- Email: [email protected]
Promiscuity, inbreeding and dispersal propensity in great tits
Animal Behaviour 84 (2012) 1363e1370
Contents lists available at SciVerse ScienceDirect
Animal Behaviour journal homepage: www.elsevier.com/locate/anbehav
Promiscuity, inbreeding and dispersal propensity in great tits Marta Szulkin*, Joanne R. Chapman 1, 2, Samantha C. Patrick 1, 3, Ben C. Sheldon Edward Grey Institute, Department of Zoology, University of Oxford, Oxford, U.K.
a r t i c l e i n f o Article history: Received 23 March 2012 Initial acceptance 24 May 2012 Final acceptance 20 August 2012 Available online 27 October 2012 MS. number: 12-00234 Keywords: dispersal extrapair paternity extrapair young great tit inbreeding depression mate choice outbreeding Parus major promiscuity
Mating with relatives leads to inbred offspring, which are likely to experience reduced fitness owing to the expression of deleterious recessive alleles. Constraints in social mate choice do not always allow individuals to avoid pairing with kin. A possible means of inbreeding avoidance is to engage in extrapair copulations with unrelated extrapair individuals. In a population of great tits, Parus major, we tested whether broods of related partners were characterized by higher rates of extrapair paternity than neighbouring outbred broods. Contrary to our expectations, broods sired by related partners had about 60% lower rates of extrapair paternity relative to outbred broods. Parental status of female birds categorized as inbreeding, outbreeding locally born or outbreeding immigrant explained 29% of the variance in the proportion of extrapair young in their broods. Outbreeding locally born females and inbreeding females did not differ in rates of extrapair paternity, but had significantly fewer extrapair young than outbreeding immigrant females. Thus, the contrasting differences in the proportion of extrapair young in inbred and outbred great tit broods do not appear to be the consequence of inbreeding per se. Instead, variation in both promiscuity and inbreeding may reflect the operation of broad-scale effects caused by variation in large-scale dispersal behaviour. Ó 2012 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Mating with a close relative can considerably decrease offspring fitness (Lynch & Walsh 1998; Szulkin et al. 2007), and reports of inbreeding depression in natural populations have been documented across a wide range of taxa (Keller & Waller 2002). If inbreeding depression is severe, selection for the evolution of inbreeding avoidance is expected. However, because of possible constraints on social mate choice, such as timing of breeding, availability of nesting sites and/or availability of unpaired individuals, it may not always be possible to avoid pairing with related individuals. One way to avoid the negative effects of pairing with a genetically related mate would be to seek extrapair copulations (EPC), resulting in an elevated proportion of extrapair young (EPY) in the nest, and a reduced rate of inbreeding at the brood level. Consistent with this prediction, several studies have reported increased rates of extrapair paternity (EPP) when mating with relatives. Blomqvist et al. (2002) showed that the prevalence of EPP
* Correspondence: M. Szulkin, Edward Grey Institute, Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, U.K. E-mail address: [email protected] (M. Szulkin). 1 These authors contributed equally to the work. 2 Present address: Plant Health and Environment Laboratory, Ministry of Agriculture and Forestry, Auckland, New Zealand. 3 Present address: Centre d’Etudes Biologiques de Chizé e CNRS, Beauvoir-surNiort, France.
increased with increasing genetic similarity between pair members in three species of shorebirds, providing support for the hypothesis that EPC functions as a means of inbreeding avoidance (but see Griffith & Montgomerie 2003 for a critique of some aspects of the study). Similarly, Eimes et al. (2005), Tarvin et al. (2005), FreemanGallant et al. (2006), Suter et al. (2007), Brouwer et al. (2011) and Varian-Ramos & Webster (2012) found increased occurrence of EPP with increasing genetic similarity of social mates in Mexican jays, Aphelocoma wollweberi, splendid fairy-wrens, Malurus splendens, Savannah sparrows, Passerculus sandwichensis, reed buntings, Emberiza schoeniclus, red-winged fairy-wrens, Malurus elegans, and red-backed fairy-wrens, Malurus melanocephalus, respectively. At the same time, several other studies (Kempenaers et al. 1996; Foerster et al. 2006; Stewart et al. 2006; Edly-Wright et al. 2007; Rubenstein 2007) failed to find such a directional relationship between EPP and social mate relatedness. Although much debate focuses on identifying factors driving the evolution of promiscuity (Arnqvist & Kirkpatrick 2005; Griffith 2007; Cornwallis et al. 2010), the original idea that female promiscuity may principally evolve as a genetic corollary of male promiscuity (Halliday & Arnold 1987; Arnold & Halliday 1988; Forstmeier et al. 2011) or other correlated traits has until recently received little attention (but see Forstmeier et al. 2011). It is not impossible that propensity for promiscuity, instead of being the direct target of selection, is in fact a corollary of dispersal
0003-3472/$38.00 Ó 2012 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.anbehav.2012.08.030
1364
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370
behaviour and/or exploratory behaviour (EB). Recent work on one great tit, Parus major, population from Wytham Woods (Oxfordshire, U.K.) has established specific links between propensity for dispersal and EB (Quinn et al. 2011), as well as EB and promiscuity (Patrick et al. 2012). Quinn et al. (2011) found a clear effect of natal origin (immigrant versus locally born) on EB, whereby EB was 11% higher among immigrants than locally born birds. In contrast, natal dispersal distance of birds born inside the study site did not influence EB, suggesting that an association between dispersal and EB existed at a between-population scale, but not within the main study site. In the same great tit population, Patrick et al. (2012) reported negative correlations between a male’s EB and the paternity lost in its social nest. Moreover, other bird studies have reported a positive association between female EB and promiscuity (Duckworth 2006; Van Oers et al. 2008). Propensity for large-scale dispersal is thus a trait that is not only influencing the genetic structure of populations, but also covarying with behavioural traits such as EB (Quinn et al. 2011), which in turn has been found to covary with promiscuity (Duckworth 2006; Van Oers et al. 2008; Patrick et al. 2012). At the same time, natal dispersal distance within the study site has been shown to reduce the likelihood of inbreeding (Szulkin & Sheldon 2008a), which all in all reflects nonrandom covariation between those behavioural traits. Given no clear understanding of the selective pressures driving promiscuity patterns in an inbreeding avoidance context, it is worth noting that dispersal is never introduced as a possible factor mediating the covariance between social-pair relatedness and promiscuity at the brood level. In the context of large fitness costs of inbreeding depression reported in the population (Szulkin & Sheldon 2007; Szulkin et al. 2007), we asked whether male and female great tits paired with relatives avoid inbreeding by increasing their proportion of EPY in the brood. We further tested whether the relationship between promiscuity and inbreeding can be explained by observed parental propensity to disperse at a between- or within-population scale; in this paper we discuss this in the light of evidence on promiscuity, dispersal and inbreeding accumulated in the population. METHODS Field Work The nestbox population of great tits from Wytham Woods (also referred to as ‘Wytham’), Oxfordshire, U.K. (1200 W 51460 N), breeds in 1021 nestboxes distributed across 385 ha of mostly deciduous forest at variable densities. It has been monitored continuously since 1947, and the number of great tit nestboxes, as well as the standard fieldwork protocol, has remained constant since 1964. More details on population characteristics can be found in Szulkin et al. (2007). All nestboxes were checked at regular intervals throughout the breeding period to establish hatch date and clutch size; parents were caught (for identification purposes) and ringed (if caught for the first time) 8e13 days after their offspring hatched, and nestlings were ringed on day 15 after hatching (hatching ¼ day 1). We define ‘locals’ as birds born in Wytham Woods and breeding in the same location. ‘Immigrants’ are birds born outside the study area, and recruited into the Wytham breeding population thereafter, which usually occurs in the autumn/winter preceding their first breeding season. While the study site is isolated from other woodland by surrounding agricultural and urban landscapes, some great tits are undoubtedly breeding around Wytham in gardens, hedgerows and small woodlots. However, it is estimated that they constitute only a small fraction (6%) of all immigrant birds coming into Wytham to breed
(Verhulst et al. 1997). Immigrant birds are easily identified, as all Wytham-born offspring are ringed before fledging (as described above); individuals caught as adults without a ring are therefore considered immigrants to the population. The majority of immigrants breeding in Wytham are expected to have dispersed >2 km (Verhulst et al. 1997), which is a noticeable distance given that the median natal dispersal distance (i.e. distance between place of birth and place of first reproduction) within Wytham is 528 m for males and 788 m for females, respectively (Szulkin & Sheldon 2008a). Movement between successive breeding sites (breeding dispersal) is limited as the majority of birds reoccupy their previous territory and range between a median distance of 50 and 143 m (Harvey et al. 1979). Birds were caught, ringed and released unharmed under British Trust for Ornithology ringing licences. Blood samples of young offspring and adults were collected (in the nest or at the time of catching, respectively) under U.K. Home Office Project Licence PPL no. 30/2409. Sampling Design A key aim of this study was to test for differences in the proportion of EPY among inbreeding and outbreeding great tit pairs, siring offspring with f 0.03125 and f ¼ 0, respectively. We further used the concept of ‘parental status’ to describe one of the following three situations: (1) inbreeding: mating with a relative and siring offspring with f 0.03125; (2) outbreeding local: born in Wytham Woods, and mating with an unrelated partner in Wytham Woods, siring offspring with f ¼ 0; (3) outbreeding immigrant: born outside of Wytham Woods, and mating with an unrelated partner in Wytham Woods, siring offspring with f ¼ 0. To identify inbred broods while still in the field, a social pedigree was built using all great tit breeding events occurring in Wytham and its vicinity from 1990 onwards. During the breeding seasons of 2005e2007 and 2009, we ran a pedigree analysis on alternate days throughout the season, using the identities of parents that had been trapped in the previous days to identify broods sired by relatives (siring offspring with an inbreeding coefficient f between 0.03125 and 0.25). Rates of EPP are known to be affected by breeding density and synchrony (Griffith et al. 2002; Westneat & Stewart 2003). Wytham Woods is a heterogeneous habitat in terms of nestbox and breeding density (Wilkin et al. 2006), food availability, soil composition and altitude (Savill et al. 2010). Breeding sites are therefore of variable quality, and hatching asynchrony has been shown to be altitude dependent in this population (Wilkin et al. 2006). In consequence, it is possible that differences in extrapair mating strategies across the site may arise. To reduce spatial and temporal variables that could confound estimates of rates of EPP in inbreeding and outbreeding birds from our population, we matched focal inbred broods with outbred broods located in close spatial and temporal proximity. Thus, for each identified focal inbred brood with an inbreeding coefficient ranging from f ¼ 0.03125 to f ¼ 0.25 (N ¼ 10), we also blood-sampled two to five neighbouring broods whose parents were found to be unrelated to each other (N ¼ 30). Restrictions for outbred broods to be included as neighbours in the study were (1) to be as closely located to a focal inbred brood as possible, and (2) to hatch within 7 days of a focal inbred brood hatching date. This resulted in an average distance of 193 m between focal and neighbour broods, which is ca. eight times lower than the average distance between nestboxes in Wytham (1495 m, Szulkin et al. 2009), and a mean difference in hatching dates between inbred and outbred nests of 1.2 days (SD ¼ 2.4; see also the Results). None of the breeding parents bred more than once in the data set.
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370
Broods were blood-sampled either on day 3 or shortly after parental identity was known (in general between days 7 and 14). Parents were blood-sampled at the nest when feeding young. A maximum of 30 ml of blood was collected from nestlings and their social parents by brachial or tibial venepuncture and stored in 800 ml of 99% ethanol. Although offspring age at the time of blood sampling varied from day 3 to 15, only broods that were sampled on day 3, or whose brood size after day 3 was not smaller than on day 3, were included in the study. Molecular Work DNA from blood samples of 395 individuals from 40 broods were Chelex-extracted and amplified as described in Patrick et al. (2012). Microsatellite loci for paternity exclusion were run from a pool of nine loci previously cloned from a range of passerine species (Patrick et al. 2012). All individuals were genotyped at between five and eight polymorphic microsatellite loci (7.7 loci were typed on average; 76% of all individuals were genotyped at eight microsatellite loci, 17%, 6%, 1% at seven, six, five loci, respectively) with a combined exclusion probability of >0.99. For more details on amplification procedures, genotyping, marker heterozygosity and HWE equilibria of microsatellites selected for paternity analyses, see supplementary material in Patrick et al. (2012). Since breeding males were not blood-sampled systematically across the study site, it was not possible to use computer-based paternity assignment methods to determine EPP. Instead, offspring genotypes were individually compared to genotypes of social parents. All chicks matched their mother’s genotypes 100%, except for three cases in which a single mismatch occurred between the maternal and offspring genotypes. We also allowed a single mismatch between paternal and offspring genotypes when assigning paternity. This single mismatch to a parental locus may be caused by mutation in the offspring, PCR errors such as slippage or allele-scoring errors. Offspring were classified as extrapair if at least two mismatches occurred between the offspring’s genotype and that of the social male, which is a standard approach for excluding paternity (see Brommer et al. 2010 and references therein); the average number of mismatches between offspring and social males for individuals assigned as EPY was 4.3 (median þ SD ¼ 4 þ 1.56, interquartile range 3e5, N ¼ 47). Statistical Analyses To overcome data set limitations, brood status was defined as either inbred (1) or outbred (0) owing to the limited number of replicates at any given level of inbreeding (see also Fig. 1a). This pooling is justified as we did not find any effect of f on the proportion of EPY in inbred broods when only the subset of broods with f > 0 was taken into account (F1,9 ¼ 0.018, P ¼ 0.90). Further analyses also included parental status, where males and females were categorized as inbreeding, outbreeding locals or outbreeding immigrants. To control for environmental heterogeneity across space and time, each set of focal inbred and neighbouring outbred nests was initially defined as a blocking treatment (N ¼ 10) and fitted as a random effect in our analyses. We used the free software R 2.12.1 (R Development Core Team 2010) with the add-on package lme4 (Bates et al. 2011). Whenever multiple fixed effects were fitted, we used AIC scores (Akaike information criterion) to select the best models (Akaike 1973). Estimated effects of variables of interest were derived from models with the lowest AIC values or from models with the smallest number of parameters if AIC scores differed by less than 2. Reduction in deviance was evaluated using
1365
likelihood ratio tests tested against chi-square and F distributions when binomial and quasibinomial error structure was used, respectively (Bolker et al. 2009). Models with normal error structure To test for the effect of parental relatedness on hatch date (HD, date of first hatched egg, standardized for year as HDstand ¼ (HDHDaverage per year)/SD, where SD was the standard deviation of HD per year), we used general linear mixed models (using the lmer function) where parental relatedness (defined as either ‘1’ when offspring f 0.03125 or ‘0’ when offspring f ¼ 0) was fitted as a fixed effect. To estimate the effect of parental relatedness on clutch size, the same model as above was used, but with standardized hatch date and year (fitted as a categorical variable) added as additional fixed effects. Models with binomial error structure Several analyses detailed below evaluating hatching success or the proportion of EPY in the brood required the use of models with binomial error structure, where data overdispersion needs to be taken into account. Because testing for overdispersion in binomial models with random effects is inherently difficult (Zuur et al. 2009), we performed model comparison where the same best model was run in (1) a mixed-model framework (using lmer in R) and (2) without random effects using the ‘quasibinomial’ family of data error structure which controls for overdispersion using the glm function in R (which does not support random effects; see also Appendix Table A1). The latter analyses yielded the most conservative outputs, and are consequently presented in the Results section. While all models gave qualitatively the same parameter estimates, SEs and P values differed somewhat, yet gave clear congruent trends for male and female parameter estimates. Models using both approaches are presented in Appendix Table A1. Model of hatching success. We tested whether choosing a relative as mate (while controlling for year, standardized hatch date, female and male age (defined as juvenile ‘0’, i.e. born in the previous breeding season, or adult ‘1’, i.e. born at least 1 year before the previous breeding season)) affects hatching success. We fitted clutch size and brood size as the binomial denominator and numerator of the model, respectively. Model of the effect of parental status on the proportion of EPY. When testing for differences in the proportion of EPY depending on parental status (inbreeding versus outbreeding, further expanded to inbreeding, outbreeding local, outbreeding immigrant), we contrasted rates of EPP using models with binomial proportions. We fitted brood size as the denominator and the number of EPY as the numerator of the model. It is likely that brood size reduction occurring during the transition from clutch size stage to brood size stage, as recorded on day 3 after hatching, affects brood reduction unequally depending on parental status, as inbreeding depression is expressed only among within-pair young of inbreeding partners. Therefore, effects of parental status on the proportion of EPY detected using brood size (and not clutch size) as the binomial denominator are highly conservative given the direction of the association between parental status and rates of EPY in the brood observed in this study (see Table 1); models run with clutch size as the binomial denominator generated even stronger estimated effects of parental status on the proportion of EPY in the brood (data not shown, but see Table 1). The following explanatory variables for initial models were used: inbreeding/outbreeding, year, standardized hatch date, male and female age. The analysis was further repeated for females and males separately, while taking into account parental status
1366
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370
(inbreeding, outbreeding local, outbreeding immigrant), year, standardized hatch date, female/male age, female/male body condition index (calculated as the ratio between body mass and wing length), pair bond duration (1 ¼ first year breeding together).
0.7
Effect of natal dispersal distance on male/female proportion of EPY in the brood. This was tested on a reduced sample of locally born inbreeding and outbreeding birds using the glm function with a quasibinomial model with clutch size as the denominator and number of EPY as the numerator. Female/male dispersal distance, age, status (inbreeding, outbreeding local), standardized HD and year were all fitted as fixed effects in the initial models.
0.4
0.6 0.5
0.3 0.2 0.1 0 f = 0.25
f = 0.125 f = 0.625 f = 0.3125 f = 0.0
0.7
RESULTS
0.6
Table 1 Percentage of extrapair young (EPY) in broods, using a reference level of brood size (BS) and clutch size (CS) EPY/BS
EPY/CS
N
17.3 8.7 23.6 19.6 13.3 6.9 59.8
16.0 8.2 21.7 18.4 12.1 5.6 65.3
30 12 18 19 11 10
(b)
0.5 EPY/BS
Broods sired by related and unrelated partners did not differ in hatch date (estimate þSE ¼ 0.25 þ 0.17, c21 ¼ 2:16, N ¼ 40, P ¼ 0.14) or clutch size (estimate þ SE ¼ 0.27 þ 0.50, c21 ¼ 0:28, N ¼ 40, P ¼ 0.60). However, parental relatedness caused a significant reduction in the number of hatched young on day 3 (F1,38 ¼ 8.78, N ¼ 40, P ¼ 0.005, parameter estimate for f (inbred)þ SE ¼ 1.23 þ 0.41). Inbreeding depression in our population can also have a substantial phenotypic effect in the form of critically reduced feather growth (see Fig. A1 for a photograph of 16-day-old offspring sired through a mothereson mating). We quantified the proportion of EPY in 10 inbred broods (average f ¼ 0.128, range f ¼ 0.03125 e f ¼ 0.25); each of them was matched by two to five outbred neighbouring broods, altogether totalling 40 nests. Of 40 broods, 23 (58%) contained EPY. Contrary to our original expectations given the apparent fitness costs of inbreeding, the proportion of EPY in inbred broods was about 60% lower than in outbred broods (estimateinbred þ SE ¼ 1.10 þ 0.59; F1,37 ¼ 4.28, N ¼ 40, P ¼ 0.045; Table 1, Fig. 1a, Appendix Table A1). In eight of 10 cases, both inbreeding parents were born in Wytham. We further tested whether breeder status (inbreeding, outbreeding local, outbreeding immigrant) influenced the proportion of EPY in their broods. Inbreeding females did not differ from outbreeding, locally born females in the proportion of EPY in their broods. In contrast, inbreeding and locally born females had significantly lower proportions of EPY in their broods relative to outbreeding females (Fig. 1b, Table 2). Inbreeding males had a marginally, nonsignificantly lower proportion of EPY in their social nests than local males, but there was no difference between inbreeding and immigrant males in the proportion of EPY in their social nests (Fig. 1c, Table 2). Overall, parental status (inbreeding, outbreeding local, outbreeding immigrant) explained 28.5% and 9.1% of the variance in EPY in the broods for females and males, respectively (Table 2). We further tested whether there was an association between increased promiscuity and natal dispersal distance among locally born birds (whose exact place of birth and place of first breeding
Outbred broods Local females Immigrant females Local males Immigrant males Inbred broods % Decrease in proportion of EPY in inbred broods (relative to outbred broods)
(a)
0.4 0.3 0.2 0.1 0 Inbreeding females
Local outbreeding females
Immigrant outbreeding females
Inbreeding males
Local outbreeding males
Immigrant outbreeding males
0.7 0.6
(c)
0.5 0.4 0.3 0.2 0.1 0
Figure 1. Proportion of extrapair young (EPY) relative to brood size (BS) among (a) inbred (f 0.03125, N ¼ 10) and outbred (f ¼ 0, N ¼ 30) broods, (b) inbreeding, outbreeding locally born and immigrant females (N ¼ 10, 12, 18, respectively) and (c) inbreeding, outbreeding locally born and immigrant males (N ¼ 10, 19, 11, respectively). Each dot represents one brood.
are known). Within-Wytham natal dispersal among outbreeding females and males did not influence the proportion of EPY in their broods (estimate for females þ SE ¼ 5.8 104 þ 5.88 104, deviance ¼ 1.51 N ¼ 21, P ¼ 0.28; estimate for males þ SE ¼ 6.4 104 þ 9.2 104, deviance ¼ 0.66, N ¼ 28, P ¼ 0.47). However, parental status (here categorized as inbreeding (1) versus local outbreeding (0)) did influence the proportion of EPY among males (estimate for males þSE ¼ 1.13 þ 0.60, deviance ¼ 5.30, N ¼ 28, P ¼ 0.041), but not females (estimate for females þ SE ¼ 0.47 þ 0.70, deviance ¼ 1.51, N ¼ 21, P ¼ 0.28; see also Fig. 1). DISCUSSION By contrasting inbreeding pairs with outbreeding neighbours breeding at similar times and in close spatial proximity, and despite
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370
1367
Table 2 Best models (GLMs with binomial proportions and quasibinomial error structure) explaining variance in the number of extrapair young (EPY) at the brood level, with EPY and brood size (BS) fitted as the binomial numerator and denominator, respectively
EPY/BS: females Female status Outbreeding local females Outbreeding immigrants Hatch date Female age Residual EPY/BS: males Male status Outbreeding local males Outbreeding immigrants Residual
Estimate
SE
t
P
0.08 1.66 0.49 1.55
0.63 0.55 0.21 0.46
0.13 3.00
0.899 0.005
1.18 0.72
0.68 0.75
1.75 0.97
df
Deviance
F
P
% Variance explained
2
21.74
9.56
<0.001
28.5
1 1 35
6.25 15.70 43.55
5.50 13.81
0.025 <0.001
8.2 20.6
2
6.99
1.96
0.155
9.1
37
69.38
0.088 0.338
Analyses were run separately for female and male breeders, whereby the effect of parental status (outbreeding local, outbreeding immigrants) on the proportion of EPY was tested relative to inbreeding individuals.
using a limited sample size of 40 broods, we found that inbreeding depression can have striking effects at the hatching stage (1.2 additional eggs on average remained unhatched among inbred broods relative to outbred neighbouring broods) and beyond (Appendix Fig. A1). These results complement findings of inbreeding depression across life history traits and environmental conditions (Szulkin & Sheldon 2007; Szulkin et al. 2007). Running pedigree software regularly throughout field seasons allowed us to identify inbred broods while offspring were still in the nest, which offered a unique opportunity for further morphological inspection. The photographs in Appendix Fig. A1 illustrate phenotypic differences in feather development that undoubtedly contribute to the fitness costs estimated quantitatively in Szulkin et al. (2007). EPY constituted 17.3% of all offspring sired across 30 outbred broods, which is consistent with rates of EPP found in a larger study of the same population (17.9%, N ¼ 128 broods, Brommer et al. 2010). Surprisingly, we found rates of EPP, measured as the proportion of EPY relative to brood size, to be about 60% lower among inbreeding pairs when compared to those outbreeding. There is thus no evidence that extrapair mating is used to reduce inbreeding depression. Female age also played an important role in explaining variation in rates of EPP (Table 2). Further study of female age-specific benefits of polyandry would be very valuable given that levels of EPP are often investigated by focusing on male traits only (but see Bouwman & Komdeur 2005). No other trait such as hatching date, parental body condition or pair bond duration was found to have any effect on rates of EPP in this data set, although such analyses would undoubtedly warrant a larger data set of sampled broods to gain greater resolution in estimating finer individual- and population-specific effects on variation in promiscuity. Further analyses revealed that locally born females had lower levels of EPY regardless of their genetic relatedness to their social partner. In contrast, female immigrants had a much higher proportion of EPY in their nests. Thus, 29% of the variance in the proportion of EPY can be explained solely by female status, that is inbreeding, outbreeding local or outbreeding immigrant. It is striking that such a large amount of variance in the proportion of EPY can be explained without invoking traditional male quality argumentation. Rather, variation in large-scale dispersal behaviour and traits associated with propensity for spatial movement is likely to explain the counterintuitive result of lower rates of EPP in the broods of inbreeding individuals. Importantly, in this and other great tit populations, large-scale or within-site dispersal behaviour was found to correlate positively with EB (Dingemanse et al. 2003; Cote & Clobert 2010; Quinn et al. 2011) and promiscuity (this study), while at the same time reducing the likelihood of inbreeding (Szulkin & Sheldon 2008a). Overall, dispersal behaviour is thus shown to overarch a large suite of interconnected traits.
This study thus adds to growing evidence suggesting that variation in promiscuity might be an epiphenomenon, or a genetic corollary of selection on other characters, such as behaviour in the other sex (Halliday & Arnold 1987; Arnold & Owens 1999; Forstmeier et al. 2011), EB (Van Oers et al. 2008; Patrick et al. 2012) or dispersal behaviour, rather than on benefits of promiscuity to the female per se. Given the increasing ease and costeffectiveness of collecting molecular data, allowing for relatively straightforward genetic characterization of the breeding population, it is perhaps surprising that there is such a dearth of accounts setting promiscuity in the context of other behavioural traits (Slagsvold & Lifjeld 1997). More studies contrasting rates of EPP against a large suite of different genetic, behavioural and life history traits, as well as against demographic parameters, would therefore be of interest. The fact that great tits in this population do not avoid inbreeding by mating with unrelated extrapair males is in agreement with a number of studies showing that parents with greater genetic similarity do not have increased rates of EPP (e.g. Kempenaers et al. 1996; Hansson et al. 2004; Foerster et al. 2006). In fact, our results even show an unusual opposite effect, as there were significantly fewer EPY in inbred broods relative to outbred broods. We suggest that these differences are likely to be mediated via large-scale dispersal behaviour rather than a directional preference for monogamy among inbreeding pairs. Counterintuitive results in studies of mate choice and genetic similarity (reporting lower rates of EPP among inbreeding broods, higher than expected genetic similarity to extrapair males, etc.) have sometimes been interpreted as evidence of kin selection (Kleven et al. 2005; Thunken et al. 2007; Wang & Lu 2011). We argue that kin selection through inbreeding is highly unlikely in species with separate sexes, even more so in noncooperatively breeding individuals where the probability of encountering a relative is low, thus further weakening any potential for selection of such behaviour. Instead, it is possible that biases suggesting preference for genetically similar mates may arise because of (1) a lack of spatial control for scenarios of random mating (Szulkin et al. 2009), (2) the number and variability of molecular markers assayed (Masters et al. 2003) or (3) genetic characteristics of breeding individuals and constraints imposed by parentage identification softwares (Slate 2009; Wetzel & Westneat 2009; Wang 2010; M. Szulkin, H. Nichols & W. Amos, unpublished data). An analysis of preference or avoidance of genetically similar mates, and possible confounding effects obscuring the interpretation of such findings, is further developed in Szulkin et al. (in press). This study does not preclude the possibility that, in some cases, particularly strong bonds between inbreeding relatives may develop during the nestling stage or the early stages of dispersal,
1368
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370
leading to altered patterns of promiscuity deviating from the population average (see e.g. Keller & Arcese 1998; Shutler et al. 2004). In our study population, we have previously shown that divorce rates in pairs siring offspring with f ¼ 0.25 are low relative to population-wide estimates (Szulkin & Sheldon 2008b). In a similar vein, siblings mate more often with each other than expected at random, even when spatial heterogeneity is controlled for (Szulkin et al. 2009). However, given that inbreeding is rare in this population (close inbreeding, generating offspring with f 0.125 was observed in 1.0e2.6% of matings depending on data set restrictiveness (Szulkin et al. 2007)), we would argue that much greater weighting should be given to (1) population density (Slagsvold & Lifjeld 1997; Westneat & Sherman 1997; Brommer et al. 2010), (2) propensity for large-scale dispersal and (3) other correlated characters such as EB (Patrick et al. 2012) as the driver for observed patterns of promiscuity, rather than selection for mating with relatives (Szulkin et al., in press). A lack of active, behavioural inbreeding avoidance is evident in this population, both when considered at the start of the breeding season (this study; Szulkin et al. 2009), or in the context of repeated matings across years (Szulkin & Sheldon 2008b). By and large, while within-site dispersal behaviour reduces the likelihood of inbreeding (Szulkin & Sheldon 2008a), large-scale dispersal behaviour may also be reflected in the rates of EPP among inbreeding individuals. Acknowledgments We thank the many generations of field workers who made this study possible, as well as the editor and two anonymous referees for constructive comments on the manuscript. This work benefited from financial support from the Christopher Welch Trust, The Queen’s College, the John Fell Fund and Magdalen College. S.C.P. was funded by a NERC Ph.D. studentship and J.R.C. by a Top Achiever Doctoral Scholarship from the Tertiary Education Commission of New Zealand. References Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle. In: Proceedings of the Second International Symposium of Information Theory (Ed. by B. N. Petrov & F. Cs’aki), pp. 267e281. Budapest: Akad’emiai Kiad’o. Arnold, K. E. & Owens, I. P. F. 1999. Cooperative breeding in birds: the role of ecology. Behavioral Ecology, 10, 465e471. Arnold, S. J. & Halliday, T. 1988. Multiple mating: natural selection is not evolution. Animal Behaviour, 36, 1547e1548. Arnqvist, G. & Kirkpatrick, M. 2005. The evolution of infidelity in socially monogamous passerines: the strength of direct and indirect selection on extrapair copulation behavior in females. American Naturalist, 165, S26eS37. Bates, D., Maechler, M. & Bolker, B. 2011. lme4: Linear Mixed-effects Models using S4 Classes. Vienna: R Foundation for Statistical Computing. Blomqvist, D., Andersson, M., Kupper, C., Cuthill, I. C., Kis, J., Lanctot, R. B., Sandercock, B. K., Szekely, T., Wallander, J. & Kempenaers, B. 2002. Genetic similarity between mates and extra-pair parentage in three species of shorebirds. Nature, 419, 613e615. Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Stevens, M. H. H. & White, J. S. S. 2009. Generalized linear mixed models: a practical guide for ecology and evolution. Trends in Ecology & Evolution, 24, 127e135. Bouwman, K. M. & Komdeur, J. 2005. Old female reed buntings (Emberiza schoeniclus) increase extra-pair paternity in their broods when mated to young males. Behaviour, 142, 1449e1463. Brommer, J. E., Alho, J. S., Biard, C., Chapman, J. R., Charmantier, A., Dreiss, A., Hartley, I. R., Hjernquist, M. B., Kempenaers, B., Komdeur, J., et al. 2010. Passerine extrapair mating dynamics: a Bayesian modeling approach comparing four species. American Naturalist, 176, 178e187. Brouwer, L., van de Pol, M., Atema, E. & Cockburn, A. 2011. Strategic promiscuity helps avoid inbreeding at multiple levels in a cooperative breeder where both sexes are philopatric. Molecular Ecology, 20, 4796e4807. Cornwallis, C. K., West, S. A., Davis, K. E. & Griffin, A. S. 2010. Promiscuity and the evolutionary transition to complex societies. Nature, 466, 969e972. Cote, J. & Clobert, J. 2010. Risky dispersal: avoiding kin competition despite uncertainty. Ecology, 91, 1485e1493.
Dingemanse, N. J., Both, C., van Noordwijk, A. J., Rutten, A. L. & Drent, P. J. 2003. Natal dispersal and personalities in great tits (Parus major). Proceedings of the Royal Society B, 270, 741e747. Duckworth, R. A. 2006. Behavioral correlations across breeding contexts provide a mechanism for a cost of aggression. Behavioral Ecology, 17, 1011e1019. Edly-Wright, C., Schwagmeyer, P. L., Parker, P. G. & Mock, D. W. 2007. Genetic similarity of mates, offspring health and extrapair fertilization in house sparrows. Animal Behaviour, 73, 367e378. Eimes, J. A., Parker, P. G., Brown, J. L. & Brown, E. R. 2005. Extrapair fertilization and genetic similarity of social mates in the Mexican jay. Behavioral Ecology, 16, 456e460. Foerster, K., Valcu, M., Johnsen, A. & Kempenaers, B. 2006. A spatial genetic structure and effects of relatedness on mate choice in a wild bird population. Molecular Ecology, 15, 4555e4567. Forstmeier, W., Martin, K., Bolund, E., Schielzeth, H. & Kempenaers, B. 2011. Female extrapair mating behavior can evolve via indirect selection on males. Proceedings of the National Academy of Sciences, U.S.A., 108, 10608e10613. Freeman-Gallant, C. R., Wheelwright, N. T., Meiklejohn, K. E. & Sollecito, S. V. 2006. Genetic similarity, extrapair paternity, and offspring quality in Savannah sparrows (Passerculus sandwichensis). Behavioral Ecology, 17, 952e958. Griffith, S. C. 2007. The evolution of infidelity in socially monogamous passerines: neglected components of direct and indirect selection. American Naturalist, 169, 274e281. Griffith, S. C. & Montgomerie, R. 2003. Why do birds engage in extra-pair copulation? Nature, 422, 833e833. Griffith, S. C., Owens, I. P. F. & Thuman, K. A. 2002. Extra pair paternity in birds: a review of interspecific variation and adaptive function. Molecular Ecology, 11, 2195e2212. Halliday, T. & Arnold, S. J. 1987. Multiple mating by females: a perspective from quantitative genetics. Animal Behaviour, 35, 939e941. Hansson, B., Hasselquist, D. & Bensch, S. 2004. Do female great reed warblers seek extra-pair fertilizations to avoid inbreeding? Proceedings of the Royal Society B, 271, S290eS292. Harvey, P. H., Greenwood, P. J. & Perrins, C. M. 1979. Breeding area fidelity of great tits (Parus major). Journal of Animal Ecology, 48, 305e313. Keller, L. F. & Waller, D. M. 2002. Inbreeding effects in wild populations. Trends in Ecology & Evolution, 17, 230e241. Keller, L. F. & Arcese, P. 1998. No evidence for inbreeding avoidance in a natural population of song sparrows (Melospiza melodia). American Naturalist, 152, 380e392. Kempenaers, B., Adriaensen, F., Van Noordwijk, A. J. & Dhondt, A. A. 1996. Genetic similarity, inbreeding and hatching failure in blue tits: are unhatched eggs infertile? Proceedings of the Royal Society B, 263, 179e185. Kleven, O., Jacobsen, F., Robertson, R. J. & Lifield, J. T. 2005. Extrapair mating between relatives in the barn swallow: a role for kin selection? Biology Letters, 1, 389e392. Lynch, M. & Walsh, B. 1998. Genetics and Analysis of Quantitative Traits. Sunderland, Massachusetts: Sinauer Associates. Masters, B. S., Hicks, B. G., Johnson, L. S. & Erb, L. A. 2003. Genotype and extra-pair paternity in the house wren: a rare-male effect? Proceedings of the Royal Society B, 270, 1393e1397. Patrick, S. C., Chapman, J. R., Dugdale, H. L., Quinn, J. L. & Sheldon, B. C. 2012. Promiscuity, paternity and personality in the great tit. Proceedings of the Royal Society B, 279, 1724e1730. Quinn, J. L., Cole, E. F., Patrick, S. C. & Sheldon, B. C. 2011. Scale and state dependence of the relationship between personality and dispersal in a great tit population. Journal of Animal Ecology, 80, 918e928. R Development Core Team 2010. R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing. Rubenstein, D. R. 2007. Female extrapair mate choice in a cooperative breeder: trading sex for help and increasing offspring heterozygosity. Proceedings of the Royal Society B, 274, 1895e1903. Savill, P., Perrins, C., Kirby, K. & Fisher, N. 2010. Wytham Woods: Oxford’s Ecological Laboratory. Oxford: Oxford University Press. Shutler, D., Hussell, D. J. T., Horn, A. G., Leonard, M. L., Shutler, R. W. & Lepage, D. 2004. Breeding between tree swallows from the same brood. Journal of Field Ornithology, 75, 353e358. Slagsvold, T. & Lifjeld, J. T. 1997. Incomplete female knowledge of male quality may explain variation in extra-pair paternity in birds. Behaviour, 134, 353e371. Slate, J. 2009. Bias in studies of heterozygosity and mate choice. Molecular Ecology, 18, 1815e1817. Stewart, I. R. K., Hanschu, R. D., Burke, T. & Westneat, D. F. 2006. Tests of ecological, phenotypic, and genetic correlates of extra-pair paternity in the house sparrow. Condor, 108, 399e413. Suter, S. M., Keiser, M., Feignoux, R. & Meyer, D. R. 2007. Reed bunting females increase fitness through extra-pair mating with genetically dissimilar males. Proceedings of the Royal Society B, 274, 2865e2871. Szulkin, M. & Sheldon, B. C. 2007. The environmental dependence of inbreeding depression in a wild bird population. PLoS One, 2, e1027. Szulkin, M. & Sheldon, B. C. 2008a. Dispersal as a means of inbreeding avoidance in a wild bird population. Proceedings of the Royal Society B, 275, 703e711. Szulkin, M. & Sheldon, B. C. 2008b. Correlates of the occurrence of inbreeding in a wild bird population. Behavioral Ecology, 19, 1200e1207. Szulkin, M., Garant, D., McCleery, R. H. & Sheldon, B. C. 2007. Inbreeding depression along a life-history continuum in the great tit. Journal of Evolutionary Biology, 20, 1531e1543.
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370 Szulkin, M., Zelazowski, P., Nicholson, G. & Sheldon, B. C. 2009. Inbreeding avoidance under different null models of random mating in the great tit. Journal of Animal Ecology, 78, 778e788. Szulkin, M., Stopher, K. V., Pemberton, J. M. & Reid, J. M. In press. Inbreeding avoidance, tolerance or preference in animals? Trends in Ecology & Evolution. Tarvin, K. A., Webster, M. S., Tuttle, E. M. & Pruett-Jones, S. 2005. Genetic similarity of social mates predicts the level of extrapair paternity in splendid fairywrens. Animal Behaviour, 70, 945e955. Thunken, T., Bakker, T. C. M., Baldauf, S. A. & Kullmann, H. 2007. Active inbreeding in a cichlid fish and its adaptive significance. Current Biology, 17, 225e229. Van Oers, K., Drent, P. J., Dingemanse, N. J. & Kempenaers, B. 2008. Personality is associated with extrapair paternity in great tits, Parus major. Animal Behaviour, 76, 555e563. Varian-Ramos, C. W. & Webster, M. S. 2012. Extrapair copulations reduce inbreeding for female red-backed fairy-wrens, Malurus melanocephalus. Animal Behaviour, 83, 857e864. Verhulst, S., Perrins, C. M. & Riddington, R. 1997. Natal dispersal of great tits in a patchy environment. Ecology, 78, 864e872.
1369
Wang, C. & Lu, X. 2011. Female ground tits prefer relatives as extra-pair partners: driven by kin-selection? Molecular Ecology, 20, 2851e2863. Wang, J. L. 2010. Do marker-based paternity assignments favour heterozygous and unrelated males? Molecular Ecology, 19, 1898e1913. Westneat, D. F. & Sherman, P. W. 1997. Density and extra-pair fertilizations in birds: a comparative analysis. Behavioral Ecology and Sociobiology, 41, 205e 215. Westneat, D. F. & Stewart, I. R. K. 2003. Extra-pair paternity in birds: causes, correlates, and conflict. Annual Review of Ecology Evolution and Systematics, 34, 365e396. Wetzel, D. P. & Westneat, D. F. 2009. Heterozygosity and extra-pair paternity: biased tests result from the use of shared markers. Molecular Ecology, 18, 2010e 2021. Wilkin, T. A., Garant, D., Gosler, A. G. & Sheldon, B. C. 2006. Density effects on lifehistory traits in a wild population of the great tit Parus major: analyses of longterm data with GIS techniques. Journal of Animal Ecology, 75, 604e615. Zuur, A. F., Leno, E. N., Walker, N., Saveliev, A. A. & Smith, G. M. 2009. Mixed Effects Models and Extentions in Ecology with R. New York: Springer.
Appendix Table A1 Robustness of parameter estimates when models with binomial error structure are run with and without random effects Response variable
BS/CS
No. of EPY/BS No. of EPY/BS
No. of EPY/BS
Variables from best model
Inbred Female age Male age Inbred Female age Female status: Outbreeding locally born Outbreeding immigrant Hatch date Female age Male status: Outbreeding locally born Outbreeding immigrant
Binomial errorsþ random effects
Quasibinomial errors (without random effects)
Effect size
SE
c2
df
P
Effect size
SE
1.32 0.75 1.13 1.09 1.13
0.41 0.41 0.46 0.57 0.45
9.00 3.16 6.43 4.11 6.25 17.94
1 1 1 1 1 2
0.003 0.075 0.011 0.042 0.012 <0.001
1.23 0.66 0.96 1.10 1.09
0.41 0.36 0.39 0.59 0.45
0.05 1.67** 0.49 1.54
0.60 0.52 0.21 0.44
0.08 1.66** 0.49 1.55
0.63 0.55 0.21 0.46
1.26 0.74
0.65 0.72
1.18 0.72
0.68 0.75
4.62 13.32 4.32
1 1 2
0.032 <0.001 0.115
F
df
P
8.78 2.66 5.13 4.28 6.60 9.56
1 1 1 1 1 2
0.005 0.111 0.030 0.045 0.014 <0.001
5.50 13.81 1.96
1 1 2
0.025 <0.001 0.155
N¼ 40. BS, CS and EPY stand for brood size, clutch size and number of extrapair young, respectively. There is a debate to what extent models with binomial proportions and random effects accurately estimate data overdispersion. We therefore contrasted all models using binomial data in this study and tested different approaches to control for overdispersion. The model run with random effects included random effects of blocking treatment (set of inbred and neighbouring outbred nests) and brood level observation (generating 40 levels for this random effect and expected to control for overdispersion in the data). The model presented in the main text used quasibinomial errors, generated only minimally different effect sizes and more conservative P values than the model run with random effects. Female and male status (outbreeding locally born, outbreeding immigrant) were tested relative to inbreeding females and males. Asterisks denote significant (P < 0.01) estimates for particular levels of female/male status when tested against inbreeding females/males using a t distribution.
1370
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370
Figure A1. Illustration of developmental abnormalities likely to contribute to inbreeding depression in the wild. Offspring born from a mothereson mating, photographed 16 days after hatching (May 2006). Two offspring (a, c) have delayed feather development (but normal body mass and size for their age); two other offspring (b, d), inbred at f ¼ 0.25, appear to be viable at this stage of their development. The photograph is reflective of the genetics underlying inbreeding depression, whereby two offspring most likely inherited the same set of recessive deleterious mutations from their mother and father, while the two other inbred within-pair young have not. The fifth offspring from this brood was sired by an extrapair male which had already fledged on day 16; its photograph is not available, but the phenotypic development on day 15 was normal for that individual.
Contents lists available at SciVerse ScienceDirect
Animal Behaviour journal homepage: www.elsevier.com/locate/anbehav
Promiscuity, inbreeding and dispersal propensity in great tits Marta Szulkin*, Joanne R. Chapman 1, 2, Samantha C. Patrick 1, 3, Ben C. Sheldon Edward Grey Institute, Department of Zoology, University of Oxford, Oxford, U.K.
a r t i c l e i n f o Article history: Received 23 March 2012 Initial acceptance 24 May 2012 Final acceptance 20 August 2012 Available online 27 October 2012 MS. number: 12-00234 Keywords: dispersal extrapair paternity extrapair young great tit inbreeding depression mate choice outbreeding Parus major promiscuity
Mating with relatives leads to inbred offspring, which are likely to experience reduced fitness owing to the expression of deleterious recessive alleles. Constraints in social mate choice do not always allow individuals to avoid pairing with kin. A possible means of inbreeding avoidance is to engage in extrapair copulations with unrelated extrapair individuals. In a population of great tits, Parus major, we tested whether broods of related partners were characterized by higher rates of extrapair paternity than neighbouring outbred broods. Contrary to our expectations, broods sired by related partners had about 60% lower rates of extrapair paternity relative to outbred broods. Parental status of female birds categorized as inbreeding, outbreeding locally born or outbreeding immigrant explained 29% of the variance in the proportion of extrapair young in their broods. Outbreeding locally born females and inbreeding females did not differ in rates of extrapair paternity, but had significantly fewer extrapair young than outbreeding immigrant females. Thus, the contrasting differences in the proportion of extrapair young in inbred and outbred great tit broods do not appear to be the consequence of inbreeding per se. Instead, variation in both promiscuity and inbreeding may reflect the operation of broad-scale effects caused by variation in large-scale dispersal behaviour. Ó 2012 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Mating with a close relative can considerably decrease offspring fitness (Lynch & Walsh 1998; Szulkin et al. 2007), and reports of inbreeding depression in natural populations have been documented across a wide range of taxa (Keller & Waller 2002). If inbreeding depression is severe, selection for the evolution of inbreeding avoidance is expected. However, because of possible constraints on social mate choice, such as timing of breeding, availability of nesting sites and/or availability of unpaired individuals, it may not always be possible to avoid pairing with related individuals. One way to avoid the negative effects of pairing with a genetically related mate would be to seek extrapair copulations (EPC), resulting in an elevated proportion of extrapair young (EPY) in the nest, and a reduced rate of inbreeding at the brood level. Consistent with this prediction, several studies have reported increased rates of extrapair paternity (EPP) when mating with relatives. Blomqvist et al. (2002) showed that the prevalence of EPP
* Correspondence: M. Szulkin, Edward Grey Institute, Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, U.K. E-mail address: [email protected] (M. Szulkin). 1 These authors contributed equally to the work. 2 Present address: Plant Health and Environment Laboratory, Ministry of Agriculture and Forestry, Auckland, New Zealand. 3 Present address: Centre d’Etudes Biologiques de Chizé e CNRS, Beauvoir-surNiort, France.
increased with increasing genetic similarity between pair members in three species of shorebirds, providing support for the hypothesis that EPC functions as a means of inbreeding avoidance (but see Griffith & Montgomerie 2003 for a critique of some aspects of the study). Similarly, Eimes et al. (2005), Tarvin et al. (2005), FreemanGallant et al. (2006), Suter et al. (2007), Brouwer et al. (2011) and Varian-Ramos & Webster (2012) found increased occurrence of EPP with increasing genetic similarity of social mates in Mexican jays, Aphelocoma wollweberi, splendid fairy-wrens, Malurus splendens, Savannah sparrows, Passerculus sandwichensis, reed buntings, Emberiza schoeniclus, red-winged fairy-wrens, Malurus elegans, and red-backed fairy-wrens, Malurus melanocephalus, respectively. At the same time, several other studies (Kempenaers et al. 1996; Foerster et al. 2006; Stewart et al. 2006; Edly-Wright et al. 2007; Rubenstein 2007) failed to find such a directional relationship between EPP and social mate relatedness. Although much debate focuses on identifying factors driving the evolution of promiscuity (Arnqvist & Kirkpatrick 2005; Griffith 2007; Cornwallis et al. 2010), the original idea that female promiscuity may principally evolve as a genetic corollary of male promiscuity (Halliday & Arnold 1987; Arnold & Halliday 1988; Forstmeier et al. 2011) or other correlated traits has until recently received little attention (but see Forstmeier et al. 2011). It is not impossible that propensity for promiscuity, instead of being the direct target of selection, is in fact a corollary of dispersal
0003-3472/$38.00 Ó 2012 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.anbehav.2012.08.030
1364
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370
behaviour and/or exploratory behaviour (EB). Recent work on one great tit, Parus major, population from Wytham Woods (Oxfordshire, U.K.) has established specific links between propensity for dispersal and EB (Quinn et al. 2011), as well as EB and promiscuity (Patrick et al. 2012). Quinn et al. (2011) found a clear effect of natal origin (immigrant versus locally born) on EB, whereby EB was 11% higher among immigrants than locally born birds. In contrast, natal dispersal distance of birds born inside the study site did not influence EB, suggesting that an association between dispersal and EB existed at a between-population scale, but not within the main study site. In the same great tit population, Patrick et al. (2012) reported negative correlations between a male’s EB and the paternity lost in its social nest. Moreover, other bird studies have reported a positive association between female EB and promiscuity (Duckworth 2006; Van Oers et al. 2008). Propensity for large-scale dispersal is thus a trait that is not only influencing the genetic structure of populations, but also covarying with behavioural traits such as EB (Quinn et al. 2011), which in turn has been found to covary with promiscuity (Duckworth 2006; Van Oers et al. 2008; Patrick et al. 2012). At the same time, natal dispersal distance within the study site has been shown to reduce the likelihood of inbreeding (Szulkin & Sheldon 2008a), which all in all reflects nonrandom covariation between those behavioural traits. Given no clear understanding of the selective pressures driving promiscuity patterns in an inbreeding avoidance context, it is worth noting that dispersal is never introduced as a possible factor mediating the covariance between social-pair relatedness and promiscuity at the brood level. In the context of large fitness costs of inbreeding depression reported in the population (Szulkin & Sheldon 2007; Szulkin et al. 2007), we asked whether male and female great tits paired with relatives avoid inbreeding by increasing their proportion of EPY in the brood. We further tested whether the relationship between promiscuity and inbreeding can be explained by observed parental propensity to disperse at a between- or within-population scale; in this paper we discuss this in the light of evidence on promiscuity, dispersal and inbreeding accumulated in the population. METHODS Field Work The nestbox population of great tits from Wytham Woods (also referred to as ‘Wytham’), Oxfordshire, U.K. (1200 W 51460 N), breeds in 1021 nestboxes distributed across 385 ha of mostly deciduous forest at variable densities. It has been monitored continuously since 1947, and the number of great tit nestboxes, as well as the standard fieldwork protocol, has remained constant since 1964. More details on population characteristics can be found in Szulkin et al. (2007). All nestboxes were checked at regular intervals throughout the breeding period to establish hatch date and clutch size; parents were caught (for identification purposes) and ringed (if caught for the first time) 8e13 days after their offspring hatched, and nestlings were ringed on day 15 after hatching (hatching ¼ day 1). We define ‘locals’ as birds born in Wytham Woods and breeding in the same location. ‘Immigrants’ are birds born outside the study area, and recruited into the Wytham breeding population thereafter, which usually occurs in the autumn/winter preceding their first breeding season. While the study site is isolated from other woodland by surrounding agricultural and urban landscapes, some great tits are undoubtedly breeding around Wytham in gardens, hedgerows and small woodlots. However, it is estimated that they constitute only a small fraction (6%) of all immigrant birds coming into Wytham to breed
(Verhulst et al. 1997). Immigrant birds are easily identified, as all Wytham-born offspring are ringed before fledging (as described above); individuals caught as adults without a ring are therefore considered immigrants to the population. The majority of immigrants breeding in Wytham are expected to have dispersed >2 km (Verhulst et al. 1997), which is a noticeable distance given that the median natal dispersal distance (i.e. distance between place of birth and place of first reproduction) within Wytham is 528 m for males and 788 m for females, respectively (Szulkin & Sheldon 2008a). Movement between successive breeding sites (breeding dispersal) is limited as the majority of birds reoccupy their previous territory and range between a median distance of 50 and 143 m (Harvey et al. 1979). Birds were caught, ringed and released unharmed under British Trust for Ornithology ringing licences. Blood samples of young offspring and adults were collected (in the nest or at the time of catching, respectively) under U.K. Home Office Project Licence PPL no. 30/2409. Sampling Design A key aim of this study was to test for differences in the proportion of EPY among inbreeding and outbreeding great tit pairs, siring offspring with f 0.03125 and f ¼ 0, respectively. We further used the concept of ‘parental status’ to describe one of the following three situations: (1) inbreeding: mating with a relative and siring offspring with f 0.03125; (2) outbreeding local: born in Wytham Woods, and mating with an unrelated partner in Wytham Woods, siring offspring with f ¼ 0; (3) outbreeding immigrant: born outside of Wytham Woods, and mating with an unrelated partner in Wytham Woods, siring offspring with f ¼ 0. To identify inbred broods while still in the field, a social pedigree was built using all great tit breeding events occurring in Wytham and its vicinity from 1990 onwards. During the breeding seasons of 2005e2007 and 2009, we ran a pedigree analysis on alternate days throughout the season, using the identities of parents that had been trapped in the previous days to identify broods sired by relatives (siring offspring with an inbreeding coefficient f between 0.03125 and 0.25). Rates of EPP are known to be affected by breeding density and synchrony (Griffith et al. 2002; Westneat & Stewart 2003). Wytham Woods is a heterogeneous habitat in terms of nestbox and breeding density (Wilkin et al. 2006), food availability, soil composition and altitude (Savill et al. 2010). Breeding sites are therefore of variable quality, and hatching asynchrony has been shown to be altitude dependent in this population (Wilkin et al. 2006). In consequence, it is possible that differences in extrapair mating strategies across the site may arise. To reduce spatial and temporal variables that could confound estimates of rates of EPP in inbreeding and outbreeding birds from our population, we matched focal inbred broods with outbred broods located in close spatial and temporal proximity. Thus, for each identified focal inbred brood with an inbreeding coefficient ranging from f ¼ 0.03125 to f ¼ 0.25 (N ¼ 10), we also blood-sampled two to five neighbouring broods whose parents were found to be unrelated to each other (N ¼ 30). Restrictions for outbred broods to be included as neighbours in the study were (1) to be as closely located to a focal inbred brood as possible, and (2) to hatch within 7 days of a focal inbred brood hatching date. This resulted in an average distance of 193 m between focal and neighbour broods, which is ca. eight times lower than the average distance between nestboxes in Wytham (1495 m, Szulkin et al. 2009), and a mean difference in hatching dates between inbred and outbred nests of 1.2 days (SD ¼ 2.4; see also the Results). None of the breeding parents bred more than once in the data set.
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370
Broods were blood-sampled either on day 3 or shortly after parental identity was known (in general between days 7 and 14). Parents were blood-sampled at the nest when feeding young. A maximum of 30 ml of blood was collected from nestlings and their social parents by brachial or tibial venepuncture and stored in 800 ml of 99% ethanol. Although offspring age at the time of blood sampling varied from day 3 to 15, only broods that were sampled on day 3, or whose brood size after day 3 was not smaller than on day 3, were included in the study. Molecular Work DNA from blood samples of 395 individuals from 40 broods were Chelex-extracted and amplified as described in Patrick et al. (2012). Microsatellite loci for paternity exclusion were run from a pool of nine loci previously cloned from a range of passerine species (Patrick et al. 2012). All individuals were genotyped at between five and eight polymorphic microsatellite loci (7.7 loci were typed on average; 76% of all individuals were genotyped at eight microsatellite loci, 17%, 6%, 1% at seven, six, five loci, respectively) with a combined exclusion probability of >0.99. For more details on amplification procedures, genotyping, marker heterozygosity and HWE equilibria of microsatellites selected for paternity analyses, see supplementary material in Patrick et al. (2012). Since breeding males were not blood-sampled systematically across the study site, it was not possible to use computer-based paternity assignment methods to determine EPP. Instead, offspring genotypes were individually compared to genotypes of social parents. All chicks matched their mother’s genotypes 100%, except for three cases in which a single mismatch occurred between the maternal and offspring genotypes. We also allowed a single mismatch between paternal and offspring genotypes when assigning paternity. This single mismatch to a parental locus may be caused by mutation in the offspring, PCR errors such as slippage or allele-scoring errors. Offspring were classified as extrapair if at least two mismatches occurred between the offspring’s genotype and that of the social male, which is a standard approach for excluding paternity (see Brommer et al. 2010 and references therein); the average number of mismatches between offspring and social males for individuals assigned as EPY was 4.3 (median þ SD ¼ 4 þ 1.56, interquartile range 3e5, N ¼ 47). Statistical Analyses To overcome data set limitations, brood status was defined as either inbred (1) or outbred (0) owing to the limited number of replicates at any given level of inbreeding (see also Fig. 1a). This pooling is justified as we did not find any effect of f on the proportion of EPY in inbred broods when only the subset of broods with f > 0 was taken into account (F1,9 ¼ 0.018, P ¼ 0.90). Further analyses also included parental status, where males and females were categorized as inbreeding, outbreeding locals or outbreeding immigrants. To control for environmental heterogeneity across space and time, each set of focal inbred and neighbouring outbred nests was initially defined as a blocking treatment (N ¼ 10) and fitted as a random effect in our analyses. We used the free software R 2.12.1 (R Development Core Team 2010) with the add-on package lme4 (Bates et al. 2011). Whenever multiple fixed effects were fitted, we used AIC scores (Akaike information criterion) to select the best models (Akaike 1973). Estimated effects of variables of interest were derived from models with the lowest AIC values or from models with the smallest number of parameters if AIC scores differed by less than 2. Reduction in deviance was evaluated using
1365
likelihood ratio tests tested against chi-square and F distributions when binomial and quasibinomial error structure was used, respectively (Bolker et al. 2009). Models with normal error structure To test for the effect of parental relatedness on hatch date (HD, date of first hatched egg, standardized for year as HDstand ¼ (HDHDaverage per year)/SD, where SD was the standard deviation of HD per year), we used general linear mixed models (using the lmer function) where parental relatedness (defined as either ‘1’ when offspring f 0.03125 or ‘0’ when offspring f ¼ 0) was fitted as a fixed effect. To estimate the effect of parental relatedness on clutch size, the same model as above was used, but with standardized hatch date and year (fitted as a categorical variable) added as additional fixed effects. Models with binomial error structure Several analyses detailed below evaluating hatching success or the proportion of EPY in the brood required the use of models with binomial error structure, where data overdispersion needs to be taken into account. Because testing for overdispersion in binomial models with random effects is inherently difficult (Zuur et al. 2009), we performed model comparison where the same best model was run in (1) a mixed-model framework (using lmer in R) and (2) without random effects using the ‘quasibinomial’ family of data error structure which controls for overdispersion using the glm function in R (which does not support random effects; see also Appendix Table A1). The latter analyses yielded the most conservative outputs, and are consequently presented in the Results section. While all models gave qualitatively the same parameter estimates, SEs and P values differed somewhat, yet gave clear congruent trends for male and female parameter estimates. Models using both approaches are presented in Appendix Table A1. Model of hatching success. We tested whether choosing a relative as mate (while controlling for year, standardized hatch date, female and male age (defined as juvenile ‘0’, i.e. born in the previous breeding season, or adult ‘1’, i.e. born at least 1 year before the previous breeding season)) affects hatching success. We fitted clutch size and brood size as the binomial denominator and numerator of the model, respectively. Model of the effect of parental status on the proportion of EPY. When testing for differences in the proportion of EPY depending on parental status (inbreeding versus outbreeding, further expanded to inbreeding, outbreeding local, outbreeding immigrant), we contrasted rates of EPP using models with binomial proportions. We fitted brood size as the denominator and the number of EPY as the numerator of the model. It is likely that brood size reduction occurring during the transition from clutch size stage to brood size stage, as recorded on day 3 after hatching, affects brood reduction unequally depending on parental status, as inbreeding depression is expressed only among within-pair young of inbreeding partners. Therefore, effects of parental status on the proportion of EPY detected using brood size (and not clutch size) as the binomial denominator are highly conservative given the direction of the association between parental status and rates of EPY in the brood observed in this study (see Table 1); models run with clutch size as the binomial denominator generated even stronger estimated effects of parental status on the proportion of EPY in the brood (data not shown, but see Table 1). The following explanatory variables for initial models were used: inbreeding/outbreeding, year, standardized hatch date, male and female age. The analysis was further repeated for females and males separately, while taking into account parental status
1366
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370
(inbreeding, outbreeding local, outbreeding immigrant), year, standardized hatch date, female/male age, female/male body condition index (calculated as the ratio between body mass and wing length), pair bond duration (1 ¼ first year breeding together).
0.7
Effect of natal dispersal distance on male/female proportion of EPY in the brood. This was tested on a reduced sample of locally born inbreeding and outbreeding birds using the glm function with a quasibinomial model with clutch size as the denominator and number of EPY as the numerator. Female/male dispersal distance, age, status (inbreeding, outbreeding local), standardized HD and year were all fitted as fixed effects in the initial models.
0.4
0.6 0.5
0.3 0.2 0.1 0 f = 0.25
f = 0.125 f = 0.625 f = 0.3125 f = 0.0
0.7
RESULTS
0.6
Table 1 Percentage of extrapair young (EPY) in broods, using a reference level of brood size (BS) and clutch size (CS) EPY/BS
EPY/CS
N
17.3 8.7 23.6 19.6 13.3 6.9 59.8
16.0 8.2 21.7 18.4 12.1 5.6 65.3
30 12 18 19 11 10
(b)
0.5 EPY/BS
Broods sired by related and unrelated partners did not differ in hatch date (estimate þSE ¼ 0.25 þ 0.17, c21 ¼ 2:16, N ¼ 40, P ¼ 0.14) or clutch size (estimate þ SE ¼ 0.27 þ 0.50, c21 ¼ 0:28, N ¼ 40, P ¼ 0.60). However, parental relatedness caused a significant reduction in the number of hatched young on day 3 (F1,38 ¼ 8.78, N ¼ 40, P ¼ 0.005, parameter estimate for f (inbred)þ SE ¼ 1.23 þ 0.41). Inbreeding depression in our population can also have a substantial phenotypic effect in the form of critically reduced feather growth (see Fig. A1 for a photograph of 16-day-old offspring sired through a mothereson mating). We quantified the proportion of EPY in 10 inbred broods (average f ¼ 0.128, range f ¼ 0.03125 e f ¼ 0.25); each of them was matched by two to five outbred neighbouring broods, altogether totalling 40 nests. Of 40 broods, 23 (58%) contained EPY. Contrary to our original expectations given the apparent fitness costs of inbreeding, the proportion of EPY in inbred broods was about 60% lower than in outbred broods (estimateinbred þ SE ¼ 1.10 þ 0.59; F1,37 ¼ 4.28, N ¼ 40, P ¼ 0.045; Table 1, Fig. 1a, Appendix Table A1). In eight of 10 cases, both inbreeding parents were born in Wytham. We further tested whether breeder status (inbreeding, outbreeding local, outbreeding immigrant) influenced the proportion of EPY in their broods. Inbreeding females did not differ from outbreeding, locally born females in the proportion of EPY in their broods. In contrast, inbreeding and locally born females had significantly lower proportions of EPY in their broods relative to outbreeding females (Fig. 1b, Table 2). Inbreeding males had a marginally, nonsignificantly lower proportion of EPY in their social nests than local males, but there was no difference between inbreeding and immigrant males in the proportion of EPY in their social nests (Fig. 1c, Table 2). Overall, parental status (inbreeding, outbreeding local, outbreeding immigrant) explained 28.5% and 9.1% of the variance in EPY in the broods for females and males, respectively (Table 2). We further tested whether there was an association between increased promiscuity and natal dispersal distance among locally born birds (whose exact place of birth and place of first breeding
Outbred broods Local females Immigrant females Local males Immigrant males Inbred broods % Decrease in proportion of EPY in inbred broods (relative to outbred broods)
(a)
0.4 0.3 0.2 0.1 0 Inbreeding females
Local outbreeding females
Immigrant outbreeding females
Inbreeding males
Local outbreeding males
Immigrant outbreeding males
0.7 0.6
(c)
0.5 0.4 0.3 0.2 0.1 0
Figure 1. Proportion of extrapair young (EPY) relative to brood size (BS) among (a) inbred (f 0.03125, N ¼ 10) and outbred (f ¼ 0, N ¼ 30) broods, (b) inbreeding, outbreeding locally born and immigrant females (N ¼ 10, 12, 18, respectively) and (c) inbreeding, outbreeding locally born and immigrant males (N ¼ 10, 19, 11, respectively). Each dot represents one brood.
are known). Within-Wytham natal dispersal among outbreeding females and males did not influence the proportion of EPY in their broods (estimate for females þ SE ¼ 5.8 104 þ 5.88 104, deviance ¼ 1.51 N ¼ 21, P ¼ 0.28; estimate for males þ SE ¼ 6.4 104 þ 9.2 104, deviance ¼ 0.66, N ¼ 28, P ¼ 0.47). However, parental status (here categorized as inbreeding (1) versus local outbreeding (0)) did influence the proportion of EPY among males (estimate for males þSE ¼ 1.13 þ 0.60, deviance ¼ 5.30, N ¼ 28, P ¼ 0.041), but not females (estimate for females þ SE ¼ 0.47 þ 0.70, deviance ¼ 1.51, N ¼ 21, P ¼ 0.28; see also Fig. 1). DISCUSSION By contrasting inbreeding pairs with outbreeding neighbours breeding at similar times and in close spatial proximity, and despite
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370
1367
Table 2 Best models (GLMs with binomial proportions and quasibinomial error structure) explaining variance in the number of extrapair young (EPY) at the brood level, with EPY and brood size (BS) fitted as the binomial numerator and denominator, respectively
EPY/BS: females Female status Outbreeding local females Outbreeding immigrants Hatch date Female age Residual EPY/BS: males Male status Outbreeding local males Outbreeding immigrants Residual
Estimate
SE
t
P
0.08 1.66 0.49 1.55
0.63 0.55 0.21 0.46
0.13 3.00
0.899 0.005
1.18 0.72
0.68 0.75
1.75 0.97
df
Deviance
F
P
% Variance explained
2
21.74
9.56
<0.001
28.5
1 1 35
6.25 15.70 43.55
5.50 13.81
0.025 <0.001
8.2 20.6
2
6.99
1.96
0.155
9.1
37
69.38
0.088 0.338
Analyses were run separately for female and male breeders, whereby the effect of parental status (outbreeding local, outbreeding immigrants) on the proportion of EPY was tested relative to inbreeding individuals.
using a limited sample size of 40 broods, we found that inbreeding depression can have striking effects at the hatching stage (1.2 additional eggs on average remained unhatched among inbred broods relative to outbred neighbouring broods) and beyond (Appendix Fig. A1). These results complement findings of inbreeding depression across life history traits and environmental conditions (Szulkin & Sheldon 2007; Szulkin et al. 2007). Running pedigree software regularly throughout field seasons allowed us to identify inbred broods while offspring were still in the nest, which offered a unique opportunity for further morphological inspection. The photographs in Appendix Fig. A1 illustrate phenotypic differences in feather development that undoubtedly contribute to the fitness costs estimated quantitatively in Szulkin et al. (2007). EPY constituted 17.3% of all offspring sired across 30 outbred broods, which is consistent with rates of EPP found in a larger study of the same population (17.9%, N ¼ 128 broods, Brommer et al. 2010). Surprisingly, we found rates of EPP, measured as the proportion of EPY relative to brood size, to be about 60% lower among inbreeding pairs when compared to those outbreeding. There is thus no evidence that extrapair mating is used to reduce inbreeding depression. Female age also played an important role in explaining variation in rates of EPP (Table 2). Further study of female age-specific benefits of polyandry would be very valuable given that levels of EPP are often investigated by focusing on male traits only (but see Bouwman & Komdeur 2005). No other trait such as hatching date, parental body condition or pair bond duration was found to have any effect on rates of EPP in this data set, although such analyses would undoubtedly warrant a larger data set of sampled broods to gain greater resolution in estimating finer individual- and population-specific effects on variation in promiscuity. Further analyses revealed that locally born females had lower levels of EPY regardless of their genetic relatedness to their social partner. In contrast, female immigrants had a much higher proportion of EPY in their nests. Thus, 29% of the variance in the proportion of EPY can be explained solely by female status, that is inbreeding, outbreeding local or outbreeding immigrant. It is striking that such a large amount of variance in the proportion of EPY can be explained without invoking traditional male quality argumentation. Rather, variation in large-scale dispersal behaviour and traits associated with propensity for spatial movement is likely to explain the counterintuitive result of lower rates of EPP in the broods of inbreeding individuals. Importantly, in this and other great tit populations, large-scale or within-site dispersal behaviour was found to correlate positively with EB (Dingemanse et al. 2003; Cote & Clobert 2010; Quinn et al. 2011) and promiscuity (this study), while at the same time reducing the likelihood of inbreeding (Szulkin & Sheldon 2008a). Overall, dispersal behaviour is thus shown to overarch a large suite of interconnected traits.
This study thus adds to growing evidence suggesting that variation in promiscuity might be an epiphenomenon, or a genetic corollary of selection on other characters, such as behaviour in the other sex (Halliday & Arnold 1987; Arnold & Owens 1999; Forstmeier et al. 2011), EB (Van Oers et al. 2008; Patrick et al. 2012) or dispersal behaviour, rather than on benefits of promiscuity to the female per se. Given the increasing ease and costeffectiveness of collecting molecular data, allowing for relatively straightforward genetic characterization of the breeding population, it is perhaps surprising that there is such a dearth of accounts setting promiscuity in the context of other behavioural traits (Slagsvold & Lifjeld 1997). More studies contrasting rates of EPP against a large suite of different genetic, behavioural and life history traits, as well as against demographic parameters, would therefore be of interest. The fact that great tits in this population do not avoid inbreeding by mating with unrelated extrapair males is in agreement with a number of studies showing that parents with greater genetic similarity do not have increased rates of EPP (e.g. Kempenaers et al. 1996; Hansson et al. 2004; Foerster et al. 2006). In fact, our results even show an unusual opposite effect, as there were significantly fewer EPY in inbred broods relative to outbred broods. We suggest that these differences are likely to be mediated via large-scale dispersal behaviour rather than a directional preference for monogamy among inbreeding pairs. Counterintuitive results in studies of mate choice and genetic similarity (reporting lower rates of EPP among inbreeding broods, higher than expected genetic similarity to extrapair males, etc.) have sometimes been interpreted as evidence of kin selection (Kleven et al. 2005; Thunken et al. 2007; Wang & Lu 2011). We argue that kin selection through inbreeding is highly unlikely in species with separate sexes, even more so in noncooperatively breeding individuals where the probability of encountering a relative is low, thus further weakening any potential for selection of such behaviour. Instead, it is possible that biases suggesting preference for genetically similar mates may arise because of (1) a lack of spatial control for scenarios of random mating (Szulkin et al. 2009), (2) the number and variability of molecular markers assayed (Masters et al. 2003) or (3) genetic characteristics of breeding individuals and constraints imposed by parentage identification softwares (Slate 2009; Wetzel & Westneat 2009; Wang 2010; M. Szulkin, H. Nichols & W. Amos, unpublished data). An analysis of preference or avoidance of genetically similar mates, and possible confounding effects obscuring the interpretation of such findings, is further developed in Szulkin et al. (in press). This study does not preclude the possibility that, in some cases, particularly strong bonds between inbreeding relatives may develop during the nestling stage or the early stages of dispersal,
1368
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370
leading to altered patterns of promiscuity deviating from the population average (see e.g. Keller & Arcese 1998; Shutler et al. 2004). In our study population, we have previously shown that divorce rates in pairs siring offspring with f ¼ 0.25 are low relative to population-wide estimates (Szulkin & Sheldon 2008b). In a similar vein, siblings mate more often with each other than expected at random, even when spatial heterogeneity is controlled for (Szulkin et al. 2009). However, given that inbreeding is rare in this population (close inbreeding, generating offspring with f 0.125 was observed in 1.0e2.6% of matings depending on data set restrictiveness (Szulkin et al. 2007)), we would argue that much greater weighting should be given to (1) population density (Slagsvold & Lifjeld 1997; Westneat & Sherman 1997; Brommer et al. 2010), (2) propensity for large-scale dispersal and (3) other correlated characters such as EB (Patrick et al. 2012) as the driver for observed patterns of promiscuity, rather than selection for mating with relatives (Szulkin et al., in press). A lack of active, behavioural inbreeding avoidance is evident in this population, both when considered at the start of the breeding season (this study; Szulkin et al. 2009), or in the context of repeated matings across years (Szulkin & Sheldon 2008b). By and large, while within-site dispersal behaviour reduces the likelihood of inbreeding (Szulkin & Sheldon 2008a), large-scale dispersal behaviour may also be reflected in the rates of EPP among inbreeding individuals. Acknowledgments We thank the many generations of field workers who made this study possible, as well as the editor and two anonymous referees for constructive comments on the manuscript. This work benefited from financial support from the Christopher Welch Trust, The Queen’s College, the John Fell Fund and Magdalen College. S.C.P. was funded by a NERC Ph.D. studentship and J.R.C. by a Top Achiever Doctoral Scholarship from the Tertiary Education Commission of New Zealand. References Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle. In: Proceedings of the Second International Symposium of Information Theory (Ed. by B. N. Petrov & F. Cs’aki), pp. 267e281. Budapest: Akad’emiai Kiad’o. Arnold, K. E. & Owens, I. P. F. 1999. Cooperative breeding in birds: the role of ecology. Behavioral Ecology, 10, 465e471. Arnold, S. J. & Halliday, T. 1988. Multiple mating: natural selection is not evolution. Animal Behaviour, 36, 1547e1548. Arnqvist, G. & Kirkpatrick, M. 2005. The evolution of infidelity in socially monogamous passerines: the strength of direct and indirect selection on extrapair copulation behavior in females. American Naturalist, 165, S26eS37. Bates, D., Maechler, M. & Bolker, B. 2011. lme4: Linear Mixed-effects Models using S4 Classes. Vienna: R Foundation for Statistical Computing. Blomqvist, D., Andersson, M., Kupper, C., Cuthill, I. C., Kis, J., Lanctot, R. B., Sandercock, B. K., Szekely, T., Wallander, J. & Kempenaers, B. 2002. Genetic similarity between mates and extra-pair parentage in three species of shorebirds. Nature, 419, 613e615. Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Stevens, M. H. H. & White, J. S. S. 2009. Generalized linear mixed models: a practical guide for ecology and evolution. Trends in Ecology & Evolution, 24, 127e135. Bouwman, K. M. & Komdeur, J. 2005. Old female reed buntings (Emberiza schoeniclus) increase extra-pair paternity in their broods when mated to young males. Behaviour, 142, 1449e1463. Brommer, J. E., Alho, J. S., Biard, C., Chapman, J. R., Charmantier, A., Dreiss, A., Hartley, I. R., Hjernquist, M. B., Kempenaers, B., Komdeur, J., et al. 2010. Passerine extrapair mating dynamics: a Bayesian modeling approach comparing four species. American Naturalist, 176, 178e187. Brouwer, L., van de Pol, M., Atema, E. & Cockburn, A. 2011. Strategic promiscuity helps avoid inbreeding at multiple levels in a cooperative breeder where both sexes are philopatric. Molecular Ecology, 20, 4796e4807. Cornwallis, C. K., West, S. A., Davis, K. E. & Griffin, A. S. 2010. Promiscuity and the evolutionary transition to complex societies. Nature, 466, 969e972. Cote, J. & Clobert, J. 2010. Risky dispersal: avoiding kin competition despite uncertainty. Ecology, 91, 1485e1493.
Dingemanse, N. J., Both, C., van Noordwijk, A. J., Rutten, A. L. & Drent, P. J. 2003. Natal dispersal and personalities in great tits (Parus major). Proceedings of the Royal Society B, 270, 741e747. Duckworth, R. A. 2006. Behavioral correlations across breeding contexts provide a mechanism for a cost of aggression. Behavioral Ecology, 17, 1011e1019. Edly-Wright, C., Schwagmeyer, P. L., Parker, P. G. & Mock, D. W. 2007. Genetic similarity of mates, offspring health and extrapair fertilization in house sparrows. Animal Behaviour, 73, 367e378. Eimes, J. A., Parker, P. G., Brown, J. L. & Brown, E. R. 2005. Extrapair fertilization and genetic similarity of social mates in the Mexican jay. Behavioral Ecology, 16, 456e460. Foerster, K., Valcu, M., Johnsen, A. & Kempenaers, B. 2006. A spatial genetic structure and effects of relatedness on mate choice in a wild bird population. Molecular Ecology, 15, 4555e4567. Forstmeier, W., Martin, K., Bolund, E., Schielzeth, H. & Kempenaers, B. 2011. Female extrapair mating behavior can evolve via indirect selection on males. Proceedings of the National Academy of Sciences, U.S.A., 108, 10608e10613. Freeman-Gallant, C. R., Wheelwright, N. T., Meiklejohn, K. E. & Sollecito, S. V. 2006. Genetic similarity, extrapair paternity, and offspring quality in Savannah sparrows (Passerculus sandwichensis). Behavioral Ecology, 17, 952e958. Griffith, S. C. 2007. The evolution of infidelity in socially monogamous passerines: neglected components of direct and indirect selection. American Naturalist, 169, 274e281. Griffith, S. C. & Montgomerie, R. 2003. Why do birds engage in extra-pair copulation? Nature, 422, 833e833. Griffith, S. C., Owens, I. P. F. & Thuman, K. A. 2002. Extra pair paternity in birds: a review of interspecific variation and adaptive function. Molecular Ecology, 11, 2195e2212. Halliday, T. & Arnold, S. J. 1987. Multiple mating by females: a perspective from quantitative genetics. Animal Behaviour, 35, 939e941. Hansson, B., Hasselquist, D. & Bensch, S. 2004. Do female great reed warblers seek extra-pair fertilizations to avoid inbreeding? Proceedings of the Royal Society B, 271, S290eS292. Harvey, P. H., Greenwood, P. J. & Perrins, C. M. 1979. Breeding area fidelity of great tits (Parus major). Journal of Animal Ecology, 48, 305e313. Keller, L. F. & Waller, D. M. 2002. Inbreeding effects in wild populations. Trends in Ecology & Evolution, 17, 230e241. Keller, L. F. & Arcese, P. 1998. No evidence for inbreeding avoidance in a natural population of song sparrows (Melospiza melodia). American Naturalist, 152, 380e392. Kempenaers, B., Adriaensen, F., Van Noordwijk, A. J. & Dhondt, A. A. 1996. Genetic similarity, inbreeding and hatching failure in blue tits: are unhatched eggs infertile? Proceedings of the Royal Society B, 263, 179e185. Kleven, O., Jacobsen, F., Robertson, R. J. & Lifield, J. T. 2005. Extrapair mating between relatives in the barn swallow: a role for kin selection? Biology Letters, 1, 389e392. Lynch, M. & Walsh, B. 1998. Genetics and Analysis of Quantitative Traits. Sunderland, Massachusetts: Sinauer Associates. Masters, B. S., Hicks, B. G., Johnson, L. S. & Erb, L. A. 2003. Genotype and extra-pair paternity in the house wren: a rare-male effect? Proceedings of the Royal Society B, 270, 1393e1397. Patrick, S. C., Chapman, J. R., Dugdale, H. L., Quinn, J. L. & Sheldon, B. C. 2012. Promiscuity, paternity and personality in the great tit. Proceedings of the Royal Society B, 279, 1724e1730. Quinn, J. L., Cole, E. F., Patrick, S. C. & Sheldon, B. C. 2011. Scale and state dependence of the relationship between personality and dispersal in a great tit population. Journal of Animal Ecology, 80, 918e928. R Development Core Team 2010. R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing. Rubenstein, D. R. 2007. Female extrapair mate choice in a cooperative breeder: trading sex for help and increasing offspring heterozygosity. Proceedings of the Royal Society B, 274, 1895e1903. Savill, P., Perrins, C., Kirby, K. & Fisher, N. 2010. Wytham Woods: Oxford’s Ecological Laboratory. Oxford: Oxford University Press. Shutler, D., Hussell, D. J. T., Horn, A. G., Leonard, M. L., Shutler, R. W. & Lepage, D. 2004. Breeding between tree swallows from the same brood. Journal of Field Ornithology, 75, 353e358. Slagsvold, T. & Lifjeld, J. T. 1997. Incomplete female knowledge of male quality may explain variation in extra-pair paternity in birds. Behaviour, 134, 353e371. Slate, J. 2009. Bias in studies of heterozygosity and mate choice. Molecular Ecology, 18, 1815e1817. Stewart, I. R. K., Hanschu, R. D., Burke, T. & Westneat, D. F. 2006. Tests of ecological, phenotypic, and genetic correlates of extra-pair paternity in the house sparrow. Condor, 108, 399e413. Suter, S. M., Keiser, M., Feignoux, R. & Meyer, D. R. 2007. Reed bunting females increase fitness through extra-pair mating with genetically dissimilar males. Proceedings of the Royal Society B, 274, 2865e2871. Szulkin, M. & Sheldon, B. C. 2007. The environmental dependence of inbreeding depression in a wild bird population. PLoS One, 2, e1027. Szulkin, M. & Sheldon, B. C. 2008a. Dispersal as a means of inbreeding avoidance in a wild bird population. Proceedings of the Royal Society B, 275, 703e711. Szulkin, M. & Sheldon, B. C. 2008b. Correlates of the occurrence of inbreeding in a wild bird population. Behavioral Ecology, 19, 1200e1207. Szulkin, M., Garant, D., McCleery, R. H. & Sheldon, B. C. 2007. Inbreeding depression along a life-history continuum in the great tit. Journal of Evolutionary Biology, 20, 1531e1543.
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370 Szulkin, M., Zelazowski, P., Nicholson, G. & Sheldon, B. C. 2009. Inbreeding avoidance under different null models of random mating in the great tit. Journal of Animal Ecology, 78, 778e788. Szulkin, M., Stopher, K. V., Pemberton, J. M. & Reid, J. M. In press. Inbreeding avoidance, tolerance or preference in animals? Trends in Ecology & Evolution. Tarvin, K. A., Webster, M. S., Tuttle, E. M. & Pruett-Jones, S. 2005. Genetic similarity of social mates predicts the level of extrapair paternity in splendid fairywrens. Animal Behaviour, 70, 945e955. Thunken, T., Bakker, T. C. M., Baldauf, S. A. & Kullmann, H. 2007. Active inbreeding in a cichlid fish and its adaptive significance. Current Biology, 17, 225e229. Van Oers, K., Drent, P. J., Dingemanse, N. J. & Kempenaers, B. 2008. Personality is associated with extrapair paternity in great tits, Parus major. Animal Behaviour, 76, 555e563. Varian-Ramos, C. W. & Webster, M. S. 2012. Extrapair copulations reduce inbreeding for female red-backed fairy-wrens, Malurus melanocephalus. Animal Behaviour, 83, 857e864. Verhulst, S., Perrins, C. M. & Riddington, R. 1997. Natal dispersal of great tits in a patchy environment. Ecology, 78, 864e872.
1369
Wang, C. & Lu, X. 2011. Female ground tits prefer relatives as extra-pair partners: driven by kin-selection? Molecular Ecology, 20, 2851e2863. Wang, J. L. 2010. Do marker-based paternity assignments favour heterozygous and unrelated males? Molecular Ecology, 19, 1898e1913. Westneat, D. F. & Sherman, P. W. 1997. Density and extra-pair fertilizations in birds: a comparative analysis. Behavioral Ecology and Sociobiology, 41, 205e 215. Westneat, D. F. & Stewart, I. R. K. 2003. Extra-pair paternity in birds: causes, correlates, and conflict. Annual Review of Ecology Evolution and Systematics, 34, 365e396. Wetzel, D. P. & Westneat, D. F. 2009. Heterozygosity and extra-pair paternity: biased tests result from the use of shared markers. Molecular Ecology, 18, 2010e 2021. Wilkin, T. A., Garant, D., Gosler, A. G. & Sheldon, B. C. 2006. Density effects on lifehistory traits in a wild population of the great tit Parus major: analyses of longterm data with GIS techniques. Journal of Animal Ecology, 75, 604e615. Zuur, A. F., Leno, E. N., Walker, N., Saveliev, A. A. & Smith, G. M. 2009. Mixed Effects Models and Extentions in Ecology with R. New York: Springer.
Appendix Table A1 Robustness of parameter estimates when models with binomial error structure are run with and without random effects Response variable
BS/CS
No. of EPY/BS No. of EPY/BS
No. of EPY/BS
Variables from best model
Inbred Female age Male age Inbred Female age Female status: Outbreeding locally born Outbreeding immigrant Hatch date Female age Male status: Outbreeding locally born Outbreeding immigrant
Binomial errorsþ random effects
Quasibinomial errors (without random effects)
Effect size
SE
c2
df
P
Effect size
SE
1.32 0.75 1.13 1.09 1.13
0.41 0.41 0.46 0.57 0.45
9.00 3.16 6.43 4.11 6.25 17.94
1 1 1 1 1 2
0.003 0.075 0.011 0.042 0.012 <0.001
1.23 0.66 0.96 1.10 1.09
0.41 0.36 0.39 0.59 0.45
0.05 1.67** 0.49 1.54
0.60 0.52 0.21 0.44
0.08 1.66** 0.49 1.55
0.63 0.55 0.21 0.46
1.26 0.74
0.65 0.72
1.18 0.72
0.68 0.75
4.62 13.32 4.32
1 1 2
0.032 <0.001 0.115
F
df
P
8.78 2.66 5.13 4.28 6.60 9.56
1 1 1 1 1 2
0.005 0.111 0.030 0.045 0.014 <0.001
5.50 13.81 1.96
1 1 2
0.025 <0.001 0.155
N¼ 40. BS, CS and EPY stand for brood size, clutch size and number of extrapair young, respectively. There is a debate to what extent models with binomial proportions and random effects accurately estimate data overdispersion. We therefore contrasted all models using binomial data in this study and tested different approaches to control for overdispersion. The model run with random effects included random effects of blocking treatment (set of inbred and neighbouring outbred nests) and brood level observation (generating 40 levels for this random effect and expected to control for overdispersion in the data). The model presented in the main text used quasibinomial errors, generated only minimally different effect sizes and more conservative P values than the model run with random effects. Female and male status (outbreeding locally born, outbreeding immigrant) were tested relative to inbreeding females and males. Asterisks denote significant (P < 0.01) estimates for particular levels of female/male status when tested against inbreeding females/males using a t distribution.
1370
M. Szulkin et al. / Animal Behaviour 84 (2012) 1363e1370
Figure A1. Illustration of developmental abnormalities likely to contribute to inbreeding depression in the wild. Offspring born from a mothereson mating, photographed 16 days after hatching (May 2006). Two offspring (a, c) have delayed feather development (but normal body mass and size for their age); two other offspring (b, d), inbred at f ¼ 0.25, appear to be viable at this stage of their development. The photograph is reflective of the genetics underlying inbreeding depression, whereby two offspring most likely inherited the same set of recessive deleterious mutations from their mother and father, while the two other inbred within-pair young have not. The fifth offspring from this brood was sired by an extrapair male which had already fledged on day 16; its photograph is not available, but the phenotypic development on day 15 was normal for that individual.