Patterns of genetic variation in rare and widespread plant congeners^†

There appears to be no recognizable correlation, either positive or negative, between the amount of genetic variation within populations of plant species and the rarity or commonness of the species as a whole.

G. L. Stebbins (1980, p. 80)

Biologists have long been fascinated with rare and narrowly endemic species. In recent years, this fascination has turned to urgency, as ever more species dwindle toward extinction, and we attempt to prevent further loss of biotic diversity. As more rare species are studied, our ignorance of them has been repeatedly highlighted. Perhaps the most crucial of our misconceptions is that, despite several arguments to the contrary (e.g., Stebbins, 1942; Drury, 1974; Rabinowitz, 1981), rare species are considered a relatively homogeneous assemblage of species with common characteristics that will become apparent given sufficient study.

One such perceived characteristic of rare species is the maintenance of low levels of genetic variation. Variously viewed as either a cause or a consequence of rarity, limited genetic diversity has been reported for many rare plant species (reviewed by Hamrick and Godt, 1989). Theories regarding the genetic properties of rare species date back at least to the 1930s when Anderson (1936, p. 496) concluded that it was an “innate invariability” that limited the range of Iris setosa var. canadensis. Almost without exception, the consensus has been that rare species have low levels of genetic variability. In contrast to his more recent view (Stebbins, 1980), Stebbins (1942) previously considered rare species to be genetically depauperate.

Stebbins' reversal seems to have been largely due to two examples of highly variable rare species (Sequoiadendron giganteum and Eriogonum apricum) and several examples of widespread species with little variation. Since Stebbins' (1980) paper and the first large compilations of isozyme data (e.g., Hamrick, Linhart, and Mitton, 1979), additional examples of highly polymorphic rare plant species have accumulated (e.g., Adenophorus periens [Ranker, 1994]; Daviesia suaveolens [Young and Brown, 1996]), suggesting that the view that rare species have less variability than more widespread ones may be an overgeneralization.

Theory predicts that low genetic variation in rare species may be tied to the effects of small population sizes (cf. Barrett and Kohn, 1991; Ellstrand and Elam, 1993). The level of expected heterozygosity (H_e) present at neutral loci in a population can be predicted by

where N_e is the effective population size and μ is the mutation rate. This equation takes into account the effects of genetic drift in small populations with lower effective population sizes. When the above equation is employed by population biologists to predict reduced genetic variation in rare species, the tacit assumption is that rare species have small population sizes, an assumption that may not hold for many rare species (e.g., Daviesia suaveolens [Young and Brown, 1996]). Ellstrand and Elam (1993) argued that rare species with large localized populations should be expected to exhibit high levels of genetic variation.

Several review papers have examined the correlations between genetic diversity and geographic range, life-history characteristics, and other aspects of plant species (e.g., Hamrick and Godt, 1989). All of these studies have identified highly significant effects of geographic range on genetic diversity at both the species and population level (Hamrick and Godt, 1989). One potential limitation of these studies is that the included taxa are treated as independent samples, ignoring their phylogenetic relatedness. Such analyses violate the assumptions of the statistical methods used to analyze the data (Felsenstein, 1985; Harvey and Pagel, 1991; Silvertown and Dodd, 1996). To examine the relationship between geographic range and genetic diversity thoroughly, phylogenetic history must be accounted for using phylogenetically independent contrasts (Felsenstein, 1985; Harvey and Pagel, 1991; Silvertown and Dodd, 1996). While it is not reasonable, at this time, to conduct a fully phylogenetic analysis of genetic diversity, Felsenstein (1985) and Silvertown and Dodd (1996) both suggest that, in the absence of a phylogeny, congeneric comparisons can be made. We can be confident that the species within a genus share a more recent common ancestor than other species in the analysis, and thus differences between congeneric species pairs are independent for each genus.

Upon reviewing the characteristics of endemic taxa, Kruckeberg and Rabinowitz (1985, p. 475) concluded that future research should include more “comparative studies to contrast the biologies of rare taxa with those of related common ones.” Following this advice, Karron (1987) compared 11 pairs of endemic and widespread congeners for both genetic variability and outcrossing rate. Outcrossing rates have generally been predicted to be low in rare species that are likely to occur in small, isolated patches. Karron's choice to compare congeners was based on the argument that studies that group together unrelated taxa with dissimilar life histories make generalizations difficult.

Karron's (1987) results for genetic diversity indicated that rare species do have significantly lower levels of genetic variation than their widespread congeners. In contrast, rare species did not exhibit significantly lower outcrossing rates than their widespread congeners (Karron, 1987). Thus, as far as the correlation between geographic range and genetic diversity, accounting for phylogenetic history did not affect the conclusions. However, at that time, genetic data for only 11 pairs of congeneric species were available.

Since Karron's (1987) paper, many more researchers have included widespread congeners when examining the genetic variation of rare species, and additional examples of highly polymorphic rare species have been reported (e.g., Ranker, 1994; Lewis and Crawford, 1995). Furthermore, in some genera both rare and widespread species have either very low or very high, but similar, levels of polymorphism (e.g., Young and Brown, 1996).

Our purpose in this paper is to summarize the data currently available on the levels of genetic variation in rare and widespread congeneric species. Using data from the literature, for up to 34 pairs of rare and widespread congeners, we have summarized levels and patterns of genetic variability at both the population and species levels.

MATERIALS AND METHODS

A literature search was conducted to identify as many cases as possible in which genetic data were reported for at least one rare and one widespread species within the same genus. Karron (1987) limited his search to species pairs studied by the same laboratory, arguing that variation in the enzyme loci examined and in the scoring of gels could confound the data. We chose to limit our study further to include only species for which data for both a rare species and a widespread congener are reported in the same paper. This removes the additional confounding influence of studies conducted for different purposes, or methodological shifts in a laboratory. Additionally, when multiple widespread species were reported in a paper, Karron (1987) used the values reported for the most widespread species. Rather than choosing one species in the absence of phylogenetic information, we used an average of the values reported for all rare species or all widespread species where more than one was reported in a paper. Although choosing the most closely related congener for comparison with a rare species may be the best approach to remove the effects of history from the comparisons, this was not usually possible as phylogenetic analyses had typically not been conducted for these genera. We believe that the mean values are the most informative representation of all available data for the species.

In all cases, we used the authors' descriptions of the relative ranges of the species. No attempt was made to reinterpret designations used by the original authors on the basis of number of individuals or range of the species. There is no widely accepted means by which a species may be termed rare, so efforts to classify rarity based on predetermined, and universally applied, limits of population size or geographic range will be arbitrary. Additionally, our use, and indeed the use in the literature, of the term “rare” includes both species that are historically rare and those that are rare due to recent population declines. Clearly there is a potential for these two types of species to differ markedly in their genetic properties. Indeed, historical population levels are only one of the many traits by which rare species can differ (see, for example, Rabinowitz, 1981; Fiedler and Ahouse, 1992).

Population-level data (calculated as the mean of the values obtained in each population examined) for percentage polymorphic loci (%P_pop), mean number of alleles per locus (A_pop), and observed heterozygosity (H_o) were recorded for each species. Species-level data (calculated from all individuals examined in the species regardless of population of origin) were gathered for percentage polymorphic loci (%P_spp), mean number of alleles per locus (A_spp), and total genetic diversity (H_T). In addition, data on population substructuring (F_ST or G_ST) were recorded (Table 1). As estimates of population substructuring, F_ST and G_ST are very similar (Hartl and Clark, 1989); they have been combined in the analyses to increase the sample size for the comparisons.

In some cases, values were calculated from raw data presented in the literature. In other cases, insufficient data were reported to allow calculation of all of the diversity parameters used in our comparisons, and no value is reported here. After Karron (1987), we plotted each measure of diversity in the rare species against that of the widespread congener. To assess the significance of differences in diversity between rare and widespread congeners, the data were analyzed with Wilcoxon signed rank tests, with rare and widespread species within a genus representing the paired samples. This nonparametric test was used, because while the congeneric comparisons are independent, it is not possible to assume that they are identically distributed, so nonparametric analyses that do not assume identical distributions must be used (Felsenstein, 1985). We also examined the correlation between level of diversity in a rare species and its widespread congener. As the issue here is not whether geographic range and diversity are correlated—where phylogeny must be accounted for—but whether diversity measures are correlated within a genus, traditional correlation statistics and significance tests (Z test using Fisher's R to Z transformation) are appropriate. All statistical tests were conducted with StatView 5.0 (SAS Institute, Cary, North Carolina, USA).

RESULTS

For both population- and species-level values, a close relationship is observed between the levels of genetic diversity in rare and widespread congeners (Figs. 1, 2). The majority of the points on the graphs appear close to the lines of equality, indicating similar levels of diversity in rare and widespread congeners. These visual interpretations are supported by the correlation analyses that show for all measures, at both the population and species levels, levels of genetic diversity are highly and significantly correlated between rare species and their widespread congeners (Table 2).

The percentage of polymorphic loci at the population level (%P_pop) is significantly lower in the rare species than in the widespread species examined here (Table 2). However, there are nine species pairs (26% of the species pairs studied) for which the rare species has a higher %P_pop than its widespread congener, and several additional rare species maintain levels of diversity that are only marginally lower than those of the widespread congeners. The range of values for %P_pop in rare species is from 0 to 86%, which is nearly equivalent to that for the widespread species (range from 0 to 84%). Additionally, values for percent polymorphic loci at the population level are highly significantly correlated within a genus (Table 2).

The mean number of alleles per locus at the population level (A_pop) is significantly lower in rare species than in widespread congeners (Table 2). However, in six species pairs (24% of the species pairs studied), the rare species has either the same number of, or more, alleles per locus than the widespread congener. The range of values in rare and widespread species is identical (1–2.8 in both rare and widespread species). Values for the mean number of alleles per locus in rare and widespread species are highly correlated within a genus (Table 2).

The rare species also have a significantly lower level of observed heterozygosity than the widespread species (Table 2). However, in seven species pairs (29% of the species pairs studied), the rare species has a higher level of observed heterozygosity than the widespread congener. The ranges of heterozygosities in rare and widespread species are very similar (0–0.27 and 0–0.28, respectively). Levels of observed heterozygosity are highly significantly correlated within a genus (Table 2).

For all three population-level measures of diversity examined, the rare species had significantly lower levels of genetic diversity than the widespread congeners. However, for all measures of diversity, 24–29% of the rare species examined were as variable as or more variable than their widespread congeners, with many additional species being almost as variable as their widespread congeners. When reasonable sampling errors are considered, close to half of the rare species examined here have as much or more genetic diversity as their widespread congeners. Additionally, the ranges in variability exhibited by rare species and their widespread congeners do not differ in any substantial way. Finally, for all measures of population-level diversity, the diversity in rare species is significantly and positively correlated with the diversity in the widespread congeners examined.

At the species level, rare species have significantly fewer polymorphic loci (%P_spp) than their widespread congeners (Table 2). As with population-level measures, five of the species examined (23% of the total) have as many or more polymorphic loci as their widespread congeners. The ranges for rare and widespread species are the same (0–93.8% for both). The values are correlated between rare and widespread species within a genus (Table 2).

Measures for mean number of alleles per locus are significantly lower in rare species than in their widespread congeners (Table 2). There are three rare species with as many alleles per locus as their widespread congeners (20% of the total), and the values are highly correlated (Table 2). The ranges are also identical (1–4.3 for both rare and widespread species).

Unlike the results for the other measures of diversity, estimates of total genetic diversity (H_T) do not differ significantly between rare species and their widespread congeners (Table 2). Five of the rare species have as much or more diversity as their widespread congeners (28% of the total number of species pairs). Diversity in rare and widespread species is highly correlated (Table 2). H_T is the one measure of diversity where the ranges of values differ markedly, with widespread species having higher values of diversity (0.033–0.327 for rare species and 0.031–0.45 for widespread species). It is possible that we simply do not have enough samples to detect significance in the trend.

Finally, rare species do not appear to partition their genetic variation within and among populations differently than do their widespread congeners. When F_ST and G_ST values are combined, mean values for rare and widespread congeners do not differ significantly from each other (Table 2). Eleven of the rare species have greater levels of substructuring than do their widespread congeners (50% of the total). Despite the lack of significant differences in degree of substructuring, there is a strong correlation in values of F_ST and G_ST in rare and widespread congeners.

DISCUSSION

Several large reviews of genetic data for plant species have provided valuable insight into the patterns of genetic variation found in plant species (e.g., Hamrick, Linhart, and Mitton, 1979; Loveless and Hamrick, 1984; Hamrick and Godt, 1989). These reviews have classified species by several criteria, including geographic range and various life-history characters, to search for trends of genetic diversity in plants with similar characteristics. Significant differences in the levels of genetic variability were detected among range categories (Hamrick, Linhart, and Mitton, 1979; Loveless and Hamrick, 1984; Hamrick and Godt, 1989). Many researchers have used data from these reviews as a point of reference for data from their own studies on rare species. Karron (1987) argued that such comparisons between a given species and a mean for all rare species neglect the evolutionary history of the species under consideration, weakening the explanatory power of the analyses. A more meaningful comparison, Karron (1987) argued, is that between a rare species and a more widely distributed congener. This form of analysis is consistent with the recommendations of Felsenstein (1985) and others who have demonstrated the importance of phylogenetically independent contrasts in looking for correlations among taxa that share evolutionary histories. Since Karron's paper, several authors have included widespread congeners when studying levels and patterns of genetic diversity in rare species, and it is apparent that several rare species are as polymorphic as or more polymorphic than their widespread congeners (e.g., Vogelmann and Gastony, 1987; Ranker, 1994; Young and Brown, 1996).

Few attempts have been made to summarize these new data. Karron (1997) used the same data set of 11 species pairs compiled for his 1987 paper. Frankham (1995) compared the levels of genetic variation in 38 endangered species to those of “related nonendangered species” (p. 310), and found that for 32 of these, the endangered species had less genetic variation. Frankham's analysis included only six plant species (none of which were those previously used by Karron [1987]), and no indication is given as to which related taxa were being used for comparison.

The data we have compiled show significant differences in the level of diversity between rare and widespread congeners based on three population-level measures (percentage polymorphic loci, mean number of alleles per locus, and observed heterozygosity) and two species-level measures (percentage polymorphic loci and mean number of alleles per locus) of genetic diversity. While these results are consistent with previous studies (e.g., Hamrick and Godt, 1989; Karron, 1987), our analyses highlight not only that there exists a high degree of correlation within a genus for all measures of diversity, but that levels of diversity for rare species encompass almost the same range as is found in widespread congeners.

Rare and widespread congeners do not differ significantly in total genetic diversity (H_T). This result is not consistent with the results of previous studies (Hamrick and Godt, 1989). One possible explanation is our smaller sample size, and thus the inability to assign significance to the trend indicating lower levels of diversity in rare species. Alternatively, this result could be “real” and perhaps a more accurate interpretation of the data, because we controlled for the nonindependence of taxa.

The genetic diversity in rare species does not appear to be partitioned differently within and among populations than that in their widespread congeners. This result is consistent with that of Hamrick and Godt (1989); however, it is still important to note that there is a moderate, yet significant, correlation between values in rare species and their widespread congeners.

The results summarized above indicate that generalizing that rare species have low levels of genetic variability is only part of the truth, echoing Hamrick's (1983) call for treating each rare species as a unique and novel entity. Our study shows that while there may be a slight reduction in genetic variation in rare species relative to their widespread congeners, it is not the case that rare species are confined to low levels of diversity. That the ranges for rare and widespread species encompass similar values indicates that there is no more an average value of diversity for a rare species than there is for a widespread one. Thus, even putting aside the issue of statistical nonindependence (Felsenstein, 1985), it makes little sense to compare a value of genetic diversity for one species to that of a mean from Hamrick and Godt's paper (1989) or this paper: a more informative comparison is to a value from a widespread congener. The high correlations between measures of genetic diversity within a genus argue that comparison to a mean is of little value, as what is important is not the mean of all rare species, but the level of diversity in other species within the genus. Finding values above or below that found in a widespread congener offers a useful point of reference from which to begin interpreting the biology of a rare species.

The use of allozyme variation has a long tradition in population genetics, and a more recent application in conservation biology. Some have questioned the appropriateness of assessing the genetic diversity of a rare species using allozyme or other genetic markers for conservation purposes, arguing that the correlation between molecular genetic variation and the phenotypic, adaptive variation that is important to the species' survival may be weak (e.g., Hamrick, Murawski, and Loveless, 1991; Karhu et al., 1996; Storfer, 1996). Several studies have compared morphological or physiological variation in rare and widespread congeners. The rare bunch grass Achnatherum (Oryzopsis) hendersonii has less phenotypic variation than its widespread congeners A. (Stipa) lemmonii and A. (Stipa) thurberiana (Rapson and Maze, 1994), consistent with traditional views of low levels of variation in rare species. In contrast, there are no differences in the photosynthetic performance of the rare Echinacea tennesseensis and its widespread congener E. angustifolia (Baskauf and Eickmeier, 1994).

Given these data suggesting that there may be little difference in physiological responses in some rare species and that about one-quarter of rare species maintain higher levels of genetic diversity than their widespread congeners, why have conservation biologists often generalized that rare species are genetically depauperate? Although some rare plant species have little or no detectable variation (e.g., Ceratophyllum echinatum [Les, 1991]; Lespedeza leptostachya [Cole and Biesboer, 1992]; Bensoniella oregona [Soltis et al., 1992]), other rare species maintain high levels of genetic diversity (e.g., Adenophorus periens [Ranker, 1994]; Agastache occidentalis [Vogelmann and Gastony, 1987]; Daviesia suaveolens [Young and Brown, 1996]; Layia discoidea [Gottlieb, Warwick, and Ford, 1985]). We suggest that the data for rare species may be overly generalized because rare species themselves are generalized. Despite repeated urgings to the contrary (e.g., Stebbins, 1942; Drury, 1974; Rabinowitz, 1981), rarity largely continues to be viewed as a single entity. Species classified as rare exhibit a diverse array of local population sizes, habitat specificities, and even geographic ranges (Rabinowitz, 1981). Thus, while biologists generalize that rare species have small effective population sizes and should have reduced genetic variabilities, this may not always be the case. Additionally, a rare species capable of surviving in several diverse habitats may maintain higher levels of genetic diversity than even a widespread species confined to one uniform habitat. Therefore, rare species with different attributes may maintain different levels and patterns of genetic diversity. Unfortunately, insufficient data are available to allow comparisons of the genetic properties of species that represent different forms of rarity.

The data summarized here clearly demonstrate that the view that rare species lack genetic variation (whether this is a cause or consequence of rarity) is an overgeneralization: while rare species do have statistically less genetic variation than their widespread congeners, there is a large range in values, and levels of diversity are highly correlated within a genus. While some rare species exhibit reduced genetic variation, others maintain levels of diversity equal to or exceeding that of widespread congeners. We hope that these results will convince conservation biologists that genetic data for a rare species are more informative when compared with data for a widespread congener or close relative. The mere fact that a rare species lacks genetic variation offers limited information for management when we do not know how this level compares to that of a widespread relative. While managers should strive to conserve as much diversity as possible, their management strategies will likely change if it is known that, for example, both a rare species and a widespread congener have similar, but low, levels of diversity. As levels of diversity may not be directly related to survival, in this example where a widespread species also exhibits low diversity, the emphasis might switch from attempting to increase diversity to focusing on other factors leading to population declines. By going a step further and examining widespread congeners, researchers can both address immediate conservation needs more effectively and begin to provide the data that will be necessary to understand the biologies of rare species.

The real asset of studying genetic variation in rare and widespread congeners is that it allows us to identify and begin investigating historical and life-history causes of rarity. Moving beyond overgeneralization, researchers can focus on other causes and consequences of rarity and explain why some species are rare and why some species exhibit reduced genetic variation. This study is only a first step toward controlling for phylogenetic history in an analysis of genetic diversity in rare species. Felsenstein (1985) indicated that comparisons of congeners, such as this study, discard much of the data and statistical power that could be obtained by using fully resolved phylogenies. As large-scale phylogenetic studies become available, levels of genetic diversity in plants should be reevaluated using phylogenetically independent contrasts. These types of studies will provide a better understanding of the phenomenon of rarity and allow conservation biologists to meet the increasingly important challenge of preserving the Earth's biotic diversity.