Altitudinal patterns for latitudinally varying traits and polymorphic markers in Drosophila melanogaster from eastern Australia

Authors

  • J. E. COLLINGE,

    1. Centre for Environmental Stress and Adaptation Research (CESAR), School of Biological Sciences, Monash University, Victoria, Australia
    Search for more papers by this author
  • A. A. HOFFMANN,

    1. Centre for Environmental Stress and Adaptation Research (CESAR), Department of Genetics, The University of Melbourne, Victoria, Australia
    Search for more papers by this author
  • S. W. MCKECHNIE

    1. Centre for Environmental Stress and Adaptation Research (CESAR), School of Biological Sciences, Monash University, Victoria, Australia
    Search for more papers by this author

Stephen W. McKechnie, Centre for Environmental Stress and Adaptation Research (CESAR), School of Biological Sciences, Monash University, Victoria 3800, Australia.
Tel.: 61 3 9905 3863; fax: 61 3 9905 5613;
e-mail: stephen.mckechnie@sci.monash.edu.au

Abstract

Altitudinal changes in traits and genetic markers can complement the studies on latitudinal patterns and provide evidence of natural selection because of climatic factors. In Drosophila melanogaster, latitudinal variation is well known but altitudinal patterns have rarely been investigated. Here, we examine five traits and five genetic markers on chromosome 3R in D. melanogaster collected at high and low altitudes from five latitudes along the eastern coast of Australia. Significant altitudinal differentiation was observed for cold tolerance, development time, ovariole number in unmated females, and the microsatellite marker DMU25686. Differences tended to match latitudinal patterns, in that trait values at high altitudes were also found at high latitudes, suggesting that factors linked to temperature are likely selective agents. Cold tolerance was closely associated with average temperature and other climatic factors, but no significant associations were detected for the other traits. Genes around DMU25686 represent good candidates for climatic adaptation.

Introduction

Because altitudinal changes occur over relatively small distances, there is generally a more rapid change in environmental conditions, especially temperature, compared with equivalent distances over latitudinal gradients (Heath & Williams, 1979; Baur & Raboud, 1988). As a consequence, for a given temperature change, higher gene flow is more likely along altitudinal gradients compared with latitudinal gradients (Blanckenhorn, 1997). Altitudinal genetic differentiation is therefore less likely to be an effect of nonadaptive processes like founder effects and can be more easily attributed to natural selection, with temperature being a strong candidate selective agent.

In Drosophila, altitudinal gradients have been studied less frequently than latitudinal gradients. Exceptions include altitudinal clines in inversion polymorphisms in D. robusta (Etges & Levitan, 2004), a cline in wing shape in D. mediopunctata (Bitner-Mathe et al., 1995), and altitude differences for wing length, oviposition activity, heat/desiccation responses and hsp70 expression in D. buzzatii (Dahlgaard et al., 2001; Sørensen et al., 2001, 2005). No differentiation was detected in D. buzzatii for eight traits over a short altitudinal transect in the Canary Islands (Bubliy & Loeschcke, 2005). In D. melanogaster from eastern Australia, latitudinal clines have been reported in cold tolerance, heat resistance, ovariole number, development time and body size (James et al., 1995; James & Partridge, 1995; Azevedo et al., 1996; Hoffmann et al., 2002). Some of these traits also show parallel clines on other continents and in other Drosophila species (Watada et al., 1986; Capy et al., 1993; Starmer & Wolf, 1997; Van't Land et al., 1999; Hallas et al., 2002), which suggests that the traits are under climatic selection. Additionally, thermal laboratory selection experiments have implicated temperature as the agent of selection for variation in clinal body size, development time and wing shape (Partridge et al., 1994a; Santos et al., 2004).

At the genetic level, there are several molecular markers on chromosomes 2 and 3 in D. melanogaster that shows latitudinal variation. These mark blocks of genes that are potentially involved with climatic adaptation (Weeks et al., 2002). Markers on chromosome 3, especially on the right arm, are of particular interest as this region contains genes that are involved with variation in thermotolerance and body size (Cavicchi et al., 1989; Partridge et al., 1994b; Bettencourt et al., 2002; Weeks et al., 2002; Anderson et al., 2003). Along the eastern coast of Australia, polymorphic genetic markers on chromosome 3R that show a latitudinal cline include hsr-omegaL/S, hsp70, DMU25686, DMTRXIII and AC008193 (Gockel et al., 2001; Bettencourt et al., 2002; Anderson et al., 2003).

Here we considered altitudinal variation in these third chromosome markers and traits by examining flies from high and low elevation sites at five locations along the eastern Australian latitudinal gradient. In this way, we could directly compare the degree of altitudinal variation relative to latitudinal variation. Population variation in traits and markers were compared with variation in climatic parameters to test if latitudinal and altitudinal patterns were associated with climate variables, irrespective of the physical distance between sites.

Methods

Collection sites

Isofemale lines were set up from field-inseminated females collected from paired locations in February and March 2002. The females were collected from high and low altitude from five different latitudes along the eastern coast of Australia (Fig. 1) using banana bait traps (Tidon & Sene, 1988). Table 1 gives site locations and the number of isofemale lines established from each site, along with climatic variables (mean temperature, rainfall and humidity). Continent-wide surfaces for these variables were interpolated from weather station data (>30 years) with the program anuclim (Houlder et al., 2000) using a 0.05° resolution digital elevation model (DEM) (Hutchinson & Dowling, 1991) and data for exact collection locations were then determined with arcview (http://www.esri.com/software). High altitude sites were cooler and tended to have a lower humidity than low altitude sites from the same latitude.

Figure 1.

Collection sites showing the five paired latitudinal locations. Bold font indicates low altitude sites.

Table 1.  Collection and climate details for five paired latitudinal collection sites.
LocationAltitude (m)LatitudeLongitudeN*(I)†Rainfall (mm)Mean temperature (°C)Humidity
  1. *number alleles scored from isofemale lines in marker analysis; †number of isofemale lines established.

Malanda80017°19′S145°31′E138 (69)161019.670.48
Innisfail1017°30′S145°60′E138 (69)349423.480.85
Springbrook103128°14′S153°16′E68 (17)216015.974.23
Kingscliff1028°17′S153°07′E172 (46)189319.575.11
Armidale129630°31′S151°41′E134 (67)73912.761.85
Coffs Harbour1430°22′S153°06′E106 (53)167118.170.75
Blackheath104633°38′S150°17′E96 (48)118710.968.83
Wollongong1034°25′S150°52′E96 (48)132516.869.62
Adaminaby102535°59′S148°46′E102 (53)7419.363.91
Bega6036°41′S149°50′E72 (36)88014.567.83

Flies were reared on potato medium under the same constant temperature (18 °C), light (12D : 12L) and humidity conditions for a number of generations in the laboratory so that environmentally induced phenotypic plasticity was diminished and genetic differences could be observed (Berven, 1982b). Mass-bred populations were founded by pooling 15 F7progeny from 17 different isofemale lines for each collection location. These populations were left for two generations as massbreds and then tested for the quantitative traits (except ovariole number) and for the molecular markers. Ovariole number was scored directly with flies from some of the isofemale lines (five lines per population) just prior to establishment of the mass-bred lines.

Traits

For all traits, the influence of altitude and latitude was tested by comparing the populations. Additional factors were tested for several of the traits, because of the presence of a hardening treatment (heat resistance), mating status effect (ovariole number) and sex effect (size).

Cold tolerance was measured on the mass-bred populations using a chill coma recovery assay similar to that described in Gibert & Huey (2001). Populations were reared at 18 °C under low-density conditions, by limiting oviposition by 25 adults to 4 days in 275 mL bottles containing 65 mL Drosophila potato medium [potato mash (2% w/v), sugar (3%), agar (0.6%), yeast (3.1%), nipagin (0.11%) and propionic acid (0.5%)]. Females were collected for chill coma 4 – 6 days post-eclosion. Ten females were placed in empty 42 mL capped glass vials for each population and randomized. The vials were placed in a 0 °C water bath (containing ethylene glycol) for 4 h. Vials were removed from the water bath and allowed to recover at 25 °C because this temperature provides a range of recovery times but still allowed several replicate experiments to be completed within a day. Recovery was scored every minute for each fly. A fly was considered recovered when it was able to stand. This process was repeated four times (four replicate ‘runs’ of 10 flies per population).

Heat resistance was measured on individual females from mass-bred populations using a heat knockdown assay (Hoffmann et al., 1997). Populations were reared under low-density at 25 °C in bottles containing Drosophila potato medium, prior to heat knockdown. This rearing temperature is higher than the temperature used for routine maintenance of the lines (18 °C) but population variation in heat resistance is not influenced by rearing temperature (Hoffmann et al., 2005). Individual mated females from these cultures (4 days post-eclosion) were placed in 5 mL capped vials and submerged in a 39 °C water bath. Heat knockdown was scored every minute and flies were considered knocked down, if unable to stand. Flies were either tested, after being hardened by a prior no-lethal heat exposure (37 °C for 1 h, 6 h prior to testing) or without hardening. Two hardened and two nonhardened flies from each of the 10 populations were tested at the same time and these were arranged in a randomized order. This resulted in 40 females being tested in an assay. The assay process was repeated 18 times (i.e. 18 ‘runs’ of four individual females per population).

To measure the ovariole number of each population, five isofemale lines from each population were chosen at random and scored for this trait. Lines were reared at 18 °C under low density (20 eggs per 45 mL vial). Two vials, each with 7.5 mL of Drosophila potato medium were set up from each isofemale line. The first vial contained four nonmated females and the second vial contained four females and four males. After 5 days, the females were collected and frozen at −20 °C until dissection. Both ovaries were extracted in Becker Ringer's solution (6.5 g NaCl, 0.14 g KCl, 0.2 g NaHCO3, 0.12 g CaCl2, 0.01 g NaH2PO4 made up to 1 L with water) and stained in saturated potassium dichromate for 4 min (Carlson et al., 1998; Wayne & Mackay, 1998). Excess stain was removed using Ringer solution and the ovarioles from both ovaries were extracted and counted. The number of ovarioles per ovary was averaged over females from a particular isofemale line (i.e. one mated and one nonmated data point per line, five data points per population).

Egg to adult development time was determined for the mass-bred populations. One hundred eggs (<8 h old) were collected from each of the 10 populations. To control density, groups of 10 eggs were placed in vials containing Drosophila potato medium, which were allowed to develop at 19 °C (note that flies were reared at 18 °C but development time was tested at 19 °C due to pragmatic reasons related to space availability). Emerging flies were collected, counted and sexed every 24 h until the end of emergence. Emergence data were used to calculate average development time for each vial.

Body size was measured as wing area, using flies that emerged from the development time experiment (i.e. on flies that had been reared at a standard density). The right wings of 10 males and 10 females per population were removed and mounted on glass microscope slides using double-sided tape. A digital image was taken of each wing using a dissecting microscope (40×) (Wild M38, Heerbrugg, Switzerland) and a PixeLINK digital camera (PL-A642) (PixeLINK, Ottawa, ON, Canada). Ten landmarks were placed on the digital image (for landmarks see Schiffer et al., 2004) using tpsDig V1.23 and tps utility program V 1.05 written by James. F. Rohlf (http://life.bio.sunysb.edu/morph/). The x and y coordinates of the 10 landmarks were used to calculate wing size as centroid size (square root of the sum of the squared inter landmark distances).

Genetic markers

Five genetic markers on the right arm of chromosome 3 were scored for altitudinal variation (Table 2) by testing one to two individuals from each isofemale line, four generations after the lines had been established from field flies. We sampled DNA from single flies using a Proteinase K method (Gloor & Engels, 1992). Following DNA extraction, samples were genotyped for two indel polymorphisms (hsr-omegaL/S and hsp70) and three microsatellite loci (DMU25686, DMTRXIII and AC008193). The hsr-omegaL/S 8 bp indel polymorphism was scored using a PCR technique (Anderson et al., 2003), while the hsp70 polymorphism was scored using the protocol described by Bettencourt et al. (2002).

Table 2.  Details of genetic markers.
LocusCytological locationPolymorphism typeNumber of segregating alleles
hsp7087A5–7(3R)139 bp indel2
DMTRXIII88B3 (3R)(CT) repeat25
hsr-omegaL/S93D6–7 (3R)8 bp indel2
DMU2568693F (3R)(AT) repeat20
AC00819394D (3R)(TG) repeat17

Microsatellites DMU25686, DMTRXIII and AC008193 were scored for the frequency of the most common allele (MCA – for each marker present at a global frequency of >15%) using PCR and described primers (Gockel et al., 2001). In addition, one primer of each pair was supplemented with an infrared labelled primer (IRDyes). Amplification was carried out with the following cycling conditions: 95 °C (2 min); 95 °C (1 min), 54 °C (1 min), 72 °C (1 min) 35 cycles; 72 °C (2 min). The reaction reagents include 2 mm MgCl2, 0.2 mm dNTPs, 0.2 pmoles μL−1 each unlabelled primer, 8 nmoles μL−1 labelled primer and 0.5 units Taq polymerase. The PCR products were run on Seqagel at 1500v for 1 h 45 min using the Li-Cor Global IR2 (Li-Cor, Lincoln, NE, USA).

Statistical analysis

anovas were undertaken to examine the effects of the latitude location along the coast and altitude on the phenotypic traits as well as the interaction between these effects. These factors were treated as fixed effects because we deliberately selected latitude points along the eastern coast and high/low altitude sites. In addition, other factors were included for different traits because these were part of the experimental design. For chill coma resistance, we included an effect of run because a number of replicate runs were undertaken with the flies. For heat resistance, we also included an effect of run as well as hardening as a fixed factor in the analysis. For ovariole number, we tested the fixed effect of mating in the anova, whereas for size we included a fixed effect of sex. The mean development time of vials may have been influenced by the proportion of adults that developed from the 10 eggs. We therefore undertook an analysis of covariance (ancova) for this trait, using the proportion of females in the flies emerging as a covariate. As well as testing for significance of all factors, we also computed effect sizes (η2) defined as SSeffect/SStotal, the proportion of the total variance that is attributed to an effect.

Multiple regression analyses were carried out in exploratory analyses to examine associations between population trait means and the three climatic variables. Under strong climatic selection, we might expect associations between trait means and climatic variables regardless of the distance between populations. Both forward selection and backward elimination were used to determine the regression model that fitted the phenotypic data and thereby suggested climatic variables associated with the traits. Only the forward selection analyses are presented as the two approaches always led to selection of the same regression model.

The association between the frequency of genotypic markers in each population and latitude as well as elevation was tested by ancova. Latitude was treated as a covariate and the effect of elevation on marker frequencies was then tested after angular transformation). Multiple regressions were also used in exploratory analyses to examine if any of the climatic variables were associated with population allele frequencies.

Results

Traits

For chill coma recovery, the anova indicated significant effects of altitude, latitude, run and an altitude by latitude interaction (Table 3). Populations from higher latitudes were more tolerant of this stress (Fig. 2) and populations from the different latitudes varied in the extent to which chill coma recovery was affected by altitude. Populations from the higher altitude sites had greater cold tolerance at temperate latitudes but not at tropical latitudes. There was a significant effect of runs as typically found for this trait (Hoffmann et al., 2002). A regression of the climatic variables onto chill coma recovery time led to the same final model regardless of whether forward selection or backward elimination was followed. Under forward (and backward) selection, fitting temperature alone led to the strongest association and a highly significant regression (F(1,8) = 13.50, P < 0.01, R2 = 0.63), but the addition of rainfall and humidity further significantly improved the fit of the regression equation (F(1,6) = 25.01, P < 0.001, R2 = 0.88). Chill coma resistance was therefore closely tied to the climatic variables.

Table 3.  Analyses of variance and covariance of trait data.
TraitSourced.f.Mean squareFPEffect size (η2)
Chill comaRun318 605.1547.84<0.0010.156
Altitude12023.585.20.0230.006
Latitude41009.482.60.0350.011
Alt by Lat41200.213.090.0150.013
Error746388.91  0.813
Heat knockdownHardening11303.8368.72<0.0010.076
Altitude167.003.530.0610.004
Latitude4165.498.72<0.0010.039
Run17117.866.21<0.0010.117
Heat by Alt14.670.250.6200.000
Heat by Lat439.342.070.0830.009
Alt by Lat436.771.940.1020.009
Heat by Alt by Lat42.970.160.9600.001
Error67518.97  0.746
Ovariole numberMating status174.4310.720.0020.090
Altitude139.405.670.0200.048
Latitude43.500.500.7330.017
Mating by Alt127.643.980.0490.034
Mating by Lat414.462.080.0910.070
Alt by Lat49.321.340.2610.045
Mating by Lat by Alt44.230.610.6580.021
Error806.94  0.675
Development timeCovariate12.1118.46<0.0010.117
Altitude11.9717.22<0.0010.109
Latitude40.776.700.0110.170
Alt by Lat40.292.520.0470.064
Error890.11  0.541
SizeSex1189.06166.51<0.0010.430
Altitude11.491.320.2530.003
Latitude45.164.550.0020.047
Sex by Lat41.461.280.2780.013
Sex by Alt10.020.020.9020.000
Alt by Lat44.103.610.0070.037
Sex by Lat by Alt40.560.500.7380.005
Error1801.14  0.464
Figure 2.

Variation in thermal tolerance traits across latitude and altitude (±SE). Closed circles: low altitude; open circles: high altitude.

For heat knockdown resistance, there was a significant effect of run, hardening and latitude on resistance (Table 3). Flies from the northernmost tropical population had the longest knockdown time and heat resistance was increased by hardening as reflected by an increase in knockdown time (Fig. 2). Altitude was marginally nonsignificant and there was a suggestion of increased resistance in the high altitude flies particularly in nonhardened flies (Fig. 2). Interaction effects were not significant in the anova. Population means for heat resistance were not associated with any of the climatic variables in regression analyses regardless of hardening.

There was a significant effect of mating as well as altitude for ovariole number, as well as a significant interaction between mating and altitude (Table 4). Mating decreased ovariole number, whereas the effect of altitude on ovariole number depended on mating status (Fig. 3). Nonmated females from low altitude sites showed relatively higher ovariole numbers, whereas there was no altitude effect on ovariole number in mated females. Ovariole number was not associated with any of the climatic variables regardless of the mating status of the females.

Table 4.  Results of anovas testing the effects of altitude on the frequency of the common allele of five genetic markers with latitude treated as a covariate.
LocusSourced.f.Mean squareFPEffect size (η2)
hsr-omegaL/SLatitude10.170429.6060.0010.795
Altitude10.00370.6350.4520.017
Error70.0058  0.188
DMU25686Latitude10.199479.7000.0000.832
Altitude10.02299.1500.0190.095
Error70.0025  0.073
AC008193Latitude10.01713.0000.1270.299
Altitude10.00010.0210.8900.002
Error70.0057  0.699
DMTRXIIILatitude10.043210.3260.0150.581
Altitude10.00180.4330.5310.024
Error70.0042  0.394
hsp70Latitude10.02772.8430.1360.289
Altitude10.00010.0100.9220.001
Error70.0097  0.710
Figure 3.

Variation in fitness-related traits across latitude and altitude (±SE). Closed circles: low altitude; open circles: high altitude.

Development time depended on the proportion of females emerging from a vial, as indicated by the significant covariate in the ancova (Table 3). There was also a significant effect of latitude and altitude on this trait and an interaction between altitude and latitude (Table 3). High altitude populations tended to develop more rapidly than low altitude populations, whereas the difference among latitudes was because of the relatively rapid development of the most southern location and slow development of the second most northerly location. The significant interaction reflected the fact, that there was less difference between altitudes at middle latitude locations compared with the most southerly and northerly locations. This trait was not associated with any of the climatic variables in the regression analyses.

There was no overall effect of altitude on wing size, but there was a significant effect of latitude and an interaction between latitude and altitude (Table 3). Flies from northern locations tended to be smaller than those from southern locations regardless of altitude. However, the significant interaction effect reflected the small size of flies from the most southerly high altitude site compared with the low altitude population at this latitude (Fig. 3). Although sex was significant, males are smaller, none of the interactions involving sex were significant and there were no significant associations of wing size with climatic variables in the regression analysis.

Markers

In the anova, there was a significant effect of latitude on allele frequency as expected for four of the five markers (Table 4). The frequency of the most common allele of DMU25686 showed a particularly strong latitudinal gradient (Fig. 4). For this marker and the other markers except hsp70, latitudinal patterns match previous published results (Gockel et al., 2002). For instance, hsr-omegaL increased with latitude consistent with published patterns (McColl & McKechnie, 1999; Anderson et al., 2003). Once the effect of attitude was removed in the anova, there was a significant effect of elevation for only one marker (DMU25686) (Table 4).

Figure 4.

Distribution of allele frequency for five molecular markers at low and high altitude (solid bar and open bar respectively) at five different latitudes. MCA, most common allele.

Two markers were associated with climatic variables in the multiple regression analyses. For DMU25686, average temperature significantly predicted the frequency of the most common allele in populations (F(1,8) = 22.75, P = 0.001, R2 = 0.74). For hsr-omega, the frequency of the most common allele was also associated with average temperature (F(1,8) = 23.62, P = 0.001, R2 = 0.75). There were no significant associations for the other markers.

Discussion

Latitudinal patterns described here are mostly consistent with those previously recorded for traits, despite the low number of locations sampled along the eastern Australian coast. This includes the decrease in chill coma resistance, but increase in heat resistance towards low latitudes (Hoffmann et al., 2002), as well as the large wing size and faster development time of high latitude populations (James et al., 1995). Nevertheless we did not find a clinal latitudinal pattern for ovariole number contrary to published data (Azevedo et al., 1996).

Flies from high altitude sites tended to be more resistant to chill coma stress, consistent with the higher levels of cold stress likely to be encountered at high altitudes. This effect was not detected at the lower latitudes, however, selection pressures in tropical latitudes will be reduced, because these locations have high average yearly temperatures with relatively low climatic variability even at high altitudes (Janzen, 1967). The strong linear association between cold resistance and the climatic variables evident from the regression analysis supports a direct role for climatic selection in generating altitudinal as well as latitudinal patterns in this trait.

In contrast to cold resistance, there was no significant difference in heat resistance between high and low altitude sites. Although there is latitudinal variation for this trait (Hoffmann et al., 2002), selection on heat resistance may be weaker than on cold resistance (Gaston & Chown, 1999). This reflects the fact that climatic conditions at the higher end of the temperature range vary less when compared with the lower end, which is reflected in patterns of variation among and within species for lower and upper thermal limits when species are collected from across thermal transects (Addo-Bediako et al., 2000; Chown & Nicolson, 2004). In contrast, D. buzzatii from Argentina show altitudinal variation for heat resistance but not for cold resistance (Sørensen et al., 2005).

Altitudinal variation in ovariole number was suggested in nonmated females but not in mated females. A higher ovariole number may be favoured in low altitude environments but only for unmated females and perhaps only at lower latitudes. Ovariole number has been shown to be higher in nonmated than in mated females in D. melanogaster (Bouletreau, 1978; Carlson et al., 1998). Although ovariole number has previously been positively correlated with latitude in D. melanogaster mated females (Capy et al., 1993; Azevedo et al., 1996), latitudinal variation in nonmated has not been considered. Ovariole number is a highly variable trait even within populations (Delpuech et al., 1995) and shows a high level of phenotypic plasticity at different developmental temperatures (David & Clavel, 1967; Delpuech et al., 1995). The reasons why low altitudes might favour an increase in ovariole number in unmated females is not clear. Changes in body size do not seem to be involved because, there was no consistent association between altitude and wing size that matched the ovariole pattern, although wing size might not closely correlate with the size of the body cavity where ovarioles are located.

The faster development time of the high altitude populations is consistent with findings from other populations of D. melanogaster (Louis et al., 1982) and from other organisms (eg. Berven, 1982a; Dingle et al., 1990). In laboratory selection experiments with D. melanogaster, rapid development has evolved at low rearing temperatures (Anderson, 1966; Partridge et al., 1994b; James & Partridge, 1995; Van't Land et al., 1999) which may help to explain the altitudinal difference for this trait. The results are also consistent with the faster development of high latitude populations as found here and in other studies (James & Partridge, 1995; Van't Land et al., 1999) although latitudinal associations tend to be weak and local selection may have a more important effect on this trait than selection along climate gradients.

We focussed on five chromosome 3R-linked genetic markers, selected from 20 markers now known to be associated with latitude along the Australian eastern coast. These 20 represent about 55% of the 36 polymorphic markers that have currently been surveyed across this full transect. Among the five markers assessed for altitude association only one, the most common allele of DMU25686, occurred at a different frequency in high altitude populations and as expected this marker showed a strong negative association with latitude (Gockel et al., 2002). Additionally, DMU25686 has been correlated with body size variation in one population (Weeks et al., 2002). However, body size did not vary with altitude and DMU25686 is likely to be associated with other traits that vary clinally and with altitude. Although these findings suggest that this microsatellite locus is a target of selection, it is likely to be only an indirect target as DMU25686 is part of a block of genes held together by In(3R)P (Weeks et al., 2002).

Most of the genetic markers scored here showed expected latitudinal patterns, with the exception of hsp70. In previous research, this marker has been shown to exhibit a relatively weak latitudinal pattern (Bettencourt et al., 2002) that may not have been detected with the limited sampling undertaken here. The hsr-omegaL/S polymorphism was strongly associated with latitude and also showed an association with average temperature, suggesting that it is influenced by weak direct or indirect climatic selection as previously postulated (McKechnie et al., 1998; Anderson et al., 2003).

In conclusion, the data indicate that cold tolerance, development time, and nonmated ovariole number vary with altitude. One of the chromosome 3R microsatellite markers also varies with altitude. In almost all cases, altitudinal differences match patterns evident at the latitudinal level. This suggests that variation in these traits and marker are directly or indirectly influenced by climatic selection, although only in the case of cold resistance and two 3R markers were tight associations with climatic variables established.

Acknowledgments

We thank B. Bernardo, J. Shirriffs, J. Griffiths and M. Schiffer for technical assistance and some of the field flies. We are grateful to Michael Kearney for running the arcview/anuclim programs to determine climatic variables at collection sites. This work was supported by the Australian Research Council via their Special Research Centre Program, and by a Systematic Infrastructure Grant from the Department of Employment, Training and Youth Affairs.

Ancillary