Phenotypic plasticity in Chrysoperla: genetic variation in the sensory mechanism and in correlated reproductive traits.
Phenotypic plasticity is a central element in the adaptation and diversification of organisms in variable environments (e.g., Bradshaw, 1965; Williams, 1966; Smith-Gill, 1983; Steams, 1983; Via and Lande, 1985; Dobson, 1989; Gillespie and Turelli, 1989; West-Eberhard, 1989; Harvell, 1990). Its evolution represents considerable biological complexity: it involves the evolution of a sensory mechanism (e.g., a switch) and at least two alternate physiological pathways, which ultimately are expressed as at least two alternative adaptive phenotypes. To unravel the steps in the evolution of phenotypic plasticity, and to understand its role in the adaptation and diversification of organisms, necessitates the demonstration of several specific genetic mechanisms--e.g., the genetic basis for responsiveness to environmental cues, genetic variation and covariation among the traits that determine the alternate phenotypes, and genotype-environment interactions among the component traits. Also, it requires an understanding of how the alternate phenotypes are expressed under natural conditions and how variation in their expression contributes to fitness.
Among insects, a ubiquitous and readily apparent form of phenotypic plasticity is the ability to enter diapause--the primary physiological adaptation to environmental heterogeneity in all the major insect groups (e.g., Lees, 1955; Danilevsky, 1961; Dingle, 1978; Tauber et al., 1986; Lee and Denlinger, 1991). Through the action of environmentally sensitive phenotypic switches, the trajectory of insect life cycles is shunted along alternate pathways--nondiapause development and/or reproduction under favorable seasonal conditions versus diapause-mediated dormancy and/or migration in anticipation of unfavorable conditions (Fig. 1). Diapause is not merely the absence of development and/or reproduction; in itself, it constitutes a complex adaptation that involves the temporal integration of a neuroendocrinologically mediated suite of behavioral, physiological, and morphological traits. Although the timing of diapause in the life cycle is species-specific, the consequences of diapause-mediated changes are pervasive throughout the organism's life history, i.e., prior to, during, and after the occurrence of dormancy (Fig. 1). Thus, diapause and its alternative (nondiapause development or reproduction) constitute ecologically important and physiologically tractable phenomena for examining how developmental switches and sets of adaptive traits are coordinated during the evolution of complex forms of phenotypic plasticity.
Experimental biologists have begun to investigate the question of genetic relationships among the numerous traits associated with the diapause syndrome (e.g., Dingle and Hegmann, 1982; Istock, 1983; Tauber et al., 1986; Gilbert, 1990). A few studies show or suggest genetic correlations between diapause-related switch mechanisms and important life-history traits that are expressed in the diapausing or nondiapausing phenotypes (e.g., Istock, 1980, 1983; Hegmann and Dingle, 1982; Groeters and Dingle, 1987; Mousseau and Roff, 1989). Such associations can impose genetic constraints on the evolution of phenotypic plasticity. In contrast, other components of life history (e.g., capacity for increase, replacement rate, fecundity, thermoregulatory wing-melanin patterns) may be unrelated genetically to the phenotypic switch, and thus they impose no constraints on the evolution of phenotypic plasticity (e.g., Hegmann and Dingle, 1982; Bradshaw and Holzapfel, 1990; Kingsolver and Wiernasz, 1991).
The studies above, and others (e.g., see Thompson, 1991), lead to further fundamental questions concerning the evolution of phenotypic plasticity: Does the evolution of complex phenotypic switches, such as those that control seasonal cycles, incur costs (as well as benefits) to the fitness of organisms? If so, do the correlated effects contribute to maintaining genetic variation in phenotypic plasticity? Does the level of intrapopulation variation in phenotypic plasticity differ among populations in a manner that attests to its adaptive nature? Do the components of complex phenotypic plasticity (i.e., the switch and the developmental pathways) evolve independently and subsequently become integrated, or do they share some common genetic bases? An understanding of the quantitative genetics underlying phenotypic plasticity is crucial to addressing each of these questions. Thus, our study focuses on the genetic structure of inter- and intrapopulation variation in the phenotypic switch and on the genetic relationship between the switch and fitness traits associated with it.
The Chrysoperla carnea species-complex
The Holarctic Chrysoperla carnea species-complex of green lacewings (Insecta: Neuroptera: Chrysopidae) provides a fine system for examining the questions above. Its seasonal cycle involves a well-defined switch that governs whether adults undertake reproduction or enter diapause-mediated aestival dormancy (see Fig. 1). It is noteworth that, in addition to aestivating, adults overwinter in a photoperiodically regulated diapause, which also involves a genetically variable switch mechanism (Tauber and Tauber, 1982). However, the two switches are independent, and our present study concentrated only on aestival diapause and voltinism (the number of generations per year).
In North America, the C. carnea species-complex currently consists of two described "species": Chrysoperla carnea ( = C. plorabunda of some authors) and Chrysoperla downesi. In eastern North America the two "species" are reproductively isolated, widely distributed, and sympatric; moreover, they consistently differ in a number of traits (habitat association, adult coloration and markings, larval markings, and courtship songs) (e.g., Tauber and Tauber, 1976, 1977, 1986, 1987; Henry, 1985). In contrast, populations in western North America exhibit considerable geographic variation in these and other traits, and the pattern of variation is much more complex (e.g., Tauber and Tabuer, 1973a, 1973b, 1982, 1986; Henry and Martinez Wells, 1990). Whether this variation characterizes distinct cryptic species remains open to question (Tauber and Tauber, 1987; Martinez Wells and Henry, 1992). Contrary to Henry (I 985; also Henry and Martinez Wells, 1990), we have no difficulty hybridizing individuals from diverse populations; consequently we retain the older name, C. carnea, and until the systematic relationships within the group are defined we treat the entire assemblage in North America as a species-complex (Tauber and Tauber, 1987).
Phenotypic Variation in the Switch. -- In eastern North America, variation in the diapause-mediated seasonal cycles is partitioned between the two reproductively isolated "species" (Tauber and Tauber, 1986). C. carnea is consistently multivoltine throughout the region; reproduction occurs under the long daylengths of late-spring and summer, and there are two or three overlapping generations per year (Tauber and Tauber, 1982, 1986). In contrast, C downesi is characterized by a photoperiodically mediated univoltine life cycle throughout the region; reproduction occurs only when individuals experience a sequence of short daylengths followed by long daylengths (Tauber and Tauber, 1976, 1986). As a consequence of this short-day/long-day requirement, reproduction typically occurs only during the spring, and the resulting offspring (which develop under the long days of late spring and early summer) enter a reproductive diapause that persists through summer, autumn, and winter.
In comparison to C carnea and C downesi in eastern North America, western populations have considerably more complex seasonal cycles (Tauber and Tauber, 1986). First, most western populations are polymorphic in their seasonal cycles and in the photoperiodic requirement for reproduction (long daylengths versus the short-day/long-day requirement). Second, in most western populations some adults can perceive and respond to the presence of the prey of their larvae. Although adults are nonpredacious, honeydew and pollen-feeders (Hagen et al., 1970), the presence of larval prey acts as a cue, or token stimulus, to promote aestival reproduction, whereas absence of prey triggers aestival diapause (Tauber and Tauber, 1973b). Third, the pattern of variation among geographic populations in the west suggests genetic and/or physiological correlations between the photoperiodic and prey components of the switch, as well as with other life-history traits (Tauber and Tauber, 1986).
Genetic Variation in the Switch. -- Two independent genetic systems underlie the variation in the photoperiodic and prey-associated responses (Tauber and Tauber, 1987). Alleles at two unlinked autosomal loci determine the photoperiodic responses, with the short-day/long-day requirement being dependent upon recessive alleles at both loci; a dominant allele at either locus results in long-day reproduction. In contrast, a polygenic mechanism with a threshhold for expression controls adult responses to the presence of larval prey. Furthermore, a dominant allele at either locus governing the photoperiodic responses suppresses the expression of responsiveness to prey. Variation in photoperiodic responses is based on homologous genes in eastern and western populations, and variation in adult responsiveness to prey is based on homologous genes throughout western populations, and it is absent from eastern populations (Tauber and Tauber, 1987).
In summary, an individual may assume any one of the following phenotypes: either (1) long-day reproducer (multivoltine, spring or summer breeder, dominant) or (2) short-day/long-day reproducer (recessive). If (2), then it can be either: (a) responsive to prey (spring breeder, facultative summer breeder) or (b) nonresponsive to prey (univoltine, spring breeder, dormant in summer).
Given the above, the next step is to elucidate relationships among the sets of traits that comprise the alternative phenotypes. Of special interest is the relationship between the diapause-regulating switch mechanism and the reproductive traits of the nondiapause phenotype, because genetic correlations between these traits should strongly influence both the trajectory and the rate of evolution of the diapause syndrome (e.g., Istock, 1983; Via and Lande, 1985; Thompson, 199 1) (Fig. 1). Therefore, we focused on four issues: (1) We examined whether the response patterns that constitute the phenotypic switch (the photoperiodic requirement for reproduction and the ability of adults to respond to prey presence) vary among families within populations. (2) We investigated whether differences exist among families in two life-history traits that influence the fitness of nondiapausing reproductive individuals, i.e., the preoviposition period and fecundity. (3) We determined whether there are correlations among familes between the two responses that make up the phenotypic switch and also between the switch and reproductive traits that influence the fitness of the nondiapausing phenotype. (4) We considered how variation in the various components of the phenotypic plasticity may be subject to selection under natural conditions.
MATERIALS AND METHODS
Ideally, quantification of the genetic relationship between phenotypic traits involves a half-sib design or selection experiments. Our system does not lend itself to such analyses because of difficulties associated with the laboratory rearing and the reproductive biology of the species. After being reared in the laboratory, C. carnea individuals may show reduced vigor; moreover, male fertility decreases unpredictably and irreversibly with serial matings (e.g., Henry and Busher, 1987). Thus, our investigation used a full-sib design -- a justifiable approach because reciprocal hybridization and backcross tests do not indicate maternal effects on the traits under study (e.g., Tauber and Tauber, 1987).
All families in our study originated from individual, field-collected females. These families probably represent full-siblings because C. carnea females usually do not remate until fertile egg production diminishes, presumably following sperm depletion (Henry and Busher, 1987). Because siblings were separated into individual vials before hatch, common family environments presented no problem.
We examined populations having a broad range of phenotypic variation in their seasonal cycles (Table 1). Two populations represented the reproductively isolated "species" from northeastern United States (the multivoltine C. carnea that inhabits fields and mixed hardwood forests -- IC, and the univoltine C. downesi that inhabits coniferous forests -- ID). Seven populations were from California; these typify the range of variation expressed by geographic populations in western North America: the Coast Range (SC), the Central Valley (DA), the foothills of the Sierra Nevada (PL), the southern Cascade Range (LS), the southwestern Sierra Nevada (AT and KC), and the eastern slope of the Sierra Nevada (TP). The number of families varied among populations depending upon our success in collecting and maintaining individuals from the field and on the availability of resources (Table 1). Voucher specimens of adults and larvae are deposited in the Cornell University Insect Collection (Lot 1158). [TABULAR DATA 1 OMITTED]
For each population, our goal was to quantify the variation in diapause-related traits among individuals from several families under two treatments of photoperiod and two treatments of prey that induce the full expression of variation in aestival diapause observed in previous studies (e.g., Tauber and Tauber, 1986). The four experimental regimens were: (i) constant long day -- individuals experienced 16 hr light (L):8 hr dark (D) for their entire lives; (ii) increase in daylength -- individuals experienced 10 hr L: 14 hr D from egg to spinning of the cocoon and 16 hr L:8 hr D from spinning onward; (iii) constant availability of larval prey -- adults received aphids plus a highly proteinaceous diet from the day of adult emergence onward; and (iv) without larval prey -- adults were provided a highly proteinaceous diet but no aphids. The two regimens, with and without aphids, were tested against the two photoperiodic treatments to give four treatments. Responses measured were (i) the induction of aestival diapause and (ii) the duration of the preoviposition period in individuals that did not enter diapause. To examine further the relationship between the phenotypic switch and the fitness of the reproductive phenotype, we determined the lifetime fecundity for nondiapausing females in one population (TP) under two conditions (an increase in daylength, with and without aphids). This population was chosen because it has high levels of polymorphism for both the photoperiodic and prey responses; the other populations were not investigated because of the large time commitment required for such tests.
Rearing, Collection of Data, and Diagnosis of Diapause
Details of larval rearing conditions were presented previously (Tauber and Tauber, 1976). Because it was impossible to experiment with all the populations simultaneously, tests were conducted in the late summer and fall of 1981-1989, except those on the eastern populations (IC and ID), which took place during the late winter and spring of 1990. To prevent cannibalism, all larvae were individually reared. Because C. carnea males may vary in the amount of sperm transferred during mating (Henry and Busher, 1987), they may contribute to overall fecundity; thus we paired females with their brothers to prevent the male component from distorting family differences. Adults were provided with water and a proteinaceous diet (a 1:1:1:1 volumetric mixture of Wheast[R], protein hydrolyzate of yeast, sugar, and honey). The prey regimen consisted of green peach aphids, Myzus persicae, on cabbage leaves.
Individuals undergoing aestival diapause were recognized by the following criteria: lack of reproduction and the assumption of a plump, waxy appearance, which indicates the lack of ovarian development and development of fat tissue. In some populations aestivating adults also undergo characteristic color changes (Tauber and Tauber, 1986). The preoviposition period of nondiapausing females constitutes the time from emergence to oviposition of the first fertile egg; this period included mating. The temperature was maintained at 24 [+ or -] 0.5 [degrees] C for all rearing and all experiments.
The induction of diapause is an all-or-nothing response, and data from each family encompass a single percentage value. Heterogeneity among familes in the proportional occurrence of aestival dispause was analyzed with three-way tables using log-linear models (unreplicated data) (Sokal and Rohlf, 1981). For each population, we considered all possible interactions, as well as individual effects of three variables (photoperiod, family, and sex, or prey regimen, family, and sex). The calculations were made with the GLIM Statistical Package (Royal Statistical Society of London).
Nondiapause preoviposition periods were examined for heterogeneity with a mixed model four-factor ANOVA using the GLM procedure (SAS Institute, 1985). Photoperiod, prey regimen, and source population were completely crossed main effects, and family groups were nested in populations and crossed with the other factors; treatments were considered fixed effects, whereas population and family were random. Variance within conditions and within populations was examined in separate two-way ANOVAs with population and family (within condition) and photoperiod and prey presence (within populations) as factors. The data on the total lifetime fecundity of the TP population were subjected to two-way ANOVA with prey and family as factors. The preoviposition periods were [log.sub.10] transformed as is often required of temporal data, and data on fecundity (count data) required square-root transformation; in both cases transformation made the distributions of the variables closer to normal.
In cases where the same tests were made on data from several populations, data sets for each population were considered to be independent, and alpha levels were 0.05. When tests were repeated on a single population, alpha levels were corrected for multiple tests (Bonferroni's adjustment for multiple tests; Snedecor and Cochran, 1989, p. 167).
Variation among Families
The Short-dayllong-day Requirement for Reproduction. -- Of the seven populations tested for responses to photoperiod, only the multivoltine population from eastern North America (IC) had no aestival diapause within any family under either constant long daylength or with an increase in daylength (Table 2). In the six other populations, a variable number of individuals within the families tested entered aestival diapause under a regimen of constant long day (no prey) (Table 2). When the insects experienced an increase in daylength, the percentage of individuals entering diapause was very low in all populations and all families. [TABULAR DATA 2 OMITTED]
Under constant long daylength (in the absence of prey), most western populations showed significant among-family heterogeneity in the incidence of aestival diapause ([two females and [two males]: PL, AT, TP; [two males] only: LS; G test, [Alpha] = 0.05) (Fig. 2). After an increase in daylength (in the absence of prey) these same families exhibited very low incidences of diapause and no significant among-family variation. Within families pooled from all populations (37 < N < 65), there was a highly significant correlation between males and females in the incidence of the short-day/long-day requirement for reproduction (Spearman rank correlation, one-tailed test, [r.sub.s] = 0.904, P [is less than or equal to] 0.0001).
In the three-way analyses of the incidence of aestival diapause, all western populations combined, and two individual populations from the Sierra Nevada Range (AT, TP), showed significant photoperiod x family interactions in the incidence of aestival diapause, and both photoperiod and family had significant individual effects on the two populations from the Central Valley and foothill regions (DA, PL) (Table 2, Fig. 2). Only one western population, that from the southern Cascade Range (LS), did not show significant variation among families; however, there was a strong effect of photoperiod and sex in this population.
In summary, the incidence of the short-day/long-day requirement for reproduction varies among families within western populations, with the expression of this variation being strongly dependent on photoperiod.
Adult Responsiveness to Prey Presence. -- Adult response to prey presence was only detected under constant long daylengths; aphids had no significant effect on the incidence of reproductive diapause when experiments were conducted under an increase in daylength (Table 2). In all populations combined, the percentage of individuals entering diapause under constant long daylengths was significantly affected by prey x family interactions (three-way analysis: G = 328.4, df = 51, P < 0.001). Furthermore, the variation among families in all populations was greater for males than females, and the responsiveness of males and females to prey was very highly correlated across families (Spearman rank correlation, one-tailed test, [r.sub.s] = 0.852, P [is less than or equal to] 0.0001).
One of the two eastern populations (the one from the multivoltine C. carnea -- IC) showed no reproductive diapause whether aphids were present or not (Table 2). Diapause in males of the other eastern population (the one from the univoltine C. downesi -- ID) was not affected by prey presence, whereas females had a significantly higher incidence of diapause in the presence of aphids than in the absence (Table 2). We have no explanation for the unusual response of the females.
The two western populations from the Central Valley and foothill regions (DA and PL) had very low levels of aestival diapause whether aphids were present or not. One of these (PL) was tested for responsiveness to prey, and it showed significant prey, sex, and family effects on the incidence of diapause (Table 2).
Four other western populations [the coastal (SC), the Cascade (LS), and two Sierran (KC and TP) populations] had relatively high degrees of adult responsiveness to prey; i.e., when aphids were provided, the average incidences of reproductive diapause were reduced by as much as 61% (Table 2). Visual assessment and three-way analyses indicated highly significant prey x family effects on the incidence of aestival diapause in three of these populations (SC, LS, and TP) (Table 2, Fig. 3). Diapause in the fourth population (KC) had significant family and prey x sex effects, and as stated above, diapause in the foothill population (PL) showed significant prey, sex, and family effects.
In general, western populations harbor considerable among-family variation for adult responsiveness to prey, and the expression of this variation is dependent on photoperiod and prey regimen.
Nondiapause Preoviposition Period. -- Among the eight populations tested, the nondiapause preoviposition period varied greatly -- from an average of four days in the multivoltine population from eastern North America (IC) to an average of twelve days in the population from the eastern Sierra Nevada (TP) (Table 3A). In general, the nondiapause preoviposition periods were not influenced by either photoperiodic condition or prey regimen, except in the LS and ID populations which showed significant photoperiodic effects (Table 3A). [TABULAR DATA 3 OMITTED]
The nondiapause preoviposition period varied significantly among populations and among families within most populations (Table 3A). For example, following an increase in daylength, the average duration of the nondiapause preoviposition period ranged from 3.7-4.4 days in families of the eastern multivoltine population (IC) and from 9.3-20.1 days in families from the polymorphic population from the eastern Sierra Nevada (TP). The ANOVA on the entire data set across photoperiod and prey regimen, as well as population and family, indicated significant population x photoperiodic regimen, population x aphid regimen, family x photoperiodic regimen, and family x prey regimen interactions on the length of the nondiapause preoviposition period (Table 3B).
In summary, there is both population-and family-specific variation in the relative length of the nondiapause preoviposition period under the photoperiodic and prey conditions tested.
Fecundity of Nondiapausing Females. -- The average lifetime fecundity of females within families from the eastern Sierra Nevada population (TP) ranged widely, but was significantly influenced by prey presence or absence and by family identity (Table 4A). However, the family x prey regimen interaction term was not significant, and thus we conclude that females from families with high fecundity in the presence of prey also have high fecundity in their absence (Table 4B). Mean family fecundity was greater in the absence ([Chi] [bar] [+ or -] SE = 628 [+ or -] 229 eggs) than in the presence of aphids ([Chi] [bar] [+ or -] SE = 383 [+ or -] 152 eggs).
Correlations among Traits
Among the polymorphic populations from western North America, the individual components of the switch were highly correlated with each other. For example, in both males and females the incidence of the short-day/long-day requirement for reproduction was significantly correlated with adult responsiveness to prey (Spearman rank correlations, one-tailed test, all populations combined, [r.sub.s] = 0.370 - 0.459, P [is less than or equal to] 0.01 for all four possible comparisons).
In addition, the components of the switch were correlated with the nondiapause life-history traits that we examined. For example, across all populations, families that had high incidences of the short-day/long-day requirement for reproduction tended, on average, to have long nondiapause preoviposition periods (Table 5A). This correlation was present when data from families were pooled across populations, and also when families were tested within individual populations. The r, values were positive in four of five populations when the mean nondiapause oviposition periods were correlated against the incidence of the short-day/long-day requirement in females; two were significant. Correlations were also positive (but not significant) in five out of six populations when the mean preoviposition period was tested against the incidence of the short-day/long-day requirement in males. The Fisher method for combining probabilities from independent tests of significance (Sokal and Rohlf, 1981, pp. 779-782) indicates that the correlation between the short-day/long-day requirement for reproduction and a lengthened nondiapause preoviposition period is a general characteristic of the C. carnea species-complex ([[Chi].sup.2] = 46.17, df = 11, P < 0.001).
Among-family variation in adult responsiveness to prey showed a similar pattern of correlation with the mean nondiapause preoviposition period (Table 5A). For example, when families from all populations were pooled, the correlations were positive and highly significant. Furthermore, the correlations among families within populations were positive for male and for female responsiveness in all populations tested; one of these correlations was significant for female responsiveness. The correlation between adult responsiveness to prey and a lengthened nondiapause preoviposition period is a general characteristic of the species-complex Fisher method for combining probabilities from independent tests of significance, [[Chi].sup.2] = 29.89, df = 8, P < 0.001).
Among families in the eastern Sierra Nevada population (TP), the mean fecundity of nondiapausing females was positively related to the incidence of the photoperiodic requirement for reproduction. The correlation was significant whether mean family fecundity was measured under conditions of aphids present or aphids absent, and for both female and male responsiveness to photoperiod (Table 5B). Among families, mean lifetime fecundity in the presence and absence of prey was also positively correlated to responsiveness to prey, but the relationship was significant only for females.
The seasonal responses of adult Chrysoperla are critical to the fitness of their offspring because once the egg is laid, development proceeds along a pathway that is uninterruped by dorinancy until the adult stage; preimaginal stages do not aestivate. As aresutl, the phenotypic switch has a profound influence, not only on adult survivorship and the seasonal timing of reproduction or aestival dormancy, but also on voltinism and larval survivorship and development. Ultimately, the switch may also have an important role in the seasonal isolation of sympatric populations (Tauber and Tauber, 1977).
The evolution of phenotypic plasticity, e.g., the evolution of alternate seasonal cycles in insects, requires the existence of genetic variation in the responses of organisms to environmental cues. Furthermore, genetic correlations between these responses and other life-history traits, as well as environmental or genetic effects on the expression of the variation, can influence the evolution of phenotypic plasticity because they may constrain (or promote) the evolution of the switch (e.g., see Via and Lande, 1985; Thompson, 1991). In Chrysoperla seasonal plasticity is based on genetically determined responses to photoperiod and prey that vary both among geographic populations and among families within populations. Moreover, as discussed below, there are significant correlations between the components of the switch and between the switch and other life-history traits that affect fitness. These correlations appear to place constraints on the expression and evolution of the switch.
The significant environment x family interactions in our tests indicate that Chrysoperla harbors considerable genetic variation in its responses to environmental heterogeneity, i.e., in the photoperiodic and prey components of the phenotypic switch that regulates aestival reproduction versus dormancy. Given that we conducted our tests with full sibs, there is a possibility that the results include nongenetic maternal or paternal effects. But we consider that this is highly unlikely because our previous experiments, which involved reciprocal crosses and backcrosses with four populations under a variety of environmental conditions, gave no hint of such effects on the key traits of interest-responses to photoperiod or prey; nor did they indicate any maternal or paternal effects on related reproductive traits, such as the length of the preoviposition period in nondiapausing females (Tauber and Tauber, 1987).
Expression of Genetic Variation in the Phenotypic Switch
Photoperiodic Component of the Switch. -- The expression of genetic variation in aestival diapause in all the Chrysoperla populations was highly dependent on photoperiod, and the very low incidence of reproductive diapause in all families subjected to an increase in daylength appears to eliminate family identity as a major source of variation in the reproductive activity of vernal adults (Fig. 2). Thus, photoperiod places a major, seasonal restriction on the evolution of the switch. However, under constant long daylengths, families show considerable differences in their reproductive states, and it appears that natural selection acts directly on the photoperiodic component of the switch during a relatively narrow period around the summer solstice -- late spring and early summer when daylengths are long and relatively constant. Despite this short period, the vital and enduring effects of diapause on the life history of the organism means that the effects of selection at this time can be expected to be very strong.
Prey Component of the Switch. -- There are significant environmental as well as genetic constraints on the natural expression of variation in adult responsiveness to the prey of their larvae. In all of the populations we have studied including the ones here, prey presence influences the aestival phenotype only under constant long daylength -- not when the organisms experience an increase in daylength (see also Tauber and Tauber, 1986). As a result, selection on the prey component of the phenotypic switch has the same seasonal restrictions as the photoperiodic component; it can occur only during late spring and early summer when daylengths are long but not increasing at a fast rate. It is noteworthy that among-family variation in the prey component is expressed whether prey are present or absent. Thus, during the photoperiodically permissive period, selection can act on adult responsiveness to prey presence whether prey are abundant or scarce.
It is also noteworthy that during the photoperiodically permissive period direct natural selection can act on only a proportion of the variation in adult responsiveness to prey. Adult responsiveness to prey, a polygenic, threshold trait, is expressed solely in individuals that are homozygous for the recessive alleles that produce the short-day/long-day requirement for reproduction (Tauber and Tauber, 1987). Variation for the trait remains unexpressed and unselected in individuals that carry a dominant allele for long-day reproduction (=hidden variation of Istock, 1983). Thus, selection on the prey component of the phenotypic switch is strongly dependent on genotype, as well as on photoperiod.
Given the genetic, as well as environmental constraints on the expression of adult responsiveness to prey presence, it appears that the trait has evolved as a modulator of the phenotypic switch's photoperiodic component. By modifying the effects of a relatively rigid, photoperiodically induced aestival reproductive dormancy, adult responsiveness to prey allows the insects to take advantage of locally or temporarily abundant prey.
Geographic Variation in the Genetic Structure of the Phenotypic Switch
The genetic structure of variation in the phenotypic switch differs markedly between eastern and western populations. In eastern North America, variation in the photoperiodic responses is, in large part, partitioned between the two reproductively isolated "species," neither of which express phenotypic plasticity in their seasonal cycles in nature. In C. carnea, the monomorphism for long-day reproduction that we found in the IC population appears to be general throughout the east (and midwest). We have never observed aestival diapause in eastern or midwestern C. carnea -- either in the field or under long-day conditions in the laboratory (e.g., Tauber and Tauber, 1977, 1982, 1986, 1987), and we conclude that eastern and midwestern C. carnea lack genetic variation for the photoperiodic component of the switch. Our earlier hybridization and backcross tests (Tauber and Tauber, 1987) suggest that these populations lack variation for the prey component of the switch as well.
Although the ID population of C. downesi in the east expressed some variability for aestival diapause in our current test, it never did so in previous tests -- virtually all individuals reared from many field-collected females entered diapause under L:D 16:8 (no prey) (e.g., Tauber and Tauber, 1976, 1986, 1987). We suspect that some modifiying factor (e.g., very sensitive thermal or humidity effects) may be responsible for the variation in the laboratory. It is unlikely that such variation is expressed in nature; we have never collected nondiapause C. downesi adults in eastern North America except in the spring and very early summer.
In contrast to eastern Chrysoperla, the species-complex in western North America expresses phenotypic plasticity in the seasonal cycle, and the degree of phenotypic plasticity varies geographically. Among the western populations, the variation appears to be strongly associated with the variability and predictability of local environments (e.g., Tauber and Tauber, 1973a, 1982). For example, in California's Central Valley and foothill regions, where irrigation practices may result in relatively stable and predictable populations of prey, Chrysoperla populations (e.g., DA, PL) have low incidences of photoperiodically determined univoltinism and low responsiveness to prey. Populations from higher elevations (e.g., LS) and the eastern and southern regions of the Sierra Nevada (e.g., KC, AT, TP), where prey conditions are likely to be very irregular, are more prone to photoperiodically determined aestivation and they carry more variation for responsiveness to prey.
At first glance, the high propensity for aestival diapause in the Coastal population (SC) is somewhat unexpected because the climatic conditions at this locale are relatively moderate. Nevertheless, prey levels tend to be low in the Strawberry Canyon area during the summer and the between-year variability in plant growth (and prey levels) appears to be very high. These factors would select for high levels of aestival diapause and for responsiveness to prey.
Genetic Correlations between the Switch and the Phenotype of Nondiapausing Reproductives
Do the genes that underlie Chrysoperla's phenotypic switch have correlated (positive or negative) effects on the nondiapause phenotype (see Fig. 1)? Our results demonstrate that, under laboratory conditions they have a negative effect on the prereproductive period of nondiapausing females and a positive effect on fecundity, but are the correlations expressed in nature?
Prereproductive Period. -- We assume that under most circumstances in nature, lengthening the nondiapause prereproductive period would reduce fitness significantly. During this period, when adults forage for food and seek mates, they are vulnerable to a variety of mortality factors, including predation. Therefore, it is important to consider how the preoviposition period of Chrysoperla is expressed in nature.
Under natural conditions the univoltine C. downesi from eastern North America (e.g., ID) does not have a nondiapause preoviposition period. Only postdiapause adults reproduce, and because of the short-day/long-day requirement that typifies this "species," all of the offspring enter reproductive diapause. Thus, variation in the nondiapause preoviposition period of the ID population is not subject to natural selection.
In contrast, the nondiapause preoviposition period is very important to the fitness of multivoltine or partially multivoltine populations, and these populations would be expected to express the correlated trait in nature. That is, populations (or families) with high incidences of the photoperiodic and prey components of the phenotypic switch should have long nondiapause preoviposition periods. This appears to be the case. Over a 20-year period, our July and August samples from coastal and Sierra Nevada populations (SC and TP, both of which have high incidences of the photoperiodic and prey requirements for nondiapause reproduction) have always yielded a substantial proportion of prereproductive adults; this suggests a relatively long prereproductive period. Alternatively, we rarely observed the occurrence of prereproductive adults in the multivoltine eastern C. carnea (which does not have either the photoperiodic or prey requirements for aestival reproduction); this pattern suggests that reproduction begins relatively quickly after nondiapausing eastern C. carnea adults emerge.
From the above comparisons, we conclude that among-family variability in the nondiapause preoviposition period is likely to be expressed in multivoltine populations. Furthermore, the benefits derived from carrying genetic variability for the phenotypic switch must be balanced against the costs of lengthening the prereproductive period.
Because of the close correlation between the two response patterns (i.e., short-day/long-day requirement for reproduction and responsiveness to prey), our data do not allow us to attribute the lengthening of the nondiapause preoviposition period to specific genes that control one or the other of the response patterns. Indeed, the ability to respond to prey occurs solely in individuals that require an increase in daylength for reproduction. Nevertheless, the univoltine eastern population (ID) provides some insight into the issue. This population has a very high incidence of the short-day/long-day requirement for reproduction and virtually no ability to avert diapause in response to prey presence. Its nondiapause preoviposition period (in the laboratory) is considerably longer than that of populations that have both very low incidences of the short-day/long-day requirement for reproduction and also no adult responsiveness to prey (e.g., IC, DA). Thus, acquiring the short-day/long-day requirement for reproduction (separate from responsiveness to prey) probably results in a lengthened nondiapause preoviposition period. Moreover, the addition of adult responsiveness to the presence of larval prey appears to lengthen the period further -- by at least two days in the LS population and up to seven days in the TP population (Table 3).
Fecundity. -- In contrast to the nondiapause preoviposition period, lifetime fecundity (in the 10 families from the eastern Sierra Nevada, TP) is significantly and positively correlated with the incidence of the short-day/long-day requirement (males and females) and adult responsiveness to prey (females) (Table 5B). We do not know if this positive correlation is significant to the fitness of nondiapausing reproductives in nature because prey regimen (Table 4B) and other environmental factors, e.g., the adult diet (Hagen et al., 1970), have a strong influence on fecundity. It is possible that environmental conditions during the reproductive period reduce or obscure family effects (e.g., see Trexler and Travis, 1990). If, however, the effects we observed in the laboratory are expressed in nature, they could be a strong positive factor in the evolution of the switch.
Our findings that the fecundity of TP females is significantly lower in the presence than in the absence of aphids appears problematic. This response may be an artifact of the experimental conditions; the very close confinement of the adult lacewings in cages with aphids may have disrupted oviposition. Nevertheless, the results indicate that fecundity is sensitive to environmental effects, and they lead us to be cautious in assigning evolutionary significance to the correlations between fecundity and components of the switch.
Postdiapause Traits. -- In addition to the patterns above, the evolution of the phenotypic switch could be influenced by genetic correlation with traits in the postdiapause phenotype (Fig. 1). For example, in Chrysoperla, correlations between the switch and postdiapause fecundity and fertility are expected to alter the evolution of the seasonal plasticity. In some, but not all insects that have been examined, there is a loss of reproductive potential during diapause (e.g., summary in Tauber et al., 1986); whether such losses are heritable and genetically corelated with the swith mechanism is presently unknown.
Maintenance of Variation in the Phenotypic Switch
Under certain restrictive conditions, temporal variation in selection can contribute significantly to maintaining genetic variability (Hedrick, 1983). Although the limitations are especially severe for traits with absolute dominance (Levins, 1968; Maynard Smith, 1982), the addition of environmentally sensitive modulators of the dominant-recessive trait can reduce their effects (Gillespie and Turelli, 1989). Such could be the case in Chrysoperla where responsiveness to prey modulates the dominant-recessive photoperiodic component of the phenotypic switch. For example, none of the 41 North American populations that we have examined express variability in the dominant-recessive photoperiodic component of the switch without also having the modulating effects of adult responsiveness to prey (Tauber and Tauber, 1986).
Negative correlations between the switch and traits in the nondiapause phenotype may also help maintain genetic variability in a phenotypic switch mechanism; such associations lead to temporal variability in the direction of selection (e.g., see Hedrick, 1983; Gillespie and Turelli, 1989). For example, in Chrysoperla, selection on the phenotypic switch may be positive during midsummer, when individuals that are responsive to environmental conditions are favored, but negative during spring when individuals with the correlated trait (a lengthened nondiapause preoviposition period) are at a disadvantage. Thus, the direction of selection varies seasonally, and the polymorphism is maintained.
By contrast, high levels of positive "autocorrelation" between subsequent environments tend to lead to diversification (fixation of gene frequencies at 0.0 or 1.0). Such appears to be the case in eastern North America, where C. carnea and C. downesi probably experience different selective regimens -- one for multivoltinism, the other for univoltinism; for each of the "species" the direction of selection is consistent across subsequent generations. In C. carnea, selection for multivoltinism simultaneously favors the correlated shortened preoviposition period in nondiapausing individuals.
In C. downesi, the evolution of univoltinism masks the expression of any correlated negative effects on the nondiapause preoviposition period; thus, univoltine populations (and individuals) can carry the phenotypic switch without a reproductive cost. Partitioning of the variability among two monomorphic, reproductive isolates in eastern North American probably reflects disruptive selection for seasonal cycles adapted to two different, but seasonally predictable patterns of prey occurrence. In C. carnea's field and meadow habitats, regularly abundant levels of prey in spring and summer probably favor multivoltine spring and summer breeding; whereas spring breeding is favored in C. downesi's coniferous habitat, where prey levels regularly decrease during summer (Tauber and Tauber, 1977)..
The above pattern of variation prompts us to suggest the sequence whereby the components of the switch mechanism evolved in western populations. We propose that adult responsiveness to prey presence, which serves to modulate the photoperiodic component of the phenotypic switch in the polymorphic western populations, evolved in an ancestral population that had a very high level of the recessive short-day/long-day requirement for nondiapause reproduction. Such a population could harbor genetic variation for responsiveness to prey with no or little reproductive cost; it would also allow the expression (and therefore selection) of responsiveness to prey. This scenario identifies the polymorphic phenotypic switch of the western populations as a relatively highly evolved, geographically variable, adaptation to environmental heterogeneity.
We are grateful to Drs. W. E. Bradshaw, D. J. Futuyma, M. D. Rausher, Y. B. Linhart, and the anonymous reviewers for their thoughtful comments on the manuscript. We also thank the National Park Service, Department of the Interior, for cooperation; L. E. Ehler, J. J. Franclemont, D. M. Helgesen, R. G. Helgesen, J. K. Liebherr, A. J. Tauber, M. J. Tauber, and P. J. Tauber for help with collecting the specimens; and B. Gollands, C. E. McCulloch, G. Churchill, and P. Davis for aid with the statistical analysis. This work was supported, in part, by NSF grants DEB-7725486, DEB-8020988, and BSR-8817822.
Bradshaw, A. D. 1965. Evolutionary significance of phenotypic plasticity in plants. Adv. Genet. 13:115-155. Bradshaw, W. E., and C. M. Holzapfel. 1990. Evolution of phenology and demography in the pitcherplant mosquito, Wyeomyia smithii, pp. 47-67. In F. Gilbert (ed.), Insect Life Cycles: Genetics, Evolution, and Coordination. Springer-Verlag, N.Y., USA. Danilevsky, A. S. 1965. Photoperiodism and Seasonal Development of Insects. Oliver and Boyd, London, UK. Dingle, H. (ed.). 1978. Evolution of Insect Migration and Diapause. Springer-Verlag, N.Y., USA. Dingle, H., and J. P. Hegmann (eds.). 1982. Evolution and Genetics of Life Histories. Springer-Verlag, N.Y., USA. Dobson, S. 1989. Predator-induced reaction norms. Bioscience 39:447-451. Gilbert, F. (ed.). 1990. Insect Life Cycles: Genetics, Evolution, and Coordination. Springer-Verlag, N.Y., USA. Gillespie, J. H., and M. Turelli. 1989. Genotype-environment interactions and the maintenance of polygenic variation. Genetics 121:129-138. Groeters, F. R., and H. Dingle. 1987. Genetic and maternal influences on life history plasticity in response to photoperiod by milkwed bugs (Oncopeltus fasciatus). Am. Nat. 129:332-346. Hagen, K. S., R. L. Taasan, and E. F. Sawall, Jr. 1970. Some ecophysiological relationships between certain Chrysopa, honeydews, and yeasts. Boll. Lab. Entomol. Agric. Filippo Silvestri Port. 28:113-134. Harvell, C. D. 1990. The ecology and evolution of inducible defenses. Q. Rev. Biol. 65:323-340. Hedrick, P. W. 1983. Genetics of Populations. Science Books International, Boston, MA USA. Hegmann, J. P., and H. Dingle. 1982. Phenotypic and genetic covariance structure in milkweed bug life history traits, pp. 177-185. In H. Dingle and J. P. Hegmann (eds.), Evolution and Genetics of Life Histories. Springer-Verlag, N.Y., USA. Henry, C. S. 1985. Sibling species, call differences and speciation in green lacewings (Neuroptera: Chrysopidae: Chrysoperla). Evolution 39:965-984. Henry, C. S., and C. Busher. 1987. Patterns of mating and fecundity in several common green lacewings (Neuroptera: Chrysopidae) of eastern North America. Psyche 94:219-244. Henry, C. S., and M. Martinez Wells. 1990. Geographical variation in the song of Chrysoperla plorabunda Neuroptera: Chrysopidae) in North America. Ann. Entomol. Soc. Am. 83:317-325. Istock, C. A. 1980. Natural selection and life history variation: Theory plus lessons from a mosquito, pp. 113-127. In R. F. Denno and H. Dingle (eds.), Insect Life History Patterns. Springer-Verlag, N.Y., USA. --. 1983. The extent and consequences of heritable variation for fitness characters, pp. 61-96. In C. E. King and P. S. Dawson (eds.), Population Biology. Columbia University Press, N.Y., USA. Kingsolver, J. G., and D. C. Wiernasz. 1991. Seasonal polyphenism in wing-melanin pattern and thermoregulatory adaptation in Pieris butterflies. Am. Nat. 137:816-830. Lee, Jr., R. E., and D. L. Denlinger. 1991. Insects at Low Temperature. Chapman and Hall, N.Y., USA. Lees, A. D. 1955. The Physiology of Diapause in Arthropods. Cambridge University Press, London, UK. Levins, R, 1968. Evolution in Changing Environments. Princeton University Press, Princeton, NJ USA. Martinez Wells, M., and C. S. Henry. 1992. The role of courtship songs in reproductive isolation among populations of green lacewings of the genus Chrysoperla (Neuroptera: Chrysopidae). Evolution 46:31-42. Maynard Smith, J. 1982. Evolution and the Theory of Games. Cambridge University Press, Cambridge, UK. Mousseau, T. A., and D. A. Roff. 1989. Adaptation to seasonality in a cricket: Patterns of phenotypic and genotypic variation in body size and diapause expression along a cline in season length. Evolution 43:1483-1496. SAS Institute. 1985. SAS user's Guide: Basics, Version 5 Edition. SAS Institute Inc., Cary, NC USA. Smith-Gill, S. J. 1983. Developmental plasticity: Developmental conversion versus phenotypic modulation. Am. Zool. 23:47-55. Snedecor, G. W., and W. G. Cochran. 1989. Statistical Methods, 8th ed. Iowa State University, Ames, USA. Sokal, R. R., and F. J. Rohlf. 1981. Biometry, 2nd ed. W. H. Freeman and Co., N.Y., USA. Streans, S. C. (ed.). 1983. The interface of life-history evolution, whole-organism ontogeny and quantitative genetics. Am. Zool. 23:1-125. Tauber, C. A., and M. J. Tauber. 1973a. Diversification and secondary intergradation of two Chrysopa carnea strains (Neuroptera: Chrysopidae). Can. Entomol. 105:1153-1167. --. 1977. A genetic model for sympatric speciation through habitat diversification and seasonal isolation. Nature 268:702-705. --. 1982. Evolution of seasonal adaptations and life history traits in Chrysopa: Response to diverse selective pressures, pp. 51-72. In H. Dingle and J. P. Hegmann (eds.), Evolution and Genetics of Life Histories. Springer-Verlag, N.Y., USA. --. 1986. Ecophysiological responses in life-history evolution: Evidence for their importance in a geographically widespread insect species complex. Can. J. Zool. 64:875-884. --. 1987. Inheritance of seasonal cycles in Chrysoperla (Insecta: Neuroptera). Genet. Res. 49:215-223. Tauber, M. J., and C. A. Tauber. 1973b. Nutritional and photoperiodic control of the seasonal reproductive cycle in Chrysopa mohave (Neuroptera). J. Insect Physiol. 19:729-736. --. 1976. Developmental requirements of the univoltine Chrysopa downesi: Photoperiodic stimuli and sensitive stages. J. Insect Physiol. 22:331-335. Tauber, M. J., C. A. Tauber, and S. Masaki. 1986. Seasonal Adaptations of Insects. Oxford University Press, N.Y., USA. Thompson, J. D. 1991. Phenotypic plasticity as a component of evolutionary change. Trends Ecol. Evol. 6:246-249. Trexler, J. C., and J. Travis. 1990. Phenotypic plasticity in the sailfin molly, Poecilia latipinna (Pisces: Poeciliidae). I. Field experiments. Evolution 44:143-156. Via, S., and R. Lande. 1985. Genotype-environment interaction and the evolution of phenotypic plasticity. Evolution 39:505-522. West-Eberhard, M. J. 1989. Phenotypic plasticity and the origins of diversity. Annu. Rev. Ecol. Syst. 20:249-278. Williams, G. C. 1966. Adaptation and Natural Selection. Princeton University Press, Princeton, NJ USA.
|Printer friendly Cite/link Email Feedback|
|Author:||Tauber, Catherine A.; Tauber, Maurice J.|
|Date:||Dec 1, 1992|
|Previous Article:||Sexual selection and fitness variation in a population of small mouth bass, Micropterus dolomieui (Pisces: Centrarchidae).|
|Next Article:||Species isolation, genital mechanics, and the evolution of species-specific genitalia in three species of Macrodactylus beetles (Coleoptera,...|