Biased Helping in a Cooperatively Breeding ... - Wiley Online Library

Ethology

Female-Biased Helping in a Cooperatively Breeding Bird: Female Benefits or Male Costs? Dean A. Williams & Amanda M. Hale Department of Biological Sciences, Purdue University, West Lafayette, IN, USA

Correspondence Dean A. Williams, Marine Genomics Laboratory, Rosenstiel School of Marine and Atmospheric Sciences, University of Miami, 4600 Rickenbacker Causeway, Miami, FL 33149, USA. E-mail: [email protected]

Received: September 6, 2006 Initial Acceptance: October 25, 2006 Final Acceptance: December 26, 2006 (S. Foster) doi: 10.1111/j.1439-0310.2007.01353.x

Abstract There is often a sex bias in helping effort in cooperatively breeding species with both male and female helpers, and yet this phenomenon is still poorly understood. Although sex-biased helping is often assumed to be correlated with sex-specific benefits, sex-specific costs could also be responsible for sex-biased helping. Cooperatively breeding brown jays (Cyanocorax morio) in Monteverde, Costa Rica have helpers of both sexes and dispersal is male-biased, a rare reversal of the female-biased dispersal pattern often seen in birds. We quantified helper contributions to nestling care and analyzed whether there was sex-biased helping and if so, whether it was correlated with known benefits derived via helping. Brown jay helpers provided over 70% of all nestling feedings, but they did not appear to decrease the workload of breeders across the range of observed group sizes. Female helpers fed nestlings and engaged in vigilance at significantly higher levels than male helpers. Nonetheless, female helpers did not appear to gain direct benefits, either through current reproduction or group augmentation, or indirect fitness benefits from helping during the nestling stage. While it is possible that females could be accruing subtle future direct benefits such as breeding experience or alliance formation from helping, future studies should focus on whether the observed sex bias in helping is because males decrease their care relative to females in order to pursue extra-territorial forays. Explanations for sex-biased helping in cooperative breeders are proving to be as varied as those proposed for helping behavior in general, suggesting that it will often be necessary to quantify a wide range of benefits and costs when seeking explanations for sex-biased helping.

Introduction Cooperative breeding is a social system in which more than two individuals contribute to rearing young. Traditionally the term ‘helper’ has been applied to both breeding and non-breeding individuals that provide parent-like behavior toward offspring that are not their own (Skutch 1935, 1961). Numerous fitness benefits of helping behavior have been proposed and these can be divided into two main categories: those based on increasing the reproduction and survival of relatives (indirect fitness) 534

and those based on benefits that accrue directly to the helper by increasing its chance to breed (Cockburn 1998; Dickinson & Hatchwell 2004). In cooperative breeders, both the sex of helpers and helping effort are often biased toward one sex (Cockburn 1998). In some species, a sex bias in helping may simply be a result of sex-biased dispersal (Greenwood 1980; Cockburn 1998). In birds, for instance, females usually disperse before their first breeding season whereas males are more philopatric and are the primary helpers. In a number of cooperatively breeding species, however, both sexes remain Ethology 113 (2007) 534–542 ª 2007 The Authors Journal compilation ª 2007 Blackwell Verlag, Berlin

D. A. Williams & A. M. Hale

philopatric through at least their first breeding season (and sometimes much longer) and thus have the opportunity to provide help (Cockburn 1998). Sexbiased helping in species with delayed dispersal by both sexes is still poorly understood (Koenig et al. 1992; Cockburn 1998). Most explanations for sex-biased helping have focused on the benefits a particular sex could receive as a result of increasing help. In several well-studied species, an association between philopatry and sexbiased helping has been linked to direct benefits that increase the chance a helper will inherit or breed on the natal territory (Woolfenden & Fitzpatrick 1984, 1986; Allaine´ et al. 2000; Clutton-Brock et al. 2002; Richardson et al. 2002). For instance, helping could increase an individual’s chance of inheriting its natal territory by augmenting group size, which in turn would increase its probability of future breeding through increased survival, increased territory size, or the production of more future helpers (Brown 1987; Kokko et al. 2001; Clutton-Brock 2002). Sex-biased helping is not exclusively associated with philopatry, however, and there are several hypotheses to explain sex-biased helping in either the philopatric or dispersive sex (e.g. Koenig et al. 1983; Cockburn 1998). A sex-specific benefit could arise if one sex has a higher variance in lifetime reproductive success (i.e. lower chance of future direct reproduction). That sex may then help more to increase its indirect component of inclusive fitness by helping relatives (Koenig et al. 1983). Alternatively, one sex may help more if helping increases its chance of current or future reproduction such as by laying eggs in a communal nest, gaining matings with the breeding female, or increasing its experience with breeding skills (Koenig et al. 1983; Li & Brown 2002; Richardson et al. 2002; Berg 2005). Another possibility that has received relatively less attention is that one sex may help less because there is a trade-off between helping effort and time spent in other activities such as extra-territorial forays that increase the chance of future successful dispersal and extra-group breeding (Clutton-Brock et al. 2002; Young et al. 2005). Brown jays (Cyanocorax morio) are cooperative breeders that provide a rare exception to the femalebiased dispersal pattern observed in many birds by exhibiting male-biased dispersal and female territory inheritance (Williams & Rabenold 2005). Brown jays live in large groups (mean ¼ 10 individuals) consisting of male and female helpers and a primary breeding female and her male consort. Secondary breeding male–female pairs with their own nests are present Ethology 113 (2007) 534–542 ª 2007 The Authors Journal compilation ª 2007 Blackwell Verlag, Berlin

Female-Biased Helping in Brown Jays

in 29% of groups (Williams 2004). Females inherit the primary breeding position or breed as secondary females in their natal territory when they are ‡5 yr of age, whereas males disperse when they are ‡2 yr of age (Williams & Rabenold 2005). Virtually all individuals within a social group contribute to feeding nestlings and fledglings and to territorial and predator defense (Skutch 1935, 1960; Lawton & Lawton 1985; Williams et al. 1994; Hale et al. 2003). Although helpers increase juvenile and yearling production within groups, it does not appear that nestling care per se is responsible for this relationship (Williams & Hale 2006). Rather, a reduction in predation because of the additional vigilance provided by helpers during the juvenile period seems to be the most likely explanation for the increase in juvenile and yearling production (Williams & Hale 2006). Helpers residing in their natal territories do gain some indirect fitness benefits as a result of their help, although these are fairly low (0.06 offspring equivalents per helper) because polyandrous matings by the primary female are common and there are a large number of helpers. The benefit of living in large groups (more than eight individuals) is substantial to breeders, however, and results in a doubling of reproductive success from an average of one to two juveniles per year (Williams & Hale 2006). There is no evidence for a relationship between group size and adult survival in this population, although survivorship is very high (>0.86 yr)1) and so detecting small but important increases in survival with group size is statistically difficult (Williams 2000). Genetic parentage analyses, determined from multilocus DNA fingerprinting, revealed that male and female helpers in their natal territories rarely gain parentage in the primary breeding female’s nest (Williams 2004; Williams & Rabenold 2005). Immigrant males commonly gain paternity by mating with primary females. Although brown jay helpers do not seem to benefit in the current breeding season by feeding nestlings, they could obtain longer-term indirect fitness benefits from their efforts by decreasing the work load of primary breeding females and presumably increasing the females’ survival (e.g. Crick 1992; Langen & Vehrencamp 1999), gaining matings with the primary female, or simply by gaining experience. These benefits could potentially vary with the sex and status of helpers. Alternatively, patterns of offspring care may also result from a reduction of care by one sex or age group. Brown jay males commonly visit neighboring groups as a prelude to dispersal and as a way to gain mating opportunities, 535


and so there may be a trade-off between helping and extra-territorial forays (Williams & Rabenold 2005). In this study, we quantified individual helper contributions to nestling care and analyzed if helping varied by the sex and dispersal status of helpers. We also analyzed whether the feeding effort of helpers decreased the effort expended by breeders. Methods Study System

The study area encompassed 14–16 brown jay social groups and was located in Monteverde, Costa Rica at an elevation of 1400–1500 m on the Pacific slope of the Cordillera de Tilara´n (1012¢N, 8442¢W). The Monteverde plateau is bounded on the east by a large expanse of cloud forest and on the south and west by cliffs. The study site is composed of considerable remnant pre-montane and lower montane moist/wet forest, windbreaks, pastures, orchards and gardens, and receives about 2500 mm of rain annually (Nadkarni & Wheelwright 2000). Nestlings (approx. 20 d old) have been banded with unique combinations of colored aluminum leg bands since 1988 and genetically sampled since 1992. In addition, 114 adults were captured from 1994 to 1996. The number of genetically unsampled adults in a group averaged two individuals from 1994 to 1996. By 1996, 78% of the adult population was genetically sampled and 84% of the adult population was banded. Brown jays are sexually monomorphic, so all genetically sampled individuals were sexed using the polymerase chain reaction (PCR)based approaches developed by Griffiths et al. (1998) and Kahn et al. (1998). Both banded and unbanded individuals can be individually identified by unique markings of black and yellow on the bill, eye rings and legs as well as by some unique plumage characteristics (Skutch 1935, 1960; Lawton & Lawton 1985; Williams et al. 1994). DNA fingerprinting techniques and parentage assignment methods have previously been described in detail (Williams 2000, 2004). As in Williams (2004), a helper was categorized as related or unrelated to the nestlings it was feeding if the helper’s average band-sharing score with all the nestlings in a brood was above or below 0.40. We also scored helpers as related to a brood if at least one pairwise band-sharing value fell above 0.40 (i.e. they were closely related to at least one nestling) and unrelated if none of the pairwise band-sharing scores fell above 0.40. 536


Behavioral Observations

During the breeding season (Feb. to Jun.) from 1994 to 1996, nests were watched for 2-h periods in the morning between 07:00 and 11:00 hours when the nestlings reached at least 14 d of age until fledging (x SE ¼ 25 0.43 d of age at fledging, n ¼ 36 broods). Previous observations of this population indicated that nestling feeding rates reached a peak around 14 d of age (D. Williams, unpubl. data). This resulted in a total of 340 h of observation and 1975 feeding visits for 30 group-years. During the observation period, we recorded which individuals fed nestlings and the amount of food brought to the nest using the index of Stallcup & Woolfenden (1978). The food brought in a single visit was given a size 1– 3, which was estimated by the size of the food item relative to the bill and the extent to which the throat pouch was distended. All the values were then summed for an individual to give the total amount of food it brought to the nestlings. It was only possible to identify the type of food brought to the nest for 765 visits (39% of 1975 total visits). Brown jays fed the young 73% arthropods and small vertebrates and 27% fruit. We also quantified an estimate of vigilance (the amount of time individuals spent within 20 m of the nest tree after feeding nestlings). Analyses of Nestling Care

For the analyses of individual helping effort, we only included helpers of known sex, age, and immigration status (total of 114 individuals and 159 individual-years) in order to adequately control for these variables in the models (see below). We did not include immigrant female helpers in our analyses because we only knew of three such individuals (representing 5 individual-years) in the groups for which we had observations of nestling care. We included five male helpers known to father young in this dataset because their behavior did not differ from other male helpers (D. Williams & A. Hale, unpubl. data). Most (58/68) of the remaining male helpers included in this dataset were excluded from parentage using DNA fingerprinting (Williams 2004). Only nests that were watched for at least four observation periods (8 h) were included in these analyses. Nestling age (days) was a potential confounding variable in the analyses of feeding rates; however, feeding rates did not vary significantly with nestling age within the range studied and observations were made over similar stages of nestling development for Ethology 113 (2007) 534–542 ª 2007 The Authors Journal compilation ª 2007 Blackwell Verlag, Berlin



all helpers (Williams 2000). Each individual was assigned an overall feeding rate (number of visits per hour) and amount of food (sum of food values per hour) for a brood. Because the total amount of food delivered per hour was significantly correlated with the number of visits to the nest per hour (r ¼ 0.97, p < 0.001), we chose the number of visits to the nest per hour as our measure of nestling feedings. Statistical Analysis

The datasets we present include individuals and groups that were repeatedly sampled across years. To control for repeated sampling with an unbalanced design, we used the method of residual maximum likelihood (REML) for analysis of linear mixed models. REML mixed models are equivalent to general linear regression except that both fixed and random effects are defined. The variance components are first estimated using residual maximum likelihood, and then the variance parameter estimates are used to form the generalized least squares estimates of the treatment effects. We included all non-collinear (non-aliased) explanatory variables in the maximal regression models and dropped terms sequentially until the models included only terms whose elimination would significantly decrease the explanatory power of the model. The significance of terms was tested using the Wald statistic, which is distributed as chi-square. All interactions between variables were examined and included in the final model where appropriate. We also tested the inclusion of group identity, individual helper identity and helper identity nested within group identity as random terms in each fitted model. The significance of the random effects was assessed by examining the variance component for each level of the random effect. If the variance component was negative, we eliminated the random effect from the final fitted model. Models were checked with diagnostic plots and variables were transformed as needed (arcsinesquare root for proportions and natural log(x + 1) for rates). Transformation did not correct for nonnormality and constancy of error terms for the proportion of time spent in the nest area; thus, the data were rank-transformed before using parametric statistical analyses (Conover & Iman 1981). Mean values and standard errors are presented unless otherwise noted. All tests were two-tailed and statistical analyses were performed using Genstat 6.1 (VSN International Ltd., Hemel Hempstead, UK) and Minitab 13.32 (Minitab Inc., State College, PA, USA). Ethology 113 (2007) 534–542 ª 2007 The Authors Journal compilation ª 2007 Blackwell Verlag, Berlin

Contributions to nestling care

To determine the influence of brood size on measures of nestling care, we analyzed data for 279 individual-years (176 individuals of known and unknown age, sex, and breeding and immigration status) at 30 nests. The helper identity nested within group identity was included as a random effect in the models for measures of nestling care (estimates of the variances SE were as follows: visits – 0.017 0.008; time in nest area – 556 560). Brood size ranged from two to five nestlings and individuals contributed more food to large broods than small broods (REML: v23 ¼ 18.78, p < 0.001). As expected, we did not detect an effect of brood size on the proportion of time spent in the nest area by individual jays (v23 ¼ 0.80, p ¼ 0.85). Therefore, we included brood size as a fixed effect in all subsequent analyses of nestling feeding rates. In the models of individual contributions to nestling care, the response variables included nestling feeding rate and time spent in the nest area. Fixed effects in these models included: year, sex, age category (young: 1 yr old; intermediate: 2–3 yr old; and old: ‡4 yr old), natal or immigrant status of the individual helper, number of feeders in the group, brood size, and relatedness of helpers to nestlings. Helper identity nested within group identity was included as a random effect in these REML models. Load-lightening

We tested for a relationship between the feeding rates of helpers, primary females and consorts within each social group and between the number of helpers and the feeding rates of primary females and consorts. Because there was no significant effect of group, individual helper identity, or helper identity nested within group identity as random terms, we present these results using generalized linear models (GLM). Results Helper contributions to nestling care

Each nestling was fed on average 1.8 0.13 (SE) times per hour by all group members (n ¼ 30 group-years). Brown jay helpers contributed a large proportion (71%) of the total nestling feedings in this population (Fig. 1). Most helpers (74%) of known sex and origin were individuals residing on their natal territories (n ¼ 41 natal females, 45 natal 537



Table 1: Results from a mixed model (REML) of nestling feeding rates by brown jay helpers of known sex and immigration status

0.9

Mean prop of nestling feedings

0.8

227

0.7

Effect

d.f.

Wald statistic (v2)

p-value

0.6

Year Brood size Sex

2 3 1

7.92 36.21 7.35

0.019 0.40 to at least one nestling in the nest) and unrelated (bandsharing 0.6 in all cases). Female helpers fed nestlings at higher rates than male helpers (Table 1, Fig. 3), and there was no effect of age on provisioning rates for helpers of 538

2

3

4

5

6

Brood size

20

Fig. 3: Relationship between mean nestling provisioning rates (visits per hour) and brood size for natal female helpers and immigrant and natal male helpers. The horizontal bars (females: black; males: gray) show predicted values from the model (Table 1). The open symbols show mean values from the data. The number of female helper-years for brood sizes 2, 3, 4, and 5 are 9, 17, 21, and 7, respectively. The number of male helper-years for brood sizes 2, 3, 4, and 5 are 18, 36, 41, and 10, respectively Table 2: Results from a mixed model (REML) of ranked proportion of time spent in the nest area by brown jay helpers of known sex and immigration status Effect

d.f.

Wald statistic (v2)

p-value

Sex Age category

1 2

8.95 15.56

0.003