Investigating latitudinal clines for life history and stress resistance traits in Drosophila simulans from eastern Australia

Authors


Carla M. Sgrò, Centre for Environmental Stress and Adaptation Research (CESAR), Department of Genetics, The University of Melbourne, Melbourne 3010, Vic., Australia.
Tel.: +61 3 9902 0332; fax: +61 3 9905 5613; e-mail: carla.sgro@sci.monash.edu.au

Abstract

Latitudinal clines have been demonstrated for many quantitative traits in Drosophila and are assumed to be due to climatic selection. However, clinal studies are often performed in species of Drosophila that contain common cosmopolitan inversion polymorphisms that also show clinal patterns. These inversion polymorphisms may be responsible for much of the observed clinal variation. Here, we consider latitudinal clines for quantitative traits in Drosophila simulans from eastern Australia. Drosophila simulans does not contain cosmopolitan inversion polymorphisms, so allows the study of clinal selection on quantitative traits that are not confounded by associations with inversions. Body size showed a strong linear cline for both females and males. Starvation resistance exhibited a weak linear cline in females, whereas chill-coma recovery exhibited a significant nonlinear cline in females only. No clinal pattern was evident for development time, male chill-coma recovery, desiccation or heat resistance. We discuss these results with reference to the role inversion polymorphisms play in generating clines in quantitative traits of Drosophila.

Introduction

Clinal variation provides the opportunity to identify traits and genes associated with environmental conditions, and, when patterns are replicated and understood in terms of effects on fitness, provide strong evidence for selection (Endler 1977). Numerous clinal studies have been performed in invertebrate species, including Drosophila (Hoffmann & Weeks, 2007). Such clinal studies not only allow insight into the actions of climatic selection on quantitative traits, but they also allow genetic variation in these traits to be linked to particular genes (Weeks et al., 2002). Many quantitative traits show latitudinal clines in Drosophila species, including body size, development time, heat and cold resistance (reviewed in Hoffmann et al., 2004; Hoffmann & Weeks, 2007). These studies have concluded that such clines represent the action of climatic selection operating on these quantitative traits. However, these clinal studies have been performed in species of Drosophila that contain common cosmopolitan inversion polymorphisms that also show clinal patterns. Recent data suggests that these inversion polymorphisms may be responsible for much of the clinal variation exhibited for several quantitative traits in Drosophila (Weeks et al., 2002;Gockel et al., 2002; Anderson et al., 2003; Calboli et al., 2003; Rako et al., 2007).

Paracentric chromosomal inversions are widespread within the genus Drosophila and are found as fixed differences between species and as polymorphisms within species, and have been most extensively studied in D. melanogaster (Krimbas & Powell, 1992; see Hoffmann et al., 2004 for a recent review). They are thought to persist in natural populations because they protect favourable allelic combinations from recombination (Dobzhansky,1970; Hoffmann et al., 2004; Kennington et al., 2006; Kirkpatrick & Barton, 2006). Evidence supporting this hypothesis comes from laboratory based studies (Dobzhansky, 1970) and, importantly, from studies of natural populations collected along latitudinal clines (Schaeffer et al., 2003; Kennington et al., 2006). Common inversion polymorphisms within Drosophila species frequently show latitudinal clines on different continents (reviewed in Hoffmann et al., 2004), and these clines reflect those found for some quantitative traits.

Body size is a classic quantitative trait that exhibits latitudinal clines in numerous organisms including Drosophila (James et al., 1995). In D. melanogaster body size shows latitudinal clines on several continents, including the east coast of Australia (James et al., 1995). Recently, clinal variation in body size has been shown to be tightly associated with an inversion polymorphism in D. melanogaster (Gockel et al., 2002; Weeks et al., 2002; Calboli et al., 2003). By crossing inbred lines of D. melanogaster derived from cline ends of Australia and using quantitative trait locus (QTL) mapping and chromosomal substitution lines, Gockel et al. (2002) showed that the largest effects on variation in body size between these lines were due to the right arm of chromosome three (∼70%). Weeks et al. (2002) used lines derived from a central field population of D. melanogaster from eastern Australia in an association analysis to link known clinal genetic polymorphisms with phenotypic traits also known to vary clinally. They also found a tight association between markers on the right arm of chromosome three and body size. These markers were all within the region spanned by the common cosmopolitan inversion In(3R)Payne which also clines with latitude. Weeks et al. (2002) postulated that this inversion was tightly associated with clinal variation in body size and confirmed that In(3R)Payne was present in the original mapping lines of Gockel et al. (2002). Calboli et al. (2003) have since performed a QTL analysis using inbred lines derived from cline ends in South America and confirmed that the largest effects on body size were due to the In(3R)Payne.

Body size has also been associated with inversions in other Drosophila species (Hoffmann et al., 2004). Although this is intriguing in itself, it presents difficulties for identifying the gene(s) responsible for clinal variation in body size in D. melanogaster and other Drosophila species as the lack of recombination in inversion heterozygotes makes it difficult to isolate the effects of a gene(s) from the inversion.

Body size is not the only trait that has been associated with inversions in Drosophila species. In D. melanogaster, inversions have been associated with heat knockdown resistance (Anderson et al., 2003), cold tolerance (Weeks et al., 2002; Anderson et al., 2003) and development time (Van Delden & Kamping, 1979; Oudman et al., 1991), traits all known to exhibit latitudinal clines. In addition, allozyme polymorphisms have also been shown to be associated with inversion polymorphisms that show clinal variation (van’t Land et al., 2000). In other Drosophila species inversions have been associated with longevity, thorax length, wing shape, development time, female fecundity, male mating success, larva to adult viability and heat resistance (Hoffmann et al., 2004). These associations could present problems for QTL mapping and association studies in natural populations because of the large blocks of genes that will co-segregate due to the presence of an inversion. Ideally, mapping lines should be devoid of or fixed for inversions (homosequential). However, this is rarely checked and if the trait of interest shows a tight association with an inversion (such as body size in D. melanogaster) then homosequential lines reflecting differences found at cline ends may be difficult to establish.

An alternative approach is to use species of Drosophila that do not contain common cosmopolitan inversion polymorphisms but do show clinal variation for the trait of interest. Drosophila simulans is closely related to D. melanogaster, has a similar geographic distribution and generally occupies similar habitats. However, it does not contain any common cosmopolitan inversion polymorphisms (Aulard et al., 2004). Drosophila simulans is found along the entire range of latitudes in eastern Australia, but to date has not been investigated for clinal variation in quantitative traits or molecular markers. Although previous researchers have examined geographic variation among populations of D. simulans (e.g. Capy et al., 1993, 1994), they studied populations collected from different continents and islands (Africa, Europe, USA, Latin America, Seychelles and Australia) so their conclusions may be confounded by differences in selection pressures between continents. In addition, none of the populations sampled represented populations collected from a latitudinal cline on any single continent. Populations of D. simulans collected along a latitudinal cline in Australia therefore provide an opportunity to investigate geographic variation at the molecular and phenotypic levels from a single continent and without the confounding effects of inversion polymorphisms.

Here, we examined populations of D. simulans collected from ten latitudes spanning a tropical-temperate gradient along the eastern coast of Australia. Using isofemale lines from each of the ten populations, we examined six quantitative traits (body size, development time, heat and cold resistance, and starvation and desiccation resistance) for clinal variation. Significant latitudinal clines were present for body size in both females and males and female starvation resistance, whereas cold resistance showed a nonlinear association with latitude in females only. No latitudinal variation was found for the remaining traits. We discuss these results in the context of climatic selection and latitudinal clines found in other species of Drosophila that contain inversion polymorphisms.

Materials and methods

Field collections of Drosophila simulans

Populations of D. simulans were collected along the East Coast of Australia in February–March 2004 at 10 locations along a transect from Sorell in southern Tasmania (latitude 42°46′S) to Maryborough in Queensland (latitude 25°33′S) (Table 1). Populations were collected as close to sea level as possible to minimize altitudinal differences between populations.

Table 1.   Collection sites for populations of D. simulans from the east coast of Australia.
PopulationsLatitudeIsofemale lines (n)
Maryborough25°33′S15
 Brisbane27°37′S12
 Kingscliff28°16′S16
 Red Rock29°59′S14
 Tuncurry32°11′S15
 Sydney33°57′S15
 Moruya35°55′S15
Greensborough37°42′S16
 Legana41°21′S14
 Sorell42°46′S15

Laboratory maintenance of populations

Up to 30 isofemale lines, generated from single inseminated field females, were initiated for each population and these were maintained under constant light conditions at 25 °C. Lines were maintained in 40 mL glass vials, each containing 10 mL of a sugar, agar, yeast and potato medium that was always treated with an antifungal agent and with antibiotics to eliminate bacterial infection. Live yeast was always added to the surface of the culture medium to stimulate oviposition. The isofemale lines were maintained under laboratory culture for four–nine generations at a population size of ∼200 flies/generation before being used to measure the six traits.

Traits measured

A generation before the experiments described below, each isofemale line was subjected to a series of short egg laying periods (between 6 and 18 h) in 40 mL vials as above to control for density effects. Approximately 50 male/female pairs of flies eclosing from these short lays were then placed into two separate empty 40 mL vials per line each containing plastic spoons filled with a treacle-yeast-agar medium covered with a layer of live yeast paste, to encourage oviposition. For the stress resistance traits experimental flies were obtained by leaving flies on spoons for approximately 20–24 h at 25 °C under constant light conditions. Eggs or 1st instar larvae from these spoons were then transferred at a density of 20–30 eggs or larvae per vial into three vials for each line. These vials were placed at 25 °C for development under constant light conditions. Flies eclosing from each vial were sexed under CO2 and transferred into holding vials for at least 24 h before being used in the stress resistance assays described below.

Development time, body size, heat resistance and cold resistance were scored at generation F7, whereas desiccation and starvation resistance assays were conducted at generation F9 of laboratory culture.

Egg-to-adult development time

Flies were placed into vials containing spoons filled with a treacle-yeast-agar medium and left to oviposit for 5 h at 25 °C under constant light. Eggs were placed at a density of 10 eggs/vial, three vials/line for each population. Twelve–fifteen isofemale lines were tested from each population. Vials were placed at 25 °C under constant light for development. Once flies started to eclose, vials were scored at 6 h intervals until most of the flies had emerged after which time they were scored twice a day until all flies had emerged. Flies that had emerged from each of the score periods were frozen at −20 °C for later scoring and sexing. The number of females and males that emerged in each vial was noted for every score period. Egg-to-adult development time was measured from the midpoint of the egg-picking period to the midpoint of adult eclosion scores.

Body size

Body size was measured as wing area on flies from the development time experiment. For each line the right wings (or left wing if the right wing was damaged) of 3–10 females and three–eight males randomly chosen from flies frozen during development time scoring, were removed with fine forceps and mounted onto microscope slides containing double sided sticky tape and covered with cover slips. Wing images were captured using a Wild M3 dissector microscope (Wild Heerbrugg, Heerbrugg, Switzerland) attached to a Nikon DS digital camera (Nikon Australia Pty. Ltd). Captured wing images were than landmarked using the tspDig version 1.2 written by F. James Rohlf. For each wing we obtained the x and y coordinates of eight landmarks following Hoffmann & Shirriffs (2002). These coordinates were than used to calculate centroid size (the square root of the sum of the squared inter-landmarked distances), giving an overall measure of wing size (Hoffmann & Shirriffs, 2002).

Heat knockdown

Heat knockdown resistance was measured on three–four males and three–four females from each line in a small vial (5 mL) knockdown assay following established procedures (Hoffmann et al., 2002). Individual males and inseminated females (7–10 days post-ecolsion including experimental time) were placed into 5 mL glass vials and were submerged into a water bath that was heated to a constant temperature of 39 °C. Heat resistance was scored as the time taken for flies to be knocked down. To ensure that variation among lines was not confounded by common environment effects, replicates from each line came from a different culture vial. For each population, 7–15 lines were tested for females and 9–15 lines for males.

Chill coma recovery

Chill coma resistance was measured on three–four females and three–four males (7–10 days post-ecolsion including experimental time) from each line following established procedures (Hoffmann et al., 2002). Individual males and inseminated females were placed into empty 40 mL glass vials which were immersed in a 10% glycol solution cooled to a constant temperature of 0 °C. After 6 h vials were removed from the cold bath and placed at room temperature and recovery time of flies was scored. This was measured as the time taken for a fly to stand upright. To ensure that variation among lines was not confounded by common environment effects, replicates from each line came from a different culture vial. For each population, 11–15 lines were tested for females and 9–15 lines for males.

Starvation resistance

Groups of three–five females (7–9 days post-ecolsion including experimental time) from each replicate vial were placed into empty 40 mL glass vials which were covered with gauze. These vials were inverted (separated by the gauze) over a second 40 mL glass vial which contained cotton wool and water. The two vials were held together and sealed using Parafilm to maintain high humidity. Vials were than placed at 25 °C under constant light conditions and mortality was scored at 6 h intervals until all flies had died from starvation. Two or three replicates were set up per line with each replicate containing flies that were reared in different culture vials, to avoid common environment effects. For each population, 10–15 lines were tested.

Desiccation resistance

Desiccation resistance was measured on three–five females (7–9 days post-ecolsion including experimental time) from each line. This was measured by placing individual inseminated females into empty 5 mL glass vials which were covered with gauze. These vials were than placed into a glass tank containing silica gel which maintained humidity at 8–10%. These glass tanks were sealed by applying petroleum jelly to the lid and edges of the tank to make them airtight. Mortality was scored every hour until all individuals had died. To ensure that variation among the lines was not confounded by common environment effects one fly from every replicate vial for each line was tested for desiccation resistance. We tested 12–15 lines per population.

Analysis of data

Prior to analysing the heat, cold, desiccation and starvation resistance data, individual values were corrected for block effects by subtracting each individual value in a block from the average for that block. These corrected values were used for all analyses. Starvation resistance was calculated using LT50’s (time taken for half the flies to die) using probit regression. All analyses were performed using spss for windows version 15.0 (SPSS Inc., Chicago, IL, USA).

For all traits measured, nested analyses of variance (anovas), with isofemale lines nested within population, were undertaken to test for differences among and within populations. To compare the relative amount of variation among and within populations the variance components were computed by restricted maximum likelihood using the VARCOMP procedure in SPSS (Hoffmann et al., 2001).

Latitudinal patterns were analysed using linear regression. For this analysis, we tested the association between latitude and the mean of each isofemale line. Since each isofemale line was founded by a different field female, each line was treated as an independent data point for any particular latitude (Hoffmann et al., 2002). Quadratic and cubic components were added to test for any curvilinear relationships when such relationships were indicated by the data. Raw data are used to plot trait means against latitude for visual purposes, with each point representing the mean trait value averaged across all isofemale lines for each population.

Results

Stress resistance traits

Nested anovas showed significant differences among populations for starvation resistance, but not for any of the other stress resistance traits (Table 2). Differences among isofemale lines within populations were significant for all stress resistance traits in both males and females, except for heat resistance. Variance components showed that line effects accounted for a higher proportion of the variance than location effects for all stress resistance traits. Line effects accounted for 1–42% of the variance. Line effects were relatively larger for starvation resistance than for the other stress resistance traits (Table 2).

Table 2.   Nested anovas testing differences among populations and isofemale lines for heat knockdown, chill coma recovery, desiccation and starvation resistance, development time and wing size.
TraitSourced.f.Mean SquareFSignificance of FVariance component (%)
Heat knockdown (females)Population97.0281.1360.3430
Line1196.1941.0530.35511.63
Error3695.882  88.37
Heat knockdown (males)Population96.2901.0890.3750.29
Line1196.7381.0730.3081.34
Error3696.192  98.37
Chill coma recovery (females)Population9435.450.9630.4740
Line127454.631.3750.0118.27
Error398330.61  91.73
Chill coma recovery (males)Population9470.610.7280.6820
Line126569.312.137< 0.00119.15
Error372303.72  80.85
Desiccation resistance (females)Population912.470.8690.5550
Line13814.4851.616< 0.00111.76
Error5398.965  88.24
Starvation resistance (females)Population9829.8876.014< 0.00120.13
Line134141.6724.238< 0.00142.40
Error25633.433  37.47
Development time (females)Population92039.481.4070.19041.22
Line1321617.414.641< 0.00131.99
Error947348.53  66.79
Development time (males)Population92373.711.4260.1821.30
Line1341930.644.826< 0.00133.41
Error945400.05  65.29
Wing size (females)Population90.044524.586< 0.0018.23
Line1280.009982.384< 0.00117.99
Error6480.00419  73.78
Wing size (males)Population90.03943.854< 0.0017.29
Line1270.01062.858< 0.00122.31
Error6690.003713  70.40

Regression analyses showed a nonlinear association between latitude and chill coma recovery for females and males (Fig. 1). Female chill-coma recovery displayed a significant cubic regression with latitude (t0.05(6) = 2.98, P = 0.025, R2 = 0.72), whereas the cubic regression of latitude on male chill coma recovery was nonsignificant (t0.05(6) = −2.14, = 0.076, R2 = 0.71). The linear regression of starvation resistance with latitude was marginally significant, with starvation resistance decreasing with latitude, but there was no significant association between latitude and any of the other stress resistance traits (Table 2; Figs 2 and 3).

Figure 1.

 Association between latitude and mean chill coma recovery time in female (a) and male (b) D. simulans populations originating from different latitudes. Cubic regression is shown. Error bars represent the standard error based on isofemale line means.

Figure 2.

 Association between latitude and mean heat knockdown time in female (a) and male (b) D. simulans populations originating from different latitudes. Error bars represent the standard error based on isofemale line means.

Figure 3.

 Association between latitude and desiccation resistance (a) and starvation resistance (b) in female D. simulans populations originating from different latitudes. Linear regression is shown. Error bars represent the standard error based on isofemale line means.

Development time

There were no significant differences among populations for development time in either sex. Differences among isofemale lines within populations were significant in both males and females (Table 2). Variance components showed that line effects accounted for a higher proportion of the variance than location effects in both sexes. Line differences were similar for males and females, with line effects accounting for 32–33% of the variance (Table 2). No latitudinal pattern in development time was evident for either sex (Table 3; Fig. 4).

Table 3.   Linear regression analyses testing for associations between latitude and all traits tested based on the mean of isofemale lines at each site.
TraitsR2± SEtP-value
Chill coma recovery (females)0.295−0.270 ± 0.148−1.830.105
Chill coma recovery (males)−0.1230.018 ± 0.1570.120.909
Heat knockdown (females)0.2070.029 ± 0.5381.830.104
Heat knockdown (males)0.065−0.025 ± 0.019−1.280.238
Desiccation (females)−0.0030.0243 ± 0.0250.990.352
Starvation (females)0.300−0.482 ± 0.199−2.210.058
Development time (females)0.0890.354 ± 0.2581.370.208
Development time (males)0.0930.373 ± 0.2691.380.204
Wing size (females)0.6500.00337 ± 0.0014.210.003
Wing size (males)0.8830.00370 ± 0.0017.75< 0.001
Figure 4.

 Association between latitude and development time in female (a) and male (b) D. simulans populations originating from different latitudes. Error bars represent the standard error based on isofemale line means.

Body size

Nested anovas showed significant differences among populations and between isofemale lines within populations in both sexes (Table 2). Variance components showed that line effects accounted for a higher proportion of the variance than location effects. Line and population differences were similar for males and females, with line effects accounting for 18–22% of the variance and population effects accounting for 7–8% of the variance (Table 2). Linear regression showed a highly significant association between latitude and wing size for both females and males (Table 3), with wing size significantly increasing with latitude (Fig. 5).

Figure 5.

 Association between latitude and body size in female (a) and male (b) D. simulans originating from different latitudes. Linear regressions are shown. Error bars represent the standard error based on isofemale line means.

Discussion

This study is the first to examine latitudinal variation for life-history and stress resistance traits in populations of D. simulans collected from a single continent. Clinal patterns were evident for female chill coma recovery and body size in both females and males. Female starvation resistance exhibited a weak linear cline with latitude, but no geographic or clinal variation was present for heat and desiccation resistance or development time in either sex.

The nonlinear cline for chill-coma recovery reported here differs from the cline found for the same trait in D. melanogaster and D. serrata found along a similar latitudinal gradient (Hallas et al., 2002; Hoffmann et al., 2002), where chill coma recovery time decreases linearly with latitude. In Australia, latitude is strongly correlated with average temperature (Hoffmann et al., 2001) and the linear cline in D. serrata is associated with minimum temperature which also shows associations with latitude (Hallas et al., 2002). The nonlinear nature of the cline found for cold resistance in this study suggests that this cline in females is likely to be the result of adaptation to climatic factors that do not show linear variation with latitude (Loeschcke et al., 2000). The patterns for cold resistance in females and males reported here suggest that the nature of selection on cold resistance differs between the sexes. Previous studies that have investigated clinal variation for cold resistance have only examined females (Hallas et al., 2002; Hoffmann et al., 2002; Magiafoglou et al., 2002), so we are unable to compare our results to previous studies.

The absence of variation within and between populations for heat resistance in this study indicates a particularly low level of genetic variation for this trait in Australian populations of D. simulans. This finding is consistent with studies on the restricted rainforest species D. birchii (Griffiths et al., 2004), who also found a lack of variation for heat resistance (measured as heat knockdown time) among and within populations of D. birchii. However, the absence of variation for heat resistance in D. simulans contrasts with the findings in its widespread sibling species D. melanogaster which exhibits a linear cline along a similar latitudinal gradient (Hoffmann et al., 2002) and higher levels of within-population variation. It is not clear how selection has generated this cline for heat resistance in D. melanogaster and not D. simulans, as they share a similarly wide geographic range. It may be the result of indirect effects of selection on cosmopolitan inversions, as the cosmopolitan inversion In(3R)Payne has previously been shown to be associated with heat resistance in Australian populations of D. melanogaster (Anderson et al., 2003), however it is also possible that D. simulans may respond to changing temperatures via plastic changes in heat resistance (David et al., 2004). In addition, we have only investigated one measure of heat resistance, and patterns for mortality following heat shock, for example, may be different (Hoffmann et al., 2002).

The lack of clinal variation for desiccation resistance in D. simulans is consistent with findings in D. simulans, D. melanogaster and D. serrata from Australia. Davidson (1990) found no geographic variation in desiccation resistance between populations of D. simulans collected from northern and southern Australia, and populations of D. melanogaster (Hoffmann et al., 2001) and D. serrata (Hallas et al., 2002) from eastern Australia do not exhibit clinal patterns for desiccation resistance. These results contrast with the findings found in drosophilids in India (Karan & Parkash, 1998), where consistent clines occur with desiccation resistance increasing with latitude. It appears that selective factors acting on desiccation resistance are inconsistent across continents, which may be due to different climatic conditions between continents (Hallas et al., 2002). Previous studies in D. simulans (McKenzie & Parsons, 1974; Davidson, 1990), D. melanogaster (Hoffmann et al., 2001) and D. serrata (Hallas et al., 2002) report abundant genetic variation for desiccation resistance within populations consistent with the levels of within-population variation detected in this study, although differences among populations of D. melanogaster account for 7.5% of variation in this trait compared to 0% in D. simulans (this study).

The weak linear cline for female starvation resistance in D. simulans reported here is consistent with the weak association between starvation resistance and latitude found in female D. melanogaster from eastern Australia (Hoffmann et al., 2001) and other drosophilds from India (Karan & Parkash, 1998). In all cases, starvation resistance decreases with latitude. However, no clinal pattern was evident for starvation resistance in populations of D. melanogaster collected from a latitudinal gradient in South America (Robinson et al., 2000), which again suggests that selection on starvation resistance may be inconsistent across continents. Interestingly, the cline in starvation resistance reported here for D. simulans is stronger than that reported by Hoffmann et al. (2001) for D. melanogaster (Z-test on regression slopes, Z = 2.396, P = 0.008), indicating that selection on this trait is stronger in D. simulans. This is also reflected by the higher levels of variation explained by differences among populations in D. simulans (20.13% this study) compared to D. melanogaster (10.5%, Hoffmann et al., 2001).

It is unlikely that the lack of clinal variation in heat knockdown and the (nonsignificant) nonlinear clinal pattern for male chill coma recovery are the result of laboratory adaptation, drift or measurement error. We followed well-established methods of laboratory culture and trait measurement that have previously been used to demonstrate latitudinal clines in D. melanogaster (Hoffmann et al., 2001; Hoffmann et al., 2002)D. serrata (Hallas et al., 2002) and D. birchii (Griffiths et al., 2004), in the absence of any effect of laboratory adaptation. Rather, it is more likely that our results reflect differences in the biology and ecology of D. simulans.

Life-history traits

No geographic or latitudinal patterns of variation were found for development time in either males or females of the populations of D. simulans examined in this study, despite significant levels of variation within populations. Similarly, Hyytia et al. (1985) found no geographic or latitudinal patterns for development time in populations of D. simulans collected from Europe, Africa and North America, although this earlier study is limited by the fact that the latitudinal gradients studied were not on a single continent. In contrast, development time decreased significantly with latitude in populations of D. melanogaster collected along the east coast of Australia (James & Partridge,1995), whereas a nonlinear cline in development time has been reported in populations of D. serrata from eastern Australia (Sgrò & Blows, 2003). Development time has been shown to be associated with clinally varying inversions in D. melanogaster (Van Delden & Kamping, 1979; Oudman et al., 1991) and D. serrata is known to have several common inversions that show clinal variation with latitude (Stocker et al., 2004). Lack of such inversions in D. simulans may in part account for the lack of geographic and population differences in development time in this study, however it is also possible that climatic selection does not act on egg-to-adult development time in this species.

This study is the first to demonstrate a strong genetic cline in body size (measured as wing size) in D. simulans, with size showing a linear increase with latitude. Previous work has found some evidence for geographic and latitudinal patterns of variation in body size (measured as thorax and wing length) in D. simulans (Tantawy & Mallah, 1961; David & Bocquet 1975; Hyytia et al., 1985; Capy et al., 1993, 1994; Morin et al., 1999; reviewed in Gibert et al., 2004), however none of these studies were based on populations of D. simulans collected from latitudinal gradients on a single continent. Climatic differences between continents make it difficult for these studies to ascertain the extent to which populations of D. simulans display geographic or latitudinal variation in body size.

Previous work in D. melanogaster from eastern Australia (James et al., 1995; Azevedo et al., 1998), South America (van ‘t Land et al., 1999; Robinson et al., 2000) and North America (Coyne & Beecham, 1987; Capy et al., 1993) has detected the same clinal pattern, with body size (measured as wing area) linearly increasing with latitude. As average temperature varies linearly with latitude, and given the linear nature of the cline in D. simulans which is consistent with findings in D. melanogaster, it is likely that average temperature is in part responsible for the size cline found in D. simulans. We find that the strength of the cline reported here in D. simulans is similar in strength to that reported for D. melanogaster from eastern Australia (James et al., 1995), with the regression slopes not differing significantly from each other (t-test on slopes, t0.05(1) = −7.502, P = 0.084). In addition, the correlations between latitude and wing size in both studies are similarly high for both sexes (D. melanogaster females r = 0.801, males r = 0.794 vs. D. simulans females r = 0.829, males r = 0.939). This contrasts with comparisons between the two species for other morphological traits (Gibert et al., 2004), which suggests that the effects of climatic selection are weaker in D. simulans. Clinal variation for body size in D. simulans in the absence of inversions indicates that body size, or a trait strongly genetically correlated to body size, is under intense direct climatic selection.

With the exception of body size and starvation resistance, much of the variability in the traits examined appears to be within populations rather than between them (except for heat knockdown where neither population nor line account for significant amounts of trait variation). Geographic location did not account for any variation in chill-coma recovery or desiccation resistance. This suggests that evolutionary forces generating differences between temperate and tropical populations are weak relative to forces generating variation within populations. Such patterns of variation are consistent with previous studies in other Drosophila species (Hoffmann et al., 2001, 2002; Hallas et al., 2002).

In conclusion, we have demonstrated genetic latitudinal clines for body size and female cold and starvation resistance in D. simulans populations from eastern Australia. These clinal patterns are partly consistent with clinal patterns found in D. melanogaster and D. serrata from the same region. Differences between these species for cold and heat resistance as well as development time may be, in part, due to the absence of chromosomal inversions in D. simulans. We find that although D. simulans is less variable than D. melanogaster for some traits (heat knockdown, chill coma recovery and development time) it exhibits similar levels of variation for other traits (desiccation resistance and body size) and even higher levels of variation for starvation resistance. Comparative functional genomic studies between D. simulans and other Drosophila species will help determine the extent to which adaptation to climatic variables occurs via the same or different genetic pathways. Studies examining the extent to which D. simulans responds to environmental change via phenotypic plasticity in stress resistance and life-history traits are also required.

Acknowledgment

We would like to thank the Australian Research Council for funding via their Special Research Centre and Fellowship (CMS) schemes.

Ancillary