Printer Friendly

An analysis of selectional response in relation to a population bottleneck.

The proximal rate of genetic differentiation among populations in quantitative traits, due to either drift or selection, is constrained by levels of additive genetic variance (Falconer 1989). The greatest amounts of additive genetic variance for such traits are expected to exist in large outbred populations; a reduction in population size would lead to a loss of additive genetic variance in direct relation to the severity of the population constriction (e.g., Lande 1980). Evolutionary theory also predicts that, as a result of selection, additive genetic variance for traits more closely associated with fitness should be minimized in equilibrium populations; consequently, such traits should be unresponsive to natural selection or genetic drift (Fisher 1930; Wright 1935; A. Robertson 1955; Istock 1983; but see Price and Schluter 1991). Even if additive genetic variance for fitness-related traits is maintained in equilibrium populations (Rose 1982; Mousseau and Roff 1987; Roff and Mousseau 1987; Price and Schluter 1991), a population bottleneck is expected to further reduce such genetic variance.

Fitness-related traits are predicted to have genetic variance sequestered by both epistatic and dominance components (e.g., Wright 1935, 1969, 1977; A. Robertson 1955). Rapid reductions in population size are predicted to cause a portion of the nonadditive genetic variance for such traits to be converted into additive genetic variance (Robertson 1952; Griffing 1960; Goodnight 1987, 1988; Cockerham and Tachida 1988; Gimelfarb 1989; Tachida and Cockerham, 1989a,b; Jiang and Cockerham 1990; Willis and Orr 1993). Conversions of this sort could then be important in adaptation for fitness traits by rendering them responsive to selection after a population bottleneck. Although the specific genetic mechanisms were not fully articulated, similar processes were originally envisaged by Mayr (1954) and Carson (1968) in their models of speciation via founder events or population bottlenecks.

In a series of papers beginning in 1986, we have used the housefly to explore the ways in which population bottlenecks may affect the levels of additive genetic variance and the patterns of differentiation of quantitative (morphometric) traits. Contrary to the traditional paradigm, we found that the additive component of genetic variance for morphometric traits increased in experimental lines that experienced bottlenecks (Bryant et al. 1986a; Bryant and Meffert 1991, 1993). This increased additive genetic variance allowed greater quantitative trait differentiation from the ancestor than predicted by the neutral expectation (Bryant et al. 1986b; Bryant and Meffert 1990), which appeared to entail a disruption of the genetic covariance relationships among traits (Bryant and Meffert 1988). We have attributed these results to the influence of nonadditive components of genetic variance in the base (ancestral) population, deriving indirect support from theory indicating that additive genetic variance can be extracted from nonadditive components through bottlenecks. This work illustrated the complex way in which founder-flush events may affect the levels of additive genetic variance for quantitative traits and, ultimately, their responses to selection. However, we did not provide direct evidence of nonadditive genetic structuring for these morphometric traits or of the altered response to selection in bottlenecked populations.

Both components of dominance and epistasis can be estimated in a large population through direct manipulations of genetic relatedness, as in a hierarchical half-sib design where multiple females are mated to individual males (Comstock and Robinson 1948; Cockerham 1956, 1963; Becker 1984; Falconer 1989) or by analysis of line crosses between genetically divergent populations, including those that have undergone artificial selection (Hayman 1958, 1960; Mather and Jinks 1982; Lynch and Walsh MS). Analysis of line crosses, for example, was used effectively by Hard et al. (1992) to show that nonadditive genetic processes, particularly epistasis, significantly influenced differentiation in photoperiodism among geographic populations of the pitcher-plant mosquito, Wyeomyia smithii.

To evaluate the potential role that bottlenecks may play in overcoming genetic constraints on differentiation in the ancestor (sensu Clark 1987a), we were interested in whether or not founder-flush events facilitate divergence along a morphological trajectory that is relatively inaccessible to the ancestral population. In a previous study (Bryant and Meffert 1990), multivariate morphometric differentiation of bottlenecked lines from the ancestor occurred disproportionately along axes of shape, as opposed to axes for general size. In contrast, differentiation was more evenly distributed across size and shape in less severely bottlenecked populations. The morphometric axes of shape thus appeared to represent avenues of differentiation that were facilitated by severe bottlenecks, and so in this experiment we select directly upon morphometric shape. As a measure of shape we use the logarithm of the ratio of wing length to thorax size, which was an indicator of divergence that was accelerated by bottleneck events in our previous studies (i.e., Bryant and Meffert 1990, Appendix). Wing/thorax ratio is also likely to be under strong stabilizing selection due to the functional constraints of flight (see Cowley and Atchley 1990), which may have a strong influence on the genetic architecture of a shape trait (e.g., Wright 1935; A. Robertson 1955). We compare selectional responses in morphometric shape in bottlenecked and nonbottlenecked populations, followed by crosses between the selected and nonselected lines, to reveal the genetic basis of shape differentiation. We therefore offer a direct evaluation of whether or not bottlenecks alter the response to selection and, as well, evaluate the role that dominance and epistasis may play in the response to artificial selection.


An outbred base population of houseflies was established in the laboratory from a single collection of several hundred flies taken from the Houston Zoological Gardens. After three generations, during which time the population flushed to a normal laboratory size of approximately 2000, we established four separate subpopulations: two nonbottlenecked lines (using a mass sampling of eggs from the base laboratory population) and two bottlenecked lines (each derived from the offspring of two male-female pairs). The four subpopulations established from the base population were designated [B.sub.1] and [B.sub.2] (bottlenecked lines) and [N.sub.1] and [N.sub.2] (nonbottlenecked lines). The bottlenecked populations were allowed to flush to the standard laboratory size before initiation of the selection protocol.

To establish each selection line, 80 females were drawn from each of the four subpopulations. Wing and thoracic measurements were taken with an ocular micrometer using a diffusion plate to immobilize flies with C[O.sub.2]. The wing measurement was the length of the first posterior cell (distance between the radial-medial cross vein and the intersection of the [radius.sub.4+5] with the exterior wing margin; West 1951); thoracic width was the distance across the dorsal transverse suture between the scutum and the scutellum. Shape for an individual fly was taken as the natural logarithm of the ratio of the wing length to the thoracic width. The females with the highest 20 of the 80 shape measurements were selected to seed the next generation of the selection line, resulting in an expected selection differential in females of 1.257 standard deviation units per generation (Falconer 1989, Appendix B). These selected females were provided with excess males from their own population and thus selection was carried out only on the females. Such selection is not expected to diminish the genetic variability substantially (Bohren 1975), so any differences between nonbottlenecked and bottlenecked populations were probably not attributable to additional inbreeding that occurred as a result of the selectional protocol. Nevertheless, a control (nonselected) line was established from each subpopulation using 20 randomly selected females provided with excess males to control for unidentified effects of finite population size that may have affected the selectional response.
TABLE 1. The matrix C giving the coefficients for the line means in
terms of the causal genetic components.

                   S    NS   [F.sub.1]   [F.sub.2]    BS      BNS

Overall mean       1     1       1           1         1       1
Additivity (a)     1    -1       0           0          .5      .5
Dominance (d)     -1    -1       1           0         0       0
a x a              1     1       0           0          .25     .25
a x d             -1     1       0           0         0       0
d x d              1     1       1           0         0       0

* S, selected line; NS, nonselected control; [F.sub.1], cross
between selected and nonselected (control) lines; [F.sub.2], result
of crosses within [F.sub.1]; BS, backcross of [F.sub.1] to selected
line; BNS, backcross of [F.sub.1] to nonselected control.

The experiment was carried out in two replicate blocks. Within each block, replicate selected and nonselected lines were established from each of the four subpopulations; the experiment thus involved eight selected lines and eight non-selected control lines arranged in two blocks. Lines are identified by two subscripts, the first identifying their source population and the second denoting the block number (e.g., [B.sub.12] would be bottlenecked population 1 from block 2). The selection protocol spanned nine generations. All lines were maintained for an additional five generations without selection, using 20 randomly chosen females supplied with excess males to seed each successive generation. Hence, the total experiment spanned 14 generations: 9 generations of selectional response followed by 5 generations of relaxed selection. All flies were reared under standard conditions of 80 eggs per 18 g CSMA larval medium at 27 [degrees] C (Bryant et al. 1986a).

In the tenth generation (the last generation of selectional response) crosses were made between the paired selected and nonselected (control) lines to estimate the components of genetic variance contributing to interline divergence. Each set of crosses included reciprocal (with regard to sex) [F.sub.1], [F.sub.2], and first order backcrosses of the [F.sub.1] to the selected or non-selected lines, yielding a total of six means (four crosses and two parental means). Twenty female flies from each cross were pinned and the wing/thorax ratios measured as before. From the vector of means across the six lines, the joint scaling test estimates the additive, dominance, and digenic epistatic (additive-by-additive, additive-by-dominance, and dominance-by-dominance) contributions to genetic differentiation, by a regression of line means onto a set of coefficients reflecting the expected contributions of genetic components (Hayman 1958; Mather and Jinks 1982). The matrix of coefficients, C, is given in table 1, from which the vector of estimates for the genetic components b is derived from

b = [([C.sup.t][V.sup.-1]C).sup.-1][C.sup.t][V.sup.-1]z,

where V is the diagonal matrix of squared standard errors for the line means, and z is a column vector of observed line means (t indicates transpose and -1 indicates inverse). Standard errors of the estimates are provided by the square root of the diagonal elements of the matrix [([C.sup.t][V.sup.-1]C).sup.-1] (Hayman 1958; Hard et al. 1992; Lynch and Walsh MS). Goodness-of-fit of the model is thus tested by the sum of squared differences between line means and values estimated from the model, each divided by the square of the standard error of the respective line mean. The goodness-of-fit statistic is distributed as a [[Chi].sup.2] with 6 - k degrees of freedom, where k indicates the number of fitted parameters (Mather and Jinks 1982). In the joint scaling method, goodness-of-fit is first carried out to determine whether the additive model alone provides an adequate fit to the vector of line means. If this minimal model is rejected, higher-order genetic parameters are included up to the full six-parameter solution, including all digenic interactions of additive and dominance components. The full six-parameter model exactly fits the observed line means; thus, these means can be retrieved exactly from the matrix product, z = bC. As a result, the efficacy of the fit cannot be tested; however, the standard errors of the individual estimates provided by the square roots of the diagonal elements of [([C.sup.t][V.sup.-1]C).sup.-1] can be used to assess the significance of the contributions of the individual genetic components to interline divergence (Mather and Jinks 1982; Hard et al. 1992).

Heritabilities and additive genetic variances were also estimated independently of the selectional protocol through parent-offspring regressions within each replicate of the subpopulations kept as large stock populations through the term of the experiment. For each line, offspring were derived from individual male-female pairs; morphometric measurements were obtained on parents and three of their female offspring. As for the selectional protocol, all flies for these tests were reared under standard larval densities. A total of 90 families were measured for the two replicate founder-flush populations (43 and 47 families) and 129 families were used for estimates on the two ancestral nonselected base populations (74 and 55 families). These estimates were compared with the realized heritabilities obtained from the selection experiment.


Selectional Response

The average selectional responses (presented as deviations from the initial population) are given in figure 1 for the nonbottlenecked lines and figure 2 for the bottlenecked lines. Morphometric shape is unlikely to drift to the extent of a neutral trait because of the potential constraints of stabilizing selection. There was no significant linear trend in the control (unselected) line means over the course of the experiment. Consequently, the selected lines could be compared directly with the overall mean values of the controls. To test statistically for selectional response in relation to these control means, one still needs to account for cumulative variance among control lines caused by drift. As for the means, there were no significant linear trends in the standard deviation among control lines over the course of the experiment. Nevertheless, to be conservative, the regression slope of the standard deviation among control lines onto generation number was used to increase incrementally the 95% confidence bounds for testing the significance of the selectional response.

On the basis of such confidence intervals around the control lines, the selected lines differed significantly from their control populations after only a few generations. As a further test of the selectional responses, the mean wing/thorax ratios for all of the nonbottlenecked and bottlenecked selected lines were tested against the means of their respective nonselected control lines in the ninth generation: responses for all lines were highly significant (P [less than] 0.001) and, overall, the lines exhibited responses of approximately seven log-units above their control. The responses of the bottlenecked and nonbottlenecked lines differed after cessation of selection: The nonbottlenecked lines drifted back toward the control level, although the regression of selected means onto generation number was not significant (b = -0.314 [+ or -] 0.130; P [greater than] 0.05). However, the rate of reversion for the bottlenecked lines was significant (b = -0.962 [+ or -] 0.238, P [less than] 0.05).

An estimate of the realized heritability (and additive genetic variance) for wing/thorax ratio is obtained from the slope of the regression of the cumulative selectional response onto the cumulative selection differential (Falconer 1989). Since only females were selected (males were taken randomly from the subpopulation) the realized heritability is then twice the value of this regression slope. The standard errors for the realized heritabilities were obtained using the weighted least-squares method outlined in Hill (1972) and in Lynch and Walsh (chap. 28, MS).

Realized heritabilities (and their standard errors) were 0.302 [+ or -] 0.028 and 0.342 [+ or -] 0.025 for the two nonbottlenecked lines and 0.441 [+ or -] 0.031 and 0.268 [+ or -] 0.025 for the two founder-flush lines [ILLUSTRATION FOR FIGURE 3 OMITTED]. The highest and lowest realized heritabilities occurred in the bottleneck lines, yet the mean realized heritability for the two bottlenecked lines was slightly higher than that for the two nonbottlenecked controls (0.354 and 0.322, respectively). Estimates of the additive genetic variances for wing/thorax ratio in the base populations can be obtained as the product of these heritabilities and the phenotypic variances. The average additive genetic variance for the two bottlenecked lines (multiplied by [10.sup.4] for convenience) was greater than that for the two nonbottlenecked lines (8.52 and 6.19 for the bottlenecked and nonbottlenecked lines, respectively).

Narrow-Sense Heritabilities

Narrow-sense heritabilities for the two nonbottlenecked lines obtained through parent-offspring covariances on the stock subpopulations were 0.670 [+ or -] 0.087 ([N.sub.1]) and 0.665 [+ or -] 0.103 ([N.sub.2]), (mean [h.sup.2] = 0.667 [+ or -] 0.067). In comparison, the parent-offspring heritabilities for the two bottlenecked lines were 0.417 [+ or -] 0.142 ([B.sub.1]) and 0.458 [+ or -] 0.132 ([B.sub.2]) (mean [h.sup.2] = 0.437 [+ or -] 0.097). Additive genetic variances (x [10.sup.4]) were 5.31 ([N.sub.1]), 7.63 ([N.sub.2]), 5.92 ([B.sub.1]), and 6.38 ([B.sub.2]). The average additive genetic variance for the two bottlenecked subpopulations (6.15) was nearly equal to that for the two base subpopulations (6.47).
TABLE 2. Mean shape, and standard errors (in parentheses) of
reciprocal [F.sub.1] crosses between selected (S) and nonselected
control (C) lines for the two blocks of the experiment. Student's
t-value is given for difference between means of reciprocal crosses.
N, nonbottlenecked lines; B, bottlenecked lines.

Line             S x C           C x S       Student's t

Block one

[N.sub.11]   0.336 (0.006)   0.340 (0.006)   0.48 NS
[N.sub.21]   0.293 (0.006)   0.300 (0.005)   0.95 NS
[B.sub.11]   0.334 (0.007)   0.331 (0.006)   0.27 NS
[B.sub.21]   0.256 (0.007)   0.257 (0.009)   0.13 NS

Block two

[N.sub.12]   0.337 (0.006)   0.320 (0.006)   0.93 NS
[N.sub.22]   0.289 (0.008)   0.277 (0.006)   1.89 NS
[B.sub.12]   0.345 (0.007)   0.316 (0.006)   2.99 P [less than] 0.05
[B.sub.22]   0.276 (0.011)   0.274 (0.008)   0.17 NS

Line Crosses

In the line crosses beyond [F.sub.1] ([F.sub.2] and first-order back-crosses) only unidirectional crosses were carried out: it is therefore important to determine whether maternal and/or sex-linked effects may confound the analyses. These effects would be evident as differences between reciprocal (by sex) [F.sub.1] crosses between selected and nonselected control lines. [TABULAR DATA FOR TABLE 3 OMITTED] Means of reciprocal [F.sub.1] crosses are given in table 2, where only one significant difference occurred, that is, between the selected and control lines for [B.sub.12]. In subsequent analyses of line crosses, means of the reciprocal [F.sub.1] lines were used; some bias, however, may confound the estimates for line [B.sub.12] (but see below).

Least-squares estimation of parameters in the additive and additive-dominance model are given in table 3, along with the goodness-of-fit of each model, for both blocks of bottlenecked and nonbottlenecked selection lines. Neither of the two nonbottlenecked populations fit the simple model of additivity, for either block, but the additive genetic model satisfactorily fits the means for all of the bottlenecked lines with the exception of [B.sub.22]. Crosses involving line [B.sub.12] fit the simple additive model, and thus there was no indication that the difference in reciprocal [F.sub.1] means generated higher-order effects. Adding simple dominance to the model did not result in a fit to the observed line means; those crosses that deviated significantly from a simple additive model continued to deviate from the additive-dominance model, so that higher-order genetic effects are likely to have been involved in the selectional responses.

Estimates for the full six-parameter model are given in table 4. Significant effects were found in the nonbottlenecked lines (in both blocks) for dominance and digenic epistasis, including additive-by-additive interactions, as well as dominance-by-dominance interactions. Components for dominance, as well as digenic epistasis, structured the genetic architecture of morphometric shape in the nonbottlenecked populations; however, such structuring was less evident in the bottlenecked lines, where only line [B.sub.22] showed some borderline significance for dominance and additive-by-additive epistasis.

Although both blocks of tests on the ancestral (nonbottlenecked) populations exhibited significant dominance and digenic epistasis, the signs of these components changed from [TABULAR DATA FOR TABLE 4 OMITTED] positive to negative indicating a larger wing/thorax ratio was dominant (and epistatic) over a smaller ratio in the first block, whereas the reverse was true for the second block. The responses of wing length and thoracic size were evaluated separately to determine whether these responded in similar fashions across blocks. Table 5 gives the ([log.sub.e] transformed) deviations for wing length and thorax size of each selected line from its nonselected control at generation 10. Morphometric shape for the nonbottlenecked lines responded primarily through increased wing length in the first block, but by decreased thoracic size in the second block; in contrast, the founder-flush lines tended not to show such differential block effects. Consequently, the disparity between the two blocks for the nonbottlenecked lines was due to wing/thorax ratio being dominant when increased wing length was primarily responsible for changes in shape (block 1). The smaller shape value was dominant when scutellum width was responsible for the selectional response (block 2). Similarly, the significant digenic epistatic component also depended on which of the two traits responded more to selection; that is, it was positive for responses in wing length but negative for reduced thorax size. Hence, larger size was dominant and epistatic over small size, with the direction of the effect depending on the trait that exhibited greater response to selection on shape.
TABLE 5. Deviations of 1n-wing length (WL) and 1n-thorax size (TS)
of selected line means from their nonselected control means, for the
two blocks of the experiments. N, nonbottlenecked lines; B,
bottlenecked lines.

                     Block one                   Block two
Line            WL             TS            WL           TS

[N.sub.1]   0.0669(***)   -0.0058        0.0281      -0.0813(***)
[N.sub.2]   0.0513(***)   -0.0393(*)     0.0260      -0.0875(***)
[B.sub.1]   0.0155        -0.0783(***)   0.0390(*)   -0.0558(**)
[B.sub.2]   0.0170        -0.0343        0.0385(*)   -0.0347

* P [less than] 0.05, ** P [less than] 0.01, *** P [less than]


Bottlenecked and nonbottlenecked lines had nearly identical additive genetic variances for morphometric shape, whether or not these were based on selectional response (realized heritabilities) or parent - offspring covariances. Thus, reducing the population to only two pairs of flies for one generation did not lower the evolutionary potential of the populations, as in the traditional view based on additive genetic effects alone (e.g., Robertson 1960; Hill and Robertson 1966; James 1971; Frankham 1980; Barton and Charlesworth 1984). Our results are anomalous with these predictions, but are consistent with our previous work on bottleneck effects, where we found that heritabilities (and additive genetic variances) of morphometric traits obtained from parent - offspring covariances were the same (or higher) in bottlenecked versus nonbottlenecked lines (Bryant et al. 1986a; Bryant and Meffert 1993).

Expected multivariate heritability along vectors depicting size and shape can be computed as the ratio of quadratic forms (Lin and Allaire 1977),

[h.sup.2] = ([a.sup.t]Ga)/([a.sup.t]Pa).

where a is a vector representing a size (or shape) axis (and [a.sup.t] its transpose), and G and P are the additive genetic and phenotypic covariance matrices, respectively, obtained through parent-offspring assays for nonselected bottlenecked and nonselected nonbottlenecked base populations. Taking the vector a = [1 1] to represent size for ([log.sub.e]-transformed) traits, the average heritabilities for the nonbottlenecked and bottlenecked lines were 0.587 and 0.165, respectively. Taking the shape vector as a = [1 -1], average heritabilities were 0.563 and 0.541 for the bottlenecked and nonbottlenecked treatments, respectively. Hence, even though heritability for morphometric shape was not altered by the bottleneck, the heritability for size would be expected to have been lowered considerably in founder-flush lines. Previously, we found that severely bottlenecked lines (reduced to a single pair) differentiated from the ancestral control primarily along avenues of morphometric shape, whereas such differentiation was more evenly distributed across morphometric axes in less severely bottlenecked populations (Bryant and Meffert 1990). The lowered average heritability for morphometric size but not shape, based on the parent-offspring covariances in severely bottlenecked lines in this study is consistent with these previous results, in that it would predict greater drift along axes of shape than size.

Upon cessation of artificial selection, the reversion of a trait toward its unselected level is interpreted as the action of natural selection directly on the trait or through a correlated response with a fitness trait (Falconer 1989). The reversion toward unselected control levels after cessation of selection was more evident in the bottlenecked lines than in the nonbottlenecked lines. If the rate of reversion were simply due to the action of natural selection, it is unclear why it would be more severe in the bottlenecked lines. However, if the response to artificial selection involved additive-by-additive epistasis, a large part of the response would be due to linkage disequilibrium that would have decayed rapidly after selection ceased (Bulmer 1985). The selectional response in the nonbottlenecked lines apparently resulted in more stable phenotypic shifts than in the bottlenecked lines, implicating possible differential influences of linkage and epistasis in the bottlenecked and nonbottlenecked treatments.

With respect to shifts occurring as a result of the bottleneck alone, the wing/thorax ratio for line [B.sub.2] was significantly lower than that for all other lines (at P [less than] 0.05), indicating that the bottleneck caused a shift in phenotypic space for this line without affecting the response to selection. In founder-flush theories of speciation, the population bottleneck is invoked precisely to facilitate phenotypic shifts that are difficult to accomplish in large populations (e.g., Mayr 1954). To the extent that morphological changes in shape may foreshadow events leading to novel differentiation and speciation, these data indicate that bottlenecks can catalyze the shifts as envisaged in these founder-flush models (see also Carson 1968, 1971; Templeton 1980).

The line-cross data provide direct evidence that the divergence of the selected lines from the unselected controls (for the nonbottlenecked lines) involved more than purely additive genetic effects. Larger trait values for either wing length or thorax size were dominant and epistatic over smaller ones. By implication, the increases in additive genetic variance for morphometric traits we observed in earlier studies (Bryant et al. 1986a; Bryant and Meffert 1993) could have been due to a conversion of nonadditive genetic variance to the pool of available additive genetic variance, as predicted by Goodnight (1988). However, F. Robertson (1955) noted that inbreeding, which cannot be entirely avoided in selection experiments, may amplify epistatic effects. Specifically, as alleles are lost during inbreeding, the remaining alleles exaggerate their segregational effects. It is unlikely that this mechanism underlies the epistatic effects observed here because the epistatic contributions to differentiation were more pronounced in the nonbottlenecked than in the bottlenecked lines, the reverse of the prediction. Therefore, we feel that the increases in additive genetic variance for morphometric traits observed earlier could well rest on the influences of epistatic components and not solely on the influences of rare recessive alleles in the base population, as suggested by Willis and Orr (1993).

In applying the line-cross method, Mather and Jinks (1982) suggested that transformations of scale can reduce or eliminate nonadditive effects. However, as the objective of our experiment was to detail the genetic structure of morphometric shape, the appropriate scale must be logarithmic (Jolicoeur 1963; Gould 1977). Even so, selectional responses on the order of 8%-10% of the original value would not be materially affected by the transformation (i.e., [log.sub.e](1 + x) [congruent] x for x less than about 0.15; Lewontin 1966; Bryant 1986). Therefore, the detection of significant contributions of dominance and epistasis to trait divergence were more likely due to fundamental nonadditive genetic architecture of morphometric shape in the ancestral population than to inappropriate transformation.

In his review, Barker (1979) found little experimental evidence of epistasis influencing quantitative traits. Coyne and Beecham (1987) and Clark (1987b) also found little evidence for nonadditive genetic variance in Drosophila for morphometric traits and viability, respectively. With the exception of lethals (Mukai et al. 1972; Simmons and Crow 1977), dominance seems to be rare for morphometric traits as well (Robertson 1967). Similarly, most studies report reduced selectional response in bottlenecked or inbred lines relative to outbred populations (Robertson [1969] reported in James 1971; Frankham 1980; Lynch and Walsh MS). Why, then, might some morphometric traits, particularly morphometric shape, be structured by nonadditive genetic components of variance? A fitness-related trait under stabilizing selection may operate epistatically (Wright 1935; Gimelfarb 1989; Hastings 1989; Turelli and Barton 1990). Adult size in Diptera, under stabilizing selectional pressure for fast larval development (small size) and high adult fecundity (large size) (Partridge and Fowler 1993), would potentially have evolved nonadditive genetic structure (Kearsey and Kojima 1967). The scaling of wing length to thorax (body) size should be critical to insect flight (e.g., Cowley and Atchley 1990) and therefore also should be a likely candidate for nonadditive effects under the influence of stabilizing selection. Thus, results may depend on the genetic architectures of the traits under a study which, in turn, may depend on the nature of their selectional histories, as suggested by Mather (1953, 1966) and F. Robertson (1955). In this regard, not all traits within the same organism are expected to have identical genetic architectures, and, therefore, the responses to selection or bottlenecks may also vary among them.

Since Wright (1931, 1940) postulated the facilitation of evolutionary differentiation among partially isolated populations by epistasis, there have been few tests either of the model (e.g., Wade and Goodnight 1991) or of potential epistatic structuring of quantitative traits (Barker 1979; Clark 1987b). The primary reason for the latter may be the logistical difficulty in performing the appropriate experiments to show such epistatic effects. As a consequence, whether or not epistatic effects are widespread remains moot. Additional experiments on the underlying genetic architectures of morphometric traits and the mechanisms responsible for selectional divergence seem imperative if we are to fully understand the evolution of quantitative traits in natural populations.


Support for this study was supplied from National Science Foundation (NSF) grant (BSR-8800977); additional support came from NSF grants BSR-9024980 and BSR-9106591 and from The University of Houston Coastal Center. We thank V. Backus, C. Goodnight, B. Walsh, and two anonymous reviewers for their comments on an earlier draft of the manuscript.


Barker, J. S. F. 1979. Inter-locus interactions: a review of experimental evidence. Theoretical Population Biology 16:323-345.

Barton, N. H., and B. Charlesworth. 1984. Genetic revolutions, founder events and speciation. Annual Review of Ecology and Systematics 15:133-164.

Becker, W. A. 1984. Manual of quantitative genetics. Academic Enterprises, Pullman, Wash.

Bohren, B. B. 1975. Designing artificial selection experiments for specific objectives. Genetics 80:205-220.

Bryant, E. H. 1986. On the use of logarithms to accommodate scale. Systematic Zoology 35:552-559.

Bryant, E. H., S. A. McCommas, and L. M. Combs. 1986a. The effect of an experimental bottleneck upon quantitative genetic variation in the housefly. Genetics 114:1191-1211.

Bryant, E. H., L. M. Combs, and S. A. McCommas. 1986b. Morphometric differentiation among experimental lines of the housefly in relation to a bottleneck. Genetics 114:1213-1223.

Bryant, E. H., and L. M. Meffert. 1988. Effect of an experimental bottleneck on morphological integration in the housefly. Evolution 42:698-707.

-----. 1990. Multivariate phenotypic differentiation among bottleneck lines of the housefly. Evolution 44:660-668.

-----. 1991. The effect of bottlenecks on genetic variation, fitness, and quantitative traits in the housefly. Pp. 591-601 in E. C. Dudley, ed. The unity of evolutionary biology. Dioscorides Press, Portland, Oreg.

-----. 1993. The effect of serial flounder-flush cycles on quantitative genetic variation in the housefly. Heredity 70:122-129.

Bulmer, M. J. 1985. The mathematical theory of quantitative genetics. Clarendon Press, Oxford.

Carson, H. L. 1968. The population flush and its genetic consequences. Pp. 123-137 in R. C. Lewontin, ed. Population biology and evolution. Syracuse University Press, Syracuse, N.Y.

-----. 1971. Speciation and the founder principle. Stadler Genetics Symposium 3:51-70.

Clark, A. G. 1987a. The quantitative genetics of evolutionary constraints. Pp. 25-45 in V. Loeschke, ed. Genetic constraints on adaptive evolution. Springer, Berlin.

-----. 1987b. A test of multilocus interaction in Drosophila melanogaster. American Naturalist 130:283-299.

Cockerham, C. C. 1956. Analysis of quantitative gene action. Brookhaven Symposium in Biology 9:56-68.

-----. 1963. Estimation of genetic variances. Statistical genetics and plant breeding. National Research Council Publication (Canada) 982.

Cockerham, C. C., and H. Tachida. 1988. Permanency of response to selection for quantitative characters in finite populations. Proceedings of the National Academy of Sciences, USA 84:6205-6209.

Comstock, R. E., and H. F. Robinson. 1948. The components of genetic variance in populations of biparental progenies and their use in estimating the average degree of dominance. Biometrics 4:254-266.

Cowley, D. E., and W. R. Atchley. 1990. Development and quantitative genetics of correlation structure among body parts of Drosophila melanogaster. American Naturalist 135:242-268.

Coyne, J. A., and E. Beecham. 1987. Heritability of two morphological characters within and among natural populations of Drosophila melanogaster. Genetics 117:727-737.

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

Fisher, R. A. 1930. The genetical theory of natural selection. Clarendon Press, Oxford.

Frankham, R. 1980. The founder effect and response to artificial selection in Drosophila. Pp. 87-90 in A. Robertson, ed. Proceedings of the international symposium on selection experiments on laboratory and domestic animals. CAB Books, London.

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 41:171-179.

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

Gould, S. J. 1977. Ontogeny and phylogeny. Belknap, Cambridge, Mass.

Griffing, B. 1960. Theoretical consequences of truncation selection based on the individual phenotype. Australian Journal of Biological Sciences 13:307-343.

Hard, J. J., Wiliam E. Bradshaw, and C. M. Holzapfel. 1992. Epistasis and the genetic divergence of photoperiodism between populations of the pitcher-plant mosquito, Wyeomyia smithii. Genetics 131:389-396.

Hastings, A. 1989. Linkage disequilibrium and genetic variances under mutational-selection balance. Genetics 121:857-860.

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.

Hill, W. G. 1972. Estimation of realized heritabilities from selection experiments. II. Selection in one direction. Biometrics 28:767-780.

Hill, W. G., and A. Robertson. 1966. The effect of linkage on limits to artificial selection. Genetical Research 8:269-294.

Istock, C. 1983. The extent and consequences of heritable variation for fitness characters. Pp. 61-96 in C. R. King and P. S. Dawson, eds. Population biology: retrospect and prospect. Columbia University Press, New York.

James, J. W. 1971. The founder effect and response to artificial selection. Genetical Research 16:241-250.

Jiang, C., and C. C. Cockerham. 1990. Quantitative genetic components within restricted populations. Crop Science 30:7-12.

Jolicoeur, P. 1963. The multivariate generalization of the allometry equation. Biometrics 19:497-499.

Kearsey, M. J., and K. Kojima. 1967. The genetic architecture of body weight and egg hatchability in Drosophila melanogaster. Genetics 56:23-37.

Lande, R. 1980. Genetic variation and phenotypic evolution during allopatric speciation. American Naturalist 116:463-479.

Lewontin, R. C. 1966. On measurement of relative variability. Systematic Zoology 15:141-142.

Lin, C. Y., and F. R. Allaire. 1977. Heritability of a linear combination of traits. Theoretical Applied Genetics 51:1-3.

Mather, K. 1953. Genetical control of stability and development. Heredity 7:297-336.

-----. 1966. Variability and selection. Proceedings of the Royal Society of London B, Biological Sciences 164:328-340.

Mather, K., and J. L. Jinks. 1982. Biometrical genetics: the study of continuous variation. Chapman and Hall, New York.

Mayr, E. 1954. Changes of genetic environment and evolution. Pp. 156-180 in J. S. Huxley, A. C. Hardy, and E. B. Ford, eds. Evolution as a process. Allen & Unwin, London.

Mousseau, T. A., and D. A. Roff. 1987. Natural selection and the heritability of fitness components. Heredity 59:181-197.

Mukai, T., S. I. Chigusa, L. E. Merrier, and J. F. Crow. 1972. Mutation rate and dominance of genes affecting viability in Drosophila melanogaster. Genetics 72:35-355.

Partridge, L., and K. Fowler. 1993. Responses and correlated responses to artificial selection on thorax length in Drosophila melanogaster. Evolution 47:213-226.

Price, T., and D. Schluter. 1991. On the low heritability of life-history traits. Evolution 45:853-861.

Robertson, A. 1952. The effect of inbreeding on the variation due to recessive genes. Genetics 37:189-207.

-----. 1955. Selection in animals: synthesis. Cold Spring Harbor Symposia in Quantitative Biology 20:225-229.

-----. 1960. A theory in limits in artificial selection. Proceedings of the Royal Society of London B, Biological Sciences 153:234-249.

-----. 1967. The nature of quantitative genetic variation. Pp. 265-280 in R. A. Brink, ed. Heritage from Mendel. University of Wisconsin Press, Madison.

Robertson, F. 1955. Selectional response and the properties of genetic variation. Cold Spring Harbor Symposia in Quantitative Biology 20:166-177.

Roff, D. A., and T. A. Mousseau. 1987. Quantitative genetics and fitness: lessons from Drosophila. Heredity 58:103-118.

Rose, M. R. 1982. Antagonistic pleiotropy, dominance, and genetic variation. Heredity 48:63-78.

Simmons, M. J., and J. F. Crow. 1977. Mutations affecting fitness in Drosophila populations. Annual Review of Genetics 11:49-78.

Tachida, H., and C. C. Cockerham. 1989a. Effects of identity disequilibrium and linkage on quantitative variation in finite populations. Genetical Research 53:63-70.

-----. 1989b. A building block model for quantitative genetics. Genetics 121:839-844.

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

Turelli, M., and N. H. Barton. 1990. Dynamics of polygenic characters under selection. Theoretical Population Biology 38:1-57.

Wade, M. J., and C. J. Goodnight. 1991. Wright's shifting balance theory: an experimental study. Science 253:1015-1018.

West, L. 1951. The housefly. Comstock, Ithaca, N.Y.

Willis, J. H., and H. A. Orr. 1993. Increased heritable variation following population bottlenecks: the role of dominance. Evolution 47:949-957.

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

-----. 1935. Evolution in populations in approximate equilibrium. Journal of Genetics 30:257-266.

-----. 1940. Breeding structure of populations in relation to speciation. American Naturalist 74:232-248.

-----. 1969. Evolution and the genetics of populations, vol. 2. University of Chicago Press, Chicago.

-----. 1977. Evolution and the genetics of populations, vol. 3. University of Chicago Press, Chicago.
COPYRIGHT 1995 Society for the Study of Evolution
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1995 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Bryant, Edwin H.; Meffert, Lisa M.
Date:Aug 1, 1995
Previous Article:Alternative routes to the evolution of competitive ability in two competing species of Drosophila.
Next Article:Artificial selection for developmental time in Drosophila melanogaster in relation to the evolution of aging: direct and correlated responses.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters