Researchers have long debated the relationship between

Transcription

1 Behavioral Ecology Vol. 10 No. 6: Sex ratios and sexual selection in socially monogamous zebra finches Nancy Tyler Burley and Jennifer Devlin Calkins Department of Ecology and Evolutionary Biology, University of California, Irvine, CA , USA An experiment was performed in which adult sex ratios of zebra finches, Taeniopygyia guttata castanotis, were varied to test possible effects of adult population sex ratios on sexual selection intensity and mating system dynamics in species with biparental care. The possibility that sex ratio influences the success of social mating patterns (leading to polygyny when males are rare and polyandry when females are rare) was not supported. Results did support the prediction of the differential allocation hypothesis that individuals of the abundant sex would increase their relative parental expenditure (PE). Although total (male female) PE did not vary between treatments, relative male PE was significantly higher in the male-biased treatment (MBT; sex ratio 64% male) than in the female-biased treatment (FBT; sex ratio 36% male). In both treatments, male PE contributions contributed to female reproductive rate. Results also supported the prediction of the differential access hypothesis that individuals of the abundant sex would experience greater intensity of selection on sexually selected attributes. Male beak color, a sexually selected trait, influenced male social parentage in the MBT but not in the FBT. Finally, broods in the FBT displayed higher hatching asynchrony and lower hatching success; we believe this was caused by early onset of incubation, a tactic used as a defense against intraspecific brood parasitism, which was much higher in the FBT. Population sex ratios may be an important factor affecting female ability to influence male parental investment patterns. Key words: differential access, differential allocation, parental investment, sex ratio, sexual selection, social monogamy, zebra finches. [Behav Ecol 10: (1999)] Researchers have long debated the relationship between adult sex ratio and population mating system. At the core of the debate is the question, do mating systems impact adult sex ratios (e.g., Koenig and Pitelka, 1981; Selander, 1965; Willson, 1984; Willson and Pianka, 1963; Wittenberger, 1976), or do adult population sex ratios cause mating patterns (e.g., Murray, 1984, 1985; Rowley, 1965; Wiley, 1974)? Frankly polygynous species ( Johnson and Burley, 1997), for example, often display female-biased sex ratios. A female-biased sex ratio may cause polygyny because surplus females become willing to accept already-mated males as mates (as in song sparrows; Smith et al., 1982). In contrast, a female-biased sex ratio may result from polygyny because polygynous males engage in mating investment that increases relative male mortality rates (e.g., Berger, 1983; Lindstrom and Kokko, 1998; Selander, 1965). These two hypotheses (mating system affects sex ratio; sex ratio affects mating system) are not necessarily mutually exclusive. Instead, there may be feedback between the two processes. Recent references to the idea that sex ratios impact mating systems typically refer to operational sex ratios (OSRs) (Emlen and Oring, 1977) rather than population sex ratios. These two concepts are not synonymous. OSR refers to the relative numbers of males and females in a population that are available for pairing/copulation at any given time. OSRs are affected by adult sex ratios (e.g., Clutton-Brock and Parker, 1992), but they also reflect parental investment patterns and tendencies toward extrapair matings. OSR is thus better viewed as a mating system component than as an independent determinant thereof (Burley and Parker, 1997). Accordingly, we focus on adult sex ratios here. To explore the role of sex ratios on mating system dynamics of species with biparental care, we investigated possible consequences of changes in the intensity of sexual selection re- Address correspondence to N. Burley. ntburley@uci.edu. Received 2 October 1998; accepted 30 March International Society for Behavioral Ecology sulting from sex ratio biases. Sex ratios may affect patterns of social mating, resulting in frankly polygynous unions when females are numerous or in polyandry when males outnumber females (e.g., Jouventin, 1982; Maynard Smith and Ridpath, 1972; Smith et al., 1982). Sex ratios may impact parental investment (PI) patterns (Breitwisch et al., 1986; Keenlyside, 1983). In species with substantial biparental care and mate choice by both sexes, individuals of the rarer sex have an advantage in mate choice by virtue of their short supply. They may tactically prefer as social mates (individuals that share offspring caregiving) those opposite-sex individuals that signal willingness to incur relatively high PI ( differential allocation ; Burley, 1986b, 1988). Strategic choosers benefit by reducing their PI, thereby increasing their residual reproductive value. The differential allocation hypothesis is based on the idea that individuals of both sexes can tactically respond to a situation, such as a locally skewed sex ratio, that impacts their mate-getting ability. This hypothesis thus predicts that the percent male contribution to offspring rearing (male PI/ total PI) should vary directly with the adult sex ratio (number of adult males)/(total number of adults). If males and females benefit equally from facultative adjustment of PI, one expects no net change in the total PI/brood, only changes in the percent male contribution. Alternatives include the possibility that only one sex facultatively adjusts PI; thus, females might have evolved set PI loads near their sustainable maximum, whereas males, with lower PI, might have greater capacity for tactical variation. If this were true, one would expect total PI/ brood to vary with sex ratio. Another alternative, of course, is the null hypothesis; lack of responsiveness to population sex ratio might occur in species with large and fluid populations in which sex ratio imbalances have seldom occurred, or where constraints limit opportunity to retaliate against a social mate that fails to provide high PI (see Discussion below). Finally, sex ratios may impact the relative intensity of sexual selection via a direct effect on differential mate access (Burley, 1977, 1983, 1986b). Differential access means that pre-

2 Burley and Calkins Sex ratio and sexual selection 627 ferred individuals have greater access to potential mates, while nonpreferred individuals must mate with any willing partner. When sex ratios are biased, effects of differential access are weakened for the rare sex and increased for the common sex. We tested the above hypotheses by establishing two captive populations of zebra finches (Taeniopygia guttata castanotis) with complementary sex ratios (64% versus 36% male; the typical tertiary sex ratio is about 52% male; Burley et al., 1989). Zebra finches are gregarious and nonterritorial socially monogamous estrildines. They feed in flocks and nest in loose colonies (Goodwin, 1982). Nonbreeders intermingle with breeders (Burley et al., 1989, unpublished data), such that information on local sex ratio appears available to birds. Additional factors that make zebra finches appropriate for this research include the following: (1) they display substantial biparental care of young; (2) they demonstrate a capacity for tactical variation in PI by both sexes (Burley, 1986b, 1988); and (3) they readily breed in captivity, tolerating a wide range of adult sex ratios. Domesticated birds resemble wild birds closely in conformation and behavior and display similar parental time budgets (Burley N, Solomon N, and Zann R, unpublished data). We expected that caregivers in the two treatments would experience different challenges to their genetic parentage: in the female-biased treatment (FBT), we expected intraspecific brood parasitism (IBP) to occur at an elevated rate; in the male-biased treatment (MBT), we expected extrapair fertilization (EPF) rates to be elevated. Accordingly, we collected data on possible tactics used to defend parentage. One possible defense against IBP is close nest attendance (e.g., Lank et al., 1989) and onset of incubation shortly after clutch initiation. Such behavior may physically prevent laying of parasitic eggs. We thus expected greater hatching asynchrony in the FBT than in the MBT. A competing hypothesis (Slagsvold and Lifjeld, 1989) is that females start incubation early to accelerate hatching and thus increase their mate s share of PI. While this hypothesis was formulated for species in which only females incubate, it could also apply to species, such as the zebra finch, in which a male s share of incubation is typically lower than his share of other parental activities. According to Slagsvold and Lifjeld s hypothesis, we should see a positive correlation between hatching asynchrony and male PI within treatments. A limitation of this study is that we lacked resources to measure EPF rates directly, although we could document a known correlate of such rates ( rapid renesting rate ; Burley et al., 1996). We were able to identify eggs and young produced through IBP using techniques previously developed (Fenske and Burley, 1995). Based on previous findings (Burley and Parker, 1997; Burley et al., 1996), we expect that our measures of female social parentage correspond closely to genetic parentage. For that reason, we focus several analyses on female fitness. METHODS Experiment initiation and husbandry protocol Birds used in these experiments were young adults from two outcrossed populations ( A and B ) of wild-type zebra finches maintained in the laboratory. In the FBT, females from population A and males from population B were used. In the MBT, males came from population A and females from population B. The FBT was initiated 2 months after the MBT. Before the MBT began, we created two pools of birds, one consisting of potential founders of the FBT and the other for the MBT. At the time experiments began, birds were selected so that the two experimental populations were similar with regard to several traits. Distributions of these traits were controlled for the following reasons: age may affect a bird s willingness to incur parental care, and beak color is used in mate choice (Burley and Coopersmith, 1987) and affects reproductive success of both sexes (Price and Burley, 1994). With the exception of male song rate (Balzer and Williams, 1998; Houtman, 1992; but see Burley and Coopersmith, in press), such effects have not been found for most other phenotypic variables thus far investigated (Burley N, unpublished data). Finally, tail stripes of birds in candidate pools were scored because many birds were missing some fraction of striped tail coverts at the time the two experimental pools were selected. There was no difference in tail stripe scores between males or females assigned to the two treatments (t test, p.20). In the 2-month interval between the start of the MBT and start of the FBT, FBT candidates were held in unisexual flights. During this interval, FBT males grew additional tail stripes. Thus, at the time treatments began (but not when birds were assigned to treatments), FBT males had higher tail stripe scores than MBT males (mean SE: FBT, ; MBT, ; t 3.27, 53 df, p.002). No other significant population differences were present at the time of experiment initiation for phenotypic traits measured. Before release into breeding aviaries, birds were randomly assigned color-band combinations for individual recognition. Each bird wore two bands per leg. Color band combinations used only colors previously shown to be of neutral attractiveness to zebra finches (Burley, 1985); moreover, the two distal bands were of identical color to minimize phenotypic asymmetry and its possible behavioral consequences (Swaddle and Cuthill, 1994). We established populations by simultaneously releasing adults into a breeding aviary. Initially, the treatments were to have reciprocal 2:1 sex ratios. Thus, the MBT was initiated using 40 males and 20 females as founders. Two weeks into the MBT, it became apparent that early nesting attempts of birds were being destroyed by conspecifics at an atypically high rate. Observations confirmed that this destruction was being performed by males lacking social mates; five such males were removed, after which pairs were better able to defend their nests. These five males are excluded from all analyses, as are two females that were accidentally released into the aviary about 4.5 months into the experiment. Thus, the sex ratio of this treatment was 63.6% male (35 males/55 total adults). Mean age ( SE) of male founders at the start of the MBT was days; that of females was days. FBT founders included 20 males and 36 females. The sex ratio of this treatment was thus 35.7% male (20 males/56 total adults). Mean age of male founders was days, and that of female founders was days. The experiment was conducted in large flights (approximately 50 m 3 ) illuminated by Vitalights and standard flourescent bulbs and under constant photoperiod (14 h light:10 h dark). Husbandry procedures closely adhered to those used previously (Burley, 1986b). Most resources (commercial finch seed mix, water, grit, cuttlebone, nestling food; straw and grass for nest building) needed for reproduction were available ad libitum. The remaining resources were added frequently (cotton batting for nest lining; fresh food). Numerous nesting sites were available; each flight contained 90 plastic nest cups set inside stainless-steel compartments. At the end of the experiment, breeding was terminated by removing nest cups after young fledged. The duration of the FBT was 162 days; the duration of the MBT was 239 days.

3 628 Behavioral Ecology Vol. 10 No. 6 Phenotype measurements Birds were hand-held while they were measured for beak and tail traits. We measured beak color using the Munsell Book of Color (Burley and Coopersmith, 1987). For analysis purposes each bird s measurements (one each for hue, value, and chroma) were transformed into a single composite numeral (Burley et al., 1992). Beak scores produced by this procedure correlate with fitness of birds in experimental colonies (Price and Burley, 1994) in ways consistent with predictions of matechoice experiments (Burley and Coopersmith, 1987, in press). Thus, these scores appear to be meaningful representations of beak color as perceived by zebra finches. We measured tail stripe scores as the percentage of the rump area covered with stripes in a fully feathered bird that was actually covered in the bird being measured. Scores ranged from 0 (a bird missing all its striped coverts) to 1 (full feather coverage). Reproductive variables We determined social parentage of a clutch by scan sampling adult attendants at active nests from a vantage point just outside the aviaries. Attendants at each nest were identified multiple times over the 5-week nesting cycle. We checked nests daily at midday and scored them for number of eggs and offspring. Eggs were scored as deriving from IBP if they were different in size or proportions from the rest of the clutch (Fenske and Burley, 1995). A deviant egg was often added on the same day as a native egg, which alerted us to inspect egg proportions. Fates of IBP eggs were monitored by frequent nest inspection at hatching time; hatchlings were marked with water-based colored pens. We banded nestlings with numbered seamless aluminum bands about 11 days after hatching. Juveniles were removed from aviaries at about 50 days of age. We standardized individual reproductive rates between treatments by dividing the number of young surviving to independence (2 weeks after fledging) produced by each social parent by the respective experimental intervals. Hatching asynchrony rate was calculated as the hatching span (in days) divided by the number of hatchlings for each nest. The mean hatching asynchrony of each social parent was calculated for all clutches with two or more hatchlings in which no IBP eggs were found. We also scored the hatching asychrony of each focal nest containing two or more hatchlings. Rapid renesting rate was the fraction of all clutch attempts of a social parent that were abandoned during or after egg laying and in which new clutches were initiated before the first possible hatch date of the abandoned clutch (12 days after the first egg date). Behavioral measurements Behavioral data were gathered by undergraduate researchers who were not informed of the test hypothesis. Procedures were similar to those previously reported (Burley, 1988). Observers were trained to conduct focal nest sampling (Altmann, 1974) of active nests. Focal sampling involved monitoring the same nesting attempt throughout the 5-week nesting cycle (except in four cases in which it was necessary, due to sampling constraints, to combine prehatch days of one clutch with the posthatch days of another for the same pair). Each day, 4 6 nests were monitored, depending on observer availability; whenever possible, both a morning and afternoon sample were collected for each focal nest (mean SE of sample number per pair: FBT, ; MBT, ). When one nest fledged, a new nest was selected for observation by locating a recently initiated nesting attempt (containing two or more eggs) of a pair not yet successfully observed. Each focal nest sample lasted 15 min, during which observers recorded the arrival and departure of social parents and all of the activities engaged in by both parents in the immediate vicinity (within 0.5 m) of the nest. Activities were scored on a laptop computer, which recorded durations of timed events. Most activities were scored as timed events. Exceptions to these were agonistic activities whose typical durations were too short to be accurately scored. These activities were scored as instantaneous events, but observers ranked them for relative intensity and noted unusually long durations. For purposes of analysis, parental activities were partitioned into four categories: active caregiving, or time spent nest building, nest inspecting, and feeding young; passive caregiving, time spent inside the nest in a position consistent with incubation/brooding; active defense, social interactions scored as instantaneous ; and passive defense, time spent perching in the vicinity of the nest ( surveying ) and time spent sitting high in the nest while looking out (see Table 1 for descriptions of behaviors). Both sexes of zebra finches typically participate in all four categories of activities. To minimize ambiguity in interpretation of results, data likely to reflect mating investment (expected to be higher in the MBT) or defense against intraspecific brood parasitism (IBP; expected to be higher in the FBT) were dropped before analysis. One ambiguous category of behavior is incubation. In these experiments, warming of eggs/young was not ascertained but was inferred from the posture of the social parent. Parents, however, attend nests even before egg laying and also sometimes later in the developmental period when young no longer require brooding. Nest attendance at these times may be associated with mating investment (Enstrom D and Burley N, unpublished data) and/or defense against IBP. Thus, time spent in passive care is presented here only for the week immediately preceding hatching of the first egg (week 2: day 7 to day 1), and the week that includes first hatch date (week 3: day 0 to day 6). During weeks 2 and 3, eggs/young require incubation/brooding for development, and competing functions appear minimal. Similarly, data on active defense are excluded for those week 1 days (day 14 to day 8) during which egg laying occurred, as defense activities occurring on those days may represent male guarding of fertile females (Birkhead and Møller, 1992) or defense against IBP (Fenske and Burley, 1995). The rationale for dividing parental expenditure (PE) into active versus passive categories is that the active behaviors, though typically of shorter duration, are expected to have higher costs (e.g., metabolic costs; risks of injury or death), and thus greater impact on fitness. Such costs, though measurable in principle, were not measured here. (Thus we refer to PE rather than PI.) Finally, the impact on fitness of some passive activities that birds engaged in commonly might be trivial. Birds resting in a nest cup, for example, scored here as incubating, may be expending no more energy than birds would necessarily spend perching somewhere else in the aviary (see also Burley, 1988). Analyses Nonparametric (Kruskal-Wallis, Mann-Whitney and Spearman rank correlation) tests were used to compare treatments for a number of phenotypic, reproductive, and behavioral attributes. Forward stepwise regression (p to enter and p to remove set at 0.15) was used to evaluate contributions of phenotypic and behavioral attributes to the reproductive rates of males and females in both treatments. Behavioral data (percent male share of four components of parental care) were arcsine transformed before inclusion in regression analyses. These tests were performed using Systat 7.0 routines (Wilkin-

4 Burley and Calkins Sex ratio and sexual selection 629 Table 1 Components of parental expenditure Table 1 Continued Category/ behavior Description (D)/ Amplification (A) Category/ behavior Description (D)/ Amplification (A) Active caregiving Nest building Nest inspect Feed D: A bird transports nesting material to the nest; adds new material to the nest; and/or arranges material already present in the nest. A: Nest material consists of straw/grass and cotton batting. Males are much more likely to transport and add grass; both sexes commonly transport and add cotton, arrange nest material. This behavior occurs during all 5 weeks of the sampling interval, albeit at varying frequencies. D: A bird perches just outside the nest entrance with upright posture and beak pointing toward entrance. (Also called look in ; Burley 1988.) A: Can occur throughout development, but common in the posthatch period. Often precedes feeding of young. D: A bird in the nest regurgitates to nestlings. A: Occurs only posthatching; values reported in Table 2, however, are averaged over all samples to preserve additivity of behaviors in this category. Feeding of very young hatchlings is difficult to detect and so is underrepresented in this sample. Time spent foraging for food to feed nestlings is also not included. Passive caregiving D: A stationary bird sits low in its nest in a position consistent with incubation/brooding. A: Actual incubation (egg warming) was not established, and nest attendance may serve multiple functions (see text). Data are included here for weeks 2 and 3 only to minimize impact of other possible functions. Active defense Passive defense Survey D: A nest attendant behaves aggressively toward (an)other bird(s) in the immediate vicinity of the nest. Behaviors scored included chase (Ch), to pursue an interloper briefly by hopping after the bird until it leaves the nest area; displace (Dis), to approach an interloper very closely, causing it to leave (but no pursuit); peck (P), to deliver a mild peck to the body of an interloper; threaten(th), any of several postures (Goodwin, 1982) which indicate imminent aggression if an interloper fails to depart; flychase (FCh), to pursue an interloper across the aviary in flight; and beakfence (BF), to repeatedly (and reciprocally) peck the beak of a combatant. A: Prior to analyses, data for all days on which the female nest attendant laid eggs were excluded. All data for aggression toward juveniles were excluded. (Parents are often aggressive to their own young in the fledging phase.) For each sample, each nest attendant was given a weighted aggression score in which each instance of Ch, Dis, P, and Th was given one point; each occurrence of FCh and BF was assigned 3 points if it was scored by the observer as intense or as lasting 50 s or longer (FCh only); otherwise an occurrence of FCh or BF received two points. An individual s aggression score was the sum of the points it was assigned as the result of aggressive acts it initiated during a sample. D: A nest attendant perches in a relaxed posture in the immediate vicinity of the nest entrance (typically 5 20 cm). Look out D: A stationary nest attendant sits high in the nest, looking out. A: This posture is typically inconsistent with incubation. If a bird stretches its neck far enough, it is sometimes possible to incubate(i) and look out (LO) at the same time. When behaviors were scored as I/LO, half the accumulated time was assigned to passive defense and the other half to passive care. son, 1996). Fisher s Exact test (Zar, 1984) was performed by consulting tables in Finney et al. (1963). The binomial exact test (Zar, 1984) was hand-calculated to measure the probability that a suite of results occurred by chance. RESULTS Brood characteristics and per-brood PE of focal nests Because parents might allocate PE based on brood characteristics (e.g., Gowaty and Droge, 1991; Patterson et al., 1980; Yasukawa et al., 1990), we examined brood characteristics of focal nests from the two treatments. No differences were found for the following traits (except where otherwise noted, sample sizes were 11 and 12 unique pairs for the FBT and MBT treatments, respectively, and values for each pair were averaged over all clutch attempts): laid clutch size (p.25), number of hatchlings (p.92), number of offspring surviving to independence (p.85), and brood sex ratio (males/ total progeny surviving to sexing age of about 45 days: FBT, n 10; MBT, n 10; p.59). There were also no population differences in these traits for all pairs in the experiment, with the exception of mean number of hatchlings per brood, which was lower in the FBT (see below). We also explored whether the total (male and female) average daily PE per nest differed between treatments. We found no significant differences between the FBT and the MBT for any of the four major categories of PE (Table 2). Within active care, the average time spent in nest inspection was marginally greater for FBT nests. Of all parental behaviors sampled, nest inspection makes the smallest contribution to the parental time budget. None of the other behav- Table 2 Per-sample parental expenditure (PE) (median) of caregiving and defensive parental behaviors Behavior FBT MBT U p Active care Feed Nest build Nest inspect Passive care Active defense Passive defense Survey Look out FBT, female-biased treatment (n 11); MBT, male-biased treatment (n 12). See Table 1 for descriptions of behaviors included in each category.

5 630 Behavioral Ecology Vol. 10 No. 6 Figure 1 Percent male contribution (median) to caregiving and nest defense in the female-biased (FBT; n 11) and malebiased (MBT; n 12) treatments. Active care: U 106, p.014; passive care: U 44.5, p.185; active defense: U 110, p.007; passive defense: U 101, p.031. iors partitioned within major categories displayed a significant difference (Table 2). Within-treatment correlation analyses were performed to determine whether relative participation across categories of PE was independent. One of 12 correlation analyses was significant. Specifically, in the FBT, there was a significantly positive correlation between participation in active and passive defense (0.74; p.02). In the reciprocal population, however, this correlation was negative ( 0.33; p.20). Overall, the direction of (nonsignificant) correlation was concordant across treatments for two of six comparisons. These results indicate that relative participation in the four categories of parental activities is not inherently linked; participation in active care, for example, does not consistently predict participation in any other category of behavior. Therefore, the four categories must be treated as independent components of PE. Differential allocation To explore whether the percentage of PE engaged in by the sexes varied as predicted by the differential allocation hypothesis, we compared the average percent male contribution (male PE/total PE) for the 5-week sample of each focal nest between treatments. Percent male contribution was significantly higher in the MBT for three of four categories of PE: active care, passive defense, and active defense (Figure 1). No difference was found in relative male contribution to passive care (incubation). The binomial probability of three or more of four comparisons being significant by chance is.0005 (Table 3); the probability that three or more of four comparisons, all with results in the same predicted direction, occurred by chance is (Table 3). In sum, results strongly support the differential allocation hypothesis: males made a greater contribution to active care and both categories of nest defense in the population (MBT) in which male access to social mates was restricted by sex ratio considerations. There were no overall differences in total PE to broods in the two treatments, however, suggesting that both sexes display similar tactical capacities to vary PE. To explore the hypothesis that females extract greater male PI by early onset of incubation, we ran correlation analyses between the four major PE categories (Table 2) and the hatching asynchrony of focal clutches for each treatment. Six of eight correlations were negative (males at nests with greater asynchrony had lower PE), but none was significant (Table 4). (The probability of six or more of eight correlations being negative by chance is.14.) Thus, hatching asynchrony does not appear to be a mechanism by which females increase male PI (see also below). Reproductive success and number of mates Reproductive rate was measured as the number of social offspring surviving to independence divided by the duration (in days) of the treatment. For females, reproductive rate was higher in the MBT (median: FBT, 0.006, n 35; MBT, 0.025, Table 3 Binomial expansion (p q) 4 for significance of differential allocation results Term A B Interpretation p 4 q Probability that all 4 tests would be significant by chance. 4p 3 q Probability that 3 of 4 tests would be significant by chance. 6p 2 q Probability that 2 of 4 tests would be significant by chance. 4p 1 q Probability that 1 of 4 tests would be significant by chance. p 0 q Probability that 0 of 4 tests would be significant by chance. A: Probability of obtaining statistically significant results for 0 4 tests of the hypothesis (p.05; q.95). B: Probability of obtaining statistically significant results, all in the expected direction, for 0 4 tests of the hypothesis (p.025; q.975). Observed results fall into the first to terms; text reports cumulative probabilities of observed results.

6 Burley and Calkins Sex ratio and sexual selection 631 Table 4 Hatching asynchrony and male parental expenditure at focal nests PE category r FBT (n 10) MBT (n 11) Active care Passive care Active defense Passive defense Spearman s test, all p.20. FBT, female-biased treatment; MBT, male-biased treatment. n 20; U 490.5, p.013). Females in the MBT also tended to have greater numbers of social mates per treatment interval (FBT, 0.006; MBT, 0.004; U 248, p.055). For males, reproductive rate also varied between treatments. Males in the FBT had higher social parentage (0.031, n 20) than those in the MBT (0.017, n 35; U 217.5, p.020). Males also obtained more social mates per unit time in the FBT (FBT, 0.006; MBT, 0.004; U 73, p.0001). We also examined the impact of the actual number of social mates (as opposed to their daily rate) on reproductive rate within experiments. In both treatments, the number of social mates influenced reproductive success of the overrepresented sex (Figure 2). This trend is attributable to the failure of individuals lacking social mates to reproduce successfully. For individuals of the underrepresented sex, however, having more than one social mate did not increase social parentage. In sum, both sexes reproduced at a higher rate when they were the underrepresented sex. This trend was caused largely by the opportunity of all individuals of the underrepresented sex to obtain social mates. Neither sex benefited from having multiple social mates. Defenses against kleptogamy We expected that caregivers in the two treatments would experience different challenges to their genetic parentage. Intraspecific brood parasitism (IBP) was expected to be higher in the FBT as a result of attempted reproduction by females lacking social mates. Extrapair fertilization (EPF) was expected to be higher in the MBT as the result of attempted reproduction by males lacking social mates. Relative incidence of IBP was measured as the fraction of parasitic eggs in the nests of each female. As expected, IBP egg rate was much higher in the FBT (median: FBT, 0.095, n 28; MBT, 0.000, n 20; U 126.5, p.0001). We compared the relative number of females in the two experiments whose nests contained IBP eggs and fledglings. At both stages, a significantly greater percentage of nests was parasitized in the FBT (Table 5). Data for hatchlings are not presented here because of the ambiguity caused by eggs that disappear at hatching time: it is not clear the extent to which such disappearances resulted from egg burial (especially of late-to-hatch eggs) or from brood reduction by hatchling eviction (e.g., Burley, 1986a). One defense against IBP eggs laid early in a clutch sequence is rapid renesting (Fenske and Burley, 1995). The rate of rapid renesting is the fraction of all nesting attempts of a female that were abandoned and replaced before the abandoned attempt could have hatched. Rapid renesting also occurs in response to EPF (Burley et al., 1996). The rate of rapid renesting did not differ significantly between treatments (FBT, 0.000, n 23; MBT, 0.167, n 20; U 268, p.34). Rapid renesting is seldom practiced in response to IBP that occurs at the end of the laying sequence (Fenske and Burley, 1995). We hypothesized that caregivers in the FBT, which were at risk for IBP, might commence incubation early to physically prevent parasitic egg laying. By this reasoning, greater hatching asynchrony of broods might be expected in the FBT. On the other hand, if nest attendance by fertile females in the MBT resulted in egg warming (see Methods), comparable levels of hatching asynchrony might be expected in the two experiments. Differences in hatching asynchrony between clutches containing IBP eggs and those without IBP eggs could occur simply because of greater laying asynchrony of clutches containing IBP eggs. To determine if birds at risk of IBP used tactics to limit its occurrence, we compared the hatching asynchrony scores for clutches not containing IBP eggs. Hatching asyn- Figure 2 Reproductive rate (median) as a function of number of social mates in the female-biased (FBT) and male-biased (MBT) treatments. A: Kruskal-Wallis H , 2 df, p.0001; B: H 3.003, 2 df, p.219; C: H 1.013, 3 df, p.798; D: H , 2 df, p.0001.

7 632 Behavioral Ecology Vol. 10 No. 6 Table 5 Relative incidence of intraspecific-brood parasitism (IBP) in the female-biased treatment (FBT) and the male-biased treatment (MBT) Egg stage a Fledgling stage b Parasitized Not parasitized Parasitized Not parasitized FBT MBT a Fisher s Exact two-tailed p.002. b Fisher s Exact two-tailed p.024 chrony of such clutches was significantly higher in the FBT (0.905, n 10) than in the MBT (0.167, n 19; U 44, p.019). Hatching asynchrony was negatively correlated with female reproductive rate in the FBT (r.598, n 10; p.05); these variables were not significantly correlated in the MBT (r.122, n 19; p.50). Finally, we explored the possible source of the different trends in the effects of hatching asynchrony on female reproductive success in the two treatments. Egg rate (number of native eggs laid divided by treatment interval) did not vary between females in the two treatments (FBT, 0.068, n 35; MBT, 0.094, n 20; U 437, p.13). Percent hatching sucess, however, was much higher for females in the MBT (0.65, n 20) than in the FBT (0.23, n 28; U 456.5, p.0001). As noted earlier (see Methods), hatching success actually reflects losses at both the egg stage (especially through burial) and the hatchling stage (through eviction of hatchlings from the nest). In the FBT, mean hatching asynchrony was negatively correlated with mean hatching success (r.730, n 10, p.017), whereas these variables were not correlated for females in the MBT (r.191, n 19, p.43). As a result, mean number of hatchlings per clutch differed between treatments (FBT, 1.00, n 28; MBT, 2.33, n 20; U 428, p.002). Females in the FBT compensated for low hatchling number by enhanced survival of nestlings. Nestling success rate (number of nestlings surviving to independence/number of hatchlings) was higher for females in the FBT (0.74, n 22) than in the MBT (0.50, n 19; U 106, p.007). Phenotype and social parentage Based on previous results (Burley and Coopersmith, 1987, in press; Price and Burley, 1994), we expected FBT females with less red beaks and MBT males with redder beaks to have greater mate-getting ability and thus higher reproductive success than their same-sex competitors. Results of a prior breeding experiment (Price and Burley, 1994) indicated that red beak color in males was a sexually selected trait unrelated to male viability, but that female beak color reflected viability as well as male mate preference. Thus, we expected results for females to be more likely to show an effect of beak color on reproduction regardless of whether they were the overrepresented sex; we expected an effect of beak color on reproduction for males only when they were overrepresented. When males were underrepresented, all males could obtain good mates, such that red-beaked males would lose their sexually selected advantage. We also explored the relationship between tail stripe score and reproductive rate. Beak color and tail stripe score were not correlated for either sex in either treatment (all p.5). In the FBT, for females, the best regression model (adjusted r 2.183, n 36, p.035) included beak score (r.002, p.042) and tail score (r.015, p.113). For males, no significant model was generated, but the best model (adjusted r 2.066, p.144) included tail score (r.049). Thus, female beak score varied as predicted, and male reproductive success was independent of beak score. For both sexes, the relationship between reproductive success and tail score was negative. In the MBT, for females, the best regression model (adjusted r 2.381, n 20, p.017) included beak score (r.003, p.066) and tail score (r.019, p.019). For males, the best regression model (adjusted r 2.232, n 35, p.006) included beak score (r.002, p.010) and tail score (r.012, p.025). Thus, for both sexes, beak scores varied in the direction predicted (females with less red beaks and males with redder beaks having higher reproductive success). Moreover, for both sexes, birds with greater numbers of tail stripes accrued higher reproductive success. In sum, female beak score was inversely proportional to reproductive rate in both experiments. Male beak score affected male reproductive rate only when males were overrepresented. Tail stripes positively predicted reproductive rate of both sexes in the MBT, but negatively predicted female reproduction in the FBT. Impact of male contribution on female fitness One additional stepwise regression analysis was performed for females of each population. In this model we added as independent variables the percent male contributions made for each category (passive and active care, passive and active defense) as well as major variables previously identified as affecting female reproductive success (beak score, tail score, hatching asynchrony). Reproductive rate remained the dependent variable. For females in the FBT, the best model generated by this approach (adjusted r 2.807, n 10, p.004) included percent male active defense (r.001, p.004), hatching asynchrony (r.045, p.008), and female tail score (r.027, p.029). For females in the MBT, the best resulting model (adjusted r 2.889, n 11, p.001) included two of the four paternal contribution rates (active care: r.001, p.026; passive care: r.001, p.043), as well as two variables found to be significant previously, hatching asynchrony (r.066, p.001) and tail score (r.034, p.000). In sum, in both treatments male PE contributed to female reproductive success. Tail scores and hatching asynchrony continued to show opposite effects on female reproductive rate in the two experiments. DISCUSSION Differential allocation Results of this experiment are consistent the main prediction of the differential allocation hypothesis. In breeding populations with adult sex-ratio biases, individuals of the overrepre-

8 Burley and Calkins Sex ratio and sexual selection 633 sented sex facultatively increased relative PE to obtain and/ or retain cooperative breeding partners. Significant differences in the predicted direction between treatments were found for three of four PE categories, including both categories (active care, active defense) in which PE is most likely to represent PI. Results of regression models indicate that female fitness (as measured by relative reproductive rate) in both populations improved as a result of greater male contribution of active care and/or defense. The FBT showed somewhat weaker trends than the MBT. We believe that this result occurred in part because of the shorter duration of the experiment. Differential allocation also occurred in a previous series of experiments in which mating attractiveness of both sexes was varied by non-neutral band-color manipulations, but in which population sex ratios were established and maintained at close to 50% male (Burley, 1988). The previous experiments were substantially longer (one ran for almost 2 years), and observed discrepancies in relative male contribution increased over the duration of the experiments, suggesting that mated individuals continually assess their ability to obtain/retain mates and adjust their PE. The fact that similar trends were observed in the current, shorter experiment indicates that birds have some ability to make appropriate assessments and adjustments of PE relatively early in their reproductive lives. Precisely how birds determine a potential mate s willingness to invest and use this as a criterion of mate choice (Trivers, 1972) has yet to be investigated. Research on this question would require a detailed analysis of consortship patterns during and after pair formation. These results also indicate that an individual of either sex may practice differential allocation in response to a variety of opportunities and constraints that he or she may encounter. Thus, individuals of both sexes attempt to manipulate each other s relative contribution of PI. This perspective is in contrast to the view, reflected in the hypotheses of Weatherhead and Robertson (1979) and Gowaty (1996), that individual males are selected to be tactically manipulative of female PI and that, while females may be selected to resist manipulation, they lack ability to extract PI from males. Specifically, Weatherhead and Robertson predicted tactical male manipulation, but not female resistance: in their view, females will opt to pair with more attractive males for the genetic benefits accruing (the sexy son hypothesis), even though such males tactically reduce PI. Gowaty s (1996) constrained female hypothesis also predicts male manipulation through tactical responses after assessment of female reproductive capacity. Thus, for example, a male may decline to provide PI if he assesses that his mate can reproduce successfully alone. Both of these hypotheses assume that females choose mates for heritable traits that affect offspring viability and/or offspring mating attractiveness, not on the basis of expected male parental care. Thus, male caregiving is determined by male assessment of its contribution to his fitness, not in response to female manipulation. The view that only males can tactically manipulate PI may be applicable to some cases in which female-only care produces offspring of quantity and quality close to that resulting from biparental care (e.g., Gowaty, 1996; Ketterson and Nolan, 1994). Even in such cases, however, females may be selected to influence male PI patterns if there is some fitness cost to females of assuming all parental care duties. A tactic possibly available to females of many species is the rewarding of parental males by conferring them relatively high paternity of subsequent broods (Freeman-Gallant, 1998). Population sex-ratio effects may also be significant here: where sex ratios are typically female biased, females should have less influence on male caregiving patterns; where sex ratios are even or male biased, male caregiving performance may influence a female s tendency to attempt more than one clutch with a given male ( Johnson and Burley, 1997). In gregarious species, females might even evaluate the tendency of neighboring males to contribute PI and use this information in subsequent mating decisions. Thus, female manipulation of male PI may take relatively subtle forms. In theory, incurring high PI should reduce residual reproductive value, most likely through decreased survivorship (Trivers, 1972). Differential allocation could lead to increased mortality of the overrepresented sex in natural populations, thereby generating a frequency-dependent process that would tend to equilibrate population sex ratios. In nature, however, other processes may cause tertiary sex ratios to deviate significantly from 50% male. In many birds, for example, mortality of females during natal dispersal is thought to generate malebiased tertiary sex ratios (Breitwisch, 1989). Thus, if femalebiased dispersal evolved early in avian lineages, differential allocation might have contributed to the evolution of substantial male care in birds (Burley N and Johnson K, manuscript in preparation). To the best of our knowledge, this is the only experimental study to investigate differential allocation in response to adult sex ratio in birds. When Keenlyside (1983) modestly varied the population sex ratio of captive cichlid fish, he found that males abandoned clutches and re-paired when faced with a surplus of females, whereas females did not abandon in a male-biased environment. His explanation for this sex difference was that the male intermating interval was inherently lower by virtue of the lower cost of male gamete production, thus suggesting that the OSR differs from the population sex ratio in this species (see Introduction). We did not find a tendency toward abandonment in zebra finches, which produce altricial young that require biparental care. Clearly, members of different taxa will experience different opportunities and constraints that shape reproductive tactics (Burley and Parker, 1997). Sexual selection and social monogamy Our ability to interpret results in light of sexual selection theory is constrained by the fact that we were unable to assign genetic parentage of offspring in these experiments. Several trends are nevertheless noteworthy. First, individuals of the underrepresented sex in both experiments were unable to capitalize on their scarcity by gaining an additional social mate (Figure 1). Had the experiments continued for a longer period of time, an increase in the occurrence of social bigamy might have been observed (see above and Burley, 1988). The failure of birds to benefit from bigamy in an environment of virtually unlimited resources clearly suggests that this species is not preadapted to evolve frank polygamy ( Johnson and Burley, 1997), even under the most permissive conditions. Variation in adult population sex ratio, then, in this species (and perhaps in estrildine finches generally) does not appear to be a viable route to major mating system evolution (as suggested by Murray, 1984; Breitwisch, 1989). Second, results reinforce earlier conclusions regarding the significance of beak color variation in zebra finches (Price and Burley, 1994): (1) male beak color is a sexually selected trait and (2) female beak color is under both sexual and natural selection. In the MBT, where opportunities for female mate choice were great, male beak color affected male reproductive success. In the FBT, however, where opportunities for female mate choice were limited, beak color did not affect male reproductive success. Female beak color influenced female reproductive success both when males had considerable opportunity for choice (FBT) and when male mate choice was minimal (MBT).

9 634 Behavioral Ecology Vol. 10 No. 6 Tail stripes are a sexually dimorphic trait in zebra finches (males have longer striped coverts with higher contrast; Burley N, unpublished data), whose possible social function has largely been unstudied. The stripes are often a target of intraspecific aggression, and loss of stripes is typically associated with high population densities. The rapid regrowth of tail stripes by FBT males in the interval between the initiation of the MBT and the FBT experiments probably resulted from the much reduced density in unisexual cages following removal of the MBT males from those cages. The MBT results, in which presence of tail stripes contributed to high reproductive success of both sexes, suggest that number of tail stripes might be used in mate choice and/or is an accurate indicator of competitive ability. The FBT results, by contrast, do not show this pattern. Instead, they suggest that presence of tail stripes is associated with low reproductive success. Further work is needed to explore the possible significance of these conflicting results. IBP and hatching asynchrony IBP rate was higher in the FBT, and in that treatment it was negatively associated with reproductive rate. Costs associated with IBP include direct (through acquisition of PI by IBP young) and indirect costs. Asynchrony patterns reflect a possible indirect cost. We hypothesize that increased asynchrony in the FBT was the result of onset of incubation shortly after egg laying began. This behavior reduced IBP but resulted in reduced brood size at hatching. Although in some species, asynchrony may reduce parental fitness by increasing the relative competitive advantage of older hatchlings (Clark and Wilson, 1981), this pattern was not observed here. Rather, brood size was reduced at such an early stage that older siblings were unlikely to have directly caused it. It may have been the case, however, that such a competitive advantage would have developed within a week or so, making it unprofitable for parents to care for late-hatched young. Thus, it is likely that parents buried slow-to-hatch eggs or evicted late-hatched young. A similar reduction in brood sizes of parasitized zebra finches was reported in a previous experimental investigation of IBP (Fenske and Burley, 1995). Future directions With the advent of modern molecular techniques, the study of animal mating systems is enjoying a much-needed renaissance, as the relationships between genetic and social components of mating systems become open to investigation (Parker and Burley, 1997). In turn, the importance of investigating sex ratio effects will increase because population sex ratios may have substantial effects on variation in the number of genetic mates that individuals of each sex obtain (Arnold and Duvall, 1994) and consequent mating tactics displayed in populations ( Johnson and Burley, 1997). The research reported here indicates that sex ratio influences multiple reproductive tactics. Further work should include investigation of effects on genetic parentage and mating success and the relationship between EPF rate and male PI under varying population sex ratios. We thank Tracey Kast, Marc Sine, and Marya Sosulski for assistance with the experiment, and Anders Brodin, Patty Gowaty, Kristine Johnson, Richard Symanski, and an anonymous reviewer for comments on earlier drafts of the manuscript. This research was supported by National Science Foundation grants BSR and IBN to N.T.B. REFERENCES Altmann J, Observational study of behaviour; sampling methods. Behaviour 49: Arnold SJ, Duvall D, Animal mating systems: a synthesis based on selection theory. Am Nat 143: Balzer AL, Williams TD, Do female zebra finches vary primary reproductive effort in relation to mate attractiveness? Behaviour 135: Berger J, Predation, sex ratios, and male competition in equids (Mammalia:Perissodactyla). J Zool 201: Birkhead TR, Møller AP, Sperm competition in birds: evolutionary causes and consequences. San Diego: Academic Press. Breitwisch R, Mortality patterns, sex ratios, and parental investment in monogamous birds. Curr Ornithol 6:1 50. Breitwisch R, Merritt PG, Whitesides GH, Parental investment by the northern mockingbird: male and female roles in feeding nestlings. Auk 103: Burley N, Parental investment, mate choice and mate quality. Proc Natl Acad Sci USA 74: Burley N, The meaning of assortative mating. Ethol Sociobiol 4: Burley N, The organization of behavior and the evolution of sexually selected traits. In: Avian monogamy (Gowaty PA, Mock DW, eds), AOU monograph 37. Washington, DC: American Ornithological Union; Burley N, 1986a. Sexual selection for aesthetic traits in species with biparental care. Am Nat 127: Burley N, 1986b. Sex-ratio manipulation in color-banded populations of zebra finches. Evolution 40: Burley N, The differential allocation hypothesis: an experimental test. Am Nat 132: Burley N, Coopersmith CB, Bill color preferences of zebra finches. Ethology 76: Burley NT, Coopersmith CB, in press. Redundant signals or multiple messages? Beak colour and song rates in zebra finches and the design of mate choice experiments. Anim Behav. Burley NT, Parker PG, Emerging themes and questions in the study of avian reproductive tactics. In: Avian reproductive tactics: female and male perspectives (Parker PG, Burley NT eds), AOU Monograph 49. Washington, DC: American Ornithological Union; Burley NT, Parker PG, Lundy K, Sexual selection and extrapair fertilization in a socially monogamous passerine, the zebra finch (Taeniopygia guttata). Behav Ecol 7: Burley NT, Price DK, Zann R, Bill color, reproduction and condition effects in wild and domesticated zebra finches. Auk 109: Burley N, Zann RA, Tidemann SC, Male EB, Sex ratios of zebra finches. Emu 89: Clark AB, Wilson DS, Avian breeding adaptations: hatching asynchrony, brood reduction and nest failure. Q Rev Biol 56: Clutton-Brock TH, Parker GA, Potential reproductive rates and the operation of sexual selection. Q Rev Biol 67: Emlen ST, Oring LW, Ecology, sexual selection and the evolution of mating systems. Science 197: Fenske B, Burley NT, Responses of zebra finches (Taeniopygia guttata) to experimental intraspecific brood parasitism. Auk 112: Finney DJ, Latscha R, Bennett BM, Hsu P, Tables for testing significance in a 2 2 contingency table. Cambridge: Cambridge University Press. Freeman-Gallant CR, Fitness consequences of male parental care in Savannah sparrows. Behav Ecol 9: Goodwin D, Estrildid finches of the world. Ithaca, New York: Comstock Press. Gowaty PA, Battle of the sexes and origins of monogamy. In: Partnerships in birds (Black J, ed). Oxford: Oxford University Press; Gowaty PA, Droge DL, Sex ratio conflict and the evolution of sex-biased provisioning in birds. Acta Congr Int Ornithol 20: Houtman AM, Female zebra finches choose extra-pair copulations with genetically attractive males. Proc R Soc Lond B 249:3 6.