Critical weight for the induction of pupariation in Drosophila ...

Critical weight for the induction of pupariation in Drosophila melanogaster : genetic and environmental variation G. H. DE MOED,* C. L. J. J. KRUITWAGEN, G. DE JONG* & W. SCHARLOO* *Population Genetics, Department of Plant Ecology and Evolutionary Biology, and Centre for Biostatistics, Utrecht University, Padualaan 14, NL-3584 CH Utrecht, the Netherlands

Keywords:

Abstract

critical weight; food level; genetic variation; insect metamorphosis; temperature.

Timing of puparium formation in Drosophila melanogaster is set by reaching a critical stage at which larvae attain the ability to pupariate. This critical stage is reached at a relatively constant size characterized by the mean critical weight, i.e. the weight at which 50% of surviving larvae pupate without further feeding. The mean critical weight might be affected by larval growth conditions. This hypothesis was tested by determining the mean critical weight in larvae raised at three temperatures and two food levels, for two isofemale lines from two populations. Pupariation probability is a function of larval weight. The two environmental variables affect pupariation probability and mean critical weight differently. Food level does not affect critical weight but affects weight-independent mortality; higher temperatures lead to a reduction of mean critical weight. Mean critical weight shows substantial differences between lines; the differences are maintained over temperatures. Genetic variation in mean critical weights has ecological and evolutionary implications.

Introduction The timing of maturation in insects is determined by reaching a critical developmental stage, at which development becomes committed to metamorphosis (e.g. Nijhout, 1981). The critical stage is a well-de®ned developmental stage, marked by a small ecdysone pulse (Berreur et al., 1979). It leads to a shift in hormonal levels, which sets the stage for further developmental changes (Riddiford, 1993). The critical stage occurs during the last larval stage, generally well before metamorphosis. Development after the critical stage is limited to a narrowly de®ned time period until the onset of pupation in holometabolous insects (Beadle et al., 1938; Bakker, 1959, 1961; Robertson, 1963; Nijhout & Williams, 1974) or adult ecdysis in hemimetabolous insects

Correspondence: G. de Jong, Population Genetics, Department of Plant Ecology and Evolutionary Biology, Utrecht University, Padualaan 8, NL-3584 CH Utrecht, The Netherlands. Tel: +31 30 2532246; fax: +31 30 2542219; e-mail: [email protected]

852

(Blakley & Goodner, 1978). The duration of this time period seems not to have been the subject of research. In a number of species, the critical stage is attained at a relatively constant size (Nijhout & Williams, 1974; Blakley & Goodner, 1978; Nijhout, 1979), indicating that the decision to enter the critical stage is based on some size-related character. Some species need an additional stimulus, like a minimal intake of nutrients (Wigglesworth, 1934; Nijhout, 1979; ZdaÂrek, 1985) or the presence of members of the opposite sex (Beck, 1971), before development will proceed. In Drosophila melanogaster, the critical stage occurs right after the second moult. After the critical stage, the developmental period is no longer affected by resource levels and can be completed in the absence of food (Beadle et al., 1938; Bakker, 1959; Robertson, 1963; Fourche, 1967). A reduction of growth rate before the critical stage increases the larval period, but does not affect adult size (Beadle et al., 1938; Robertson, 1963). This observation indicates that the critical stage coincides with the attainment of a relatively constant weight. The size of a larva at reaching the critical stage of

J. EVOL. BIOL. 12 (1999) 852±858 ã 1999 BLACKWELL SCIENCE LTD

Larval weight and pupariation in Drosophila

commitment to pupariation is referred to as its critical weight. The critical weight is a symptom of the underlying physiology and need not imply a direct relation between size and the decision to pupariate. Under a symmetric distribution of critical weights over larvae in a strain or population, the mean critical weight is the weight at which 50% of the larvae attain the ability to enter the next developmental stage. The mean critical weight has been regarded as the characteristic critical weight for a Drosophila strain (Bakker, 1959, 1961, 1969; Robertson, 1963; Royes & Robertson, 1964). Larval critical weight will partly determine the way age and size at maturity respond to environmental variation (Bernardo, 1993), and is therefore important in life history evolution. The larval developmental period is determined by the time needed to reach the critical weight and the time from the critical stage to pupation. Final body weight is determined by the critical weight and the possibilities for additional growth before the onset of pupariation, as determined by the availability of resources (e.g. Robertson, 1963). This basic model, ®rst proposed by Bakker (1959), seems to explain the widely observed reduction of body weight and increase in development time at reduced food levels in Drosophila (e.g. Robertson, 1960; Bakker, 1961; Gebhardt & Stearns, 1993). Increased temperatures during the preimaginal stages in Drosophila lead to a reduction in adult size (e.g. Ray, 1960; David et al., 1983; de Moed et al., 1997) and development time (e.g. David et al., 1983) and an increase in larval growth rate (e.g. Partridge et al., 1994; de Moed et al., 1998). The way growth and development interact in response to temperature, leading to smaller body size at higher temperature, is still unclear. The smaller adult size may result from a stronger increase in development rate as compared to growth rate (van der Have & de Jong, 1996), or from a lower critical weight, at higher temperatures. The main assumption of Bakker's model, the presence of a mean critical weight which is unaffected by growth conditions, has as not yet been tested. All direct measurements of the mean critical weight or pupariation fraction were performed under optimal food conditions (Bakker, 1961, 1969; Church & Robertson, 1966; Fourche, 1967; Partridge et al., 1994) and no studies estimate the mean critical weights as affected by temperature. In this study we tested the constancy of the mean critical weight over various temperatures and food levels. The effect of environmental conditions on the mean critical weight was determined on isofemale lines from two populations from different latitudes known to differ in adult size (Noach et al., 1996; de Moed et al., 1997). We tested whether these differences in size involve genetic differences in mean critical weight, and whether these genetic differences are maintained over environments.


853

Materials and methods Lines Fertilized females were collected at Montpellier, France (Fra), by the authors, and at Nijengzi, Tanzania (Tan), by Dr J. J. M. van Alphen of Leiden University, and used to start isofemale lines. These lines were cultured in replicate 50-mL bottles on a standard corn meal medium at 17.5 °C for 60 generations, at a minimal population size of about 500. Two of these isofemale lines were used from each population (Fra11, Fra24, Tan6, Tan71). Experimental conditions Growth conditions of the larvae were varied by changing food quality and ambient temperature. Vials were kept at either 17.5 °C, 22.5 °C or 27.5 °C at 60% relative humidity. Different food qualities were created by varying the concentration of dead yeast in the medium. The experimental media contained 20 g sucrose, 19 g agar, 1 mg nipagine and either 8 g or 32 g dead yeast powder in 1 litre tap water. Larvae were grown under axenic conditions to get reproducible and well-de®ned diets. Eggs were sterilized in a 3% hypochlorite solution for 15 min (Sang, 1956). Vials (diameter 3.5 cm) containing 10 mL media were inoculated with 150 5 eggs per vial. Mean critical weight for pupariation The mean critical weight for pupariation was estimated by determining the probability that a larva of known weight is able to pupariate when deprived of food. For each environmental condition and line, larvae were collected The larvae were collected at different ages during the 2nd and 3rd larval stage: at six ages at 27.5 °C, and at eight ages at 22.5 °C and 17.5 °C; ages were chosen to represent more or less comparable larval stages. Larvae of each age were collected from a separate vial; this results in 36±48 data points from which to estimate mean critical weight for each line and environmental condition. At each age larvae were collected and sorted by eye into six replicate groups; a group was composed of eight larvae with approximately equal body weights. Each group of larvae was weighed (Mettler ME22 microbalance), transferred to 2% agar medium to prevent dehydration, returned to their treatment temperature and scored for fraction pupariation. Weight variation within each group is expected to be small. The gradual increase in pupariation fraction with mean weight will therefore represent variation in critical weight between larvae. The mean critical weight will be found at 50% pupariation, in the absence of other mortality factors, if the distribution of critical weights is symmetrical over larvae.

854

G. H. DE MOED ET AL.

Analysis The relation between larval fresh weight and the fraction pupariation per group of eight larvae was analysed by generalized linear modelling, using maximum likelihood estimates. Because a fraction of even the largest larvae will not pupariate, we introduced an age- and weightindependent probability of pupariation (1 ) d). The mechanisms determining d are assumed to act directly before pupariation. Estimation of the models was done in two stages: one to determine the weight-independent probability of pupariation (1 ) d), and one to determine the weight-dependent probability of pupariation p, where larval weight was log-transformed to improve the ®t of the model (full model: ddev. 15.25). Any age-dependent mortality factors are inextricable from weight-dependent factors. The probability of pupariation p depending on larval weight W was modelled using a logit transformation (Finney, 1971): logp=1 ÿ p b0 b1 W and p p1 ÿ d where p is the expected pupariation fraction based on larval weight, d equals the fraction larvae unable to pupariate independent of weight and p* equals the overall pupariation fraction. The mean critical weight (W at p 0.5) was determined from the regression parameters, as W50 )b0/b1. The full model includes the in¯uence on fraction pupariation of Weight, the environmental treatments Temperature and Food Level, the genetical treatment Lines and all interactions between them. Treatment effects and their interactions were tested by removing terms consecutively from the model. As the full model showed no overdispersion there was no need for using a heterogeneity factor (see Table 1). The difference in deviance from consecutive models was tested against a chi-square distribution with degrees of freedom equal to the difference in number of parameters of the models. Analyses were performed using S-plus v3.3 for Windows.

Results The relation between larval weight and the fraction of larvae pupariating under different environmental conditions is shown in Fig. 1. Under all conditions, some fraction of even the largest larvae never pupariate. Under poor food conditions, the fraction pupariation in heavy larvae is lower than under rich food conditions. This effect was attributed to a difference in weightindependent mortality d. In the logit analysis, larval weight (W), food levels (F), temperatures (T) and lines (L) and all their interactions are used in a full model ®tted to all data. Weight is always retained in the model; other factors are consecutively removed from the full model ®tted to all data (Table 1). The logit analysis shows

Table 1 Logit regression analysis testing differences in the weightindependent mortality d between different lines (L), temperatures (T) and food levels (F). Factors, including all interactions with other factors, were tested by removal from the full model. Remaining factors (italics) were tested by subsequent removal from the reduced model. Signi®cance of each step is indicated by *P < 0.05, **P < 0.01, ***P < 0.001. Model Full model Factor

d.f. 943 dd.f.

Deviance 1006.799 dDevà

±L ±F ±T

18 3 4

21.50 34.94 16.29

±T ±F ±L

16 4 6

27.38 24.55 10.41

* ***

±F ±L ±T

12 9 8

40.24 16.20 11.69

***

P ns

*** ** ns

ns ns

Deviance full model 1006.799 with 943 d.f. (1015 observations ± 72 parameters). P(v2[943]) > 1006.799 = 0.0732; ns. Heterogeneity 1.0676, i.e. no overdispersion in the data; no heterogeneity factor is needed. dd.f., increase in degrees of freedom when factors are removed from the model. àdDev, increase in deviance when factors are removed from the model.

that removing the factor food level F from the model always leads to a highly signi®cant change in the ®t of the model, independent of the order of removal of the factors. This implies a signi®cant effect on d between food levels. No consistent effect of temperature on d could be shown. Lines do not differ in d at any of the environmental conditions (Table 1). The weight-independent mortality d was over twice as high on poor food (Rich food: d 0.051; Poor food: d 0.131). Larvae with a similar weight, but grown on different food levels, therefore show physiological differences which affect their ability to pupariate. The estimated logit regression lines of pupariation fraction to log larval weight for each line and each environmental condition, using a constant d within each food level, are plotted in Fig. 1. The slopes of the logit regression lines (Fig. 1) indicate that substantial variation in critical weight exists among larvae within each line (CV 38 17%; see Materials and methods). The mean critical weight was estimated as the weight at 50% pupariation after correction for weight-independent mortality d, i.e. at 47.5% and 43.5%. The logit regression analysis testing for differences in the relation between log larval weight and the fraction pupariation due to different food levels, lines and temperatures after the weightindependent mortality d has been removed is presented in Table 2. Changes in the ®t of the model with low statistical signi®cance will not be interpreted. The level of food does not affect the mean critical weight for each of the lines, as indicated by the nonsigni®cant Food and



855

Fig. 1 Fraction of larvae pupariating plotted against larval fresh weight for four lines, three temperatures and two food levels. For presentation purposes, observations were paired according to mean weight and points represent means per pair. Logit regression lines, based on log-transformed weights, are shown for each temperature using a common natural mortality rate d per food level, of d 0.051 for rich food and d 0.131 for poor food.

Food*Line effects. Food level affects weight-independent mortality, not weight-dependent mortality. The mean critical weight differs consistently between lines at each temperature, as can be seen from the high signi®cance of the main effect Line after removal of the Food*Line and Temperature*Line interactions (Table 2). The Tanzanian lines show a lower mean critical weight than the French lines (Fig. 2), pointing to a genetic difference due to the geographical origin of the lines. Temperature effects on mean critical weight are relatively small compared with the genetic differences in critical weight, but are highly signi®cant (Table 2, main effect T). A reduction of temperature leads to a higher mean critical weight, especially at the lowest temperature. This effect of temperature appears to be more


pronounced in the two French lines (Fig. 2). In order to investigate whether the temperature effect on pupariation differs between the France lines and the Tanzania lines, a series of logit regression models was implemented by starting from the model containing only main effects but no interactions (Table 3, model 1). To this model, a Line*Temperature interaction was added for the two France lines together, as well as a Line*Temperature interaction for the two Tanzania lines together, leading to model 2 (Table 3). The deviance is signi®cantly lowered by offering two degrees of freedom: the temperature effect on fraction pupariation therefore differs between the France lines on the one hand and the Tanzania lines on the other hand, as is re¯ected in Fig. 2. Ignoring the subdivision of the four lines in two Tanzania and two

856


Table 2 Logit regression analysis testing differences in the relation between log larval weight and the fraction pupariation for different lines (L), temperatures (T) and food levels (F). Models include separate parameters for weight-independent mortality d for both food levels. The main factors were tested after consecutive removal of all higher order interactions containing that factor. Signi®cance of each step is indicated by *P < 0.05, **P < 0.01, ***P < 0.001. Model Full model Factor

d.f. 965 dd.f.

Deviance 1031.04 dDevà

±T*F*L

12

37.45

±T*F ±T*L ±T

4 12 4

11.95 11.74 48.94

±F*L ±F*T ±F

6 4 2

11.56 10.02 3.04

±L*T ±L*F ±L

12 6 6

11.78 13.23 323.05

P

*** * ns

*** ns

*

ns ns

* ***

Deviance full model 1031.044 with 965 d.f. (1015 observations ± 50 parameters). P(v2[965]) > 1031.044 = 0.0687; ns. Heterogeneity 1.0684, i.e. no overdispersion in the data; no heterogeneity factor is needed. dd.f., increase in degrees of freedom when factors are removed from the model. àdDev increase in deviance when factors are removed from the model.

France lines (model 5 in Table 3) does not lead to any signi®cant improvement from keeping the lines subdivided by geographical origin: lines do not differ within populations. In the Tanzania lines some temperature effect is present; model 4 in which no Line*Temperature interaction for the Tanzania lines was implemented had a signi®cantly worse ®t than models 2 and 5 (Table 3). A marginally better model can be found in model 3, which assumes the in¯uence of 22.5 °C on pupariation to be the same as that of 27.5 °C. Although model 3 is not signi®cantly different from model 2, which assumes all three temperatures to have different effects, the balance of low parameter number and low deviance indicates a slightly better ®t to the data (Table 3). Mean critical weight was also estimated using a weight-independent mortality d that is speci®c for each line; the difference in mean critical weight found by this method and the mean critical weight estimated using a constant d per food level (Fig. 2) proved marginal, indicating robustness of the estimates.

Discussion This study shows that in Drosophila melanogaster the probability of reaching the pupal stage depends upon larval weight and larval physiology. Larval physiology, independent of larval weight, is affected by feeding conditions. The probability of reaching the larval stage as mediated by larval weight is genetically variable and

Fig. 2 Critical weight at three temperatures and two food levels in four lines from France and Tanzania, estimated by logit regression on log transformed larval weights; a common natural mortality rate d per food level, of d 0.051 for rich food and d 0.131 for poor food, is used.

affected by the temperature conditions during larval development. For each strain, the larval chance of reaching the pupal stage is summarized in the mean critical weight, i.e. the weight at which half the larvae that subsequently survive attain their ability to pupariate without further feeding. The mean critical weight is lower at higher temperatures. However, a reduction of food level does not affect the mean critical weight, thus con®rming Bakker's (1959) basic model for larval growth within set temperatures. Marked genetic differences were observed in mean critical weight between lines of different geographical origin, differences which are maintained with temperature. The effects of temperature on the mean critical weight are relatively small as compared with the observed genetic differences. Few data are available on the effect of growing conditions on the mean critical weight for pupariation. Partridge et al. (1994) observed that in larvae with a similar weight, the fraction pupariation is higher in larvae that grow faster. We observed a similar reduction in the pupariation fraction with the reduction of food level; the reduction of food level causes a 30% reduction in growth rate (De Moed et al., 1998). However, the reduction in pupariation fraction is maintained in larvae with a high weight, indicating that its mechanistic basis is unrelated to



857

Table 3 Logit regression analysis testing the effect of line combinations and temperature on the fraction pupariation. The main factors log larval weight (W), food (F), temperature (T) and line (L) were tested ®rst. All models contain separate parameters for the main effect of Lines. Models include separate parameters for weight-independent mortality d for both food levels. Model

Parameters

Deviance

Full model: 1 main effects only: independence of temperature and line

50 9

1031.04 1115.287

2

adding L*T for France lines adding L*T for Tanzania lines all temperatures separate 3 adding L*T for France lines adding L*T for Tanzania lines temperature: effect 22.5 °C = effect 27.5 °C 4 adding L*T for France lines no L*T for Tanzania lines all temperatures separate

11

5

15

adding L*T over all lines separately

9

9

1108.524

1109.085

1116.522

dd.f.

dDev

1 vs. 2: 2 1 vs. 5: 6 2 vs. 5: 4

1 vs. 2: 6.723 1 vs. 5: 11.609 2 vs. 5: 4.846

3 2 3 6 4 2 4 6

3 vs. 2: 0.561 3 vs. 5: 6.407 4 vs. 2: 7.998 4 vs. 5: 12.844

vs. 2: vs. 5: vs. 2: vs. 5:

P

0.025 < P < 0.05 ns ns

ns ns 0.01 < P < 0.025 0.025 < P < 0.05

1103.678

the attainment of the mean critical weight. Within each food level, fast larval growth and high mean critical weight go together in our lines. Larval growth rates of the two France lines are higher than larval growth rates of the two Tanzania lines (De Moed et al., 1998). The response of the mean critical weight to temperature parallels the generally observed reduction of adult body size with temperature (e.g. Ray, 1960; David et al., 1983). The observed reduction in mean critical weight largely agrees with the reduction in adult body weight of some 10% between 17.5 °C and 27.5 °C (e.g. David & Clavel, 1967). This does not necessarily imply a causal relation between the weights at the two developmental stages. Both mean critical weight and mean ®nal larval weight, and thereby adult weight, result from the interaction between growth rate and development rate. A stronger increase in development rate with temperature as compared to growth rate results in a reduction of weight with temperature (van der Have & de Jong, 1996). A similar effect of temperature on growth rate and development rate throughout larval development will result in a similar response of mean critical weight, i.e. the average weight at the critical stage, and larval weight at pupariation. The observed genetic differences in mean critical weight suggest that substantial genetic variation exists in mean critical weight between natural populations of Drosophila melanogaster. This raises the question how critical weight is related to ®tness and adaptive strategies to cope with environmental variation. It has been suggested that the critical size is determined by the minimal weight that will allow the production of a functional animal (Slansky & Scriber, 1985). In Drosophila melanogaster, adults emerging from larvae of low critical weight show a low fecundity and are often


infertile (personal observation). Furthermore, mating success in males is strongly related to male body size (e.g. Partridge & Farquhar, 1983), suggesting that the minimal size of the larva to produce a competitive adult is substantially larger than the critical size. The critical weight seems to play a more general role in the individual's life-history. The critical weight is pivotal in the trade-off between age and size at maturity, determining the strategy to respond to varying growth conditions. Drosophila funebris and D. immigrans are similar sized as adults, but the critical weight is smaller and the critical stage is set at an earlier stage in development in D. funebris than in D. immigrans (1.5 mg vs. 2.38 mg; Royes & Robertson, 1964). Development time will therefore be less affected by poor growth conditions in D. funebris. As a consequence, body weight in D. funebris is more sensitive to the growth conditions than in D. immigrans (Royes & Robertson, 1964). A postponed critical stage may be selected in species where the extension of the developmental period under poor food conditions may have little ®tness consequences, as in univoltine insects, and the maintenance of a large size under poor conditions gains a larger selective advantage. In species inhabiting ephemeral habitats, such as Drosophila, the developmental period cannot be extended to a large degree, especially under deteriorating conditions, owing to the risk of being unable to complete development. The critical stage must be set at a relatively early stage. A lower mean critical weight also allows the larvae to develop at lower food rations, showing a higher survival rate and faster development. Royes & Robertson (1964) showed that D. funebris, which shows a smaller mean critical weight than D. immigrans, is able to develop under poorer food conditions. Similar patterns were observed in

858


Oncopeltus, where larger (sub)species with higher mean critical weights show a stronger increase in mortality and developmental period in response to poor growth conditions (Blakley & Goodner, 1978; Blakley, 1981). However, the mean critical weight in D. melanogaster (Church & Robertson, 1966; Bakker, 1969) and in Oncopeltus (Blackley, 1981) is positively correlated over strains with adult body size. Both cases suggest that the mean critical weight may have evolved in response to the trade-off between a larger adult size and the ability to develop under poor growth conditions.

Acknowledgments We thank Herman van der Klis for technical assistance and Dick Smit for help with the ®gures. This research was supported by the Netherlands Organization for Scienti®c Research (NWO ± SLW grant 805.36.187).

References Bakker, K. 1959. Feeding period, growth, and pupation in larvae of Drosophila melanogaster. Entomol. Expis Appl. 2: 171±186. Bakker, K. 1961. An analysis of factors which determine success in competition for food among larvae of Drosophila melanogaster. Arch. NeÂerland. Zool. 14: 200±281. Bakker, K. 1969. Selection for rate of growth and its in¯uence on competitive ability of larvae of Drosophila melanogaster. Neth. J. Zool. 19: 541±595. Beadle, G., Tatum, E. & Clancy, C. 1938. Food level in relation to rate of development and eye pigmentation in Drosophila melanogaster. Biol. Bull. 75: 447±462. Beck, S.D. 1971. Growth and retrogression in larvae of Trogoderma glabum (Coleoptera: Dermestidae). 2. Factors in¯uencing pupation. Ann. Entomol Soc. Am. 64: 946±949. Bernardo, J. 1993. Determinants of maturation in animals. Trends Ecol. Evol. 8: 166±173. Berreur, P., Poncheron, P., Berreur-Bennefant, J. & Simpson, P. 1979. Ecdysone levels and pupariation in a temperature sensitive mutation of Drosophila melanogaster. J. Exp. Biol. 210: 333±373. Blakley, N. 1981. Life history signi®cance of size-triggered metamorphosis in milkweed bugs (Oncopeltus). Ecology 62: 57±64. Blakley, N. & Goodner, S.R. 1978. Size-dependent timing of metamorphosis in milkweed bugs (Oncopeltus). & its life history implications. Biol. Bull. 155: 499±510. Church, B.B. & Robertson, F.W. 1996. Biochemical analysis of genetic differences in the growth of Drosophila. Genet. Res. 7: 383±407. David, J.R., Allemand, R., van Herrewege, J. & Cohet, Y. 1983. Ecophysiology: Abiotic factors. In: The Genetics and Biology of Drosophila (M. Asburner, H. L. Carson & J. N. Thompson, eds), pp. 106±154. Academic Press, London. David, J.R. & Clavel, M.F. 1967. In¯uence de la tempeÂrature subie au cours du deÂveloppement sur divers caracteÁres biomeÂtriques des adultes de Drosophila melanogaster. J. Insect Physiol. 13: 717±729. Finney, D.J. 1971. Probit Analysis, 3rd edn. Cambridge University Press, Cambridge.

Fourche, J. 1967. Le determinisme des mues et des metamorphoses chez Drosophila melanogaster: in¯uence de jeune et de la fourniture d'ecdysone. Arch. Anat. Microsc. 56: 141±152. Gebhardt, M.D. & Stearns, S.C. 1993. Phenotypic plasticity for life history traits in Drosophila melanogaster. I. Effect on phenotypic and environmental correlations. J. Evol. Biol. 6: 1±16. van der Have, T.M. & de Jong, G. 1996. Adult size in ectotherms: Temperature effects on growth and differentiation. J. Theoret. Biol. 183: 329±340. de Moed, G.H., de Jong, G. & Scharloo, W. 1997. The phenotypic plasticity of wing size in Drosophila melanogaster: the cellular basis of its genetic variation. Heredity 79: 260±267. de Moed, G.H., de Jong, G. & Scharloo, W. 1998. The energetics of growth in Drosophila melanogaster: effect of temperature and food conditions. Neth. J. Zool. 48: 169±188. Nijhout, H.F. 1979. Stretch-induced moulting in Oncopeltus fasciatus. J. Insect Physiol. 25: 277±281. Nijhout, H.F. 1981. Physiological control of moulting in insects. Am. Zool. 21: 631±640. Nijhout, H.F. & Williams, C.M. 1974. Control of moulting and metamorphosis in the tobacco hornworm, Manduca sexta (L.): growth of the last-instar larva and the decision to pupate. J. Exp. Biol. 61: 481±491. Noach, E.J.K., de Jong, G. & Scharloo, W. 1996. Phenotypic plasticity in morphological traits in two populations of Drosophila melanogaster. J. Evol. Biol. 9: 831±844. Partridge, L., Barrie, B., Fowler, K. & French, V. 1994. Thermal evolution of pre-adult life history traits in Drosophila melanogaster. J. Evol. Biol. 7: 645±664. Partridge, L. & Farquhar. 1983. Lifetime mating success in male fruit¯ies (Drosophila melanogaster) is related to their size. Anim. Behav. 31: 871±877. Ray, C. 1960. The application of Bergmann's and Allen's rules to the poikilotherms. J. Morph. 106: 85±108. Riddiford, L.M. 1993. Hormone receptors and the regulation of insect metamorphosis. Receptor 3: 203±209. Robertson, F.W. 1960. The ecological genetics of growth in Drosophila. 1. Body size and developmental time on different diets. Genet. Res. 1: 288±304. Robertson, F.W. 1963. The ecological genetics of growth in Drosophila. 6. The genetic correlation between the duration of the larval period and body size in relation to larval diet. Genet. Res. 4: 74±92. Royes, V.W. & Robertson, F.W. 1964. The nutritional requirements and growth relations of different species of Drosophila. J. Exp. Zool. 156: 105±136. Sang, J.H. 1956. The quantitative nutrition requirements of Drosophila melanogaster. J. Exp. Biol. 33: 45±72. Slansky, F. & Scriber, J.M. 1985. Food consumption and utilization. In: Comprehensive Insect Physiology, Biochemistry and Pharmacology (G. A. Kerkut & L. I. Gilbert, eds), Vol. 4, pp. 87± 163. Pergamon Press, Oxford. Wigglesworth, V.B. 1934. The physiology of ecdysis in Rhodnius prolixus. II. Factors controlling moulting and metamorphosis. Q. J. Microsc. Sci. 77: 191±222. ZdaÂrek, J. 1985. Regulation of pupariation in ¯ies. In: Comprehensive Insect Physiology, Biochemistry and Pharmacology (G. A. Kerkut & L. I. Gilbert, eds), Vol. 8, pp. 301±334. Pergamon Press, Oxford. Received 11 August 1998; accepted 16 December 1998
