Heredity 65 (1990) 169—177 The Genetical Society of Great Britain Received 9 February 1990

Antagonistic pleiotropy and the evolution of wing dimorphism in the sand , firm us

D. A. Roff Department of Biology, McGill University, 1205 Dr Penfield Avenue, Montreal, Quebec, Canada, H3A 1BI.

At 30°C the micropterous females of the sand cricket, Gryllus firmus, begin reproduction at an earlier age after eclosion and have a larger cumulative fecundity than macropterous females. These reproductive costs may offset the advantages of being macropterous and hence capable of migration. The evolutionary significance of this phenotypic trade-off, which is characteristic of wing dimorphic in general, is contigent on the traits being genetically correlated. The genetic basis of the phenotypic tradeoff between flight capability and reproduction in the sand cricket, Grylius firmus, was examined by selecting for increased and decreased incidence of macroptery, and measuring the age schedules of fecundity of macropterous and micropterous females in the selected and control lines. The two traits, wing dimorphism and age schedule of reproduction, are shown to be genetically correlated. Although the mean fecundity within the selected populations changed the fecundities of macropterous and micropterous forms remained constant, suggesting that the age schedule of reproduction may itself be a threshold trait with respect to the continuously varying character controlling the expression of wing form. The relevance of antagonistic pleiotropy to the maintenance of genetic variation for wing form and the age schedule of reproduction is discussed.

INTRODUCTION 1975, 1990a). But though there is abundant evidence of a phenotypic trade-off between repro- Migrationby flight is an important aspect of the duction and wing morph, there are no studies life history of many species, permitting them demonstrating that these tradeoffs have a genetic to colonize highly ephemeral habitats (Southwood, basis and are thus examples of antagonistic 1962; Johnson, 1969; Harrison, 1980; Dingle, pleiotropy. 1985). However, the benefits of migration may be The purpose of the present study was to test offset in part both by the energetic cost of flight the hypothesis that in the sand cricket, Gryllus and by the cost of producing the machinery of firmus, there are negative genetic correlations flight, viz the wing muscles, wings etc (Roff, 1977, between wing morph and reproductive traits. It 1986a; Roff and Fairbairn, 1990; Inglesfield and has previously been shown that in this species the Begon, 1983; Denno et a!., 1989). The latter cost macropterous (flight capable) morph has a delayed has been demonstrated by experiments with wing age at first reproduction and a decreased total dimorphic insects, (i.e., species in which some fecundity (Roff, 1984, 1989). Furthermore, the individuals within a population or family have heritability of wing morph is large (approximately functional wings and are capable of flight, while 0.65, Roll, 1986b), and hence selection for others have reduced wings and are incapable of increased or decreased incidence of macroptery flight). Within such species there is a consistent should produce rapid changes, a prediction pattern of delayed reproduction and reduced verified by artificial selection (Roll, 1990b). The fecundity of the fully winged morph (Roff, 1986a; existence of genetic correlations between wing Denno et a!., 1989; Roll and Fairbairn, 1990). If morph and reproductive characters was investi- genetically based, these trade-offs could sig- gated by measuring correlated changes in the age nificantly influence the evolution of the incidence schedule of reproduction during the course of this of wing dimorphism within a population (Roll, selection experiment. 170 D. ROFF

MATERIALS AND METHODS tion design was followed from the outset and no initial group of single pair matings constructed. Experimental protocol To estimate the age schedule of fecundities individual pairs were set up as described in Roll Detailsof the species and rearing methods are (1984). Adults were fed Purina© rabbit chow and given in Roff (1986b), and only the salient points eggs removed weekly for a period of four weeks. are presented here. The stock of G. firmus used in In pairs from the lines selected for macroptery the present study was derived from approximately both males and females were macropterous, while 40 individuals (approximate sex ratio 1: 1) from a in the pairs from the micropterous lines both adults single location in northern Florida in 1981. They were micropterous. For the control lines, the male are maintained in diapause averting conditions and female within a pairing were of the same (25-30°C, no set photoperiod but the laboratory morph, one half of the pairings comprising both lights ensure a relatively long light period), with macropterous adults and one half comprising both a breeding stock of between 100—300 individuals. micropterous adults. Approximately 20 replicates For the selection experiments individuals were were used for each combination, the exact number raised in batches of 60 individuals per disposable depending upon numbers emerging, space and mouse cage, as described in Roff (1986b). Food labour limitations. The fecundity of the microp- was provided, ad libitum and comprised Purina© terous females in the LW lines and the fecundity rabbit chow and fresh lettuce leaves. of the macropterous females in the SW lines were Previous studies were conducted at 30°C and not measured because by generation 5, when a photoperiod of 17 hrs L: 7 hrs D: under these fecundity trials were initiated, there were too few conditions the proportion of macropterous males to obtain a sufficient sample: as explained below and females is about 76 per cent and 64 per cent, this does not interfere with the statistical analysis. respectively. To reduce this percentage to around In the first selection experiment, measurements 50% in the females, in the present experiment were made at generations 5 and 15, and in the crickets were reared at 28°C, 15 hrs L; 9 hrs D. The second experiment at generations 5, 6 and 11. The first selected lines were initially constructed as total sample size was 419, comprising 216 macrop- follows: eggs were obtained from the stock culture terous and 203 micropterous females. In addition and the nymphs raised under the experimental to weekly egg counts, daily counts to day 14 were conditions. From this group 20 pairs were extracted made in generation 11 of the second selection comprising ten pairs LW x LW (macropterous x experiment. macropterous) and tenpairs SW x SW (micropterous x micropterous). This procedure was adopted to obtained an initial estimate of the Statisticalanalysis heritability of wing dimorphism (Roll, 1990b). Themeasurement of the generic correlation To establish a macropterous line (hereafter between a threshold trait and some other character referred to as the LW line) 200 adults (100 males, presents particular problems. To understand these, 100 females) from the ten LW x LW matings were first consider the usual situation of two con- mixed together, with approximately equal rep- tinuously varying characters X and Y (e.g., body resentation from each family. Similarly, a and size and fecundity) for which a phenotypic micropterous line (SW) was started by mixing relationship exists (fig. 1). Suppose X is increased together the offspring from the SW x SW crosses. or decreased by artificial selection; if the genetic A Control line (C) was formed by mixing the correlation, rg, between X and Y is 1 then the offspring from all crosses. For each of the lines, correlated trait Y will "slide" along the regression six cages containing 60 newly hatched nymphs per to the point A, but if rg equals 0 there will be no cage were established. Males and females of a change in Y with a change in X (point B in fig. 1). desired morph were selected upon eclosion into If rg lies between 0 and 1 trait Y will respond by adults until approximately 100 of each sex were moving to a point between A and B. obtained. The relatively large number of parents For a threshold trait we can distinguish only was used to prevent inbreeding depression. The discrete phenotypes, which are, in this case, the mass selection procedure outlined above was fol- macropterous (LW) and micropterous (SW) lowed on all generations subsequent to the first. morphs. However, we assume that there is an To provide a replicate, a second series of lines was underlying variable that shows continuous vari- set up some months after the first, the protocol ation, genotypes exceeding the threshold value followed being identical except that a mass selec- producing one phenotype and those below the ANTAGONISTIC PLEIOTROPY AND WING DIMORPHISM 171

of SW. The two scenarios can be distinguished by measuring Y at several generations during selec- tion. Note that even if the value of Y for the LW and SW morphs stay the same under selection, the mean value of Y in the selected population will change by virtue of the changing frequencies of macropterous and micropterous individuals. Y The basic statistical model used in the present analysis was, Y= a+bX1+cX2+dX3 + interaction terms + error where X1 is "wing morph" (LW or SW), X2 is "treatment" (selected or control), X3 is "trial", Selected Base population Selected and a, b, c, d are constants. Effects due to "trial" may reflect systematic differences resulting from x selection or "error" differences due to "random" Figure1 Hypothetical phenotypic relationship between two variation between trials; for this reason I used a continuously varying characters X and Y (e.g.,bodysize categorical variable for "trial". fecundity). If the genetic correlation, r,equals1 selection on X will produce a correlated response to Y to A. If Tg= 0 there will be no response (B), otherwise Y will move to a point between A and B. RESULTS

Theresponse to selection was very rapid, and by threshold producing the alternate (Falconer, 1981). generation 5 the frequency of macropterous Because it is not possible to measure the actual crickets had risen to over 80 per cent in the LW value of the underlying factor, X, an unambiguous lines and decreased to less than 10 per cent in the plot of the phenotypic relationship between X and SW lines (fig. 3). An asymmetrical response was Y in the unselected population is not possible. If obtained with the incidence of microptery increas- the relationship is linear we can plot three points ing more rapidly in the SW lines than the incidence along the proposed line, two corresponding to the of macroptery in the LW lines (fig. 3, Roff, 1990b). mean values for the two morphs within the unselec- Given the large changes in frequency, if a corre- ted population and the third corresponding to the lated response in fecundity and/or age at first mean value of the whole population (fig. 2(a)). On reproduction exists, it should be evident by gener- the other hand, the relationship between X and Y ation 5 when the first measurements were made. might itself correspond to a threshold trait, in The overall mean egg production in the first which case the mean value for the base population week was 774 (SE =6.6)eggs for the micropterous will not, in general, lie on the curve (fig. 2(b)). In females and 395 (55) for the macropterous the first case, if rg= 1, artificial selection for females. Many females did not lay eggs in the first increased per cent LW will decrease Y (e.g.,fecun- week and hence for this week a logarithmic trans- dity), while selection for increased per cent SW formation (log {eggs+ 1}) was used (the analysis will increase Y (points A in fig. 2(a); note that this was also done with the untransformed data without result is the same as shown in fig. 1). However, in significant differences in the results). In week 1 the second scenario there will be no change in Y there was a significant effect of wing morph (X1) for a particular morph (fig. 2(b)) though the mean and trial (X3), but no effect of treatment (X2, table value of Y for the population as a whole will 1), and no significant interaction effects. The sig- change as the frequency of each morph changes. nificant effect due to trial was due to the fecundities We can conclude that if the value of Y of the LW of females in all lines at generation 15 of the first morph in the line selected for increased frequency selection experiment being low, a reduction that of LW either remains the same or decreases then was also found in the remaining three weeks. While there is a correlated response and rg is greater than inbreeding depression cannot be entirely ruled out, 0: a similar argument can be applied to the SW the large number of males and females contributing morph in the line selected for increased frequency to each generation, and the fact that there has been 172 D. ROFF

Increased LW Increased SW a A r =1 9 SW O r 1 B 9 r =0 r=0 V g g r '1 B 9 LW r =1 9 A

A A A Selected Base population Selected

Increased LW Increased SW b A • r=1 SW 9 0'r1 B g r =0 Vr=O • -I— 9 B 0r19

r =1 9 w w LW A

A A A Selected Base population Selected x

Figure 2 Two hypothetical phenotypic relationships between a continuously varying character, X, which determines the expression of a threshold character (e.g.,wingdimorphism) and a second character Y (e.g.,fecundity).The average value of the two morphs (LW and SW) will lie on the line but the mean value for an individual from the population need not do so (e.g., panel b). If rg= 1selection for a change in the incidence of macroptery will cause a correlated response of Y to A: if Tg= 0,the mean value of the population will remain constant (B), otherwise LW and SW will move to values between A and B. AITAGONISTIC PLEIOTROPY AND WING DIMORPHISM 173 no obvious reduction in hatchling numbers during Females the selection experiments, suggests that this reduc- tion was due to some other cause. The above results indicate, (1) that the fecun- dity of a macropterous or micropterous female did not depend upon whether it came from a selected line or the control line, and (2) that the SW morph (I) lay significantly more eggs than the LW morph. 0 Thus, the mean 1st week fecundity of the LW line decreased as the incidence of macroptery a) increased, while that of the SW line increased as 0 the incidence of microptery increased. The lack of C.) a significant interaction term suggests that the relationship between fecundity and the factor con- a) trolling wing morph is more like a threshold trait 0) (fig. 2(b)) than a continuous linear function (fig. C 2(a)). ci The increased fecundity of the SW morph could 0 be due to a decrease in the age at first reproduction a) and/or a higher rate of egg production. The first 0 possibility was tested in two ways. First, I com- pared the proportion of females (arcsine trans- formed) that began oviposition in the first week (i.e., fecundity >0) with respect to wing morph, treatment and trial. As in the previous analysis there was a significant effect of wing morph (P = 6 8 10 OO4;i.e., more micropterous females layed eggs in the first week than macropterous females), and Generation trial (P =0.001),but no effect of treatment (P = Figure3 Response to selection for increasing and decreasing 0.52),and no significant interaction effects. incidence of macroptery in Gryllus firmus.Line1 —; Secondly, I compared the ages at first reproduction Line 2———. obtained from the daily egg counts made on line 2 at generation 11 (fig. 4). The statistical model used was, Y =a+ bX1 + cX2 + dX1X2 + error, where P>0.05) and no significant effect of treatment Y =dayat first reproduction, X1 =wingmorph, after adjusting for wing morph (F194= 014, P> X2 =treatment,and a, b, c, d are fitted constants. 0.5), but a highly significant effect of wing morph There was no significant interaction (F194 =4.59, (F1,94= 165, P

Table I Correlation coefficients for the regressions of fecundity on wing morph, treatment and trial (cols. 1—3). Because there was no effect of treatment only the adjusted R value of the additive model comprising X1 and X3 (col. 4), and the saturated model (X1, X2, X3, X1X2, X1X3, X2X3, X1X2X3)arepresented. Total sample size is 419. Model

Week X1 X2 X3 X1,X3 Saturated Week la 000 03O Ø.37** Ø.37** Week 2 000 000 Ø.33** Ø.33** Ø.37** Week 3 0.11* 000 0.36** 0.38** Ø.37** Week 4 0.11* 0'OO 0.20** 0.24** 0.23* Total 000 0.00 036** 0.36** 035**

X1: Wing morph (SW or LW). X2: Treatment (Selected or Control). X3: Trial (6 trials). a: based on log (x + 1) transformation. *P<0.01.**P<0.001. 174 0. ROFF

100

- - -.- - SW, Control 80 0) • SW, Selected C C) •060 5-0 a)40

- --.-- LW, Control 20 • LW, Selected

0

3 + Days since eclosion Figure 4 Cumulative proportion of females laying as a function of age since eclosion in G. firmus. reproduction of micropterous females was 655 In summary, differences in the age schedule of days (±0.14) and for macropterous females 745 reproduction were found between wing morphs days (±018). These two analyses indicate, (1) that but not between crickets of the same morph from micropterous female crickets begin reproducing at different lines. The micropterous morph had a an earlier age than macropterous crickets, and (2) higher egg production in the first week after that this difference is maintained under selection eclosion than the macropterous morph but the total for changes in the incidence of macroptery. They fecundity of the two morphs was the same. The do not rule out the possibility of additional higher egg production in the first week of the differences in rate of egg production. micropterous morph was, in part at least, a con- Egg production in the second week was not sequence of an earlier age at first reproduction. correlated with wing morph or treatment, but was The maintenance of differences in the age significantly correlated to trial (table 1). Ignoring schedules of reproduction between micropterous the effects of trial, the overall mean number of and macropterous crickets means that as the eggs laid was 3898 (265) for micropterous females frequency of the macropterous morph changes in and 3660 (18.15) for macropterous. In weeks 3 response to selection on the incidence of macrop- and 4 there was a significant effect of wing morph tery there will be a correlated response in the mean on egg production (table 1) but, contrary to egg age schedule of reproduction in the population. production in week 1, macropterous females pro- Thus we can conclude that the phenotypic relation- duced more eggs than micropterous females (in ship between the age schedule of reproduction and week 3 micropterous females laid 3450 [19.5] eggs wing morph observed in the unselected population compared to 4124 [19.74] laid by macropterous is in part due to a genetic correlation between the females: results for week 4 were 2869 [16.5] and traits. 3481 [17.33], respectively). Furthermore, in con- trast to the results obtained at 30°C and a photo- period of 17L:7D, the total egg production of DISCUSSION micropterous females was not significantly different from that of macropterous females (table Themicropterous morph of G.firmus begins repro- 1, means of 1099 [540] and 1166 [47.28] for ducing after eclosion sooner than the macrop- micropterous and macropterous females, respec- terous morph and lays more eggs in the first week tively). of adult life. The results of the selection experiment ANTAGONISTIC PLEIOTROPY AND WING DIMORPHISM 175 indicate that these differences in reproduction are of flight propensity in the selected and control genetically correlated to wing morph. Contrary to stocks of G. firmus suggest that even among the the finding at 30°C (Roff, 1984, 1989), when the macropterous morph these exists a continuum of crickets are raised at 28°C there is no difference in migratory propensity (Fairbairn, 1986; Butler, total fecundity between the two morphs. The 1987; Fairbairn and Desranleau, 1987; Fairbairn absence of a difference is caused by a "crossing- and Roff, 1990). Therefore, it would be premature over" of the m curves (i.e., the age schedules of to presume a threshold response with respect to reproduction) in the second week. Such an inter- m: further experiments using lines that are section was noted by Fairbairn (1988) for Gerris vitually pure breeding and larger sample sizes are remigis, Limnoporus caniculatus, (data from Zera, required to examine this question. 1984), and the cricket species, Gryllus firmus and In the laboratory culture there is clearly no Allonemobius fasciatus, (data from RoIl, 1984). It advantage to retaining flight capability for migra- also occurs in Javesella pellucida (Mochida, 1973), tion; therefore, why is the morph maintained under Cavelerius saccharivorus (Fujisaki, 1986), and Hor- laboratory culture? The proportion of vathiolius gibbicollis (Soibreck, 1986), but not in micropterous males and females in the stock cul- the aphid species Setobion avenae, Metopolophium ture are presently 973 per cent and 8056 per cent, dirhodum (Wratten, 1977) and Aphisfabae (Dixon respectively. Since the percentage of each morph and Wratten, 1971). A reversal in egg production in the stock culture has not been monitored it is later in adult life does not necessarily mean that possible that the incidence of macroptery has the egg production of both morphs will eventually declined. However, since neither the temperature be equal; for example, the intersection of the m nor photoperiod are monitored the high incidence curves in G.firmus at 30°C occurs late in reproduc- of microptery may be due to environmental condi- tive life and could not lead to the cumulative tions. Recent extractions from the stock indicate fecundity of the LW morph eventually catching up that the incidence of macroptery under the tem- to that of the SW (Roff, 1984, 1989). An intersection perature and photoperiod conditions of the selec- of the m curves may be typical in wing dimorphic tion experiment produce aproximately 50 per cent species, but this does not generally offset the fecun- macroptery, as found at the initiation of the selec- dity advantage of the SW morph (Roff, 1986a; tion experiment (Roff, unpublished data). A Roff and Fairbairn, 1990). That the fecundity second reason why macroptery may be maintained advantage is negated under one environmental is that the fecundity advantage of micropterous regime but not another in G. firmus suggests crickets may disappear at lower temperatures. The that more data are needed on the relationship present experiments indicate that the cumulative between the age schedule of reproduction and fecundity advantage of micropterous females is different environmental conditions among insect found at 30°C but not at 28°C, though this could species. be a consequence of differences in temperature The fact that the difference in fecundity between and/or humidity since the experiments were run the same morphs from the selected and control in different incubators. Data presently being ana- lines did not change in any systematic way with lysed also suggest that macropterous crickets the generation of selection suggests that the develop faster than micropterous crickets at lower relationship between m and the factor controlling but not higher temperatures: the shorter develop- the expression of wing form is more like a threshold ment time of the macropterous morph could function (fig. 2(b)) than a linear function (fig. 2(a); counterbalance the longer period between final the lack of a change in the selected lines with eclosion and first reproduction. Furthermore, there generation cannot be used as sole evidence since may be no fitness advantage to microptery in the after generation 5 little further response to selec- male sex (Roff and Fairbairn, 1990). These factors, tion occurred: the important observation is the lack the environmental conditions of culture, a change of a difference between selected and control lines). in relative fitness with environmental factors and Because there was significant variation between a lack of difference in fitness between the morphs trials which may have obscured small changes this within the males, will at least slow down the rate conclusion must be considered tentative. It is not at which the micropterous morph will be expected unreasonable to postulate a threshold type of to increase in frequency. The consequences of response in the present case if the wing form these factors on the relative fitness of the two delineates a migratory and non migratory morph. morphs is under study. However, interspecific comparisons of the flight The hypothesis that there are trade-oIls propensity of various gerrid species and an assay between traits is central to most theories of the 176 D. ROFF evolution of life history variation (Stearns, 1976, of traits but because it may also play a role in 1977; Maynard Smith, 1978). The experiment pre- preserving genetic variation (reviewed in Rose, sented in this paper demonstrates that, in principle, 1983, 1985). In single locus models some domin- life history variation in wing dimorphic insects may ance is required for a genetic polymorphism to be be generated by the genetically determined trade- sustained by antagonistic pleiotropy (Rose, 1982). off between migratory ability and reproduction. Single locus mechanisms for the determination of Macropterous individuals can migrate and hence wing morph are found in the insects, particularly the loss of fitness accruing from a delayed age at among those with holometabolous metamorplosis first reproduction and, under some conditions a (Roff and Fairbairn, 1990). It is difficult to evaluate reduced fecundity, is compensated by the advan- explicitly the relative fitness values of macrop- tage of migration in an ephemeral habitat. The terous and micropterous morphs, but simulations genetic correlation between these two traits means using single locus models with dominance demon- that, at least in the short term, evolutionary changes strate that in a heterogeneous environment poly- in one will be modulated by the decreased fitness morphisms for migration can be preserved as a necessarily induced by the other (antagonistic consequence of the differences in fitness of the two pleiotropy). morphs (Roff, 1975). Dominance is also necessary In some species, such as Pyrrhocoris apterus but less important in polygenic systems (Rose, (Honèk, 1976), Gerris remigis (Fairbairn, 1986) 1985), but without further numerical analysis, it is and Chorthippus parallelus (Ritchie et a!., 1987), not possible to define the exact conditions required the macropterous morph is found only at very low to maintain variation of the incidence of macrop- frequencies and does not fly. Ritchie et a!. (1986) tery in a heterogeneous environment. speculated that the macropterous form of Chorthip- Several factors may operate in conjunction with pus parallelus, though showing the same delay in antagonistic pleiotropy to maintain variation. reproduction and cumulative fecundity as other First, the expression of wing form both in G.firmus wing dimorphic species, may possess fitness advan- and wing dimorphic insects in general, is highly tages other than the ability to disperse. However, dependent on environmental conditions such as Fairbairn (1986) argued that in G. remigis popula- temperature and photoperiod (Honék, 1976; Rofi, tions from Eastern Canada the occurrence of unpublished data). Therefore, though the herita- macropters is a result of unusual environmental bility of wing form is high under constant condi- conditions and not a balance between fitness com- tions (approximately 0.65; Roff, 1986b, 1990), it ponents of the apterous and macropterous morphs: may be very low under the highly variable condi- this mechanism could also account for the tions that may occur in the field, and phenotypic occassional appearance of macropterous Chorthip- differences in wing form and the age schedule of pus parallelus (Ritchie, personal communication). reproduction may not reflect genetic variation. Selection against the appearance of non migratory Thus much of the genetic variation may be macropterous forms at low frequency would be effectively "hidden" from the action of natural very weak and macroptery could persist for a long selection. Secondly, the optimal level of migration time as a "genetic relic": it should not be supposed and hence wing dimorphism may vary between that all populations are at genetic equilibrium (Ber- regions; as there will undoubtedly be migration yen and Gill, 1983; Fairbairn, 1986; Fairbairn and between regions there may exist no level of migra- Roff, 1990). Macroptery is not a "genetic relic" in tion for both regions that is most fit and selection G. firmus, the incidence of macroptery in the field may thus tend to preserve variation, or at least being as high as 23 per cent (Veazey et a!., 1976) erode it at a very slow rate. Future research must and the macropterous form being capable of, and address the question of genotype by environment willing to fly (Fairbairn and Roff, 1990; Roff, per- interactions and the pattern of environmental sonal observation). For this species, for other heterogeneity found in the field. North American Gryllus species and North American Orthopera in general, the hypothesis that the dimorphism is maintained, in part at least, by environmental heterogeneity appears to be the Acknowledgements I am most grateful for the constructive most attractive at present (Walker and Sivinski, criticisms of Drs D. J. Fairbairn, H. Dingle, J. Endler, T. Prout and M. Turelli. Catherine Helick and Sharon David provided 1986; Roff, 1990c). indispensable technical assistance. This work was supported Antagonistic pleiotropy is important not only by an operating grant from National Science and Engineering because it can determine the optimal combination Council of Canada. ANTAGONISTIC PLEIOTROPY AND WING DIMORPHISM 177

REFERENCES ROFF, D. A. 1977. Dispersal in dipterans: its costs and con- sequences. J. Anim. EcoL, 46, 443-456. ROFF, D. A. 1984. The cost of being able to fly: a study of wing T. 1987. Population dynamics and dispersal in water- BUTLER, polymorphism in two species of crickets. Oecologia, 63, striders: a comparative approach. M.Sc. Thesis, Concordia 30-37. University, Quebec, Canada. ROFF, D. A. 1986a. The evolution of wing dimorphism in insects. DENNO, R. F., OLMSTEAD, K. L. AND McCLOUD, E. S. 1989. Evolution, 40, 1009-1020. Reproductive cost of flight capability: a comparison of life ROFF, D. A. 1986b. The genetic basis of wing dimorphism in history traits in wing dimorphic planthoppers. EcoL Ent., the sand cricket, Gryllus firmus and its relevance to the 14, 31-44. DINGLE, H. 1985. Migration. In Kerkut, G. A. and Gilbert, evolution of wing dimorphisms in insects. Heredity, 57, 221—231. L. I. (eds), Comprehensive Insect Physiology, Biochemistry ROFF, D. A. 1989. Exaptation and the evolution of dealation in and Pharmacology. Vol. 9, Behaviour. Pergamon Press, New insects. J. evol. Biol., 2, 109—123. York. ROFF, D. A. 1990a. Understanding the evolution of insect life DIXON, A. F. G. AND wRArrEN, S. D. 1971. Laboratory studies on aggregation, size and fecundity in the black bean aphid, cycles: the role of genetical analysis. In, Gilbert, F. (ed), Genetics, Evolution and Coordination of Insect Ltfe Cycles. Aphis fabae Scop. BulL ent. Res., 61, 97-111. Springer-Verlag, Berlin (In press). FAIRBAIRN, D. i. 1986. Does alary dimorphism imply dispersal ROFF, D. A. 1990b. Selection for changes in the incidence of dimorphism in the waterstrider, Gerris remigis? Ecol. Ent., wing dimorphism in Gryllusfirmus. Heredity, 65, 163—168. 11, 355—368. ROFF. D. A. 1990c. The evolution of flightlessness in insects. FAIRBAIRN, D. .i. 1988. Adaptive significance of wing dimorph- ism in the absence of dispersal: a comparative study of Ecol. Monogr., (In press). ROFF, D. A. AND FAIRBAIRN, D. j.1990.Wing dimorphisms wing morphs in the waterstrider, Gerris remigis. Ecol. Ent., and the evolution of migratory polymorphisms among the 13, 273—28 1. Insecta. Am. ZooL, (In press). FAIRBAIRN, D. J. AND DESRANLEAU, L. 1987. Flight threshold, ROSE, M. R. 1982. Antagonistic pleiotropy, dominance and wing muscle histolysis, and alary polymorphism: correlated genetic variation. Heredity, 48, 63—78. traits for dispersal tendency in the Gerridae. EcoL Ent., 12, ROSE, M. R. 1983. Theories of life-hostory evolution. Am. ZooL, 13—24. 23, 15—23. FAIRBAIRN, D. J. AND ROFF, D. A. 1990. Genetic correlations ROSE, M. R. 1985. Life history evolution with antagonistic among traits determining migratory tendency in the sand pleiotropy and overlapping generations. Theor. Pop. BioL, cricket, Gtyllusfirmus. Evolution, (In press). 28, 342—358. FALCONER, Ii S. 1981. Introduction to Quantiative Genetics. SOLB RECK, C. 1986. Wing and flight muscle polymorphism in 2nd ed. Longman, London and New York. a lygaeid bug, Horvathiolus gibbicollis: determinants and FUJISAKI, K. 1986. Reproductive properties of the oriental life history consequences. EcoL Ent., 11, 435-444. chinch bug, Cavelerius saccharivorus Okajima (Heterop- SOUTHWOOD, T. R. E. 1962. Migration of terrestrial tera: Lygaeidae), in relation to its wing polymorphism. in relation to habitat. BioL Rev., 37, 171—214. Res. Pop. EcoL, 28, 43-52. STEARNS, S. C. 1976. Life-history tactics: a review of the ideas. HARRISON, R. G. 1980. Dispersal polymorphism in insects. A. Q. Rev. Biol., 51, 3-47. Rev. Ecol. Syst., 11, 95—118. STEARNS. S. C. 1977. The evolution of life-history traits: a HONEK, A. 1976. Factors influencing the wing polymorphism critique of the theory and a review of the data. Ann. Rev. in Pyrrhocoris apterus (Heteroptera, Pyrrhocoridae). Zool. Ecol. Syst., 8, 145-171. Jb. Syst. Bd., 103, 1-22. VEAZEY, J. N., KAY, C. A. R., WALKER, T. J. AND WHITCOMB, INGLESFIELD, C. AND BEGON, M. 1983. The ontogeny and W. H. 1976. Seasonal abundance, sex ratio, and macroptery cost of migration in Drosophila subobscura Collin. BioL J. of field crickets in Northern Florida. Ann. ent. Soc. Amer., Linn. Soc., Lond., 167, 76-96. 69, 374-380. JOHNSON, C. G. 1969. Migration and Dispersal of Insects by WALKER, T, AND SIVINSKI, J. M. 1986. Wing dimorphism in Flight. Methuen, London. field crickets (Orthopera: : Gryllus). Ann. ent. MAYNARD SMITH, J. 1978. Optimization theory in evolution. Soc. Amer., 79, 84-90. A. Rev. EcoL Syst., 9, 31-56. wRArrEN, S. D. 1977. Reproductive strategy of winged and MOCHIDA, o. 1973. The characters of the two wing-forms of wingless morphs of the aphids Setobion avenae and Javesella pelucida (F.) (Homoptera: Delphacidae) with Metopolophium dirhodum. Ann. appl. BioL, 85, 319-331. special reference to reproduction. Trans. R. ent. Soc. Lond., ZERA, A. j.1984.Differences in survivorship, development rate 125, 177—225. and fertility between the long-winged and wingless morphs RITCHIE, M. G., BUTLIN, R. K. AND HEWITr, G. M. 1987. Causa- of the waterstrider, Limnoporus canaliculatus. Evolution, tion, fitness effects and morphology of macropterims in 38, 1023—1032. Chorthippus parallelus (: Acrididae). Ecol. Ent., 12, 209—218. ROFF, D. A. 1975. Population stability and the evolution of dispersal in a heterogeneous environment. Oecologia, 19, 2Y1—237.