SEARCH

SEARCH BY CITATION

Keywords:

  • cline;
  • Drosophila serrata;
  • geographical variation;
  • seasonal variation;
  • stress

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Geographical patterns for quantitative traits in Drosophila and other insects are commonly used to investigate climatic selection. They are usually determined from comparisons of populations over extensive areas and based on one collection per population. Here we consider patterns in the Australian endemic species Drosophila serrata established over a shorter transect with repeated sampling. Summer (prewinter) and spring (post-winter) collections were made from 10 to 14 localities, incorporating the southern border of D.serrata and extending approximately 1000 km northwards along the eastern coast of Australia. Linear or curvilinear associations with latitude were evident for development time, viability and cold resistance but patterns differed between collections. Some geographical (population) and genetic associations between traits were found and these also tended to differ between collections. Results confirm the importance of cold stress resistance over winter to the southern border of this species. Microsatellite markers were developed for D.serrata. These indicated a low level of genetic differentiation between populations, high levels of gene flow and no evidence that the most southerly populations were isolated. The results suggest that selection generated geographical patterns in cold resistance, development time and viability, and that substantial gene flow may prevent adaptation at the border to conditions beyond the current distribution of D. serrata.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Geographical patterns in quantitative traits in Drosophila are usually detected by population comparisons over a substantial distance, typically spanning more than 20° in latitude (Watada et al., 1986; Coyne & Beecham, 1987; Imasheva et al., 1994; Worthen, 1996; van't Land et al., 1999; Hoffmann et al., 2001). Clines in traits examined in these studies may be associated with climatic factors such as temperature and humidity patterns. These patterns can be consistent across continents and/or species (Karan et al., 1998a) suggesting that they are produced by selection. For instance wing length and other size related traits show a well known and studied cline that has been associated with mean temperature and other parameters (David & Bocquet, 1975; Imasheva et al., 1994; Karan et al., 1998b; Gilchrist & Partridge, 1999; van't Land et al., 1999).

Whereas such clinal patterns can indicate selection, they are correlative and do not constitute direct evidence. They also do not indicate the intensity of selection on traits as even weak selection can cause geographical differences to develop among populations, particularly if gene flow is limited. The presence of variation among populations from a similar latitude (e.g. Sokoloff, 1965; Bitnermathe et al., 1995; Hoffmann et al., 2001) suggests that local processes can also generate differences among populations. The impact of local processes is evident from genetically based trait changes in a single locality sampled in different seasons (Sokoloff, 1965; McKenzie & Parsons, 1974; Bouletreau-Merle et al., 1992; Bitnermathe et al., 1995).

Geographical patterns in traits can be useful for investigating evolutionary limits particularly in the context of species borders. Variation among populations can help to identify putative selective factors that limit the range of species as these are expected to increase or decrease towards a border (Hoffmann & Parsons, 1997). The steepness of clines at borders will depend on gene flow which on a broad scale can result in ‘smooth’ clines (Kimura et al., 1993) but on a local scale can prevent differentiation between populations and even lead to maladaptation (Ichijo, 1986; Hoffmann & Parsons, 1997). Local selective forces need to be strong enough to counteract the diluting effects of gene flow and this is unlikely at borders when there is unidirectional gene flow from central to border sink populations (Garciaramos & Kirkpatrick, 1997; Kirkpatrick & Barton, 1997).

In this study we consider geographical variation over a latitudinal gradient of 7° in Drosophila serrata. A number of studies on this species have been undertaken to understand evolutionary processes that determine species borders (Blows & Hoffmann, 1993; Jenkins & Hoffmann, 1999, 2000, 2001) and the evolution of sexual isolation (Blows & Allan, 1998; Blows, 1999; Higgie et al., 2000). The southern border of D. serrata does not appear to be constrained by resource availability (Jenkins & Hoffmann, 2001; van Klinken & Walter, 2001) but is correlated with cold stress levels over winter (Jenkins & Hoffmann, 2001). Species comparisons indicate that D. serrata is more susceptible to cold than species with a more southerly distribution (Jenkins & Hoffmann, 1999). Seasonal variation for cold resistance suggests selection for increased resistance at the southern border; cold resistance in a border population after winter is relatively higher than in some northerly populations whereas this pattern is not evident before winter (Jenkins & Hoffmann, 1999) perhaps because of gene flow or trade-offs between cold resistance and other traits.

Here we expand on the work of Jenkins & Hoffmann (1999) in several ways. We considered geographical variation in a number of populations extending northwards from the southern border of D. serrata and sampled at two times (before and after winter). Variation in development time, viability and size was assessed to test if populations show clinal patterns for a range of traits, and to test if any clinal patterns are stable across time. We also reassessed patterns for cold resistance first noted by Jenkins & Hoffmann (1999). To examine the role of gene flow on geographical patterns, we isolated eight polymorphic microsatellite markers for D. serrata, and used these to characterize patterns of variation among populations before and after winter. The results reinforce the notion that changes in geographical patterns occur at this scale and suggest a role for gene flow in limiting the southern border of D. serrata. They also indicate that clinal patterns in a number of traits in D. serrata change between collections.

Stocks

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Two collections of flies along the eastern coast of Australia were made, the first before winter in February 1999 incorporating 14 sites, and the second after winter in September 1999 incorporating 10 sites. Both collections started from the most southerly point of the D. serrata range at Wollongong, and extended approximately 1000 km north to the Brisbane area. Unfortunately, as observed previously (Jenkins & Hoffmann, 1999), it was difficult to obtain D. serrata near Wollongong after winter (we only caught a single fly) and the southernmost site tested was therefore at Terrigal. Care was taken to ensure that all flies were collected at a similar low altitude (<100 m above sea level) and when possible at the same locations in the two collections. Populations were named according to the town nearest the collection site. Table 1 lists the sites, numbers of isofemale lines established for each population and the number of field flies typed for microsatellite markers.

Table 1.  Number of isofemales lines initiated for each population and number of field individuals typed for microsatellites in each collection period with corresponding latitude.
Collection sitesNumber of isofemale linesNumber of field individuals typed for microsatellitesLatitude (°S)
PrewinterPost-winterPrewinterPost-winter
Redcliffe 6  5 27 : 12
Keppera Park1010262127 : 24
Fingal Head 6 6 91528 : 12
Kingscliff1010242428 : 15
Brunswick Heads1010252228 : 31
Ballina1010212028 : 51
Grafton10 28 29 : 40
Red Rock1010212930 : 01
Nambucca Heads8 16 30 : 37
Cresent Head1010271931 : 10
Manning Point1010242331 : 58
Seal Rocks10619832 : 12
Terrigal78121933 : 26
Wollongong10 29 34 : 24

In the laboratory, stocks were maintained as isofemale lines at 25 ± 1 °C under constant light in 40 mL vials containing 15 mL of dead yeast–sucrose medium. To control for environmental carryover effects but minimize the effects of laboratory adaptation, life-history traits were measured soon after lines were initiated at the F2, F3 and F4 laboratory generations. To reduce inbreeding, a minimum of three replicate vials was maintained for each isofemale line; based on other experiments with isofemale lines (Hoffmann et al., 2001), inbreeding depression is unlikely to influence the traits within this interval of laboratory culturing. When rearing flies for the experiments, larval density was maintained at approximately 30 per vial to minimize any confounding effects as a result of overcrowding.

Development time and viability

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

F1 adults were collected over a 5-day period and aged for 6 days. At this age, maternal carryover effects that can influence offspring viability in D. serrata are not yet evident (Hercus & Hoffmann, 2000) and fecundity is at a peak. For each strain, two females were left for 13 h overnight to oviposit on a small spoon containing medium and eggs were transferred to two replicate vials (10 eggs per vial). Vials were placed at 25 °C and their position was shuffled every second day to counter any minor temperature differences within the cabinet. To assess development time, flies emerging from vials were collected every 8 h. Vials were scored until no new adults emerged for >48 h. At this time the number of pupae in each vial was also counted to obtain pupal viability data. Adult size was also measured on the emerging flies (five females per strain) using wing length (along the third longitudinal vein from the wing tip to its intersection with the anterior crossvein).

Cold resistance

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

F4 females were collected over a 2-day period and aged for 7 days. A minimum of three replicate vials was set up per strain, each vial containing 10 females. To assess cold resistance, flies were placed in glass vials and subjected to a cold stress of −2 °C for 1 h in a water bath of ethylene glycol as described in Watson & Hoffmann (1995). Mortality was scored 48 h following the stress as few females died after this period. Wing length was also scored on a sample of the females (five per strain) measured for cold resistance. Note that although we refer to this test as measuring cold resistance, it is more correctly described as a measure of cold shock resistance.

Microsatellite libraries

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Three partial genomic libraries were constructed for the isolation of microsatellite loci in D. serrata and D. birchii (a sibling species). Purified genomic DNA was isolated from each species using a standard phenol chloroform based extraction method incorporating an RNase step (Sambrook et al., 1989). Approximately 2 μg of purified genomic DNA was double digested with restriction enzymes to construct each library. Drosophila birchii genomic DNA was double digested with Sau 3 A I and Rsa I, whereas D. serrata genomic DNA was digested with Sau 3 A I and Hae III in addition to Sau 3 A I and Rsa I to increase the library coverage. Digests were run on a 1% agarose gel and digest products between 300 and 800 bp were excised from the gel. Size-selected digest products were ligated into Bam H I and Hinc II digested pUC 19 vector. Ligations were desalted and transformed into electrocompetent Escherichia coli JM109 cells (Promega, Annandale, Australia). Colonies were lifted onto N+ hybond membranes. Microsatellite probes (AC)10 and (AG)10 were end-labelled with [γ33P] adenosine triphosphate (ATP) and hybridized to membranes overnight at 55 °C for D. birchii and50 °C for D. serrata. Membranes were exposed toautoradiograph film and film was aligned back tocolonies after exposure. Positive colonies were re-screened to confirm their status.

DNA extraction and PCR

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Both male and female flies were extracted for microsatellite analysis using a modification of the Chelex protocol of Walsh et al. (1991). Flies were frozen with Liquid N2 in a 0.5-mL microcentrifuge tube and crushed with a pestle. To each crushed fly, 150 μL of 5% Chelex solution was added; samples were mixed and incubated a 90 °C for 15 min. Prior to polymerase chain reaction (PCR), samples were centrifuged at 17 900g for 2 min.

Polymerase chain reactions were carried out in 10 μL volumes, with 2 μL of template DNA, 167 μm dNTPs, 1.5 mm MgCl2, 0.1 μm of unlabelled forward primer, 0.03–0.06 μm of forward primer end-labelled with [γ33P] ATP, 0.4 μm of unlabelled reverse primer, 5 μg of bovine serum albumin (BSA), 1×Taq buffer and 0.4 units of Taq polymerase (Promega). Thermocycling was the same for all primer pairs with an initial denaturation at 94 °C for 3 min, followed by 35 cycles of 94 °C for 30 s, 47–57 °C (see Table 4 for annealing temperatures) for 30 s and 72 °C for 30 s. PCR products were electrophorized on 5% polyacrylamide denaturing gels and initially sized with a pUC19 sequence.

Analysis of quantitative data

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Before analysis of quantitative data, outliers were detected through standardized residuals in a regression against latitude. Cases with standardized residuals greater than three standard deviations away from the mean were considered outliers. Data sets were also tested for equality of variances through the Scheffe–Box test, and the Kolmogorov–Smirnov test was used to test deviations from normality. If normality was not detected after outliers were removed, data were transformed appropriately prior to analysis. All analyses were conducted using SPSS V10.0 for Windows.

To test for genetic differentiation between populations, anovas were utilized. As each population is comprised of a number of isofemale lines, population was treated as the main factor with lines nested within this factor. The nested term permits identification of genetic differentiation within populations although it can be difficult to directly relate this variation to heritability (Hoffmann & Parsons, 1988).

Associations with latitude were also considered by regressing trait values against latitude included as a linear and quadratic term. Isofemale lines were considered to be independent data points for each latitudinal position as these had been established from the offspring of different field females. For the microsatellite data, regressions of the most common allele against latitude were undertaken to test for any clinal patterns.

Comparisons of linear and/or curvilinear regressions across collections were undertaken. When linear equations were fitted a method outlined in Sokal & Rohlf (1995) was utilized to test for homogeneity of regression coefficients. For the curvilinear associations with latitude, we used the F probability distribution to test for the degree of diversity across entire curves based on (Motulsky, 1999)

  • image

Correlations to detect associations between traits were computed at two different levels, population and isofemale line. Associations at the population level may reflect the outcome of correlated selective forces acting on a geographical scale and genetic correlations among traits. To detect these associations, populations were considered as data points. Trait associations were also tested among isofemale line means, to reflect underlying genetic correlations. In this case each isofemale line mean was considered as a data point in the analysis. To account for potential population differences that would obscure these associations, isofemale lines were standardized for population differences prior to analysis (Hoffmann et al., 2001). The Dunn–Sidak method was utilized to adjust significance levels for multiple comparisons (Sokal & Rohlf, 1995).

Analysis of microsatellite data

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

All microsatellite loci were tested for genotypic linkage disequilibrium using Fisher's method and were subjected to exact tests using the Markov Chain method (500 batches, 10 000 iterations) to estimate unbiased P-values for Hardy–Weinberg (H-W) equilibrium using the program GENEPOP (version 3.1) (Raymond & Rousset, 1995). Measures of heterozygosity were made using gentetic data analysis (GDA) (Lewis & Daykin, 1999). Genetic differentiation was examined with FST estimates comparing allelic variation between populations. Overall estimates of FST were made for each seasonal collection and pairwise FST values were computed for sites within collections and were tested for significance (based on 10 000 permutations) using FSTAT version 1.2 (Goudet, 1995; Goudet, 2000) which does not rely on Hardy–Weinberg assumptions and corrects for multiple comparisons.

To test for isolation by distance, genetic and geographical distances were compared as an FST/(1 − FST) matrix with a geographical distance matrix (in km) using a Mantel test (10 000 permutations) in GENEPOP. To investigate the temporal population changes, pairwise FST estimates were made between populations present in both collections. These comparisons were tested for significance using the genic differentiation test in GENEPOP that uses the Markov chain method (100 batches, 1000 iterations). P-values were subjected to Bonferroni adjustment and a combined probability test, the Fisher method, was employed to determine if there was overall genetic change between collections (Sokal & Rohlf, 1995).

Gene flow was estimated as Nm and defined as the absolute number of migrant individuals or gametes that move into a subpopulation each generation (Slatkin, 1985; Slatkin & Barton, 1989; Hartl & Clark, 1997) with N being effective population size and m the probability each gamete is an immigrant. Measures of gene flow were made using two methods: private alleles (Slatkin, 1985) (using GENEPOP) and from FST values where Nm = [(1 − FST) − 1]/4 (Hartl & Clark, 1997). These measures are considered to provide reliable estimates of gene flow, although when migration is high FST may overestimate gene flow (Slatkin & Barton, 1989).

Development time

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Genetic differences among populations for this trait were evident for both seasons and both sexes (Table 2). Differentiation among isofemale lines within populations as indicated by the nested term was evident in some collections. For females, this term was significant only before winter, whereas the opposite was true for the males.

Table 2.  Nested anova s comparing population differences for development time, viability, size and cold resistance.
TraitPopulation MS (d.f.)Line (population) MS (d.f.)Error MS ( d.f.)
  • *

    P  < 0.05;

  • **

    P  < 0.01;

  • ***

    P  < 0.001.

Development time
 Female – prewinter249.064 (13)*** 77.807 (109)*50.792 (85)
 Female – post-winter511.878 (9)***123.245 (76)92.844 (72)
 Male – prewinter255.849 (13)*126.509 (110)90.807 (87)
 Male – post-winter628.430 (9)***125.985 (77)***57.336 (66)
Viability
 Pupal to Adult – prewinter  0.298 (13)***  0.080 (113) 0.075 (123)
 Pupal to Adult – post-winter  0.0985 (9)  0.074 (77) 0.056 (83)
 Egg to Adult – prewinter  0.183 (13)**  0.068 (113) 0.052 (123)
 Egg to Adult – post-winter  0.069 (9)  0.068 (77) 0.056 (83)
Size
 Female – prewinter 22.720 (13) 13.055 (105) 9.331 (71)
 Female – post-winter  8.043 (9) 16.296 (75)*** 6.110 (67)
 Male – prewinter 14.223 (13) 16.297 (103)** 8.952 (73)
 Male – post-winter  5.230 (9) 10.470 (75)** 5.224 (57)
Cold resistance
 Female – prewinter  0.199 (11)  0.195 (57)*** 0.030 (281)
 Female – post-winter2383.378 (9)1757.980 (77)***257.701 (245)

Regression analyses (Table 3) indicated significant quadratic relationships for both sexes in both collections, much of it because of the development time of the southern populations. However, the form of the curve differed among collections (see Fig. 1). In the prewinter data set, isofemale lines derived from the highest and lowest latitudes recorded the slowest development times with intermediate latitudes showing the fastest times (Fig. 1). In contrast, in the post-winter data set, development times were reduced at the highest latitudes with the intermediate latitudes recording the slowest times. It is unclear if this switch is the result of a change in performance of flies collected from intermediate latitudes or those derived from both the higher and lower latitudes; because the two sets of flies were scored in experiments carried out at different times, minor variation in environmental conditions such as temperature and humidity may have influenced average development time in the two experiments.

Table 3.  Regression analysis associating traits with latitude. Both linear and quadratic components of latitude are considered. Each isofemale line mean is treated as a data point in the analyses.
TraitLinear component (×103)Quadratic component (×105) Intercept
b  ± SE P -value b  ± SE P -value
Prewinter
 Development time – female−10.276 ± 2.698<0.0010.005 ± 0.0010.000826.133
 Development time – male −7.306 ± 3.2210.0250.003 ± 0.0010.022670.006
 Viability – pupal to adult0.008 ± 0.0030.004 −0.273
 Viability – egg to adult0.005 ± 0.0020.019 −0.155
 CV pup. to adult viability −1.140 ± 0.3850.012162.922
 Cold resistance – female −0.270 ± 0.2970.366 75.517
Post-winter
 Development time – female13.783 ± 5.9750.024−0.007 ± 0.0030.019−438.896
 Development time – male18.210 ± 6.3480.005−0.009 ± 0.0030.004−675.872
 Viability – pupal to adult −0.005 ± 0.0030.118 1.227
 Viability – egg to adult −0.006 ± 0.0030.030 1.265
 CV pup. to adult viability0.987 ± 0.4570.063−79.458
 Cold resistance – female0.802 ± 0.3320.018−54.268
image

Figure 1. Associations between latitude and mean female and male development time at 25 °C in the two collections. Curvilinear regression lines are shown. Error bars represent ±1 standard error based on isofemale line means.

Download figure to PowerPoint

To test for differences in latitudinal associations across collections, a comparison of curvilinear regression lines among collections was conducted. For females, the combined SS (26 425, d.f. = 206) and separate SS (13 949, d.f. = 203) led to a significant (P < 0.001) Fratio (60.52). For males, the F ratio (44.66) was also significant (P < 0.001) based on a combined SS of 31 128 (d.f. = 208) and separate SS of 18 824 (d.f. = 205). The association between development time and latitude was therefore altered between the two collections.

Viability

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

This trait was scored at the pupal and adult stages, to enable egg to pupal viability to be separated from pupal to adult viability. Egg to pupal viability did not vary among populations or seasons (results not presented). For pupal to adult (and egg to adult) viability there was differentiation among populations in the pre-winter collection which was not evident after winter (Table 2). For both seasons and for both measures of viability, there was no significant variation among lines within populations. This suggests that heritable differences in viability within populations are relatively small when compared with variation among populations.

Latitudinal associations with viability were also considered. Egg to adult viability in prewinter collections showed a positive linear association with latitude (Table 3), increasing at higher latitudes. This was associated with latitudinal changes in pupal to adult viability but not egg to pupal viability. In the post-winter collection, there was a significant negative association between viability and latitude, viability decreasing towards the southern border (Fig. 2). Regression coefficients for these opposing relationships differed significantly between collections for both pupal to adult viability (t210 = 3.00, P = 0.003) and egg to adult viability (t210 = 3.13, P = 0.002).

image

Figure 2. Associations between latitude and mean viability as well as coefficient of variation (CV) of viability at 25 °C in the two collections. Linear regression lines are shown. Error bars represent ±1 standard error based on isofemale line means.

Download figure to PowerPoint

Population variances measured as the coefficient of variation (CV) among isofemale line means were also related to latitude. For pupal to adult viability the CVs showed a latitudinal pattern, becoming smaller towards the border for the prewinter collection (Table 3, Fig. 2). A comparison of regression slopes across collections was significant (t20 = 3.48, P = 0.002). There were no latitudinal patterns for this trait in the post-winter collection.

Wing length

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Nested anovas provided no evidence for differentiation among populations for this trait in either collection (Table 2). There were differences among lines in some comparisons; for females, line differences were significant in the collection after winter, and for males line differences were evident in both collections. Regressions against latitude provided no evidence for clinal patterns for either sex (data not presented).

Cold resistance

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Nested anovas yielded similar results in the two collections. Whereas population differences were not significant, there was significant variation among lines within populations (Table 2). Despite the lack of population variation, female cold resistance did show a clinal association after winter (Table 3, Fig. 3) because of an increase in cold resistance with latitude. This increase towards the border is consistent with other published data (Jenkins & Hoffmann, 1999). The comparison of regression coefficients between collections was significant (t152 = 2.40, P = 0.018).

image

Figure 3. Female cold resistance plotted against latitude in the two collections. Linear regression lines are shown. Error bars represent ±1 standard error based on isofemale line means.

Download figure to PowerPoint

Trait correlations

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Selective forces operating on a geographical scale can lead to correlations among traits when populations are compared, although these correlations may also arise from genetic associations among traits within populations. We tested the former by comparing traits across populations and the latter by considering correlations among isofemale line means within populations. At the population level, only two correlations were significant. Male size was negatively associated with viability in the summer collection (r15 = −0.633, P = 0.015) and female size was correlated significantly with cold resistance (r10 = 0.728, P = 0.017) at the geographical level in the post-winter collection. This suggests that common selective factors may influence size and viability or that genetic correlations occur among these traits. However, neither association was evident in the other collection or at the isofemale line level.

There were also some associations at the isofemale line level that were not evident in comparisons of populations. Female development time and size were positively correlated in the prewinter collection (r118 = 0.223, P = 0.015). Male development time and egg to pupal viability were also positively associated in this collection (r124 = 0.318, P = <0.001), as were pupal to adult viability and female size (r119 = 0.242, P = 0.008). However in none of these cases were there consistent associations across the collections, suggesting spurious or weak genetic correlations between the traits.

Microsatellite isolation and variation

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Around 30 000 colonies were screened for microsatellites in D. serrata and D. birchii. From these, 16 and 90 positive colonies were isolated for D. serrata and D. birchii, respectively. A number of positive colonies were sequenced and primers designed (nine pairs for D. birchii and 21 pairs for D. serrata). Eight primer pairs which were autosomal and polymorphic were used in this study (Table 4) and the clone sequences can be found at GenBank (Accession no. AF448356-64).

Table 4.  Characteristics of microsatellite loci. The locus derived from D. birchii is denoted Dbir and loci from D. serrata as Dser.
Locus* Primer (5′-3′)Annealing tem- perature (°C)Repeat sequence in cloneNo. of allelesPrewinter HEPost-winter HEAllele size range (bp)
Dbir3F:TTTAACACTCATACGCCCTTTG52(CA)13270.8410.825241–299
R:AGCTACGGAAGTATGACGAACA      
Dser6F:GAGCAAATCGTGGCAGAAGAG50(CT)10380.9190.920124–170
R:CTCCACCCCCAGCACAAG      
Dser10F:TGTTCCCTTGATACCCTCCCC50(GT)6100.1120.176131–143
R:TCTGGCGTTGAGTGTTAGTGG      
Dser13F:GGATCTTTCTCGCAATTCGG55(CA)7200.7060.627222l–264
R:ACTAACTAACCAACGAAAGCCG      
Dser15F:GGTCTGCGGTTGATTTTTATGG57(CA)12330.8940.878166–217
R:CTGGGACTGAGGCTGGGACT      
Dser16F:TCTCAAGTGGGGTATGCCTGG47(AC)11140.7010.667128–151
R:CGGTAGAGAAGATTCGGACGG      
Dser18F:TTGACGGCCACAGGATTTATTT50(TG)6A(GT)6110.3400.34686–134
R:ATGGAAAGGCTCGATTGTCTC (GA)2(GT)4    
Dser34F:GAGCGAGAACTGGTTTTAC47(AG)15A(AG)6430.9540.951120–204
R:AACTGATACTACTCTTTGTGG      
Average overall loci  24.5 0.6820.674 

All loci were highly variable with allele numbers ranging from 10 to 43, and an average of 24.5. Heterozygosities for each collection ranged from 0.112 to 0.954 with an average of 0.682 prewinter and 0.674 post-winter (Table 4). Fisher tests for linkage disequilibrium indicated that all loci were assorting independently. Exact test for H-W equilibrium revealed that all loci in the prewinter sample were in H-W proportions. However, not all loci conformed to H-W proportions in the post-winter sample. Two loci, Dser15 and Dser34, showed an excess of homozygotes (P < 0.05) in one site for each locus (Ballina for Dser15 and Keperra Park for Dser34).

Population structure and gene flow

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Population substructuring was low in the prewinter sample (FST = 0.002 ± 0.001), although the FST value was significantly different from zero (P = 0.018). The level of population substructuring was also low in the post-winter sample (FST = 0.002 ± 0.0002) and the FST value again differed significantly from zero (P = 0.004). To determine which populations contributed to substructuring, pairwise FST estimates were computed. None of the population pair comparisons were significantly different from zero after correcting for multiple comparisons, with the exception of Seal Rocks and Cresent Heads (FST = 0.0193, P < 0.05) in the post-winter collection. This suggests that the significance in overall FST is a result of small differences among many populations rather than large differences among a few populations. These very low FST values are unlikely to be biologically significant despite statistical significance. Nevertheless to determine whether the genetic differences could be associated with geographical distance, FST/(1 − FST) was plotted against geographical distance (Fig. 4). Mantel tests showed that there was no significant correlation between this measure and geographical distance in either collection (prewinter: positive correlation P = 0.1822 and negative correlation P = 0.8178; post-winter: positive correlation P = 0.1558 and negative correlation P = 0.8443). Populations did not differ significantly between the two collections with all FST estimates close to zero (combined probability, P = 0.399). Regressions of the most common allele at each locus vs. latitude indicated no clinal pattern for any of the loci (data not presented).

image

Figure 4. Correlation between paired FST /(1 −  FST ) values and geographical distance between collection sites of D. serrata: (a) prewinter (b) post-winter.

Download figure to PowerPoint

Overall gene flow estimates were obtained for both collections. Estimates from the FST and private allele methods both indicated very high gene flow, as expected from the low level of genetic differentiation observed in FST measures. The FST method estimated Nm as 125 for both the pre and post-winter collections and the private allele method estimated Nm as 6.75 for the prewinter collection and 6.01 for the post-winter one. The higher estimates obtained by the FST method were consistent with expectation that when gene flow is high, FST will overestimate gene flow (Slatkin & Barton, 1989).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

Jenkins & Hoffmann (1999) previously noted a seasonal influence on geographical patterns for cold stress. In their study differences in cold resistance among populations developed after winter but were absent prior to winter. It was postulated that gene flow in summer eliminated differences among populations whereas selection re-established them during winter, accounting for the higher level of resistance to cold in southern populations. The current results concur in part with these earlier findings. Whereas the anova s provided no evidence ofsignificant population differences, cold resistance increased towards the southern border in the post-winter collection. A comparison of the two regression slopes across collections showed a significant difference, reinforcing the notion that cold stress increases at the southern border over the winter period.

The microsatellite data support the notion of high levels of gene flow among the populations. FST values were extremely low and there was at best only weak evidence for any population substructuring. Nm measures indicated that approximately six individuals were exchanged between populations per generation across the southern range of this species. These estimates are comparable with those from some studies with other Drosophila species. For instance in D. pseudoobscura and D. melanogaster gene flow (Nm) was estimated at two (Schaeffer & Miller, 1992) and six (Bubli et al., 1996) migrants per generation, respectively.

Whereas D. serrata can only be collected in low numbers at the border after winter, there is no evidence that the southern populations have become genetically isolated from more northern populations. Given the high level of gene flow in D. serrata, it is possible that unidirectional gene flow from central to marginal populations may contribute to the loss of clinal patterns for cold resistance prior to winter and that differences in cold resistance among populations are re-established each season.

The geographical patterns for viability and development time suggest that despite high levels of gene flow, differences among populations can develop. For both traits, cline direction changed between collections suggesting that cold resistance is not the only trait influenced by changing selection pressures. Additional collections across seasons are needed to confirm that shifts in the shape of the cline for these traits are related to seasonal selection. In other Drosophila species, clinal variation for development time usually involves an increase in developmental rate towards the tropics (James et al., 1995; Robinson & Partridge, 2001) although there are inconsistencies (Worthen, 1996). The nonlinear nature of these clines suggests that selective factors other than those correlated with latitude are involved. C. Sgró and M. W. Blows (unpublished data) recently showed nonlinear changes in genetic variances among populations of D. serrata for development time suggesting complex selection patterns on this trait.

The size data provide no evidence for clinal (or seasonal) changes in this trait. In a broad geographical survey, Hallas et al. (2002) showed that sharp changes in size only occurred in D. serrata around the tropics but not in southern areas where the current samples were obtained, in agreement with the present results. Size therefore does not appear to be under strong selection at the border in contrast to cold resistance and life-history traits. The lack of population differentiation for size could also reflect a low level of heritable variation for this trait although Jenkins & Hoffmann (1999, 2000) found significant field heritability in D. serrata for morphological traits including wing length.

Viability across the southern distribution of D. serrata shows strong clinal patterns and seasonal shifts. The direction of the cline changed significantly across seasons with border populations demonstrating a higher viability relative to more central populations in the summer collection and the opposite trend for the post-winter collection. There was also a clinal reduction in variation for pupal to adult viability towards the border suggesting directional selection for altered viability. A reduction in genetic variation towards the borders has been hypothesized to contribute to constraints in species border expansion (Hoffmann & Parsons, 1997). However in this study, a reduction in variation towards the border was not maintained across seasons, nor was it evident for the other tested traits. Diapause has been shown to influence the expression of fitness-related traits in Drosophila (Tauber et al., 1986; Williams & Sokolowski, 1993; Tatar et al., 2001). This may be particularly relevant for seasonal studies but does not appear to be evident in D.serrata. Recent field work demonstrated that both field and laboratory reared females who are either held or collected under winter field conditions at the southern border of their range can readily produce viable eggs (A.Magiafoglou, unpublished data).

Trait associations at the genetic (isofemale) level in this study are dependent on collection season, and provide no evidence for strong negative genetic associations among the traits that might influence adaptive shifts. The lack of correlation among traits at the population level suggests that traits respond to selection independently around the southern border area in D. serrata. This contrasts with earlier data suggesting some association between cold resistance and fecundity (Jenkins & Hoffmann, 1999) which was not measured in our study.

In summary, the results suggest that life-history traits as well as cold resistance may also be under selection in D. serrata in populations near the border as geographical patterns are evident despite substantial gene flow. The different patterns we obtained in the two collections suggest that selection pressures vary over time but these remain to be identified. Changing seasonal selection pressures for life-history traits coupled with substantial gene flow may combine to limit further expansion of the southern distribution of D. serrata.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References

This research was supported by the Australian Research Council via their Special Research Centre scheme. We thank Carla Sgrò, Michele Schiffer and two anonymous reviewers for comments on an earlier draft and Miriam Hercus for experimental advice.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Stocks
  6. Development time and viability
  7. Cold resistance
  8. Microsatellite libraries
  9. DNA extraction and PCR
  10. Analysis of quantitative data
  11. Analysis of microsatellite data
  12. Results
  13. Development time
  14. Viability
  15. Wing length
  16. Cold resistance
  17. Trait correlations
  18. Microsatellite isolation and variation
  19. Population structure and gene flow
  20. Discussion
  21. Acknowledgments
  22. References