Inbreeding depression in the wild

Abstract

Despite its practical application in conservation biology and evolutionary theory, the cost of inbreeding in natural populations of plants and animals remains to a large degree unknown. In this review we have gathered estimates of inbreeding depression (δ) from the literature for wild species monitored in the field. We have also corrected estimates of δ by dividing by F (coefficient of inbreeding), to take into account the influence that the variation in F will have on δ. Our data set includes seven bird species, nine mammal species, four species of poikilotherms (snakes, fish and snails) and 15 plant species. In total we obtained 169 estimates of inbreeding depression for 137 traits; 81 of those estimates included estimates of F. We compared our mammalian data (limited to those traits related to juvenile mortality) to the estimates for captive zoo species published by Ralls et al. (1988) to determine if, as predicted from the literature, natural estimates of inbreeding depression are higher than captive estimates. The mean δ ± SE (significantly different from zero and not corrected for F) for homeotherms was 0.509 ± 0.081; for poikilotherms, 0.201 ± 0.039; and for plants, 0.331 ± 0.038. Levels of inbreeding depression this high in magnitude will be biologically important under natural conditions. When we limited our data set to mortality traits for mammals and corrected for F=0.25 (as is the case for the Ralls et al. data set), we found a significant difference between the two data sets; wild estimates had a substantially higher mean cost of inbreeding at F = 0.25: 2.155 (captive species: 0.314). Of the 169 estimates of δ, 90 were significantly different from zero, indicating that inbred wild species measured under natural conditions frequently exhibit moderate to high levels of inbreeding depression in fitness traits.

Introduction

Inbreeding depression is the decline in the value of a trait as a direct consequence of inbreeding (Wright, 1977; Shields, 1987). The most common estimates of inbreeding depression involve traits that are closely related to fitness, such as reproductive traits (e.g. number of eggs laid, number of young surviving), or metric traits indirectly associated with fitness (e.g. ejaculate volume, plant height). The reduction of fitness after close inbreeding can be caused by a number of genetic factors: the unmasking of recessive deleterious alleles (Lande, 1994; Lynch et al., 1995), increased homozygosity and/or reduced allozyme variability (Falk & Holsinger, 1991; Brock & White, 1992; Pray et al., 1994; Vrijenhoek, 1994). Whatever the genetic mechanism, inbreeding depression is a real phenomenon that has received a substantial amount of attention in the literature (Ralls & Ballou, 1983, 1986; De Bois et al., 1990; Lacy et al., 1993; see Frankham, 1995a and Roff, 1997 for reviews).

Most of the literature concerning inbreeding depression has concentrated on domestic or captive-bred wild species (Ralls & Ballou, 1986; for a review see Lacy et al., 1993) because of the obvious difficulties of making estimates on wild species in nature. One of the most comprehensive data sets is that of pedigrees from zoo populations (Ralls et al., 1988). Forty captive populations belonging to 38 species show an average increase in mortality of 33% for inbred matings (Ralls et al., 1988). Ralls et al. (1988), p. 191 suggest that ‘the total costs of inbreeding in natural populations are probably considerably higher’, which would make the cost of inbreeding in natural populations of substantial evolutionary consequence. The implications of high levels of inbreeding depression to population extinction are obvious (Lande, 1988; Caro & Laurenson, 1994; Caughley, 1994). However, the degree of inbreeding depression in wild populations remains controversial (see Frankham, 1995a for a discussion). The two most commonly suggested reasons why inbreeding effects in natural populations may not be significant are: (i) animals in the wild avoid close inbreeding, and therefore do not manifest the deleterious fitness effects; and (ii) even if inbreeding does occur, animals are able, either behaviourally or physiologically, to deal with the deleterious genetic effects before they are manifest on a phenotypic level, whereas captive species, because of the conditions of captivity, cannot respond in such a manner. Although evidence of inbreeding depression in wild species has been published (see Frankham, 1995a for a short review), the lack of a comprehensive review across species has led to the remaining scepticism about its existence in natural populations (Caro & Laurenson, 1994).

The objective of the present study was to estimate the average inbreeding depression for wild species measured under natural conditions. We are concerned here not with whether inbreeding occurs in the wild (although we report the coefficient of inbreeding, F, for those studies for which it was available) but rather the consequences of inbreeding on characteristics of organisms living in the wild. Specifically, we attempt to answer two questions: (i) is inbreeding depression in wild populations of sufficient magnitude to be biologically important should inbreeding occur? and (ii) does the cost of inbreeding differ between natural and captive populations?

Methods

The data set

We obtained 169 estimates of inbreeding depression from the literature. The data set includes 35 species (20 animals, 15 plants) and 137 traits (see Appendix). We included only species that were sampled from wild populations or species that were artificially inbred in the laboratory, or glasshouse, for one generation and their progeny released, or grown, in the area from which their parents originated. Where more than one estimate was given for a particular trait, we included all estimates in the analysis.

To standardize relative differences in fitness traits, we used the coefficient of inbreeding depression δ (Lande & Schemske, 1985):

where X_I=inbred trait value and X_O=outbred trait value. To standardize estimates of δ further, we changed traits such as juvenile mortality (where it is expected that X_O < X_I) to juvenile survivorship (so that X_O > X_I). This way all estimates are ‘positive’, and the a priori prediction is that outbred values should be greater than inbred ones. Certain traits (e.g. sperm abnormalities in lions) that could not be modified because they were not expressed as portions of a total, were not used in the analysis. We included traits that were either directly related to fitness, e.g. total number of eggs laid, or traits indirectly related to fitness, e.g. juvenile weight.

Because the magnitude of inbreeding depression will vary with the inbreeding coefficient of the inbred individuals studied, F (Falconer, 1989), we corrected δ estimates by standardizing with respect to F. The change in trait value because of inbreeding is

where b is the slope of the relationship between trait value and F. Because X_O will vary among traits, we scale by dividing throughout by X_O giving

Because 1 − X_I/X_O is the measure of inbreeding depression, δ, we can simplify the equation to

Since $\delta =b_{X_{\text{o}}}F$, the standardized slope is equivalent to inbreeding depression when F=1. Therefore, dividing the estimates of δ by F allows for a standardized comparison of the cost of inbreeding. We obtained 81 estimates of F from 14 studies. We called the F-corrected data set $b_{X_{\text{O}}}$ (includes negative values because of X_O < X_I).

Statistical analysis

All statistical analyses were carried out using SYSTAT (Wilkinson, 1991). We divided the data set into estimates of δ that were significant and those that were nonsignificant, to determine how often significant levels of inbreeding depression were detected.

To determine if natural conditions increase inbreeding depression relative to captive conditions, we compared the mean inbreeding depression we obtained from the literature with the data set included in the Ralls et al. (1988) review of inbreeding depression in captive-bred populations of wild species. Ideally the most appropriate test would be a comparison between natural and captive conditions for the same traits in the same species (paired comparison). We were not able to conduct paired comparisons because of the lack of use of the same species between this and the Ralls et al. study. Ralls et al. (1988) calculated the slope of ln(survival) vs. inbreeding and then predicted the cost of inbreeding for a level of inbreeding of F=0.25. Because the Ralls et al. data set was limited to survival of offspring of mammals only, we limited our data set to traits directly related to survival in mammals. Our estimates were obtained from the δ data set and corrected for F=0.25. We used a Student’s t-test to determine if significant differences exist between mean estimates of the cost of inbreeding at F=0.25 between the two data sets.

Results

Magnitude of inbreeding depression

Theory suggests that females should not mate with their closest relatives unless the cost of inbreeding is less than 0.33 (Smith, 1979). In addition, an increased probability of extinction occurs just below intermediate levels (F=0.30–0.40) of inbreeding (Frankham, 1995b). We found very high mean estimates of inbreeding depression for species measured in the wild. For δ estimates, mean inbreeding depression ranged from 0.197 (poikilotherms) to 0.268 (homeotherms) (Table 1; 30% of estimates >0.33). In addition, mean (±SE) δ estimates that were significantly different from zero were 0.509 ± 0.081 for homeotherms, 0.201 ± 0.039 for poikilotherms and 0.331 ± 0.038 for plants. Most of these estimates of inbreeding depression are sufficiently high in magnitude (>0.33) to be considered biologically important (see Smith, 1979; Frankham, 1995b). In addition, most of the traits (80%), are directly associated with fitness.

Table 1 Mean inbreeding depression and standard errors for homeotherms, poikilotherms and plants using the two estimates δ and $b_{X_{\text{o}}}$

For $b_{X_{\text{o}}}$ estimates, mean inbreeding depression corrected for F ranged from 0.552 for plants to 0.818 for homeotherms (Table 1).

Wild and captive comparisons of the cost of inbreeding at F=0.25

The comparison of our data set (limited to only those inbreeding depression estimates of mortality of mammals and corrected for F=0.25, i.e. δ/0.25) with that of Ralls et al. (1988) revealed a highly significant difference between mean estimates for juvenile mortality (our data set (mean ± SE): n=9, x¯=2.155 ± 0.482; Ralls et al. data set: n=40, x¯=0.314 ± 0.044; t₄₇=7.687, P=0.0001). Even without correcting for F=0.25, our estimate was significantly higher than the Ralls et al. estimate (x¯=0.539 ± 0.121; t₄₇=2.061, P=0.04). Although inbreeding depression normally ranges between 0 and 1 (unless the survival of inbreds exceeds that of outbreds), our calculated mean cost of inbreeding of 2.155 results from the correction using F=0.25. As predicted by Ralls et al. (1988), wild estimates of the cost of inbreeding at F=0.25 are substantially higher than captive estimates.

Discussion

We found that statistically significant levels of inbreeding depression in the wild are detected ≈54% of the time when species are known to be inbred. When significant, mean inbreeding depression (not corrected for the coefficient of inbreeding, F) ranged from 0.20 in poikilotherms to 0.51 in homeotherms. When corrected for F, mean inbreeding depression for all estimates ranged from 0.55 in plants to 0.82 in homeotherms. The analysis using only mammals revealed significantly greater estimates of the cost of inbreeding at F=0.25 from free-ranging mammals than estimates from captive populations (2.16 and 0.31, respectively). Therefore, as predicted by Ralls et al. (1988), conditions experienced in the wild increase the cost of inbreeding (similar findings have been made for plants; reviewed by Roff, 1997).

Although we have demonstrated that the cost of inbreeding under natural conditions is much higher than under captive conditions, we lack sufficient data to determine which environmental factors cause such an increase. Inbreeding depression is typically more severe in harsher environments (Falk & Holsinger, 1991; Hoffmann & Parsons, 1991; Latter et al., 1995; for a review see Miller, 1994). Environmental factors such as unpredictable rainfall, fluctuating temperatures and limiting resources to feed young are all likely to have a significant effect on juvenile mortality. Weak inbred young that would normally die in the wild would most likely survive in captivity with veterinary assistance (Ralls et al., 1988). Some studies have shown that individuals with relatively low allozyme heterozygosity and/or with a high number of lethal equivalent alleles are much more susceptible to factors that may not affect “normal” individuals (Pierce & Mitton, 1982; O’Brien et al., 1985; Mitton et al., 1986; Murphy et al., 1987; Ralls et al., 1988; Fritz & Simms, 1992; for examples in which no effects are observed see review in Roff, 1997). Although most of our arguments suggest reasons why inbreeding depression will be higher in the wild, inbreeding depression in captivity can be biased upwards as a result of poor husbandry or as an artifact of captive breeding. It has been argued that a reduction in fitness traits is to be expected in animals that have greatly dissimilar genetic backgrounds (because of the breakdown of coadapted gene complexes), which may be a common occurrence in captive populations (Smith, 1979); in such cases outbreeding depression may have been misdiagnosed as inbreeding depression (Templeton, 1987). The ongoing debate concerning whether the seriously reduced reproductive capacity of cheetahs in captivity is caused by genetic factors or incorrect captive conditions, is a good example of the difficulty of determining the cause of low fitness even for an individual species (Caro & Laurenson, 1994; Merola, 1994; O’Brien, 1994). However, poor husbandry techniques for captive species may increase inbreeding depression, which means that in a situation where an inbred population is maintained under ideal conditions, the inbreeding depression in survival will probably be lower than the Ralls et al. (1988) estimate of 0.31, and will be substantially lower than our calculated mean of 0.539.

There are a number of important implications of high levels of inbreeding depression in wild species. Populations that experience high levels of inbreeding and subsequent inbreeding depression may in future generations have significantly lower levels of inbreeding depression even if closely inbred, because of the purging of deleterious recessive alleles expressed during inbreeding (Wright, 1977; Lorenc, 1980; Bryant et al., 1990; Barrett & Charlesworth, 1991; Ribble & Millar, 1992; Hedrick, 1994). Nevertheless, although the expected effects of purging deleterious alleles have been documented to some extent (see Husband & Schemske, 1996 for a review), the degree of purging is questionable (see Frankham, 1995a for a discussion), and an accelerated rate of inbreeding in populations can potentially drive a population towards extinction (Gilpin & Soulé, 1986). Although the susceptibility of most populations of animals and plants to high levels of inbreeding and inbreeding depression is poorly known, our results show that inbred organisms in the wild do exhibit inbreeding depression and that the costs of inbreeding in the wild are substantially higher than previously thought (Ralls et al., 1988). The importance of inbreeding depression for wild populations depends not only on the magnitude of the effect when it occurs but also the likelihood of inbreeding. Although high levels of inbreeding have been observed in some populations of animals and plants (Thornhill, 1993; Husband & Schemske, 1996), much more data are needed to ascertain its frequency.