Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
1By randomly dividing adult females of the butterfly Bicyclus anynana, reared in a common environment, among high and low temperatures, it is demonstrated that oviposition temperature induces a plastic response in egg size. Females kept at a lower temperature laid significantly larger eggs than those ovipositing at a higher temperature.
2Cross-transferring the experimentally manipulated eggs between temperatures and investigating hatching success showed that a lower rearing temperature is more detrimental for the smaller eggs produced at a higher temperature than for the larger eggs produced at a lower temperature, supporting an adaptive explanation.
3However, when examining two potential mechanisms for an increased fitness of larger offspring (higher desiccation resistance of larger eggs and higher starvation resistance of larger hatchlings), no direct link between egg size and offspring fitness was found. Throughout, i.e. even under benign conditions, larger offspring had a higher fitness.
4Therefore, egg size should be viewed as a conveniently measurable proxy for the plastic responses induced by temperature, but caution is needed before implying that egg size per se is causal in influencing offspring traits.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Elaborating on these previous results, we set out here to disentangle some of the causal mechanisms promoting increased fitness of larger eggs by testing the following hypotheses. If the fitness advantages of larger eggs were causally related to their size, one would predict an enhanced ability to withstand food shortage for the larger hatchlings originating from larger eggs based on their presumably higher nutritional storage reserves. Furthermore, a higher hatching success of larger compared with smaller eggs, when exposed to low relative humidity (based on an improved volume–surface ratio reducing water loss and therefore enhancing desiccation resistance), is expected. We examine these predictions and also show that the plastic response of egg size to temperature is highly reproducible and reversible, and that the relationship between egg size and offspring performance differs among temperature environments.
Materials and methods
study organism and experimental population
Bicyclus anynana is a tropical, fruit-feeding butterfly distributed from southern Africa to Ethiopia (Larsen 1991). The species exhibits striking phenotypic plasticity (two seasonal morphs), which is thought to function as an adaptation to alternative wet–dry seasonal environments and the associated changes in resting background and predation (Brakefield 1997). A laboratory stock population of B. anynana was established at Leiden University in 1988 from over 80 gravid females collected at a single locality in Malawi. Butterflies from the stock population were used for this study.
All butterflies used in this study were reared for at least two generations in a climate room at 27 °C, high humidity and a photoperiod of 12L : 12D. Larvae were fed on young maize plants. Following adult eclosion, males and females were kept apart for 2 days, and afterwards put together for mating for 3 days. After the mating period, females were randomly divided between two cages containing host plants for oviposition. One cage remained at the larval rearing temperature of 27 °C, whereas the other was transferred to 20 °C (high humidity and 12L : 12D throughout) to induce a plastic increase in egg size. The oviposition plants within the cages were replaced daily. The temperatures used are similar to those at which the larvae of the wet and dry seasonal forms, respectively, develop in the field (Brakefield & Mazzotta 1995). Throughout the experiments, butterflies had access to moist banana for adult feeding. As the eggs of B. anynana are nearly perfectly spherical, egg size was measured as cross-sectional area (mm2), using a digital camera (Leica DC200) connected to a binocular microscope. The resulting images were analysed using Scion Image public software (Scion Corporation 2000). The mean of about 10 eggs was used to calculate egg size for individual females (for details see Fischer, Zwaan & Brakefield 2002; Fischer et al. 2003).
Experiment 1: egg hatching success at different temperatures
Butterflies for this experiment were reared in two replicate populations at 27 °C, each of which was divided among oviposition temperatures. Sixty-six females were used per replicate and oviposition temperature (i.e. 264 in total). On days 0 (i.e. after assigning females to oviposition temperature groups, but before they were actually transferred), 7, 11, 13 and 22 following the division among temperatures, egg samples were carefully removed from the oviposition plants and subsamples were measured. To analyse hatching success at 20 °C and 27 °C, eggs collected on days 0 and 12/13 were used. We established 10 replicate Petri dishes with about 30 (min. 20, max. 40) eggs each per treatment group and egg-rearing temperature. Hatchlings were counted and removed daily until no more hatchings occurred. On day 13 (after the eggs for this experiment had been collected), the females of the second replicate population were cross-transferred between oviposition temperatures to test whether the plastic response of egg size to temperature is reversible, while the other replicate remained at its original temperature.
Experiment 2: starvation resistance of hatchlings
For this experiment the replicate population from Experiment 1 that was not cross-transferred was used. Eggs were collected on days 17–22 of the oviposition period and subsequently stored in plastic pots at 27 °C. The pots were checked daily early in the morning (i.e. soon after hatching) for hatchlings, which were randomly divided between two treatments. There was no indication of a difference in development time between large and small eggs in an earlier experiment (K. Fischer et al., unpublished data). The hatchlings were either transferred to Petri dishes containing fresh maize cuttings (control), or starved for 24 ± 2 h (i.e. maize cuttings were added to the dish the following day). Ten hatchlings per Petri dish and 30 replicates per treatment were used. Maize cuttings were replaced daily. Once we had enough hatchlings for this part of the experiment, we performed a small additional experiment using a similar set up. However, hatchlings were starved for 48 ± 2 h instead of 24 h. Twelve replicates but no controls were used per oviposition temperature group. The number of surviving larvae was scored on day 4 of the experiment, a time period sufficient to assess which larvae were able to establish successfully on the maize cuttings and which ones had died.
Experiment 3: egg hatching success at different humidities
For this experiment we used the same B. anynana stock population as before, but a few generations later. Butterflies were reared at 27 °C and divided between oviposition temperatures as outlined above. Fifty mated females were used per oviposition temperature. Egg samples were collected and measured on days 0, 8, 11 and 14 following the division between temperatures. To analyse hatching success at different relative humidities, eggs collected on days 11–13 were used. Eggs from both oviposition temperature groups were kept at 27 °C, where they were randomly divided among three relative humidity treatments. To ensure large differences in relative humidity, eggs were placed on Petri dish bottoms, which were placed into tightly sealed plastic pots (500 ml). Depending on treatment, these pots contained a c. 1 cm layer of silica gel (dry), a layer of soaking wet paper towels (wet) or nothing (control). Ten replicates with 10 eggs each were used per treatment. Hatchlings were counted and removed daily until no more hatchlings were found.
The plastic response of egg size to temperature was analysed by nested repeated measurements anovas, with oviposition temperature and time (i.e. day within oviposition period) as factors and replicates (if applicable) nested within oviposition temperature. Data on egg hatchability and larval survival were analysed using nominal logistic regressions on binary data with oviposition temperature and treatment (e.g. rearing temperature, relative humidity) as factors. If applicable, replicates were again nested within oviposition temperature. Throughout all means are given ± 1 SE.
temperature-mediated plasticity in egg size
Egg size showed a clear plastic response to oviposition temperature in both generations of the B. anynana population under investigation. A repeated measurements anova on days 0, 7, 11, 13 of the previous generation confirmed significant effects of oviposition temperature (F1,343 = 202·4, P < 0·0001), time (F3,341 = 4·9, P = 0·0025) and the interaction between the two (F3,341 = 49·4, P < 0·0001), with replicates behaving in a similar, though not exactly the same, way (F2,343 = 5·7, P < 0·0038; Fig. 1). For the replicates remaining at their original oviposition temperature, egg size still differed substantially among groups on day 22 (i.e. after all eggs used in this study had been collected; 20 °C: 0·737 ± 0·008 mm2, n = 67 eggs; 27 °C: 0·627 ± 0·006 mm2, n = 61). This difference in area is equivalent to 27·4% in volume. The replicates switched between temperatures on day 13 of the oviposition period showed a clear response in the expected direction to their new environment on day 22, indicating that the plastic response to temperature is reversible (Fig. 1).
The data gained from the later generation showed a closely similar plastic response (repeated measurements anova: oviposition temperature F1,158 = 209·2, P < 0·0001; time F3,156 = 15·7, P < 0·0001; interaction term F3,156 = 33·2, P < 0·0001). Eventually (i.e. on day 14 of the oviposition period), the difference between oviposition temperature groups amounted to 22·8% in volume (20 °C: 0·759 ± 0·005 mm2, n = 99 eggs; 27 °C: 0·662 ± 0·004 mm2, n = 100).
egg hatching success at different temperatures
On the day of division among temperatures, egg sizes did not differ substantially between oviposition temperature groups (Fig. 1). Nevertheless, a nominal logistic regression revealed a significant (though fairly small) difference in egg hatchability between oviposition temperature groups ( = 13·4, P = 0·0003; Fig. 2a), while effects of egg-rearing temperature ( = 2·2, P = 0·14), of replicates ( = 2·1, P = 0·36) and the interaction between oviposition and egg-rearing temperature were non-significant ( = 1·1, P = 0·30). At both temperatures, the groups subsequently remaining at 27 °C had a higher hatching success than the ones transferred to 20 °C (c. 82·5–73·5% at 20 °C, c. 83·0 compared to 79·0% at 27 °C; Fig. 2a).
After an acclimation period of 12 days, however, when oviposition temperature groups differed substantially in egg size (Fig. 1), all factors affected egg size significantly. Hatching success was on average slightly higher at a rearing temperature of 27 °C ( = 17·8, P < 0·0001; Fig. 2b), and was higher for the large eggs produced at the lower temperature ( = 9·3, P = 0·0023). The latter difference was mainly due to a considerably higher hatching success of these eggs at a rearing temperature of 20 °C as compared with the small eggs produced at the higher temperature, resulting in a highly significant interaction between oviposition and egg-rearing temperature ( = 29·6, P < 0·0001; Fig. 2b). Differences among replicates were small, though significant ( = 7·5, P = 0·0239).
In order to control for initial differences in hatching success among groups, we set the initial values from before temperature change to 100% and present the later results relative to this value. In this way it becomes more obvious that at an egg-rearing temperature of 27 °C all groups had a somewhat lower (by 7–18%) hatching success in the second round (i.e. after an acclimation period of 12 days; Fig. 2c). This decrease seems to be slightly higher for the groups ovipositing at 27 °C, which show at this stage, in contrast to the initial situation (i.e. prior to temperature change), a marginally lower hatching success than the groups ovipositing at 20 °C. At the lower egg-rearing temperature of 20 °C, the small eggs originating from the higher oviposition temperature suffered a substantial reduction in hatching success (by c. 30% compared with the initial value), whereas the performance of the large eggs produced at the lower temperature remained similar or was even slightly improved (Fig. 2c).
starvation resistance of hatchlings
As before, the eggs used for this experiment differed substantially in size among oviposition temperature groups (Fig. 1). Here, the replicates that were not cross-transferred among temperatures were used (see Materials and methods). Regardless of treatment (i.e. starved vs control), the hatchlings originating from the larger eggs produced at 20 °C had a higher probability of survival than those from the smaller eggs produced at 27 °C ( = 20·3, P < 0·0001; Fig. 2d). A starvation period of 1 day prior to first feeding reduced survivorship in both groups substantially and to a similar extent ( = 119·9, P < 0·0001; interaction term = 0·5, P = 0·47). When hatchlings were starved for 2 days, none of those originating from the small eggs produced at 27 °C survived, whereas at least 8·3% (i.e. 10 out of 120) of those originating from the large eggs produced at 20 °C survived (no statistics can be given because of zero values only in the 27 °C group).
egg hatching success at different humidities
This experiment used butterflies from the same B. anynana stock population as before, but a few generations later. Again, the experiment was carried out when eggs differed considerably in size (see above). Across all three humidity treatments, the larger eggs produced at 20 °C had a higher hatching success compared with the smaller ones produced at 27 °C ( = 26·7, P < 0·0001; Fig. 2e). There was a clear trend towards a lower hatching success under dryer conditions, but this was not significant ( = 5·3, P = 0·0694). Both oviposition temperature groups were affected by the different treatments in a comparable way (interaction term = 1·8, P = 0·40).
As expected, temperature induced a plastic response in egg size in both generations of B. anynana studied. The females kept as adults at a lower temperature laid significantly larger eggs than those kept at a higher temperature (Fig. 1), confirming earlier results on B. anynana (Fischer et al. 2003) and other insects (e.g. Avelar 1993; Huey et al. 1995; Ernsting & Isaaks 1997; Blanckenhorn 2000). Since all animals in our study were kept in the same environment as larvae and young adults (during premating and mating), the divergence in egg size is clearly and directly attributable to differences in oviposition temperature, but not to temperature-related effects on body size or physiology during development. This is confirmed by the fact that the plastic increase in egg size at the lower temperature is reversible by switching females between temperatures (Fig. 1; see also Fischer et al. 2003; Blanckenhorn 2000), although the group transferred from 27 to 20 °C did not reach the egg size of those females kept continuously at 20 °C. However, this latter observation may indicate resource depletion due to the advanced female age (see Giron & Casas 2003). Thus, the phenomenon we are investigating is evidently acclimation in the strict sense, defined as a reversible, facultative response to changes in a single environmental variable in the adult stage (Willmer, Stone & Johnston 2000; Wilson & Franklin 2002).
To test for any differences in offspring performance associated with the plastic changes to the phenotype, we cross-transferred egg samples from both oviposition temperature groups between high and low temperature. Unexpectedly, the egg samples collected just before the females were divided among temperatures showed significant differences in hatching success. At both egg-rearing temperatures, the eggs from the replicates subsequently remaining at 27 °C had a slightly higher survival probability than those from the groups subsequently transferred to 20 °C. As we have no explanation for this, we assume it to be a chance effect of allocation to treatments.
After an acclimation period of 12 days, when oviposition temperature groups differed substantially in egg size, the larger eggs produced at a lower temperature were more likely to hatch than the small ones produced at a higher temperature. Although hatching success was similar in both groups at the higher rearing temperature, the larger eggs from 20 °C had a clearly better performance at the lower rearing temperature than the small ones from 27 °C. The latter had a clearly reduced hatching success when reared at 20 °C as compared with 27 °C, whereas the eggs from 20 °C performed equally well at both rearing temperatures or even slightly better at 20 °C. This pattern becomes particularly clear when data were corrected for initial differences in hatching success among groups (Fig. 2c). The resulting interaction between oviposition temperature and egg-rearing temperature confirmed that the lower temperature was more detrimental for smaller than for larger eggs, i.e. the relationship between egg size and offspring performance differed between environments. Interestingly, a recent study on Drosophila yielded comparable results regarding body size, with the fitness benefits of large males being significantly larger at low temperatures (Reeve et al. 2000).
Our findings support the notion that temperature-mediated plasticity in egg size (or some other covarying factors, see below) may be an adaptation to the prevailing temperature conditions in B. anynana, and therefore support the beneficial acclimation hypothesis. Our previous experiments (Fischer et al. 2003) did not reveal such an interaction, probably because they involved a much smaller data set with limited statistical power. In line with our previous results (Fischer et al. 2003), hatching success was, overall, slightly higher at a rearing temperature of 27 °C, suggesting that a constant temperature of 20 °C falls below the optimum for growth and early development in this tropical butterfly (cf. Brakefield & Reitsma 1991).
In summary, the above results support our earlier interpretations (Fischer et al. 2003). Based on differential survival probabilities for small and large offspring at a lower temperature, it seems to pay to produce fewer but larger eggs at a low temperature, resulting in increased offspring performance. At a higher average temperature, however, which offers a high chance of survival regardless of the investment per offspring, highest returns should be gained by producing as many small eggs as possible. Consequently, females would be expected to adjust their egg size to the temperature experienced during oviposition, and when this temperature provides a predictable cue for the environmental conditions experienced by the offspring in early life.
In assessing these results, it should be borne in mind that egg size per se is not necessarily the cause of the differences in offspring performance, as other traits such as nutrient composition may covary with egg size and temperature. Therefore, we tested two mechanisms that could potentially promote increased fitness of larger offspring. First, we examined whether hatchlings originating from larger eggs have an enhanced ability to withstand food shortage. Our previous results showed that the hatchlings originating from larger eggs had a higher probability of reaching adulthood, perhaps because of a higher starvation resistance and/or ability to establish themselves on the food plant (Fischer et al. 2003). A higher starvation resistance of hatchlings originating from larger eggs provides a straightforward explanation based on a tight correlation between egg size and hatchling size in B. anynana (and other Lepidoptera; see Fischer et al. 2002; Fischer et al. 2003), and the presumably higher storage reserves of larger hatchlings (Azevedo, French & Partridge 1997).
As expected, the hatchlings originating from larger eggs in our study had a higher survival probability than those from smaller eggs. However, although a starvation period of 1 day reduced survival probabilities substantially, hatchlings from large and small eggs were affected to the same extent. Ideally, one would expect a similar performance under benign conditions (where storage reserves are less important), but a pronounced difference under food stress. The fact that the performance of the hatchlings originating from larger eggs was enhanced throughout (i.e. in controls and under starvation) may suggest that the ability of a hatching larva to establish itself on the food plant is even more important than periods of food shortage (see Nakasuji & Kimura 1984; Nakasuji 1987; Braby 1994). On the other hand, our preliminary data using a longer starvation period of 2 days do suggest that hatchling size is important, as none of the small hatchlings, but at least a few of the large ones, survived.
Second, we tested whether the higher hatching success of larger eggs is directly related to their size. Based on an enhanced volume–surface ratio reducing water loss and thereby presumably increasing desiccation resistance (see also Kennington et al. 2003), we predicted that large eggs would perform better than small eggs under dry conditions. However, as before, larger eggs showed a higher hatching success throughout (i.e. in all treatments), and the difference between large and small eggs was not larger at a low relative humidity. Surprisingly, treatment did not significantly affect hatching success, although there was a clear trend in the predicted direction. These data do not support the hypothesis that the increased hatching success of the eggs produced at a lower temperature is causally related to their larger size, or at least size seems not to be the only factor.
When interpreting the differences in offspring performance among oviposition temperature groups one has to consider potentially confounding effects of differences in physiological age. Recently, not only egg size but also egg energy content was shown to decline with age in a parasitoid (Giron & Casas 2003). However, for the following reasons we think that it is unlikely that differences in female physiological age influenced substantially any of the observed patterns. First, all individuals were reared at the same temperature, and subsequently also spent the initial part of their adult life span at the same temperature. Second, egg samples were taken before any age effects were evident (cf. Fischer et al. 2003). Third, female longevity does not differ significantly between 20 and 27 °C in B. anynana, suggesting that rates of ageing are similar across groups (Fischer et al. 2003). Fourth, even if one assumed that the slightly more marked decline in hatching success at 27 °C for the groups ovipositing at the higher temperature between the first and second measurement is related to age effects, this would not qualitatively change any of the observed patterns. Fifth, effects of increased maternal age on offspring fitness generally appear to be weak in insects (Karlsson & Wiklund 1984; Wiklund & Karlsson 1984; Fox & Dingle 1994).
In summary, our results show an acclimation response of egg size to the prevailing temperature during oviposition in B. anynana. As confirmed by the present and previous results (Fischer et al. 2003), the plastic response is highly reproducible and reversible. The evidence to date suggests that temperature-mediated plasticity in egg size may be adaptive in B. anynana, supporting the beneficial acclimation hypothesis. However, we were not able to reveal conclusive evidence that egg size per se is the causal and/or decisive factor promoting increased fitness of offspring produced at a lower temperature. Contrary to expectations, egg size did not interact with relative humidity during egg rearing, and hatchling size did not interact with starvation. Nevertheless, we did find some evidence that hatchling size is important for the establishment on the food plant and for starvation resistance, but the mechanisms causing a higher hatching success of the larger eggs produced at a lower temperature remain unclear. Based on our results it appears unlikely that there is a direct causal link between hatching success and egg size. Identifying the crucial factor influenced by temperature will be subject to future research. As a next step we will focus on the chemical composition of eggs, which has been found to differ among small and large eggs in a parasitoid (Giron & Casas 2003). For the time being, egg size should be viewed as a conveniently measurable proxy for the plastic responses induced by temperature, but it is necessary to be extremely careful in assuming that the plastic changes in egg size per se are causal in influencing offspring performance.
We thank Casper J. Breuker for writing a macro to automate egg measurements, and Niels Wurzer and Mariel Lavrijsen for the supply of maize plants. Two anonymous referees kindly provided constructive criticism. We acknowledge financial support from the German Research Council (DFG grants no. Fi 846/1–1 and Fi 846/1–2 to K.F.) and the Netherlands Organization for Scientific Research (NWO 811–34·005 to B.J.Z.).