Low Potential for Climatic Stress Adaptation in a Rainforest Drosophila Species

See allHide authors and affiliations

Science  04 Jul 2003:
Vol. 301, Issue 5629, pp. 100-102
DOI: 10.1126/science.1084296


The ability of sensitive rainforest species to evolve in response to climate change is largely unknown. We show that the Australian tropical rainforest fly Drosophila birchii exhibits clinal variation in desiccation resistance, but the most resistant population lacks the ability to evolve further resistance even after intense selection for over 30 generations. Parent-offspring comparisons indicate low heritable variation for this trait but high levels of genetic variation for morphology. D. birchii also exhibits abundant genetic variation at microsatellite loci. The low potential for resistance evolution highlights the importance of assessing evolutionary potential in targeted ecological traits and species from threatened habitats.

Many species are currently experiencing novel conditions because of habitat fragmentation and climate change, resulting in rapid shifts in species distributions as well as genetic changes in quantitative traits (13). It is often assumed that populations have abundant genetic variation in quantitative traits for adaptation (4), but this is on the basis of studies of genetic variation in generalist model species with broad distributions.

Desiccation resistance is associated with the distribution of many species of insects. Species restricted to tropics, particularly rainforests, have low levels of resistance, reflecting the continuous high humidity encountered in these habitats (5). As conditions become drier and as fragmentation generates edge effects that alter the microclimate within fragments, desiccation stress experienced in rainforests is likely to increase (6). Desiccation resistance has a high heritability in populations of widespread Drosophila species, reflecting a high evolutionary potential and allowing populations with increased levels of resistance to evolve rapidly (710). However, these studies have been undertaken on a restricted set of generalist species that are relatively widespread and easily cultured under artificial conditions, and their relevance to species from restricted humid habitats remains to be assessed.

Drosophila birchii, which is restricted to rainforests of northern Australia and New Guinea (11), is sensitive to desiccation, unlike its closely related sibling species, D. serrata, that has a broader distribution (12, 13). In northern Queensland, D. birchii is normally collected only from rainforest pockets, including a region near Yeppoon where an isolated population marks the southern distribution limit of D. birchii (Fig. 1).

Fig. 1.

Regions of northeastern Australia where D. birchii are found (shaded areas). The locations of the four sites sampled repeatedly to test for laboratory adaptation (small circles) and two major cities (medium circles) are also indicated.

To examine geographic variation in desiccation resistance in D. birchii, we characterized isofemale lines from 12 sites, including Yeppoon, along the eastern coast for resistance (14). There was a weak clinal pattern of resistance increasing with higher latitudes (Fig. 2), in agreement with results from several other Drosophila species (15, 16). This geographic pattern is likely to reflect climatic selection because of reducing levels of precipitation with higher latitudes, which is related to water-loss rates in insects (17). Climatic averages obtained from the Australian Bureau of Meteorology weather stations near the collection sites north of Townsville (latitude 19.3°S) indicate a mean annual rainfall exceeding 2000 mm and more than 120 rain days per year, compared to a rainfall below 1500 mm and less than 100 rain days per year at sites south of Townsville.

Fig. 2.

Clinal variation in desiccation resistance (hours to 50% mortality) in D. birchii. Points represent the means of isofemale lines from 12 locations. Error bars are standard errors based on isofemale line means. A regression of desiccation resistance onto latitude based on isofemale line means is significant [F(1,67) = 11.20, P = 0.001, r2 = 0.14], as is the equivalent regression based on location means [F(1,10) = 7.61, P = 0.02, r2 = 0.43], reflecting an increase in resistance at higher latitudes. The Finch Hatton population used for selection is indicated.

To test the ability of D. birchii to respond to selection for increased desiccation resistance, we exposed flies to a desiccation stress until 80 to 90% had died and then bred from the survivors. This process was repeated for 30 cycles over 50 culture generations (14). The Finch Hatton population was used for these experiments because it had a high level of resistance in the geographic comparison. Selection failed to increase resistance substantially (Fig. 3), suggesting a lack of genetic variation for increased resistance that is contrary to expectations and findings from other Drosophila species selected with the use of similar methods (710). Comparisons of the lengths of time taken for 50% of flies to die (LT50s, estimated by linear interpolation) indicate no evidence for a significant selection response in a direct comparison of all selected and control lines after 28 cycles of selection [nested analysis of variance (ANOVA), F(1,4) = 1.92, P = 0.24]. Two of the selected lines may have exhibited slightly higher resistance (Fig. 3), but differences between the two sets of lines were small (11.6% on average) as compared to increases of two- to fivefold in D. melanogaster, D. simulans, and D. serrata [from data in (7, 8, 10)]. Comparisons of the 10% and 90% mortality points also indicated that the response to selection was not significant.

Fig. 3.

Desiccation resistance of lines during selection and results of replicate line crosses. (A) Desiccation resistance scored at 24° to 26°C averaged across lines as hours taken to 80to 90% mortality (solid line). There were 30 cycles of selection (solid diamonds) over 50culture generations. The anticipated responses to selection (hatched lines) based on heritabilities (7, 9) obtained from selection on D. melanogaster (crosses), D. simulans (open triangles), and D. serrata (open circles) are also presented. (B and C) Time (hours) to 50% mortality in lines and crosses among replicate lines tested at the same time at 25°C after 28 cycles of selection. Means and standard errors (bars) are based on five replicates of 10females per line scored at 1-hour intervals. Selected lines are represented by S-; control lines by C-.

Selection responses can be confounded by several factors, including inbreeding and laboratory adaptation (18, 19). The independent replicate lines were crossed to ensure that the absence of a selection response was not a consequence of inbreeding depression influencing desiccation resistance (14). Because flies derived from crosses between lines had a similar level of resistance to those derived from crosses within lines (Fig. 3), there was no evidence that inbreeding effects countered any selection response. Further selection was also attempted on synthetic populations that had been established by combining flies from the independently maintained replicate lines in an attempt to increase genetic variation available for selection. However, a further six generations of selection on these synthetic populations failed to increase desiccation resistance. Because neither inbreeding nor limited genetic variation within replicate lines appeared to have influenced the selection response, genes increasing desiccation resistance do not appear to be segregating in the Finch Hatton population.

We carried out parent-offspring comparisons to directly measure heritable variation for desiccation resistance in a new mass-bred population established from flies collected at Finch Hatton (14). This population contained abundant genetic variation; an analysis of variation at five microsatellite loci for 60 females indicated an average heterozygosity level (H) of 0.65 ± 0.03 and an average number of alleles per locus (A) of 8.4 ± 2.4. These are similar to values obtained for H (0.56 ± 0.06) and A (7.4 ± 1.4) from field Finch Hatton D. birchii and are also comparable to H and A in widespread species, including D. melanogaster (20). Narrow-sense heritability estimates for desiccation resistance based on single-parent or midparent comparisons were zero (Table 1), reflecting that there was little resemblance between parents and their offspring for this trait and that there was a low level of heritable variation for desiccation resistance in this population. By contrast, heritabilities for wing size and wing aspect in the same population were intermediate to high (Table 1) and consistent with heritability estimates for morphological traits in widespread Drosophila species (21, 22).

Table 1.

Parent-offspring comparisons for desiccation resistance and wing traits in D. birchii from the Finch Hatton population.

Parental comparison Number of families Regression coefficient Probability Narrow-sense heritability Standard error of heritability
Desiccation resistance
Female 122 <0.001 0.992 0 0.090
Male 121 -0.058 0.453 0 0.154
Midparent 113 -0.037 0.64 0 0.095
Wing size (centroid)
Female 66 0.353 0.003 0.706 0.230
Male 66 0.193 0.054 0.386 0.108
Wing aspect
Female 66 0.411 <0.001 0.821 0.198
Male 66 0.340 <0.001 0.680 0.158

Because laboratory adaptation can influence desiccation resistance as well as other traits in Drosophila (23, 24) and confound selection responses, we tested for changes in resistance in lines maintained in the laboratory for different lengths of time. Isofemale lines were established from field flies collected from four sites in 2000 as well as from the same sites in 2001 and 2002. These lines were evaluated simultaneously for desiccation resistance in 2002 (14). Lines from the most northern population were consistently less resistant than those from the southern populations (Fig. 4), suggesting stable population differences despite varying lengths of time in laboratory culture. An ANOVA indicated significant differences among populations [F(3,65) = 9.07, P < 0.001] but no effect of collection year [F(2,65) = 1.84, P = 0.167] or interaction between year and population [F(6,65) = 0.71, P = 0.639]. The population differences match those evident from the more extensive clinal collection, including a slight reduction in resistance in Yeppoon as compared to Finch Hatton (compare in Fig. 3). Laboratory culturing, therefore, did not influence the desiccation resistance of D. birchii and confound the selection results.

Fig. 4.

Effects of laboratory adaptation on desiccation resistance (hours to 50% mortality). Lines were collected from four sites on each of three occasions and held in culture for 6, 19, or 31 generations before testing. Means and standard errors (bars) are on the basis of means of isofemale lines. For locations, see Fig. 1.

Differences in desiccation resistance among D. birchii populations suggest that there has been a past history of selection on this trait. Yet, low levels of genetic variation for desiccation resistance appear to be preventing any further increases in resistance in this rainforest species despite ample genetic variation in other traits and at neutral markers as evident from the microsatellite results. Our results show that genetic variation in neutral markers can provide an incomplete picture of the evolutionary potential of populations, consistent with the weak association between genetic diversity as measured by quantitative methods and that measured by molecular methods (25). The absence of a selection response for traits linked to climatic stress in this study and in a few other cases (26) suggests that levels of variation must be evaluated for ecologically relevant traits in those species that are threatened by climate change and fragmentation, including endangered species (27).

Supporting Online Material

Materials and Methods

References and Notes

View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article