Fundamental Evolutionary Limits in Ecological Traits Drive Drosophila Species Distributions

See allHide authors and affiliations

Science  04 Sep 2009:
Vol. 325, Issue 5945, pp. 1244-1246
DOI: 10.1126/science.1175443


Species that are habitat specialists make up much of biodiversity, but the evolutionary factors that limit their distributions have rarely been considered. We show that in Drosophila, narrow and wide ranges of desiccation and cold resistance are closely associated with the distributions of specialist and generalist species, respectively. Furthermore, our data show that narrowly distributed tropical species consistently have low means and low genetic variation for these traits as compared with those of widely distributed species after phylogenetic correction. These results are unrelated to levels of neutral variation. Thus, specialist species may simply lack genetic variation in key traits, limiting their ability to adapt to conditions beyond their current range. We predict that such species are likely to be constrained in their evolutionary responses to future climate changes.

Specialist species within particular climatic zones and/or dependent on particular hosts make up much of biodiversity (1), but generally investigations have not considered whether evolutionary factors may limit the distributions of these species (2, 3). Models predicting climate change suggest that it is crucial to determine how species will evolve and contract or expand from their current distributions, particularly in the case of threatened climate specialists. Distributional patterns may be explained by the fact that species simply lack the appropriate genetic variation in key traits necessary to adapt to conditions beyond their distributions and thus are constrained by a fundamental genetic limit (4). Such a lack of genetic variation contradicts the theory that traits are highly variable (5), which originated from quantitative genetic theory that shows mutational variance to be large enough to maintain genetic variation within traits (6).

In particular, the idea that genetic variance might limit evolutionary responses was supported by the fact that certain plant species exhibited phenotypic variation for resistance to soil contamination and could successfully colonize mine tailings contaminated with heavy metals (7). Species or populations unable to evolve higher levels of metal resistance were limited by low levels of genetic variation for this trait. Few studies had gone on to further test this prediction until it was suggested that a lack of appropriate variation for desiccation resistance might be linked to the distribution of two rain forest Drosophila species (3, 8). These desiccation-sensitive species did not respond to persistent selection for increased desiccation resistance, whereas cross-generation comparisons suggested that this was due to very low genetic variation for this trait (8, 9) in contrast to other widespread Drosophila species (10).

Environmental variables are tightly associated with the distribution of Drosophila and insects more generally (11). It is well documented that species distributions become narrower toward the tropics (Rapaport’s rule) (12). One hypothesis to explain this pattern and which has empirical support is the climatic variability hypothesis (12), which proposes that species with greater physiological tolerance to climatic variables will be able to extend their distribution to higher latitudes. This leaves species that are unable to increase their tolerance restricted to the tropics.

We extend the climatic variability hypothesis and suggest that levels of genetic variation in key ecological traits are driving differences in physiological tolerances between tropical and widespread Drosophila species. To examine this, we initially compared patterns of mean resistance to two stresses, desiccation and cold, for 30 species (13). Both resistance traits have been closely linked to differences in species distributions [cold resistance in (1416) and desiccation resistance in (15, 17)]. We found that species restricted to the tropics showed low levels of resistance to both desiccation and cold in comparison with more widespread temperate species (Fig. 1). Consequently, there was a strong association between levels of resistance and tropical versus widespread distribution, with the exception of Drosophila funebris, which is adapted to cold environments (Fig. 1). These data are in accord with the climatic variability hypothesis and with studies that also show that levels of cold resistance in species match expectations on the basis of climatic data (18, 19). Furthermore, these results confirm that cold and desiccation are important variables influencing distributional patterns in Drosophila species. Non-drosophilid tropical insects also have a much narrower range of thermal optima (12).

Fig. 1

Standardized cold resistance plotted against standardized desiccation resistance for Drosophila species. ♦ indicates species with a widespread distribution; ◊ indicates species with a restricted/tropical distribution. an, ananassae; ari, arizonensis; bip, bipectinata; bi, birchii; bus, busckii; bun, bunnanda; buz, buzzatii; er, erecta; eni, enigma; fun, funebris (cold adapted); hyd, hydei; im, immigrans; in, inornata; k, kikkawai; lat, lativittata; leb, lebanonensis; mal, malerkotliana; m, mauritiana; mel, melanogaster; nit, nitidithorax; p, paulistorum; ps, pseudoananassae; rep, repleta; sec, sechellia; ser, serrata; sim, simulans; sub, subobscura; t, teisseri; y, yakuba; and sul, sulfurigaster. The outlier noted in the text is circled.

To establish whether low genetic variation in resistance traits influences tropical and widespread distributions, we estimated genetic variation (mostly as hn2, the narrow sense heritability) in five tropically restricted and five widespread Drosophila species. For most traits, pedigree information was analyzed with statistical animal models (13, 20) to generate small SEs around heritability estimates. Patterns of genetic variation in the traits of cold and desiccation resistance were consistent with our hypothesis; low levels of genetic variation were found in all tropical species, as indicated by low heritability estimates (Fig. 2). Those cases in which we observed only small and nonsignificant estimates of hn2 were unlikely to reflect a lack of power because we obtained low SEs for most comparisons. In contrast, we observed significant levels of variation in all widespread species, and the difference in hn2 between tropical and widespread species overall was greater than twofold for both desiccation and cold resistance (table S1). Patterns of additive genetic variance (VA) also matched that of hn2, highlighting that specialization to tropical environments may reflect low genetic variation in key ecological traits. Fundamental limits in genetic variation therefore appear to be driving distribution limits in these Drosophila species.

Fig. 2

Species comparison with trait mean and genetic variance estimates. –, D. melanogaster; ●, D. simulans; ▲, D. repleta; ■, D. hydei; ♦, D. serrata; ○, D. birchii; Δ, D. bunnanda; □, D. pseudoananassae; ×, D. sulfurigaster; and ◊, D. bipectinata. (A) hn2 versus desiccation resistance. (B) VA versus desiccation resistance. (C) hn2 versus cold resistance. (D) VA versus cold resistance. Error bars reflect one SE.

The most commonly invoked explanation for such observed low levels of genetic variation in traits is that they are subject to directional selection that causes the fixation of favored alleles (21). However, if this is true then low genetic variation should be coupled with high (directionally selected) trait means if high resistance levels are favored. Instead, if genetic variation is directly involved in limiting species distributions, we would expect low genetic variation to be coupled with low trait means, whereas species with high levels of genetic variation should not be constrained in this way (22). Under these predictions, the mean expression of a distribution-limiting trait and levels of genetic variation for that trait should be correlated, as was observed in this study (Fig. 2). If experimental error was swamping estimates of hn2, particularly in the restricted species, an association between environmental variance and trait mean would be expected; this was not the case for either desiccation (r2 = 0.05, P = 0. 51) or cold (r2 = 0.26, P = 0.14) resistance.

To ensure that phylogenetic history was not responsible for the observed patterns in trait mean, hn2, and species distribution, we generated phylogenetic independent contrasts (PICs) (13, 23). PICs generate statistical independence within a data set by standardizing trait values by the branch length in two sister species. Each comparison results in an independent contrast, and consequently the number of contrasts equals n – 1. We used a highly resolved phylogenetic tree (fig. S1), created on the basis of two mitochondrial (COII and ND5) and two nuclear (ADH and hb) genes, to phylogenetically correct our estimates of hn2, VA, and trait mean. The strength of the relationships decreased slightly after correction for phylogeny but were still significant (Table 1). When PICs for cold and desiccation resistance were plotted, a strong positive and linear association remained. Thus, trait mean was closely associated with levels of genetic variation (Fig. 2 and table S1), emphasizing that levels of genetic variation for cold and desiccation resistance, rather than phylogeny, are driving species distributions.

Table 1

Regression analyses for trait means onto genetic variance estimates for desiccation and cold resistance. Regressions were computed on both the original unstandardized data set and the PICs.

View this table:

Population processes such as inbreeding or drift could alter levels of genetic variation and explain the observed patterns. However, if this was true we would expect low levels of genetic variation in all traits. To test this, we estimated hn2 for wing size (13). In contrast to the stress resistance traits, we saw no consistent association between hn2 or VA and trait mean between restricted and widespread species. All estimates of both hn2 and VA differed significantly from zero in all species (table S1). We observed no relationship between mean wing size and hn2 (r2 = 0.001, P = 0.92). We also tested for a relationship between wing size and mean resistance because body size could influence trait resistance (24); however, we did not detect a relationship between any of the traits (for desiccation, r2 = 0.29, P = 0.10; for cold, r2 = 0.10, P = 0.36). Furthermore, in some species microsatellite repeat variation has been characterized. For these species, it was high and comparable for species that are widespread and those that are climatically specialized (25).

It is unclear what mechanisms produce these patterns. Hypotheses involving asymmetrical gene flow that might account for evolutionary stasis do not predict that low levels of genetic variation will limit evolutionary responses (22). The rate of mutation seems unlikely to limit variation in our tropical species because we would expect equivalent levels of variation to arise through mutation if desiccation resistance is influenced by similar genes across species. Our results may be caused by a loss of genetic variance through DNA decay (22); if selection on these traits is weak or essentially neutral, theory suggests that mutations will accumulate that could potentially result in loss of function of genes, and hence traits (26). The hypothesis that the loss of underlying genetic variability through DNA decay leads to ecological specialization has been proposed to occur in some bacterial systems (27). The tropical species may be evolutionarily derived, or else the more widespread species are derived after activation of resistance, perhaps after gene duplication or the development of novel resistance pathways. Tradeoffs might also be involved if they produce strong directional selection for sensitivity to cold and desiccation in tropically restricted species.

Our analysis supports the hypothesis that specialization to tropical environments in Drosophila species reflects fundamental genetic limits in ecologically important traits. We show that genetic variation in tropical Drosophila species for both desiccation and cold resistance is consistently low, which is in contrast to widely distributed species. Whether other tropically restricted species share a similar evolutionary fate remains to be determined. However, evidence suggests that an absence of genetic variation may be limiting distributions in other systems (2). Further work examining the genetic architecture of traits that may be linked to species distributions will be important for the assessment of the future evolutionary potential of species. If species distributions are commonly driven by genetic limits, the consequences of climate change on biodiversity may be greater than previously assumed. It is worth considering heat resistance as well as cold and desiccation resistance within this context, particularly because recent research has highlighted the pending threat of increases in temperature on the persistence of tropical species (12, 28, 29).

Supporting Online Material

Materials and Methods

Fig. S1

Tables S1 and S2


References and Notes

  1. Materials and methods are available as supporting material in Science Online.
  2. We thank the Australian Research Council and the Commonwealth Environmental Research Fund for financial support. We also thank M. Blacket for molecular work; A. Miller for assistance with the phylogenetic analysis; K. Mitchell, C. Doig, T. Kristensen, J. Overgaard, L. Carrington, K. Lein, and X. Du for technical support; and reviewers for constructive comments.
View Abstract

Stay Connected to Science

Navigate This Article