Printer Friendly

Reproductive characteristics of the flower breeding Drosophila hibisci bock (drosophilidae) in Eastern Australia: genetic and environmental determinants of ovariole number.

In his classic paper on the correlation between relatives, R. A. Fisher (1918) introduced the notion of "dual epistacy" to account for deviations from linearity due to interactions between distinct genes involved in determining quantitative characters. Fisher did not elaborate on the role of epistasis in determining quantitative phenotypes and implied that the second-order contribution of gene-gene interactions supplied only small quantities that could be generally ignored. Sewall Wright (1968), however, recognized the universality of interaction effects (i.e., epistasis) as a generalization of his model of a netlike relationship between the genome and complex characters. The question of the importance of epistasis in the evolution of quantitative characters remains a central problem in understanding adaptation and genetic divergence in natural populations. Other nonlinear effects such as genotype-by-environment interactions in conjunction with geographical population structure and fluctuating population size are also potentially important determinants of quantitative features of organisms. These factors along with epistasis, environmental influences, and the traditional additive genetic contribution provide a fairly complete scheme for understanding and dissecting the causes of phenotypic variability.

The degree of interaction and independence of genetic and environmental determinants of quantitative phenotypes may influence the mode of adaptation and also may constrain or facilitate the types of solutions to ecological problems (Hard et al. 1993). Many ecological-genetic studies identify phenotypes that are relevant to their question (e.g., evolutionary response to predation pressure). They then examine whether phenotypic variation has a genetic basis and how much environmental conditions influence the organism's phenotype (Barker et al. 1990). The potential importance of epistasis and genotype-by-environment interaction generally is recognized to be relevant to ecological-genetic questions, but the difficulty in determining their significance and estimating their magnitude often precludes including them in ecological-genetic studies (Barker 1974, 1979; Hedrick et al. 1978). The search for, and identification of, cases in which epistasis is present has been more common since tests for detecting it were devised (Mather 1949; Hayman 1958, 1960).

Theoretical considerations of the role of epistasis in the evolution of natural populations include the following features. Epistasis (1) can, in conjunction with genetic drift, enhance the rate of genetic diversification (Bryant et al. 1986; Goodnight 1987, 1988, 1995); (2) can mask genotypes and thus hide additive variation, thereby limiting the range of phenotypes exposed to the selective process (Gimelfarb 1989); (3) may be involved in the expression and increase in the additive genetic variation after founding events (Templeton 1980; Goodnight 1988); (4) can be involved in producing heterosis (Falconer 1989; Schell and Cockerham 1992); (5) is important for the operation of interdemic selection because it facilitates the generation of between-deme genetic variation (Wright 1931, 1969, 1978; Wade and McCauley 1984); and (6) is central to Wright's shifting balance theory of evolution, where gene interactions are the foundation for promoting coadaptation of genes (Wright 1935, 1952, 1969; Goodnight 1995).

Genotype-by-environment interactions also can enhance genetic diversification in isolated populations, especially when the "best" genotype varies among environments (Falconer 1989; Gillespie and Turelli 1989; Pray et al. 1994). Just as epistasis can be the basis for "coadapted gene complexes" within a population, genotype-by-environment interactions can be the foundation for adaptations realized by organisms living in different environments. In addition, the expression of epistasis can depend on the environment (Yermano and Allard 1961; Jinks et al. 1973; Blows and Sokolowski 1995) and the degree of expression of a quantitative trait (Blows and Hoffman 1996).

Populations that are sufficiently separated and experience different environments are suitable for investigation of genetic, environmental, and genotype-by-environment interactions because they are likely to experience selective pressures that lead to genetic divergence. Phenotypic clines are useful in this context because the phenotypes often experience regular changes in environmental variation over short and long distances.

Clines are usually interpreted as resulting from either an evolutionary response to environmental conditions that have persisted over relatively long periods of time or from short-term processes such as recent hybridization and subsequent spread of intermediate phenotypes from the hybrid zone (Endler 1977). Phenotypic clines can also result from constant genotypes along the cline when differential phenotypic expression is a consequence of short-term ecological conditions. We have been investigating a latitudinal cline in the number of ovarioles carried by females of the Hibiscus flower-breeding Drosophila hibisci Bock. Drosophila hibisci uses Hibiscus section Furcaria DC (Malvaceae) flowers at all stages of its life cycle (Cook et al. 1977). The plants grow as small, spatially isolated, normally monospecific stands of usually less than 50 trees, often less than 10 trees. The flowers are used for larval and adult nutrition, as a site for courting and mating (Polak et al., in press) and as shelter. Hibiscus flowers used by D. hibisci are one-day flowers, opening in the morning and senescing and closing in the evening. These closed blossoms sometimes remain on the plant, but usually detach by the next day and fall to the ground beneath the parent tree or bush. Adult females insert their eggs into the petal tissue inside the open flower. Depending on the temperature, larvae take about a day to hatch. Larvae feed on flower tissues as well as the microbes associated with the decaying blossom. Pupation probably occurs in surrounding debris or in the soil.

A survey of 36 populations ranging from 20.8 [degrees] to 34.4 [degrees] South latitude along the east coast of Australia showed that females from northern populations have fewer ovarioles (10-13) than females in southern populations (15-20 ovarioles). Mean ovariole number was not correlated with species of Hibiscus used as a larval substrate or with floral or habitat characteristics of the local host Hibiscus populations (Starmer et al., in press). These characters included density of trees or bushes, flower weight, petal length, microorganisms present, density of flies, and density of associated beetles. Mean ovariole number of the 36 populations was significantly correlated with mean thorax length, mean local temperature, and mean daily rainfall (5-19 days prior to collection). These results suggest that the larval abiotic environment may influence the reproductive potential of the population by determining the ovariole number of the resulting females, but do not exclude the possibility that the populations have adapted to different predictable larval environments present along the latitudinal gradient.

As a first step in determining what environmental and genetic factors are responsible for the population differentiation of ovariole number, we have experimentally investigated the effect of larval and pupal developmental temperature in conjunction with quantitative genetic analysis between two populations that differ in their mean ovariole number. These experiments were conducted to ascertain (1) the influence of developmental temperature; (2) the magnitude of genetic determination, including maternal effects and digenic epistasis; and (3) the degree of genetic by environment interaction.


Flowers used in the laboratory were obtained daily from H. heterophyllus plants growing in a greenhouse at the University of New England, Armidale, New South Wales.

Two parental strains of D. hibisci were obtained from two sites 1157 km apart. The northern strain, designated "n," was derived from adults reared from 131 flowers of H. heterophyllus collected in Eungella National Park, Queensland (21.139 [degrees] S, 148.507 [degrees] E). The other strain, "m," originated from 181 flowers of H. heterophyllus collected from the middle of the geographic distribution of D. hibisci near Port Macquarie, New South Wales (31.500 [degrees] S, 152.804 [degrees] E).

Original parental strains of D. hibisci used in self and cross generations were reared in the laboratory at 25 [degrees] C from eggs deposited in field-collected flowers. Adults were reared by placing individual flowers with eggs into open plastic bags inside glass milk bottles containing dry, autoclaved sand. After pupae started to form, 10 mL of water was added to each bottle to moisten the sand. Emerging adults were aspirated from the bottles each day. Sexes were separated and placed either in sugar vials (5% sucrose, 1.5% agar) or into cages containing flowers inserted into containers of water resting on dry sand. The cages received one to three newly opened flowers every day. Old flowers were left in the cages or removed and placed in bottles at the appropriate temperature for larval development. Cages were used to mature and maintain females as virgins either for crosses or for dissection of their reproductive structures.

Test Lines

The sources for the lines and crosses used in this study are listed in Table 1. Adults reared from the field-collected flowers incubated at 25 [degrees] C were allowed to mature, then selfed and crossed reciprocally to form four lines (m[cross product], n[cross product], [F.sub.1], [F.sub.1r]). These matings were conducted at 25 [degrees] C in duplicate cages each with 20 pairs of females and males. Flowers with eggs were collected from these eight cages daily over a 15-day period, put into bottles, split daily into three groups, and placed at three different temperatures (25.0 [degrees], 21.5 [degrees], and 18 [degrees] C). Adults of the four lines that were reared at 25 [degrees] C were used to form five segregating lines composed of the four backcrosses and the [F.sub.2] generation as detailed in Table 1 and Figure 1. Each of these lines was obtained by placing 18-20 pairs of males and females in cages at 25 [degrees] C. After five days, each reciprocal backcross cage was supplemented with 10-12 pairs of males and females. Flowers containing eggs of these five lines, collected daily over a 10-day period, were placed in bottles and incubated at 25 [degrees] C.
TABLE 1. Source of genetic lines used in the analyses.

Line                    Source

m                       Reared at 25 [degrees] C from field
                        collected flowers, near Port Macquarie

m[cross product]        Self of m

m*                      Pooled m and m[cross product] reared at 25
                        [degrees] C

n                       Reared at 25 [degrees] C from field
                        collected flowers, Eungella National Park

n[cross product]        Self of n

n*                      Pooled n and n[cross product]

[F.sub.1]               n females x m males

[F.sub.1r]              m females x n males

[F.sub.1]*              Pooled [F.sub.1] and [F.sub.1r] used in

[F.sub.1p]              Pooled [F.sub.1] and [F.sub.1r] used in

[F.sub.1]**             Pooled [F.sub.1]* and [F.sub.1p]

[F.sub.2]               Self of [F.sub.1p]

Bcm                     [F.sub.1p] females x m[cross product] males

Bcmr                    m[cross product] females x [F.sub.1p] males

Bcn                     [F.sub.1p] females x n[cross product] males

Bcnr                    n[cross product] females x [F.sub.1p] males

For females in each test generation, adult thorax length was measured from the anterior end of the thorax to the posterior end of the scutellum, and the abdomen was dissected in Ringer's solution. The ovaries were fixed for about 1 minute in FAA (6:16:1:30, formalin:absolute alcohol:glacial acetic acid:distilled [H.sub.2]O), and the number of ovarioles in each ovary was counted. Asymmetry in number of ovarioles was calculated as the difference in ovariole number between the two ovaries. If an egg was present in the uterus, its length and width were measured. Egg volume was approximated by calculating the volume of a prolate spheroid as (1/6)[Pi](egg length)[(egg width).sup.2] (Atkinson 1979; Montague et al. 1981).

Line means and standard errors for ovariole number were estimated by adjusting the means with an analysis of covariance using thorax length as the covariate. The distributions of ovariole number for the parental lines were tested for normality with the Shapiro-Wilk statistic, W. Both distributions fit the normal distribution (n: W = 0.96, P = 0.11; m: W = 0.97, P = 0.24), and neither square-root nor logarithmic transformations of ovariole number improved the fit to the normal. We thus used a linear scale for ovariole number because it conforms with the normality assumption for tests of significance and it permits comparisons with other studies on the genetic determination of ovariole number that used the same scale.


Two main models based on generation means analysis (Mather and Jinks 1982) were used to estimate genetic and environmental contributions to the observed means. The first was a genetic model using the self, [F.sub.1], backcross, and [F.sub.2] lines reared as larvae at 25 [degrees] C. This model was used to evaluate two basic modes of genetic activity, one with and one without maternal effects. The model without maternal effects included the following composite genetic effects (additive, d; dominance, h; digenic epistasis, i, j, and 1), whereas the maternal model involved gene activity of progeny (additive, d; maternal additive, dm; maternal dominance, hm; and digenic epistasis between maternal and progeny genotypes,, Other digenic epistasic parameters such as progeny dominance by maternal dominance ( are not possible to estimate for the lines produced in this study because they either cannot be discriminated or their coefficients are linear combinations of the coefficients for other simpler terms in the model. Given that we observed nine means, models with more that eight parameters could not be evaluated. As a consequence, a full digenic epistatic model with all parameters of progeny, maternal and maternal by progeny interaction was not possible to evaluate for the lines raised at 25 [degrees] C.

The second model was used to estimate the contribution of additive, dominance, and maternal effects in addition to environmental effects of larval developmental temperature. The self and [F.sub.1] lines reared at the three temperatures were used for this model. Estimates for composite genetic (d, h, dm) and environmental effects (e1, linear; e2, deviation from linear) and the interactions of genetic and environmental effects (d.el, d.e2, h.e1, h.e2, dm.e1, and dm.e2) were obtained.

In all models, least-squares procedures were used to estimate model parameters contained in vector Y and their variances from the diagonal of their variance covariance matrix S (Mather and Jinks 1982; Lynch and Walsh 1997). The estimates of Y and S are obtained as

[Mathematical Expression Omitted] (1)


[Mathematical Expression Omitted], (2)

such that C is the coefficient matrix for the contribution of effects to each line mean, V is the diagonal matrix of the error variances of each line mean, and X is the vector of observed line means. All parameters of each model were estimated in a full model to provide a guide to which parameters might be important in reduced models (Mather and Jinks 1982). Goodness of fit of each model was tested by a [[Chi].sup.2], calculated as

[Mathematical Expression Omitted]. (3)

(Hayman 1958). The degrees of freedom for this [[Chi].sup.2] is the number of line means minus the number of parameters estimated in the model.

F-statistics were used to evaluate the improvement in the goodness of fit for other parameters (Graybill 1961). This was accomplished by using

[Mathematical Expression Omitted], (4)

where n is the number of observations, [p.sub.1] is the number of parameters in the basic model, and [p.sub.2] is the number of parameters of the expanded model.

In the analysis of generation means (or means of line crosses) the loci involved are assumed to be in Hardy-Weinberg and linkage equilibrium. In addition, loci that differ between the two parental lines are assumed to be unlinked. The parental lines can carry the same alleles but because the analysis estimates the net difference between additive effects, average dominance, and the net directional epistasis, the alleles that are the same only need to differ in frequency (Lynch and Walsh 1997). The composite effects of the model summarize the total effects over all loci such that some alleles will contribute positive and others negative increments to the phenotype. In essence, the models are arbitrary for the number and distribution of loci involved. Estimates of parameters are used to predict the line means as the algebraic sum of the contribution of each parameter associated with the expected genotype of that line (Mather and Jinks 1982).

Field Collections

Two field collections were made one week apart (October 10 and 17, 1996) at one site (BEL; 30 [degrees] 25.155 [minutes] S, 152 [degrees] 49.425 [minutes] E; near Bellingen, NSW) for comparison of ovariole number in field-caught females with their progeny reared in the laboratory. We conducted this study because the mean ovariole number of laboratory reared flies of the two populations used in the crosses (m and n) differed significantly from the numbers estimated from field-caught females in the previous year (Starmer et al., in press). Females from the two BEL collections were returned to the laboratory, [TABULAR DATA FOR TABLE 2 OMITTED] [TABULAR DATA FOR TABLE 3 OMITTED] measured, dissected, ovarioles counted, and an egg, if present in the uterus, was measured. Recently laid eggs contained in 0-1-day-old field flowers were collected at the same time as the females. These were returned to the laboratory and placed at 18.0 [degrees], 21.5 [degrees], and 25.0 [degrees] C on the same day. Adults reared from these flowers were matured in cages with flowers, measured, dissected, and ovarioles counted for comparison with the field-caught females.


The average thorax length of females reared at 25 [degrees] C varied among lines and each line exhibited a significant (P [less than] 0.05) positive correlation between thorax length and ovariole number (Table 2). Among lines (Table 2; m*, n*, [F.sub.1], [F.sub.1r], [F.sub.2], Bcm, Bcmr, Bcn, Bcnr), mean thorax length also was correlated positively with mean ovariole number ([Mathematical Expression Omitted], df = 7, P [less than] 0.001). Therefore, thorax length was used as a covariate in an analysis of covariance to adjust the mean ovariole number for each line. Table 3 lists the least-squares means for parental, [F.sub.1]s, [F.sub.2], and backcross generations reared at 25 [degrees] C. Mean ovariole number (corrected for thorax length) for parental and self generations of m did not differ significantly ([F.sub.1,61] = 0.00, P = 0.98) so the data for parental and self were pooled. This was also the case with the n line ([F.sub.1,63] = 1.05, P = 0.46). However, the means estimated from the pooled parental data for the two strains m and n did differ from one another ([F.sub.1,127] = 25.2, P [less than] 0.0001).

The adjusted means and variances ([SE.sup.2]) for the nine lines (pool parental in Table 3) were used in a generation means analysis to estimate the composite additive genetic effect (d) in the model ova = m + d, where m is the mean over all populations (model C, Table 4). The regression line resulting from using this estimate is plotted along with the observed generation means in Figure 2. The departure from the regression line and the goodness of fit for the expected generation means ([[[Chi].sup.2].sub.7] = 8.27, P = 0.30) show the additive model to be adequate but estimates of other parameters (model A, Table 4) indicates the epistatic parameter (j) is significantly [TABULAR DATA FOR TABLE 4 OMITTED] different from zero. An F-test was used to evaluate the improvement in the goodness of fit for this and other parameters (Graybill 1961). Only the epistatic parameter (j) showed a significant reduction in the [[Chi].sup.2] (model B, Table 4). The F-statistic for this test is obtained by comparing the goodness of fit [[Chi].sup.2] of model B ([[[Chi].sup.2].sub.6] = 4.28, Table 4) with that of the additive model (model C, [[[Chi].sup.2].sub.7] = 8.27, Table 4); [F.sub.1,6] = (8.27 - 4.28) / (4.28 [divided by] 6) = 5.59, (P = 0.056). An analysis of pooled reciprocals (Table 3) produced essentially the same results with the ova = m + d + j model describing the line means adequately ([[[Chi].sup.2].sub.3] = 1.84, P = 0.64) and with both d and j contributing significantly (P [less than] 0.05).

Evaluation of the model with maternal effects (Table 4) showed maternal effects and progeny-by-maternal epistasis were not significant. All F-tests used to evaluate the significance of adding parameters to model C were not significant (P [less than] 0.05). When only maternal additive (dm) and maternal dominance (dh) were used in the model (model F, Table 4), the maternal additive effect was significant but the goodness of fit of the model was not adequate ([[[Chi].sup.2].sub.6] = 14.2, P = 0.03) to describe the means.

Thorax length and ovariole number were positively correlated (P [less than] 0.05) between individuals within lines (r tl.ova, Table 2). The correlation between thorax length and ovariole number was also present between the means of these two variables among lines raised at the three temperatures (25 [degrees] C: m*, n*, [F.sub.1], [F.sub.1r]; 21.5 [degrees] C: m[cross product], n[cross product], [F.sub.1], [F.sub.1r]; 18 [degrees] C: m[cross product], n[cross product], [F.sub.1], [F.sub.1r]; [Mathematical Expression Omitted], df = 10). The mean ovariole number for each line was therefore adjusted by using thorax length as a covariate in an analysis of covariance. The resulting least-squares means for line ovariole number at the three temperatures are listed in Table 5. These means and their variances were used to estimate genetic and environmental parameters that might explain the differences in ovariole number (Table 6).
TABLE 5. Least-squares means of ovariole number corrected for
correlation with thorax length for the parental, [F.sub.1], and
[F.sub.1r] generations of crosses between m and n. Larvae were
reared at 25.0 [degrees], 21.5 [degrees] and 18.0 [degrees] C.
[cross product] = self generation of original laboratory reared
line. * = pooled generation.

                           Unpooled      ([F.sub.1] and [F.sub.1r])
Line                n      ova (SE)               ova (SE)

25.0 [degrees]

m*                 64   21.236 (0.342)         21.269 (0.344)
n*                 66   18.612 (0.331)         18.616 (0.333)
[F.sub.1]          43   20.073 (0.425)
[F.sub.1r]         24   21.069 (0.575)
[F.sub.1]*         67                          20.363 (0.355)

21.5 [degrees]

m[cross product]   21   22.444 (0.587)         22.442 (0.589)
n[cross product]   42   21.502 (0.424)         21.546 (0.425)
[F.sub.1]          40   21.040 (0.426)
[F.sub.1r]         23   21.625 (0.579)
[F.sub.1r]         63                          21.240 (0.343)

18.0 [degrees]

m[cross product]   30   21.113 (0.499)         21.066 (0.501)
n[cross product]   43   20.542 (0.410)         20.542 (0.412)
[F.sub.1]          31   21.188 (0.497)
[F.sub.1r]         31   22.628 (0.483)
[F.sub.1]*         62                          21.943 (0.350)
TABLE 6. Estimates of parameters for three models (A-C) for ovariole
numbers at different temperatures. Bold numbers indicate significant
contributions for that parameter ([Alpha] = 0.05).

                      A                B                C

m               20.376 (0.201)   20.423 (0.195)   20.379 (0.196)
d                0.190 (0.320)
h                0.864 (0.320)    0.848 (0.312)    0.881 (0.312)
dm               0.609 (0.249)    0.714 (0.130)    0.642 (0.132)
e1              -0.452 (0.201)   -0.557 (0.154)   -0.520 (0.154)
e2               1.597 (0.414)    1.627 (0.396)    1.648 (0.396)
d.e1             0.624 (0.320)    0.551 (0.195)
h.e1            -0.217 (0.320)
dm.e1           -0.111 (0249)
d.e2            -0.011 (0.602)
h.e2            -1.504 (0.602)   -1.439 (0.577)   -1.471 (0.577)
dm.e2           -0.317 (0.437)
[[Chi].sup.2]    -               10.45             2.46
df               0                6                5

Investigation of several models resulted in an adequate description of the 12 line means by using seven parameters: mean, m; progeny dominance, h; maternal additive, dm; environmental effects (e1, linear; e2, deviations from linear); and interactions of genetic and environmental effects, d.el and h.e2). All of these estimates were significant (P [less than] 0.05) and the model expectation for each generation mean was close to the observed means ([[[Chi].sup.2].sub.5] = 2.45, P = 0.78). These results revealed significant dominance but no significant additive effects of the progeny genotypes. However, maternal genotypes had an additive effect. In addition, environmental effects were strong, including a nonlinear environmental effect (e2) that interacted with progeny dominance (h.e2). The e2 effect is apparent in Table 5 as a higher ovariole number at the intermediate temperature (21.5 [degrees] C).


Analysis of asymmetry in ovarioles (ovasym [+ or -] SE) showed the parental strains m (1.094 [+ or -] 0.101) and n (0.909 [+ or -] 0.099) to be statistically equivalent ([F.sub.1,128] = 1.71, P = 0.19). Egg volume ([mm.sup.3] [+ or -] SE) for strain m (0.0133 [+ or -] 0.0004) was statistically equivalent ([F.sub.1,51] = 0.56, P = 0.46) to strain n (0.0129 [+ or -] 0.0004). Therefore, the genetic analysis of these two traits was not carried out.

The comparison of field-caught females with laboratory reared progeny of the field-caught female generation is listed in Table 7. Two features of these data are apparent. The field-collected females have significantly fewer ovarioles (means adjusted for thorax length) than the laboratory-reared progeny regardless of larval rearing temperature (1st week [F.sub.1,58] = 31.3, P [less than] 0.0001; 2nd week [F.sub.1,94] = 17.8, P [less than] 0.001) and the two collections of field females (one week apart) have significantly different ovariole numbers ([F.sub.1,56] = 9.4, P = 0.003) but the flies reared in the laboratory are not different between weeks ([F.sub.1,96] = 0.13, P = 0.72).


The relationship between ovariole number and fitness has not been well established within any Drosophila species. Wayne et al. (1997) studied 45 isogenic lines of D. melanogaster derived from nature but found no correlation between ovariole number and fitness. This result differed from that found by David (1970) and Bouletreau-Merle et al. (1982), who observed a positive correlation between ovariole number and the maximum daily rate of oviposition. The difference between these studies could be based on differences in how female fecundity was measured and further experiments are required to explain the discrepancy (Wayne et al. 1997).

The wide range of ovariole number (one per ovary to as many as 100 per female) in the Hawaiian drosophilids fits a pattern of low fecundity to high fecundity as the ovariole number increases. This pattern is coincident with the variation in the abundance and nutrient quality of the larval niche in which rich and abundant substrates are hosts to species with a large number of ovarioles and nutritionally poorer substrates are hosts to species with a lower number of ovarioles (Kambysellis and Heed 1971). Flower-breeding species are on the lower end of the "fecundity potential" spectrum where Scaptomyza (Exalloscaptomyza) caliginosa Hardy (breeds in morning glory blossoms on the island of Hawaii) has two ovarioles per female, and laboratory studies show lifetime reproductive output of 8 to 12 progeny per female (Starmer and Wolf, unpubl. data). Drosophila hibisci breeds in larger flowers and has a higher number of ovarioles, ranging from means of 10 to 20 in natural populations over a latitudinal range of 13.6 [degrees] (Starmer et al., in press). The generation means analysis reported here was directed toward understanding the potential genetic and environment determinants of ovariole number between populations of D. hibisci. A within population analysis would be valuable for comparison but has yet to be conducted.

The inheritance of ovariole number in D. hibisci at 25 [degrees] C is adequately described by an additive model but additive-by-dominance epistasis significantly improves the fit of the model. This result indicates that digenic epistasis probably is involved in determining the ovariole number of D. hibisci, at least in this interpopulational study. The relationship of the additive (d) and epistatic (j) components is such that they are of opposite sign but equal in magnitude. The consequence of this form of epistasis is to reduce the variance in the [F.sub.2] and backcross generations and to reduce the genetic variation on which natural selection can operate. For example, in a hypothetical case, if only two loci each with two alleles are responsible for the phenotype (the actual number of loci is not known in our study), this would produce a 14:1:1 phenotypic ratio in the [F.sub.2] of a dihybrid cross and a 3:1 ratio in the backcross generations. This is analogous to classical dominant epistasis that produces a 12:3:1 ratio of phenotypes in the [F.sub.2] of a dihybrid cross. Linkage among additive genes also might account for deviations from linearity. An adequate test for linkage would require additional segregating generations (e.g., [F.sub.3], [F.sub.4]).

Studies of other species of Drosophila have also uncovered significant epistatic influences on ovariole number. R'Kha et al. (1991) conducted a similar genetic analysis by crossing two populations representing two closely related species in the D. melanogaster species complex (D. sechellia and D. simulans). They raised the parental, [F.sub.1], and backcross generations on medium with and without the addition of morinda fruit. Both conditions revealed strong genetic components in the determination of ovariole number. We reanalyzed their data using generation means analysis and confirmed their findings of significant additive effects and no dominance. In addition, we detected a significant additive-by-additive epistatic component when the means for populations reared without the addition of morinda fruit were used in the analysis. Chakir et al. (1995) also found evidence of epistatic effects on ovariole number in D. melanogaster strains collected in France and the Congo. Their study analyzed the effects of chromosomal substitutions between the two populations and found significant interactions between chromosomes 2 and 3. Chakir et al. (1995) compared their results with other investigations (Thomas-Orillard 1975, 1982) on French and Japanese populations and found them consistent in terms of chromosomal involvement. They suggest that genetic determination of ovariole number is due to the action of homologous genes in these cases and that the homology may extend to other species. Our results are similar only with respect to the mode of gene activity because we have no data on chromosomal effects.

In D. hibisci, parental and [F.sub.1] generations raised at different temperatures revealed dominance (h), maternal effects (dm), strong linear and nonlinear environmental components (el, e2), and genotype-by-environment interactions (d.el and h.e2). A significant maternal effect on ovariole number also has been reported in D. melanogaster by Thomas-Orillard (1984). She found that a picornavirus in the cytoplasm is responsible for an increase in ovariole number as well as affecting other quantitative traits such as developmental time and fresh weights of females. Delpuech et al. (1995) also reported nonlinear effects of temperature on ovariole number in temperate and tropical populations of D. melanogaster. In both studies, ovariole number was higher at intermediate temperatures than at higher and lower temperatures.

Because the segregating generations were not raised at the different temperatures, it was not possible to estimate digenic epistasis or interactions of epistatic effects with environmental effects. It is nevertheless significant that nonlinear terms involving genetic effects and environmental effects are prevalent in the analysis of parents and their [F.sub.1]s. These results in conjunction with our observations on differences between ovariole number of laboratory-reared females and field-caught females (corrected for body size) indicate the phenotype is very sensitive to environmental change. Even females collected from the same site one week apart showed significant changes in ovariole number (Table 7). This was also true for sites at which D. hibisci were collected almost one year apart (Starmer and Barker, unpubl. data). Bouletreau (1978) found results similar to ours for the ovariole number of D. melanogaster females dissected from field collections and for their progeny reared in the laboratory. She concluded that field-collected females have fewer ovarioles because of a trade-off with selective pressures for survival.

Fly reproductive potential undoubtedly is modified by environmental fluctuations. The analysis of data from D. hibisci shows that temperature of larval development is an important influence on ovariole number (Table 6). However, other factors that were not tested that could be important include: (1) larval density or competition; (2) temperature fluctuations experienced during larval development; (3) flower nutrient quality; (4) flower moisture content; and (5) influence of nematode (Howardula sp.) parasites that are found in the natural populations. For example, Kambysellis et al. (1979) showed the ovariole number of D. mimica varied over seasons and attributed this variation to changes in body size that most likely resulted from larval competition.

We agree with Hard et al. (1992, 1993) that "models assuming strictly additive genetic effects, even if appropriate at the within-population level, may tell us little about the actual genetic divergence of populations." We might add that assuming other nonlinear effects such as genotype-by-environment interaction can be ignored may also be misleading. The nature of the genetic architecture and its relationship to the environment is the basis for understanding and evaluating theories on evolution. If Sewall Wright's shifting balance theory is plausible, then complex characters will be controlled by epistatic genes and influenced by other nonlinear components of the genetic system. Revealing the nature of the genetic structure is thus an important step in evaluating the process leading to adaptation and divergence.


This study was supported by grants from International Programs of the National Science Foundation and the Australian Research Council and by funds from Syracuse University. We thank J. Davies, A. Edmonds, B. Laidler, and Y. Wang for their assistance. We appreciate the help of S. Pearson for shipment of flowers from Eungella National Park. Comments from the corresponding editor and two reviewers helped improve the manuscript.


ATKINSON, W. C. 1979. A comparison of the reproductive strategies of domestic species of Drosophila. J. Anim. Ecol. 48:53-64.

BARKER, J. S. F. 1974. The state of information concerning deviations from additivity of gene effects. Proc. 1st World Congr. Genet. Appl. Livestock Prod. 1:373-383.

-----. 1979. Inter-locus interactions: a review of experimental evidence. Theor. Popul. Biol. 16:323-346.

BARKER, J. S. F., W. T. STARMER, AND R. J. MACINTYRE. 1990. Ecological and evolutionary genetics of Drosophila. Plenum Press, New York.

BLOWS, M. W., AND A. A. HOFFMAN. 1996. Evidence for an association between nonadditive genetic variation and extreme expression of a trait. Am. Nat. 148:576-587.

BLOWS, M. W., AND M. B. SOKOLOWSKI. 1995. The expression of additive and non-additive genetic variation under stress. Genetics 140:1149-1159.

BOULETREAU, J. 1978. Ovarian activity and reproductive potential in a natural population of Drosophila melanogaster. Oecologia 35:319-342.

BOULETREAU-MERLE, J., R. ALLEMAND, Y. COHET, AND J. R. DAVID. 1982. Reproductive strategy in Drosophila melanogaster: significance of a genetic divergence between temperate and tropical populations. Oecologia 53:323-329.

BRYANT, E. H., S. A. MCCOMMAS, AND L. M. COMBS. 1986. The effect of an experimental bottleneck upon quantitative genetic variation in the housefly. Genetics 114:1191-1211.

CHAKIR, M., J. R. DAVID, E. PLA, AND P. CAPY. 1995. Genetic basis of some morphological differences between temperate and equatorial populations of Drosophila melanogaster. Experientia 51:744-748.

COOK, R. M., P. A. PARSONS, AND I. R. BOCK. 1977. Australian endemic Drosophila. II. A new Hibiscus-breeding species with its description. Aust. J. Zool. 25:755-763.

DAVID, J. R. 1970. Le nombre d'ovarioles chez la Drosophila: relation avec la fecondite et valeur adaptive. Arch. Zool. Exp. Gen. 111:357-370.

DELPUECH, J.-M., B. MORETEAU, J. CHICHE, E. PLA, J. VOUIDIBIO, AND J. R. DAVID. 1995. Phenotypic plasticity and reaction norms in temperate and tropical populations of Drosophila melanogaster: ovarian size and developmental temperature. Evolution 49:670-675.

ENDLER, J. A. 1977. Geographic variation, speciation and clines. Monogr. Popul. Biol., no. 10. Princeton Univ. Press, Princeton, NJ.

FALCONER, D. S. 1989. Introduction to quantitative genetics. 3d ed. Wiley, New York.

FISHER, R. A. 1918. The correlation between relatives on the supposition of Mendelian inheritance. Trans. R. Soc. Edinburgh 52:399-433.

GILLESPIE, J. H., AND M. TURELLI. 1989. Genotype-environment interactions and the maintenance of polygenic variation. Genetics 121:124-138.

GIMELFARB, A. 1989. Genotypic variation for a quantitative character maintained under stabilizing selection without mutations: epistasis. Genetics 123:217-227.

GOODNIGHT, C. J. 1987. On the effect of founder events on epistatic genetic variance. Evolution. 41:80-91.

-----. 1988. Epistasis and the effect of founder events on the additive genetic variance. Evolution 42:441-454.

-----. 1995. Epistasis and the increase in additive genetic variance: implication for phase 1 of Wright's shifting-balance process. Evolution 49:502-511.

GRAYBILL, F. A. 1961. An introduction to linear statistical models. Vol. I. McGraw-Hill, New York.

HARD, J. J., W. E. BRADSHAW, AND C. M. HOLZAPEEL. 1992. Epistasis and the genetic divergence of photoperiodism between populations of the pitcher-plant mosquito, Wyeomyia smithii. Genetics 131:389-396.

-----. 1993. The genetic basis of photoperiodism and its evolutionary divergence among populations of the pitcher-plant mosquito, Wyeomyia smithii. Am. Nat. 142:457-473.

HAYMAN, B. I. 1958. The separation of epistatic from additive and dominance variation in generation means. Heredity 12:371-390.

-----. 1960. The separation of epistatic from additive and dominance variation in generation means. II. Genetica 31:133-146.

HEDRICK, P., S. JAIN, AND L. HOLDEN. 1978. Multilocus systems in evolution. Evol. Biol. 11:101-182.

JINKS, J. L., J. M. PERKINS, AND H. S. POONI. 1973. The incidence of epistasis in normal and extreme environments. Heredity 31:263-269.

KAMBYSELLIS, M. P., AND W. B. HEED. 1971. Studies of oogenesis in natural populations of Drosophilidae. I. Relation of ovarian development and ecological habitats of the Hawaiian species. Am. Nat. 105:31-49.

KAMBYSELLIS, M. P., W. T. SWARMER, G. SMATHERS, AND W. B. HEED. 1979. Studies of oogenesis in natural populations of Drosophilidae. II. Significance of microclimatic changes on oogenesis of D. mimica. Am. Nat. 115:67-91.

LYNCH, M., AND B. WALSH. 1997. Genetics and analysis of quantitative traits. Sinauer, Sunderland, MA.

MATHER, K. 1949. Biometrical genetics. 1st ed. Methuen, London.

MATHER, K., AND J. L. JINKS. 1982. Biometrical genetics: the study of continuous variation. 3d ed. Chapman and Hall, New York.

MONTAGUE, J. R., R. L. MANGAN, AND W. T. SWARMER. 1981. Reproductive allocation in the Hawaiian Drosophilidae: egg size and number. Am Nat. 118:865-871.

POLAK, M., W. T. SWARMER, AND J. S. F. BARKER. In press. A mating plug and male mate choice in Drosophila hibisci Bock. Anim. Behav.

PRAY, L. A., J. M. SCHWARTZ, C. J. GOODNIGHT, AND L. STEVENS. 1994. Environmental dependency of inbreeding depression: implications for conservation biology. Conserv. Biol. 8:562-568.

R'KHA, S., P. CAPY, AND J. R. DAVID. 1991. Host-plant specialization in the Drosophila melanogaster species complex: a physiological, behavioral and genetic analysis. Proc. Nat. Acad. Sci. USA 88:1835-1839.

SCHELL, F. W., AND C. C. COCKERHAM. 1992. Multiplicative vs. arbitrary gene action in heterosis. Genetics 131:461-469.

STARMER, W. T., L. L. WOLF, J. S. F. BARKER, J. M. BOWLES, AND M. A. LACHANCE In press. Reproductive characteristics of the flower breeding Drosophila hibisci Bock (Drosophilidae) along a latitudinal gradient in eastern Australia: relation to flower and habitat features. Biol. J. Linn. Soc. 62.

TEMPLETON, A. R. 1980. The theory of speciation via the founder principle. Genetics 94:1011-1038.

THOMAS-ORILLARD, M. 1975. Tentative de localisation des genes qui president au determinisme du nombre d'ovarioles chez la Drosophile. Arch. Genet. (Zur.) 48:116-127.

-----. 1982. Structure d'effet du systeme multifactoriel qui modele la morphogenese ovarienne de la Drosophile. Arch. Zool. Exp. Gen. 122:455-465.

-----. 1984. Modifications of mean ovariole number, fresh weight of adult females and developmental time in Drosophila melanogaster induced by Drosophila C virus. Genetics 107:635-644.

WADE, M. J., AND D. E. MCCAULEY. 1984. Group selection: the interation of local deme size and migration in the differentiation of small populations. Evolution 38:1047-1058.

WAYNE, M. L., J. B. HACKETT, AND T. F. C. MACKAY. 1997. Quantitative genetics of ovariole number in Drosophila melanogaster. I. Segregating variation and fitness. Evolution 51:1156-1163.

WRIGHT, S. 1931. Evolution in Mendelian populations. Genetics 16:97-159.

-----. 1935. Evolution in populations in approximate equilibrium. J. Genet. 30:257-266.

-----. 1952. The genetics of quantitative variability. Pp. 5-41 in E. C. R. Reeve and C. H. Waddington, eds. Quantitative inheritance. Her Majesty's Star. Office, London.

-----. 1968. Evolution and the genetics of populations. Vol. 1. Genetic and biometric foundations. Univ. of Chicago Press, Chicago.

-----. 1969. Evolution and the genetics of populations. Vol. 2. The theory of gene frequencies. Univ. of Chicago Press, Chicago.

-----. 1978. Evolution and the genetics of populations. Vol. 4. Variability within and among natural populations. Univ. of Chicago Press, Chicago.

YERMANO, D. M., AND R. W. ALLARD. 1961. The detection of epistatic gene action in flax. Crop Sci. 1:307-310.
COPYRIGHT 1998 Society for the Study of Evolution
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1998 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Starmer, William T.; Polak, Michael; Wolf, Larry L.; Barker, J.S.F.
Date:Jun 1, 1998
Previous Article:Conflicts of interest in social insects: male production in two species of Polistes.
Next Article:Estimating the heritability of fluctuating asymmetry in field Drosophila.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters