SEARCH

SEARCH BY CITATION

Keywords:

  • extinction;
  • density dependence;
  • environmental stochasticity;
  • genetic variation;
  • inbreeding–stress interaction;
  • population dynamics;
  • wolf spiders

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Ample evidence exists that an increase in the inbreeding level of a population reduces the value of fitness components such as fecundity and survival. It does not follow, however, that these decreases in the components of fitness impact population dynamics in a way that increases extinction risk, because virtually all species produce far more offspring than can actually survive. We analyzed the effects of the genetic quality (mean fitness) of individuals on the population growth rate of seven natural populations in each of two species of wolf spider in the genus Rabidosa, statistically controlling for environmental factors. We show that populations of different sizes, and different inbreeding levels, differ in population dynamics for both species. Differences in population growth rates are especially pronounced during stressful environmental conditions (low food availability) and the stressful environment affects smaller populations (<500 individuals) disproportionately. Thus, even in an invertebrate with an extremely high potential growth rate and strong density-dependent mortality rates, genetic factors contribute directly to population dynamics and, therefore, to extinction risk. This is only the second study to demonstrate an impact of the genetic quality of individual genotypes on population dynamics in a wild population and the first to document strong inbreeding–environment interactions for fitness among populations. Endangered species typically exist at sizes of a few hundred individuals and human activities degrade habitats making them innately more stressful (e.g. global climate change). Therefore, the interaction between genetic factors and environmental stress has important implications for efforts aimed at conserving the Earth's biodiversity.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The erosion of genetic variation, the loss of population fitness and their combined effect on extinction risk is a central topic in conservation biology (Reed & Frankham, 2003; Reed, 2005; Ouborg, Vergeer & Mix, 2006; Vilas et al., 2006). Several lines of evidence suggest that genetic factors may impact extinction risk. Severely inbred populations have been rescued from long-term decline via the introduction of genes from one or a few individuals from outbred populations of the same species (Westemeier et al., 1998; Madsen et al., 1999; Richards, 2000; Vilàet al., 2003; Pimm, Dollar & Bass, 2006) and in highly inbred populations of butterflies and plants genetic variation (heterozygosity levels) has been shown to be directly correlated with extinction risk independent of population size (Newman & Pilson, 1997; Saccheri et al., 1998; Vilas et al., 2006). However, it has been suggested that genetic factors might be negligible in all but the most inbred populations; populations so small that they are likely doomed to extinction in the near future due to demographic or environmental stochasticity regardless of their genetic constitution (Lande, 1988).

Suggestions that genetic factors are important only to very small populations are countered by recent research. Fitness components (primarily fecundity measures) are positively and consistently correlated in a log-linear fashion with population size and/or linearly with measures of genetic diversity (Reed & Frankham, 2003; Reed, 2005, 2007; Reed, Nicholas & Stratton, 2007), even in populations exceeding 10 000 individuals. Furthermore, computer simulations have demonstrated significant impacts of inbreeding depression on extinction risk in populations with carrying capacities of up to 2000 individuals (O'Grady et al., 2006).

Another, stronger, argument against the widespread impact of genetic factors on extinction risk in natural populations is the ubiquity of density-dependent population dynamics (Turchin, 1995; Lande et al., 2002; Heering & Reed, 2005; Sibly et al., 2005; Brook & Bradshaw, 2006). Density-dependent mortality rates suggest that decreases in fecundity are compensated for by increases in survival rates for those offspring born into the less-competitive environment. Virtually all organisms produce far more offsprings than can actually survive to sexual maturity. Thus, one must ask: Do reductions in fecundity or survival typically associated with increasing inbreeding coefficients have direct consequences on population dynamics and, therefore, on extinction risk? Models examining genetic impacts on extinction risk often incorporate only a general and simplified form of density dependence (e.g. a ceiling carrying capacity) that is often limited to reproductive rates and not mortality rates and/or assumes that the genetic risks are due only to the effects of increased homozygosity through inbreeding (Reed et al., 2003b; Henle, Sarre & Wiegand, 2004; Reed, 2004; Sabo, Holmes & Kareiva, 2004; O'Grady et al., 2006).

Genes express their effects on individuals; however, the effect of variation in the genetic quality (fitness) of individuals on population dynamics has seldom been explored. A notable exception is a recent paper demonstrating an effect of genetic variation at a single locus on population growth (Hanski & Saccheri, 2006). To address the need for data examining the direct effects of genetic factors on population dynamics in natural environments, we conducted a 3-year field study on two semelparous species of wolf spider Rabidosa punctulata and Rabidosa rabida. Seven populations of each species were assayed for population size, genetic diversity levels, parasitism rates, prey capture rates, fecundity, survival rates from emergence to adulthood and the annual population growth rate. We were interested in finding in whether the mean fitness (genetic quality) of individuals impacts the realized population growth rate despite density-dependent population dynamics. The mean fitness of individuals differs among populations of different long-term effective populations sizes, primarily because this determines the degree to which the fate of alleles is determined by stochastic versus deterministic processes (Reed, 2005). We measure both genetic variation and population size and use the two as surrogates for the average genetic quality of individuals in the population. Mean fecundity of females in these populations, of both species, is highly correlated with population size and with measures of genetic variation (Reed et al., 2007). This is one of a very small number of studies to examine overall genetic effects on the population dynamics of wild populations and is the first study to look for them outside highly inbred populations.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

General methods

We identified seven field sites, in northern Mississippi (USA), that contained populations of both R. punctulata and R. rabida. The two species are habitat specialists limited to meadows and fields. Study sites were isolated from each other by forest and/or a matrix of human-dominated habitat (e.g. residential neighborhoods, roads, commercial centers), with between-population distances ranging from 1.3 to 62.5 km. Considering the patchiness of the habitat and the hostility of the habitat matrix surrounding the habitat patches, we suspected that gene flow among sites was very limited or non-existent. Analysis of molecular data corroborated this assumption for R. rabida. We genotyped 15–25 individuals of this species, from each population, at 15 polymorphic microsatellite loci and estimated <1.5 effective migrants per generation among all pairs of populations (Reed et al., 2007).

Our sites varied considerably in vegetation cover and relief, but most sites were dominated by broom sedge Andropogon virginicus and many contained wild blackberry Rubus sp. or kudzu Pueraria lobata. Although the two species of wolf spider are morphologically very similar, have nearly identical diets and are congeners, their phenologies and life histories differ considerably. Rabidosa punctulata matures from mid-September to mid-October and the inseminated females delay reproduction until March. Rabidosa rabida matures in mid-June through July and reproduction occurs in July and early August. When mature, both species are among the largest spiders found in their habitat. However, adult female R. rabida weigh c. 600 mg and adult female R. punctulata around 415 mg. Both species are semelparous with R. punctulata typically producing a clutch of 100–200 offsprings of relatively large size (1.35 mg) while R. rabida typically produces 300–400 offsprings with a mean weight of 1.05 mg each. These wolf spiders make excellent model organisms for the study of population dynamics because of their discrete generations and the ease of estimating fecundity and survivorship.

Fecundity

Female R. punctulata and R. rabida carry their offspring for 30–40 days in a silken egg sac attached to their spinnerets. Following this period, the offspring emerge from the egg sac and spend c. 14 days clinging to specialized hairs located on the female's abdomen. Females found with an egg sac were brought into the laboratory and once the offspring emerged and dispersed, they were counted and weighed. Fecundity (fec) was estimated as the mean number of live offspring produced per adult female in a population for a given year. The total number of female R. punctulata sampled for fecundity assays was 362 or >17 per population per year. The total number of R. rabida sampled for fecundity assays was 285 or >13 per population per year.

Survival and population growth rate

The total number of offsprings produced in a population was estimated by multiplying the mean fecundity of females at time t by the number of adult females in the population at time t. The number of individuals reaching sexual maturity in the following year (Nt+1) divided by the number of offspring born the year before provides an estimate of the proportion of offspring surviving. Paired t-tests were used to test whether survival rates differed between years.

The population growth rate is equal to

  • image(1)

where N is the population size of all sexually mature adults and t the time measured in years (generations).

Prey capture rates

The density of potential prey items at each site, for R. punctulata and R. rabida, was estimated from random quadrat sampling in 2003. As absolute prey density did not explain differences in fecundity among populations, we converted absolute prey density to a relative prey density, by dividing by the density of spiders, which resulted in an improved fit. However, the measure of prey capture with the most explanatory power proved to be the proportion of spiders captured with prey, which we call the prey capture rate. After 2003, we used the proportion of spiders captured with prey as our only estimate of the prey capture rate.

During the course of the experiment, we captured 5211 individual R. punctulata and 7039 individual R. rabida. At the time of capture, individuals were assayed for the presence or absence of a food item. The proportion of individuals with a prey item was the prey capture rate for that year at that site. It should be emphasized that these estimates of the prey capture rate were made by sampling repeatedly from each population during the growing season for each species and, therefore, reflects the effects of food limitation on growth and survival.

Population size

The number of adult spiders in each population was estimated using mark–recapture techniques. At night, R. punctulata and R. rabida are easily collected using a flashlight to locate individuals by the light reflected back from the tapetum of their eyes. Because juveniles routinely molt, they cannot be given permanent identification. Thus, to estimate total population size, we use repeated mark–recapture estimates. Juvenile spiders were captured, clearly marked with non-toxic paint and then released. The following night, the same site was searched again and the population size estimated using the Lincoln–Peterson method. The unbiased estimator, when individuals are replaced once captured, is

  • image

where M is the number of individuals marked in the first sample, C the total number of individuals captured in the second sample and R the number of individuals in the second sample that are marked. All assumptions of the model are met in this study. The assumption of a closed population of constant size is ensured by the extremely low levels of gene flow and the short time interval between samples. Captures are almost certainly random and unaffected by being marked, as individuals are located by their eyeshine, and marks are permanent as adult R. punctulata can be easily identified 6 months after marking. While we cannot show that the marks did not affect survival, we do know that in six tests (2 species × 3 years) only once was there a significant difference between the proportion of marked and unmarked individuals captured with prey and that instance favored the marked individuals. It is also worth noting that there were positive and highly significant relationships between our estimates of population size from mark–recapture procedures and the amount of genetic diversity present in both species (Reed et al., 2007).

Mark–recapture estimates were repeated three to five times, at each site, during the course of the growing season. Once populations consisted of a significant (>15%) proportion of mature (reproductive) individuals, the mark–recapture estimates were stopped. Linear regression using time as the independent variable and estimated population size as the dependent variable was then performed to make a best estimate of population size at the onset of the reproductive season.

Genetic variation

Details of our methods are provided in Reed et al. (2007). Briefly, in order to estimate levels of neutral genetic variation, corroborate our estimates of population size and support our assumptions of genetic isolation, we genotyped 15–25 individuals from each R. rabida population at 15 polymorphic microsatellite loci. Expected heterozygosity in these seven populations was positively and significantly correlated with the harmonic mean of population size (r2=0.785, P<0.01).

Broad-sense heritabilities were estimated for six quantitative traits in the seven populations of R. punctulata using the correlation of phenotypic values among siblings. Heritabilities were estimated from 12 to 15 families per population from known parents that were captured when penultimate (the molt before reaching sexual maturity) and mated in the laboratory. The quantitative traits for which heritabilities were estimated included abdomen length, abdomen width, carapace width, growth rate, number of black spots on the under side of the abdomen and proportion of the underside of the abdomen that was black. A mean heritability for the population was estimated as the arithmetic mean of the heritability estimates for each of the six traits individually. Mean heritability in these seven populations was positively and significantly correlated with the harmonic mean of population size (r2=0.820, P=0.005).

Statistical analyses

The combined effects of long-term effective population size (genetic diversity), changes in prey capture rates, fecundity and the interaction between genetic diversity and changes in prey capture rates on survivorship were examined using Akaike's information criterion adjusted for sample size (AICc). The same statistical methods were used for examining the combined effects of genetic diversity, changes in prey capture rates and the interaction between genetic diversity and changes in prey capture rates on the population growth rate. The change in the prey capture rate (ΔPCR) is defined as ln (PCRt+1/PCRt) and the population growth rate (r) is defined as ln (Nt+1/Nt), where N is population size and t represents the time in discrete generations (1-year interval).

We used AICc values as an objective means for selecting the best approximating model for survival to sexual maturity and the overall population growth rate. For each model, Akaike weights (wi) were calculated for each predictor variable used in a candidate model. A given wi is considered as the weight of evidence in favor of model i being the best approximating model or can be viewed as measuring the relative importance of a predictor variable across a set of candidate models. Models were constructed using some measure of genetic variation (heritability, or expected heterozygosity), the prey capture rate and the interaction between them. For survivorship, fecundity in the prior year was also included. Only models with AICc differences (Δi)<7.00 (compared with the best-fit model) were included as candidate models. The absolute fit of the best model was described using the R2 value from multiple regression adjusted for the number of independent variables used in the model.

Because of the small number of populations used in this study and the lack in some cases of a clearly best single model, we summed Akaike weights across all models for all years (separately for each species) to determine the relative importance of the variables in the model. For both species the summed Akaike weights are presented as relative weights with the most important variable receiving a score of one and the other variables scaled as a proportion of the most important variable.

To emphasize the effects of population size on population dynamics, we perform linear regression using log10 population size as the independent variable and population growth rate (r) as the dependent variable, after having accounted for the effects of ΔPCR (i.e. we used the residuals from the regression of r onto ΔPCR as the dependent variable).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Survival rates

For R. punctulata, a clearly best single model describing survival rates could be identified for both years (Table 1). The proportion of the variation in survival rates among populations explained by multiple regression, adjusted for the number of independent variables used, was very high (R2=0.957 and R2=0.881) and the overall model was statistically significant. The best-fit model for survivorship of spiderlings born in 2003 was the relative change (log PCRt+1/PCRt) in the prey capture rate (ΔPCR), fecundity in the prior year (fect−1), heritability and the interaction between ΔPCR (defined as PCRt+1/PCRt for purposes of the interaction) and H2. The best-fit model for survivorship of spiderlings born in 2004 was ΔPCR, fect−1 and H2. Thus, the best-fit models for both years were similar, differing only in the exclusion or inclusion of the interaction term.

Table 1.   Results of model selection, for survival rates, using Akaike's information criterion (AICc)
ModelAICcΔiwi
  1. The relative weight of evidence for each model is provided by the wi value.

  2. fec, fecundity; HE, expected heterozygosity; PCR, prey capture rate.

Rabidosa punctulata 2003–2004
 ΔPCR+fect−1+H2+ΔPCR ×H2−80.480.001.000
R. punctulata 2004–2005
 ΔPCR+fect−1+H2−51.460.001.000
Rabidosa rabida 2003–2004
 ΔPCR+fect−1+HE−63.170.001.000
R. rabida 2004–2005
 ΔPCR+HE−22.160.000.445
 ΔPCR+HE+ΔPCR ×HE−21.960.200.404
 ΔPCR−20.002.160.151

All seven populations of R. punctulata experienced reduced prey capture rates and decreased population size from 2003 to 2004 and all populations experienced increased prey capture rates and increased population size from 2004 to 2005. The interaction term was included in the best-fit model during a time of decreasing resources and increased competition. The inclusion of the prior year's fecundity in the best-fit model also points to density-dependent mortality due to food limitation, with high fecundity in the prior year having a negative effect on survivorship of that cohort during the growing season. Mean survival among populations of the 2003 cohort until maturity in 2004 was 1.12±0.10% as compared with the 2004 cohort's survival rate of 4.90±0.76% (paired t-test, d.f.=6, t=5.02, P<0.01). Clearly the first year was more stressful (competitive) than the second year.

For R. rabida, a single clearly best-fit model describing survivorship could be identified in only one of the 2 years (Table 1). The amount of variation in survival rates among populations explained by multiple regression, adjusted for the number of independent variables used, was very high for both years (R2=0.953 and R2=0.863) and the overall model was statistically significant. The best-fit model for survivorship of the 2003 cohort was ΔPCR, fect−1 and HE, having more than 90 times as much support as the second best model. The best-fit model for survivorship of the 2004 cohort was ΔPCR and expected heterozygosity (HE). However, there is a nearly identical level of support for a model consisting of ΔPCR, HE and the interaction between them.

All seven populations of R. rabida experienced increased prey capture rates and increased population size from 2003 to 2004 and five of seven populations experienced decreased prey capture rates and decreased population size from 2004 to 2005. Survival of the 2003 cohort until maturity in 2004 was 1.28±0.09% as compared with the 2004 cohort's survival rate of 0.47±0.06% (paired t-test, d.f.=6, t=6.32, P<0.005). Clearly the second year was more stressful (competitive) than the first year. Again, the interaction term appears important during times of environmental stress.

Population growth rates

Trivially, population growth rates can be accurately predicted from the fecundity and survival estimates. However, we wished to establish what ecological and/or genetic factors were responsible for variation among populations in the fecundity and survival rates, and thus contributed to variation in the overall population growth rate.

For R. punctulata, a single clearly best model describing the population growth rate could be identified for both years (Table 2). The proportion of variation in growth rates among populations explained by multiple regression was very high (R2=0.945 and R2=0.866) and the overall model was highly significant. The best-fit model for population growth between 2003 and 2004 was ΔPCR+H2. This mode had more than six times the support of the second best model, which was ΔPCR and ΔPCR ×H2. The best-fit model for population growth from 2004 to 2005 was ΔPCR alone, having a little over three times as much support as the second best model (H2). Heritability and its interaction with prey capture rates were more important during times of environmental stress (2003–2004).

Table 2.   Results of model selection, for population growth rates (r), using Akaike's information criterion (AICc) adjusted for small sample sizes
ModelAICcΔiwi
  1. The relative weight of evidence (likelihood) for each model is provided by the wi value.

  2. fec, fecundity;HE, expected heterozygosity; PCR, prey capture rate.

Rabidosa punctulata 2003–2004
 ΔPCR+H2−31.860.000.864
 ΔPCR+ΔPCR ×H2−28.163.700.136
R. punctulata 2004–2005
 ΔPCR−29.040.000.539
 H2−26.622.420.161
 ΔPCR+ΔPCR ×H2−26.522.520.153
 H2+ΔPCR ×H2−26.442.600.147
Rabidosa rabida 2003–2004
 ΔPCR+HE+ΔPCR ×HE−26.970.001.000
R. rabida 2004–2005
 ΔPCR+HE−29.380.000.470
 ΔPCR+ΔPCR ×HE−28.590.790.317
 HE−27.751.630.213

For R. rabida, a single clearly best-fit model describing the population growth rate could be identified in only one of the 2 years (Table 2). The amount of variation in population growth rates among populations explained by multiple regression was very high for both years (R2=0.737 and R2=0.916) and the overall model was highly significant. The best-fit model for population growth from 2003 to 2004 was ΔPCR, HE and the interaction term ΔPCR ×HE, having over 60 times as much support as the second best model. The best-fit model for population growth from 2004 to 2005 was ΔPCR and HE. However, there is a similar level of support for a model consisting of ΔPCR and ΔPCR ×HE and for HE alone.

Relative importance of predictor variables

Changes in the prey capture rate were the most important variable impacting survival rates for both species (Table 3). This is not surprising as numerous studies have shown spiders to be food-limited (Wise, 2006). However, genetic variation was nearly as important a factor in explaining variation among populations in survival rates for both species. The positive relationship suggests that genetic factors impact this fitness component even when environmental factors are taken into consideration. Fecundity in the previous year was also very important in determining survivorship. In this case, the relationship was negative. High levels of fecundity in the previous year set the stage for increases in competition for food and strongly density-dependent mortality rates, potentially through starvation and cannibalism (Wise, 2006). The interaction between ΔPCR and population size was of less importance, particularly in R. rabida.

Table 3.   The relative importance of different predictor variables for survival rates and population growth rates
 Rabidosa punctulataRabidosa rabida
  1. Importance was determined by averaging Akaike weights across years for all predictor variables.

  2. PCR, prey capture rate.

Relative importance (survival)
 ΔPCR1.001.00
 Genetic variation1.000.92
 Fecundity (fect−1)1.000.70
 ΔPCR × genetic variation0.500.20
Relative importance (population growth rate)
 ΔPCR1.001.00
 Genetic variation0.690.94
 ΔPCR × genetic variation0.260.74

Changes in the prey capture rate were also the most important factor determining the overall population growth rate. However, genetic variation was 69% as important as changes in prey capture rates in R. punctulata and 94% as important in R. rabida. The interaction between ΔPCR and population size was also important, being c. 50% as important as ΔPCR. Thus, the effects of genetic diversity as well as their interaction with environmental conditions play a significant and important role in determining population dynamics.

The effect of population size on population growth rate

To further explore the effects of long-term effective population size on population growth rates, linear regression was performed using ΔPCR as the independent variable and population growth rate (r) as the dependent variable. Thus, ΔPCR was treated as a covariate. The residuals from this regression were then regressed against population size (log10 N). This provides a measure of population growth rate with the variation due to differences in prey capture rates removed. Covariate analysis using r as the dependent variable gave contrasting results depending on whether prey capture rates were decreasing or increasing. In years where the prey capture rate was increasing, the regression was positive but not significant (R. punctulata, r2=0.035, P=0.69; R. rabida, r2=0.378, P=0.14). In years where the prey capture rate was decreasing relative to the prior year, the regression was positive and significant for both species (R. punctulata, r2=0.822, P<0.005; R. rabida, r2=0.609, P<0.05) (Fig. 1a and b). Thus, population size and relative inbreeding levels become more important to population dynamics under stressful conditions. At population sizes <500, the population growth rate becomes strongly negative under stressful environmental conditions and will greatly increase the probability of extinction (Fig. 1a and b) in both species.

image

Figure 1.  Linear regression of the population growth rate adjusted for ΔPCR (residuals of a regression using r as the dependent variable and ΔPCR as the independent variable) onto population size for Rabidosa punctulata (a) and Rabidosa rabida (b). During stressful/competitive environmental conditions, the residuals become positively associated with log10 population size, demonstrating that larger populations have greater population growth rates under similar environmental conditions. PCR, prey capture rate.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We have shown that the genetic quality (mean fitness) of individuals in populations of two species of wolf spider directly impact the dynamics of those populations. Specifically: (1) This study demonstrates an impact of genetic health of a population on its dynamics across a wide range of population sizes. Population size and genetic diversity impact population dynamics, and therefore extinction risk, even when those populations consist of thousands of adult individuals. (2) Population size and genetic diversity impact population dynamics in the two species assayed, despite the fact that they are highly fecund invertebrates with strongly density-dependent mortality rates under field conditions. (3) Genetic effects were strongest during stressful environmental conditions, when prey capture rates were low and competition among spiders high. These inbreeding–environment interactions are known extensively from laboratory experiments (reviewed by Armbruster & Reed, 2005); however, population level impacts on wild populations have strong ramifications for the conservation of biodiversity. We elaborate on these three points below.

Inbreeding and population dynamics

Long-term effective population size is expected to impact population fitness and components of fitness by influencing the degree to which the fate of an allele is deterministic versus stochastic and by limiting the number of beneficial mutations available for adaptive evolution. However, there is widespread skepticism concerning the importance of genetic factors to population persistence in populations smaller than a few tens or possibly hundreds of individuals. We demonstrate that for seven populations of two different species of wolf spider, variation among populations in the proportion of spiderlings surviving to sexual maturity is attributable to changes in prey capture rates and long-term effective population size in almost equal parts. Elsewhere, we have shown that variation in mean fecundity among these same populations of spiders is attributable to population size and prey capture rates in about equal measures (Reed et al., 2007). The effect of long-term effective population size on the overall population growth rate is slightly smaller than for each of the fitness components individually, being only 80% as important as differences in the prey capture rate. This is not unexpected. Because of the negative correlation between fecundity among adult females in the prior year and survival within the following year's cohort, the two inbreeding effects partially cancel each other out. Despite this, inbreeding levels had a strong and significant effect on overall population growth rate in these populations. Furthermore, the relationship between the population growth rates, adjusted for the prey capture rate, was log linear with respect to population size and showed no signs of reaching an asymptote despite assaying populations of >20 000 adult individuals. Thus, inbreeding levels impact population dynamics over a range of population sizes.

Genetic effects despite density-dependent mortality rates

Previous work on the link between genetic factors and extinction risk have relied on correlations between fitness components and population size or genetic variation (Reed & Frankham, 2003; Reed, 2005), on laboratory experiments (Frankham, 1995; Reed, Briscoe & Frankham, 2002), or on computer simulations (Reed et al., 2003b; O'Grady et al., 2006) to implicate inbreeding as a component of extinction risk. The first method assumes that reductions in fitness components translate into decreases in population growth rate. This is not necessarily true given the ubiquity of density-dependent mortality and fecundity rates and the fact that reductions in population growth rates may not increase extinction risk substantially except over a small range of stochastic growth rates near zero (Lande, 1995; Reed et al., 2003b; Reed, 2007). Conclusions drawn from the second and third methods rely on necessarily simplified versions of reality and therefore need to be documented in natural populations. It has been pointed out (Lande, 1998; Puurtinen et al., 2004; Pimm et al., 2006) that the true gauge of any factor impacting extinction risk will be its effect on population dynamics.

Evidence that these populations are regulated in part by density-dependent mechanisms comes from the fact that the best-fit model for survival for both species always includes changes in the prey capture rates and also includes the mean fecundity of adult females in both years for R. punctulata and 1 year for R. rabida. Thus, fluctuations in population size are strongly correlated with changes in the availability of prey and the survival rate is also strongly influenced by the density of spiderlings competing for the available prey. Despite the food-limited and density-dependent regulation of population dynamics and the fact that females of both species produce hundreds of offspring, some measure of genetic diversity also appeared in the best-fit models of survival in both years for both species. Thus, long-term effective population size impacted the overall population growth rate strongly. It affected it most powerfully in years when prey availability was decreasing. Thus, density-dependent mechanisms of population regulation did not mask genetic effects on population dynamics, but made them more pronounced.

Environment–inbreeding interaction

The results of this study have important implications for the conservation and management of endangered species. Human activities typically degrade habitats by fragmenting them, polluting them and introducing foreign species as competitors and predators. This results in species facing increasingly stressful environmental conditions. Prominent examples include disease, climate change and the interaction between them. In about 60% of wildlife epidemics, the causative pathogen is likely of exotic origin (Dobson & Foufopoulos, 2001) and there is increasing evidence that decreased genetic variation is associated with increased susceptibility to infectious diseases and other parasites (Coltman et al., 1999; Meagher, 1999; Cassinello, Gomendio & Roldan, 2001; Acevedo-Whitehouse et al., 2003; Reid, Arcese & Keller, 2003; Spielman et al., 2004; Whitman et al., 2006; Hale & Briskie, 2007). Global climate change is forcing many species to adapt to warming temperatures and increased drought periods at an unprecedented pace (Bradshaw & Holzapfel, 2001; Parmesan & Yohe, 2003) and there is considerable consternation over whether many threatened species will have sufficient genetic variation to track rapidly changing environmental conditions (Bürger & Lynch, 1995; Gomulkeiwicz & Holt, 1995; Boulding & Hay, 2001; Etterson & Shaw, 2001; Stenseth & Mysterud, 2002; Reed et al., 2003a). Furthermore, deteriorating environmental conditions and habitat fragmentation can interact with disease threats to increase the threat of extinction further (Kiesincher & Blaustein, 1995; Reed et al., 2003b; Reed, 2004; Pounds et al., 2006). Threatened populations continue to be reduced in size and face increasing levels of stress.

The negative interaction between inbreeding and stress and its impact on population growth rates and population dynamics demonstrated in this study suggest that emphasis should be placed on preserving genetic variation in threatened species not only because of its role in the long-term evolutionary potential of populations but also because of its short- to mid-term impacts on extinction risk. Currently the major categorization systems for ranking priority conservation actions (e.g. IUCN categorization system, the US nature conservancy system) implicitly acknowledge small population size as elevating extinction risk. However, they do not account for the interactions between environmental quality and the genetic quality of the population. Populations containing fewer than 500 individuals may experience significantly elevated risks of extinction due to inbreeding–environment interactions and the fact that fluctuations in environmental quality are temporally correlated. Thus, years with low-quality environmental conditions are more likely to be followed by another year with poor environmental conditions than good. Such interactions are important and should be considered when estimating the relationship between population size and extinction risk (Reed et al., 2003c) and when designing reserve networks for the long-term persistence of biodiversity.

Increased risk of extinction

Robert (2006) suggests that negative environmental perturbations may actually improve population persistence due to more efficient purging of the mutation load. However, our empirical results are in agreement with most experimental results suggesting that small populations are not purged of their genetic load efficiently enough to counter the effects of the fixation rate for effectively neutral alleles (e.g. Reed & Bryant, 2001). If inbreeding was effective at purging the genetic load of populations, the smaller populations in our study should have performed equally well, especially during stressful environmental episodes, rather than having lower fitness and this being particularly true during stressful environmental conditions. Part of the problems likely stems from the fact that Robert (2006) assumes that there is only one type of environmental disturbance, which would increase the efficacy of purging greatly. In fact, populations face multiple negative environmental perturbations and the genetic correlations among various stresses are probably low (Armbruster & Reed, 2005; Swindell & Bouzat, 2006; Reed, 2007).

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The University of Mississippi provided funding for this research. Allison Derrick, Christian Felton, Pat Miller, Alex Teoh and Winter Williams helped collect spiders. Juan Bouzat, Steven Brewer, Richard Frankham and Julian O'Grady provided comments on a previous draft of this paper.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References