For some time, the reversibility of evolution was primarily discussed in terms of comparative patterns. Only recently has this problem been studied using experimental evolution over shorter evolutionary time frames. This has raised questions of definition, experimental procedure, and the hypotheses being tested. Experimental evolution has provided evidence for multiple population genetic mechanisms in reverse evolution, including pleiotropy and mutation accumulation. It has also pointed to genetic factors that might prevent reverse evolution, such as a lack of genetic variability, epistasis, and differential genotype-by-environment interactions. The main focus of this perspective is on laboratory studies and their relevance to the genetics of reverse evolution. We discuss reverse evolution experiments with Drosophila, bacterial, and viral populations. Field studies of the reverse evolution of melanism in the peppered moth are also reviewed.

Corresponding Editor: W. T. Starmer

It is possible to estimate the rate of diversification of clades from phylogenies with a temporal dimension. First, I present several methods for constructing confidence intervals for the speciation rate under the simple assumption of a pure birth process. I discuss the relationships among these methods in the hope of clarifying some fundamental theory in this area. Their performances are compared in a simulation study and one is recommended for use as a result. A variety of other questions that may, in fact, be the questions of primary interest (e.g., Has the rate of cladogenesis been declining?) are then recast as biological variants of the purely statistical question—Is the birth process model appropriate for my data? Seen in this way, a preexisting arsenal of statistical techniques is opened up for use in this area: in particular, techniques developed for the analysis of Poisson processes and the analysis of survival data. These two approaches start from different representations of the data—the branch lengths in the tree—and I explicitly relate the two. Aiming for a synoptic account of useful theory in this area, I briefly discuss some important results from the analysis of two distinct birth-death processes: the one introduced into this area by Hey (1992) is refitted with some powerful statistical tools.

Corresponding Editor: M. Whitlock

Natural hybrid zones between related species illustrate processes that contribute to genetic differentiation and species formation. A common viewpoint is that hybrids are essentially unfit, but they exist in a stable tension zone where selection against them is balanced by gene flow between the parent species. An alternative idea is that selection depends on the environment, for example, by favoring opposite traits in the two parental habitats or favoring hybrids within a bounded region. To determine whether selection of hybrids is environment dependent, we crossed plants of naturally hybridizing Ipomopsis aggregata and I. tenuituba in the Colorado Rocky Mountains and reciprocally planted the seed offspring into a suite of natural environments across the hybrid zone. All types of crosses produced similar numbers and weights of seeds. However, survival of the offspring after 5 years differed markedly among cross types. On average, the F₁ hybrids had survival and growth rates as high as the average for their parents. But hybrid survival depended strongly on the direction of a cross, that is, on which species served as the maternal parent. This fitness difference between reciprocal hybrids appeared only in the parental environments, suggesting cytonuclear gene interactions that are environment specific. These results indicate that complex genotype-by-environment interactions can contribute to the evolutionary outcome of hybridization.

Corresponding Editor: S. Tonsor

Genetic change is a necessary component of speciation, but the relationship between rates of speciation and molecular evolution remains unclear. We use recent phylogenetic data to demonstrate a positive relationship between species numbers and the rate of neutral molecular evolution in flowering plants (in both plastid and nuclear genes). Rates of protein and morphological evolution also correlate with the neutral substitution rate, but not with species numbers. Our findings reveal a link between the rate of neutral molecular change within populations and the evolution of species diversity.

Corresponding Editor: D. Baum

Barriers to gene flow between species result from selection against foreign linkage blocks in hybrids. When the geographic ranges of taxa meet at multiple locations, the opportunity exists for variation in the genetic architecture of isolating barriers. Hybrid zones between two sunflower species (Helianthus annuus and H. petiolaris) in Nebraska and California exhibited remarkably similar patterns of introgression of mapped molecular markers. Congruence among hybrid zones may result from limited intraspecific variation at loci contributing to isolation and from similar selective effects of alleles in the heterospecific genetic background. The observed consistency of introgression patterns across distantly separated hybrid zones suggests that intrinsic forces predominate in determining hybrid zone dynamics and boundaries between these sunflower species.

Corresponding Editor: O. Savolainen

We tested for adaptive differentiation between two natural populations of Impatiens capensis from sites known to differ in selection on plasticity to density. We also determined the degree to which plasticity to density within a site was correlated with plastic responses of experimental immigrants to foreign sites. Inbred lines, derived from natural populations in an open-canopy site and a woodland site, were planted reciprocally in both original sites at naturally occurring high densities and at low density. The density manipulation represents environmental variation typically experienced within the site of a given population, and the transplant manipulation represents environmental differences between sites of different populations. Internode elongation, meristem allocation, leaf length, flowering date, and total lifetime fitness were measured. Genotypes originating in the open site, where selection favored plasticity of first internode length and flowering time (Donohue et al. 2000a), were more plastic in those characters than genotypes originating from the woodland site, where plasticity was maladaptive. Therefore, these two populations appear to have responded to divergent selection on plasticity. Plasticity to density strongly resembled plasticity to site differences for many characters, suggesting that similar environmental factors elicit plasticity both to density and to overhead canopy. Thus, plasticity that evolved in response to density variation within a site influenced phenotypic expression in the foreign site. Plastic responses to site caused immigrants from foreign populations to resemble native genotypes more closely. In particular, immigrants from the open site converged toward the selectively favored early-flowering phenotype of native genotypes in the woodland site, thereby reducing potential fitness differences between foreign and native genotypes. However, because genotypes from the woods population were less plastic than genotypes from the sun population, phenotypic differences between populations were greatest in the open site at low density. Therefore, population differences in plasticity can cause genotypes from foreign populations to be more strongly selected against in some environments than in others. However, genetic constraints and limits to plasticity prevented complete convergence of immigrants to the native phenotype in any environment.

Corresponding Editor: S. Tonsor

The biogeographic pattern of chloroplast DNA (cpDNA) haplotypes in Eucalyptus on the island of Tasmania is consistent with reticulate evolution, involving at least 12 Tasmanian species from the subgenus Symphyomyrtus. Intraspecific cpDNA polymorphism in 14 of 17 species is coupled with extensive sharing of identical haplotypes across populations of different species in the same geographic area. Haplotype diversity is lowest in central regions of Tasmania formerly occupied by alpine vegetation during glacial intervals and in northern regions that were periodically linked to continental Australia by land bridges. The observed distribution of several cpDNA haplotypes unique to Tasmania coincides with modeled locations of glacial refugia in coastal areas of Tasmania and shows the power of cpDNA in unraveling the complex history of past distributions of Eucalyptus. The results suggest that the model of evolution of the eucalypts should be reassessed to allow for the anastomosing effects of interspecific hybridization and introgression.

Corresponding Editor: O. Savolainen

Fluctuating asymmetry (FA) has been used as a measure of developmental stability and may indicate individual phenotypic or genotypic quality. Using water boatmen (Callicorixa vulnerata) from a natural population, we examined the relationship between tarsal FA (tarsal spine number, tarsal length) and indices of body condition in two habitats. We used body weight and residual body weight (controlling for body length) as indices of condition because experimental food deprivation in water boatmen led to a reduction in each. We detected a negative relationship between FA and both indices of condition in two ecologically distinct pond habitats. We predicted this association was due to a negative relationship between FA and competitive feeding ability. Consequently, we examined associations between survival time and tarsal FA in C. vulnerata under resource-limited laboratory conditions. Univariate analyses revealed a negative correlation between survival and tarsal FA in each trait. Inclusion of survival time, body length, gender, tarsal spine number, tarsal length, and measures of FA into multivariate analyses revealed a negative correlation between survival and FA. Individuals with the greatest survival had higher nutritional condition than individuals that succumbed early in the experiment. Asymmetric individuals may suffer a foraging handicap as a result of the use of tarsi in feeding or they may be of poor genetic quality. Our results suggest elevated FA may limit resource acquisition and are consistent with the use of FA as a measure of fitness.

Corresponding Editor: D. Roff

Female mate preferences are a major cause of diversity and elaboration in male sexual traits. Here we characterize the shape of female preference functions for pulse length and carrier frequency of the courtship song of Drosophila montana by fitting both parametric and nonparametric functions to the incidence of female receptive gestures to synthetic song. Preference functions for both traits are strongly directional. That for pulse length is linear and favors short pulses, whereas that for carrier frequency is stabilizing in shape, but would exert directional preferences favoring males with high carrier frequency. The preference for carrier frequency has probably evolved under sexual selection, but reasons for the preference for short pulses are less apparent. We also examine the effect of ambient temperature on the carrier frequency of male song and on the preference function for carrier frequency. For many similar acoustic communication systems, temperature coupling, a compensatory effect of temperature on preference functions, is thought to maintain coordination between preferences and signals. However, although the carrier frequency of D. montana song is highly dependent on environmental temperature, there is no temperature coupling of the female preference function. We suggest that temperature coupling may often arise due to a common effect of temperature on song and preference, rather than be an advantageous characteristic whose function is to maintain coordination in temperature-affected communication systems.

Corresponding Editor: H. L. Gibbs

The sex-ratio trait, known in several species of Drosophila including D. simulans, results from meiotic drive of the X chromosome against the Y. Males that carry a sex-ratio X chromosome produce strongly female-biased progeny. In D. simulans, drive suppressors have evolved on the Y chromosome and on the autosomes. Both the frequency of sex-ratio X and the strength of the total drive suppression (Y-linked and autosomal) vary widely among geographic populations of this worldwide species. We have investigated the pattern of Y-linked drive suppression in six natural populations representative of this variability. Y-linked suppressors were found to be a regular component of the suppression, with large differences between populations in the mean level of suppression. These variations did not correspond to differences in frequency of discrete types of Y chromosomes, but to a more or less wide continuum of phenotypes, from nonsuppressor to partial or total suppressor. We concluded that a large diversity of Y-linked suppressor alleles exists in D. simulans and that some populations are highly polymorphic. Our results support the hypothesis that a Y-chromosome polymorphism can be easily maintained by a balance between meiotic drive and the cost of drive suppression.

Corresponding Editor: H. A. Orr

Highland populations of several Drosophila species in Argentina were active early in the afternoon in the field as opposed to populations from a much warmer lowland site, where flies were mainly active in the early evening prior to sunset. For one of these species, Drosophila buzzatii, we tested for a genetic component of activity differences by carrying out crosses within and between populations and measuring oviposition activity of the progeny in the laboratory. We found that activity in the highland population exceeded that in the lowland one during the midafternoon, whereas activity in the lowland population exceeded that in the highland one prior to the beginning of the dark period. Oviposition activity for the period corresponding to the field observations was regressed on the proportion of the genome derived from the highland population. This variable significantly predicted oviposition activity between 1400 and 1600 and between 2000 and 2200 h. Activity of both reciprocal crosses was intermediate and not significantly different from each other, suggesting that nuclear genetic, rather than cytoplasmic factors contribute to differences in oviposition activity between the populations. Two morphological, one genetic, and one stress resistance trait were also scored to examine whether temperature differences between environments were associated with other differences between populations. Wing length of wild-caught and laboratory-reared flies from the highland population significantly exceeded that in the lowland. Thorax length of laboratory-reared flies from the highland population also significantly exceeded that from the lowland. Chromosomal inversion frequencies differed significantly between the two populations with a fivefold reduction in the frequency of arrangement 2st in the highland as compared to the lowland population. This arrangement is known for its negative dose effect on size, and thus, the highland population has experienced a genetic change, perhaps as a result of adaptation to the colder environment, where body size and the frequency of arrangement 2st have changed in concert. Finally, a heat knockdown test revealed that the lowland population was significantly more resistant to high temperature than the highland one. In conclusion, we suggest that temperature has been an important selective agent causing adaptive differentiation between these two populations. We also suggest that the activity rhythms of the two populations have diverged as a consequence of behavioral evolution, that is, through avoidance of stressful temperatures as a mean of thermal adaptation.

Corresponding Editor: L. Stevens

In spite of the predicted genetic and ecological costs of sex, most natural populations maintain sexual reproduction, even those capable of facultative parthenogenesis. Unfertilized eggs from natural populations of Drosophila mercatorum occasionally develop into viable adults, but obligately parthenogenetic populations are unknown in this species. To evaluate the microevolutionary forces that both favor and constrain the evolution of parthenogenesis in D. mercatorum, we have measured parthenogenetic rates across a natural, sexually reproducing population and characterized the life-history changes that accompany the transition from sexual to parthenogenetic reproduction in laboratory strains. A highly significant difference in parthenogenetic rate was found between two populations in close geographic proximity, with increased rate found with lower population density. Laboratory strains of parthenogenetic females suffered increased mortality and reduced egg viability relative to their virgin counterparts from a sexual strain. Lifetime egg production was similar across all strains, but a shift in peak egg production to an earlier age also occurred. The combination of these life-history traits resulted in a higher net reproductive value for sexual females, but because they also had a longer generation time, intrinsic rate of increase was not as dramatically different from parthenogenetic females. In environments with high early mortality, there may be no fitness disadvantage to parthenogenesis, but the predicted ecological advantage of a twofold increase in intrinsic rate of increase was not realized. These results support the theory of Stalker (1956) that parthenogenesis is favored in environments in which sexual reproduction is difficult or impossible.

Corresponding Editor: C. López-Fanjul

Evolutionary biologists typically assume that the number of eggs fertilized or developing embryos produced is correlated with an individual's fitness. Using microsatellite markers, we document for the first time estimates of realized fitness quantified as the number of offspring surviving to adulthood in an insect under field conditions. In a territorial damselfly whose males defend tree hole oviposition sites, patterns of offspring survivorship could not be anticipated by adults. Fewer than half of the parents contributing eggs to a larval habitat realized any reproductive success from their investment. The best fitness correlate was the span over which eggs in a clutch hatched. Among parents, female fecundity and male fertilization success were poor predictors of realized fitness. Although body size was correlated with female clutch size and male mating success, larger parents did not realize greater fitness than smaller ones. The uncoupling of traditional fitness surrogates from realized fitness provides strong empirical evidence that selection at the larval stage constrains selection on mated adults.

Corresponding Editor: C. Boggs

Host-associated mating is crucial in maintaining the partial reproductive isolation between the host races of Eurosta solidaginis (Diptera: Tephritidae), a fly that forms galls on Solidago altissima and S. gigantea. (We refer to flies reared from S. gigantea as gigantea flies and those reared from S. altissima as altissima flies.) We measured the host preference of males and females of both host races, F₁ hybrids between the host races, F₂, and backcrosses to both host races. Male and female altissima flies and female gigantea flies had high host fidelity, whereas male gigantea flies had low host fidelity. This result suggests that there may be gene flow between the host races due to nonassortative mating that occurs when male gigantea mate with altissima females on S. altissima. This indicates assortative-mating mechanisms in addition to host-associated mating are required to produce the partial reproductive isolation between the host races that has been observed. Nongenetic factors had no influence on host preference. Larval conditioning did not influence host preference: reciprocal F₁ hybrids reared in S. altissima and S. gigantea both preferred S. gigantea. Adult experience had no impact on host preference: females preferred their natal host plant regardless of which host they encountered first as an adult. The hypothesis that maternal effects influence preferences was rejected because male and female flies did not show a consistent preference for the host plant of their mother. We also found no evidence that preference was a sex-linked trait because F₁ and backcrosses to the host races with different combinations of X chromosomes from the two host races preferred S. gigantea. Our results indicate that host preference is not determined by a large number of genes because preference of hybrids did not correspond to the proportion of the genome derived from each host race. The strength of the ovipuncture preference for S. gigantea by gigantea females, the females of both reciprocal F₁ hybrids, the backcross to gigantea, and F₂s indicates that preference is inherited nonadditively at a limited number of loci. The F₁ female hybrids, however, had a weaker host preference for S. gigantea than the pure gigantea host race, indicating that there may be incomplete dominance or modifier loci. Males had different host preference patterns than females, with individual male gigantea and male F₁ hybrids usually exhibiting preference exclusively for S. gigantea or S. altissima. One hypothesis explaining the difference in host preference between males and females is that the same gene influences both female and male host preference, but it is a sex-influenced gene. Thus, males carrying the gene for S. gigantea preference have an intermediate host preference, whereas females have a strong host preference to S. gigantea. In summary, we found that the host preference that produces host-associated mating is inherited nonadditively at a relatively small number of loci on autosomal genes. This mode of inheritance meets the assumptions of models of sympatric speciation, indicating that the host races could have evolved in sympatry.

Corresponding Editor: W. T. Starmer

Two general patterns that have emerged from the intense studies on insect–host plant associations are a predominance of specialists over generalists and a taxonomic conservatism in host-plant use. In most insect–host plant systems, explanations for these patterns must be based on biases in the processes of host colonizations, host shifts, and specialization, rather than cospeciation. In the present paper, we investigate changes in host range in the nymphalid butterfly tribe Nymphalini, using parsimony optimizations of host-plant data on the butterfly phylogeny. In addition, we performed larval establishment tests to search for larval capacity to feed and survive on plants that have been lost from the female egg-laying repertoire. Optimizations suggested an ancestral association with Urticaceae, and most of the tested species showed a capacity to feed on Urtica dioica regardless of actual host-plant use. In addition, there was a bias among the successful establishments on nonhosts toward plants that are used as hosts by other species in the Nymphalini. An increased likelihood of colonizing ancestral or related plants could also provide an alternative explanation for the observed pattern that some plant families appear to have been colonized independently several times in the tribe. We also show that there is no directionality in host range evolution toward increased specialization, that is, specialization is not a dead end. Instead, changes in host range show a very dynamic pattern.

Corresponding Editor: C. Boggs

Physiological causes of genetic differences in cannibalism were examined to gain a better understanding of constraints on behavior evolution. Cannibalism has complex population level consequences in Tribolium confusum, including dramatic effects on population size. Laboratory strains with low and high cannibalism rates, obtained through inbreeding, have maintained distinct levels of cannibalism for over two decades even in the absence of artificial selection to maintain the differences. Why strains differ in their cannibalism rates was examined by measuring: (1) the nutritional benefit from cannibalism in both nutritionally good and poor environments, and (2) the possibility that eggs are an important source of water. How strains achieve differences in cannibalism was examined by testing for differences between strains in their ability to find eggs and in their tendency to eat eggs.

Beetles from both strains survive equally well in a nutritionally good environment, but they accomplish this in different ways. The low cannibalism strain has high survivorship with and without cannibalism. The high cannibalism strain has low survivorship when not fed eggs and survivorship equivalent to the low cannibalism strain when fed eggs, suggesting it compensates for poor nutritional adaptation by eating eggs. The strains also differ in feeding behavior; beetles from the high cannibalism strain have a higher appetite for eggs. Beetles from the two strains did not differ in locomotor activity, search efficiency, or need for water. The observed behavioral and nutritional differences may contribute to the maintenance of different levels of cannibalism.

Corresponding Editor: D. Roff

Many examples of cryptic marine species have been demonstrated with biochemical and molecular studies. In most cases, a broadly distributed taxon is actually a group of sibling species that can be distinguished (upon closer examination) by ecological or morphological characters. Fishes of the family Albulidae constitute a notable exception. Bonefish (Albula spp.) morphology and ecology are highly conserved around the globe, and their extended pelagic larval stage could allow population connections on a vast geographic scale. Based on this perceived homogeneity, bonefishes were classified as a single pantropical species, A. vulpes. However, allozyme studies of Hawaiian populations indicated that two sympatric species (A. glossodonta and A. neoguinaica) are included in the synonymy of A. vulpes. To ascertain the number and distribution of evolutionary partitions in Albula, we surveyed 564 bp of mitochondrial DNA (mtDNA) cytochrome b from 174 individuals collected at 26 locations. Sequence comparisons reveal eight deep lineages (d = 5.56–30.6%) and significant population structure within three of the four lineages that could be tested (ϕ_ST = 0.047–0.678). These findings confirm the genetic distinctiveness of the three species noted above and invoke the possibility of five additional species. Clock estimates for mtDNA indicate that these putative species arose 4–20 million years ago. Distinct evolutionary lineages coexist in several sample locations, yet show little morphological or ecological differentiation in sympatry. Thus, bonefish species seem to defy the evolutionary conventions of morphological differentiation over time and ecological displacement in sympatry. Despite multiple cases of sympatry, sister-taxa relationships inferred from mtDNA indicate that divergence in allopatry has been the predominant speciation mechanism in Albula. Stabilizing selection in the homogeneous habitat occupied by bonefishes (tropical sand flats) could promote the retention of highly conserved morphology and ecology.

Corresponding Editor: Louis Bernatchez

Models for the evolution of plasticity predict that individuals having phenotypes induced by exposure to enemies should experience relatively low fitness when enemies are absent. However, costs of induced phenotypes have been difficult to find in both plants and animals, perhaps because costs are expressed at later stages in the life cycle. We searched for delayed costs of an induced defense in larvae of the water frog Rana ridibunda, which exhibits strong phenotypic responses to predators. Tadpoles grew to metamorphosis in outdoor artificial ponds, in either the presence or absence of Aeshna dragonfly larvae confined within cages. We collected metamorphs at forelimb emergence, estimated their development rate until tail resorption was complete, and measured their body and leg shape and hopping performance. Development rate through metamorphosis reflects the duration of a transitional period during which metamorphs are especially vulnerable to predators, and hopping performance may reflect ability to escape predators. Froglets from the dragonfly treatment lost mass through metamorphosis significantly faster than those from predator-free ponds, but they resorbed their tails at about the same rate, despite the fact that their tails were relatively large to begin with. Froglets developing from predator-induced tadpoles had shorter, more muscular legs, and hopped 5% longer distances (difference not significant). Therefore, producing an induced defense against insect predators during the tadpole stage did not exact a cost during or immediately after metamorphosis; if anything, tadpoles with the predator-induced phenotype gave rise to more vigorous froglets. These results focus attention on other costs of the induced phenotype, as well as alternative explanations for plasticity that do not rely on direct fitness trade-offs.

Corresponding Editor: T. Garland

While it is widely recognized that the manner in which organisms adjust their timing of reproduction reflects evolutionary strategies aimed at minimizing offspring mortality or maximizing reproductive output, the conditions under which the evolutionarily stable strategy involves synchronous or asynchronous reproduction is a matter of considerable discord. A recent theoretical model predicts that whether a population displays reproductive synchrony or asynchrony will depend on the relative scales of intrinsic regulation and environmental disturbance experienced by reproducing individuals. This model predicts that, under conditions of negligible competition and large-scale environmental perturbation, evolution of a single mixed strategy will result in asynchronous reproduction. We tested this prediction using empirical data on large-scale climatic fluctuation and the annual timing of reproduction by three species of flowering plants covering 1300-population-years and four degrees of latitude in Norway. In agreement with model predictions, within populations of all three species reproductive asynchrony increased with the magnitude of large-scale climatic perturbation, but bore no relation to the strength of local density dependence. These results suggest that mixed evolutionarily stable strategies can arise from the interplay of combinations of agents of selection and the scale at which they operate; hence it is fruitless to associate synchronous versus asynchronous timing with particular single factors like climate, competition, or predation.

Corresponding Editor: M. Morgan

Several recent studies suggest that interactions with conspecific males can reduce the longevity of female Drosophila melanogaster or support the idea that male and female fitness components are involved in antagonistic interactions. Here we report that males from third-chromosome isogenic lines demonstrated significant genetic variation in male reproductive performance and in the longevity of their mates. Increased male performance was marginally significantly associated with one measure of increased female survival rate. However, there was no indication of trade-offs or negative correlations between male reproductive success and female survival. We discuss alternative hypotheses for the cause of the induced variation in female longevity.

Corresponding Editor: W. T. Starmer

To assess whether, while overwintering, natural populations of Drosophila pseudoobscura are likely to experience substantial bottlenecks in their numbers and genotypes, laboratory tests of the cold sensitivities of each stage of the life history and reproduction were undertaken. Three genetically distinctive lineages established from flies caught at high elevation were used for testing in temperatures likely to persist in protected pockets of fermenting deciduous leaf fall in overwintering sites. Sensitivities to cold of each stage in development were measured as frequencies of survival to adulthood following a period in 5°C in a particular stage. The cold sensitivity of adults was measured as the survival in and following cold stays in adulthood. It was found that cold sensitivity decreases as development progresses, but that only adults (females more than males) are able to withstand long periods in the cold. The cold sensitivity of reproductive capacity of males was scored as their success in mating following a two-month cold stay, and of females as the numbers laying fertile eggs following periods of months in the cold. Both males and females maintain reproductive capacity. Of particular significance, however, is that even after six months in the cold females are able to restart production of eggs and these eggs may be fertilized by the sperm of matings prior to their cold stay. Thus, a substantial proportion of overwintering genomes must be those of adult females and those of the sperm carried by females from matings in the previous summer. This simple finding strongly suggests that populations are not likely to suffer substantial bottlenecks while overwintering. Further, it indicates how arrays of genetic variation may be maintained through winters and largely avoid winter selective pressures. Frequent migration between populations is therefore not required to maintain the variation commonly found in populations throughout the species range.

Corresponding Editor: H. Allen Orr