METHODOLOGICAL ADVANCES (Papers I-IV)
The first part of this thesis focuses on the methodological advances I have developed with my colleagues for studying biology and conservation of brown bears. Of course these methods are applicable to other species as well.
DNA amplification methods from fecal samples
We proposed a new method (Paper I) and redesigned specific bear primers (Paper II) to amplify DNA from fecal samples. DNA extracted from noninvasive samples is often the only source of genetic material available for many wild animals. However, it is typically of low quality and quantity and is therefore prone to poor amplification success, allelic dropout, false alleles and contaminations. Dealing with these problems is a challenge for researchers.
The “multiplex pre-amplification method” (Paper I) proposes two distinct steps in the amplification of DNA products. The first step simultaneously amplifies all microsatellite loci to be subsequently genotyped. The second step uses post-amplification aliquots as template in locus-specific PCRs to genotype individuals. Compared to conventional PCR approaches, this method allows to increase amplification rates, reduce genotyping error rates (false alleles and allelic dropouts) and improve the readability of the microsatellite profiles, particularly in species for which DNA is in limiting conditions or degraded. Moreover, the amount of DNA template required is reduced, which is particularly useful in studies with low-quantity DNA samples. The multiplex pre-amplification PCR method worked well in two different laboratories and for four different species, both carnivores and herbivores, although the results for bears were not shown in Paper I. This new approach can be considered as a major advance in the field of noninvasive genotyping.
This multiplex pre-amplification method employed together with a nested PCR approach and re-designed microsatellite primers (Paper II) allowed us to optimize the amplification of DNA from bear feces. Most of the original microsatellite primers (Paetkau & Strobeck 1994; Taberlet et al. 1997) were redesigned in order to obtain smaller amplicons and more similar annealing temperatures among primers. We also designed new sex-identification primers, amplifying a short fragment more specific to carnivores, in order to avoid amplifications from most of the prey DNA coextracted from a bear feces. The strategy of combining the multiplex pre-amplification method and a semi-nested PCR could be transposed to other species where conventional PCR approaches yield low success due to limiting DNA concentration and/or quality.
Genotyping errors in population genetics
We reviewed methods for tracking and assessing genotyping errors in population genetics
(Paper III). In spite of their widespread occurrence, genotyping errors usually remain ignored in population genetics studies, except when analyzing samples suspected to be problematic (e.g. Paetkau 2003). Using four different datasets differing in their sampling strategies (noninvasive or traditional), in the type of organism studied (plant or animal) and the molecular markers used (microsatellites or AFLPs), we showed how prevalent genotyping errors are, in a wide population genetics context, and identified their main causes. This study points out for the first time the necessity of estimating and reporting genotyping error rates in population genetics studies. Tracking genotyping errors and identifying their causes are necessary to clean up the datasets and validate the final results according to the precision required. We proposed guidelines for practioners, at each step of the experimental process (from sampling to analysis) to limit and estimate the genotyping error rate.
Parentage analysis software: PARENTE
We developed a software to analyze parentage from genetic data (Paper VI). Reliable parentage assignment is a first important step in the study of mating systems and social organization. Accurate parentage assignment allows one to determine the genetic “payoff” for behavioral strategies, and lifetime reproductive success (Hugues 1998). In many species, including bears, field data can only provide information on the mother (due to the absence of paternal care), and the only way to identify the father is from genetic information, i.e. from multilocus genotypes. However, there is a clear lack of parentage assignment softwares that are able to consider several generations of individuals and that allow the determination of both parents without any prior assumptions. The software we developed has several advantages over previously available softwares. First, it enables one to find fathers, mothers, as well as both parents simultaneously for an individual, based on multilocus genotypes, without any prior assumptions. It is possible to consider a defined number of genetic incompatibilities between the offspring and parent(s) in order to take into account the error rate in the genetic dataset (e.g. genotyping errors, mutations). Second, it considers dates of birth and death of individuals to restrict the analysis to possible parents only. In this way the input file is able to consider all genotyped individuals, even with overlapping generations and there is no need to define different files for parents and offspring. Third, this software is able to handle large datasets (more than 1000 genotypes). Finally, parentage probabilities are calculated using a Bayesian approach, considering the error rate in the genetic dataset and the sampling rate of the population, which are important factors in parentage assignment. Simulated data showed that this software was reliable (Cercueil 2004; Annex 3). Effectively, PARENTE was able to find 97% and 83% of correct parentage relationships with a probability P≥0.95 and 0.8<P<0.95 respectively. Those performances were compared with the results obtained from another commonly used parentage analysis software, CERVUS (Marshall et al. 1998), using the same simulated dataset, and 87% and 55% of correct parentage relationships were found, for the same probabilities respectively. Using the Scandinavian brown bear dataset, we also verified that known relationships from field data were identified by PARENTE (without any prior assumption). In 65% and 77% of the cases, the relationships mother-offspring were identified with a probability P> 0.99 and P>0.95 respectively. This software permitted the construction of pedigree data (Annex 4) from individual multilocus genotypes (Annex 2).
APPLICATIONS FOR MANAGEMENT, CONSERVATION AND UNDERSTANDING BIOLOGY
Population size estimates (Papers V-VI)
The first genetic studies of the Scandinavian brown bear (Taberlet & Bouvet 1994; Taberlet et al. 1995; Waits 2000; Manel et al. 2004) constituted an important step in the management of this population. However, there was a great need to determine the size of the present population (Naturvảrdsverket 2003). It is difficult to census populations of bears, due to the behavior of this solitary carnivore that occurs at low densities, but it is necessary for proper management and conservation, allowing, for instance, managers to set hunting quotas. Previous population size estimates were based on conventional field methods, such as helicopter surveys or observations in the field (Bjärvall & Sandegren 1987; Sandegren & Swenson 1997; Selander & Fries 1943; Swenson et al. 1994; 1995). However, the reliability of those estimates has never been evaluated.
Estimates using genetic methods
The noninvasive genetic approach developed in this thesis to estimate population sizes, includes three phases: the collection of fecal samples in the field, the laboratory phase, and the analytical phase, i.e. population size estimates derived from fecal genotypes. The laboratory phase has been described earlier in Papers I and II. The two other phases (sampling and analysis) were tested in Paper V.
We compared four methods for estimating population sizes based on genotyping of fecal samples. Two methods used rarefaction indices (Kohn et al. 1999; Eggert et al. 2003), one was based on a closed population capture-mark-recapture (CMR) method (MARK; White & Burnham 1999), and the last one used a Lincoln Peterson CMR method (Seber 1982), combining genetic and field data. We evaluated the accuracy and precision of the estimates based on a known minimum number of radiomarked bears in a subsection of our sampling area. We concluded that the MARK method, based on a principle of capture-mark-recapture of individuals within the fecal sampling, performed the best. We obtained a population size estimate of 550 (482-648 95% confidence intervals) brown bears in Dalarna and Gävleborg counties. In addition, we give recommendations concerning the collection of fecal samples in the field and the use of the best performing population size estimate method. We also suggest how to avoid some important biases in the estimates.
Comparing the performance of genetic and field methods
In Paper VI we compared the performance of the genetic method recommended in Paper V (MARK method) with three other methods based on conventional field data, to estimate population sizes. The field methods were based on (1) observations of females with cubs; (2) data from hunter killed bears; and (3) a capture-mark-recapture principle, considering the proportion of marked estrous females with adult males during the mating season, using helicopters. We found that the three field methods tended to underestimate the true population size and that the noninvasive genetic method seemed to perform the best. In addition, a cost/benefit analysis, in terms of time and money, showed that the genetic method was 4 to 5 times cheaper than the best performing field method, i.e. the mark recapture census using helicopters.
Based on these two studies, and considering hunting statistics in other parts of Sweden and a net population growth rate of about 4.7% annually, it was possible to extrapolate the population size estimate to the entire Sweden; approximately 2200 bears (1600-2800 95% CI) in 2004 (Kindberg et al. 2004). We conclude that the present management of the population has been successful and bears in Sweden can be considered as being in a good conservation status.
These results are important for the human-dimension perspective of wildlife management. In the long-term, the survival of bears depends on how they are accepted by humans. Much of the human-bear conflict arises from the fact that local people and managers/researchers have different answers to common questions such as “how many bears?”. The involvement of the local human population in the project (e.g. collection of fecal samples) helped to increase the acceptance of the results. It was also very important that those volunteer helpers, and other people interested, obtain feedback from the research project. In cooperation with Jonas Kindberg, we made the results from the fecal analysis available on the following website of the Swedish Association for “Hunting and Wildlife Management” (http://www.jagareforbundet.se/forsk/bjornspillningsinventering/default.asp). On this website, it is possible to see maps of the different hunting areas with the geographical localization of fecal samples corresponding to male and female bears or without amplifiable DNA, and a number is assigned to each genetically identified individual. This allows people to perceive the utility and importance of the collection of fecal samples and to see the results from the analysis, e.g. what is the minimum number of bears that has been in their region before or during the sampling period.
The brown bear mating system (Papers VII-VIII)
In general, the mating system of bears is poorly known. Only a few studies have addressed this question with a limited amount of field and genetic data (Craighead et al. 1995b; Craighead et al. 1998; Kovach & Powell 2003; Schenk & Kovacs 1995). We studied the mating system of the brown bear with the help of molecular markers. From parentage analysis and pedigrees (Annex 4), we examined mating strategies employed by both sexes in relation to SSI and approached the subject of mate selection by females.
Sexually selected infanticide
SSI is probably a major factor influencing mating systems in several species (Hrdy 1979; Hrdy & Hausfater 1984) and has been found to be an important factor affecting cub survival in the Scandinavian brown bear population (Swenson et al. 1997; 2001). In paper VII, we examined the presence of the three requirements for SSI in our brown bear population, based on documented cases of infanticide and genetic data collected in the field. We found that mothers that lost all cubs had their next litter more than one year earlier than those with surviving cubs. From genetic analyses of samples found at the sites of infanticide, we determined that the infanticidal males were not the fathers of the killed cubs and that they had a high probability of fathering the female’s next litter. Interestingly, most of those males were residents, suggesting that they probably recognize the female they mated with the year before and do not attempt to kill cubs that are potentially their own. We concluded that all three prerequisites of SSI were met and that SSI might be an adaptive male mating strategy in brown bears. Multiple paternities were genetically demonstrated and suggested that females may mate with several males in order to confuse paternities, i.e. as a counterstrategy to SSI. This study was the first to genetically document mating strategies in relation to SSI in a nonsocial species. We predict those findings may also be applicable to other social and nonsocial species with SSI and we encourage future studies to obtain paternity estimates in order to assess the reproductive benefits to infanticidal males and to evaluate the counterstrategies employed by females.
Female mate selection
In paper VIII, we investigated different factors possibly influencing female mate selection in the brown bear. We compared different characteristics of the males “chosen” to be fathers, as determined by paternity tests using microsatellite polymorphism, to those known to be available in the vicinity of the female. We assumed that SSI might influence female mate selection in this species, as females may seek mating counterstrategies to SSI. Consequently, we tested the following predictions: (1) females would select high quality males based on their age, body size, heterozygosity, or relatedness, in order to maximize their reproductive output and avoid inbreeding; (2) females would adopt a strategy to minimize the risk of SSI, and rather mate with potentially infanticidal males, i.e. the geographically closest males. We found evidence that both geographical distance and male morphological, genetic and age criteria significantly influenced female choice. Female brown bears might mate with the closest males as a counter-strategy to infanticide and exercise a post-copulatory cryptic choice, using morphological traits or dominance status as indicators of male genetic quality (“good gene” hypothesis, Brown 1997). Mate selection was apparently not a mechanism to avoid inbreeding or outbreeding, as our results supported a random mating scheme in relation to genetic relatedness. Instead, inbreeding avoidance may be a natural consequence of sex-biased dispersal (McLellan & Hovey 2001).
Table of contents :
I. The use of molecular markers in conservation and management
I. 1. Estimating population size
I. 2. Understanding a species biology and behavior
II. The framework: the Scandinavian Brown Bear Research Project
III. Objectives of the thesis
IV. Material and methods
IV. 1. The study species
IV. 2. The study population
IV. 3. The study areas
IV. 4. Field methods
IV. 5. Genetic methods
V. Main results and discussion
V. 1. METHODOLOGICAL ADVANCES (Papers I-IV)
V. 1. 1. DNA amplification methods from fecal samples
V. 1. 2. Genotyping errors in population genetics
V. 1. 3. Parentage analysis software: PARENTE
V. 2. APPLICATIONS FOR MANAGEMENT, CONSERVATION AND UNDERSTANDING BIOLOGY
V. 2. 1. Population size estimates (Papers V-VI)
V. 2. 1. 1. Estimates using genetic methods
V. 2. 1. 2. Comparing the performance of genetic and field methods
V. 2. 2. The brown bear mating system (Papers VII-VIII)
V. 2. 2. 1. Sexually selected infanticide
V. 2. 2. 2. Female mate selection
ANNEX 1: Map of bear distribution in Scandinavia
ANNEX 2: Multilocus genotypes from all sampled Scandinavian brown bears
ANNEX 3: Results from parentage analysis simulations.
ANNEX 4: Pedigrees of the sampled Scandinavian brown bears