Discovering genetic changes important for adaptation along environmental gradients is a critical goal. These genes can teach us about the size, number, and types of genetic changes involved in historical adaptive shifts as well as the importance of gene function and genetic network structure in constraining evolutionary responses. From these parameters, we can gain predictive power for assessing how species may respond to global change. Both sets of authors chose candidate genes predicted to affect a clinally varying trait as a point of departure. McKechnie et al. focused on Dca because a previous study found a strong association between this gene and wing size variation in a single population (Rako et al. 2007). InR was chosen by Paaby et al. because, like high-latitude populations compared to low-latitude populations, mutants in InR and other insulin signalling pathway genes have longer life spans, increased stress tolerance, and decreased reproductive success compared to wild-type flies (Schmidt et al. 2005; Tatar et al. 2001). For both InR and Dca, reciprocal clines in allele frequency for common alleles were found in Australia. Paaby et al. also found strikingly similar trends for InR alleles in US populations.
While these correlations implicate InR and Dca in clinal adaptation, correlations can arise for a multitude of reasons and several questions must be addressed to develop the full evolutionary story. First, how finely can the latitude-associated gene region be delimited, allowing spurious associations caused by population structure or selection at linked loci to be excluded? Second, does the allelic variation directly cause changes in a clinally varying phenotype? Finally, is there evidence of selection on either the gene or its associated phenotype by a spatially varying environmental condition?
With respect to the first question, both studies benefit from D. melanogaster’s dispersal abilities. Unlike many species, population structure is largely absent even at continental scales, and the vast majority of loci show no association with latitude (Turner et al. 2008). Selection at linked loci is a more pressing concern. Both InR and Dca are tightly linked to a chromosomal inversion polymorphism and thus additional genes and markers whose allele frequencies also vary with latitude (Weeks et al. 2002). Each study demonstrates their variants are independent of this inversion; their genes still exhibit allele frequency clines when flies with or without the inversion are separately considered.
Paaby et al. cleverly took advantage of geographic replication and model system genetics to further delimit their latitude-associated gene region to a candidate nucleotide change and define its function. The authors were able to show through extensive sequencing and genotyping studies that only an indel variant in the first exon exhibits robust allele frequency clines on both continents. Linkage disequilibrium rapidly declines to either side of this variant, further supporting the inference that this is an independent site. Observation of replicate clines of the same variant on two continents is compelling evidence for natural selection on InR since other evolutionary processes like genetic drift or migration are highly unlikely to reproduce the same pattern.
To test the functional relevance of the InR variation, Paaby et al. isolated chromosomes containing the two clinally varying alleles in a common genetic background. After several generations of recombination pared away associations with other loci, flies carrying different InR alleles exhibited significant differences for several phenotypes in directions consistent with geographical patterns. Females with the allele at high frequency in high-latitude populations lived longer, recovered faster from cold stress, and laid fewer eggs than females carrying the allele at high frequency in low-latitude populations. If, as the authors propose, the indel contributes to all these phenotypic differences, then that would be a very interesting result because it would provide a concrete mechanistic basis for tradeoffs between correlated life history traits. However, as the authors note, the crossing scheme did not completely dissociate SNPs within InR and nearby genes from the indel, and these results cannot decisively address this matter. Moreover, different nucleotides within another pleiotropic Drosophila gene independently affect natural variation in different life-history traits (Carbone et al. 2006). Nevertheless, the authors have laid solid groundwork toward confirming their hypothesis that this single coding mutation has pleiotropic effects contributing to clinal adaptation.
McKechnie et al. followed an alternative experimental path toward defining a role for their gene, Dca, in clinal variation of their trait of interest, wing size, because no previous molecular work had demonstrated that Dca directly controls wing size. Like a previous study that found an association between variation in the Dca promoter and wing size (Rako et al. 2007), the authors used a complex crossing scheme that preserved a considerable amount of natural variation to obtain the same result in a second Australian population. Notably, this work helped disentangle Dca’s effects from those of the linked chromosomal inversion. While variation in Dca was not associated with thorax size, the inversion was associated with changes in both wing and thorax size, indicating that changes in Dca may specifically affect wing size and alter the wing : thorax size ratio. McKechnie et al. then took advantage of Drosophila’s genetic toolkit and overexpressed Dca in transgenic flies. Dca overexpression reduced wing size but not thorax size, confirming Dca specifically functions in wing size regulation.
Based on these results, the authors predicted that if cis-regulatory variation at Dca contributes to the cline in wing size, then the Dca allele associated with smaller wing size should have higher expression. Indeed, expression analysis showed that Dca expression increases as the frequency of the allele associated with smaller wing size increases. Thus, the combination of functional and expression studies performed by the authors yielded strong correlative evidence consistent with a causative role for these regulatory mutations in clinal variation. While McKechnie et al. provide no population genetic evidence for selection, wing size is known to affect dispersal performance in field release trials (Hoffman et al. 2007). Flies with large wings relative to their thorax disperse farther than flies with relatively small wings, and selection on dispersal caused by latitudinal differences in resource distribution or other factors could drive the observed clines in wing size.
Both papers make substantial progress in bridging individual genetic variants, clinally varying phenotypes, and natural selection, but further work is required to fully develop the evolutionary picture in either case. From a molecular perspective, additional transgenic or complementation studies are necessary to directly prove that the InR coding change or any of the several regulatory polymorphisms in Dca cause changes in life history traits or wing size. From an ecological viewpoint, the changing relationship between these phenotypes and fitness at different latitudes needs further experimental study to link specific selective pressures to organismal variation. The results from these current works will make such studies more feasible as clever genetic manipulation of Dca or InR could generate flies well suited for such tests.
While generalizations about the genetics of clinal adaptation will require identification of more genes, it is tempting to draw inferences based on the authors’ findings. For instance, clinal adaptation appears to involve a plurality of mutation types; these studies and others have implicated both coding and regulatory changes (Caicedo et al. 2004; Hoekstra et al. 2004; Collinge et al. 2008; Fry et al. 2008; Mullen & Hoekstra 2008; Schmidt et al. 2008).
Both sets of authors speculate that only certain genes will harbour variation capable of effectively responding to selection along environmental gradients, and Paaby et al. provide some data consistent with this hypothesis. Other genes in the insulin signalling pathway show similar mutant phenotypes as InR, including chico (Clancy et al. 2001); however, unlike InR, the authors found that chico lacks clinal variation and exhibits no evidence of long term selection. Only the chico coding region was surveyed, however, and given the short distance over which linkage disequilibrium breaks down, a potential role for cis-regulatory change cannot be excluded. While the different patterns of amino acid evolution in InR and chico over phylogenetic time may be attributable to different functional constraints, different clinal variation patterns could also be a consequence of differences in the sampling of ancestral genetic variation. The US and Australian clines have evolved over only the past few hundred years and the same indel mutation is associated with clinal adaptation on both continents, suggesting evolution occurred by selection on standing variation acquired from the source population. Thus, differences in ancestral variation at InR and chico or the sampling of this variation in founding populations could also yield divergent evolutionary outcomes for the two genes. Further population genetics studies of InR, chico, and other insulin pathway genes in African populations will help resolve these possibilities.
The sophisticated crossing scheme employed for the association study by McKechnie et al. allowed the authors to determine how much of the total variation in wing size in a population is controlled by promoter variation in Dca. They estimate that Dca controls a large amount (>20%) of the heritable variation in wing size, suggesting that mutations of large effect do play roles in clinal adaptation. From an empirical perspective, this is a cause for optimism. This finding indicates that clines are not solely governed by small changes in allele frequency in many genes with vanishingly small effects, and that additional genes involved in clinal adaptation are likely identifiable. Future studies that creatively integrate diverse experimental strategies like these two papers will undoubtedly be successful at doing so.