Cross-generation effects due to cold exposure in Drosophila serrata

Authors


†Author to whom correspondence should be addressed. E-mail: A.Magiafoglou@latrobe.edu.au

Summary

  • 1Environmental variation experienced in the parental and grandparental generation can affect progeny phenotype, performance and response to selection. Here the effects of parental and grandparental exposure to a non-lethal cold shock are considered in Drosophila serrata Malloch. Development time, viability and early age productivity were measured in flies originating from border and central locations in the distribution of this species that had been held under two separate laboratory maintenance schedules.
  • 2Cross-generation effects were detected for several traits. Development time usually decreased following maternal and/or grandmaternal cold exposure. Parental cold exposure negatively influenced viability while grandparental effects on viability were negligible. Early female productivity showed opposing responses depending on generation; maternal cold exposure increased progeny productivity while grandmaternal exposure decreased it. Male parental and grandparental exposure to cold shock decreased male productivity, although this pattern may have been partly confounded by size effects.
  • 3Population effects, reflecting geographical origin, were limited to development time, while selective background effects were demonstrated for most traits. The influence these factors had on the expression of cross-generation effects was negligible, given interactions with treatment were not evident. These responses suggest that environmental variation experienced in preceding generations can influence progeny phenotype in a manner that is complex and difficult to predict.

Introduction

The expression of progeny phenotype is determined by a range of factors that not only include inherited nuclear and cytoplasmic genes and the developmental environment of progeny, but also environmental effects derived from one or both parents. Parental environment can affect progeny phenotype and performance in a variety of ways, including via mate choice (reviewed in Qvarnstrom & Price 2001), morphological variation (Fox & Savalli 1998; Hunt & Simmons 2000) and life-history variation (Carriere 1994; Schmid & Dolt 1994; Fox, Waddell & Mousseau 1995; Galloway 2001). Environmental factors may act across more than one generation (Miao, Bazzaz & Primack 1991; Fox & Savalli 1998) and encompass a diversity of effects (Mousseau & Dingle 1991; Rossiter 1996; Mousseau & Fox 1998). As selection acts on the phenotype, environmental variation experienced in the parental generation has the potential to influence the direction and magnitude of selective processes acting across generations (Kirkpatrick & Lande 1989; Lande & Kirkpatrick 1990; Mousseau & Dingle 1991; Jablonka et al. 1995; Watson & Hoffmann 1996; Wolf & Brodie 1998; McAdam et al. 2002; Rauter & Moore 2002).

Environmentally induced maternal effects are particularly well documented across a range of species (Schmid & Dolt 1994; Fox et al. 1995; Reznick, Callahan & Llauredo 1996; Mousseau & Fox 1998). Maternal effects are typically more common than paternal effects owing to the tendency for females to invest a greater proportion of their resources in production or care of offspring. However, paternal effects due to environmental experiences have also been demonstrated. For instance, paternal light environment in the plant Campanula americana influences offspring characters (Galloway 2001), paternal rearing host in the Seed Beetle Stator limbatus affects offspring life-history traits (Fox et al. 1995), while paternal provisioning influences morphology in the Dung Beetle Onthophagus taurus (Hunt & Simmons 2000). Furthermore, the direction of cross-generation response to environmental stimuli may be dependent on parental sex. In the plant Solidago altissima, maternal and paternal environment affected progeny fitness in opposing directions (Schmid & Dolt 1994), while in Drosophila, cold exposure in females led to an increase in progeny resistance while the paternal effect resulted in a reduction in progeny resistance (Watson & Hoffmann 1995). Thus, when determining the impact of cross-generation effects, it appears important to consider the relative contribution each parental sex has on progeny fitness, as well as the potential interactions that may exist among the sexes.

While cross-generation effects are environmentally induced, there is growing evidence in the literature suggesting that the expression of these effects may be under some level of genetic control. For instance, Lacey (1996) found that parental temperature effects in the plant Plantago lanceolata depended on parental genotype, and Galloway (2001) found family differences in progeny germination of Campanula americana after exposure to maternal and paternal treatments. The presence of genetic factors influencing the expression of carry-over effects suggests that populations may evolve to alter expression of cross-generation effects in progeny.

In this study we consider the impact of parental and grandparental cold shock on early progeny fitness in Drosophila serrata Malloch. Studies on the effects of temperature extremes on organisms have usually focused on costs and benefits of acclimation within a generation (e.g. Krebs & Loeschcke 1994; Kelty & Lee 2001; Hoffmann, Sorensen & Loeschcke 2003). A fitness difference is often generated by rearing flies at different temperatures or by exposing organisms to sublethal stress levels that cue biochemical changes to increase resistance to future lethal stresses. But how do acclimation effects affect the following generation? In Drosophila melanogaster, parental developmental temperature influenced male territorial success (Zamudio, Huey & Crill 1995). In addition, Watson & Hoffmann (1995, 1996) showed that acclimation in D. melanogaster prior to selection for cold resistance altered selection responses in progeny, and that maternal and paternal acclimation influenced resistance in opposing directions. Therefore acclimation effects have the ability to extend across generations.

Cold stress is of particular relevance to D. serrata. Species comparisons show D. serrata has a greater susceptibility to cold than other Drosophila species with a more southerly (= cooler) Australian distribution (Jenkins & Hoffmann 1999). Seasonal variation for cold resistance among populations indicates cold selection for increased resistance at the southern border (Jenkins & Hoffmann 1999; Magiafoglou, Carew & Hoffmann 2002). Winter trials with field cages placed at or below the southern border of D. serrata suggest that cold temperature shock may also impact on a variety of traits such as male mating success, early and lifetime productivity, fecundity and viability (Jenkins & Hoffmann 1999; A. Magiafoglou unpublished data). These are direct effects of cold shock within a generation, and additional cross-generation effects could influence the ability of populations to recover in favourable spring conditions.

The following questions are considered here: (1) Are there cross-generation effects of cold shock at the maternal and/or grandmaternal stages on progeny fitness? (2) Do paternal and/or grandpaternal effects on progeny fitness also occur? (3) Is there an interaction among generations for environmental effects of cold shock? Field studies have shown that winter field conditions may be experienced by one to two generations of flies, and it is possible that effects of cold shock may be compounded across parental and grandparental generations. (4) Do cross-generation effects depend on genetic background? Lines derived from two geographical regions and with a different history of culture in the laboratory were considered.

Materials and methods

stocks and maintenance

Drosophila serrata stocks were initiated using field females collected from two localities in April 1998: Wollongong (34°40′ S, 150°56′ E) at the southern species border, and Coffs Harbour (30°01′ S, 153°10′ E), located approximately 600 km north of the southern border. To establish mass bred populations for each locale, equal numbers of progeny from 25 inseminated field females were combined. These populations, termed Coffs Harbour and Wollongong, were held at 25 °C for 20 generations. At this time replicate groups were created for each population, and held under two separate selection schedules for approximately 2·5 years.

selection background

19 °C Selection

The first selection schedule represents typical Drosophila laboratory maintenance techniques. Populations (Coffs Harbour and Wollongong) were cultured at 19 °C constant light, and maintained in a minimum of four replicate 200-ml culture bottles, with each bottle housing approximately 150 flies. Adult flies were transferred to fresh bottles every three days until about 14 days of age, at which time adults were discarded. Emerging adult progeny were combined then collected over 2–3 days from the replicate bottles (for one or two of the transfer periods). Adults were randomly distributed across bottles of fresh culture media to initiate the next generation. Prior to the present study, stocks from 19 °C had undergone about 40 generations of culture.

7–18 °C SELECTION

The second schedule involved conditions designed to reflect average minimum and maximum temperatures and photoperiod parameters that D. serrata individuals experience at the southern species border over the coldest winter month (Australian Bureau of Meteorology –www.bom.gov.au). Temperature fluctuated daily from 7 °C (12 h) to 18 °C (12 h), with a photoperiod of 10L : 14D. Adults were also subjected to a non-lethal cold shock to reflect the natural occurrence of occasional short periods at 0 °C. Flies from both populations (Coffs Harbour and Wollongong) were exposed to this environment, with each population comprising two replicate lines that were held independently over the course of the selection period. Emerging adults were collected over 5–6 days and combined before being distributed across four 200-ml culture bottles at densities of approximately 150 flies per bottle. Adults were then held at 19 °C for 1–2 days prior to the cold shock, after which they were transferred to empty 200-ml bottles which were immersed in a 20% ethylene glycol solution cooled to 0 °C. After 1 h, bottles were removed and placed at 19 °C. Adults were subsequently transferred to fresh culture bottles until the same time on the following day when the cold shock treatment was repeated. After the cold shock treatment, flies were transferred to fresh bottles every 3–4 days until they were about 14 days posteclosion, at which time they were discarded. After each transfer, bottles containing eggs and some larvae were transferred to the fluctuating temperature environment. Populations from the fluctuating environment had only been cultured for 21 generations under these conditions because of the slower generation time compared with the 19 °C treatment.

experimental cross-generation treatments

To investigate cross-generation effects of cold shock on progeny fitness, all stocks were reared at 19 °C for two generations. Flies used to initiate the parental and grandparental generations were collected as virgins over a 2-day period. The following day, flies were sexed using CO2 and groups of 25 single sex individuals were held in 40-ml vials containing culture medium. When flies were 7–8 days old, they were exposed to the cold shock treatment (see Fig. 1 for experimental treatment outline).

Figure 1.

Experimental design testing for cross-generation effects. CS denotes individuals exposed to cold shock. All individuals exposed to cold shock were mated with control flies from the same strain.

The cold shock involved exposure to 0 °C for 3 h. This was the maximum period stocks could be held at this temperature without mortality. Four replicates of 25 individuals (single sex) held in 40-ml glass vials without food were immersed into 20% ethylene glycol cooled to 0 °C for the duration of the cold shock. To allow recovery of cold shocked adults, crosses were set up at least 18 h after the cold shock. In all cases, crosses involved stressed and control flies of the same population. For each cross (see Fig. 1), two replicate bottles were set up, each containing 50 pairs of males and females. Adults oviposited in these bottles for 5 days at 19 °C (constant light) and were left under these conditions for development.

traits

To determine the influence of cross-generation effects on progeny fitness, development time, viability, productivity and size were considered. To score viability and development time, two replicate bottles of parental adults (each with 50 pairs of males and females) from each treatment group were set up on spoons containing treacle–semolina–yeast–agar medium to stimulate oviposition. To ensure that virgin flies were fertile, flies were crossed 18 h after cold shock treatment and 6 h prior to being placed on medium. Mating was frequently observed within the 6-h period. Eggs were collected over an 18-h period and transferred to vials each containing 15 ml of medium, 10 eggs per vial and four vials per treatment. Vials were then randomly arranged in a container and held in a 19 °C cabinet. The position of tins in the cabinet was altered every second day to counter minor temperature variation within the cabinet. To determine development time, flies emerging from vials were collected every 8 h. Vials were scored until no new adults emerged for >48 h.

Productivity was measured on emerged offspring. Flies were collected as virgins over 2 days and sexed under CO2. Eight replicate vials per treatment and population were set up, each vial containing two females and four males. Maternal and grandmaternal effects were only tested on female productivity, and paternal/grandpaternal effects were only tested on male productivity. For female productivity, female progeny from the treatments were crossed with males from the same population whose parents/grandparents had not been shocked. For male productivity, male progeny from treatments were crossed with females derived from unshocked parents/grandparents from the same population. Adults were left in vials for 4 days and productivity was measured as the total number of offspring emerging from the vials.

Wings of individuals from treatment groups were mounted on slides to measure adult wing length. Images of wings were captured using a Wild M38 microscope attached to a Panasonic WV-GP460 digital camera. Wing length was determined by placing landmarks at the intersection of the anterior cross vein with the third longitudinal vein, and at the wing tip of the third longitudinal vein, using tpsDig version 1·2 written by F. James Rohlf. This program expresses landmarks as x and y coordinates in a Cartesian space, from which wing length was calculated.

analysis of quantitative data

Outliers were detected using the Grubbs test through GraphPad Prism version 3·02 for Windows (GraphPad 2000). Grubbs test determines the Z ratio as the difference between the outlier and mean divided by the standard deviation of the sample. To test for deviations from normality, the Kolmogorov–Smirnov statistic was computed. If normality was not detected after outliers were removed, data were transformed prior to analysis. Before anovas were run, data were tested for equality of variances through the Scheffé-Box test (Sokal & Rohlf 1995). All analyses were conducted using SPSS V10·0 for Windows (SPSS 1999).

Female development time and viability data showed skewed distributions and transformation failed to overcome this problem. Despite this, anovas were still undertaken as the validity of these analyses are not greatly affected by skewed distributions (Zar 1996), and variances among groups were homoscedastic. Nevertheless we did also undertake non-parametric analyses of the data (Kruskal–Wallis tests) to verify any significant differences among groups (data not presented).

Differences among populations, selection background and cross-generation treatment were tested using anovas. Three-way anovas were run for each trait, with population, selection and treatment as the main factors. To investigate homogeneous subsets of the main factors (population, selection and treatment), Tukey B post hoc tests were performed. Strains were pooled within population for the 7–18 °C stocks as differences among strains were small and only demonstrated for male development time (Wollongong population, strains one and two, one-way anova, F1,71 = 4·948, P = 0·03).

Because development time can be confounded by viability, analyses of covariance were used to correct for differences in viability among vials. Where significant associations due to covariates were evident, results are presented with the covariate included, although usually inclusion of a covariate did not alter patterns in the results.

Results

female development time

Population, selection and treatment differences were evident in all anovas, and there was an interaction between selection and population, but interactions involving the treatment term were not significant (Table 1). Because interactions between treatment and selection/population were not significant for this trait (or any other trait), only the average effect of the treatments are presented, while population effects are presented separately for the two backgrounds. Among populations, Coffs Harbour flies developed faster than flies from Wollongong, while 19 °C selection resulted in faster development to flies selected in the fluctuating environment (7–18 °C) (Fig. 2a). Significant cross-generation treatment differences were evident with maternal, grandmaternal treatments reducing development time, while paternal and grandpaternal treatments acted to increase it (Fig. 2b). An interaction between selection background and population was evident, suggesting population-specific adaptation to differing laboratory maintenance schedules for this trait. Viability as a covariate in this analyses was significant (Table 1) suggesting variation in female development time is partly influenced by viability. Despite this, no significant correlation between these two measures was evident (Spearman's correlation rs = 0·061, N = 223, P = 0·364).

Table 1. anovas examining differences among populations, selection histories and cross-generation treatments. Mean squares (MS), degrees of freedom (df) and probablilities (P) reported
 Development timeViabilityProductivity
FemaleMale  FemaleMale
MS (d.f.)PMS (d.f.)PMS (d.f.)PMS (d.f.)PMS (d.f.)P
  • *

    Viability as covariate.

Covariate* 350·833 (1) 0·039 824·614 (1)<0·001
Population1938·293 (1)<0·0013066·567 (1)<0·001194·671 (1)0·307   62·745 (1) 0·794   764·300 (1) 0·321
Selection3402·769 (1)<0·0015118·526 (1)<0·001237·661 (1)0·26016 556·862 (1)<0·00115 882·274 (1)<0·001
Treatment 357·693 (9)<0·001 216·091 (9) 0·002425·742 (9)0·01812 478·592 (3)<0·001 4 236·885 (3) 0·001
Population × Selection 405·927 (1) 0·026 135·436 (1) 0·171107·294 (1)0·448 2 924·055 (1) 0·076   405·254 (1) 0·469
Population × Treatment  52·269 (9) 0·756 120·973 (9) 0·095 44·398 (9)0·988 1 772·214 (3) 0·126   720·325 (3) 0·425
Selection × Treatment  96·401 (9) 0·302  80·528 (9) 0·349142·7614 (9)0·646  464·550 (3) 0·678 1 004·625 (3) 0·275
Population × Selection × Treatment 145·945 (9) 0·070  96·544 (9) 0·216195·088 (9)0·402 1 517·286 (3) 0·178   149·431 (3) 0·900
Error  80·829 (182)   71·723 (182) 185·918 (198)    916·138 (172)    770·185 (174) 
Figure 2.

Female development time. (a) Comparison of Coffs Harbour and Wollongong populations selected under the 7–18 °C and 19 °C regimes. Treatment effects are pooled within population and selection background. (b) Comparison of cross-generation treatments. Data pooled across population and selective background. Error bars represent ± 1 standard error.

male development time

Differences at the population, selection and treatment level were evident for male development time (Table 1). Similar to female development time, males from Coffs Harbour developed faster than those from Wollongong, which may be attributable to initial geographical differences among populations. In addition, stocks selected in the 19 °C environment demonstrated a faster development than those from 7 to 18 °C (Fig. 3a). Cross-generation treatment differences were significant (Table 1), mostly because of differences between maternal and combined grandmaternal–grandpaternal treatments (Fig. 3b). Maternal cold exposure tended to decrease male development time while grandparental exposure acted to increase this trait. Viability was included as a covariate in the analysis of male development time, as there was an increase in development time with decreasing egg-to-adult survival (Spearman's correlation rs = 0·133, N = 223, P = 0·048).

Figure 3.

Male development time. (a) Comparison of Coffs Harbour and Wollongong populations selected under the 7–18 °C and 19 °C regimes. Treatment effects are pooled within population and selection background. (b) Comparison of cross-generation treatments. Data pooled across population and selective background. Error bars represent ± 1 standard error.

viability

Significant differences among populations or selection background were not evident for viability, although cross-generation treatment effects were significant for this trait (Table 1, Fig. 4). Generally, viability fitness declined if flies of either sex were exposed to cold in the parental generation, regardless of the cross-generation treatment. Cross-generation effects of cold on viability therefore do not persist across generations.

Figure 4.

Cross-generation treatment effects of cold shock exposure on viability. Population and selection background were pooled within treatment. Untransformed data presented. Error bars represent ± 1 standard error.

productivity

For female productivity, there was no population effect, although selection background and treatment effects were evident (Table 1). Generally, populations from the fluctuating selection environment showed a reduced productivity compared with those from the 19 °C environment (Fig. 5a). Maternal cold shock increased productivity, while grandmaternal cold shock decreased it (Fig. 5b). Exposure to cold in the maternal generation therefore had a positive effect on offspring productivity, but exposure a further generation removed was detrimental. Size has the potential to influence productivity measurements, with larger female D. serrata producing greater numbers of offspring (Hercus & Hoffmann 1999). However, size did not influence the current results as the correlation coefficient between size (wing length) and productivity at the treatment level was not significant (Pearson correlation r = −0·103, N= 16 treatments, P = 0·704). Female size did not differ among populations, but significant differences were demonstrated among selection regimes and treatments (F1,101 = 9·543, P = 0·003; F1,101 = 2·876, P = 0·040). An interaction among selection and population was evident for female size (F1,101 = 8·520, P = 0·004), suggesting initial geographical differences or population-specific adaptation to laboratory maintenance schedules may have influenced size.

Figure 5.

Comparisons of female productivity. For female productivity, only maternal, grandmaternal and combined maternal, grandmaternal effects were tested. (a) Comparison of Coffs Harbour and Wollongong populations selected under the 7–18 °C and 19 °C regimes. Treatment effects are pooled within population and selection background. (b) Comparison of cross-generation treatments. Data pooled across population and selective background. Error bars represent ± 1 standard error.

Male productivity was not influenced by population, although both selection and treatment were significant (Table 1). As effects on productivity were separately tested for each sex, it is difficult to compare patterns across sexes. Despite this, the ranking of selection regime is similar to that for female productivity; flies maintained in the fluctuating environment showed a lower productivity than those from the 19 °C environment (Fig. 6a). Generally, exposure to cold in the paternal or grandpaternal generations reduced the productivity of male progeny (Fig. 6b). Again size was considered for its potential influence on productivity. Unlike in females, size was negatively correlated with male productivity (Pearson correlation r = −0·646, N = 16 treatments, P = 0·007), with larger males producing relatively fewer offspring. In an anova on male size, selection and the interaction between selection and population were significant (F1,104 = 16·220, P < 0·001; F1,104 = 13·127, P < 0·001, respectively).

Figure 6.

Comparisons of male productivity. For male productivity, only paternal, grandpaternal and combined paternal, grandpaternal effects were tested. (a) Comparison of Coffs Harbour and Wollongong populations selected under the 7–18 °C and 19 °C regimes. Treatment effects are pooled within population and selection background. (b) Comparison of cross-generation treatments. Data pooled across population and selective background. Error bars represent ± 1 standard error.

Discussion

The results indicate that cold shock can have carry-over effects that may influence the population dynamics of D. serrata. For instance, cross-generation treatment effects were demonstrated for development time, viability and early age productivity. Moreover, the expression of carry-over effects and outcome on fitness may depend both on the sex and generation exposed to cold shock. These responses suggest that environmental variation experienced in preceding generations can influence progeny phenotype in a manner that is complex and difficult to predict.

Progeny viability in this study was negatively influenced by parental carry-over effects, but this failed to persist across generations. Maternal cold exposure also results in reduced progeny viabilities in D. melanogaster (Watson & Hoffmann 1996). As egg-to-adult viability incorporates multiple developmental stages including egg hatch, larval and pupal viability, it is possible that cold specifically influences one or more stages of development. Viability in this study was scored as overall egg to adult survival, hence it was not possible to determine the stage of development affected. Maternal age in D. serrata has been shown to affect several preadult developmental stages including both egg hatch and larval to adult viability, while grandmaternal effects only influenced egg hatch (Hercus & Hoffmann 2000).

The productivity of females showed opposing responses that were dependent on the generation exposed to cold shock. A maternal treatment increased productivity while grandmaternal exposure decreased this trait. As a fitness measure, productivity is a rather complex trait, incorporating a number of life-history components that can affect productivity output. Body size, reproductive ability (including sperm and egg quality), fecundity and developmental viability may all impact on this measure. Variation in offspring productivity may therefore reflect any of these traits. Size effects appear unimportant as no association between productivity and female size was evident in the analysis. Cold exposure can have an immediate effect on fecundity in insects (Rinehart, Yocum & Denlinger 2000; Broufas & Koveos 2001), but it is unclear how maternal cold exposure may affect fecundity levels in the following generation. The increase in offspring productivity following maternal cold shock is in the opposite direction to that observed in selection responses, where increased cold resistance is associated with decreased early fecundity (Watson & Hoffmann 1996).

Paternal and grandpaternal exposure to cold consistently and negatively influenced male productivity. While larger females may often be associated with a higher fecundity, it is evidently not the case for males here, because larger males produced fewer offspring. A cost of male size on female egg production has been noted in D. melanogaster (Pitnick 1991; Pitnick & Garcia-Gonzalez 2002) and appears important here, although other factors such as sperm quality may also be involved. In D. melanogaster paternal temperature environment may impact on offspring fecundity levels (Huey et al. 1995).

Treatment effects were evident for the development time of both sexes. Grandmaternal effects tended to decrease progeny development time while paternal and grandpaternal treatments tended to increase it. However, both population and selection background may influence the development of female progeny, as evident from the interaction between these factors. Suggestive evidence of carry-over effects in D. melanogaster and D. simulans following cold exposure was found by Watson & Hoffmann (1996). Cold selected lines, continuously selected, were relaxed for a generation leading to an increase in progeny development times. However, their results may have been confounded by an association between the selected trait and development time. A reduction in development time may be an artefact of the differences observed among viability measures (Watson & Hoffmann 1996). To test if viability influenced carry-over effects for development times, this measure was included as a covariate in the analyses but did not influence the conclusions.

The relative influence of geographical origin and selective background on the life-history traits was tested in this study. Population effects reflecting geographical origin were limited to development time, while selective background effects were demonstrated for most traits. The influence of these factors had on the expression of cross-generation effects was negligible, as interactions between these factors and treatment were not evident. While other studies have demonstrated that genetic factors can shape the extent to which cross-generation effects are expressed in progeny (Van Hinsberg 1998; Wulff et al. 1999), this was not the case here.

Cross-generation effects of temperature treatments have been previously described in Drosophila. In particular, ambient temperature studies show how early fecundity, male territorial success and offspring fitness, measured as per capita rate of increase, can be influenced by parental temperature (Huey et al. 1995; Zamudio et al. 1995; Gilchrist & Huey 2001). Morphological traits can be influenced by parental temperature, as can offspring heat resistance (Crill, Huey & Gilchrist 1996). Therefore, the expression of cross-generation effects in response to parental temperature may be more widespread than previously thought.

The adaptive significance of cross-generation effects is debatable. Fitness consequences of cross-generation effects may depend on the type of environmental effect that elicits the response and/or the environment in which the response is tested (Rossiter 1996). For instance, if organisms are exposed to predictable changes in the environment, cross-generation effects of parental environment may act to enhance progeny fitness, as in the case for diapause induction. With respect to this investigation and the known ecology of D. serrata, exposure to cold shock is likely to occur toward the species southern border over the winter period. Moreover, field investigations across the southern distribution suggest flies tend to overwinter as adults and may undergo quiescence, where females are able to resume reproduction immediately after conditions improve (Jenkins & Hoffmann 1999; A. Magiafoglou unpublished data). The consequences of cross-generation effects of cold in this species are therefore likely to impact on the early springtime generation. In light of this, an increase in female progeny productivity after maternal cold exposure may be considered an adaptive response, particularly if coupled with faster development, although paternal effects may negate this.

In conclusion, the experiments indicate that parental and, to a lesser extent, grandparental carry-over effects from cold exposure in D. serrata can impact progeny fitness. Age effects may also be important in this species (Hercus & Hoffmann 2000), particularly as flies appear to overwinter as adults. This indicates that environmental changes as well as age effects can generate carry-over effects in D. serrata. Cross-generation effects therefore have the potential to be important from an ecological perspective. They can also be important in the evolution of life-history traits in response to stressful conditions as they have the potential to push selection in opposing directions to what would be expected through genetic inheritance alone (Kirkpatrick & Lande 1989).

Acknowledgements

We thank Carla Sgrò for experimental advice, and both Michele Schiffer and Paul Umina for critical comments. We also thank two anonymous reviewers for their helpful and highly constructive comments. This work was supported by an Australian Postgraduate Research Scholarship and a grant from the Australian Research Council via their Special Research Centre scheme.

Ancillary