Sex-biased initial eggs favours sons in the slightly size-dimorphic Scops owl (Otus scops)

Transcription

1 Biological Journal of the Linnean Society, 2002, 76, 1 7. With 3 figures Sex-biased initial eggs favours sons in the slightly size-dimorphic Scops owl (Otus scops) G. BLANCO 1 *, J. A. DÁVILA 1, J. A. LÓPEZ SEPTIEM 2, R. RODRÍGUEZ 3 and F. MARTÍNEZ 4 1 Instituto de Investigación en Recursos Cinegéticos (C.S.I.C-U.CL.M), Ronda de Toledo s/n, Ciudad Real, Spain 2 Sociedad para la Conservación de los Vertebrados, Apdo. 270, Majadahonda, Madrid, Spain 3 Department of Applied Biology, Estación Biológica de Doñana, C.S.I.C., Sevilla, Spain 4 Departamento de Ecología, Facultad de Biología, Universidad Complutense, 28040, Madrid, Spain Received 25 June 2001; accepted for publication 3 December 2001 We assessed the potential control of female Scops owls (Otus scops) over the production of daughters and sons in relation to their order in the laying sequence, and investigated the possible adaptive value of such control on the survival and growth of nestlings. The population sex ratio at fledging was not significantly biased. A relationship between ovulation order and egg sex was found, which was mostly due to the strong male biased sex ratio among initial eggs, as 17 of 18 first-hatched chicks (94%) from first-laid eggs were males. Asynchronous laying and subsequent asynchronous hatching resulted in a decreasing survival of chicks with increasing hatching order. Female chicks fledged with a higher mass than males and fledging mass decreased with hatching order due to male biased hatching order and its influence in the growth of the single first-hatched female. These results suggest that female Scops owls may be able to adjust the sex of their offspring at egg production to invest differentially in their survival and growth.. ADDITIONAL KEYWORDS: hatching order nestling sex. INTRODUCTION Evidence that birds may control the sex composition of the brood has increased since molecular techniques for sexing nestlings have started to be extensively used (Quinn, 1999). Mechanisms of facultative sexratio manipulation after hatching have been investigated in relation to brood reduction derived from hatching asynchrony (Slagsvold, Røskaf & Engen, 1986; Breitwisch, 1989; Slagsvold, 1990). There is now evidence to suggest that females may adjust both the sex composition and sex sequence of the clutch (Heinsohn, Legge & Barry, 1997; Arnold et al., 2001; Komdeur et al., 1997; Kilner, 1998; Ledge et al., 2001), although the mechanisms for sex-ratio adjustment at egg production or laying remain poorly understood (Krackow, 1999). *Corresponding author. gblanco@irec.uclm.es Females may adjust the sex ratio of their offspring according to diverse social, parental and environmental conditions (Bensch, 1999). When individuals of one sex cost more to produce, as in highly dimorphic species, sex ratios may differ from equality because of sex-biased mortality beyond or under parental control (Slagsvold et al., 1986; Bensch, 1999). Mortality may be skewed due to different susceptibility to starvation in times of food shortage or because of different competing ability for food of offspring of a sex compared with their different-sex siblings. Given that survival probability decreases with hatching order, parents should be able to control the sequence of sexes within a clutch to achieve an optimal offspring sex ratio that minimizes parental costs and maximizes offspring survival (Bortolotti, 1986). Facultative sex-ratio manipulation after hatching may lead to a fine-tuning adjustment according to current environmental conditions (Slagsvold et al., 1986; Wiebe & Bortolotti, 1992). 1

2 2 G. BLANCO ET AL. Optimal sex composition and sequence of the brood may be more difficult to decide at egg production than after hatching if based in unpredictable environmental conditions because of the long time span between egg production and offspring independence from parental care compared to the shorter time from hatching. Sex manipulation should be expected to occur at egg production, as the cost of such manipulation after hatching can be expensive or risky. Alternatively, some kind of co-ordination between sex manipulation both before and after hatching would lead to an optimal sex brood composition and sequence aimed to maximize fitness (Heinsohn et al., 1997). Owls (order Strigiformes) exhibit variable degrees of reversed sexual dimorphism in size and show high levels of hatching asynchrony and brood reduction, so they are candidates to adopt some kind of mechanism for sex allocation (Olsen & Cockburn, 1991). Data on the sex-ratio of owls are scarce, but two recent papers show different sex-allocation patterns (Appleby et al., 1997; Hörnfeldt et al., 2000). Appleby et al. (1997) have showed that tawny owls, Strix aluco, laid femalebiased clutches on territories with more abundant prey at the time of fledging. Therefore, tawny owls appear to predict future food abundance for chick fledging and then adjust sex composition of the brood. This mechanism was argued to be adaptive because the reproductive success of adult females was positively related to prey abundance in the territory on which they were reared as chicks (Appleby et al., 1997). The only study where sex sequence has been studied in an owl species found no relationship between hatching order and nestling sex, although only sex ratios of earlier (first to third) and later (fourth to eighth) chicks were presented (Hörnfeldt et al., 2000). Therefore, whether female owls co-ordinate sex ratio adjustment at oviposition and later, via brood reduction, to achieve an optimal sex composition and sequence of the brood remains unknown. In this paper we assessed the potential control of female Scops owls over the production of daughters and sons in relation to laying order, and investigated the possible adaptive value of such control on the survival and growth of nestlings. METHODS THE STUDY SPECIES The Scops owl is a small (90g) trans-saharan migrant that breeds mainly in tree cavities and also accepts nest boxes. Sexes look alike, but females are heavier and slightly larger than males (Cramp, 1985). In our study area, breeding females (N = 41) were 3.1 ± 1.4% larger than males (N = 8) for lengths of head, bill to forehead, tarsus, wing, third primary and tail. Females were 10.9% heavier than males while feeding chicks of about 15 days, which is when the female mass has stabilized since the peak near to laying. Two to five eggs are laid at two-day intervals, and incubation (by the female only) began with the first laid egg (own data). Chicks hatch asynchronously over 2 4 days depending of clutch size, and brood reduction usually occurs (Koening, 1973; this study). Chicks leave the nest after days and become independent at days (Koening, 1973; this study). DATA COLLECTION The study was carried out during the breeding seasons of in riparian forests and neighbour gardens and agricultural fields near Madrid, central Spain. In the study area, most Scops owls bred in nest boxes ( cm), provided by us. Nest boxes were monitored for egg laying and hatching order. Incubating females were captured in the nest, and banded, during daylight just after laying the first egg and sometimes just before. We did not record partial predation of clutches. Eggs were not marked to determine laying sequence because this would have implied frequent visits to the nest during laying, which usually causes nest desertion in this species (pers. obs.). The laying and hatching order of particular eggs within nests roughly coincided, as females laid eggs at two-day intervals and incubation begins with the first egg, which results in a strong hatching asynchrony (Koening, 1973). Given that we checked hatching of each chick with an accuracy of 0 1 days, we are confident that hatching order reflects laying order in the nests considered for the analyses. Nests where a discrepancy between dates of laying and hatching was associated with the loss of some egg after the first hatching were not considered in the analyses (N = 2). Nestlings were marked after hatching by painting their claws with unique combinations of colours using nail polish, and banded afterwards at days of age. Brood reduction and the identity of dead or disappeared nestlings were recorded in each visit to the nests. We did not record losses of complete broods or partial predation of broods. During the breeding seasons of , nestlings were weighed with Pesola balances to the nearest 1g and measured with callipers for tarsus length to the nearest 0.1mm every 3 5 days. Common growth models were fitted using the least square iterative method provided by the nonlinear curve-fitting program in the SPSS statistical package. Increase in mass and tarsus length fitted a logistic curve (mass: R 2 ± SD = 98.5 ± 1.4, tarsus: R 2 ± SD = 99.0 ± 1.4, N = 46) better than other common growth curves (Ricklefs, 1968). The asymptotes of growth curves of particular chicks were used as estimates of the con-

3 SEX-BIASED INITIAL EGGS IN THE SCOPS OWL 3 sidered measures at the time of fledging. The asymptote of tarsus length can be considered as a measure of the definitive skeletal size of chicks because tarsus growth stabilize when chicks have 20 days of age, while still in the nest (this study). The rate constant (k) of the logistic equation was used as a measure proportional to the time span required to grow from hatching size to asymptotic fledging size, while the maximum growth rate occurring at the inflection point was used as a measure of the faster instantaneous growth rate (Ricklefs, 1967). Nestling sex was determined by molecular procedures (Ellegren, 1996) using DNA extracted from blood. Only feathered nestlings were sampled for blood by brachial venipuncture at an age of 15 days to avoid any disturb potentially interfering with their normal development or survival. Dead nestlings disappeared rapidly from the nests, so they can not be sampled in order to sex them. Sex identification using this method was first tested using control individuals of known sex (N = 7). DATA ANALYSES Overall, we sexed 60 fledglings from 18 broods in , and measured the growth of 46 of these chicks from 14 broods in To examine whether sex in relation to laying/hatching order differed from parity, we used two-tailed binomial tests separately by laying order. To examine the influence of laying order in sex determination of hatchlings, multiple logistic regression analyses were performed controlling for the potential confounding effects of hatching date (days from 1 May) included as a continuous independent variable and year and number of hatchlings as categorical ones. Fledglings of one particular female were sexed in two different years, therefore, to avoid pseudoreplication, we also randomly excluded from the analysis the nestlings of this female in one year. R statistic was presented as an indicator of the relative contribution of each variable to the model. Fledgling success was analysed using logistic regression analysis where the success (1) or failure (0) of each nestling to fledge was the dependent variable. Hatching date (days from 1 May) of each chick was included as a continuous independent variable, while year and number of hatchlings were considered as categorical predictors. In addition, hatching order was included as an ordinal independent variable because of the predictable increase in mortality with hatching order. General linear models were performed to test for the effects of hatching order and sex on the asymptote, growth rate constant and maximum growth rate of mass and tarsus length. Year, number of hatchlings and hatching date were also included as independent variables to control for their potential influence in chick growth. We started with fully saturated models and sequentially removed non-significant interactions (with P > 0.05) to create a minimal adequate model. We used the chick as analysis unit because potential pseudoreplication on growth of siblings was avoided by including hatching order as a factor on which we wish to obtain conclusions. When the effect of hatching order was not significant, we used the nest as the analysis unit by considering the mean values for siblings of the same sex to avoid data non-independence. Fifth order of hatching was not considered in the analyses because only one chick hatched in this order succeeds to fledge. All analyses were performed with an SPSS statistical package. Overall, we obtained data on hatching order and fledging success of chicks from 22 nests. Given that four complete broods were not sexed and several sexed broods were not measured for growth, sample sizes varied slightly among analyses. All P-values refer to two-tailed tests. Mean ± SD are reported except when indicated otherwise. RESULTS SEX AND HATCHING ORDER Most first-hatched chicks were males (17 of 18 chicks, 94.4%). This trend occurs in all four years (1996: 80% of first hatched were males, N = 5; 1997: 100%, N = 4; 1998: 100%, N = 5; 1999: 100%, N = 4). The proportion of each sex among first-hatched chicks differed strongly from parity pooling years (binomial test, P < , N = 18). Sex of second-hatched chicks (38.9% males, N = 18), third-hatched (37.5% males, N = 16) and fourth-hatched chicks (42.9% males, N = 7) did not differ from parity (binomial test, P = 0.48, 0.45 and 1.00, respectively). All first-hatched chicks from clutches where all eggs produced a fledgling were males (binomial test, P = 0.008, N = 8), while the sex ratio of chicks hatched in subsequent orders did not differ from parity (binomial test, all P > 0.05). Overall, the proportion of sons and daughters did not significantly differ from a binomial distribution, both considering all broods (57% males, binomial test, P = 0.37, N = 60) and broods where all eggs produced a fledgling (53% males, binomial test, P = 0.86, N = 30). A logistic regression analysis revealed significant effects of hatching order in sex determination (c 2 1 = 6.90, P = 0.009, R = 0.24, Fig. 1) independently of year (c 2 3 = 1.51, P = 0.68, R = 0.00) hatching date (c 2 1 = 1.81, P = 0.11, = 0.09) and number of hatchlings (c 2 3 = 0.04, P = 0.99, R = 0.00). Similar results were obtained when nestlings of a female captured in two years were included only for a single year (hatching

4 4 G. BLANCO ET AL. Figure 1. Best-fit curve adjusted to a logistic regression of sex ratio in relation to hatching order of Scops owl fledglings. Numbers are sample sizes. Figure 2. Fledging success expressed as percentage of hatchlings that produced a fledgling, in relation to hatching order. Numbers above bars are sample sizes. order: c 2 1 = 5.69, P = 0.017, R = 0.20, all remainder variables were not significant at P > 0.15). FLEDGING SUCCESS A total of 71 chicks from 22 broods succeeded to fledge. Brood reduction occurred in 54.5% of the nests with hatchlings (N = 22). All first- and second-hatched chicks survived to fledge, while some mortality occurred among chicks hatched later (Fig. 2). Hatching order explained a major proportion of the variability in fledgling success of particular chicks (c 2 1 = 27.13, P < , R = 0.58) independently of the effect of year (c 2 3 = 9.69, P = 0.02, R = 0.17; the remainder variables were not significant at P > 0.19). WEIGHT GAIN AND TARSUS GROWTH IN RELATION TO SEX AND HATCHING ORDER Female chicks fledged with a higher mass than males (F 44,1 = 60.85, P < ; Fig. 3A), and fledging mass decreased with hatching order (F 44,3 = 6.66, P < 0.001) mostly due to the highest fledging mass of the single first-hatched female (Fig. 3A). Excluding this female, the analysis revealed significant effects of sex (F 43,1 = 26.73, P < ) but not of hatching order (F 43,3 = 0.66, P = 0.58) or the interaction between sex and hatching order (F 43,2 = 2.06, P = 0.14). The analysis pooling values for siblings of the same sex to avoid pseudoreplication revealed that females fledged a 10.3 ± 6.7% heavier than their male siblings did (t-test for related samples, t = 4.57, d.f. = 10, P = 0.001). The growth constant (k) for mass was higher for males than females (F 44,1 = 14.28, P < 0.001; Fig. 3B), and differed among chicks hatched in different orders (F 44,3 = 3.29, P = 0.03) mostly due to the short time required to grow from hatching to asymptotic fledgling size of the single first-hatched female (Fig. 3B). Excluding this chick, we found a significant effect of sex (F 43,1 = 4.79, P < 0.036) but no effect of hatching order (F 43,3 = 1.58, P = 0.21) or their interaction (F 43,2 = 2.83, P = 0.08). The maximum growth rate that occurs at the point of inflection did not differ between sexes (F 44,1 = 0.43, P = 0.52), but was significantly lower for chicks hatched in fourth order compared with chick hatched before (F 44,3 = 3.69, P = 0.02, all P < for posthoc comparisons between fourth and the remaining hatching orders according to Tukey test). We found no effects of hatching order (F 44,3 = 0.53, P = 0.67), or the interaction of hatching order sex (F 44,3 = 0.16, P = 0.93) in the asymptote of tarsus while controlling for the remaining independent variables. When excluded hatching order from the model and pooled values for siblings of the same sex to avoid pseudoreplication, we found that females fledged with a larger tarsus than males (F 25,1 = 7.14, P = 0.02). Within nests, females fledged with a tarsus 4.5 ± 1.8% larger than their male siblings (t-test for related samples, t = 2.52, d.f. = 10, P = 0.03). The time required to reach fledgling tarsus size was similar to males and females hatched in any order as indicated the analysis of k values (sex: F 45,1 = 1.80, P = 0.19, hatching order: F 45,3 = 2.55, P = 0.07, interaction: F 45,3 = 0.84, P = 0.48). Tarsus grows at a similar rate in both sexes (F 45,1 = 0.67, P = 0.41) but chicks hatched in fourth order show lower growth rates of tarsus than chick hatched before (F 45,3 = 3.74, P = 0.02, Tukey test: first

5 SEX-BIASED INITIAL EGGS IN THE SCOPS OWL 5 Figure 3. Effects of hatching order and sex (, males;, females) on (A) asymptotic fledging mass and (B) growth constant (k values) of Scops owl nestlings. to fourth order P = 0.049, second to fourth order P = 0.026, third to fourth order P = 0.10, all remaining comparisons P > 0.88). DISCUSSION In this study we report sex-biased laying order in the Scops owl. This bias was mostly due to the strong male biased sex ratio among initial eggs, as 17 of 18 first-hatched chicks from first-laid eggs were males. This agrees with the hypothesis that the control over sex allocation in species that lay multiple-egg clutches may be achieved by biasing the sex of the initial egg in the desired direction and then relying on the chance of meiosis (Emlen, 1997). This seems to be also the case in several species where the sex ratio of first-laid eggs is the only significantly biased or at least more biased than that of later-laid eggs (Bortolotti, 1986; Bednarz & Hayden, 1991; Leroux & Bretagnolle, 1996; Arnold, Griffith & Goldizen, 2001; but see Kilner, 1998). In this sense, the bias towards males in firstlaid Scops owl eggs is one of the most skewed so far reported in birds and, to our knowledge, the first evidence of sex-biased laying order in owls. Alternatively, the sex bias in the first egg, as a common observation, might not be the result of female sex manipulation, but rather an epiphenomenon, as female behavioural and hormonal changes before first and after first laying might cause differential sexes in the first vs. subsequent eggs (G. Bortolotti, pers. com.). In any case, these changes affect sex determination of the first egg differently, as initial eggs may more frequently produce males (Bednarz & Hayden, 1991; this study) or females (Bortolotti, 1986; Leroux & Bretagnolle, 1996) among different species with reversed sexual dimorphism. Asynchronous laying and subsequent asynchronous hatching resulted in a decreasing survival of eggs/ chicks with increasing laying and hatching order. Given that all first-laid eggs produced a fledgling, a male-biased sex ratio would be the result of some kind of mechanism of facultative sex ratio manipulation at initial egg production or laying rather than of differential mortality. Female preference for sons over daughters for her initial eggs may be adaptive in the Scops owl if this preference maximizes fitness returns per unit of parental investment. In such a case, variance in reproductive success should differ between the sexes (Charnov, 1982). No sufficient data are available to test this hypothesis, but evidences of female-female competition for nest sites (authors unpublished data) and male tendency to polygyny (Koening, 1973; authors unpublished data) suggest differential reproductive value of the sexes in the observed direction. Hatching sequence starting with a male may have the greatest chance of producing at least one male in good condition if females hatching in first order otherwise impose a competitive hierarchy with negative consequences for their siblings or, at least, for brothers. The decision to adjust sex ratios of broods prior to laying may prevent a more costly subsequent readjusting of brood sequence via infanticide to optimize size hierarchy in relation to sex (see also Heinsohn et al., 1997). The possibility that first-hatched females can impose a negative influence to their siblings survival or future viability can not be assessed accurately in this study because the extreme sex-biased hatching sequence led to the production of a single female vs.

6 6 G. BLANCO ET AL. 17 males hatched in first order across four years. However, this single female influenced the overall analysis of mass growth, suggesting that the hypothesis of competition between sexes in relation of hatching order is plausible (see also Bortolotti, 1986; Bednarz & Hayden, 1991). Alternatively, females may not be favouring sons, but by laying a male egg first in the laying sequence they may be maximizing both survival and numbers of all the chicks in the brood (Bortolotti, 1986). When individuals of one sex cost more to produce, as occurs in highly dimorphic species, costs for parents and offspring survival may not be independent of the sex composition and sex sequence of the brood (Bortolotti, 1986). The Scops owl is not highly sexually dimorphic in size, in contrast to other species where sex-biased laying sequence has been recorded (Howe, 1977; Bortolotti, 1986; Bednarz & Hayden, 1991; Olsen & Cockburn, 1991; Dzus, Bortolotti & Gerrard, 2000; but see Kilner, 1998). The establishment of a competitive hierarchy within the brood due to hatching asynchrony may, however, promote that the larger sex further increase its competitive superiority if hatched in first order (Bortolotti, 1986; Bednarz & Hayden, 1991). Hatching span might be sufficient for first-hatched females to gain advantage of their slightly larger size given the fast growth and short nestling period of Scops owls, probably constrained by their small size and cavity nesting (O Connor, 1984). At any case, outstanding growth of the exception, i.e. the single initial female, confirm the rule of a possible maternal preference towards initial sons to presumably minimize competition for food among different-sex siblings. This is further supported by the single first-hatched female being reared together with two sisters hatched later, which should be strong competitors, but no brother. In conclusion, our results agree with those of Bortolotti (1986) and Bednarz & Hayden (1991), suggesting that females may bias the sex of the first egg through an unknown mechanism to prevent a maladaptive disparity in the size of offspring during the nestling period. Alternatively, female physiological changes at first vs. subsequent laying may results in male-biased first eggs, promoting by chance an apparently advantageous brood sex sequence. ACKNOWLEDGEMENTS We thank J. A. Fargallo, J. de la Puente, J. A. Donázar, L. M. Carrascal, J. Viñuela and J. A. Godoy for their help in different stages of the study. G.B. and J.A.D. were supported by grants from the ME.C. of Spain. We thank G. Bortolotti, K. Lessells, T. Hipkiss and an anonymous referee for valuable comments to the manuscript and C. Lakunza for improving the English. REFERENCES Appleby RM, Petty SJ, Blakey JK, Rainey P, Macdonald DW Does variation of sex ratio enhance reproductive success of offspring in tawny owls (Strix aluco)? Proceedings of the Royal Society B 264: Arnold KE, Griffith SC, Goldizen AW Sex-biased hatching sequences in the cooperatively breeding Noisy Miner. Journal of Avian Biology 32: Bednarz JC, Hayden T Skewed brood sex ratio and sex-biased hatching sequence in Harris s hawks. American Naturalist 137: Bensch S Sex allocation in relation to parental quality. In: Adams NJ, Slotow RH, eds. Proceedings of the 22 International Ornithology Congress Durban. Johannesburg: BirdLife South Africa, Bortolotti GR Influence of sibling competition on nestling sex-ratios of sexually dimorphic birds. American Naturalist 127: Breitwisch R Mortality patterns, sex ratios, and parental investment in monogamous birds. In: Power DM, ed. Current ornithology, Vol. 6. New York: Plenum Press, Charnov EL The theory of sex allocation. Princeton: Princeton University Press. Cramp S The birds of the western Palearctic, Vol. IV. Oxford: Oxford University Press. Dzus EH, Bortolotti GR, Gerrard JM Does sexbiased hatching order in bald eagles vary with food resources? Écoscience 3: Ellegren H First gene on the avian W chromosome (CHD) provides a tag for universal sexing of non-ratite birds. Proceedings of the Royal Society B 263: Emlen ST When mothers prefer daughters over sons. Trends in Ecology and Evolution 12: Heinsohn R, Legge S, Barry S Extreme bias in sex allocation in Eclectus parrots. Proceedings of the Royal Society B 264: Hörnfeldt B, Hipkiss T, Fridolfsson A-K, Eklund U, Ellegren H Sex ratio and fledging success of supplementary-fed Tengmalm s owl broods. Molecular Ecology 9: Howe HF Sex-ratio adjustment in the common grackle. Science 198: Kilner R Primary and secondary sex ratio manipulation by zebra finches. Animal Behaviour 56: Koening L Das Aktionssystem der zwergohreule Otus Scops Scops (Linné 1758). Fortschritte der Verhaltensforschung 13: Komdeur J, Daan S, Tinbergen J, Mateman AC Extreme adaptive modification in sex ratio of the Seychelles warblers eggs. Nature 385: Krackow S Avian sex ratio distortions: The myth of maternal control. In: Adams NJ, Slotow RH, eds. Proceedings of the 22 International Ornithology Congress Durban. Johannesburg: BirdLife South Africa, Ledge S, Heinsohn R, Double MC, Griffiths R, Cockburn A Complex sex allocation in the laughing kookaburra. Behavioral Ecology 12:

7 SEX-BIASED INITIAL EGGS IN THE SCOPS OWL 7 Leroux A, Bretagnolle V Sex ratio variations in broods of Montagu s Harriers, Circus pygargus. Journal of Avian Biology 27: O Connor RJ The growth and development of birds. Chichester: Wiley & Sons. Olsen P, Cockburn A Female-biased sex allocation in peregrine falcons and other raptors. Behavioural Ecology and Sociobiology 28: Quinn TW Sex and the single chromosome. In: Adams NJ, Slotow RH, eds. Proceedings of the 22 International Ornithology Congress Durban. Johannesburg, BirdLife South Africa, Ricklefs RE Patterns of growth in birds. Ibis 4: Slagsvold T Fisher s sex ratio theory may explain hatching patterns in birds. Evolution 44: Slagsvold T, Røskaft E, Engen S Sex ratio, differential cost of rearing young, and differential mortality between the sexes during the period of parental care: Fisher s theory applied to birds. Ornis Scandinavica. 17: Wiebe K, Bortolotti GR Facultative sex ratio manipulation in American kestrels. Behavioural Ecology and Sociobiology 30: