Lisa M. Meffert

The effect of serial founder-flush cycles on quantitative genetic variation in the housefly

Experimental populations of the housefly, initiated from a single outbred natural population, were subjected to five serial founder events of one, four, or 16 pairs of flies. After each bottleneck the populations were allowed to flush to approximately 1000 pairs, at which time they were assayed for additive genetic variance for eight morphometric traits using parental–offspring covariances. Additive genetic variance for all bottleneck sizes rose above the level of the outbred control in response to the first bottleneck and remained comparable to or higher than that of the control over most of the successive bottleneck episodes; no bottleneck size exhibited additive genetic variance significantly below the level of the control throughout the experiment. Such changes in additive genetic variance in response to bottlenecks are inconsistent with a model of neutral additive gene action, suggesting that non-additive components of genetic variance are likely to have affected the traits. Two models of non-additive genetic variance, pure dominance and additive-by-additive epistasis, provided reasonable fits to our data, but were not distinguishable from each other. Both empirical and theoretical results suggest that additive genetic variance for quantitative traits can remain high despite repeated reduction of the population size to as low as a single mated pair. Historical bottlenecks cannot be accurately inferred from levels of additive genetic variance for complex quantitative characters that are affected by non-additive components of genetic variance.

MATING PROPENSITY AND COURTSHIP BEHAVIOR IN SERIALLY BOTTLENECKED LINES OF THE HOUSEFLY

The efficacy of bottlenecks to stimulate divergence in courtship behavior and consequent premating isolation was tested by serial founder‐flush episodes of three sizes (one, four, or 16 pairs) on a population of houseflies established in the laboratory from a single field population. After the fifth founder‐flush episode, intraline and interline crosses were performed to detect divergence in mating propensities and patterns of assortative mating. Videotapings of intraline courtships for the bottleneck lines and the control were evaluated for changes in courtship repertoire. All bottleneck lines showed significant divergence from the control in male and/or female mating propensity and in courtship behavior. Divergence from the control was bidirectional for both male and female mating propensities as well as for courtship element utilization. Out of 15 tests for assortative mating between bottleneck lines and between bottleneck lines and the control, only two cases of positive assortative mating and one case of negative assortative mating were detected. Because some bottleneck lines showed increased courtship element utilization and because decreased courtship utilization in some bottleneck lines was related to higher male mating success, the mechanisms behind the Kaneshiro model (which is based upon ancestral females discriminating against bottleneck males that had “lost” courtship elements) were not supported in general. A partitioning effect of the bottlenecks upon the intrinsic variation in the ancestral population for courtship pattern appeared to explain a large component of the directions of divergence from the control. Still, the pattern of divergence of some bottleneck lines apparently was not constrained by the intercorrelation structure of courtship behaviors detected in the control. Because previous studies showed that the bottleneck lines had rebounded from inbreeding depression to fitness levels of the control, this study documents nondebilitating differentiation in the courtship repertoire that can account for divergent mating propensities and premating isolation.

Fitness Rebound in Serially Bottlenecked Populations of the House Fly

Critics of theories of speciation via bottlenecks have argued that the loss of genetic variation as a result of a bottleneck would be sufficient toretard adaptation in founder populations or, in terms of Wrights adaptive-topography model, would lower the chance of a peak shift in relation to an outbred population (B. Charlesworth and Smith 1982; Barton and Charlesworth 1984; B. Charlesworth and Rouhani 1988). However, we have shown that additive genetic variation can actually increase in experimentally bottlenecked lines of the house fly (Bryant et al. 1986). More specifically, when dominance and epistasis contribute to the genetic variance, a portion of this nonadditive variance can be converted into additive genetic variance during a bottleneck to increase the overall level of additive genetic variance available for selectional response (Robertson 1952; Goodnight 1987, 1988). As a result, additive genetic variation would become available to a founder population when selection may be most intense in the invasion of a new territory. The major problem of such a scenario is that several measures of fitness in our bottleneck lines declined in relationship to the control line (Bryant et al. 1986, fig. 3). If this lowered fitness were to persist, these lines would not be viable progenitors of future populations (Hedrick 1987). In making such an assessment, however, it is important to follow the fitness of bottleneck lines for a longer period. Not all individuals within an inbred population are necessarily inbred to the same degree (Avery and Hill 1977, 1979; Franklin 1977; Weir et al. 1980; Cockerham and Weir 1983), and if there is a correlation between degree of inbreeding and fitness within these populations, natural selection could quickly restore fitness. This note reports on the changing fitness of our bottleneck lines through five successive bottleneck cycles. Although inbreeding depression is well documented in many organisms, particularly for Drosophila and Mus (review in D. Charlesworth and Charlesworth 1987), few studies have followed fitness changes of individual ines (Dobzhansky and Spassky 1947; Hollingsworth and Maynard Smith 1955); thus, recovery from inbreeding depression in most cases has been due to differential survival among inbred lines and not due to changes within

Nonadditive genetic effects in animal behavior.

Heritabilities, commonly used to predict evolutionary potential, are notoriously low for behaviors. Apart from strong contributions of environmental variance in reducing heritabilities, the additive genetic components can be very low, especially when they are camouflaged by nonadditive genetic effects. We first report the heritabilities of courtship traits in founder‐flush and control populations of the housefly (Musca domestica L.). We estimated the heritability of each male and female display through the regression of the courtships involving daughters and sons (with randomly selected mates) onto the “midparental” courtship values of their parents. Overall, the average heritability was significantly (ndocumentclass{aastex}nusepackage{amsbsy}nusepackage{amsfonts}nusepackage{amssymb}nusepackage{bm}nusepackage{mathrsfs}nusepackage{pifont}nusepackage{stmaryrd}nusepackage{textcomp}nusepackage{portland,xspace}nusepackage{amsmath,amsxtra}nusepackage[OT2,OT1]{fontenc}nnewcommandcyr{nrenewcommandrmdefault{wncyr}nrenewcommandsfdefault{wncyss}nrenewcommandencodingdefault{OT2}nnormalfontnselectfont}nDeclareTextFontCommand{textcyr}{cyr}npagestyle{empty}nDeclareMathSizes{10}{9}{7}{6}nbegin{document}nlandscapen

AN ANALYSIS OF SELECTIONAL RESPONSE IN RELATION TO A POPULATION BOTTLENECK

P=.012

Divergent ambulatory and grooming behavior in serially bottlenecked lines of the housefly

nend{document} ) higher for the parent‐daughter assays than for the parent‐son assays. We attributed the low (even negative) heritabilities to genotype‐by‐environment interactions whereby the male’s behavior is influenced by the “environment” of his mating partner’s preferences for the display, generating epistasis through indirect genetic effects. Moreover, bottlenecked lines had up to 800% of the heritability of the controls, suggesting “conversion” of additive genetic variance from nonadditive components. Second, we used line‐cross assays on separate populations that had been selected for divergence in mating behavior to identify dominance and epistasis through heterosis and outbreeding depression in courtship. Finally, our literature review confirms the prevalence of such low heritabilities (i.e., a conservative mean of 0.38) and nonadditive genetics in other behavioral repertoires (64% of the studies). We conclude that animal behavior is especially prone to the gamut of quantitative genetic complexities that can result in negative heritabilities, negative selection responses, inbreeding depression, conversion, heterosis, and outbreeding depression.

MORPHOMETRIC DIFFERENTIATION IN SERIALLY BOTTLENECKED POPULATIONS OF THE HOUSEFLY

Three measures of multivariate integration were derived from both additive genetic covariance and correlation matrices estimated from parent‐offspring covariances to investigate the effect of bottlenecks of different sizes on genetic integration of morphological traits in the housefly, Musca domestica L. Bottleneck lines were initiated with one, four, or 16 pairs of flies sampled from a natural outbred (control) population. Bottlenecks of intermediate size significantly increased the average genetic correlation among traits, resulting in nearly isomorphic variation among all traits in these lines. Single‐pair bottlenecks significantly disrupted the trait interrelationships, and the suites of traits identified by principal components of the additive genetic correlation and covariance matrices for the control population were no longer evident in these bottleneck lines. The alteration of the genetic relationships among traits as a result of a bottleneck suggests that nonadditive components of genetic variation affecting these traits were present in the control line. We discuss the implications of nonadditive gene action, particularly epistasis, for speciation via bottlenecks.

A test of how well the repeatability of courtship predicts its heritability

Selection for increased morphometric shape (ratio of wing length to thorax width) was compared between control (nonbottlenecked) populations and bottlenecked populations founded with two male–female pairs of flies. Contrary to neutral expectation, selectional response was not reduced in bottlenecked populations, and the mean realized heritabilities and additive genetic variances were higher for the bottlenecked lines than for the nonbottlenecked lines. Additive genetic variances based on these realized heritabilities were consistent with independent estimates of genetic variances based on parent–offspring covariances. Joint scaling tests applied to the crosses between selected lines and their controls revealed significant nonadditive components of genetic variance in the ancestor, which were not detected in the crosses involving bottlenecked lines. The nonbottlenecked lines responded principally by changes in one trait or the other (wing length or thorax width) but not in both, and regardless of which trait responded, larger trait size was dominant and epistatic to smaller size. Stabilizing selection for morphometric shape in the ancestor likely molded the genetic architecture to include nonadditive genetic effects.