Does cooperation increase helpers' later success as breeders? A test of the skills hypothesis in the cooperatively displaying lance-tailed manakin


Correspondence author. E-mail:


  1. Experience improves individual performance in many tasks. Pre-breeding cooperation may provide important experience that improves later success as a breeder, offering one compelling explanation for why some individuals delay reproduction to help others breed (the ‘skills hypothesis’). However, confounding effects of age, quality and alternative selective benefits have complicated rigorous tests of this hypothesis.
  2. Male lance-tailed manakins perform cooperative courtship displays involving partnerships between unrelated alpha and beta males, and alphas monopolize resulting copulations. Beta males therefore do not receive immediate direct or indirect fitness benefits, but may gain skills during cooperation that increase their later success as an alpha. To date, however, the effect of cooperative experience on later success as a breeder has never been tested in any cooperatively displaying taxon.
  3. The effects of prior cooperative experience on reproductive success of alpha lance-tailed manakins were analysed in a mixed model framework using 12 years of information on cooperative experience and annual and lifetime genetic reproductive success for 57 alpha males. Models included previously identified effects of age and alpha tenure. Individual-level random effects controlled for quality differences to test for an independent influence of beta experience on success.
  4. Males accumulated up to 5 years of beta experience before becoming alphas, but 42·1% of alphas had no prior beta experience. Betas became alphas later in life, and experienced significantly lower reproductive success in their final year as alpha than males that were never beta, but did not have higher lifetime success or longer alpha tenures. Differences in patterns of annual siring success were best explained by age-dependent patterns of reproductive improvement and senescence among alphas, not beta experience.
  5. Cooperative experience does not increase relative breeding success for male lance-tailed manakins. Importantly, beta cooperation seems to be an alternative reproductive tactic that yields fitness payoffs equivalent to a non-cooperative route to alpha status, if population growth rate is stable.


Prior experience leads to improved performance in many aspects of vertebrate social systems, including offspring care (Weladji et al. 2006; Angelier et al. 2007), courtship (Freeberg 2000), foraging (Krebs & Inman 1992), and defence against predators (McLean, Lundie-Jenkins & Jarman 1996). Most individuals obtain the experience necessary for improved performance through personal experience, but animals may also improve performance through observation or by participating as a ‘helper’ in critical activities (Komdeur 1996; Kendal et al. 2005). Cooperative reproduction is a situation in which some individuals forgo breeding and instead assist in the reproductive efforts of others (Brown 1987; Koenig & Dickinson 2004). Cooperative reproduction has been documented at all stages of the reproductive process, including territory defence (Sherman 1995), courtship (Foster 1977; Krakauer 2005), nest construction (Wilson & Hölldobler 1980) and offspring care (Skutch 1961; Brown 1987; Koenig & Dickinson 2004). Studies of ‘cooperative breeding’ or ‘helping at the nest’, in which helpers assist with offspring care, have figured most prominently in the investigation of cooperation (Koenig & Dickinson 2004). However, the hypotheses developed from cooperative breeding systems also are directly applicable to other types of reproductive cooperation.

In general, fitness benefits that select for cooperative behaviour could come about through three major mechanisms (Brown 1987): immediate direct benefits if cooperators receive some share of reproduction during their cooperative tenures (Nonacs & Hager 2011), indirect fitness benefits if they help closely related breeders and pass on their genes indirectly (Hamilton 1964), or delayed direct benefits if they improve their chances of future reproductive success (Wiley & Rabenold 1984). Of the three, quantifying the delayed direct benefits of cooperative behaviour has proven particularly challenging as it requires longitudinal data on individuals and may be a less powerful selective force than immediate direct or indirect benefits when more than one class of benefits are present simultaneously. Delayed direct benefits may come from increases in survival probability, likelihood of attaining a breeding position, or in success once a breeding position is attained. The ‘skills hypothesis’ addresses this latter possibility, and proposes that cooperating individuals benefit by gaining skills or experiences that improve their later reproductive success (Skutch 1961; Selander 1965). To date, this hypothesis has not been tested in types of cooperative reproduction other than cooperative breeding.

Even in cooperative breeding systems, empirical support for the skills hypothesis remains scarce. A comparison of the fledging success of same-aged novice breeders with and without helping experience revealed no detectable differences for red-cockaded woodpeckers Picoides borealis; (Khan & Walters 1997) or white-fronted bee-eaters Merops bullockoides; (Emlen & Wrege 1988). Provisioning rates as a helper were unrelated to later reproductive success in the acorn woodpecker Melanerpes formicivorus; (Koenig & Walters 2011). In Western bluebirds Sialia mexicana, individuals that had been helpers actually fledged fewer chicks than non-helpers in their first year as breeders (Dickinson, Koenig & Pitelka 1996).

Support for the skills hypothesis comes from two key studies of cooperative breeders. Seychelles warblers Acrocephalus sechellensis with helping experience outperformed those without helping experience following translocation to an unoccupied island (Komdeur 1996). Similarly, long-tailed tits Aegithalos caudatus that helped in years when their own breeding attempt failed changed their nest construction behaviour and increased their success in the following year (Hatchwell et al. 1999). However, findings from both of these studies have been recently called into question as new molecular and morphological analyses revealed unexpected effects of direct reproduction and individual quality that could influence helper success (Richardson, Burke & Komdeur 2002; Meade & Hatchwell 2010). Analyses of genetic paternity suggest that Seychelles warbler helpers frequently may be co-breeders (Richardson, Burke & Komdeur 2002), and so may gain direct as well as indirect breeding experience before breeding independently. Likewise, only long-tailed tits in relatively good physical condition become helpers following a nest failure, raising the possibility that differences in individual quality rather than experience gained while helping could explain the previous results (Meade & Hatchwell 2010).

To date, tests of the skills hypothesis have focused on species that experience a high degree of indirect fitness benefits from cooperation. However, evidence for benefits of delayed direct breeding is more likely to be found in species in which helpers are not close relatives and so do not receive additional (immediate direct or indirect) benefits from cooperation (Koenig & Walters 2011). The cooperatively displaying lance-tailed manakin Chiroxiphia lanceolata offers an excellent opportunity to investigate the role of cooperative experience in influencing success as a breeder. Male lance-tailed manakins display for females in a lek mating system, and form long-term male-male partnerships to cooperate in complex courtship displays (DuVal 2007b). In these partnerships, dominant alpha breeders are assisted by subordinate beta helpers in both song and dance displays. Previous work in this system has shown that male partners are not close relatives and that, with extremely rare exceptions, only alpha males mate (DuVal 2007a; DuVal & Kempenaers 2008). Importantly, this excludes confounding effects from immediate direct or indirect benefits of helping which have complicated tests of the skills hypothesis in other species. Alpha males increase their average siring success with each year of alpha tenure, and this effect is distinct from an independent influence of individual age on reproduction (DuVal 2012). The positive influence of alpha experience on reproductive success suggests that cooperation as a non-breeding beta might provide an alternative way to gain such experience. At advanced ages, reproductive gains from increased experience are offset by senescence in siring success (DuVal 2012). Although most alphas have beta partners (DuVal 2007c), not all males serve as betas before attaining alpha status. Though previous work has suggested that beta experience increases males' chances of becoming an alpha (DuVal 2007a), it remains unclear whether and how beta experience affects success once males attain alpha status.

To investigate how experience as a cooperating beta male affects realized success as a breeder, this study combined 11 consecutive years of information on individual social status, age, and genetic siring success from a longitudinal study of a colour-marked population of lance-tailed manakins. Analysis of the influence of cooperative experience on later breeding success can be confounded by variation in success due to age, territory quality, and individual quality. For example, helpers in most species are older than individuals that do not help (Brown 1987). I therefore used generalized linear mixed models with age as a covariate and individual identity as a random effect to examine the role of previous beta experience in determining annual and lifetime success as an alpha male, separate from an influence of individual age and controlling for variation among individuals over time (van de Pol & Verhulst 2006). Previous tests of the skills hypothesis generally have focused on the influence of cooperative experience on success during individuals' first breeding season, rather than on lifetime success, and may therefore fail to detect longer term benefits (Komdeur 1996; Koenig & Walters 2011). To address this possibility, this study examined both annual and lifetime reproductive success of alpha male lance-tailed manakins in relation to their cooperative history. Finally, because alpha lance-tailed manakins exhibit senescence in siring success and early display performance is associated with decreased success later in life in another lekking species (Preston et al. 2011), analyses considered the possibility that cooperating males experience changes in reproductive potential early or late in their reproductive tenures.

Materials and methods

Study species and field techniques

Lance-tailed manakins are small (ca. 15·5–22 g) Neotropical passerines. Like most manakins (family Pipridae), they are sexually dimorphic, and males perform elaborate courtship displays at traditional display sites. In this species, male display areas are separated by approximately 50 m (an ‘exploded lek'), so that they are in auditory but not visual contact (DuVal 2007b). Females fly among male display areas to observe courtship displays and choose a mate, but nest outside mates' territories (DuVal & Kempenaers 2008). Females receive no assistance in rearing their offspring. Males attain adult plumage in their fourth year after hatching, and three age-specific plumages before adulthood allow unambiguous age determination for any male captured before his fourth year (DuVal 2005).

Male lance-tailed manakins form long-term male-male partnerships and display cooperatively for females, similar to other Chiroxiphia manakins (Snow 1971; Foster 1977, 1981; McDonald 1989). Males cooperate in duet songs, sung by two males perched side-by-side in a tall tree, and in two-male courtship dances. Social status was determined following previously established criteria (DuVal 2007c). Alphas were distinguished by consistent presence at display areas, the performance of solo attraction (‘pip flight’) and courtship displays for females, and by alpha-specific components of cooperative courtship displays (e.g. the piercing, metallic eek call that ends a bout of two-male leapfrog dancing for females, and butterfly-like slow flight displays that continue after the departure of the subordinate partner). Beta males were identified as alphas' most common duet partners, and the only males to participate with alphas in multi-male displays for females. Although betas may eventually become alphas, dominance relationships within an alliance were never reversed (i.e. alphas did not become betas). Adult males that are neither alphas nor betas are referred to here as ‘non-territorial’ males. Each display area was occupied by one alpha and generally one beta male, and a variable number of non-territorial associates that did not participate in courtship displays (DuVal 2007c).

Study site

The study was conducted on a 46 ha area of secondary growth dry tropical forest on the eastern tip of Isla Boca Brava, a 3000 ha island in the Gulf of Chiriquí, Republic of Panamá. Data considered here were collected between 2000 and 2011, with survival information from 2012 used to identify the end of alpha tenure for two individuals. On average there were 27 active lance-tailed manakin display areas on this study site each year, each occupied by one alpha and various other adult males (DuVal 2007c). Hour-long behavioural observations were conducted three times weekly at a core group of 12–18 male display areas during each breeding season to identify males active in the area, and to quantify rates of duet singing and dance displays performed for females (in 2000–2003, observations were 2 h periods twice a week; 9063 h of observation in 11 years). All additional display areas were observed for at least 10 h to establish social status of individuals present.

Cooperative experience

Cooperative experience was quantified as number of years (breeding seasons) in the beta role, and tenure as alpha or beta was years that a male held that status. For the full duration of a male's beta tenure to be known, the male had to first be observed as neither alpha nor beta in the study population (n = 51 individuals with completely observed alpha tenures, and 14 males alive at the study's end with partial alpha tenures observed). Analyses that considered beta experience as a binary value (ever vs. never observed as beta) included all males for which status was identified in the year prior to attaining alpha status (n = 57 males). This approach allows accurate assessment of beta experience as only one beta that ascended to alpha status did not do so directly after his beta tenure. This unusual male was observed as beta for 1 year, then classified non-territorial for 2 years before becoming alpha, and is treated here as having 1 year of beta experience. Beta partners were identified for nearly all alphas (see DuVal 2007c for information on solo alpha males), so it is unlikely that betas were incorrectly classified as non-territorial individuals. Reliable interactions were necessary to establish beta status and so very short beta tenures (< 1 month) may have gone undetected. Males first banded as alpha individuals were excluded from all analyses, as their prior beta experience was not known. Alpha tenures ended when males disappeared from the study population and were presumed dead. Disappearances are unlikely to be dispersal events because alphas usually maintained the same territory throughout their tenure, and were never assigned to offspring sired after they disappeared.

Reproductive success

Reproductive success was quantified as genetic siring success. Lifetime reproductive success was the sum of the scaled yearly siring values for an individual across his full breeding tenure. A scaling factor is necessary because researcher success at finding manakin nests improved over the course of this study. Because more chicks were sampled in later years, raw siring values could overestimate the relative success of individuals breeding later in the study. To correct for this possible bias, number of chicks sired was standardized within each year as the number of chicks sired divided by the number of chicks assigned to any male that year, and multiplied by the average number of chicks assigned in 2005–2009 (years in which sampling success was high and consistent; average = 95·6 ± 5·6 chicks). The 2004 field season was particularly short with few nests sampled, so applying this correction to that year's data likely overestimates success of males with reproduction detected that year. Values from 2004 were corrected further by subtracting three chicks, so that the mean number of chicks sired by successful males in that year was comparable to other years of the study. This adjustment affected the annual success measures of four individuals with known beta histories (one that sired chicks in 2004 and three that did not). When data from 2004 were excluded from analyses, results were unchanged.

Paternity was assessed for 1110 chicks from 674 different nests, representing reproductive output of 258 individual females. Females lay no more than two eggs per nest. Genetic techniques and details of assignment success are described in detail elsewhere (DuVal & Kempenaers 2008; DuVal 2012). In short, genomic DNA was extracted from blood or tissue samples of adults, chicks, and unhatched eggs and used to amplify 20 variable microsatellite loci. Paternity was assigned using maternal identity observed directly from female behaviour at the nest (and genetically confirmed), or parental pair assessment in CERVUS 3·0 when maternal information was unavailable (Kalinowski, Taper & Marshall 2007). Candidate sires were all adult-plumaged males alive in that year, regardless of social status. The proportion of candidates sampled was estimated in each year as the number of identified adult males present in that year, divided by that number plus the number of unbanded adult males captured in the following year (0·83 in 2000; 0·95–0·99 in other years). Chicks were assigned paternity only when they matched a sire with ≥ 95% confidence. Confidence values represent the difference in log-likelihood scores of the two most likely candidate sires, based on 10 000 simulation cycles and given locus-specific population allele frequencies. Using these criteria, 89·9% of sampled offspring were assigned to a known male in the study population.

Statistical analyses

Analyses were conducted in the Program R 2·15·2 (R_Development_Core_Team 2011), and all means are reported ± s.d. All tests are two-tailed unless otherwise noted. Analyses were restricted to alpha males because (i) the goal of this study was to investigate the effects of beta experience on success as an alpha and (ii) previous work has demonstrated that reproduction is essentially limited to alpha males in the study population (DuVal 2007a; DuVal & Kempenaers 2008). Not all alphas sired offspring in a given year, and unsuccessful alphas are also included in all analyses.

Initial comparisons used two-tailed tests (t-tests and Wilcoxon Rank Sums tests) to assess whether males' cooperative experience was related to age and siring success in the first and last year of alpha tenure. Males' first and last years as breeders are candidate time periods in which reproductive differences between individuals may be most apparent. Individuals in a range of species have decreased or highly variable breeding success in their first breeding effort (Curio 1983; Rebke et al. 2010) and others suggest effects of cooperative experience may be most pronounced in the initial breeding effort (Komdeur 1996; Koenig & Walters 2011). The final year of reproductive tenure is of interest because trade-offs between early courtship display and accelerated senescence have been documented in another lekking species (Preston et al. 2011), and lance-tailed manakins exhibit senescence in siring success with age (DuVal 2012). Betas participate in extended periods of intense acrobatic courtship display before becoming alphas, and therefore may experience different patterns of senescence than non-beta males.

Changes in social status normally occurred between field seasons, but in nine of 32 cases alpha males in their final year as alpha were known or suspected to have died before the end of the breeding season (two with remains found after predation and identity confirmed genetically; seven absent from display areas and never resighted). These males were included in analyses of annual reproductive success (ARS) because they were present in the study population during the first ca. 2 months of the observed breeding seasons, when the majority of matings occurred, and excluding these records did not influence results.

To examine whether differences in patterns of ARS between alphas with and without beta experience could be explained by different life-history trajectories rather than beta experience per se, a series of biological hypotheses was tested using generalized linear mixed models (GLMMs) and model support assessed using AICc, the small sample modification of AIC, including one d.f. for random effect (Vaida & Blanchard 2005). Key variables previously identified as influential for annual success of alpha males were individual age, the quadratic term for age, years of experience in the alpha role, and a binary variable indicating whether it was the male's last year of life (DuVal 2012). Tested models therefore assessed (i) whether beta experience was related to ARS (i.e. as a simple main effect); (ii) whether beta experience was related to ARS given male age and alpha tenure; (iii) whether the relationship of age or alpha tenure with ARS varied for males with different levels of beta experience and (iv) whether males with beta experience had different levels of success in critical time periods (i.e. their first and/or last year as alpha), separate from effects of age and alpha tenure. Pairwise correlation coefficients revealed collinearity between the binary first-year term and the continuous term for years of alpha tenure (r = −0·7) which precluded simultaneous consideration of these two variables. Models specifically testing for first-year effects (i.e. 4a and 4c in Table 1) therefore exclude the tenure term. Correlations among other fixed factors in GLMM analyses were low to moderate with the exception of the expected relationship between age and the quadratic term for age (Schielzeth 2010). Two additional models were considered to provide a context for the major comparisons of beta experience: the model previously identified as the best-fit model explaining the relationship of life-history variables to alpha male ARS, without reference to beta history (DuVal 2012), and a model including only the random effect. Models were run separately for binary and continuous measures of beta experience, and all included individual identity as a random effect. As males usually remained at the same display territory throughout their tenure as alphas, this random effect variable includes variation due to both individual quality (e.g. genetic or environmental differences) and display territory quality. GLMMs were conducted with glmmadmb in package glmmADMB, using a zero-inflated negative binomial error distribution (with variance = ϕμ)(Fournier et al. 2012; Skaug et al. 2012). Strong support for a particular model was inferred when the AICc value was at least two AICc units less than the next-best model (Burnham & Anderson 2002). Confidence intervals for parameter estimates of the best-fit model were generated from the 2·5th and 97·5th percentiles of unmanipulated parameter estimates from 10 000 nonparametric bootstraps, resampling cases in the repeated-measures model (Hox 2002). Analyses of ARS included yearly data from all alpha males of known age for which complete beta tenure was quantified, including those still alive at the end of the study.

Table 1. Models evaluating the relationship of past beta experience with alpha male annual reproductive success. Nine models tested four explicit hypotheses about the relationship of beta experience with alpha annual reproductive success (ARS), and two additional models for comparison assessed first the influence of alpha age, tenure, and final year without beta status (i.e. the best-fit model from DuVal 2012), and second a constant term to assess fit of the random effect alone. Reported metrics include log-likelihood values, the estimated number of parameters (k), and AICc values. Beta experience (B) was parameterized as (a) a continuous variable (years as beta prior to alpha status) and (b) a binary variable (any beta experience), and models for each parameterization are ranked by ΔAICc, the difference in AICc between each model and the best-fit model. AICc weights (wi) indicate the relative support for each model compared with other models in the tested set. Additional parameters considered were individual age (A); the quadratic term for age (A2); current year of alpha tenure (T); and binary variables indicating whether it was the male's initial (I) and/or final (F) year of alpha status. Models 4a and 4c did not include T because of collinearity with I (see methods). Asterisks denote interactions. Models in bold had approximately equivalent support based on AICc values. Data were 122 records of annual reproductive success for 48 individuals. All models were run in the R package glmmADMB using a zero-inflated negative binomial distribution and including male ID as a random effect
Hypothesis testedModel(a) Years as beta(b) Binary beta experience
Rank−logLik k AICcΔ AICc w i Rank−logLik k AICcΔ AICc w i
1) Beta experience is directly related to ARS
a. B11−233·15476·727·4<0·00111−233·25476·929·2<0·001
2) Beta experience is related to ARS when age and alpha tenure are also considered
a. A + A2 + T + B4−217·88452·83·5·074−217·78452·75·0 0·04
3) Former betas experience different relationships of age or alpha tenure with ARS than do males that were never betas
a. A + A2 +T × B5−217·39454·24·9·036−217·59454·77·0 0·01
b. A + A2 × B + T7−217·59454·75·4·037−217·69454·87·1 0·01
c. A × B + A2 + T6−217·59454·65·3·035−217·59454·66·9 0·02
d. A × B + A2 × B + T9−217·310456·57·2·018−216·710455·47·7 0·005
4) Success of former betas varies in critical time periods, separate from age and alpha tenure effects
a. A + A2 + I × B8−218·29455·96·7·019−218·09455·57·8 ·01
b. A + A2 + T + F × B 1 213·6 10 449·3 0 ·40 1 212·9 10 447·7 0 ·49
c. A + A2 + I × B + F × B3−213·311451·52·2·13 2 212·3 11 449·1 1·4 ·25
The relationship of alpha age, tenure and reproductive stage with ARS when beta experience is not considered
A + A2 + T + F 2 216·3 8 450·0 0·7 ·28 3−216·38450·02·3 ·16
Constant term only

The relationship of beta tenure with lifetime reproductive success (LRS) was first examined using generalized additive models (GAM) with a quasi-Poisson error structure and a user-defined k corresponding with the number of discrete alpha or beta tenure lengths assessed. Measures of LRS in this population did not conform to any standard sampling distribution, necessitating the use of a nonparametric approach. GAMs are nonparametric spline-fitting functions, with confidence estimates produced through Bayesian likelihood procedures, and can be employed to identify nonlinear effects of independent variables without requiring the user to define a hypothesized functional form for the analysis (Wood 2006). GAMs were conducted in package mgcv in the Program R (Wood 2011). Subsequent two-sample comparisons assessed the longevity, years of alpha tenure and LRS of alpha males with and without beta experience.

Analyses of LRS were restricted to males whose complete alpha and betas histories were known (i.e. both the male's social status prior to becoming alpha, and the end of that male's alpha tenure had been recorded). LRS analyses included males of unknown age because males must be observed as subadults to determine exact age, but were often present in the population as adult-plumaged birds for many years before they attained either beta or alpha status. Reproduction was limited to alpha individuals, and males generally did not regress in social status once they attained alpha status. Exceptions to this were two males that regressed to beta status after holding alpha status for one season. Their beta experience was classified as years as a beta prior to their time as alpha. Restricting analyses to males whose complete alpha and beta tenures had been observed may introduce a bias if males excluded because they were alive at the end of the study period were particularly long-lived or otherwise differed from males included in the analysis. Analyses of LRS excluded 14 alpha males in their second through their ninth year as alpha in 2011 (4·14 ± 2·31 years observed as alpha; 51 males with quantified LRS and fully observed beta tenures averaged 2·53 ± 1·58 years as alpha). These surviving alphas were slightly older than males for which complete lifetime success was quantified: 12 were of known age and were in their 6th through 14th year after hatching (10·25 ± 2·56th year after hatching, while the 31 known-aged males with quantified LRS lived 8·64 ± 2·29 years after hatching). The 14 excluded males had 0–4 years of beta experience (1·80 ± 1·23 years of experience among excluded males with any beta history, 4 individuals without beta experience). To examine the influence of these exclusions on the GAM results, models were also run including these 14 individuals with cumulative reproductive success as the response variable, and years observed as alpha (rather than total alpha tenure) as an independent variable.


Variation in cooperative experience prior to alpha status

Of 57 alpha males for which complete alpha status was observed and beta history was known, 24 (42·1%) never held beta status. Alphas males that were never observed as betas moved directly to an alpha position from non-territorial status. Males that were beta before becoming alpha served an average of 2·1 ± 1·1 years in the beta role (n = 27 former betas) (Fig. 1).

Figure 1.

Alpha male lance-tailed manakins had 0–5 years of beta experience. One male observed as beta for 6 years before becoming alpha was excluded from this study because his start of beta tenure was not observed. Data are 51 males with years of alpha and beta tenure known.

Timing of alpha status and patterns of annual reproductive success

Alpha males that had been betas entered alpha status at significantly older ages than alpha males that had no beta experience (t = 4·50, P < 0·0001, n = 28 former betas, 20 without beta experience; Fig. 2a). Males with beta experience had higher levels of reproductive success in their first year as alphas (Wilcoxon Z = −2·04, P = 0·04, n = 28 former betas, 20 without beta experience; Fig. 3a), but did not differ significantly in siring success in their final year of alpha status (Wilcoxon Z = 1·70, P = 0·09, n = 17 former betas, 15 without beta experience; Fig. 3b).

Figure 2.

Males with prior beta experience became alphas at significantly older ages (a) did not differ in total years of alpha tenure, (b) and had longer lifespans, (c) than males that advanced to alpha status without first being betas. There was no difference in lifetime reproductive success of alpha males with and without prior beta experience and (d) Thick lines indicate median values, boxes enclose the first through third quartiles of the data, and whiskers indicate the values within 1·5 times the interquartile range. Data falling outside of these bounds are indicated by open circles, and asterisks denote significant differences. Sample sizes differ among comparisons because (a) and (c) include only males for which exact ages were determined by capture in a pre-definitive plumage, and (c) includes only known-aged males that had disappeared from the study population by the end of the study.

Figure 3.

When considered without information on male age, annual siring former betas was significantly higher than that of males without beta experience in the first year as alpha (a) but not significantly different in their final year and (b) Thick lines indicate median values, boxes enclose the first through third quartiles of the data, and whiskers indicate the values within 1·5 times the interquartile range. Data falling outside of these bounds are indicated by open circles, and asterisks denote significant differences.

The patterns reported above may be influenced by age differences between males with different cooperative histories rather than beta experience itself. To investigate whether the patterns in siring success reported above could be explained by variation in individual age or changes experienced during alpha tenure, I assessed the fit of nine GLMM models testing four general hypotheses about the relationship of beta experience to annual reproductive success of alphas (Table 1). AICc comparisons revealed strongest support for a model including an interaction of beta experience with a binary variable indicating the male's final year of life, in addition to variables describing male age and years of alpha experience (Table 1, model 4b). Parameter estimates for this model indicated that alpha males that were formerly betas sired fewer chicks in their final year of life than did males without beta experience (Table 2). Results were comparable when beta experience was quantified as a continuous or binary variable (Table 2). This model was not more strongly supported than a model excluding any term for beta experience when beta history was quantified as years of beta experience (Table 1, ΔAICc = 0·7), while the comparable model for a binary assessment of beta history had marginally stronger support (Table 1, ΔAICc = 2·3). A model that parameterized tenure using interactions of beta experience with binary terms for males' first and last years as alpha received approximately equivalent support to the best-fit model (Table 1, model 4c for binary beta experience). However, collinearity between the T and I terms prevented simultaneous consideration of these variables, and the similar log likelihoods of models 4b and 4c suggest that using additional variables to describe alpha tenure in 4c did not explain additional variance in the response. If that model was discarded, 4b received 65% of the Akaike weight in the model set with a binary parameterization of beta experience.

Table 2. Parameter estimates and bootstrapped confidence intervals for the best-fit model. Parameter estimates from the best-fit generalized linear mixed model describing the relationship of beta experience to annual reproductive success once males reach alpha status. Data were 122 records of annual reproductive success for 48 alpha individuals, analysed in a generalized linear mixed model with zero-inflated negative binomial error structure (glmmADBM, Program R). A random effect of male identity was included to account for repeated observations of individuals. Bootstrapped 95% confidence intervals for the fixed effects were generated by sampling 10 000 times with replacement to create data sets of 48 individuals' reproductive histories, including all sampled years for each selected individual. Bold type highlights significant terms. Results are shown for continuous and binary assessments of beta experience
Assessment of beta experienceExplanatory variablesEstimate + SE Z P 95% CI from bootstrap
ContinuousIntercept−4·83 ± 1·74 −2·77 0·006 −10·17 −2·09
Age1·06 ± 0·37 2·83 0·005 0·48 2·18
Age2−0·05 ± 0·02 −2·59 0·01 −0·12 −0·02
Year of alpha tenure0·27 ± 0·11 2·51 0·01 0·13 0·58
Years as beta0·03 ± 0·140·210·83−0·090·52
Final year (Y)0·11 ± 0·320·360·72−0·430·83
Years as beta × Final year (Y)−0·60 ± 0·29 −2·09 0·04 −1·64 −0·23
BinaryIntercept−4·47 ± 1·67 −2·68 0·007 −9·57 −2·00
Age0·95 ± 0·36 2·63 0·009 0·35 2·07
Age2−0·05 ± 0·02 −2·58 0·01 −0·11 −0·02
Year of alpha tenure0·29 ± 0·10 2·92 0·004 0·15 0·58
Ever beta (Y)0·42 ± 0·321·290·20−0·051·43
Final year (Y)0·40 ± 0·341·170·24−0·231·38
Ever beta (Y) × Final year (Y)−1·41 ± 0·50 −2·84 0·005 −2·84 −0·71

Longevity and patterns of lifetime reproductive success

There was no detectable difference in the length of alpha tenure of males with beta experience (Wilcoxon Z = 0·96, P = 0·33, n = 33 former betas, 24 without beta experience; Fig. 2b). In keeping with the observation that beta males became alpha later in life but held alpha statuses of comparable length to alphas that were not betas, alpha males with beta experience had significantly longer lives than males without beta experience (t = 2·18, P = 0·04, n = 17 former betas, and 14 without beta experience; Fig. 2c). Length of beta tenure was not correlated with alpha longevity (rs = 0·32, P = 0·08, n = 31) or years of alpha tenure (rs = −0·13, P = 0·36, n = 51).

Alphas with beta experience did not sire more chicks in their lifetimes than alphas that were never betas (Wilcoxon Z = −0·16, P = 0·87, n = 33 former betas, 24 without beta experience; Fig. 2d). Although some former betas achieved very high levels of lifetime siring success, variance in siring success was not significantly different between males with different cooperative histories (Levene's test equality of variances: F(1,55) = 3·77, P = 0·06). Among males that had been betas, there was no correlation between years as a beta and lifetime reproductive success (rs = 0·04, P = 0·80, n = 27 alphas with any beta experience and lifetime success measured). A generalized additive model revealed no significant relationship between years as a beta and lifetime reproductive success (Table 3, model 1). To consider potential biases introduced by the exclusion of alpha males still alive at the end of the study, a second GAM model investigated the role of beta experience on cumulative reproductive success with both years of beta tenure and years to date of alpha tenure as explanatory variables (i.e. males alive in 2012 were included in analyses). This model reconfirmed the strong relationship of alpha tenure with cumulative siring success, but showed no relationship of beta tenure (Table 3, model 2).

Table 3. GAM assessments of relationships between beta experience and lifetime or cumulative siring success. Beta tenure had no independent influence on lifetime reproductive success (model 1; 0·12% of deviance explained), and no significant influence on cumulative siring success when considered in a model with years of alpha tenure (model 2; 56·6% of deviance explained). Results shown are from two generalized additive models with quasi-Poisson error structure examining the relationship between lifetime (model 1) or cumulative (model 2) reproductive success and beta tenure. Data for model 1 came from 51 alpha males for which years of beta tenure and lifetime reproductive success were known. Data for model 2 came from 65 alpha males with known beta histories and included those still alive at the end of the study (i.e. the response variable was cumulative but not necessarily lifetime reproductive success). The nonparametric smoothing term is designated by an s before the smoothed variable
  Parameter estimates
1. Beta experienceInterceptEstimate (SE) t P
1·73 (0·16)11·03<0·0001
s(Beta tenure years)e.d.f. F P
2. Alpha and beta experienceInterceptEstimate (SE) t P
1·59 (0·14)11·77<0·0001
s(Alpha tenure years)e.d.f. F P
s(Beta tenure years)1·000·030·87


Contrary to the predictions of the skills hypothesis, beta cooperation did not result in higher success of cooperators once they attained alpha status compared with males that did not serve as cooperative subordinates. Cooperatively displaying lance-tailed manakins exhibited substantial variation in whether and for how long males participated as beta helpers before attaining alpha (breeding) status. Males that were betas became alphas later in life than males that skipped the beta step, and beta experience did not increase males' annual reproductive success when differences in age and time as an alpha were also considered. Former betas furthermore showed no significant differences in length of tenure once they became alpha, or in lifetime reproductive success. These findings suggest there are distinct but equally successful cooperative and non-cooperative routes to attaining alpha status in the study population.

Performance increases from cooperative experience are expected to be most evident early in individuals' tenures as breeders. For example, Seychelles Warblers with helping experience had greater success in their initial breeding attempt, but did not improve with additional breeding experience (Komdeur 1996). An apparent advantage in siring success observed for beta lance-tailed manakins in the first year of alpha tenure was explained by age differences between alphas with and without beta experience: males without beta experience attained alpha status at younger ages, and young alphas are relatively unattractive to females (DuVal 2012).

In contrast, unexpected negative effects of beta experience were observed in males' final year of alpha status. Although more males with beta experience were present as alphas at advanced ages (Fig. S1), the observed decrease in siring success exceeded senescence-related declines predicted by age (DuVal 2012), occurred across the range of male ages, and persisted when analyses were repeated with the subset of males no older than the oldest alphas without beta experience (< 12 years). Recent evidence in the lekking houbara bustard Chlamydotis undulata indicates that males exhibiting higher levels of display behaviour early in life later experience accelerated senescence (Preston et al. 2011). Lance-tailed manakins with beta experience participate in up to five breeding seasons of acrobatic courtship display before attaining alpha status, and so late-life declines in success may reflect trade-offs between early display performance and later success, or intensified senescence in this group of males. Although these results reject the hypothesis that cooperating lance-tailed manakins acquire skills as a direct result of helping that increase their later reproductive success, social interaction may still be critical for the learning of displays. Male lance-tailed manakins interact extensively in the absence of females, and display skills may be developed through alternate ‘practice’ interactions without the need for long-term cooperative alliances.

Why do some males serve as cooperatively displaying betas while others do not? Differences in individual quality or in the quality of available territories may influence whether males cooperate before becoming alphas. Helpers in several cooperatively breeding species apparently help when breeding in a given year is not possible (e.g. due to territorial availability) or not likely to be productive (Dickinson, Koenig & Pitelka 1996; Hatchwell et al. 2004). Variation in individual quality could influence cooperation if beta experience offers a means for low-quality males to improve later success, for example by increasing the quality of their display performance. Although there was no detectable increase in success of males with beta experience compared to non-beta males, individual males may have improved their performance relative to that expected if they had not been betas. Opportunities for obtaining display areas may also be an important factor influencing cooperation in this species. While previous work has established that lance-tailed manakins do not form strict queues for status at display areas (DuVal 2007a), associations with an alpha partner may create territorial opportunities through a combination of inheritance and alliances that allow new territories to be established near existing areas. It remains to be determined whether the territories of alphas that do not serve as betas are of poorer quality (i.e. have lower food resources or higher predator exposure) than those of alphas with beta experience.

This analysis examined the effects of beta experience on males' performance in the alpha role, but helpers in systems with cooperative reproduction may also gain delayed direct benefits from increased survival (Lehmann & Keller 2006) or increased success in obtaining a breeding position relative to non-cooperators (Wiley & Rabenold 1984). Beta cooperative behaviour may allow some individuals avoid costs of becoming alpha at a young age (e.g. physiological stress or mortality risk) by delaying entry into the breeding population, and these individuals may experience different reproductive trajectories than early recruits (Aubry et al. 2009). This study detected no difference in length of alpha tenures for males with and without beta experience, but former betas lived longer suggesting cooperative experience may influence males' likelihood of surviving until they become alphas. Furthermore, if survival is low for alphas, and expected alpha success is related to age, males may maximize reproductive opportunities by delaying alpha tenure until an age when the first years as alpha are expected to be most productive. Timing of reproduction in relation to population growth rate is an important determinant of evolutionary fitness to be considered in future analyses: earlier reproduction may be favoured despite equivalent annual or lifetime siring success in an expanding population.

Although patterns of reproductive success reported here seem typical for the majority of alphas in the study population, it remains possible that beta experience may contribute to the success of a minority of extremely long-tenured individuals. These analyses necessarily excluded males that were already alpha at the start of the field study, as beta history of these individuals was unknown. Finally, while this study found no significant difference in variance in lifetime reproductive success between alphas with and without beta experience, this comparison approached significance. Additional sampling may reveal greater variation in siring success for former betas. Beta status may be beneficial if it carries increased possibility of large reproductive payoffs, though such extreme success is very rare.

This study provides the first evidence that cooperative experience as a beta does not increase males' success in the alpha role. Instead, cooperation as a beta is an alternative route to reproductive success in lance-tailed manakins, and equivalent fitness benefits can be achieved by males that move directly into the breeding role without forming cooperative alliances. The data presented here demonstrate that the non-cooperative route to attaining breeding status is a viable and widespread strategy, with nearly half of males not serving as betas before becoming alphas (though most alphas have beta partners; DuVal 2007c). The genetic, social and ecological factors influencing whether males cooperate before becoming alpha are promising areas for future research.


Twelve dedicated volunteer field crews assisted in data collection. This project was supported by the National Science Foundation (IOS-0843334), and utilized long-term data collected with support from sources including the Max Planck Institute for Ornithology and the University of California, Berkeley. B. Inouye and D. Houle contributed useful advice on analyses. M. Jones, R. Sardell, C. Stahala, C. Vanderbilt, and two anonymous reviewers provided constructive comments on earlier drafts. F. Koehler and Y. Pinzon provided field site access. Staff of ANAM Panama and the Museo de Vertebrados, Universidad de Panama provided assistance with permits.