Inbreeding avoidance mechanisms: dispersal dynamics in cooperatively breeding southern pied babblers

Authors

  • Martha J. Nelson-Flower,

    Corresponding author
    1. Percy FitzPatrick Institute of African Ornithology, DST/NRF Centre of Excellence, University of Cape Town, Rondebosch 7701, South Africa
    Search for more papers by this author
  • Philip A. R. Hockey,

    1. Percy FitzPatrick Institute of African Ornithology, DST/NRF Centre of Excellence, University of Cape Town, Rondebosch 7701, South Africa
    Search for more papers by this author
  • Colleen O’Ryan,

    1. Department of Molecular and Cell Biology, University of Cape Town, Rondebosch 7701, South Africa
    Search for more papers by this author
  • Amanda R. Ridley

    1. Percy FitzPatrick Institute of African Ornithology, DST/NRF Centre of Excellence, University of Cape Town, Rondebosch 7701, South Africa
    2. Department of Biological Sciences, Macquarie University, Marsfield, NSW 2122, Australia
    Search for more papers by this author

Corresponding author. E-mail: marthajn@hotmail.com

Summary

1. Breeding with kin can reduce individual fitness through the deleterious effects of inbreeding depression. Inbreeding avoidance mechanisms are expected to have developed in most species, and especially in cooperatively breeding species where individuals may delay dispersal until long after sexual maturity. Such potential mechanisms include sex-biased dispersal and avoidance of kin known through associative learning.

2. The investigation of inbreeding avoidance through dispersal dynamics can be enhanced by combining fine-scale population genetic structure data with detailed behavioural observations of wild populations.

3. We investigate possible inbreeding avoidance in a wild population of cooperatively breeding southern pied babblers (Turdoides bicolor). A combination of genetic, geographic and observational data is used to examine fine-scale genetic structure, dispersal (including sex-biased dispersal) and inheritance of dominance in cooperatively breeding groups.

4. Unusually, sex-bias in dispersal distance does not occur. Rather, individuals appear to avoid inbreeding through two routes. First, through dispersal itself: although both males and females disperse locally, they move outside the range within which genetically similar individuals are usually found, going twice as far from natal groups as from non-natal groups. Second, through avoidance of familiar group members as mates: individuals inherit a dominant position in the natal group only when an unrelated breeding partner is present.

5. This study uses spatial genetic analyses to investigate inbreeding avoidance mechanisms in a cooperative breeder and shows that individuals of both sexes can avoid inbreeding through a dispersal distance mechanism. While it appears that dispersal allows most individuals to move beyond the range of closely related kin, matings may still occur between distant kin. Nevertheless, any costs of breeding with a distant relative may be outweighed by the benefits of local dispersal and the immense fitness gains available from attaining a breeding position.

Introduction

Breeding with close relatives can expose individuals to a variety of problems, including the unmasking of deleterious recessive alleles and a decrease in overall fitness, a phenomenon termed inbreeding depression (Pusey 1987; Pusey & Wolf 1996; reviewed by Keller & Waller 2002). In many cooperatively breeding species, breeders and helpers live year-round in highly related social groups (Emlen 1997; Koenig & Haydock 2004). In some species, both sexes remain in the natal group well beyond reaching sexual maturity, at which point the only available breeding partners in the group may be close relatives (Cockburn 1998; Ekman et al. 2004). Additionally, many cooperatively breeding species are highly sedentary and disperse over short distances (Zack 1990), resulting in high local densities of close relatives. Most cooperatively breeding birds avoid inbreeding (Koenig & Haydock 2004; but see Williams & Rabenold 2005; Townsend et al. 2009), but how they avoid relatives as mates both in the natal group and after dispersal is still debated.

Inbreeding avoidance mechanisms are expected to have developed in naturally out-crossing populations or species (Keller & Waller 2002). One such mechanism is natal dispersal, whereby individuals simply move away from the area of their birth, thus avoiding close relatives as mates (Wright 1943; Howard 1960; Greenwood 1980; Szulkin & Sheldon 2008). A second proposed inbreeding avoidance mechanism is sex-biased dispersal, which occurs when one sex disperses more frequently or further than the other, with the net result that opposite-sex relatives become spatially separated (Greenwood 1980). A third proposed mechanism is for individuals to avoid breeding with ‘natal-familiar’ partners, that is, those that were present in the social group during their dependent period (Hamilton 1964; Zahavi 1990; Komdeur & Hatchwell 1999). Many cooperatively breeding birds learn to recognize kin during the dependent period (Komdeur & Hatchwell 1999; Baglione et al. 2003; Richardson, Burke & Komdeur 2003; Koenig & Haydock 2004; Sharp et al. 2005) which, for these birds, is prolonged (Langen 2000).

In most field studies, it is difficult to distinguish between individuals that have dispersed beyond the limits of the study area and those that have died (Koenig, van Vuren & Hooge 1996). Genetic data should therefore be used alongside observational data to investigate population structure, gene flow and dispersal (Peakall, Ruibal & Lindenmayer 2003; Harris et al. 2009). These analyses allow characterization of average dispersal distance (Woxvold, Adcock & Mulder 2006; Beck, Peakall & Heinsohn 2008) and genetic similarity among neighbours (Double et al. 2005). Such genetic data also make it possible to examine sex differences in gene flow and relatedness in populations (including where dispersal distances are so short that both sexes appear philopatric) to assess the potential function of sex-biased dispersal as an inbreeding avoidance mechanism. For example, correlations between genetic and geographic distances between individuals have revealed female-biased dispersal in cooperatively breeding birds (Double et al. 2005; Temple, Hoffman & Amos 2006; Woxvold, Adcock & Mulder 2006) and male-biased dispersal in fish (Cano, Mäkinen & Merilä 1998; Stiver et al. 2007).

We use genetic data and behavioural observations to explore evidence of inbreeding avoidance mechanisms in a population of cooperatively breeding southern pied babblers (Turdoides bicolor). Inbreeding is avoided in this species: relatives do not court one another, breeding pairs are as related to one another as would be expected by chance, and inbreeding coefficients are non-significant (Nelson-Flower et al. 2011). To investigate inbreeding avoidance mechanisms, we first investigate dispersal and spatial patterns of relatedness in the babbler population as a whole using observational, genetic and geographic data. We then assess sex-biased dispersal as an inbreeding avoidance mechanism by examining observational data for sex-linked differences in the occurrence of dispersal, dispersal distance and immigration into non-natal groups. We again use genetic and geographic data to investigate population genetic structure for evidence of sex-biased dispersal. Finally, in an exploration of whether birds avoid natal-familiar partners, we examine the circumstances under which individuals inherit dominance.

Materials and methods

Study Site and Study Population

Comprehensive observations of southern pied babbler life histories and breeding attempts were collected from July 2003 to April 2009 at the Kuruman River Reserve, South Africa (26°58′S; 21°49′E; see Raihani & Ridley 2007 for information on climate and vegetation). Twenty-nine wild babbler groups were habituated to the presence of a human observer at a distance of 2–3 m; allowing observational data to be collected without disturbing natural behaviour (see Ridley & Raihani 2007a for habituation techniques). GPS coordinates of groups were recorded every 15 min throughout observation sessions. Each individual in the population was ringed with a unique combination of coloured leg-rings. Blood was collected from 319 individuals that hatched before May 2008; these birds were genotyped at nine polymorphic microsatellite loci (Nelson-Flower et al. 2011) and were sexed using the CHD-W and CHD-Z markers (Griffiths et al. 1998).

Southern pied babbler groups typically comprised a dominant pair with subordinate helpers, the status of which could be determined by the observation of dominance assertions (pecks and other attacks), to which subordinates responded with submissive behaviour (Raihani 2008). Babbler groups ranged in size from two to 11 adults (individuals over 12 months old; Ridley & Raihani 2008) with a mean group size of 4·3 ± 0·2 adults. Habituated groups were adjacent to one another, and defended year-round territories of 75–150 ha that were arranged in a cluster (Golabek 2011). Mean distance between centres of nearest neighbouring territories was 588 ± 29 m (N = 88 measurements over 6 years). The dominant pair monopolized over 95% of reproduction; subordinate helpers were usually the offspring of the dominants, although immigrant subordinate helpers also rarely joined groups (4·1 ± 0·7% of subordinates lived in non-natal groups each year; Nelson-Flower et al. 2011). Individuals that attempted to immigrate into non-natal groups (‘prospectors’) were initially chased and attacked by members of the target group (Raihani et al. 2010). Most prospectors appeared to leave their home groups of their own accord, but some individuals were forced to leave by other group members and not allowed to return (eviction; A.R. Ridley, unpublished data). Prospectors immigrated into groups singly or in pairs (A.R. Ridley, unpublished data). Prospectors could eventually immigrate into non-natal groups as either a breeding dominant or helpful subordinate: older birds were more likely to immigrate as dominants while younger birds were more likely to immigrate into subordinate positions, regardless of sex (Raihani et al. 2010). Sex did not affect the age at which an individual first prospected (471 ± 35 days, N = 37 individuals) or successfully dispersed (565 ± 65 days, N = 23 individuals; Raihani et al. 2010).

Dispersal

Dispersal was considered to have occurred if a ringed individual from the study population left one group, joined another group and stayed there for at least 2 weeks (Ridley, Raihani & Nelson-Flower 2008). Movements of individuals returning to their natal groups from non-natal groups were not considered to be dispersal, nor were movements of individuals that followed a parent to a new group when the natal group dissolved (N = 3 of 61 movements between groups). Following Greenwood (1980), dispersal was categorized as ‘natal’ if the individual left its natal group to enter another group, and ‘breeding’ if the individual went from one non-natal group to another. When an individual dispersed, the distance between the group it left and the group it entered was measured. Group location was determined by finding a mean coordinate using the final GPS coordinate per observation session over the month that the individual was last seen in the group it left, and the month that it was first seen in the new group. Disperser sex and dispersal type (from a natal or breeding group) were investigated to understand how these affected dispersal distance. To do this, a linear mixed model was performed within r version 2.12.1 (R Development Core Team 2010), with (log-transformed) dispersal distance as the response variable. The sex of the disperser and the type of dispersal (natal or breeding) were included as explanatory terms, and the disperser’s identity nested within its group were included as random terms. Data include 58 dispersal events by 46 individuals from 18 groups.

Because recorded dispersal events and distances are usually an underestimate of true dispersal incidences and distances (Koenig, van Vuren & Hooge 1996), genetic data are useful as an alternative measure of population structure and can clarify dispersal patterns. Measures of genetic similarity between individuals were based upon microsatellite DNA variation (Nelson-Flower et al. 2011). A series of spatial analyses were performed to compare genetic similarity with geographic distance. In a continuous population, short dispersal distances are expected to result in ‘positive genetic structure,’ such that genetic similarity is inversely correlated with the distance between individuals (Wright 1943). Positive genetic structure indicates whether most dispersal is local (Beck, Peakall & Heinsohn 2008). Spatial autocorrelation was calculated in genalex 6.41 (Peakall & Smouse 2006) to inspect fine-scale genetic structure (as detailed in Peakall, Ruibal & Lindenmayer 2003; Double et al. 2005; Beck, Peakall & Heinsohn 2008; Blackmore, Peakall & Heinsohn 2011). Spatial autocorrelation uses geographic distance and genetic distance matrices. It estimates an autocorrelation coefficient, r, for each group of individuals within a specified geographical range and produces a graphical representation of genetic and geographic distance. The coefficient r is bounded by −1 and +1 and has a mean of zero when there is no autocorrelation (Smouse & Peakall 1999). Significance is tested by 999 random permutations of the data set, which provide 95% confidence intervals about the null hypothesis of no spatial genetic structure. Significant positive or negative genetic structure is indicated if r falls outside the 95% confidence interval bounding the null hypothesis (Smouse & Peakall 1999). To examine spatial genetic structure for the population as a whole, spatial autocorrelation was first investigated for all individuals (N = 311 individuals in 25 groups), and then for dominants only (N = 55 individuals in 19 groups). Distance classes used for spatial autocorrelation analyses were 600 m as this was the average distance between nearest neighbours.

Sex-Biased Dispersal

Sex-biased dispersal leads to the physical separation of related potential mates (Greenwood 1980). First, sex bias in dispersal itself was examined using binomial tests within r version 2.12.1 (R Development Core Team 2010). The population sex ratio of adults (males/males + females) was compared to the sex ratio of dispersing individuals (both natal and breeding) to determine whether any observed biases in dispersal were simply the result of an unequal population sex ratio. Second, sex bias in immigration by all birds into previously established study groups was measured. Immigration includes all individuals (both ringed and previously unknown birds) entering previously established groups. It may be a better measure of the consequences of dispersal than movement of individuals out of groups because when a bird leaves a group but is not seen again, it is impossible to determine whether this represents a dispersal event (to a receiving group outside the study area) or mortality (Koenig, van Vuren & Hooge 1996). Immigration occurred when individuals entered established groups in the study population and stayed at least two weeks. Immigration could occur as a consequence of a breeding vacancy due to death or emigration, or through aggressive overthrow of a dominant bird (Raihani et al. 2010).

Population genetic structure can reveal the existence or extent of sex-biased dispersal (Beck, Peakall & Heinsohn 2008). Subordinates in groups are typically highly related to one another, and groups are genetically differentiated from one another (Nelson-Flower et al. 2011). However, if males and females differed in their between-group genetic variation, this could indicate sex-biased dispersal. For example, if males were typically related to males in several other groups, but females were not, this would indicate female-biased dispersal. Several tests are available to detect evidence of sex-biased dispersal from genetic data; these include hierarchical analysis of molecular variance (amova), corrected assignment index (AIc) and combined spatial autocorrelation analysis for males and females (Beck, Peakall & Heinsohn 2008; Blackmore, Peakall & Heinsohn 2011). A hierarchical amova was performed within genalex 6.41 (Peakall & Smouse 2006): this allowed a simultaneous examination of the genetic differentiation among sexes and groups (Blackmore, Peakall & Heinsohn 2011). Males and females were included in the same data set, with males and females grouped separately and each sex further partitioned into social groups (in genalex 6.41 terminology, males and females were called ‘regions’, with social groups called ‘populations’). Under strong sex-biased dispersal, significant differentiation among the sexes may be expected when the dispersing sex comes from genetically differentiated neighbouring populations (Blackmore, Peakall & Heinsohn 2011). Again, random permutation (1000 permutations) was used to test for significance. This analysis required two or more males or females per group (N = 141 males and 163 females in 23 groups). Genetic data and an assignment test-based procedure (the corrected assignment index or AIc) were then used to investigate the prevalence of sex-biased dispersal (performed within genalex 6.41; Peakall & Smouse 2006). AIc determines the expected frequency of each individual’s genotype in the population (Favre et al. 1997; detailed in Beck, Peakall & Heinsohn 2008 and Blackmore, Peakall & Heinsohn 2011). Overall, AIc values will average to zero for the entire population, but under sex-biased dispersal, a significant difference exists between male and female means (assessed using a Mann Whitney U-test). In addition, negative AIc values with larger variances are expected for the dispersing sex (Mossman & Waser 1999). Only individuals that had been typed at all nine polymorphic microsatellite loci could be used in AIc analysis. The analysis was carried out first with all individuals (N = 302 individuals: 138 males and 164 females) and then with dominants only (N = 54 individuals: 29 males and 25 females). Spatial analyses were also used to investigate sex-biased dispersal. A pattern of local positive genetic structure for males but not females would indicate that males tended towards philopatry while females dispersed more randomly (Double et al. 2005). A combined spatial autocorrelation analysis for male and female dominants (detailed in Beck, Peakall & Heinsohn 2008) was used to determine whether sex differences exist in fine-scale genetic structure as a result of sex-biased dispersal (N = 55 individuals in 19 groups).

Avoiding Natal-Familiar Breeding Partners

Breeding vacancies and how they were filled were investigated to determine whether birds avoided familiar relatives as mates. For each breeding vacancy that occurred, the sex of the missing bird, the source of the individual that filled the vacancy (natal or immigrant) and whether the group already contained an unrelated adult subordinate of the appropriate sex to fill the breeding vacancy were recorded. Binomial tests in r version 2.21.1 (R Development Core Team 2010) were used to determine whether vacancies were filled by immigrants more often when only related subordinates could inherit, compared to when an unrelated potential inheritor was present.

Results

Dispersal

Dispersal was recorded for 36 females and 22 males that moved into non-natal groups and remained there for at least two weeks. The type of dispersal (natal vs. breeding) strongly influenced dispersal distance (F1,35 = 8·93, = 0·005), although disperser sex had no effect (F1,35 = 0·15, = 0·697). While males and females moved a similar distance from their previous groups, natal dispersal was nearly twice as far as breeding dispersal (Table 1). Although observed dispersal distance is an underestimate (Koenig, van Vuren & Hooge 1996), both sexes dispersed significantly further from their natal groups than they did from non-natal groups. Spatial autocorrelation analysis for all individuals combined revealed significant, positive genetic structure from the within-group level (0–0·6 km) up to 1·2 km away (i.e. for two territory widths; Fig. 1a). This pattern of significant, positive genetic structure was still detectable using the dominants-only data set when subordinates were removed (Fig. 1b), showing that most dispersal is over short distances. Taken together, these results indicate that a) most groups comprise genetically similar members, and b) short-distance dispersal is common.

Table 1.   Breakdown of mean dispersal distance (km) and sample size per sex and type of dispersal. Overall means ± SEM are also shown
Disperser sexBreeding dispersal (km)NNatal dispersal (km)NOverall mean: sex (km)N
Female0·96121·78241·51 ± 0·1936
Male0·6731·53191·41 ± 0·1922
Overall mean: dispersal type (km)0·91 ± 0·13151·67 ± 0·1743  
Figure 1.

 Spatial autocorrelation analyses of (a) all individuals in the population, (b) dominants only and (c) dominant males vs. dominant females. Dotted lines in (a) and (b) indicate 95% confidence intervals around the null hypothesis of no spatial genetic structure.

Sex-Biased Dispersal

Relative to the population sex ratio, there was no sex bias in dispersal overall, or in natal or breeding dispersal when considered separately, although a small sample size for breeding dispersal may have precluded finding a significant difference (Table 2). There was also no sex bias in immigration overall or in subordinate immigration, but when considering dominants alone, a significant female bias in immigration was revealed (Table 2). Hierarchical analysis of molecular variation (amova) indicated no genetic differentiation between the sexes (FRT = −0·006, P = 1·0), showing that genetic structure among males and females was similar. This analysis did find significant same-sex group-based differentiation (FSR = 0·047, P = 0·001), supporting previous findings that groups consisted mainly of individuals with higher relatedness (Nelson-Flower et al. 2011). Assignment indices (AIc) also did not differ significantly from one another for all males and all females (Mann–Whitney U-test: Z = 1·033, P = 0·302), or for only dominant males and dominant females (Mann–Whitney U-test: Z = −1·156, P = 0·256). Thus, neither amova nor AIc analyses of population genetic structure provided support for the occurrence of sex-biased dispersal. Comparative spatial autocorrelation analysis of male and female fine-scale genetic structure indicated no significant differences between the sexes (Fig. 1c), again failing to support sex-biased dispersal. This result confirms that the positive genetic autocorrelation for dominants (Fig. 1b) was not driven by high local relatedness for one sex alone. Overall, the genetic data indicate that genetic structures for males and females are similar, and therefore that spatial separation of opposite-sex relatives through sex-biased dispersal does not occur.

Table 2.   Sex ratio (males/ males + females) of the pied babbler population, dispersal (overall, natal and breeding) and immigration (overall, dominant and subordinate) between September 2003 and April 2009
 TotalSex ratioP-value
  1. P-values indicate the significance of the sex ratio of dispersal or immigration compared with the overall population sex ratio.

Population3020·46 
Overall dispersal580·380·344
Natal dispersal430·440.982
Breeding dispersal150·200·091
Overall immigration680·350·154
Dominant immigration420·260·026
Subordinate immigration260·500·828

Avoiding Natal-Familiar Breeding Partners

Individuals were never observed to inherit dominance in groups where a relative was already dominant. There were 33 breeding vacancies recorded between July 2003 and May 2009 (N = 79 group-years) which arose when groups lost a dominant through emigration or death. There was no sex bias in the availability of breeding vacancies: of the 33 vacancies, 15 were male and 18 were female (Binomial test: = 0·728). Of these 33 vacancies, 11 were inherited by natal subordinates and 22 filled by unrelated immigrants. There was no sex bias in the likelihood of these outcomes: 6 of 11 subordinate inheritors were male (Binomial test: = 1·0) and 9 of 22 immigrants were male (Binomial test: = 0·522). A far better predictor of whether a vacancy would be filled by immigration or by inheritance was whether or not the group contained an adult subordinate of the correct sex that was unrelated to the remaining dominant. Of the 33 vacancies, 13 occurred in groups where a subordinate could inherit without inbreeding (i.e. was unrelated to the opposite-sex dominant). Of these 13 vacancies, 11 were inherited by a natal subordinate, and only two were filled by immigrants. Twenty vacancies occurred in groups that did not contain an unrelated adult subordinate of the correct sex. All of these vacancies were filled by unrelated immigrants. Thus, whether vacancies were filled by immigration or inheritance depended on whether or not the group contained an unrelated inheritor (N = 2/13, 20/20; Binomial test, = 0·001).

Discussion

Southern pied babblers avoid inbreeding using two primary mechanisms. First, both sexes use dispersal to dilute the risk of breeding with close relatives. Dispersal is an important inbreeding avoidance mechanism (Szulkin & Sheldon 2008), but can only function as such when individuals disperse beyond the bounds of local populations (Arcese 1989), thereby reducing the risk of a by-chance meeting and mating with a relative. Although spatial genetic analysis of the pied babbler population revealed that dispersal distances are generally short, dispersers move twice as far from natal groups (1·7 km) as from non-natal groups (0·9 km), taking them beyond the range within which most individuals with higher relatedness are found (c. 1·2 km). A comparable dispersal strategy is displayed by female superb fairy-wrens (Malurus cyaneus), with birds dispersing farther from natal groups than from non-natal groups, presumably to avoid pairing with related males living near the natal group (Cockburn et al. 2003). Observational data revealed that females immigrate as dominants into established study groups more often than do males. Breeding vacancies are not sex-biased; the sex-bias in dominant immigration is probably due to the fact that females (but not males) aggressively oust one another from dominance positions, which provides them with an additional route to dominance (Raihani et al. 2010). This is reflected in breeding dispersal that was close to significantly female-biased, while natal dispersal was not. Because breeding dispersal distances were very short, the increased movement of dominant females did not result in the spatial separation of opposite-sex relatives. There was no evidence for a sex difference in babbler dispersal distance, nor did spatial analyses of genetic population structure reveal differences between males and females. Therefore, although dispersal distance is used as an inbreeding avoidance mechanism by this species, sex-biased dispersal distance is not. The second inbreeding avoidance mechanism used by southern pied babblers is the avoidance of natal-familiar breeding partners: they do not inherit a dominant breeding position when the potential breeding partner is a relative, nor do they court relatives within the group (Nelson-Flower et al. 2011).

Although dispersal typically carries individuals beyond their close kin, most dispersal by southern pied babblers is local, as indicated by both the pattern of positive spatial genetic structure and behavioural observations. Local dispersal has also been established using behavioural observations and genetic data in apostlebirds (Struthidea cinerea; Woxvold, Adcock & Mulder 2006). In other cooperative species, local dispersal has been inferred by behavioural observations but refuted by genetic evidence (e.g. grey-crowned babblers Pomatostomus temporalis; Blackmore, Peakall & Heinsohn 2011) or by radiotracking data (e.g. acorn woodpeckers Melanerpes formicivorus; Koenig, van Vuren & Hooge 1996). Southern pied babblers may select local dispersal over the high cost of longer distance dispersal (Koenig, Haydock & Stanback 1998) or floating (when individuals have no fixed territory and remain alone for extended periods; Ridley, Raihani & Nelson-Flower 2008). Dispersal to remote groups requires prospectors to travel long distances and compete with residents on multiple territories, which may be costly (Daniels & Walters 2000). For example, extended periods of prospecting result in chronic elevation of stress hormones in meerkats (Suricata suricatta; Young & Montfort 2009). Local dispersal allows the option of returning to the natal group between prospecting forays. The natal territory can be used as a ‘safe haven’ where relatives may facilitate access to food (Ekman, Bylin & Tegelström 2000) or increase foraging efficiency through cooperative vigilance (Hollén, Bell & Radford 2008). In southern pied babblers, staying in the natal territory before dispersing may be important for maintaining body condition, itself an important predictor of successful dispersal (Ridley, Raihani & Nelson-Flower 2008; Raihani et al. 2010).

Although most southern pied babblers disperse outside the range of genetically similar individuals and avoid breeding with known relatives, some do disperse within this range and may mate with unknown kin. The chance of mating with distant kin is higher when most dispersal is over short distances (Zahavi 1990) and/or if individuals are long-lived. Some babblers in this study were breeding dominants in 2003, when the study site was established, and were still alive in 2011 (10% of 2011 dominants were present in 2003). Given these slow turnover rates, dispersers may regularly encounter, and potentially mate with, distant unknown kin. However, the costs of breeding with a distant relative may be outweighed by the benefits of short-distance dispersal and the very substantial fitness gains available from attaining dominance (dominants monopolize almost all breeding activity; Nelson-Flower et al. 2011). Theoretical work has indicated that inbreeding may be accepted when the costs of dispersal are high (Lehmann & Perrin 2003) as in the case of red-cockaded woodpeckers (Picoides borealis; Daniels & Walters 2000). As the pedigree for this southern pied babbler population becomes more detailed, future work will have the opportunity to investigate the prevalence and fitness consequences of matings between distant kin.

In southern pied babblers, sex differences in dispersal distance and genetic population structure were not found; in the sense of the spatial separation of opposite-sex relatives, sex-biased dispersal does not occur. Only two other species of cooperatively breeding birds have been genetically determined to lack sex-biased dispersal: white-winged choughs Corcorax melanorhamphos (Beck, Peakall & Heinsohn 2008) and grey-crowned babblers (Blackmore, Peakall & Heinsohn 2011). Both of these species, similarly to the southern pied babbler, live in groups that have long-term pair bonds between dominants, helpers that appear to greatly enhance the reproductive success of breeders, and new breeding opportunities that arise rarely (Heinsohn 1992; Blackmore & Heinsohn 2007; Ridley & Raihani 2007b, 2008; Beck, Peakall & Heinsohn 2008; Blackmore, Peakall & Heinsohn 2011). The paucity of breeding positions may explain the lack of sex differences in dispersal distance in southern pied babblers. The inheritance of breeding vacancies occurs rarely, and the main route to breeding is costly dispersal (Raihani et al. 2010); in the closely related Arabian babbler (Turdoides squamiceps), for example, such a dispersal cost is reduced body mass (Ridley 2011). Rather than dispersing a long way as a matter of course (as when sex-biased dispersal is used as an inbreeding avoidance mechanism), southern pied babblers of both sexes may benefit more from acquiring dominance when vacancies arise in the local sub-population and risk breeding with unknown kin.

Pied babblers also reduced the likelihood of inbreeding by avoiding reproduction with familiar natal group members. Neither subordinates nor dominants court relatives within the natal group (Nelson-Flower et al. 2011) and subordinates never inherit dominance when the potential breeding partner is related. Additionally, babblers are never observed to disperse to become dominant in groups with natal-familiar breeding partners. Avoidance of mating with group members known from dependence would be likely to prevent inbreeding between full siblings and between parents and offspring because groups are highly kin-structured (Nelson-Flower et al. 2011). Kin recognition based on associative learning has been detected and tested in cooperatively breeding long-tailed tits (Aegithalos caudatus; Sharp et al. 2005) and Seychelles warblers (Acrocephalus sechellensis; Komdeur, Richardson & Burke 2004) and is under investigation in southern pied babblers (D. Humphries, unpublished data).

Subordinate southern pied babblers of both sexes inherit dominant positions in their groups only when the opposite-sex dominant is unrelated. Failure to take dominance alongside relatives in the natal group does occur in a few other cooperatively breeding birds (Koenig, Stanback & Haydock 1999; Daniels & Walters 2000; Hatchwell et al. 2000; Haydock, Koenig & Stanback 2001; Cockburn et al. 2003; Blackmore & Heinsohn 2008) but strict avoidance of a related social mate is unusual. For example, subordinates may inherit dominance positions in groups containing only relatives, but breed with extra-group partners (Rowley, Russell & Brooker 1986; Amos, Schlotterer & Rautz 1993; Cockburn et al. 2003; Charpentier et al. 2005; Spong et al. 2008; Blackmore & Heinsohn 2008). Among southern pied babblers, dominants may occasionally breed outside the pair bond, but they never breed outside the group (Nelson-Flower et al. 2011), perhaps because groups are intensely territorial (Golabek 2011), limiting access to unrelated extra-pair breeding partners. Also unusually, dominant females rarely abandon breeding positions following the death of the dominant male (N = 1 of 15 male vacancies). If the only breeding partners for the widowed dominant female are relatives, male vacancies are rapidly filled by immigrants. Among other cooperatively breeding birds, abdication by the dominant female is more common and is itself interpreted as an inbreeding avoidance mechanism (Woolfenden & Fitzpatrick 1984; Walters, Doerr & Carter 1988; Hatchwell et al. 2000; Yáber & Rabenold 2002). Because southern pied babblers face such intense competition for dominance positions (A.R. Ridley, unpublished data), widowed females that abdicated would find it difficult to acquire dominance in another group.

In summary, fine-scale genetic structure analyses, coupled with detailed behavioural observations, can allow valuable windows into the mechanisms whereby animals avoid inbreeding. Both male and female southern pied babblers disperse at the same rate and over roughly the same distance, which exceeds the range of the locally related population. This strategy, combined with avoidance of known relatives as mates within the group, has proved effective in avoiding inbreeding (Nelson-Flower et al. 2011). While inbreeding within the natal group is completely avoided, errors may still be possible at the larger spatial scale of the subpopulation, resulting in occasional pairings of kin. Such pairings could be costly, but this might be outweighed by the low cost of local dispersal and the sizeable fitness gains available from attaining the dominant position.

Acknowledgements

We are very grateful to Professor T. H. Clutton-Brock and also the staff at the Kuruman River Reserve for support in setting up and maintaining the Pied Babbler Research Project. The Northern Cape Conservation Authority permitted research on southern pied babblers and the Kotzes and the de Bruins kindly allowed us access to babbler groups on their land. We are grateful to Dr. Deborah Dawson at the NERC Biomolecular Analysis Facility – Sheffield, University of Sheffield, UK for supplying test primers. Katharine Bradley and Rebecca Rose provided valuable help in the field. Matt Bell, Lucy Browning, Krys Golabek, Sarah Knowles, Jenny Oates, Andy Radford, Nichola Raihani and Helen Wade helped with maintaining habituation of babbler groups. We thank Tom Flower, Sinead English, Stuart Piertney and three anonymous reviewers for providing helpful comments, and Jenny Guthrie for logistical aid at a crucial moment. Fieldwork received ethical clearance from the University of Cape Town’s Animal Ethics Committee. This work was supported by the Department of Science and Technology and National Research Foundation Centre of Excellence at the Percy FitzPatrick Institute for African Ornithology, University of Cape Town.

Ancillary