Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2011 Aug 1.
Published in final edited form as: Evolution. 2010 Aug 19;64(10):2921–2934. doi: 10.1111/j.1558-5646.2010.01039.x

THREE SELECTIONS ARE BETTER THAN ONE: CLINAL VARIATION OF THERMAL QTL FROM INDEPENDENT SELECTION EXPERIMENTS IN DROSOPHILA

David M Rand 1,2, Daniel M Weinreich 1, Daniel Lerman 1,3, Donna Folk 4,5, George W Gilchrist 4,6,7
PMCID: PMC3148135  NIHMSID: NIHMS304268  PMID: 20497214

Abstract

We report the results of two independent selection experiments that have exposed distinct populations of Drosophila melanogaster to different forms of thermal selection. A recombinant population derived from Arvin California and Zimbabwe isofemale lines was exposed to laboratory natural selection at two temperatures (TAZ: 18°C and 28°C). Microsatellite mapping identified quantitative trait loci (QTL) on the X-chromosome between the replicate “Hot” and “Cold” populations. In a separate experiment, disruptive selection was imposed on an outbred California population for the “knockdown” temperature (TKD) in a thermal column. Microsatellite mapping of the “High” and “Low” populations also uncovered primarily X-linked QTL. Notably, a marker in the shaggy locus at band 3A was significantly differentiated in both experiments. Finer scale mapping of the 3A region has narrowed the QTL to the shaggy gene region, which contains several candidate genes that function in circadian rhythms. The same allele that was increased in frequency in the High TKD populations is significantly clinal in North America and is more common at the warm end of the cline (Florida vs. Maine; however, the cline was not apparent in Australia). Together, these studies show that independent selection experiments can uncover the same target of selection and that evolution in the laboratory can recapitulate putatively adaptive clinal variation in nature.

Keywords: Artificial selection, circadian rhythm, experimental evolution, photoperiod, thermotolerance

Some of the most compelling examples of natural selection are patterns of parallel responses to environmental stress in independent populations. Repeatable shifts in allele or phenotype frequency resulting from a common selective pressure are difficult to explain under most neutral scenarios. Classic examples are the latitudinal clines at the Adh locus in Drosophila melanogaster in North America and Australia where the Adh-F allele is common at higher (colder) latitudes (Oakeshott et al. 1982). More recent studies have confirmed that these parallel latitudinal clines are quite common in Drosophila, extending to different species, phenotypic traits, and loci (Huey et al. 2000; Calboli et al. 2003; Balanya et al. 2006; Sawyer et al. 2006; Turner et al. 2008). Nevertheless, counter examples do exist indicating that clinal selection is not the same everywhere (Weeks et al. 2006).

Because temperature is a key environmental factor that covaries with latitude, much attention has been focused on thermal selection in Drosophila. In ectothermic organisms such as small insects, body temperature varies with air temperature, resulting in daily and seasonal changes in metabolic rate and correlated processes such as enzyme activities (Lanciani et al. 1990; Berrigan and Partridge 1997; Hochachka and Somero 2002). Natural selection has shaped the thermal biology of drosophilids, creating a range of phenotypes from desert-adapted species such as Drosophila mojavensis that can withstand temperatures greater than 40°C (Stratman and Markow 1998) to geographic variation in temperature dependent male sterility (David et al. 2005) to cold adapted species such as Drosophila subobscura that are capable of flight at air temperatures well under 10°C (Gilchrist and Huey 2004). In Australian D. melanogaster heat tolerance decreases, and cold tolerance increases with latitude along the east coast cline (Hoffmann et al. 2002). In North America, however, heat and cold tolerance both increase with latitude and are likely associated with variation in diapause and stress resistance (Schmidt et al. 2005a,b; Schmidt and Paaby 2008). Despite evidence for parallel North–South differentiation in many nucleotide polymorphisms in both North America and Australia (Turner et al. 2008), it is not yet clear if parallel clines are controlled by selection on the same phenotypes or the same set of interacting nucleotide polymorphisms. Studies seeking to dissect the genetic bases of these clinally varying traits would be generally informative about the repeatability of natural selection in the wild (Paaby et al. 2010).

One approach to this problem is to map quantitative trait loci (QTL) for thermal traits. Several studies have identified QTL in D. melanogaster that are correlated with population differences in thermotolerance. Loeschcke and colleagues have identified thermal tolerance QTL segregating between two geographically distinct populations of D. melanogaster that had been selected for high versus low resistance to thermal knockdown (Norry et al. 2004; Vermeulen et al. 2008a,b). Morgan and colleagues used two different inbred laboratory stocks that happened to differ in thermotolerance to identify QTL for both high and low temperature resistance (Morgan and Mackay 2006). Comparing these two studies, there is some potential overlap of QTL for high-temperature resistance spanning the Hsp70 region on chromosome 3R, but the QTL windows are sufficiently broad that it is difficult to say that the same genes are implicated in these two different mapping populations (Norry et al. 2004; Morgan and Mackay 2006). Verifying QTL in independent populations is a critical way forward in establishing meaningful associations between genotype and phenotype.

Given the relative ease of selection for thermal traits in the laboratory (Cavicchi et al. 1985; McColl et al. 1996; Gilchrist and Huey 1999), and the ample evidence for clinal variation in thermal traits in nature (Umina et al. 2005; Hoffmann and Weeks 2007; Schmidt and Paaby 2008), the Drosophila model provides an opportunity to test the hypothesis that thermal QTL discovered from selection on thermotolerance are associated with clinal variation. The repeatability of evolution remains an open question in current research (Simoes et al. 2008), and comparisons between laboratory and natural selection provide one means of seeking general properties that may have contributed to adaptive variation. In this article, we present results from linkage mapping of artificially selected lines from two rather different thermal selection experiments and examine the clinal variation of the QTL identified. The striking correspondence of QTL from these two experiments, and the direction of clinal variation of the QTL in nature, strongly suggests a common target for natural and artificial selection. The genomic region identified further implicates a connection between the perception of photoperiod and resistance to thermal stress.

Materials and Methods

LABORATORY NATURAL SELECTION

The “TAZ” recombinant populations were established from isofemale lines of D. melanogaster collected in Arvin, CA (North America; line Arv 3) and Zimbabwe (southern Africa, line Zim 6). Both lines were provided by Chip Aquadro and had been maintained in vial culture for several years rendering them inbred to an unknown degree. Virgin males and females (~50 of each) from each line were separated, reciprocally crossed between Arvin and Zimbabwe, and egg collections were made. Equal numbers of eggs from each direction of the cross were seeded into four replicate populations (AZ1–AZ4); each population was reared for ~25 discrete 18-day generations at 25°C. This initial phase of laboratory evolution allowed for chromosomes to recombine and segregate. This recombination phase will reduce linkage disequilibrium and should allow loci harboring thermally sensitive allelic variation to respond to selection more independently than if the temperature selection been imposed soon after hybridization.

Two of these populations (AZ1 and AZ3) were each divided to form four distinct replicate lines for laboratory natural selection. Two lines each from the AZ1 and AZ3 populations were reared at 18°C for ~25 generations (hereafter referred to as AZ1C1, AZ1C2, AZ3C1, AZ3C2; the “C” referring to “cold” and the final number to replicate line within that temperature). The other two lines from the AZ1 and AZ3 populations were reared at 28°C for ~45 generations (hereafter referred to as AZ1H1, AZ1H2, AZ3H1, AZ3H2; the “H” referring to “hot” and the final number to the replicate line). The difference in the number of generations was due to faster generation time at 28°C.

These culture conditions imposed “laboratory natural selection” due to differential growth, survival, and mating of individual genotypes in replicate experimental populations at two temperature treatments. All populations were maintained in population cages provisioned with four 180 mL food bottles containing 30 mL of cornmeal/yeast extract/dextrose/agar food, seeded with a pinch of dry yeast. The Cold (18°C) lines were maintained on a 24-day cycle, with 20 days from egg to adult and four days on new food bottles to continue mating and laying eggs. From day 17 to 20, the caps of the food bottles were removed allowing flies to escape into the cage and mate freely among culture bottles within a cage. At day 20, the old bottles were removed, fresh bottles were added to the cage, and egg laying continued for four days. The Hot (28°C) were maintained on a 12-day cycle with nine days from egg to adult and two days to continue mating a laying eggs. At day nine, caps were removed from the bottles allowing flies to escape into the cage and mate freely. At day 10, four fresh bottles were placed in each cage, and egg laying continued until day 12. These schedules were maintained to allow for high, and approximately equal, population densities at the two temperatures but no effort was made to maintain populations at specific sizes. For both the Hot and Cold cages, at the end of egg laying, all adults were tapped out of each culture bottle, the bottles were labeled and the post-reproductive adults were frozen for later DNA analysis. The first several generations were frozen, and then samples were frozen every five generations thereafter.

KNOCKDOWN ARTIFICIAL SELECTION

The flies used in this experiment were sampled from a large, out-bred population (~1000 isofemale lines) of D. melanogaster that was collected by Larry Harshman and Michael Turelli at Escalon, California in 1991. Selection was performed in a Weber column with a water jacket connected to a circulating heater (Thermo Scientific, Haake DC10), as described previously (Gilchrist and Huey 1999). The column is stabilized at 30°C and then approximately 1000 flies from one line are poured into the column from the top. Water temperature is then ramped up to 50°C and pumped through the water jacket, gradually raising the temperature within the column at a rate of about 0.4°C per minute. As the temperature rises, the flies begin to fall out of the column, where they are collected at 0.5°C intervals (TKD) for sorting by sex prior to selection.

This study reports linkage disequilibrium mapping of replicate High and Low TKD lines. For the High TKD lines, 25% of the population from the highest temperatures was retained such that there were approximately equal numbers of males and females. The Low TKD lines were established by retaining the 25% that fell out at temperatures from ~36 to 37°C. The ~250 selected individuals were divided between two 180-mL bottles (30-mL food) and held one week to ensure re-mating. The next day, eggs were collected and placed in vials in groups of 50 for rearing to adulthood for the next generation. All of the selection lines were maintained at 25°C, 12L:12D on cornmeal/molasses/agar medium at uniform densities (50 eggs per vial) totaling ~1000 flies per line prior to selection.

NATURAL POPULATIONS

To compare our laboratory-selected flies with clinal variation in nature, we obtained D. melanogaster from North American and Australian populations. Eight populations in the eastern United States (four samples of 50 lines each from Florida, two samples of 50 lines each from Pennsylvania and New Jersey, and two samples of 50 lines each from Maine) were provided by Paul Schmitt as isofemale cultures and are described in Schmidt et al. (2005a) and Schmidt and Conde (2006). Sample sizes of microsatellite alleles for Florida, Pennsylvania/New Jersey, and Maine were 84, 92, 74 alleles, respectively. Twelve populations from eastern Australia were provided by Ary Hoffmann as variable numbers of isofemale lines per population and are described in Umina et al. (2006). An average of 40 alleles was sampled for estimating allele frequencies (range 10–82; two populations were excluded with low sample sizes: “I” and “T” from Umina et al.). Both collections had been in the laboratory for about four years at the time of scoring.

MICROSATELLITE ASSAYS

A set of 26 microsatellites spanning the three major chromosomes was used for linkage mapping (Schug et al. 1998). Four were monomorphic in all of the populations assayed leaving the 22 shown in Table S1. For the TAZ selection, alleles were scored simply as “Arvin” or “Zimbabwe” as these parental lines carried diagnostic alleles. Eleven of the 22 loci were polymorphic in the TAZ populations. For the TKD selection lines, alleles were scored by size; 20 of the 22 loci were polymorphic. Flies were prepared using a Tris/NaCl/Proteinase K buffer following the procedures described in Rand et al. (1994). Microsatellites were amplified in 25-mL reactions using 2 mL of fly homogenate, 2.5 mL of Promega Taq buffer, 2.0 mL of 10 mM dNTPs, 2.5 mM MgCl2, 0.3 mL of each primer (from a 10 pM/mL stock), and 0.1 mL of Promega Taq polymerase. Amplification was for 35 cycles of 95°C for 30 sec, 54°C for 1 min, 72°C for 1 min, followed by 10 min at 72°C. For the initial screen of TAZ and TKD populations, amplified microsatellites were run on an ABI 377 automated sequencer according to the manufacturer’s specifications. Fragment sizing was done using ABI GenScan software. All clinal analyses were done on an ABI 3730 at the Hospital for Sick Children in Toronto, Canada. Slight differences in allele sizes were perceivable between these platforms (~1–2 bp across runs), but common alleles allow us to place different datasets in register. At the DROSEV2 locus, the common alleles were binned into size classes 140 and 152, based on the 3 bp repeat unit and comparison of allele frequency data from the laboratory and cline samples centered on different lengths (e.g., 139 and 151 vs. 141 and 153).

Samples of flies were scored from each replicate population from the TAZ lines at generation 0, and 18 for the Cold populations and generation 0 and 35 for the Hot populations. For some X-linked microsatellite markers, samples were scored at generations 6, 12, 20, and 25 to confirm frequency trajectories between the starting and ending values. Allele frequencies were estimated from samples of 18 females per population, providing a maximum sample size of 36 alleles for each microsatellite. Some individual PCR reaction failed resulting a range of sample sizes from 36 to 12 with an average of 27 alleles across loci and populations (low values were for loci other than band 3 markers that became the candidate region). For the TKD lines, samples of five females from each of three replicate populations, in each of the two temperature treatments (High and Low) were scored for 20 microsatellite markers after 40 generations of selection. The lines had been maintained as unselected cultures for more than a year at the time of initial microsatellite sampling. In 2004, the TKD lines were reselected and subsequently re-sampled for microsatellite variation, and no significant differences were observed between the initial sample and the 2004 re-sample in the candidate QTL region.

Our initial surveys identified cytological band 3 on the X-chromosome as a candidate QTL region. Other studies in D. melanogaster have implicated population differentiation or thermal selection at loci near band 3 (Sawyer et al. 1997, 2006; Harr et al. 2002) so we scored the selection lines for single nucleotide polymorphisms (SNPs) in these loci. DNA was amplified with specific primers, and digested with a diagnostic restriction enzyme (available upon request). Frequencies of “cut” versus “not cut” were scored in each of the Low and High TKD populations. For samples from natural populations, allele frequencies were estimated using available samples from North America and Australia (see above), in the band 3 region. Wider sampling was not done in these samples, but a more extensive study of clinal variation across the D. melanogaster genome has been reported (Turner et al. 2008).

STATISTICAL ANALYSIS

A significant association between a marker and selection regime was inferred using several different tests. First, simple G-tests of allele counts in the Hot versus Cold TAZ lines, and in the High versus Low TKD lines, were performed with correction for multiple tests across the 11 polymorphic TAZ markers or the 20 polymorphic TKD markers. Second, a two-level analysis of variance (ANOVA) was performed with microsatellite marker being a random effect, and temperature being a fixed effect. The response variable was arcsin-transformed allele frequencies in each population at the latest sample date. A significant interaction effect of a marker × temperature indicates a locus-specific response to the selection regime. Third, different general linear models were tested in the TAZ and TKD populations that account for differences in allele counts in these different experiments. Because only two alleles per locus were scored for the TAZ lines, we used a binomial GLM with a logit link function, testing selection treatment as a fixed variable and line within treatment as a random effect. For the TKD lines, we used a binomial general linear model (GLM) with a logit link function.

For the fine-scale mapping around the focal QTL, G-tests or Fisher’s exact tests were used, comparing the replicate High versus Low TKD populations. Because there are three Low (L1–L3) and six High (E1–E6) populations, G-tests were done across replicate Low and replicate High populations to test for heterogeneity, and were also done for the allele counts pooled across High versus pooled across Low populations. For some markers, certain alleles show no significant allele frequency difference between the Low and High populations, whereas other alleles showed very significant differentiation. When these patterns were clearly replicated across populations (as revealed by G-tests), this was taken as evidence for selection acting on that marker or some factor linked to it. Fixation indices were also used to quantify the net effect of thermal selection. If the fixation index comparing all High to all Low populations (FHL) was greater than any individual FST across replicate High or replicate Low populations, this was taken as evidence for an effect of thermal selection on that locus.

Results

LINKAGE DISEQUILIBRIUM MAPPING OF SELECTED POPULATIONS

Two independent thermal selection experiments were conducted using populations of D. melanogaster. Laboratory natural selection was carried out at 18°C and 28°C on replicate populations of TAZ flies established from advanced generation samples of a reciprocal cross between two inbred isofemale lines collected in Arvin, CA (North America; line Arv 3) and Zimbabwe (southern Africa, line Zim 6). Disruptive selection for High versus Low resistance to thermal knockdown (TKD) was carried out on an out-bred stock of flies from a wild California sample. Microsatellite variation was scored in replicate “Hot” and “Cold” populations of the TAZ experiment and the replicated “High” and “Low” TKD populations. A time course of microsatellite allele frequencies for the TAZ populations is shown in Figure 1, and end-point analyses are shown in Figure S1. A clear separation of Hot and Cold populations is evident at the most distal marker on the X (SGG), but not at other loci. From ANOVA of population allele frequencies, using locus, temperature and date as main effects, there is a highly significant effect of locus and a locus × temperature interaction (see Table S2). Other markers on the autosomes show no clear pattern of replicated differentiation between the Hot and Cold populations (Fig. 1). In the end-point analyses (Fig. S1), a marker on chromosome 2L and two markers on 3R showed significant differentiation in pooled allele frequencies, but the degree of allele frequency shift was not as dramatic and consistent as that shown at the SGG markers, and is due in part to heterogeneity across replicates (compare Figs. 1 and S1). Notably, the genotypes across individual flies at SGG and an adjacent marker, DMZW, showed very strong linkage disequilibrium (data not shown), as might be expected for closely linked markers in experimental populations.

Figure 1.

Figure 1

Open in a new tab

Real-time QTL analysis of thermal selection in experimental populations. Data are for the TAZ populations only. Top: Map of the X-chromosome. Middle: Map of the second chromosome. Bottom: Map of the third chromosome. Each small panel shows the frequency of the Arvin, CA allele on the y-axis, plotted as a function of time in generations on the x-axis, for an individual marker locus. Approximate chromosomal locations are identified on the map above the panels, with cytological band location shown above the chromosomes. Locations for each marker are listed in Table S1. Numbers on each line represent replicate cage number, as designated in the legend (top panel, see Materials and Methods). A highly significant difference between replicate Hot and Cold cages is evident in band 3 (SGG microsatellites), but not at the other markers on the X, or autosomes (see Table S2 and Fig. S1).

Microsatellite variation in three replicate High and three replicate Low populations of the TKD lines showed that markers on the X-chromosome had the clearest response to selection (defined as replicate populations showing parallel frequency shifts—see Fig. 2). The clearest differences between temperature treatments lie at the DMSGG3 and DROSEV2 markers. The DMZW marker is significantly differentiated in the TAZ population, but not in the initial screen of the TKD populations. Conversely, the DROSEV2 marker is significantly differentiated in the TKD populations, but not in the TAZ populations. A marker on chromosome 3R (DROABDB) is significantly differentiated in both the TAZ and TKD populations, but not to the degree of differentiation seen in at DMSGG3. In Figures S1 and 2, the allele frequencies at the latest time point are shown, with significance indicated by asterisks (* = P < 0.05; ** = P < 0.01; ** = P < 0.001; using a binomial GLM model for the TAZ populations, and a Poisson GLM model for the TKD populations, corrected for multiple tests; see Materials and Methods for details). These analyses establish that, in both base populations, a gene (or genes) in linkage disequilibrium with the DMSGG3 microsatellite marker harbors alleles whose fitness(es) depend on temperature. We acknowledge that the rather small sample of markers may have missed other thermal QTL, but even with a very dense screen, common loci in independent populations would warrant further study.

Figure 2.

Figure 2

Open in a new tab

Map of allele frequency differences for the TKD populations. Data shown are the frequency of the common allele in each replicate population (shown as adjacent bars) for each marker locus labeled under each set of bars. The red bars directed upward are for the High populations; blue bars directed downward are for the Low populations. Significance of the difference is identified with asterisks (* = P < 0.05; ** = P < 0.01; ** = P < 0.001; using a Poisson GLM model, corrected for multiple tests. (See Materials and Methods for details).

FINER SCALE MAPPING IN THE SHAGGY-PER REGION

The initial scans shown in Figure 2 were done on the TKD population in 2001 approximately two years after the original study (Gilchrist and Huey 1999). In 2004, these lines had been maintained without selection for several years and three generations of knockdown selection was performed on the three replicate Low (L1–L3) and six replicate High populations (E1–E6) that were extant. We performed a finer scale SNP mapping analysis around the SGG marker region on these re-selected lines.

Table 1 and Figures 3 and S2 show the results of finer scale mapping in the shaggy gene region. Notably, the two flanking markers at chromosome band 2F4 and 4C10 show no consistent differentiation between the replicate High and Low TKD lines (for 2F4, pooled G = 1.1, n.s; for 4C10, pooled G = 19.44 ≪0.0001, but replicate Low populations show higher levels of differentiation, so no net differentiation is supported; see Table 1). These define the maximum breakpoints within which the true QTL lies.

Table 1.

Allele frequencies and genetic differentiation among High and Low knockdown populations.

Population L1 L2 L3 E1 E2 E3 E4 E5 E6 GHvsL FstL FstH FHL FHL/Fst
2F4 microsat (n) 20 20 20 20 18 20 20 18 18
136 allele 0.60 0.85 0.65 0.90 0.94 0.20 0.85 0.78 1.00 0.056 0.417 0.225 0.539
G 2.01 1.20 0.00 0.00 14.44 2.20 1.10
P 0.16 0.27 1.00 1.00 0.00 0.10 0.10
3A7 microsat (n) 20 20 20 20 16 20 20 18 16
188 allele 0.05 0.00 0.05 0.35 0.88 0.95 0.65 0.22 0.13 0.017 0.402 0.553 1.375
G 0.00 5.80 5.80 8.04 0.31 0.52 34.66
P 1.00 0.04 0.01 4.5E-03 0.90 0.66 1.6E-08
SGG microsat (n) 40 38 40 40 40 40 40 38 38
118 allele 0.00 0.00 0.00 0.55 0.85 0.75 0.10 1.00 0.79 – 0.380 0.694 1.829
G 0.00 0.00 0.00 7.20 15.00 0.00 206.5
P 1.00 1.00 1.00 6.6E-03 5.3E-05 3.0E-57
DMZW microsat (n) 40 40 40 40 40 40 40 38 40
104 allele 0.00 0.00 0.00 0.55 0.00 0.60 0.50 0.97 0.00 – 0.483 0.628 1.299
G 0.00 0.00 0.00 27.65 0.45 70.24 40.33
P 1.0 1.0 1.0 8.4E-09 5.0E-01 1.5E-21 1.5E-10
Per repeats (n) 20 20 20 20 20 20 20 18 20
24 allele 0.00 0.00 0.00 0.00 0.00 0.70 0.10 0.33 0.00 – 0.432 0.491 1.137
G 0.00 0.00 0.00 0.00 12.60 5.32 11.10
P 1.00 1.00 1.00 1.00 2.4E-04 1.0E-02 8.6E-04
Syx4 3B4 (n) 20 20 20 20 20 20 19 20 19
cut allele 1.00 1.00 0.95 0.90 0.90 0.30 0.89 0.00 0.26 0.034 0.545 0.644 1.182
G 0.00 0.00 0.00 0.00 11.89 3.50 34.19
P 1.00 1.00 1.00 1.00 5.6E-04 2.0E-02 5.3E-10
Pon 4C10 (n) 20 20 20 20 20 20 19 20 20
cut allele 1.00 1.00 0.00 1.00 1.00 0.85 0.74 1.00 1.00 1.000 0.165 0.833 0.833
G 0.00 36.00 36.00 0.00 0.72 0.00 19.44
P 1.00 1.5E-11 1.5E-11 1.00 4.5E-01 1.00 8.9E-06
Open in a new tab

L1–L3 are three replicate Low TKD lines selected for knockdown at ~37°C.

E1–E6 are six replicate High TKD lines selected for knockdown at ~42°C.

GHvsL is the G-test statistic for allele counts pooled across High versus across Low populations.

FstL is Fst measured among the three replicate Low lines.

FstH is Fst measured among the six replicate High lines.

FHL is Fst measured between pooled High and pooled Low lines.

FHL/Fst is the FHL standardized by the larger of the two Fst values from the high or low populations.

Figure 3.

Figure 3

Open in a new tab

Differentiation of the SGG marker in thermal selection and latitudinal clines. (A) The SGG 118 allele increased in frequency in the TKD High populations. (B) The SGG marker shows the strongest net differentiation between High and Low selection regimes across the shaggy region. The y-axis of panel B is the ratio of total FST between High versus Low selection to the FST value among replicate populations within one selection regime (see Table 1 for statistical analyses). (C) The SGG-118 allele shows significant clinal variation with higher frequency in Florida (latitude 25) than in Maine (latitude 44; Florida-Maine: G = 8.55, P < 0.01; Florida-Central: G = 4.80, P < 0.05; Central-Maine: G = 0.82, n.s.). The slope of the regression of SGG 118 allele frequency on latitude is significantly different from zero in North America (R2 = 0.65, P < 0.02). (D) The Australian cline does not replicate the North American pattern and shows no significant cline.

The five other markers in Figure 3 (3A7, SGG, DMZW, per repeats, and Syx4) show varying levels of allele frequency differentiation between High and Low populations. As in the initial screen, the SGG microsatellite shows highly significant differentiation, with the 118 allele more common in the High populations and the 109 allele more common in the Low populations (pooled G = 206, P < 10−57, although replicate High lines do show significant differences in allele frequency; see Table 1). Table 1 presents population differentiation values (using traditional FST values) at different levels of variation: FstL describes differentiation among replicate Low populations, and FstH describes differentiation among replicate High populations. FHL describes the net differentiation among High and Low population, using pooled allele counts from within each replicate population. Because there is differentiation among replicate populations for some markers, the net differentiation of a marker is quantified as the ratio: FHL/Fst, where Fst is the larger of FstL or FstH. This value (FHL/Fst) is reported as differentiation in Figure 3B.

The markers on either side of SGG show statistically significant differentiation based on pooled G-tests of individual allele frequencies, but the repeatability across replicate populations is less clear. Marker 3A7 to the left of SGG has a significant increase in the 188 allele in the High populations, but there are individual High populations that are not significantly different from Low populations (L1, L2, L3 are not significantly different from the E5 and E6; see Fig. S2 and Table 1). The DMZW marker lies just to the right of SGG and maps within the transcription unit of the shaggy gene. The 92 and 102 alleles at DMZW show no significant differentiation, but the 104 allele shows a significantly higher frequency in the High populations based on pooled data. However, as with 3A7, there is significant allele frequency variation among the replicate High populations, suggesting that selection at DMZW was not consistent (see Table 1, Fig. S2). The difference between this result and the initial screen (Fig. 2) lies in the greater number of High populations scored. To the right of SGG and DMZW, the Thr-Gly repeats in the period locus show significant differentiation of the 24 allele, but only when data are pooled across replicate populations. The other alleles do not show significant differentiation between High and Low populations. At the syntaxin4 locus, an Nco I RFLP shows significant differentiation between High and Low when alleles are pooled across all replicates, and when L1–L3 are compared to E1–E3 or E4–E5. High populations E3, E5, and E6 are significantly differentiated from each Low population, but there is clear heterogeneity across High populations in allele frequency.

Collectively, there is strong evidence for a selective event in the region spanning chromosome location 3A7 to 3B4. The SGG marker has the clearest signature of selection as the allele frequency differentiation between High and Low is the greatest and the variation among replicate High and replicate Low populations is lowest (Fig. 3B; Table 1). It is unclear whether selection favored specific alleles in the High populations or eliminated them from the Low populations. Because Low populations generally lack the allele that appears differentiated, whereas High populations show more variation, selection to remove alleles from the Low populations is more apparent, but the greater number of High replicates makes a firm conclusion difficult.

In an effort to determine if the shaggy gene itself was part of this response, we compared the thermal knockdown profiles of the sgg1 mutation to a wild-type allele. The mutant has a slightly lower TKD than wild type, but the variance among flies in the mutants was very different from wild type or the High and Low lines, suggesting other background effects (data not shown). Shaggy has many alternatively spliced transcripts and the sgg1 mutation may not capture the variation among High and Low lines.

CLINAL ANALYSIS OF SGG AND OTHER MARKERS

The repeatability of the response of the SGG marker among independent experimental population (TAZ and TKD) suggests that these samples from natural populations harbor common variation that is sensitive to thermal selection. To test the hypothesis that the same alleles that responded to artificial selection in the laboratory also show clinal variation with latitude in natural populations, we surveyed allele frequency at the SGG microsatellite, and a subset of other loci in the band 3 region, in North American and Australian samples that span latitudinal gradients (see Methods). In North America, SGG is clinal in a direction that mirrors the response to selection for elevated TKD in the laboratory: the SGG-118 allele is significantly increased in frequency in low-latitude samples from Florida compared to higher-latitude samples from Pennsylvania and New Jersey, or from Maine; see Fig. 3C. The other SGG microsatellite alleles do not show significant clinal variation as the mid latitude samples (Pennsylvania and New Jersey) are inconsistent with a trend. In Australia, the SGG marker is not clinal, as inferred from samples at several locations (see Fig. 3D).

The DROSEV2 microsatellite marker showing differentiation in the initial screen (Fig. 2) shows clinal variation consistent with the TKD selection: the 152 allele was enriched in the High populations and is more common at low latitudes in both North America and Australia (North–South frequencies of North American DROSEV2–152 are: Maine [latitude 44°N] = 0.10, Pennsylvania/New Jersey [latitude 40°N] = 0.28, Florida [latitude 25°N] = 0.40; G-test Maine vs. Florida with n = 40 per population: Gadjusted = 6.3; P < 0.02. South-North frequencies of Australian DROOSEV2–152 are HF [latitude 18] = 0.26, HY [latitude 18] = 0.11, Sorell [latitude 42] = 0.11. G-test HF vs. Sorell with n = 40 per population: Gadjusted = 2.7; ns.). We acknowledge, however, other alleles at DROSEV2 do not show clear North–South differentiation, and heterogeneity among locations within latitudes prevent a significant regression of allele frequency on latitude. In line with these variable results, DROSEV2 was not significantly differentiated in the TAZ experiment, but it does fall in the thermal QTL uncovered by (Norry et al. 2007), suggesting it has some weak connection to thermal selection.

For the DMZW (band 3A), Period and Pon markers, some individual alleles showed clinal trends in North America or Australia, none of the relationships were significant as inferred from regression of allele frequency on latitude, or using G-tests of allele frequency differences. We acknowledge that the sample sizes for some loci were limited resulting in low power to detect trends, but in general, those alleles showing weak clinal trends in wild samples were not the same alleles that showed evidence for differentiation between the divergent TKD populations.

The background genomic level of North–South differentiation has been estimated in the same samples of lines we used for our clinal analyses (Turner et al. 2008). This study used whole genome hybridization to tiling arrays in a North versus South comparison for separate samples from North American and from Australia and identified thousands of markers with North–South differentiation, but only104 with strong evidence of differentiation in both populations. Two probe regions in the shaggy gene were significantly differentiated in Australia, but not in North America (cf. Turner et al. 2008, Table S2). Notably, the closest flanking markers to the left (distal) and right (proximal) that also show North–South differentiation lie outside the map locations implicating SGG in our study (see Figs. 3 and S2) suggesting that independent loci in this larger genomic region can show latitudinal variation. However, the number of significant probes flanking the SGG region varies as expected with different false discovery rates (cf. Turner et al. 2008, Table S2).

Discussion

Artificial selection in the laboratory for increased performance in alternative thermal environments has uncovered significant QTL on the X-chromosome of D. melanogaster. Notably, two independent selection experiments have identified the same QTL in the shaggy gene region despite rather different genetic backgrounds and distinct selection regimes (laboratory natural selection vs. high-temperature knockdown: TAZ vs. TKD). The specific alleles of the QTL marker that were increased in frequency in the high-and low-selection lines of thermal knockdown experiment show parallel frequency differences between Maine (latitude 44) and Florida (latitude 25), suggesting that artificial selection in the laboratory has recapitulated some aspect of selection maintaining clinal variation in nature. In short, two selection experiments are better than one, and experimental evolution can be informative about a third line of evidence for evolution: clinal variation in nature.

The significance of these results is underscored by the contrast of the two experiments. Although temperature-sensitive performance was the target of selection in both cases, the standing pool of genetic variation and the specific phenotypes selected upon were very different. The TAZ lines, derived from reciprocal crosses between two inbred lines, should have high heterozygosity but low allelic diversity (~2 alleles per locus, 1 allele each from the two founding inbred lines from Arvin California or Zimbabwe). The TKD population should have relatively high heterozygosity and high allelic diversity for alleles common in California populations. The TAZ population was allowed to recombine in laboratory populations for ~40 generations, but this should still have higher levels of linkage disequilibrium than the wild population used in the TKD experiment. Data from the microsatellite mapping experiments confirm these expectations: only two alleles were found at all but one locus in the TAZ samples and multiple alleles were found at each locus in the TKD. Moreover, linkage disequilibrium was complete between the adjacent markers SGG and DMZW in the TAZ population, but these markers showed no clear linkage association in the TKD population. Because North American and African populations of D. melanogaster are differentiated for a variety of genetic markers (Hale and Singh 1987; Singh and Rhomberg 1987; Begun and Aquadro 1995; Anderson et al. 2008), the functional variation in the TAZ and the TKD population is likely to be quite distinct.

In the TAZ populations, the phenotype of selection was overall competitive ability in population cages, maintained at different temperature (18°C vs. 28°C). Since this comprises all aspects of fitness across the life cycle, including egg-to-adult development time, mating speed or proficiency, and fecundity, it is hard to identify the specific target of selection. The TKD experiment involved 25% truncation selection for different temperatures at which flies lost the ability to hold on to an inclined surface in the range of ~37°C to 42°C. The selection protocol did not involve reproduction per se, because all flies in the 25% tail that was selected were allowed to re-mate and leave offspring. Thus, the TKD selection is more sharply focused on physiological traits related to adult high-temperature thermotolerance. We further acknowledge that both assays may have involved selection on desiccation or other stress-related traits, even though temperature was the focal manipulation. It is thus quite surprising that the TAZ and TKD experiments would uncover any QTL in common. As shown in Figures 1 and 2, mapping data uncovered QTL that are unique to each experiment (DROMHC, DROTROPI1, DROABDB). These QTL are certainly of interest for future experiments, and given the low resolution of our mapping data, may be independent evidence for the broad thermal QTL reported for the second and third chromosome of D. melanogaster (Anderson et al. 2003; Norry et al. 2004; Morgan and Mackay 2006; Vermeulen et al. 2008a). But taken together, two more general questions emerge form the current study: why are X-linked thermal QTL in common, and what functions linked to the SGG marker might be strong candidates for thermal selection?

WHY X-LINKED THERMAL QTL?

Recent QTL mapping studies have identified thermal QTL in D. melanogaster, but X-linked regions appear underrepresented and none of the studies has identified QTL in band 3 as we report here (Morgan and Mackay 2006; Norry et al. 2007, 2008; Vermeulen et al. 2008a). One possible explanation for this discrepancy is the random sampling of alleles that is inherent in QTL mapping experiments. If a single pair of parental lines is used, allelic variation is very limited and will not capture most functional variation segregating in nature. It may simply be good luck that the two sets of lines we happened to use in our experiments carried meaningful functional variation on the X-chromosome, and even carried the same QTL in the band 3 region.

A second possible explanation for the greater evidence for X-linked QTL in our study is that the phenotypes we studied were particularly sensitive to recessive variation that may have been exposed in males. The hemizygous state of males could allow expression of thermally sensitive traits that were masked in females, and some aspect of our laboratory natural selection (TAZ) or knockdown selection (TKD) drove a response in males. There are significant differences between males and females in thermotolerance phenotypes as well as in the induction of Hsp70 and Hsc70 (Folk et al. 2006). However, the difference between High and Low TKD populations is only significant in females (Folk et al. 2006). Thus, it seems unlikely that the X-linked map location of our QTL was due largely to exposure of recessive thermally sensitive phenotypes in males.

ARE THERE GOOD CANDIDATE GENES IN BAND 3?

Hitchhiking mapping studies in D. melanogaster have implicated the genomic region near band 3 as possible sites of selective sweeps associated with colonization out of Africa (Harr et al. 2002; Pool et al. 2006). From the Harr et al study, the period or syntaxin4 genes become promising candidates. Studies have indicated that the period locus is under clinal selection related to temperature, and that the Thr-Gly repeat region is a potential target of selection within period (Sawyer et al. 1997, 2006). In our mapping populations, neither period nor syntaxin4 provides as compelling support for the target of selection as does the SGG marker. Moreover, the Thr-Gly repeats do not show clinal variation in Australia in our samples. The DMZW marker, which maps to the 3′ end of the SGG locus less than 500 bp to the right of the SGG microsatellite, also shows significant but inconsistent differentiation between High and Low TKD populations (Figs. S2 and 3). The robust evidence for selection at the SGG marker compared to the closely linked DMZW and per repeats raise the possibility that a recombination event occurred during the selection process that separating markers with inconsistent patterns (DMZW and period) from selected markers (SGG) in the shaggy locus. This could also explain the evidence for differentiation at the Syx4 marker further to the right (proximal) on the X-chromosome (see Figs. S2 and 3).

To the left (distal) along the X, markers localize the breakpoint as a maximum of 300 kb away in chromosome band 2F4 where no differentiation is evident. A marker closer to SGG at band 3A7, ~30 kb to the left of shaggy shows significant allelic differentiation between High and Low lines, but the inconsistent pattern among the High replicates raises questions about whether the marker was influenced by selection. Between the 2F4 and SGG markers are ~30 genes with varied functions (see Fig. S2). One notable gene is pdfr, a G-protein coupled receptor identified as the Drosophila pigment dispersing factor (PDF) receptor. PDF is a neurotransmitter that is required for normal circadian activity rhythms (Yoshii et al. 2009), and may have functional connections to the activity of period and shaggy.

The shaggy gene is a serine-threonine kinase that is involved in the regulation of the circadian clock by phosphorylating the TIMELESS protein and promoting the nuclear localization of the TIMELESS/PERIOD heterodimer (Martinek et al. 2001). Shaggy is also a homolog of glycogen synthase kinase-3 (GSK-3) that phosphorylates heat shock factor-1 (HSF-1), which regulates the expression of heat shock proteins. Thus, shaggy itself remains a very strong candidate for the target of selection in our experiments.

WHAT IS THE CONNECTION BETWEEN PHOTOPERIODISM AND THERMOTOLERANCE?

Within the limits of resolution provided by the current mapping data, it is noteworthy that the region carries three genes involved in circadian rhythms (pdfr, shaggy, and period). Shaggy’s function in the regulation of heat shock proteins provides and logical phenotype for the action of selection. HSF-1 is maintained as an inactive monomer by phosphorylation and the SGG protein is a source of this kinase activity. When de-repressed, HSF-1 trimerizes, enters the nucleus and initiates transcription of Hsps. Overexpression of SGG (or its homolog GSK-3beta) can reduce the transcriptional activity of HSF-1 and Hsp protein levels (He et al. 1998). If our TKD selection acted upon varied levels of Hsps, then this could have been mediated though variation in the ability of SGG alleles to keep HSF-1 in the inactive state. It is not clear whether this would be a polymorphism affecting the protein or the expression of various shaggy transcripts (at least 13 transcripts and polypeptides are known (see flybase.org). A test of this prediction would be to examine the expression levels of shaggy transcripts under conditions similar to the knockdown assay used for the selection experiments. This work is underway and will be reported elsewhere.

It is also possible that our selection experiments acted indirectly on functions related to the perception of photoperiod. Although we did not manipulate photoperiod in our experiments, a connection between photoperiod, temperature, and clinal variation has long been recognized (Bradshaw and Holzapfel 1977; Pittendrigh et al. 1991). Indeed, pdfr, shaggy, and period could all have been involved in the response to selection based on the known relationship between temperature compensation and the clock cycle in Drosophila. Flies can maintain normal circadian activity profiles at different temperatures (Zimmerman et al. 1968), and period plays a role in this regulation (Sawyer et al. 1997). Flies maintained in constant darkness still show circadian activity patterns, but these are reduced significantly in constant daylight. However, the circadian rhythms can be restored in constant daylight if temperature cycles are imposed (Yoshii et al. 2005). Notably, PDF is critical in adjusting the cycle to different stimuli (Yoshii et al. 2009). Transcription of the period locus is regulated by negative feedback system involving the PER protein itself and, in part, its regulation by SHAGGY (Young and Kay 2001). The regulation of splicing of period RNAs is temperature dependent and is associated with temperature dependent adjustment in the period of evening activity (Chen et al. 2007). The various mechanisms of the clock system suggest that it is important in adaptation to the variety of environmental variables that change on daily and seasonal scales (Stoleru et al. 2007).

Together, the apparent coincidence of genes associated with photoperiodism in the QTL region we have uncovered by artificial selection on thermal traits supports existing studies that are pursuing such a connection (Bradshaw and Holzapfel 1977; Bradshaw et al. 2003, 2004). Gene expression profiles in Drosophila thermal selection lines have uncovered an excess of phototransduction genes being upregulated (Nielsen et al. 2006). Photoperiodism and phototransduction are not the same process, but the connection to thermal stress suggests an important functional overlap. Studies in plants have revealed clear associations between genes involved in the perception of photoperiod, flowering time, and clines tracking latitudinal or climatic gradients (Stinchcombe et al. 2004; Zhang et al. 2008). For a small ectotherm such as Drosophila, where dealing with heat or cold stress requires upregulation of expensive chaperones such as Hsp70, perhaps the best cue a fly can have for when, and how much, of this defense system to marshal is to measure the photoperiod. It would be inefficient to maintain constitutively high levels of Hsp70, and much more efficient to express this at the time of day, or season, when it is most needed. Short day lengths occur in colder winter months and long day lengths in hotter summer months, so an adaptive prediction is that the expression of Hsp70s, other chaperones, or diapause states would be tuned to the predicted temperature range that is correlated photoperiod (Sandrelli et al. 2007; Tauber et al. 2007; Schmidt et al. 2008). Although the current study cannot establish these relationships, it provides strong set of candidates for testing hypotheses about the parallel pathways that regulate fitness in variable environments.

Supplementary Material

Supplementary Figures & tables

Figure S1. Map of allele frequency differences for the TAZ populations.

Figure S2. Fine scale map of the region surrounding the thermal QTL in the TKD populations.

Table S1. Microsatellite markers used for disequilibrium mapping.

Table S2. ANOVA for allele frequency variation across the X-chromosome in the TAZ selected populations.

Acknowledgments

We would like to thank three anonymous reviewers for comments that improved the manuscript significantly, Lietta Nicolaides and Johanna Kowalko for scoring some SNPs, and Dawn Abt for expert technical assistance in the laboratory. This work was supported by NSF grant DEB 0343464 to GWG and DMR.

LITERATURE CITED

  1. Anderson AR, Collinge JE, Hoffmann AA, Kellett M, McKechnie SW. Thermal tolerance trade-offs associated with the right arm of chromosome 3 and marked by the hsr-omega gene in Drosophila melanogaster. Heredity. 2003;90:194–202. doi: 10.1038/sj.hdy.6800220. [DOI] [PubMed] [Google Scholar]
  2. Anderson JA, Song YS, Langley CH. Molecular population genetics of Drosophila subtelomeric DNA. Genetics. 2008;178:477–487. doi: 10.1534/genetics.107.083196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Balanya J, Oller JM, Huey RB, Gilchrist GW, Serra L. Global genetic change tracks global climate warming in Drosophila subobscura. Science. 2006;313:1773–1775. doi: 10.1126/science.1131002. [DOI] [PubMed] [Google Scholar]
  4. Begun DJ, Aquadro CF. Molecular variation at the vermilion locus in geographically diverse populations of Drosophila melanogaster and D. simulans. Genetics. 1995;140:1019–1032. doi: 10.1093/genetics/140.3.1019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Berrigan D, Partridge L. Influence of temperature and activity on the metabolic rate of adult Drosophila melanogaster. Comp Biochem Physiol A Mol Integr Physiol. 1997;118:1301–1307. doi: 10.1016/s0300-9629(97)00030-3. [DOI] [PubMed] [Google Scholar]
  6. Bradshaw WE, Holzapfel CM. Interaction between photoperiod, temperature, and chilling in dormant larvae of the tree-hole mosquito, Toxorhynchites rutilus Coq. Biol Bull. 1977;152:147–158. doi: 10.2307/1540555. [DOI] [PubMed] [Google Scholar]
  7. Bradshaw WE, Quebodeaux MC, Holzapfel CM. Circadian rhythmicity and photoperiodism in the pitcher-plant mosquito: adaptive response to the photic environment or correlated response to the seasonal environment? Am Nat. 2003;161:735–748. doi: 10.1086/374344. [DOI] [PubMed] [Google Scholar]
  8. Bradshaw WE, Zani PA, Holzapfel CM. Adaptation to temperate climates. Evolution. 2004;58:1748–1762. doi: 10.1111/j.0014-3820.2004.tb00458.x. [DOI] [PubMed] [Google Scholar]
  9. Calboli FC, Kennington WJ, Partridge L. QTL mapping reveals a striking coincidence in the positions of genomic regions associated with adaptive variation in body size in parallel clines of Drosophila melanogaster on different continents. Evolution Int J Org Evolution. 2003;57:2653–2658. doi: 10.1111/j.0014-3820.2003.tb01509.x. [DOI] [PubMed] [Google Scholar]
  10. Cavicchi S, Guerra D, Giorgi G, Pezzoli C. Temperature-related divergence in experimental populations of Drosophila melanogaster. I. Genetic and developmental basis of wing size and shape variation. Genetics. 1985;109:665–689. doi: 10.1093/genetics/109.4.665. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chen WF, Low KH, Lim C, Edery I. Thermosensitive splicing of a clock gene and seasonal adaptation. Cold Spring Harb Symp Quant Biol. 2007;72:599–606. doi: 10.1101/sqb.2007.72.021. [DOI] [PubMed] [Google Scholar]
  12. David JR, Araripe LO, Chakir M, Legout H, Lemos B, Petavy G, Rohmer C, Joly D, Moreteau B. Male sterility at extreme temperatures: a significant but neglected phenomenon for understanding Drosophila climatic adaptations. J Evol Biol. 2005;18:838–846. doi: 10.1111/j.1420-9101.2005.00914.x. [DOI] [PubMed] [Google Scholar]
  13. Folk DG, Zwollo P, Rand DM, Gilchrist GW. Selection on knockdown performance in Drosophila melanogaster impacts thermotolerance and heat-shock response differently in females and males. J Exp Biol. 2006;209:3964–3973. doi: 10.1242/jeb.02463. [DOI] [PubMed] [Google Scholar]
  14. Gilchrist GW, Huey RB. The direct response of Drosophila melanogaster to selection on knockdown temperature. Heredity. 1999;83:15–29. doi: 10.1038/sj.hdy.6885330. [DOI] [PubMed] [Google Scholar]
  15. Gilchrist GW, Huey RB. Plastic and genetic variation in wing loading as a function of temperature within and among parallel clines in Drosophila subobscura. Integr Comp Biol. 2004;44:461–470. doi: 10.1093/icb/44.6.461. [DOI] [PubMed] [Google Scholar]
  16. Hale LR, Singh RS. Mitochondrial DNA variation and genetic structure in populations of Drosophila melanogaster. Mol Biol Evol. 1987;4:622–637. doi: 10.1093/oxfordjournals.molbev.a040466. [DOI] [PubMed] [Google Scholar]
  17. Harr B, Kauer M, Schlotterer C. Hitchhiking mapping: a population-based fine-mapping strategy for adaptive mutations in Drosophila melanogaster. Proc Natl Acad Sci USA. 2002;99:12949–12954. doi: 10.1073/pnas.202336899. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. He B, Meng YH, Mivechi NF. Glycogen synthase kinase 3beta and extracellular signal-regulated kinase inactivate heat shock transcription factor 1 by facilitating the disappearance of transcriptionally active granules after heat shock. Mol Cell Biol. 1998;18:6624–6633. doi: 10.1128/mcb.18.11.6624. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hochachka PW, Somero GN. Biochemical adaptation: mechanism and process in physiological evolution. Oxford Univ. Press; New York: 2002. [Google Scholar]
  20. Hoffmann AA, Anderson A, Hallas R. Opposing clines for high and low temperature resistance in Drosophila melanogaster. Ecol Lett. 2002;5:614–618. [Google Scholar]
  21. Hoffmann AA, Weeks AR. Climatic selection on genes and traits after a 100 year-old invasion: a critical look at the temperate-tropical clines in Drosophila melanogaster from eastern Australia. Genetica. 2007;129:133–147. doi: 10.1007/s10709-006-9010-z. [DOI] [PubMed] [Google Scholar]
  22. Huey RB, Gilchrist GW, Carlson ML, Berrigan D, Serra L. Rapid evolution of a geographic cline in size in an introduced fly. Science. 2000;287:308–309. doi: 10.1126/science.287.5451.308. [DOI] [PubMed] [Google Scholar]
  23. Lanciani CA, Giesel JT, Anderson JF. Seasonal change in metabolic-rate of Drosophila simulans. Comp Biochem Physiol A. 1990;97:501–504. doi: 10.1016/0300-9629(90)90117-b. [DOI] [PubMed] [Google Scholar]
  24. Martinek S, Inonog S, Manoukian AS, Young MW. A role for the segment polarity gene shaggy/GSK-3 in the Drosophila circadian clock. Cell. 2001;105:769–779. doi: 10.1016/s0092-8674(01)00383-x. [DOI] [PubMed] [Google Scholar]
  25. McColl G, Hoffmann AA, McKechnie SW. Response of two heat shock genes to selection for knockdown heat resistance in Drosophila melanogaster. Genetics. 1996;143:1615–1627. doi: 10.1093/genetics/143.4.1615. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Morgan TJ, Mackay TF. Quantitative trait loci for thermotolerance phenotypes in Drosophila melanogaster. Heredity. 2006;96:232–242. doi: 10.1038/sj.hdy.6800786. [DOI] [PubMed] [Google Scholar]
  27. Nielsen MM, Sorensen JG, Kruhoffer M, Justesen J, Loeschcke V. Phototransduction genes are up-regulated in a global gene expression study of Drosophila melanogaster selected for heat resistance. Cell Stress Chaperones. 2006;11:325–333. doi: 10.1379/CSC-207.1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Norry FM, Dahlgaard J, Loeschcke V. Quantitative trait loci affecting knockdown resistance to high temperature in Drosophila melanogaster. Mol Ecol. 2004;13:3585–3594. doi: 10.1111/j.1365-294X.2004.02323.x. [DOI] [PubMed] [Google Scholar]
  29. Norry FM, Sambucetti P, Scannapieco AC, Gomez FH, Loeschcke V. X-linked QTL for knockdown resistance to high temperature in Drosophila melanogaster. Insect Mol Biol. 2007;16:509–513. doi: 10.1111/j.1365-2583.2007.00747.x. [DOI] [PubMed] [Google Scholar]
  30. Norry FM, Scannapieco AC, Sambucetti P, Bertoli CI, Loeschcke V. QTL for the thermotolerance effect of heat hardening, knockdown resistance to heat and chill-coma recovery in an intercontinental set of recombinant inbred lines of Drosophila melanogaster. Mol Ecol. 2008;17:4570–4581. doi: 10.1111/j.1365-294X.2008.03945.x. [DOI] [PubMed] [Google Scholar]
  31. Oakeshott JG, Gibson JB, Anderson PR, Knibb WR, Anderson DG, Chambers GK. Alcohol dehydrogenase and glycerol-3-phosphate dehydrogenase clines in Drosophila melanogaster on different continents. Evolution. 1982;36:86–96. doi: 10.1111/j.1558-5646.1982.tb05013.x. [DOI] [PubMed] [Google Scholar]
  32. Paaby AB, Blacket MJ, Hoffmann AA, Schmidt PS. Identification of a candidate adaptive polymorphism for Drosophila life history by parallel independent clines on two continents. Mol Ecol. 2010;19:760–774. doi: 10.1111/j.1365-294X.2009.04508.x. [DOI] [PubMed] [Google Scholar]
  33. Pittendrigh CS, Kyner WT, Takamura T. The amplitude of circadian oscillations: temperature dependence, latitudinal clines, and the photoperiodic time measurement. J Biol Rhythms. 1991;6:299–313. doi: 10.1177/074873049100600402. [DOI] [PubMed] [Google Scholar]
  34. Pool JE, Bauer DuMont V, Mueller JL, Aquadro CF. A scan of molecular variation leads to the narrow localization of a selective sweep affecting both Afrotropical and cosmopolitan populations of Drosophila melanogaster. Genetics. 2006;172:1093–1105. doi: 10.1534/genetics.105.049973. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Rand DM, Dorfsman ML, Kann LM. Neutral and non-neutral evolution of Drosophila mitochondrial DNA. Genetics. 1994;138:741–756. doi: 10.1093/genetics/138.3.741. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Sandrelli F, Tauber E, Pegoraro M, Mazzotta G, Cisotto P, Landskron J, Stanewsky R, Piccin A, Rosato E, Zordan M, et al. A molecular basis for natural selection at the timeless locus in Drosophila melanogaster. Science. 2007;316:1898–1900. doi: 10.1126/science.1138426. [DOI] [PubMed] [Google Scholar]
  37. Sawyer LA, Hennessy JM, Peixoto AA, Rosato E, Parkinson H, Costa R, Kyriacou CP. Natural variation in a Drosophila clock gene and temperature compensation. Science. 1997;278:2117–2120. doi: 10.1126/science.278.5346.2117. [DOI] [PubMed] [Google Scholar]
  38. Sawyer LA, Sandrelli F, Pasetto C, Peixoto AA, Rosato E, Costa R, Kyriacou CP. The period gene Thr-Gly polymorphism in Australian and African Drosophila melanogaster populations: implications for selection. Genetics. 2006;174:465–480. doi: 10.1534/genetics.106.058792. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Schmidt PS, Conde DR. Environmental heterogeneity and the maintenance of genetic variation for reproductive diapause in Drosophila melanogaster. Evolution Int J Org Evolution. 2006;60:1602–1611. [PubMed] [Google Scholar]
  40. Schmidt PS, Matzkin L, Ippolito M, Eanes WF. Geographic variation in diapause incidence, life-history traits, and climatic adaptation in Drosophila melanogaster. Evolution Int J Org Evolution. 2005a;59:1721–1732. [PubMed] [Google Scholar]
  41. Schmidt PS, Paaby AB, Heschel MS. Genetic variance for diapause expression and associated life histories in Drosophila melanogaster. Evolution Int J Org Evolution. 2005b;59:2616–2625. [PubMed] [Google Scholar]
  42. Schmidt PS, Paaby AB. Reproductive diapause and life-history clines in North American populations of Drosophila melanogaster. Evolution. 2008;62:1204–1215. doi: 10.1111/j.1558-5646.2008.00351.x. [DOI] [PubMed] [Google Scholar]
  43. Schmidt PS, Zhu CT, Das J, Batavia M, Yang L, Eanes WF. An amino acid polymorphism in the couch potato gene forms the basis for climatic adaptation in Drosophila melanogaster. Proc Natl Acad Sci USA. 2008;105:16207–16211. doi: 10.1073/pnas.0805485105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Schug MD, Wetterstrand KA, Gaudette MS, Lim RH, Hutter CM, Aquadro CF. The distribution and frequency of microsatellite loci in Drosophila melanogaster. Mol Ecol. 1998;7:57–70. doi: 10.1046/j.1365-294x.1998.00304.x. [DOI] [PubMed] [Google Scholar]
  45. Simoes P, Santos J, Fragata I, Mueller LD, Rose MR, Matos M. How repeatable is adaptive evolution? The role of geographical origin and founder effects in laboratory adaptation. Evolution. 2008;62:1817–1829. doi: 10.1111/j.1558-5646.2008.00423.x. [DOI] [PubMed] [Google Scholar]
  46. Singh RS, Rhomberg LR. A comprehensive study of genic variation in natural populations of Drosophila melanogaster. II. Estimates of heterozygosity and patterns of geographic differentiation. Genetics. 1987;117:255–271. doi: 10.1093/genetics/117.2.255. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Stinchcombe JR, Weinig C, Ungerer M, Olsen KM, Mays C, Halldorsdottir SS, Purugganan MD, Schmitt J. A latitudinal cline in flowering time in Arabidopsis thaliana modulated by the flowering time gene FRIGIDA. Proc Natl Acad Sci USA. 2004;101:4712–4717. doi: 10.1073/pnas.0306401101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Stoleru D, Nawathean P, Fernandez MP, Menet JS, Ceriani MF, Rosbash M. The Drosophila circadian network is a seasonal timer. Cell. 2007;129:207–219. doi: 10.1016/j.cell.2007.02.038. [DOI] [PubMed] [Google Scholar]
  49. Stratman R, Markow TA. Resistance to thermal stress in desert Drosophila. Funct Ecol. 1998;12:965–970. [Google Scholar]
  50. Tauber E, Zordan M, Sandrelli F, Pegoraro M, Osterwalder N, Breda C, Daga A, Selmin A, Monger K, Benna C, et al. Natural selection favors a newly derived timeless allele in Drosophila melanogaster. Science. 2007;316:1895–1898. doi: 10.1126/science.1138412. [DOI] [PubMed] [Google Scholar]
  51. Turner TL, Levine MT, Eckert ML, Begun DJ. Genomic analysis of adaptive differentiation in Drosophila melanogaster. Genetics. 2008;179:455–473. doi: 10.1534/genetics.107.083659. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Umina PA, Hoffmann AA, Weeks AR, McKechnie SW. An independent non-linear latitudinal cline for the sn-glycerol-3-phosphate (alpha- Gpdh) polymorphism of Drosophila melanogaster from eastern Australia. Genet Res. 2006;87:13–21. doi: 10.1017/S0016672306007919. [DOI] [PubMed] [Google Scholar]
  53. Umina PA, Weeks AR, Kearney MR, McKechnie SW, Hoffmann AA. A rapid shift in a classic clinal pattern in Drosophila reflecting climate change. Science. 2005;308:691–693. doi: 10.1126/science.1109523. [DOI] [PubMed] [Google Scholar]
  54. Vermeulen CJ, Bijlsma R, Loeschcke V. A major QTL affects temperature sensitive adult lethality and inbreeding depression in life span in Drosophila melanogaster. BMC Evol Biol. 2008a;8:297. doi: 10.1186/1471-2148-8-297. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Vermeulen CJ, Bijlsma R, Loeschcke V. QTL mapping of inbreeding-related cold sensitivity and conditional lethality in Drosophila melanogaster. J Evol Biol. 2008b;21:1236–1244. doi: 10.1111/j.1420-9101.2008.01572.x. [DOI] [PubMed] [Google Scholar]
  56. Weeks AR, McKechnie SW, Hoffmann AA. In search of clinal variation in the period and clock timing genes in Australian Drosophila melanogaster populations. J Evol Biol. 2006;19:551–557. doi: 10.1111/j.1420-9101.2005.01013.x. [DOI] [PubMed] [Google Scholar]
  57. Yoshii T, Heshiki Y, Ibuki-Ishibashi T, Matsumoto A, Tanimura T, Tomioka K. Temperature cycles drive Drosophila circadian oscillation in constant light that otherwise induces behavioural arrhythmicity. Eur J Neurosci. 2005;22:1176–1184. doi: 10.1111/j.1460-9568.2005.04295.x. [DOI] [PubMed] [Google Scholar]
  58. Yoshii T, Wulbeck C, Sehadova H, Veleri S, Bichler D, Stanewsky R, Helfrich-Forster C. The neuropeptide pigment-dispersing factor adjusts period and phase of Drosophila’s clock. J Neurosci. 2009;29:2597–2610. doi: 10.1523/JNEUROSCI.5439-08.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Young MW, Kay SA. Time zones: a comparative genetics of circadian clocks. Nat Rev Genet. 2001;2:702–715. doi: 10.1038/35088576. [DOI] [PubMed] [Google Scholar]
  60. Zhang Q, Li H, Li R, Hu R, Fan C, Chen F, Wang Z, Liu X, Fu Y, Lin C. Association of the circadian rhythmic expression of GmCRY1a with a latitudinal cline in photoperiodic flowering of soybean. Proc Natl Acad Sci USA. 2008;105:21028–21033. doi: 10.1073/pnas.0810585105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Zimmerman WF, Pittendrigh CS, Pavlidis T. Temperature compensation of the circadian oscillation in Drosophila pseudoobscura and its entrainment by temperature cycles. J Insect Physiol. 1968;14:669–684. doi: 10.1016/0022-1910(68)90226-6. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Figures & tables

Figure S1. Map of allele frequency differences for the TAZ populations.

Figure S2. Fine scale map of the region surrounding the thermal QTL in the TKD populations.

Table S1. Microsatellite markers used for disequilibrium mapping.

Table S2. ANOVA for allele frequency variation across the X-chromosome in the TAZ selected populations.

NIHMS304268-supplement-Supplementary_Figures___tables.pdf (160.1KB, pdf)

RESOURCES