ13C also vary along latitudinal gradients (Peterson & Fry 1987; Kelly 2000; Connolly et al. 2004). δ34S do not vary between consumers and prey and δ13C vary little with increasing trophic level (generally less than 1 ‰, see review in Peterson & Fry 1987). In contrast, 15N is preferentially accumulated in Ωthe tissues of the consumers relative to their diet, therefore an average enrichment of 3 to 4 in δ15N (15N/14N) is generally observed with each increasing trophic level (see review in Kelly 2000). δ15N is therefore used as an indicator of trophic position. It can also reflect feeding areas in some ecosystems (e.g. inshore vs offshore in the Bay of Biscay, Chouvelon et al. 2012).
The turn-over rate of stable isotopes in a given tissue depends on the tissue metabolic rate. Therefore, stable isotopes are integrated over different time scales in different tissues (Tieszen et al. 1983; Hobson & Clark 1992). For example, in plasma, stable isotopes will inform on the diet and habitat use during the last few days preceding the tissue sampling (e.g. Podlesak et al. 2005) and in skin or muscle during several weeks to months (e.g. Tieszen et al. 1983; Browning et al. 2014). In hard tissues, like teeth, bones, whiskers or baleen plates, stable isotopes are integrated over the entire life of the individuals (e.g. Best & Schell 1996; Estrada et al. 2006; Mendes et al. 2007; Kernaleguen et al. 2012). The integration time of a specific soft tissue can also vary according to the species considered as metabolic rates are also species-specific (MacAvoy et al. 2006). One drawback of this method is that interpretation might be difficult especially if the baseline values of the ecosystems are not known (reviewed in Ramos & Gonzalez-Solis 2012). For instance, similar stable isotope signatures could be the result of a similar diet in the same habitat or a dissimilar diet in distinct habitats that have the same baseline values.
Stable isotopes have numerous applications in ecology and environment studies. To cite only a few examples, stable isotopes have been used to identified foraging habitats and migration patterns in a wide range of taxa (i.e. insects, fish, birds or mammals, see review in Rubenstein & Hobson 2004). By comparing stable isotopes in consumers and potential prey, or applying stable isotope mixing models on predator and prey data, it is possible to estimate the diet of a predator (e.g. Cherel et al. 2008; Huckstadt et al. 2012; Watt et al. 2013). As stable isotopes reflect habitat use and diet composition, stable isotope analyses can also help to determine population structure (e.g. Rooker et al. 2008a; Olin et al. 2012; Rioux et al. 2012; Wilson et al. 2012). Stable isotope signatures could be used as proxies of ecological niches (Newsome et al. 2007; Jackson et al. 2011). The ecological niche has been defined by Hutchinson (1957, 1978) as an n-dimensional hyper-volume with biotic and abiotic environment and resource variables as axes. These axes may be quantified by stable isotope signatures of δ15N and δ13C (or others such as δ34S) as they inform on either or both trophic level and environment and resource uses (Bearhop et al. 2004; Newsome et al. 2007; Jackson et al. 2011). Although the limits of stable isotope analyses should be recognized (i.e. see
above and complex physiological processes may influence stable isotope tissue composition), isotopic niches can therefore be used to investigate ecological niches (Newsome et al. 2007). In Chapters 4 and 6 stable isotopes are used as indicators of foraging ecology and habitat use (i.e. ecological niches) as well as tools to investigate population structure.


Morphometrics is the quantitative analysis of the size and/or the shape of an organism. They can be used, together with other morphological characters (e.g. coloration patterns) and genetic analyses, to separate species (e.g. short-finned and long-finned pilot whales are distinguished with the ratio of the length of the pectoral fin to the total length of the body along with the number of teeth per half jaw, Van Bree 1971; Robineau 2005). They can also be used in evolutionary ecology studies to understand how environmental conditions might influence morphological traits such as body size, size of appendices or cranial traits on short to evolutionary time scales (e.g. Grant & Grant 2002; Viaud-Martinez et al. 2007; Berner et al. 2010; Rode et al. 2010). For instance, body length can strongly be constrained by environmental conditions. Decreased body length in a polar bear population over two decades was correlated with a decline in sea-ice habitat availability (Rode et al. 2010). A rapid increase in body length in a population of fur seals may be the result of selective processes, in particular as bigger individuals have higher reproductive success (Authier et al. 2011). Resource polymorphism may also shape morphological traits (Smith & Skúlason 1996). The shape and size of beaks of darwin’s finshes varied across years probably because of variations in the availability of their food, i.e. seeds of different sizes, and the presence of competitors (e.g. Grant & Grant 2002, 2006). Although morphological trait evolutions for the latter examples were rapid, morphological divergence may be constrained by time. For example, Canada lake and stream threespine sticklebacks, that originated thousands years ago, are highly morphologically differentiated. In contrast, European lake and stream individuals were weakly morphologically distinct, possibly as a result of time constraints on divergence, as they originated less than 150 years ago. Nevertheless, at least some traits have evolved on a contemporary basis (Berner et al. 2010). In cetaceans, morphological variations are observed for example between open oceans and enclosed seas such as “dwarfism” for bottlenose dolphins and harbor porpoises in the Black Sea (Perrin 1984; Viaud-Martinez et al. 2007; Viaud-Martinez et al. 2008), which may have evolved on an evolutionary time scale. Thus, variations in morphological characters may reveal adaptations to different resource use both in terms of habitats and diets, and can therefore be an indicator of population structure (Perrin 1984). For instance, offshore and coastal bottlenose dolphin ecotypes in the North-East Pacific and in the North-West Atlantic differ in skull features (Hoelzel et al. 1998b; Perrin et al. 2011). In addition, apical tooth wear differ between weakly genetically differentiated killer whale specialists and generalists in the North-East Atlantic (e.g. Foote et al. 2009).
In Chapter 6, morphometric analyses are carried out to characterize bottlenose dolphin ecotypes in the North-East Atlantic.

Molecular markers: mitochondrial DNA and microsatellites

Mitochondrial DNA is a small circular molecule which is present in numerous copies in animal cells. It is haploid and mostly maternally inherited although heteroplasmic individuals (i.e. for which mitochondrial DNA was biparentally inherited) can be observed in different proportions in some taxa (e.g. Zouros et al. 1994; Vollmer et al. 2011). As it is haploid, there is generally no recombination (but see Eyre-Walker 2000; Ujvari et al. 2007). Evolution rate is five to ten times faster than nuclear DNA in mammals (Moritz et al. 1987), with an average mutation rate of 1 x 10-8 per site per year, making it useful in population genetics and phylogenetic studies. Mitochondrial DNA is composed by different regions which have different evolution rates including the control region which is the most variable and rapidly evolving part and thus of interest for population genetic studies. Estimates of mutation rates for the control region of cetaceans vary from 0.5 x 10-8 to 1.3 x 10-6 per site per year (Hoelzel et al. 1991; Harlin et al. 2003; Alter & Palumbi 2009; Fontaine et al. 2010).
As it is haploid and maternally inherited, effective population size at mitochondrial loci is four times lower than at nuclear loci. Mitochondrial genome is therefore more sensitive to genetic drift and integrates demographic events like population expansions or bottlenecks* since a longer time than nuclear markers. Polymorphism in the sequence is detected through sequencing. Each haplotype is a unique sequence. Different haplotypes differ by one or more nucleotides because of substitutions, deletions or insertions.

Non-bayesian clustering methods

Both TESS and STRUCTURE rely on genetic model assumptions (e.g. Hardy- Weinberg and Linkage Equilibria) and are therefore based on an “idealized” population model. With large datasets, they may require long computational times, due to the nature of MCMC simulations, in particular for STRUCTURE. For example, the MCMC may need tens of thousands of steps to reach convergence. In addition, an initial portion of the MCMC should be discarded to avoid the influence of initial values on the posterior distributions. DAPC (Discriminant Analysis of Principal Components) is an alternative method that does not rely on any genetic model assumptions (Jombart et al. 2010). It tries to cluster individuals based on genetic similarity, with genotypes being treated like a classical multivariate dataset. In DAPC, the number of clusters is first determined using a K-means method that aims at determining populations of individuals by minimizing within-population genetic variation. As in the Bayesian clustering methods, the K-means algorithm is ran with different numbers of putative populations. BIC (Bayesian Information Criterion) is used to determine the most likely number of populations. Then, the data are transformed using a Principal Component Analysis which summarizes the overall variability among individuals both among and within populations. This step ensures that the numbers of variables (i.e. alleles) are lower than the number of individuals and that the variables are not correlated. The Discriminant Analysis is applied on the Principal Components; it aims at partitioning genetic variation so that among-population variation is maximized while within-population variation is minimized. Individuals are assigned probabilistically to each population. DAPC has the advantage to have a fast computational time, even for large datasets. In addition, it has been shown to be as efficient as STRUCTURE (Jombart et al. 2010). DAPC also provides a visual representation of the structure between the populations, i.e. the scatterplots, which helps to understand the patterns of genetic structure (see Figure 2.6, Jombart et al. 2010).

READ  Yellow Vests, Pessimistic Beliefs, and Carbon Tax Aversion 

Coalescent theory and population demographic history analyses

Coalescent theory is the base of numerous methods or models that aim at reconstructing the past history of populations such as their size, growth rate, gene flow or their patterns and times of divergence using molecular markers. Here, I will explain the general theory and the specific method that was used in this dissertation to reconstruct the demographic history of bottlenose dolphins in the North-East Atlantic in Chapter 6.
Classical population genetics is a prospective approach which aims at predicting the future of allele frequencies in populations. In contrast, coalescent theory is a retrospective approach which aims at reconstructing the genealogy of a sample of genes going backwards in time to the Most Recent Common Ancestor (MRCA, Figures 2.7a to 2.7c, reviewed in Nordborg 2001). It should be noted that in a coalescent framework, we work with genes, not individuals. In any population, the probability for two genes to coalesce follows an exponential probability distribution. As we get backwards in time, the number of genes will decrease and the time to the next coalescent event (represented by the branch length) will increase. As most mutations can be considered neutral, they can be added afterwards following a Poisson distribution with parameter the length of branches. Figure 2.7. Principle of the coalescent theory. a) The complete genealogy of a population of 10 genes. b) Genealogy of a sample of genes (n=3), here highlighted in black, back to a single common ancestor. c) The genealogy of the sampled genes. It starts form n genes at present back to a single gene in the past, the Most Recent Common Ancestor (MRCA), through coalescent events at different times in the past (source: Leblois, 2010, “La théorie de la coalescence et ses applications”, diapositives de cours, ENS Lyon).
For neutral markers, the gene genealogy is only based on the demography of the population. The topology of the coalescent tree (i.e. the branch lengths and times of coalescent events) can thus inform us about the demography of the population (Figure 2.8).
The coalescent theory allows the probabilistic simulation of genetic variability expected under different demographic scenarios. Simulation is made easier as it is based on samples of genes instead of the whole population. However, the number of possible gene genealogies is infinite. Therefore, numerical approaches (that will not be detailed here) have been developed to explore the relatively more probable genealogies. These methods can be named “coalescent samplers” (reviewed in Kuhner 2009). To find the most likely genealogy (i.e. the probability that the data have evolved under this genealogy and mutation model), the sampler can implement either or both likelihood-based or Bayesian approaches using Markov Chain Monte Carlo (MCMC). However, the computation of the likelihood function is notoriously difficult, as the search space for parameters is infinite, which limits the possibilities of scenarios to test. Hence, mostly simple scenarios, which generally involve a low number of populations, can be tested. Although recent developments allow to include more populations (e.g. IMa2, Hey 2010), computation times are long (several months) and MCMC might never reach convergence as the parameter space is very large.

Confidence in the scenario choice and in the parameter estimates

For each scenario, a few hundred datasets are simulated using parameters values drawn from the prior distribution specified in the first step. Posterior probabilities are computed and used to estimate the Type-I and Type-II error rates in choosing each scenario. For instance, Type-I error rate for scenario A is estimated as the proportion of simulated datasets generated under scenario A that supports other scenarios. Type-II error rate for scenario A is estimated as the proportion of datasets simulated under all the other scenarios that supports scenario A.


The “goodness-of-fit” of a scenario according to the observed dataset, that is how well a scenario can reproduce the observed dataset, can be computed. It measures the consistency between a scenario and its parameter posterior distributions (i.e. “the posterior predictive distributions”) and the observed dataset using summary statistics. Summary statistics should also include statistics that have not been included previously in the inference step; otherwise the quality of the fit may be overestimated. In practice, data are simulated under each scenario using parameter values drawn from parameter posterior distributions. DIYABC allow testing visually, through a Principal Component Analysis, if the observed data are in the range of the values generated using the posterior predictive distributions. The probability that the simulated data do not encompass the observed data could be estimated for each summary statistics.

Table of contents :

1) Interaction between social, ecological and genetic structures
2) Drivers of structure
a) Social structure
b) Ecological structure
c) Genetic structure
3) Conservation implications
4) Study model: bottlenose dolphin and research questions
a) Studying cetacean population structure: interest and challenges
b) Why studying bottlenose dolphins?
c) Taxonomy and variations in ecology, morphology and genetic structure
d) Life-histories and social structure
e) Bottlenose dolphins in the North-East Atlantic, distribution and conservation status .
f) Research questions
g) Manuscript organization
1) A combination of approaches: from recent to evolutionary time scales
a) Photo-identification
b) Ecological and diet indicators
c) Morphometrics
d) Molecular markers: mitochondrial DNA and microsatellites
2) Statistical analyses of molecular markers
a) Bayesian statistics
b) Genetic structure
c) Coalescent theory and population demographic history analyses
1) Introduction
2) Material and methods
a) Surveys and photo-identification
b) Social structure
c) Abundance
3) Results
a) Survey effort and photo-identification
b) Social structure
c) Community size
4) Discussion
a) A fission-fusion social structure
b) Possible ecological drivers of large group sizes
c) Division in three social clusters
d) Abundance
e) Monitoring and conservation
1) Introduction
2) Material and methods
a) Boat surveys, biopsy sampling and photo-identification
b) Social structure
c) Genetic analyses
d) Genetic population structure
e) Ecological population structure
f) Influence of relatedness, sex and ecology on association patterns
3) Results
a) Biopsy sampling
b) Genetic population structure
c) Ecological population structure
d) Influence of relatedness, sex and ecology on association patterns
4) Discussion
a) Three social and ecological clusters but a single population
b) Ecology but not kinship influences social structure
c) Influence of phylogeography on social structure
d) Drivers of social structure and interest of combining approaches
1) Introduction
2) Material and methods
a) Sample collection, DNA extraction and sexing
b) Microsatellite genotyping and validity
c) Mitochondrial DNA sequencing
d) Population structure
e) Nuclear genetic differentiation and diversity
f) Mitochondrial DNA differentiation and diversity
g) Recent migration rates
h) Effective population sizes
3) Results
a) Microsatellite validity
b) Drift prediction model
c) Population structure
d) Genetic differentiation and genetic diversity in the NEA
e) Recent migration rates
f) Effective population sizes
4) Discussion
a) Hierarchical structure
b) Possible drivers of population structure
c) Effective population size estimates: small coastal vs large pelagic populations
d) Management implications
e) Ecotype delineation and future directions
1) Introduction
2) Material and methods
a) Genetic inference of the population demographic history
b) Ecological and morphological characterization of ecotypes
3) Results
a) Genetic inference of the population demographic history
b) Morphometric analyses
c) Stable isotope analyses
d) Stomach content analyses
4) Discussion
a) Ecologically-driven demographic history of bottlenose dolphins in the North-East Atlantic
b) Niche specializations maintain genetic divergence between coastal and pelagic ecotypes
c) Absence of strong influence of ecology on external morphological traits
d) Possible differential stage of speciation in the North Atlantic
1) Synthesis of the results
a) Bottlenose dolphin social, ecological and genetic structures in the Normano-Breton gulf
b) Bottlenose dolphin population structure in the North-East Atlantic
2) Structuring patterns of bottlenose dolphins and other mobile social predators: interaction between ecology, sociality and genetics
a) The central role of ecology in shaping the structure of populations
b) Social behavior likely strengthens the influence of ecology on genetic structure
c) Influence of evolutionary history on social structure
3) Combination of scales and approaches to study the structure and evolution of populations
a) Combination of spatial scales
b) Combination of approaches
4) Implications for conservation
a) Conservation of bottlenose dolphins in the Normano-Breton gulf
b) Conservation of bottlenose dolphins in the North-East Atlantic
c) Management implications beyond bottlenose dolphins
5) Perspectives


Related Posts