The Journal of Heredity 2001:92(2)
© 2001 The American Genetic Association 92:180-189
Group-Structured Genetic Models in Analyses of the Population and Behavioral Ecology of Poikilothermic Vertebrates
From the Department of Fisheries and Wildlife, Michigan State University, 13 Natural Resources Building, East Lansing, MI 48824-1222 (Scribner) and Savannah River Ecology Laboratory and Department of Genetics, University of Georgia, Aiken, South Carolina (Chesser). Dr. Chesser is currently at the Department of Biological Sciences, Texas Tech University, Lubbock, TX 79409.
Address correspondence to Kim T. Scribner at the address above or e-mail: scribne3{at}pilot.msu.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsGeneral Characteristics of...Conclusions and Conservation...References |
|---|
Estimates of gene correlations among individuals within and among populations are frequently derived from statistical analyses of genetic data (e.g., F statistics). These measures can be important tools in molecular ecology and conservation, and offer important insights into population breeding structure. Using recently derived theory developed for group-structured populations, we show that fixation indices, when combined with basic population ecological and demographic data can be used to investigate population mating systems and to predict dispersal rates, trajectories and asymptotic levels of fixation indices, and effective population size. Four case studies of poikilothermic vertebrates are used to demonstrate the broad utility of evolutionary and ecological inferences afforded by group-structured models.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsGeneral Characteristics of...Conclusions and Conservation...References |
|---|
The dynamics of genes in natural populations center on the movement of genes within and among populations. Accordingly, dispersal, aspects of the mating system, and variances in male and female reproductive success are all critically important parameters in understanding how genetic variation is partitioned and maintained. Prior evolutionary models that quantify such partitioning of levels of genetic variation within and among subpopulations or breeding groups differ in assumptions and in the relative importance placed on these parameters. Wright's formulations were based on an "island" model, and principally involve interactions between drift and gene flow. Wright (1943) addressed gene correlations within subpopulations as the accrual of homozygosity, coancestry, or identity by descent due to the effects of small effective population size. These models assume that there are equal reproductive contributions among breeding adults and that subpopulations are of constant, equal size. Wright further assumed a large number of subpopulations, each contributing dispersers to the pool of migrants with equal and constant probability.
More recently Wright's models have been recast using different methodologies or by emphasizing the importance of different evolutionary forces. Cockerham (1969) demonstrated that Wright's (1965) fixation indices could be determined using gene correlations 
, where F is the correlation between genes within individuals, 
or coancestry is the correlation between genes of different individuals within subpopulations, and f is the correlation of genes within individuals in the total population. Cockerham (1969, 1973) used an analysis of variance (ANOVA) approach for the derivation of genetic variance components in subdivided populations. Slatkin (1991) described relationships among probabilities of identity by descent, the distribution of coalescence times of pairs of genes, and Wright's measure of degree of population differentiation (FST). The assumption of an infinite number of subpopulations was subsequently relaxed (Crow and Aoki 1984). Other researchers (Kaj and Lascoux 1999; Maruyama and Tachida 1992; McCauley 1993; Whitlock and Barton 1997) relax some of the assumptions of Wright's island model in a metapopulation context, highlighting the importance of different birth-immigration-death processes to equilibrium probabilities of identity by descent, to Wright's fixation indexes, and to estimates of effective population size.
Chesser and coworkers (Chesser 1991a,b, 1998; Chesser and Baker 1996; Sugg and Chesser 1994) also have shown that assumptions of Wright's models are seldom met in nature. Group-structured models introduced by these authors expand on Wright's formulations to incorporate aspects of organismal demography, life history, and behavioral ecology to estimate gene correlations at different levels: within individuals (F), between individuals within a breeding group (), and between individuals from different breeding groups (
). By parameterizing the models of how genes move within and among groups, greater accuracy and predictability can be obtained when estimating ecological and behavioral features of populations.
The increased availability of markers and methods of statistical inference have afforded unparalleled opportunities for population analysis (Avise 1994; Burke et al. 1992). Although the empirical examples offered in this volume speak to the power of molecular genetic markers, there is still relatively little information on mating systems and parentage for poikilothermic vertebrates. However, fisheries biologists and herpetologists frequently possess a wealth data on population demography, female reproductive success, and life-history characteristics which may be used in conjunction with genetic data and group-structured models. The parameters necessary to generate estimates of gene diversity within and among populations for different modes of inheritance (e.g., maternal, paternal, biparental) are shared among the models. Therefore calculation of fixation indexes and effective sizes for one mode of inheritance will often permit the estimation of variables for other types of genetic characters. Here we employ such techniques to predict the dynamics of genetic markers heretofore not assayed for the populations and species we examine.
Such approaches can also be relevant to issues of management and conservation. Understanding the processes that affect gene correlations of individuals within and among populations can be critical in predicting changes in gene correlations and effective population size for natural populations. Awareness of the plausible trajectories of population change is increasingly important in situations of accelerated anthropogenic change in population connectivity, demographic structure, and abundance.
Our objectives are to introduce concepts underlying group-structured models (Chesser 1991a,b; Chesser and Baker 1996; Sugg and Chesser 1994), using empirical case studies from our previous research on four species of poikilothermic vertebrates. We show that parameterization of group-structured models allows for incorporation of biologically relevant aspects of species life history and behavioral ecology, and thus offers a greater range of ecological and evolutionary inferences than is afforded by "classical" population genetic models.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsGeneral Characteristics of...Conclusions and Conservation...References |
|---|
Coancestry, or the correlation of genes within individuals, is a measure of genetic variation. When the correlation of genes increases at a particular level of population structure, the variation decreases. Gene correlations are of three forms: the correlation of genes within individuals (F), between individuals within the same group (
), and between individuals from different groups (). One can calculate fixation indexes from gene correlations as
 |
 |
 |
= 1 - HS and = 1 - HT.
The accumulation of coancestry among individuals within groups is affected by the type of matings, the mean and variance in the number of progeny produced by each mating, and the movement of individuals among groups. In the absence of inbreeding, the most extreme value of coancestry is achieved by male polygyny together with female philopatry (Chesser 1991a,b). In such cases, progeny born within groups are half-siblings achieving an ultimate coancestry of 1/6. The probability that a random pair of progeny born within a group share the same parent (female or male) is
 |
k2 is the variance in progeny number, and n is the mean number of parents (either male or female). High variance leads to higher coancestry because there is a higher proportion of full siblings relative to other, less-related offspring. Dispersal among groups increases the coancestry among groups () at the expense of that within groups () and usually will decrease inbreeding (F). Unlike Wright's traditional models, the group structured models do not assume a very large number of equally sized breeding groups. Variance in group size can increase the probability that random pairs of individuals from the same native group are mates, even if they disperse. The probability that a random pair of individuals is from the same native group is
 |
n2 is the variance in number of parents and s is the number of groups (Chesser 1998; Chesser and Baker 1996). Input of these parameters into population models permit the incorporation of behavioral information that can have substantial effects on gene dynamics within and among groups. Although it can be argued that additional parameters create additional complications, Wright's models involve several unrealistic assumptions: zero variance in group size, a very large number of groups (s), equal male and female dispersal rates, and random mating within groups. When such assumptions are incorporated into our group structured models, our outputs do indeed converge on the predicted values of Wright (Chesser et al. 1993). |
General Characteristics of Poikilothermic Vertebrates |
|---|
|
TopAbstractIntroductionMaterials and MethodsGeneral Characteristics of...Conclusions and Conservation...References |
|---|
Group-structured models are expected to have great utility for poikilothermic vertebrates, which often share numerous commonalties in habitat requirements, demographics, and behavioral ecology (Table 1). For most fish, turtle, and anuran species, breeding is not random, but occurs in a large number of defined breeding groups of variable size and demographic composition. Further, mating systems lead to asymmetries in male and female reproductive success. Aquatic breeding habitats can be quite ephemeral. Of importance, all species have different breeding and movement ecologies that lead to varying expectations for how gene correlations (and thus genetic variation) are partitioned and maintained over time.
|
Case Study 1: Breeding Ecology and Effective Population Size in the Common Toad
Comparatively little is known of the effects of reproductive ecology and behavior on spatial genetic diversity in natural amphibian populations. Yet an understanding of amphibian movements and breeding ecology related to male and female reproductive success and to effective population size is critical in amphibian conservation. Amphibians are of particular conservation concern, as numbers and diversity have declined precipitously in many portions of the world (Alford and Richards 1999). Many species are susceptible to environmental perturbations because their complex life histories include both aquatic and terrestrial components.
Breeding populations of the common toad Bufo bufo were studied in an intensively managed agricultural landscape in central Great Britain. We used demographic and genetic data for eight populations in close proximity, as part of a larger study of 20 populations throughout an agricultural landscape in Leistershire (Scribner et al. 2001).
Common toads spend most of the year in terrestrial habitats but return for a brief period each spring to breed in small aquatic environs (Slater et al. 1985). Recapture studies (Reading et al. 1991) reveal that both males and females exhibit high philopatry to natal ponds. Recently Scribner et al. (2001) showed that populations of B. bufo in central Great Britain are highly structured based on spatial heterogeneity in minisatellite allele frequency. Significant spatial autocorrelation (significant nonindependence in minisatellite allele frequency) was documented over interbreeding pond distances of less than 2 km (Scribner et al. 2001). We consider the group of breeding ponds to represent a single metapopulation.
During the spring breeding season, males enter the breeding ponds and establish territories in water near the pond's edge or in adjacent terrestrial habitat. Males remain in the pond and are reproductively active throughout the breeding season. Each female then comes to a pond typically for a single evening during which time the operational sex ratio is male-biased (average >5:1; Scribner et al. 2000). Thus there is likely to be intense competition for females and high variance in male breeding success (Eibl-Eibesfeldr 1950). Large males are known to attain a disproportionately high number of matings (Davies and Halliday 1980).
Genetic analyses were based on individual genotypes at three microsatellite loci for approximately 50 breeding adults from each pond (Scribner et al. 1994). The F statistics (FIT = 0.153; FST = 0.024*, FIS = 0.132*; *P < .05 Bonferroni corrected) suggest that populations were significantly differentiated in allele frequency and that some level of substructuring or nonrandom mating exists. Demographic data for the breeding populations were also collected (see Scribner et al. 2001 for details). Estimates of breeding population size varied from 800 to 12,000 individuals. Based on these estimates, the probability that two randomly chosen breeding adults come from the same population (
) is 0.222.
Using the "temporal method" (Waples 1989), estimates of effective population size were made for three other breeding populations in close proximity to those surveyed. This method uses the adult-progeny variance in allele frequency to estimate the effective number of breeding adults (Scribner et al. 1997). In our cases, estimates of the effective number of breeders were approximately 1% of the total census size. Perhaps a high degree of polygyny or high variance in male and female reproductive success were contributing factors for this dramatic result. Here we extend previous analyses using group-structured models as applied to the genetic data to estimate levels of gene flow, male reproductive performance, and effective population sizes for uniparentally and biparentally inherited genes in this metapopulation.
We used transition equations (equations 27 and 28; Chesser and Baker 1996) to converge on rates of male and female dispersal (the probability that males and females bred in a locale other than of their natal origin) and degree of polygyny that gave the same FST using the other available demographic parameters. Across s = 8 populations, the mean and variance in breeding population numbers were n = 1919 and 2n = 3,272,813. Male and female dispersal rates among the eight breeding groups were estimated to be dm = df = 0.4. Convergence was possible only when some level of polygyny was incorporated into the transition equations. Our estimate of the frequency of male polygyny (
m = 0.1) was consistent with the genetic and demographic data. Higher proportions of polygyny also were possible from the models, but required higher dispersal rates to maintain the observed level of FST.
Estimates of levels and partitioning of gene diversity reflected in fixation indexes for these populations were based on a single time period and may not reflect the equilibrium state given the demographic characteristics for this metapopulation. Using group structured models (Chesser 1998, equations 31 and 32), we were also able to predict asymptotic values of F statistics for maternal, paternal, and biparentally inherited genes (Figure 1). Asymptotic estimates of spatial variance in paternally inherited genes (FLSY) was high (Figure 1). A nonzero estimate of polygyny implies that some males are contributing disproportionately to reproduction, a result consistent with direct observations in other studies (Davies and Halliday 1980). In general, variation in male reproductive success increases both the coancestries within breeding groups and the variance among them. High predicted levels of male coancestry and low effective population size for male genes effectively counter the relatively high rates of gene flow. In this case, there was a low predicted effective population size for male genes (Figure 2A), indicating that relatively few males were contributing genes to the study populations. In our model, estimates of F statistics are expected to asymptote in relatively few generations (Figure 1). Changes in gene correlations (
F, , ) are most affected by rates of dispersal and population size (Chesser and Baker 1996). The relatively small and highly variable population sizes coupled with high estimated rates of dispersal affect the rates of change to asymptotic estimates of effective population size.
|
|
For the toad metapopulation, estimates of effective population size for diploid (Ne(diploid)) and unisexually (Ne(mtDNA), Ne(paternal)) inherited genes differ dramatically (Figure 2), and estimates of male coancestry and intergroup effective size are extremely small (Nem
= Nem 
46). For biparental and maternal genes, estimates of effective population size are comparatively higher (Ne(diploid) and Ne(mtDNA), respectively; Figure 2B,C). These results support general conclusions of the Ne:N ratios as calculated by the other methods (Scribner et al. 1997). Thus, based on group-structured models, estimates of Ne:N for the entire metapopulation (0.0078) are comparable to those for single breeding populations (0.012) as estimated using adult-progeny variances in allele frequency. Polygyny also has the effect of dramatically lowering the effective sizes for paternal and diploid genes, which already had Ne(mtDNA):N ratios of less than 1% (Figure 2). For maternal genes, the Ne:N ratio (35%), as expected, is considerably higher than for nuclear genes and paternal genes, the much lower value for the latter being most plausibly explained by the high variance in n (the number of parents) among populations, and to male polygyny.
Methods of estimating effective population size vary greatly (see reviews in Frankham 1995; 1996; Nunney and Elam 1992; Whitlock and Barton 1997). The methods employed here differ markedly from those that are based on rates of patch occupancy, patch extinction, and patch recolonization as advocated for metapopulation analyses by Hedrick and Gilpin (1997). However, in situations where populations now or in the past have experienced appreciable gene flow, calculation of Ne for each breeding group separately is theoretically impossible and biologically uninformative. In reality, the effective size for a group will approach the effective size of the entire array of populations if immigration and emigration are taken into account (Chesser et al. 1993).
Case Study 2: Gene Dynamics and Effective Population Size in Highly Exploited Sockeye Salmon
The complex life histories of anadromous salmonids, coupled with their ability to home with high fidelity to natal streams after years in the ocean (Quinn 1990), has produced a complex array of ecologically and genetically distinct populations within each species. This unique life history has made salmonids particularly attractive to ecologists and evolutionary biologists.
For sockeye salmon (Oncorhynchus nerka), intraspecific diversity must have arisen recently because much of the current range was colonized after the last glacial retreat. Based on contemporary distributions and analyses of population genetic structuring, sockeye salmon exhibit great heterogeneity in (1) the number and sizes of spawning aggregations (i.e., subpopulations) within individual aquatic systems; (2) the number, location, and timing of spawning runs; and (3) the extent of morphological, life history, and genetic variability within drainages (Burgner 1991; Wood 1995).
The accumulation of genetic data for the species has been motivated primarily to address questions of population differentiation and harvest allocation. Little attempt has been made to relate breeding ecology and population demography to spatial variances in gene frequency. Furthermore, inferences from genetic sampling typically have been based on one point in time, whereas salmon populations typically experience large and episodic changes in size that can affect gene dynamics.
We focus our analyses on spawning aggregations of sockeye salmon in Tustamena Lake in southcentral Alaska. This is an oligotrophic, glacially turbid, geologically young (<2000 ybp) lake draining approximately 1375 km2. Returning adult salmon can be distinguished by life-history differences in run timing and spawning habitat: "early run" sockeye spawn in lake tributaries, and "late run" sockeye spawn in areas of groundwater upwelling along shorelines and in the river outlet. These life-history patterns are common throughout much of the species native range (Burger et al. 1997; Burgner 1991).
Estimates of adult population size are available for most of the major spawning aggregations within the Tustamena Lake system over an 18-year period (1975–1992; Alaska Department of Fish and Game, unpublished data). The mean total yearly number of breeding adults was estimated to be 135,270. Parameter estimates for group-structured models include the number of breeding groups (s = 9 primary spawning sites) characterized genetically, and annual estimates of total adult spawners per location (range in population means 800–68,260). We assumed the adult sex ratio to be equal or slightly skewed toward males (Burgner 1991). The average number of breeding females and males in each group (averaged over 18 years) is n = m = 7680.
Cytonuclear data are available for each of the nine breeding populations. Analyses of seven microsatellite loci (Scribner KT, et al., unpublished data) and mtDNA data (Burger et al. 1997) suggest that populations are spatially structured. Analyses were conducted with and without the Kasilof River, as this outlet-spawning population is very divergent in mtDNA haplotype frequency and microsatellite allele frequency (F statistics with Kasilof River, F = 0.007, = 0.015*, f = -0.008; mtDNA ST = 0.139*; F statistics without Kasilof River, F = 0.001, = 0.007*, f = -0.006; mtDNA ST = 0.002; *P < 0.05 Bonferroni corrected).
The objectives of our study were to combine estimates of demographic parameters with cytonuclear estimates of gene diversity to estimate male and female dispersal rates, and nuclear and mitochondrial effective population sizes (as per Chesser 1998; Chesser and Baker 1996). This case study exemplifies the effects of human exploitation. For Tustamena sockeye salmon, the numbers of breeding adults that pass the gauntlet of commercial and sport fishermen fluctuate dramatically, primarily as a function of human interference.
Over 18 years, breeding salmon numbers in the Tustamena system fluctuated from 55,500 to 278,000 fish. For nine spawning aggregations, the mtDNA data yielded ST = 0.139. However, when the Kasilof River population was excluded, the differentiation was only ST = 0.002. Chesser and Baker (1996; equation 21) showed that the differentiation among populations using mtDNA could be expressed as
 |
is the probability that any two randomly chosen individuals are from the same population and f is the probability that any two randomly chosen offspring from the same breeding group were produced by the same mother. We chose to estimate the initial conditions using the following parameter definitions:
 |
 |
 |
 |
 |
 |
 |
ST = 0.139, solving equation 1 for df, yields a female dispersal rate of 0.0017. When the Kasilof River breeding population was excluded, df was estimated to be 0.131, or 75 times higher! The initial conditions defined above and the derived female dispersal rates were used as the starting point for a numerical solution to derive gene correlations, fixation indexes, and effective population sizes over time. We further assumed that male and female dispersal rates were equal. After each iteration of the transition matrix (equations 27 and 28; Chesser and Baker 1996), we allowed the population census number to be regulated randomly (by "fishing"), but with population numbers not permitted to drop below 55,500 in any one time period. Population growth via recruitment occurred prior to the fishing, but was not allowed to exceed 278,500 fish. The condition that the variance of group size was equal to the group size was maintained. All other parameters remained constant. Given the estimates of fixation indices and ranges in population escapement over time, we estimate the mean and variance in female reproductive success to be k = 3 and var(k) = 6.
In Figure 3, results of the transition models are shown in terms of asymptotic estimates of fixation indices and asymptotic estimates of effective population size for uniparental and biparental genes (Figure 4). The gradual decline in the effective numbers for mtDNA is expected (Chesser 1998; Chesser and Baker 1996; Chesser et al. 1993). Also shown are the expectations for the effective numbers for diploid genes and paternally inherited genes, although these were not empirically estimated prior to the study. However, the fact that we have defined sufficient parameters to solve not only mtDNA characters, but also those necessary for dynamics of diploid and uniparental characters, permits us to predict a priori the distribution of these genes as well.
|
|
Estimates of asymptotic effective population size and the trajectories of these estimates over time differ appreciably depending on whether the highly diverged Kasilof River breeding population is included in the analyses (Figure 4A versus B). Fluctuations in the numbers of salmon recruited subsequent to fishing caused considerable shifts in estimates of effective sizes and genetic fixation indexes especially when the Kasilof population was omitted. High dispersal (dm = df = 0.131) among the remaining (s = 8) groups accentuates the changes in Ne with fluctuating census size. However, these fluctuations were insufficient to divert the populations from approaching equilibrium values over time. Genetic differentiation for all modes of inheritance is shown. Because males were assumed to mate randomly (
m = 0) and dm = df, the effective size and ST for mtDNA (maternal) and paternal characters (if present) are equal. The ultimate ST attained was not equal to those empirically estimated or derived from equation 1 because of the random fluctuations in population numbers.
Overall the empirical data show that much of the genetic variance among populations is accounted for by inclusion of the Kasilof River. Presumably this is because substantially fewer salmon migrate between the Kasilof River and the other sampled breeding populations. In any event, loss of the spawning grounds for salmon in the Kasilof River would result in the loss of more than one-eighth of the total genetic variance found in salmon in this region. The dynamics of the remaining populations are shown by Figure 3. The combination of relatively high dispersal rates and randomly changing population numbers (after recruitment) causes substantial fluctuation in effective sizes. However, it should be noted that without inclusion of the Kasilof River, the ultimate effective population sizes of the salmon (100,000 for diploid, 50,000 for haploid) were about two-thirds of their respective values when that population was included.
Case History 3: Metapopulation Dynamics and Environmental Change in Slider Turtles
Most poikilothermic vertebrates are closely tied to their environments (Table 1), and thus are particularly susceptible to environmental changes. For species that spend at least a portion of the life cycle in freshwater, fluctuations in breeding habitat can have particularly dramatic effects. However, most reptiles and amphibians also have the capacity to disperse over land when environmental conditions deteriorate. Thus environmentally mediated dispersal can be a natural event, and it is of interest to investigate its importance on population ecology and breeding structure. Again, genetic markers and group-structured evolutionary models can be employed to this end.
The slider turtle (Trachemys scripta) is a highly aquatic species, well adapted to ephemeral habitats in the southeastern United States (Gibbons 1990). This species is characterized by long generation intervals (Gibbons and Semlitch 1982; Gibbons 1990) and high adult survivorship (Frazer et al. 1990). Adult longevity coupled with low but variable rates of recruitment suggest that little temporal variation in population genetic characteristics might be expected for this abundant species.
Episodic drought conditions in the Southeast can result in the total drying of even "permanent" aquatic habitats. Gibbons et al. (1983) documented the influence of drought on migration and reproduction for T. scripta on the Savannah River Site (SRS) as part of a 30-year study of the species ecology. Long-term studies for one aquatic site during the drought of 1980–1981 documented a 86% reduction in total numbers of clutches, and a 640% increase in rates of emigration relative to more normal times. Further, no emigrating females had oviductal eggs, suggesting total reproductive failure during that year. Genetics data were collected during and following this period (Scribner et al. 1986; Scribner et al. 1995; Smith and Scribner 1990), and form the empirical basis for further analyses employing group-structured models.
During 1983 and 1984, 299 individuals were sampled from three slider turtle populations on the SRS, each inhabiting a site with permanent water. These sites did not completely dry during the drought, but likely received immigrants from others which did. The 1984 (but not the 1983) sample included juveniles that would have been produced in the year following the 1980–1981 drought. In 1984 we resampled from adult populations surveyed in 1983 as well as offspring produced the year after the admixture event. Based on F statistics (Weir and Cockerham 1984), we observed a dramatic change in the apportionment of gene correlations among individuals within and among populations across the 2 years (1983, FIS = 0.237, FIT = 0.244, FST = 0.010; 1984, FIS = 0.089, FIT = 0.099, FST = 0.010), leading Scribner et al. (1995) to conclude that the drought had indeed caused substantial population admixture.
Numerical methods were used to provide solutions for changes in gene correlations that might account for these changes in the F statistics. Given that the 1984 sample was a combination of adults from 1983 plus their progeny born after the drought, we can express the fixation indexes as
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
, , and F that yield the fixation indexes for 1984. However, these indexes are estimates, not exact. Comparisons of fixation indexes for 1983 and 1984 show that the most robust estimates of change in gene correlations are F 
-0.146 and 0. Incorporation of these changes yields estimates close to the 1984 fixation indexes.
The fixation indexes for 1983 need to be scrutinized closely to interpret what transpired biologically. The highly positive FIS value (indicating a deficit of heterozygotes) suggests that the adult population in that year was already comprised of a mixture of animals from genetically divergent populations, producing a Wahlund effect (Chesser 1991a,b). Furthermore, because the FIT value was not very different from the FIS (and because FST was low), the admixture appears to have been random and fairly complete. Such population mixture results in very low coancestry values within and among groups. Then, upon mating, outcrossing among breeding adults repartitions the Wahlund variance, producing the -F. Using the F value as an index to mixing among the former breeding groups, we can estimate that the inbreeding coefficient in the 1983 population (prior to mixing) was 0.244–0.146 = 0.098. Also we estimate the value of intragroup coancestry () to have been 0.010 + 0.146 = 0.156. Therefore estimates for the fixation indexes for populations prior to the admixture event that produced the 1983 adults are F'ST(1983) = 0.156, F'IS(1983) = -0.0675, and F'IT(1983) = 0.098.
By 1984 the adults and their young carried admixtures of the genes that produced the 1983 adults. Thus no substantial changes in coancestry within or among the breeding groups are to be expected. Clearly the drought has brought about significant changes in the breeding structure in the turtles on the SRS. The Wahlund effect seen in adults and in the F for their progeny would not have been possible unless there was substantial differentiation among the populations prior to the drought period.
The take home message from this case study is that genetic data collected over time and across demographic groups can provide insights into the effects of environmental processes on levels and apportionment of gene correlations within and among individuals and breeding groups. This study also exemplifies the importance of metapopulation structure for poikilothermic vertebrates. The ability to emigrate from inhospitable environments is significant in terms of the immediate fitness of the individual and in restructuring gene diversity. Dispersal and mating patterns can have adaptive consequences through their effects on reordering genetic variation and influencing levels of inbreeding.
Such analyses also have conservation implications. Increased habitat fragmentation is likely to inhibit dispersal and may predispose populations to increased levels of inbreeding or loss in viability. We would caution, however, that dispersal within metapopulations is not a panacea. Reapportionment of genetic variation (as described in this example) is possible only if populations are spatially differentiated to begin with. Thus in the absence of mechanisms for accrual of variance among breeding groups, the effect of admixture on within-population levels of variability will reach a point of diminishing returns. The efficacy of episodic admixture as a conservation strategy should be investigated on a case-by-case basis.
Case Study 4: Reproductive Ecology and Gene Dynamics in Painted Turtles
Long-term ecological studies are also indispensable to an understanding of the spatiotemporal dynamics of populations and of the biotic and abiotic actions that affect such change. Long-term studies are of particular importance for long-lived organisms such as freshwater turtles which exhibit delayed sexual maturity and long generation times (Congdon and Gibbons 1990).
For the past 45 years, extensive studies on painted turtles (Chrysemys picta) have been conducted on the E. S. George Reserve, a 560 ha research area located in southeastern Michigan (Congdon and Gibbons 1996). These efforts have afforded unparalleled opportunities to use genetic markers in conjunction with data on individual female reproductive history to reconstruct genealogical relationships within breeding populations and to examine the relative importance of female reproductive ecology on gene dynamics. Scribner et al. (1993) proposed that variance in female reproductive success within identifiable breeding groups could increase genetic correlations among individuals within groups and thus increase genetic variance among groups. Such expectations contrast with classical interpretations that invoke gene flow and drift (as related to Ne) as the primary mechanisms partitioning genetic variation. Here we review the general findings of Scribner et al. (1993).
The E. S. George Reserve is characterized by numerous semi-isolated shallow aquatic habitats surrounded by upland hardwoods and old fields. Turtles remain in aquatic habitats for the majority of the year, and females nest in discrete areas surrounding each marsh. Information on individual movements were based on 12,683 captures of 2203 adult males and females over 15 years. Based on recapture data, females show a high degree of philopatry to breeding areas, with 88.4% of all females nesting on upland areas within the same marsh and 62.8% of all females consistently nesting at the same site. Muscle biopsies and blood samples from 429 individuals (representing approximately one-third of the adult population) were obtained during 1986–1990. Eighteen allozyme loci were screened to apportion genetic variation among marshes, among nest locations within marshes, and among individuals within nesting areas. All individuals were aged, so genetic characteristics [e.g., heterozygosity (H) and inbreeding coefficients (F)] could be estimated for each cohort.
Reproductive data on numbers of nests, nest success, and numbers of hatchlings produced from each nest were obtained by monitoring gravid females. Data from successfully nesting females and the number of progeny successfully recruited from each nesting area for each year were used to calculate average coancestries (mean ) as described by Chesser (1991a). Briefly, the number of individuals in a family (b), number of families (n), and total numbers of individuals within a nesting area (N) were used to estimate the average coancestry of the jth nesting area as
 |
 |
 |
= 0.069; Scribner et al. 1993). Thus the magnitude of gene correlation among individuals within and among populations as inferred from allozyme data roughly matched that developed from the demographic information, irrespective of levels of gene flow or genetic drift.
Mean estimates of F, , and f can provide a summary of average gene correlations across genetically heterogeneous demographic groups but fail to capture the true nature of annual variation in ecological factors leading to variation in breeding parameters (e.g., mean and variance in reproductive success, etc.). For C. picta, within-year estimates of H, F, and were all correlated with reproductive characteristics of nesting females (Scribner et al. 1993). However, estimates of mean gene correlations among individuals within nesting areas varied greatly across age cohorts ( = 0.032–0.171; Scribner et al. 1993), and the highest levels of were estimated for years in which relatively few females reproduced (r = -0.73, P = .036; Scribner et al. 1993). Estimates of heterozygosity (H) and inbreeding coefficients (F) also varied greatly among age cohorts (range in H: 0.041–0.096; range in F: -0.215–0.135).
Annual variation in cohort H and F is not the result of year-to-year differences in variance effective population size (Nev) due to large annual variation in female reproductive success. Effective population size is indeed inversely proportion to reproductive variance, and low effective population size should lead to year-to-year changes in cohort genetic characteristics. However, this variation should be expressed as differences in allele frequency, not genotypic composition. We observed little variation in allele frequency among cohorts, in contrast to the large variation in genotypic composition as demonstrated by variation in H and F. Differences in H and F among cohorts clearly reflect year-to-year changes from relatively outbred to inbred matings. These changes go beyond expectations of yearly variation in that might be explained by female-offspring and sib-sib correlations, or stochastic drift relative to variance effective population size (Scribner et al. 1993).
Estimates of coancestry based on allozyme data (0.046) and female reproductive data (0.069) were also used to examine the issue of multiple paternity in female clutches. Scribner et al. (1993) hypothesized that if allozyme data accurately described the apportionment of genetic variation within and among groups, then a lower value of Fnm (0.046) compared with (0.069) should register the presence of half sibs (rather than full sibs exclusively) within some clutches. Thus such an outcome would reflect instances of multiple paternity. Using allozyme estimates of mean coancestry, the proportion of singly sired (P) and multiply sired (1—P) clutches was approximated as
 |
|
Conclusions and Conservation Implications |
|---|
|
TopAbstractIntroductionMaterials and MethodsGeneral Characteristics of...Conclusions and Conservation...References |
|---|
Statistics generated from genetic data on natural populations have limitations when used as the sole source of ecological or evolutionary inference (e.g., Whitlock and McCauley 1999). Aggregations of individuals used to estimate genetic characteristics from geographic locales frequently do not comprise a randomly mating population, but rather are structured into breeding groups which are not necessarily stable or isolated, but rather can have considerable fluidity. Poikilothermic vertebrates have evolved suites of life-history, demographic, and behavioral characteristics (Table 1) such that episodic changes in gene correlations are likely to be the norm. Anthropogenic effects are likely to further impact the gene dynamics of natural populations by increasing population fragmentation and lowering total metapopulation size.
Findings from these four case histories provide several general perspectives. First, breeding groups within populations of poikilothermic vertebrates are extremely dynamic. Stochasticity imposed by environmental heterogeneity or anthropogenic factors can have appreciable effects on population size and reproductive characteristics. These factors do not necessarily affect asymptotic levels of gene correlations or effective size, but will prolong times to attain asymptotic values. Second, variance in group size is an important parameter affecting levels and partitioning of gene correlations within and among breeding units. Due to their complex life histories and reliance on multiple habitats during different life-history stages, poikilothermic vertebrates are particularly susceptible. Third, breeding groups are not demographically or genetically independent. Parameterization of models which more fully incorporate aspects of organismal life history and behavioral ecology will facilitate greater interpretative power and predictive potential.
The biology of poikilothermic vertebrates is complex, involving ontogenetic and seasonal shifts in habitat use, movement patterns, and in behavioral repertoires employed. Genetic data when interpreted within a highly parameterized statistical framework (such as group-structured models) can assist in understanding and managing this complexity.
|
Acknowledgments |
|---|
Numerous individuals contributed ideas and data to this presentation. Special thanks go to A. Arntzen, T. Burke, J. Condgon, J. W. Gibbons, M. H. Smith, G. Hoelzer, K. Sage, C. Burger, E. Knudsen, and numerous other colleagues for their efforts in collecting and interpreting the empirical data presented. Financial support for this project was provided by the Department of Fisheries and Wildlife and Michigan State University (MSU); the Partnership for Research and Management (PERM) program between the Michigan Department of Natural Resources and MSU; the Natural Environmental Research Council's Joint Agriculture and the Environment Program; and the Savannah River Ecology Laboratory under contract DE-FC09-96SR18546 from the U.S. Department of Energy to the University of Georgia Research Foundation. The article benefited greatly from editorial comments by J. Avise and two anonymous reviewers.
|
Footnotes |
|---|
This paper was delivered at a symposium entitled "DNA-Based Profiling of Mating Systems and Reproductive Behaviors in Poikilothermic Vertebrates" sponsored by the American Genetic Association at Yale University, New Haven, CT, USA, June 17–20, 2000.
Corresponding Editor: John C. Avise
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsGeneral Characteristics of...Conclusions and Conservation...References |
|---|
-
Alford RA and Richards SJ, 1999. Global amphibian declines: a problem in applied ecology. Annu Rev Ecol Syst 30:133–165.[Web of Science]
Avise JC, 1994. Molecular markers, natural history and evolution. New York: Chapman & Hall.
Burger CV, Spearman WJ, and Cronin MA, 1997. Genetic differentiation of sockeye salmon subpopulations from a geographically young lake system. Trans Am Fish Soc 126:926–938.
Burgner RL, 1991. Life history of sockeye salmon (Oncorhynchus nerka). In: Pacific salmon life histories (Groot C and Margolis L, eds). Vancouver: University of British Columbia Press; 1–118.
Burke T, Rainey WE, and White TJ, 1992. Molecular variation and ecological problems. In: Genes in ecology (Berry RJ, Crawford TJ, and Hewitt GH, eds). Oxford: Blackwell; 229–254.
Chesser RK, 1991a. Gene diversity and female philopatry. Genetics 127:437–447.[Abstract]
Chesser RK, 1991b. Influence of gene flow and breeding structure on gene diversity within populations. Genetics 129:573–583.[Abstract]
Chesser RK, 1998. Heteroplasmy and organelle gene dynamics. Genetics 150:1309–1327.
Chesser RK and Baker RJ, 1996. Effective sizes and dynamics of uniparentally and diparentally inherited genes. Genetics 144:1225–1235.[Abstract]
Chesser RK, Rhodes OE Jr, Sugg DW, and Schnabel A, 1993. Effective sizes for subdivided populations. Genetics 135:1221–1232.[Abstract]
Cockerham CC, 1969. Variance of gene frequencies. Evolution 23:72–84.
Cockerham CC, 1973. Analysis of gene frequencies. Genetics 74:679–700.
Congdon JL and Gibbons JW, 1990. The evolution of turtle life histories. In: Life history and ecology of the slider turtle (Gibbons JW, ed). Washington, DC: Smithsonian Institution Press; 45–54.
Congdon JD and Gibbons JW, 1996. Structure and dynamics of a turtle community over two decades. In: Long-term studies of vertebrate communities (Cody ML and Smallwood JA, eds). New York:Academic Press; 137–159.
Crow JF and Aoki K, 1984. Group selection for a polygenic behavioral trait:estimating the degree of population subdivision. Proc Natl Acad Sci USA 81:6073–6077.
Davies NB and Halliday TR, 1980. Competitive mate searching in male common toads, Bufo bufo. Anim Behav 27:1253–1267.
Eibl-Eibesfeldt I, 1950. Ein beitrag zur paarungsbiologie der erdkrote (Bufo bufo L.). Behavior 2:217–236.
Frankham R, 1995. Effective population size/adult population size ratios in wildlife: a review. Genet Res 66:95–107.[Web of Science]
Frankham R, 1996. Relationship of genetic variation to population size in wildlife. Conserv Biol 10:1500–1508.
Frazer NB, Gibbons JW, and Greene JL, 1990. Life tables of a slider turtle population. In: Life history and ecology of the slider turtle (Gibbons JW, ed). Washington, DC: Smithsonian Institution Press; 183–200.
Gibbons JW, 1990. The slider turtle. In: Life history and ecology of the slider turtle (Gibbons JW, ed). Washington, DC: Smithsonian Institution Press; 1–18.
Gibbons JW, Green JL, and Congdon JD, 1983. Drought-related responses of aquatic turtle populations. J Herpetol 17:242–246.
Gibbons JW and Semlitch RD, 1982. Survivorship and longevity of a long-lived vertebrate species: How long do turtles live? J Anim Ecol 51:523–527.
Hedrick PW and Gilpin ME, 1997. Genetic effective size of a metapopulation. In: Metapopulation biology: ecology, genetics, and evolution (Hanski IA and Gilpin ME, eds). New York: Academic Press; 165–181.
Kaj I and Lascoux M, 1999. Probability of identity by descent in metapopulations. Genetics 152:1217–1228.
Maruyama K and Tachida H, 1992. Genetic variability and geographical structure in partially selfing populations. Jpn J Genet 67:39–51.
McCauley DE, 1993. Genetic consequences of extinction and recolonization in fragmented habitats. In: Biotic interactions and global change (Kareiva PM, Kingsolver JG, and Huey RR, eds). Sunderland, MA: Sinouer Associates; 217–233.
Nunny L and Elam DR, 1992. Estimating the effective population size of conserved populations. Conserv Biol 8:175–184.
Quinn TP, 1990. Current controversies in the study of salmon homing. Ethol Ecol Evol 2:49–63.
Reading CJ, Loman L, and Madsen T, 1991. Breeding pond fidelity in the common toad, Bufo bufo. J Zool 225:201–211.
Scribner KT, Arntzen JW, and Burke T, 1994. Comparative analysis of intra- and inter-population genetic diversity in Bufo bufo using allozyme, single locus microsatellite and minisatellite, and multilocus minisatellite data. Mol Biol Evol 11:737–748.[Abstract]
Scribner KT, Arntzen JW, and Burke T, 1997. Effective number of breeding adults in Bufo bufo estimated from age-specific variation at minisatellite loci. Mol Ecol 6:701–712.[Medline]
Scribner KT, Arntzen JW, Cruddace N, Oldham RS, and Burke T, 2001. Environmental correlates of toad abundance and population genetic diversity. Biol Conserv 98:201–210.
Scribner KT, Congdon JD, Chesser RK, and Smith MH, 1993. Annual differences in female reproductive success affect spatial and cohort-specific genotypic heterogeneity in painted turtles. Evolution 47:1360–1373.
Scribner KT, Evans JE, Morreale SJ, Smith MH, and Gibbons JW, 1986. Genetic divergence among populations of the yellow-bellied slider turtle (Pseudemys scripta) separated by aquatic and terrestrial habitats. Copeia 1986:691–700.
Scribner KT, Morreale SJ, Smith MH, and Gibbons JW, 1995. Factors contributing to temporal and age-specific genetic variation in the freshwater turtle Trachemys scripta. Copeia 1995:970–977.
Slater FM, Gittins SP, and Harrison JD, 1985. The timing and duration of the breeding migration of the common toad (Bufo bufo) at Llandrindod Wells lake, mid-Wales. Br J Herpetol 6:424–426.
Slatkin M, 1991. Inbreeding coefficients and coalescence times. Genet Res 58:167–175.[Web of Science][Medline]
Smith MH and Scribner KT, 1990. Population genetics of the slider turtle. In: Life history and ecology of the slider turtle (Gibbons JW, ed). Washington, DC: Smithsonian Institution Press; 74–81.
Sugg DW and Chesser RK, 1994. Effective population sizes with multiple paternity. Genetics 137:1147–1155.[Abstract]
Waples RS, 1989. A generalized approach for estimating effective population size from temporal changes in allele frequency. Genetics 121:379–391.
Weir BS and Cockerham CC, 1984. Estimating F-statistics for the analysis of population structure. Evolution 38:1358–1370.[Web of Science]
Whitlock MC and Barton NH, 1997. The effective size of a subdivided population. Genetics 146:427–441.[Abstract]
Whitlock MC and McCauley DE, 1999. Indirect measures of gene flow and migration: Fst = 1/(4Nm+l1). Heredity 82:117–125.
Wood CC, 1995. Life history variation and population structure in sockeye salmon. In: Evolution and the aquatic ecosystem: defining unique units in population conservation (Nielsen J and Powers DA, eds). Symposium no. 17. Bethesda, MD: American Fisheries Society; 195–216.
Wright S, 1943. Isolation by distance. Genetics 28:114–138.
Wright S, 1965. The interpretation of population structure by F-statistics with special regard to systems of mating. Evolution 19:395–420.[Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||