Genet. Res., Camb. (1993), 62, pp. 213-222 With 1 text-figures Copyright © 1993 Cambridge University Press 213 Artificial sexual selection alters allometry in the stalk-eyed fly Cyrtodiopsis dalmanni (Diptera: Diopsidae) GERALD S.WILKINSON Department of Zoology, University of Maryland, College Park, Maryland 20742, USA (Received 26 February 1993 and in revised form 6 July 1993) Summary Selection for increased and decreased ratio of eye span to body length was exerted on male stalk- eyed flies (Cyrtodiopsis dalmanni) from Malaysia using replicate selected and unselected lines. Response to selection was symmetrical. After 10 generations high line male eye span increased to 1-3 body lengths while low line male eye span declined to 11 body lengths. Realized heritabilities for eye span to body length ratio, estimated using regressions of deviations from unselected controls on cumulative selection differentials, were greater than zero for all four selected lines with average h2 = 0-35 + 006. The static linear allometric relationship between eye span and body length diverged between selected lines and rotated among selected line males in the same direction as among males in other sexually dimorphic diopsid species. Crosses between lines after 13 generations of selection indicate that the genes which influence relative eye span combine additively and do not exhibit sex linkage or maternal effects. The genetic correlation between the sexes, 0-29 + 005 as estimated by the regression of female on male change in eye span, did not prevent sexual dimorphism in eye span from diverging between lines. These results suggest that the exaggerated eye span of male C. dalmanni is maintained by natural selection opposing sexual selection rather than by lack of or asymmetry in additive genetic variation. Furthermore, the variation in sexual dimorphism for eye span-body length allometry observed among extant diopsid species is consistent with sexual selection of variable intensity acting on relative eye span. additive genetic variation. Thus, continued elabor- 1. Introduction ation of sexually-selected traits, at least in these Exaggerated sex-limited weapons or ornaments species, is not prohibited by lack of genetic variation. favored by sexual selection may attain a limit for one However, breeding experiments do not reveal if of two reasons. Either additive genetic variation for response to selection will be symmetrical. continued elaboration is exhausted or natural selection If sexually dimorphic traits have experienced pro- opposes further exaggeration and results in stabilizing longed directional selection, then less additive genetic selection where additive genetic variation is main- variation for increased than for decreased elaboration tained by a balance between mutation and selection can be expected for at least three reasons. Mutations (Fisher, 1958; Lande, 1976; Lande, 1981). Either may reduce the expression of a complex sexually- persistent female choice of a male character or selected trait more often than increase it (Pomian- elaboration of a structure used to settle male contests kowski et al. 1991), the frequencies of genes that could lead to either genetic outcome (Lande, 1981; influence the trait may be near their upper limits Charlesworth, 1984; Maynard Smith & Brown, 1986). (Falconer, 1981), and genes that influence the trait Although field estimates of heritability (Prout & may exhibit directional dominance (Fisher, 1958). Barker, 1989; Riska et al. 1989) have not been Artificial selection on a male display or combat reported, breeding experiments on a variety of insects character should, therefore, result in more rapid (Carson & Lande, 1984; Butlin & Hewitt, 1986; evolutionary response toward reduced expression than McLain, 1987; Simmons, 1987; Wilkinson, 1987; toward increased elaboration. Unfortunately, sym- Hedrick, 1988; Moore, 1989; Simmons & Ward, metry of response cannot be evaluated in the two 1991) and fish (Houde, 1992; Bakker, 1993) indicate studies (Cade, 1981; Carson & Teramoto, 1984) that sexually-selected characters typically possess where artificial selection was exerted on sexually- G. S. Wilkinson 214 selected traits because selection only proceeded for a species gather into aggregations on root threads under few generations and the response was not measured as overhanging embankments along forest streams in a deviation from controls. However, 24 of 30 attempts Malaysia. Males compete for control of groups of to select divergently on a reproductive fitness trait females by comparing eye spans and escalate confron- have shown higher realized heritabilities for dimin- tations by grappling with their forelegs when eye span ished than improved performance (Frankham, 1990), is similar (de la Motte & Burkhardt, 1983). Winners of as expected if breeding values are not normally contests invariably displace smaller males prior to distributed around phenotypic means. darkness and at dawn mate with most of the females In this study I exerted divergent selection on an in the aggregation (Lorch et al. 1993). C. whitei exaggerated male trait to assess symmetry of response females prefer to alight near model males with and the extent to which genes influencing that trait act experimentally elongated eye span (Burkhardt & de la additively. The study organism is the Malaysian Motte, 1988) suggesting that female choice, in addition stalk-eyed fly, Cyrtodiopsis dalmanni, and the trait is to male competition, may favor exaggerated eye relative male eye span, i.e. the ratio of the distance span in these flies. between the tips of a male's eyes and his body length (Fig. 1). Sexual dimorphism in eye span is pronounced in this and the sympatric congener, C. whitei, because 2. Materials and methods the slope of the regression of eye span on body length, (i) Laboratory procedures or what I will refer to as linear static allometry (Cock, 1966; Klingenberg & Zimmermann, 1992), is greater Over 20O C. dalmanni were netted along small forest than one in males but less than one in females streams near Kuala Lumpur, Malaysia, transported (Shillito, 1971; Burkhardt & de la Motte, 1985). to Maryland, and housed in a 40 x 40 x 120 cm Consequently, differences in eye span magnify dif- population cage in January 1989. This cage is kept in ferences in body size in males. a 25 °C constant temperature room on a 12 h light- The static allometric relationship between eye span dark cycle with a 30 min dawn/dusk period provided and body length also varies greatly between males of by a 25 W incandescent light. Humidity is kept high different diopsid species (Burkhardt & de la Motte, by lining the cage with moist cotton and blotting 1985). For example, C. quinqueguttata and the more paper. Twice each week fresh ears of corn are ground primitive Sphyrecephala brevicornis (Feijen, 1989) and provided in disposable dishes as food for adults. exhibit no dimorphism in eye span and no difference Mold is inhibited by adding 5 ml of a 10% solution of between the sexes in eye span-body length allometry methylparaben in 90 % ethanol to each liter of corn (Burkhardt & de la Motte, 1985; unpublished data). pulp prior to autoclaving for 30 min. Because flies Because the eye span on body length regression for C. both feed and oviposit on the processed corn, larvae quinqueguttata is similar in slope to that of female C. are reared by introducing food in plastic cups rather whitei and C. dalmanni (Burkhardt & de la Motte, than dishes. Twice each week these food cups are 1985), the elevated slope of male C. whitei and C. transferred into larger 500 ml containers lined with dalmanni eye span on body length regressions probably damp cotton and plugged with foam stoppers to represents an evolutionarily derived state. Further- permit the larvae to climb out of the cups and pupate more, elongated eye spans among male C. whitei and in the cotton. Newly eclosed flies are returned to the C. dalmanni do not represent a correlated response to cage as necessary to maintain the population at selection for larger size (Lande, 1979) because males approximately 300 flies. Because the minimum gener- of these species are typically smaller in length than ation time is eight weeks and adult flies live six months male C. quinqueguttata (Burkhardt & de la Motte, or more in captivity (unpublished data), the base 1985). Similar increases in the slope of male eye span- population had been in captivity for less than 7 body length regressions have occurred between generations prior to the onset of selection in March sexually monomorphic and dimorphic species in at 1991. least three other genera of diopsids (Burkhardt & de Artificial sexual selection was exerted on the ratio la Motte, 1985; unpublished data). Thus, in addition of eye span to body length in males for four replicate to determining eye span response to selection, I also lines. Two lines each were selected for high and low examine male and female allometric relationships for ratios by taking either the 10 highest or 10 lowest any correlated responses to selection. A response in ratios of 50 measured males and housing them with 25 eye span-body length allometry to sexual selection on randomly selected virgin females in a 40 x 40 x 40 cm relative eye span would demonstrate additive genetic cage. Two unselected lines were maintained at the variation for static allometry and provide a mechanism same time by randomly selecting 10 males and 25 for the diversification of male morphology observed females. Larvae were reared in incubators at 25 + 1 °C in this group of unusual flies. in cups containing 50 ml of corn pulp. Any pupae Available evidence indicates that sexual selection remaining in cups were transferred to the cotton could influence sexual dimorphism for eye span in C. before eclosion. Three cups of food were provided dalmanni and C. whitei. Every evening flies of these twice each week for two weeks to obtain flies for the Sexual selection on allometry 215 considers measurement error variance, not drift variance.