University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

Anthony Zera Publications Papers in the Biological Sciences

March 1983

Genetic and Environmental Determinants of Wing Polymorphism in the Waterstrider canaliculatus

Anthony J. Zera University of Nebraska - Lincoln, [email protected]

David J. Innes State University of New York, Stony Brook

Margaret E. Saks State University of New York, Stony Brook

Follow this and additional works at: https://digitalcommons.unl.edu/bioscizera

Part of the Microbiology Commons

Zera, Anthony J.; Innes, David J.; and Saks, Margaret E., "Genetic and Environmental Determinants of Wing Polymorphism in the Waterstrider " (1983). Anthony Zera Publications. 21. https://digitalcommons.unl.edu/bioscizera/21

This Article is brought to you for free and open access by the Papers in the Biological Sciences at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Anthony Zera Publications by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. Published in Evolution, 31(3), 1983, pp. 513–522. Copyright © 1983 Society for the Study of Evolution; published by Blackwell Publishing. Used by permission.

Genetic and Environmental Determinants of Wing Polymorphism in the Waterstrider Limnoporus canaliculatus

Anthony J. Zera, David J. Innes, and Margaret E. Saks

Department of Ecology and Evolution, State University of New York, Stony Brook, New York 11794 Submitted July 1981; revised August 1982.

Wing polymorphism commonly occurs in under polygenic control in G. asper. Results of many , especially species of Orthoptera, subsequent genetic studies in G. lacustris and Coleoptera, Homoptera, and (Harri- G. lateralis by Vepsäläinen (1974a) and G. la- son, 1980). The polymorphism consists of dis- custris and G. asper by Guthrie (1959) were crete differences in wing length with morphs purported to corroborate Poisson’s single lo- exhibiting fully developed, reduced, or to- cus model in a modifi ed form. Although recog- tally absent wings. In addition to differences in nizing that morph determination is infl uenced wing length, morphs often differ in a number by environmental factors such as photoperiod of other characteristics such as degree of fl ight and temperature, both Guthrie (1959) and Vep- muscle development, duration of nymphal de- säläinen (1974a) reported that under certain en- velopment, time to fi rst reproduction, fertility vironmental conditions results of their crossing and diapause (Anderson, 1973; Vepsäläinen, experiments were consistent with a single lo- 1978; Harrison, 1980). cus or supergene (Vepsäläinen, 1974a) model. Wing polymorphism is an attractive sys- However, results of these studies are open to tem for investigating the evolution of disper- question. Vepsäläinen (1974a) used fi eld-col- sal in natural populations (Vepsäläinen, 1978; lected females in his study. Since these fe- Denno and Grissell, 1979; Harrison, 1980). A males may have mated in the fi eld with males key step in such studies is the identifi cation of a morph type different from those used in of the environmental and genetic components the laboratory crosses, his data are equivocal. of morph determination. Numerous studies Similarly, Guthrie (1959) did not report tak- of insects from several different orders have ing any precautions to insure that virgin fe- clearly demonstrated that environmental vari- males were used in laboratory crosses. Stud- ables such as photoperiod, temperature and ies by Poisson (1924), Ekblom (1941, 1949) density may strongly infl uence the develop- and Guthrie (1959) did not adequately con- ment of an individual into a particular morph trol environmental variables such as photope- (see references in Harrison, 1980). However, riod and temperature which are now known or the genetic component of morph determina- are strongly suspected to infl uence morph de- tion is poorly understood. termination (Vepsäläinen, 1974a). In addition, In a number of studies, attempts have been in all studies which claimed to demonstrate made to identify the genetic basis of determi- that morph type was inherited as a monogenic nation in species of waterstriders (: trait, progeny ratios of crosses were, in many Hemiptera), and the results used to formulate cases, inconsistent with this model. Conse- models of the evolution of winglessness. Pois- quently, there are no reliable genetic data on son (1924) claimed that results of crossing ex- morph determination for any gerrid species, periments with lacustris were consistent and models of the evolution of winglessness with a single-locus two-allele model, with the based on assumptions of a particular mode of allele for short wings dominant to the allele for inheritance derived from these studies are also long wings. Ekblom (1941, 1949), on the other questionable (see Discussion). hand, concluded that wing polymorphism was In this study we focused on the genetic in- 513 514 ZERA, INNES, & SAKS IN EVOLUTION 31 (1983)

TABLE 1. Frequency of the long-winged (MAC) morph in populations of Limnoporus canaliculatus sampled in 1978. Populations are as follows: PT = Petersham, Mass; PR = Pink Ravine, Stows, CT.; NI = Niantic, CT.; NISS = Nissequogue River State Park, Long Island, N.Y. Numbers in parentheses refer to sample sizes.

fl uences of morph determination in the wing- winged morph in October, 1978 = .04, N = polymorphic waterstrider, Limnoporus can- 251; Table 1). Adults were pair-crossed in the aliculatus. We wished to determine if the laboratory, and 1–2 generations reared at 24.5 inheritance of morph type could be explained ± 1.5 °C, 16 light:8 dark, constant photope- by the single locus (or supergene) model pro- riod, prior to the start of crossing experiments. posed by most workers for other gerrid species, This was done 1) to obtain virgin females, 2) or whether a more complex genetic explana- to eliminate potential maternal effects and tion was required. We used “split brood” exper- 3) to obtain a parental generation which had iments to determine if the mode of inheritance been reared under known conditions for cross- was different under different photoperiods as ing experiments. Mating pairs and develop- well as to assess the infl uence of photoperiod ing eggs were maintained at 24.5 ± 1.5 °C and on morph determination. 16L:8D constant photoperiod. 24.5 °C was chosen as the experimental temperature since MATERIALS AND METHODS L. canaliculatus had previously been shown to breed and develop in a manageable period of Species Studied time (approximately 35-40 days) at this tem- Limnoporus canaliculatus is a semi-aquatic perature (Zera, 1981b). Adults were stored at hemipteran which occurs on the surface of 13 °C, continuous light until needed to make ponds and streams. The distribution and life cy- appropriate crosses. Females were used in cle of this species has previously been reported crossing experiments only if they had emerged (Calabrese, 1979; Zera, 1981a). Populations of in containers without adult males. L. canaliculatus consist of various proportions For pair crosses, a male and female L. can- of fully winged, wingless, and (very rarely) aliculatus were placed in a 1 liter plastic bea- short-winged individuals. Morph frequencies ker half fi lled with distilled water containing exhibit considerable temporal and spatial vari- a small piece of styrofoam as a substrate for ation. For example, populations in New York, egg laying. Each day adults were fed freshly Massachusetts, and Connecticut sampled in killed housefl ies or Drosophila melanogas- July, 1978 consisted almost exclusively of ter. Approximately every third day, the sty- wingless individuals while the same popula- rofoam pieces containing attached eggs were tions sampled in September and October were removed from the beakers and placed in sepa- wing-polymorphic (Table 1). rate containers fi lled with distilled water. Each day newly hatched nymphs were counted and Crossing Experiments placed in appropriate containers. Nymphs were The breeding scheme used in crossing ex- fed freshly killed D. melanogaster until 3rd in- periments is given in Figure 1. Adults were col- star, and housefl ies or D. melanogaster there- lected at the Nissequogue River State Park, after. Both the amount and type of food fed to Smithtown, New York (frequency of the long- developing nymphs were kept constant. Except WING POLYMORPHISM IN WATERSTRIDER L. CANALICULATUS 515

ferred to as long-constant and short-decreasing, respectively. A decreasing photoperiod treat- ment was included since both change and di- rection of change of day-length are reported to infl uence morph determination in at least one gerrid species (Vepsäläinen, 1974a). As men- tioned previously, winged morphs are produced in the fall in L. canaliculatus when absolute daylength is short and decreasing. A “split brood” design, that is, rearing por- tions of progeny of a particular cross under dif- ferent treatments, was used to assess the infl u- ence of photoperiod on morph determination. This design is essential in comparing morph ra- tios of progeny reared under different treatments Figure 1. Experimental design used in crossing ex- since it eliminates confounding infl uences due periments of Limnoporus canaliculatus. All mating to intercross variation. In the split brood experi- pairs were maintained under 16L:8D constant photo- ments, newly hatched individuals were counted period. See Materials and Methods for details. daily and randomly divided between the two treatments. Individuals were added to a partic- ular container over no more than a six day pe- for the fi rst three days of development, den- riod. This was done to eliminate cannibalism of sity of progeny was kept between 3–7 individ- older nymphs on younger nymphs and for other uals per 100 cm2. We considered it unlikely reasons mentioned below. that differences in density for a short period Pair crosses were set up in two series (Fig. of time early in development would have an 1). In series I crosses, all parents had been effect on morph determination. No associa- reared under long-constant photoperiod, while tion between density and morph frequency in series II crosses, all parents had been reared was observed in samples of over 30 natu- under short-decreasing photoperiod. Broods of ral populations of L. canaliculatus suggest- series I crosses were split between long-con- ing that density does not infl uence morph de- stant and short-decreasing photoperiod or were termination (Zera, unpubl.). Jarvinen and reared exclusively under long-constant pho- Vepsäläinen (1976) argue that density per se toperiod. All progeny of series II crosses were is unlikely to infl uence morph determination reared under a short-decreasing photoperiod. in gerrids during any stage of development. Thus, in series I crosses, both parents and at Crosses were set up in all possible morph least one-half of their progeny were reared un- combinations in order to determine the con- der long-constant photoperiod, while in se- tribution of genetic differences to variation in ries II crosses, both parents and progeny were morph ratios. Progeny of crosses were raised reared under short-decreasing photoperiod. The under two different photoperiod regimes: 1) parents of series II crosses were progeny of se- 16L:8D, constant, or 2) 16L:8D light phase de- ries I crosses which had emerged as adults over creasing 15 minutes every 48 hours. The de- a six day period (13:15L to 12:30L). creasing photoperiod treatment, of necessity, Except for cross B, all progeny were placed differed from the 16L constant photoperiod in in the short-decreasing photoperiod treatment having both a decremental change in daylength during a 10 day period when photoperiod was as well as a shorter absolute daylength. Any decreased from 16L to 14:45L. Since progeny difference in morph frequencies between these of cross B hatched later than those from other two treatments could be due to either or both of crosses, their initial exposure to the decreasing these factors. Thus, the 16L constant and 16L photoperiod treatment began between 15 and decreasing treatments will henceforth be re- 14 hours of light. 516 ZERA, INNES, & SAKS IN EVOLUTION 31 (1983)

TABLE 2. Results of Series I crosses of Limnoporus canaliculatus. Parents had been reared under a long-con- stant photoperiod.

* MAC and APT morphs were scored according to the classifi cation of Brinkhurst (1959), MAC = long-winged morph, APT = wingless morph. Survivorship from egg hatch to at least fi fth instar for all crosses except “L” in long-constant photoperiod was greater than 30% (avg = 49%, range = 31%–71%). Survivorship of cross “L” in long-constant photoperiod was 23%. ** Two MIC ♂ produced in short-decreasing photoperiod; not included in the analysis. δ 1 MIC ♂ produced; not included in the analysis.

As mentioned above, siblings were added to experiments and was kept at 24.5 ± 1.5 °C for the same container over no more than a six day fi ve weeks prior to use. This enabled us to en- period. This design resulted in a temporal se- sure that the experiments were not contam- ries of replicates from each mating pair which inated with gerrids produced from fi eld laid could be compared for differences in morph ra- eggs. No L. canaliculatus hatched from fi eld tios. Temporal trends among replicates within a collected water. cross in either photoperiod would indicate in- fl uence of rank order of eggs or some correlate on morph determination. Differences in morph TABLE 3. Results of Series II crosses of Limnopo- ratios among replicates in only short-decreas- rus canaliculatus. Parents were F1 progeny of Series ing photoperiod would indicate the infl uence of I crosses reared under short-decreasing photope- riod. All progeny were reared under short-decreas- absolute daylength on morph determination. ing photoperiod. In this study we wished to rear individuals on a homogeneous reproducible medium such as distilled water. However, since constitu- ents in pond water (e.g., emerging insects used as food by gerrids) might affect morph deter- mination, results obtained using distilled wa- ter might be artifi cial and not directly related to morph determination in natural populations. Consequently, as a control, broods of sev- eral series I crosses reared under long-constant * MAC and APT morphs were scored according to the classifi cation photoperiod were split between pond and dis- of Brinkhurst (1959), MAC = long-winged morph, APT = wingless tilled water. Pond water was collected at the morph (see Materials and Methods) Survivorship from egg hatch to at least fi fth instar ranged from 42%-61% (avg = 51%) for all same locality as individuals used in crossing crosses. WING POLYMORPHISM IN WATERSTRIDER L. CANALICULATUS 517

TABLE 4. Photoperiod dependent association between sex and morph type in progeny of crosses of Limnoporus canaliculatus.

* Number in parentheses refers to the total number of adult progeny scored for morph type. ** n.d. = no data δ Results of two tailed sign test (ties ignored), H0 frequency of MAC ♂ = frequency of MAC ♀. For a particular cross, the number of progeny reported in this table does not necessarily correspond to that reported In Tables 2 and 3. Tables 2 and 3 contained unsexed progeny which were scored as fi fth instars as well as progeny scored as adults (see Materials and Methods).

The two common morphs were classifi ed three of the 858 adults scored in this study were according to the method of Brinkhurst (1959). MIC, the probability of misscoring a fi fth instar Adults whose wings extended past the begin- was very low. Data for micropterous progeny ning of the seventh abdominal tergite were clas- are given in Table 2. Because of their low fre- sifi ed as macropters (MAC) while wingless in- quency, they are not dealt with further. dividuals were classifi ed as apters (APT). Rare individuals with wing lengths not reaching RESULTS the anterior end of the seventh abdominal ter- Results of series I crosses are given in Table gite were classifi ed as micropters (MIC). Fifth 2. In each photoperiod treatment, the propor- instar nymphs which died before molting to tion of MAC progeny produced by each MAC adults were classifi ed as MAC or APT, depend- × MAC cross was greater than the proportion ing upon the presence or absence of wing pads. produced by each APT × APT cross. In general, Of hundreds of individuals observed in this and the frequency of MAC progeny of MAC ♂ × other studies, MAC adults always molt from APT ♀ and APT ♂ × MAC ♀ crosses ranged fi fth instars with wing pads while APT adults between the frequency produced by MAC always molt from fi fth instars with no wing × MAC and APT × APT crosses. Both MAC pads. There is a slight uncertainty in scoring and APT progeny were produced in nearly all fi fth instars because rare individuals may molt MAC × MAC and APT × APT crosses in both into micropters (MIC). However, since only photoperiods. 518 ZERA, INNES, & SAKS IN EVOLUTION 31 (1983)

Table 5. Number (Proportion) of MAC and APT progeny produced in “early” and “late” replicates of individ- ual crosses of L. canaliculatus in long-constant photoperiod. See Materials and Methods for additional details

Results of series II crosses are given in Ta- creasing photoperiod, also was the only cross ble 3. The same general pattern was observed which produced a higher frequency of MAC in these crosses as in series I crosses: the fre- ♂♂ than MAC ♀♀ in short-decreasing and quency of MAC progeny produced by each MAC ♀♀ than MAC ♂♂ in long-constant MAC × MAC cross was greater than that pro- photoperiod. duced by each APT × APT cross; both MAC No signifi cant difference in morph ratios and APT progeny were produced in each MAC was observed in any of the “split brood” ex- × MAC and APT × APT cross. periments in which progeny were reared on In eight of nine comparisons of the prog- distilled versus pond water, nor was any trend eny of the same cross reared under different observed (P > .1 for all χ2 homogeneity tests; photoperiods (split-brood crosses), the fre- average sample size per treatment was 27, sam- quency of MAC progeny was signifi cantly ple size ranged from 11 to 48). higher in the short-decreasing than in the long- A comparison among temporal repli- constant photoperiod (P < .05, two-tailed sign cates within a cross could be made for eight test). Sex ratio did not differ from 1:1 in either crosses whose progeny were reared in long- photoperiod. constant photoperiod (Table 5). In one cross A signifi cant association between sex and (B), the frequency of winged progeny was sig- 2 morph type, as well as a sex × morph × photo- nifi cantly greater in the later replicate (χ (1) = period interaction was observed (Table 4). Ig- 8.43, P < .005; see Table 5). Of the remaining noring ties, 10 of 12 crosses (B, E, 3, 14, 7, seven crosses, fi ve exhibited non-signifi cant 8, H, L, 12, 20) produced a higher frequency increases in the frequency of winged progeny of MAC ♀ than MAC ♂ progeny in short-de- in late versus early replicates. Although late creasing photoperiod (P < .05, two-tailed sign replicates tended to have a higher frequency test). Cross P, which was the only cross which of winged morphs, the difference over all produced a higher frequency of MAC than crosses was non-signifi cant (P > .1, two-tailed APT morphs in long-constant versus short-de- sign test; Table 5). WING POLYMORPHISM IN WATERSTRIDER L. CANALICULATUS 519

Similar comparisons were made among creasing photoperiod treatment exhibited no temporal replicates of crosses reared in short- signifi cant difference in morph ratios. How- decreasing photoperiod (crosses B, E, H, Q, 7, ever, more detailed studies are needed to ade- 20). In no case was a signifi cant difference ob- quately determine whether different genotypes served, nor was any trend observed (P > .1 for exhibit differential threshold responses to ab- all homogeneity tests; average sample size per solute daylength or to decremental change in replicate was 17, sample size ranged from 5– daylength. 27). Development of an individual L. canalic- ulatus into a MAC or APT adult is also sex- infl uenced, and the infl uence of sex is pho- DISCUSSION toperiod-dependent (Table 4). Associations between sex and morph type observed in short- The development of an individual Lim- decreasing photoperiod are consistent with noporus canaliculatus into a macropterous data obtained from natural populations of L. (MAC) or apterous (APT) adult is determined canaliculatus and other gerrid species. In the by a complex interaction between genetic and laboratory, under short-decreasing photope- non-genetic components. In both series I and II riod, nearly all crosses of L. canaliculatus pro- crosses, each MAC × MAC cross produced a duced a higher frequency of MAC ♀♀ than higher frequency of MAC offspring than each MAC ♂♂ (Table 4). Similarly, adults sam- APT × APT cross reared under the same pho- pled from 21 of 22 populations of this species toperiod (Tables 2 and 3). This demonstrates from Maine to Florida in September–October, a strong genetic component in morph deter- 1978–1979, exhibited a higher frequency of mination. Since both MAC and APT prog- MAC ♀♀ than MAC ♂♂ (P < .005, two-tailed eny were produced in both MAC × MAC and sign test; Zera, unpubl.). Moreover, 7 of 8 fall APT × APT crosses, the results are inconsis- and spring samples of G. lacustris exhibited a tent with a single-locus, two-allele model and higher frequency of MAC ♀♀ than MAC ♂♂ require more complex multilocus or multial- (Fig. 1 of Anderson, 1973). Since overwinter- lelic models. In this study, as in many others ing occurs in the adult stage in gerrids, indi- (reviewed by Harrison, 1980), photoperiod ex- viduals sampled in the spring had developed hibited a strong infl uence on morph determina- under short-decreasing photoperiod, like adults tion (Table 2). The higher frequency of winged sampled in the fall. Similar results for fall sam- progeny produced in the short-decreasing pho- ples of Gerris lacustris were also obtained by toperiod may have resulted from 1) the lower Vepsäläinen (1974b; Tables 8–10, 21–22). The absolute daylength or, 2) decrease in day- higher frequency of long-winged females pro- length experienced by developing nymphs. duced in natural populations of L. canalicula- With respect to absolute daylength, the differ- tus suggests that dispersal rates may differ be- ences between the two treatments are consis- tween the sexes. tent with seasonal changes in morph frequen- A higher frequency of winged males than cies in natural populations of L. canaliculatus winged females was produced in the long-con- (Table 1). A higher frequency of MAC individ- stant photoperiod treatment. This treatment is uals was observed in fall versus summer sam- most similar to early summer (late June–July) ples from each of four populations of L. can- with respect to absolute daylength. However, aliculatus (Table 1). Since absolute daylength as mentioned previously, populations of L. decreases after the summer solstice (June 22), canaliculatus are composed exclusively or al- adults emerging later in the season developed most exclusively of wingless morphs during under lower absolute daylengths than adults this time (Table 1). Thus, this sex × morph as- emerging earlier in the season. Slight differ- sociation, if it occurs in natural populations of ences in absolute daylength appeared to have L. canaliculatus, is unlikely to be biologically no infl uence on morph determination since important. “early” versus “late” replicates in the short-de- 520 ZERA, INNES, & SAKS IN EVOLUTION 31 (1983)

In long-constant photoperiod, no over- An understanding of the mode of inheri- all difference in morph ratios was observed tance of wing polymorphism is a prerequisite between “early” and “late” replicates (Ta- for understanding the genetic basis of dispersal ble 5). However, the highly signifi cant dif- ability. Knowledge of the genetic basis of dis- ference between replicates in one cross (B), persal ability is, in turn, critically important in coupled with the trend in the same direction the development of realistic models of the evo- in other crosses, suggests that such factors as lution of dispersal. For example, Roff (1975) rank order of eggs or maternal age may infl u- formulated a population model of dispersal in ence morph determination. Thus, these factors which the probability of dispersing was a func- should be taken into account in future studies tion of genotype. In this model, the mode of in- of wing polymorphism in gerrids. Infl uence of heritance of dispersal ability infl uenced both maternal age on morph determination has pre- the stability of the dispersal polymorphism and viously been documented in (MacKay, the effect of environmental stability on the pro- 1977). In this study, we did not investigate portion of dispersers maintained in the popu- the infl uence of other environmental variables lation. These results were obtained for a wide which are likely to infl uence morph determi- range of dispersal and survival probabilities. nation. For example, in many species, Thus, realistic genetic mechanisms of dispersal temperature is known to strongly infl uence the ability must be incorporated if dispersal models development of an individual into a winged are to accurately refl ect processes occurring in or wingless adult (Johnson, 1966; Lamb and natural populations. White, 1966; Schaefers and Judge, 1971). In all single locus models of wing polymor- Temperature also strongly infl uences the pene- phism, the allele for short wings (or wingless- trance of the vestigal winged trait in Drosoph- ness) is dominant to the allele for long wings ila melanogaster (Stanley, 1931). Thus, morph (Harrison, 1980). Since short-winged gerrids determination in L. canaliculatus is likely to are unable to fl y (Brinkhurst, 1958; Ander- be even more complex than we have demon- son, 1973), the only genotype capable of dis- strated. persal by fl ight is the homozygous recessive Since there are no reliable genetic data for long-winged morph. If adult females are not in- other wing-polymorphic gerrids, we do not seminated prior to dispersal, the allele for short know whether our results are typical for spe- wings must be introduced into newly colonized cies of this family or are unique to L. canalic- habitats either through mutation or by gene ulatus. Genetic studies of wing polymorphism fl ow dependent upon dispersal mechanisms have been done in only a few other insects other than fl ight (Harrison, 1980). Vepsäläinen (summarized in Harrison, 1980). In the coleop- (1978) recognized this problem and speculated teran Sitona hespidula, there is unambiguous that predispersal mating rather than mutation evidence that wing polymorphism is controlled was a more probable way of introducing the al- by a single locus (or supergene) with two al- lele for short wings into newly colonized habi- leles. Two other studies of wing polymor- tats. However, he provided no evidence for in- phism in coleopterans suggest a similar mode semination prior to dispersal and, in fact, stated of inheritance. However, studies in other that premigration mating is probably very rare groups (e.g., Orthoptera and Hemiptera) are in gerrids. If the mode of inheritance of wing either inconclusive or demonstrate that wing polymorphism is shown to differ from the sin- polymorphism is more complex than a sim- gle-locus, two-allele model mentioned above, ple Mendelian trait (Harrison, 1980). Although the introduction of the allele for winglessness the number of species examined is small, these into newly colonized habitats is no longer prob- data suggest that different taxonomic groups lematical. In this study, wingless progeny were may exhibit different modes of inheritance of produced in all crosses in which both parents wing polymorphism. were winged (Tables 2 and 3). WING POLYMORPHISM IN WATERSTRIDER L. CANALICULATUS 521

Consequently, alleles for winglessness may toperiod-dependent morph × sex association be transmitted into newly colonized habitats was also observed. In general, results were con- via normal dispersal by fl ight, even if females sistent with morph frequencies observed in nat- are not inseminated. Important aspects of many ural populations of L. canaliculatus, as well as hypotheses of the evolution of winglessness in other gerrid species. These results are at vari- gerrids are critically dependent upon the as- ance with previous reports of a single-locus (or sumption that wing polymorphism is controlled supergene), two-allele mode of inheritance of by a single locus with two alleles. For exam- this trait in gerrids and have implications for ple, Vepsäläinen’s model (1978) of the evolu- currently developing models of the evolution of tion of dominance of short wings (or wingless- winglessness and dispersal. ness) is based on this assumption. Hypotheses which attempt to explain the maintenance of wing polymorphism in natural populations of Acknowledgments gerrids are also dependent upon the assumption We wish to acknowledge the cooperation of of a monogenic inheritance of wing polymor- the staff of the Nissequogue River State Park phism. For example, several investigators have especially Mr. Greg Mertz. We also thank the evoked heterozygote superiority in attempting following persons who critically commented on to explain wing polymorphism as a balanced an earlier draft of the manuscript: Drs. Diane polymorphism (Brinkhurst, 1959; Guthrie, M. Calabrese, Hugh Dingle, Richard G. Harri- 1959). Heterozygote superiority is a suffi cient son, Kari Vepsäläinen and two anonymous re- condition for the maintenance of polymor- viewers. This study was supported by funds phism only for a single locus with two alleles from the Department of Ecology and Evolution, (Crow and Kimura, 1970). The hypotheses State University of New York at Stony Brook. mentioned above assume a particular mode of This is contribution number 430 from the De- inheritance for which there are no reliable data. partment of Ecology and Evolution, State Uni- If the mode of inheritance for other gerrid spe- versity of New York at Stony Brook. cies is similar to that of L. canaliculatus, the hypotheses are based on an erroneous assump- tion. Until more detailed information for both Literature Cited the genetic and environmental components of wing polymorphism become available, it will Anderson, N. M. 1973. Seasonal polymorphism and not be possible to formulate realistic models of developmental changes in organs of fl ight and the evolution of winglessness. reproduction in bivoltine pondskaters (Hem. Gerridae). Entomol. Scand. 4:1-20. Brinkhurst, R. O. 1958. Alary polymorphism in the SUMMARY Gerroidea (Hemiptera-). Nature 182:1461-1462. Genetic and environmental factors infl u- encing wing polymorphism in the waterstrider, ———. 1959. Alary polymorphism in the Gerroi- Limnoporus canaliculatus, were analyzed by dea (Hemiptera-Heteroptera). J. Anim. Ecol. 28: rearing progeny of single pair crosses in dif- 211-230. ferent photoperiods. There was a strong ge- Calabrese, D. M. 1979. Pterygomorphism in 10 netic component to morph determination, but Neartic species of Gerris. Amer. Midl. Natur. results were inconsistent with a single-locus, 101:61-68. two-allele mode of inheritance under either Crow, J. F., & M. Kimura. 1970. An Introduction to photoperiod. Photoperiod strongly infl uenced Population Genetics Theory. Harper and Row, morph determination, with a higher proportion N.Y. of long-winged morphs being produced in the Denno, R. F., & E. F. Grissell. 1979. The adaptive- short-decreasing photoperiod treatment. A pho- ness of wing-dimorphism in the salt marsh-in- 522 ZERA, INNES, & SAKS IN EVOLUTION 31 (1983)

habiting , Prokelisia marginata (Ho- Schaefers, G. A., & F. D. Judge. 1971. Effects of moptera: ). Ecology 60: 221-236. temperature, photoperiod, and host plant on Ekblom, T. 1941. Untersuchungen über den Flügeld- alary polymorphism in the aphid, Chaetosiphon imorphismus bei Gerris asper L. Not. Entomol. fragaefolii. J. Ins. Physiol. 17:365-379. 21:49-64. Stanley, W. F. 1931. The effect of temperature upon ———. 1949. Neue Untersuchungen über den vestigal wing in Drosophila melanogaster, with Flügelpolymorphismus bei Gerris asper Fieb. temperature-effective periods. Physiol. Zool. Not. Entomol. 29:1–15. 4:394-408. Guthrie, D. M. 1959. Polymorphism in the surface Vepsäläinen, K. 1974a. Determination of wing- water bugs (Hemipt.-Heteropt.:Gerroidea). J. length and diapause in waterstriders (Gerris Anim. Ecol. 28:141-152. Fabr., Heteroptera). Hereditas 77:163-175. Harrison, R. G. 1980. Dispersal polymorphisms in ———. 1974b. The life cycle and wing lengths of insects. Ann. Rev. Ecol. Syst. 11:95-118. Finnish Gerris Fabr. species (Heteroptera, Gerri- dae). Acta. Zool. Finn. 141:l-73. Jarvinen, O., & K. Vepsäläinen. 1976. Wing dimor- phism as an adaptive strategy in waterstriders ———. 1978. Wing dimorphism and diapause in (Gerris). Hereditas 84:61-68. Gerris: determination and adaptive signifi cance, p. 218-253. In H. Dingle (ed.), Evolution of In- Johnson, B. 1966. Wing polymorphism in aphids. IV. sect Migration and Diapause. Springer Verlag, The effect of temperature and photoperiod. Ento- N.Y. mol. Exp. App. 9:301-3 13. Zera, A. J. 1981a. Genetic structure of two species of Lamb, K. P., & D. White. 1966. Effect of tempera- waterstriders (Gerridae: Hemiptera) with differ- ture, starvation and crowding on production of ing degrees of winglessness. Evolution, 35:218- alate young by the cabbage aphid (Brevicoryne 225. brassicae). Entomol. Exp. Appl. 9:179-184. ———. 1981b. Extensive variation at the α-glyc- Mackay, P. A. 1977. Alate-production by an aphid: erophosphate dehydrogenase locus in species of the “interval timer” concept and maternal age ef- waterstriders (Gerridae: Hemiptera). Biochem. fects. J. Ins. Physiol. 23:889-893. Genet. 19:797-812. Poisson, R. 1924. Contribution a 1’étude des Hemip- teres aquatiques. Bull. Biol. Fr. Belg. 58:49-305. Roff, D. A. 1975. Population stability and the evo- lution of dispersal in a heterogeneous environ- ment. Oecologia 19:217-237. Corresponding Editor: G. B. Johnson