Breeder aggression does not predict current or future cooperative group formation in a cooperatively breeding bird

Abstract

In cooperatively breeding species, subordinates forgo reproduction to assist breeders in raising offspring. When cooperative breeding is facultative, breeders from the same population may differ in whether they are assisted by non-breeding helpers. Predation risk is a major source of nest failure and assistance during nest defense is often an overlooked, yet important, way helpers assist breeders. A breeder’s aggressive response to a nest predator could have important implications for whether they form cooperatively breeding groups. We investigated the hypothesis that breeder aggression towards a nest predator is related to current and future helper recruitment. We tested the prediction that less aggressive breeders were more likely to form cooperative groups, which could occur if these breeders benefit from helper assistance during nest defense. We also considered the possibility that more aggressive breeders were more likely to form cooperative groups. We assessed the effects of partnerships and tested whether aggression exhibited by breeding partners was correlated. We conducted this work in the facultative, cooperatively breeding brown-headed nuthatch (Sitta pusilla). We measured breeder aggression in response to a taxidermy model of a nest predator to determine whether breeders’ aggression correlated with their current or future helper recruitment. We found no evidence of a sex difference in aggression among breeders and aggression scores of breeding partners were not significantly correlated. Aggression scores for both breeding males and breeding females were unrelated to whether they formed cooperative groups in the current year. We followed most of the breeding males, though not breeding females, across years and found that breeding males’ aggression scores were unrelated to helper recruitment the following year. Our results suggest that breeders’ responses to nest predators are not related to cooperative group formation in this species and that males and females showed comparable levels of aggression towards a nest predator.

A major challenge in the study of cooperative behavior is understanding why individuals vary in their cooperative tendencies (Komdeur, 2006). In cooperatively breeding species, subordinates (i.e. “helpers”) forgo reproduction to assist breeders in raising offspring (Skutch, 1935; Skutch, 1961). In species where cooperative breeding is facultative, breeders within a single population often differ in whether they are assisted by a non-breeding helper (e.g. Emlen & Wrege, 1988; Mumme, 1992; Russell & Hatchwell, 2001; Cox & Slater, 2007; Dias et al., 2015). Previous studies have revealed that differences in an individual’s sex, age, relatedness to group members, and the conditions breeders experience while breeding may explain why individuals vary in cooperative behavior (e.g. Hamilton, 1964; Koenig et al., 1992; Komdeur, 2006; West et al., 2007). However, breeders that are similar in these characteristics (i.e. are the same age or found in comparable environments) regularly differ in whether they are assisted by a helper. Theoretical (Barta, 2016) and empirical studies (Le Vin et al., 2011) suggest individual variation in behaviors, like aggression, could influence the occurrence of cooperation, but how individual variation in behavioral traits relates to variation in cooperation remains unclear. Aggressive tendency is a well-documented behavioral type (Sih et al., 2004); individuals often differ in their aggressive response compared to others in the population and those differences can be consistent across time and context (e.g., Sih et al., 2004; Duckworth, 2006; Cain et al., 2011). Here we investigated whether variation in a breeder’s aggressiveness in response to a nest predator was related to whether the breeder formed cooperative breeding groups with nonbreeding helpers.

Research on cooperative breeding has often focused primarily on a single aspect of helping - offspring provisioning (Teunissen et al., 2020a). However, helpers may also provide critical assistance to breeders by participating in nest or territory defense (e.g. Feeney et al., 2013; Groenewoud et al., 2016). Predation is a major source of nest failure in many species (e.g., Martin, 1995; Innes & Johnston, 1996; Hidalgo Aranzamendi, 2017) and aggression towards a nest predator is an important behavior helpers perform in cooperative groups. In fact, cooperative nest defense could be one of the more important collective actions cooperative species perform (Shen et al., 2017). The degree of predation risk can also influence the formation of cooperative groups (Vehrencamp, 1978; Innes & Johnston, 1996; Groenewoud et al., 2016). Individuals within a population can differ consistently from one another in their responses to a predator (e.g., Vrublevska et al., 2014; Cain et al., 2011) and this variation could affect the likelihood that some individuals form cooperative groups. For example, breeders that are less aggressive when defending a nest may be more likely to breed in cooperative groups if helpers assist in nest defense (e.g. Arnold et al., 2005; Feeney et al., 2013; Groenewoud et al., 2016). Breeders that are highly aggressive may be less likely to form cooperative groups because they are able to defend their nest without assistance. Alternatively, aggressive breeders may be more likely to form cooperative groups if they have higher reproductive success and helpers are offspring from previous nests (e.g., Ekman et al., 2004; Preston et al., 2016). More aggressive breeders may also be more likely to form cooperative groups because helpers contribute to offspring care. In some avian species, more aggressive breeders contribute less to parental care behaviors (e.g. provisioning) than less aggressive breeders (e.g. Knapton and Falls, 1983; Kopachena and Fall, 1993; Mutzel et al., 2013) and spend more time engaged in agonistic interactions. For example, in blue tits (Cyanistes caeruleus), a species that displays bi-parental care, more aggressive males spent less time provisioning offspring compared to less aggressive males; females paired with more aggressive males provisioned young more often to compensate for their partner’s reduction in parental care (Mutzel et al., 2013). If aggressive males or females in cooperatively breeding species also contribute less to parental care, they could disproportionately benefit from the contribution of helpers that offset the reduction care provided. Considering how individual differences in breeder aggression relate to variation in helper recruitment could provide insights into why breeders in a single population vary in cooperative group formation.

In this study we investigated the relationship between breeders’ aggressive tendencies and cooperative group formation in a wild population of brown-headed nuthatches (Sitta pusilla). The brown-headed nuthatch is a small (~10g) passerine that exhibits facultative cooperative breeding and nests in mature pine forests of the southeastern United States. The study population was color-banded (~92% of individuals) and has been extensively monitored for over a decade (Cox & Slater, 2007; Haas et al., 2010; Han et al., 2015; Cusick et al., 2018; Cox et al., 2019). In this population, helpers are typically second-year males, but only a subset of males assist in their second year, while others attempt independent breeding. The occurrence of cooperative breeding also varies among breeders in this population: approximately 20–30% of breeding pairs are assisted by one or more helpers (Cox & Slater, 2007; Cox et al., 2012; Han et al., 2015; Cusick et al., 2018; Cox et al., 2019).

Variation in cooperative behavior among breeders could be affected by differences in aspects of territory quality (e.g. Pasinelli & Walters, 2002; Koenig et al., 2011), due to resource defense benefits (Shen et al., 2017), or variation in fledging success as helpers may be previous offspring (e.g. Ekman et al., 2004; Nelson-Flower et al., 2018). Previous studies in this nuthatch population have rejected these possibilities. Neither fledging success (Cusick et al., 2018) nor whether breeders were assisted by helpers (Cusick, 2019) were related to pine cover, a measure of territory quality. Similarly, whether breeders were assisted by helpers was not related to previous nesting success (Cusick, 2019). Some breeders were assisted by helpers to which they were not closely related (Han et al., 2015; Cusick, 2019) and some breeders were assisted by helpers they did not raise (Cusick, 2019). Brown-headed nuthatches are primary cavity nesters and often interact aggressively with nest predators, conspecifics, and secondary cavity nesting competitors, highlighting the importance of nest defense in the study population. Collectively, these previous findings, coupled with natural variation in cooperative breeding behavior, enabled us to investigate the possibility that breeder aggression related to helper recruitment.

With the goal of understanding the underexplored role of aggression in cooperative group formation, we investigated the hypothesis that breeder aggression towards a nest predator is related to current and future helper recruitment. We tested the prediction that less aggressive breeders would be more likely to form cooperative groups, which might be the case if these breeders benefit from helpers’ assistance during nest defense. We also tested the prediction that more aggressive breeders would be more likely to form cooperative groups, which could occur if breeders that differ in their aggressive response differ in their needs for assistance with offspring care. In addition, to assess potential effects of partnerships among male and female breeders, we tested whether the aggression exhibited by breeding partners was correlated. To do this, we quantified individual differences in breeder aggression in response to a nest predator. Assessments of breeder aggression occurred when late-stage young were present (i.e., nests had accrued high levels of adult investment). The predator model used posed a threat nestlings, but not to adults. These conditions should evoke a strong aggressive response from group members (e.g. Teunissen et al., 2020b).We assessed whether breeders’ aggression scores related to (1) differences in cooperative group formation in the year they were tested (i.e., whether they had a helper in the present year) and (2) variation in cooperative group formation in the following year. We further investigated whether the aggression exhibited by breeding partners was correlated.

Methods

Study Area and Study Population

The study was conducted February-May of 2015–2017 at Tall Timbers Research Station (30.66° N 84. 22° W; hereafter TTRS) in north Florida. TTRS comprises over 800 ha of mature pine forests (Crawford & Brueckheimer, 2012) and is dominated primarily by short leaf (Pinus echinata) and loblolly (P. taeda) pines. Brown-headed nuthatches excavate new cavities during each breeding season in decaying tree boles (hereafter “snags”, Cox & Slater, 2007). Artificial nest boxes were used by some breeding pairs during the study period as part of a concurrent study (Cox et al., 2019). A single nest box was placed near snags where excavation behavior had been observed and then removed if the box was not actually used. This avoided an artificial increase in territory numbers (Cox et al., 2019).

General Field Methods

Individuals were color-marked as nestlings (two plastic color bands on one leg and a single anodized federal band on the other leg) approximately 8–12 days post-hatching (nestling period lasts 18 days). Adults were captured using mist nests and marked similarly. Individuals were also measured and DNA samples were collected. We accessed nestlings in natural cavities by removing the face of the cavity and then repairing the cavity face using wood putty, duct tape, and staples (Ibarzabal & Tremblay, 2006; Cox & Slater, 2007; Cox et al., 2019). In contrast, nest boxes were easily accessible. Nest searching began in mid-February; potential nests sites were identified by observing excavation and nests were confirmed to be active by the presence of eggs and nestlings (see Cox & Slater, 2007; Cusick et al., 2018; Cox et al., 2019 for more details). Nests were visited regularly (i.e., every 2–4 days) and video recorded to confirm nest status (e.g. presence of eggs vs. nestlings) and to identify the adults tending the nest (Cusick et al., 2018).

Sex of marked individuals in our population was determined using DNA samples collected when individuals were banded (Han et al., 2015; Cox et al., 2019). Sex for unbanded birds was assigned by using sex-specific calls (Norris, 1958) and behaviors (e.g., only females incubate) that have been validated (≥95% accuracy) using DNA determinations (Han et al., 2015; Cox et al. 2019). The social status of individuals (breeders vs. helper) was confirmed using behavioral observations (e.g. breeders and helpers differ in hourly provisioning rates, cavity excavation, and other behaviors Cusick et al., 2018), sex or age (helpers are primarily males in their first year post-fledging, Haas et al., 2010; Cox & Slater, 2007, Cox et al., 2019). These methods have been corroborated using paternity assessments in previous studies (Han et al., 2015). “Cooperative nests” were tended by a breeding pair and at least one helper and “pair-only nests” were attended by just the breeding pair.

Aggression Trial Procedures

Aggression trials were conducted in 2015 and 2016 when nestlings were 10–12 days post-hatching. We quantified individuals’ aggression by measuring their response to freeze-dried specimens of a red-headed woodpecker (Melanerpes erythrocephalus), a nest predator of the brown-headed nuthatch (JACusick, pers. obs, Weiss, 1999, Figure 1a). We used two red-headed woodpecker specimens (hereafter “model”) in these experiments and randomly selected the model used for each trial. Each trial was video recorded using a high-definition video camera (Canon VIXIA HFM500 or HF20) mounted ≤8 m from the nest. Trained observers monitored the nuthatches’ responses during each trial using a spotting scope while sitting in a camouflaged blind approximately 7 m from the nest (Hunter’s Specialty Leaf Blind, Mossy Oak Infinity Camo; Gander Mountain, St. Paul, Minnesota). A detailed account of the events during the trial was also collected by the observer using a voice recorder (Olympus VN-5200PC or SONY ICRecorder), which was synchronized with the video camera.

Figure 1.

Set-up of the brown-headed nuthatch aggression trial. (A) Red-headed woodpecker model affixed below and to the right of the nest cavity entrance. (B) Camouflage cloth placed at the nest in the same orientation as the model two days before the trial to habituate nuthatches to the cloth. Here an individual fed offspring while the cloth was on the snag. This cloth was used to cover the model during the pre-exposure period of the aggression trial.

We conducted a “habituation period” prior to each trial to familiarize individuals with the camouflage cloth used to cover the model during the pre-exposure period of the aggression trial (Figure 1b). We stapled the camouflage cloth to the snag, directly below and to the right of the cavity opening and then watched the nest for 20 minutes to ensure that adults continued to provision offspring. If provisioning continued, the cloth was left on the snag until the onset of the aggression trial two days later. If adults did not resume normal provisioning within 20 minutes, we removed the cloth from the nest.

Aggression trials were documented using synchronous recordings of video cameras and voice recorders. After initiating recordings, we removed the camouflage cloth and secured the woodpecker model to the snag. A black zip-tie was placed underneath the feathers and wrapped around the middle of the body of the model and tightened so the excess zip-tie remained on the ventral side of the model. We then stapled the zip-tie directly to the snag so the model was directly below and to the right of the cavity entrance (Figure 1a). The model was covered with the camouflage cloth and a brown string was attached to the cloth using a safety pin (Figure 1b). To ensure the cloth remained in position, but could be easily removed by pulling the string, we loosely stapled the cloth to the snag. We then extended the string to the observation blind. Observers sat on the ground, covered themselves with the camouflage blind, and began the trial.

Trials were 30 minutes and consisted of (1) a 10-minute “pre-exposure period” where the model was hidden by the camouflage cloth, (2) a 10-minute “exposure period” where the model was exposed by removing the cloth, and (3) a 10-minute “post-exposure period” where the model was removed from the snag. During the pre-exposure period, we confirmed the adults were present based on their unique color bands and that they were provisioning nestlings or removing fecal sacs (Cusick et al., 2018). For a small number of trials (N=3) adults were not observed tending nestlings during the pre-exposure period. In these cases, we stopped the trial without exposing the model under the assumption that if adults were not providing care, they were not in the immediate area. We removed the model and replaced the camouflage cloth and returned the following day. In all three cases, the trial was successfully completed on the second day.

The exposure period began when the model was revealed. Observers documented the aggressive behaviors performed by individuals on the voice recorder (see Table 1 and Figure 2 for description of behaviors). Because of the synchrony among nests and limited number of camouflage cloths available, the cloth was not used on some nests. For all cases where the cloth was not used, the observer quickly approached the nest, stapled the model to the nest, and then returned to the blind and began the 10-minute exposure period. We found no significant difference in the maximum aggressive response recorded at nests where the cloth was used (3.91±2.09, N=11 nests) compared to the nests where it was not used (4.09±2.31, N=16 nests; Mann-Whitney U Test W=359, p=0.90, N= 27 nests). We also did not observe significant differences in aggression scores among breeding males and breeding females based on whether or not the cloth was used (breeding males: Mann-Whitney U Test W=121.5, p=0.08, breeding females: Mann-Whitney U Test W=64.5, p=0.21).

Table 1.

Brown-headed nuthatch aggression scale (modified from Duckworth 2006). Each individual was assigned an aggression score based on the frequency of defense behaviors displayed and proximity to the heterospecific model. Figure 2 provides visual examples of individuals assigned an aggression score of two (Figure 2a), three (Figure 2b), and six (Figure 2c). Individuals were assigned the highest observed score even when behaviors categorized as less aggressive were also performed.

Aggression Score	Qualitative Description	Quantitative Assessment of Aggression
1	No aggressive response	Either not present or was observed in pines, but no other vocalizations or behaviors observed
2	Low aggression	Remained in pine trees – did not approach, but vocalizations, wing venting¶ were observed
3	Moderate aggression	Proximity from model >1 model length Flew above snag 1–5 times and/or landed on snag 1–5 times
4	Moderate aggression	Proximity from model >1 model length Flew above snag ≥6 times and/or landed snag ≥6 times
5	High aggression	Proximity from model <1 model length Dived at model, but no contact 1–5 times and/or landed on snag 1–5x
6	Very high aggression	Proximity from model <1 model length Dived at model ≥6 times and/or, landed on snag ≥6 times and/or made physical contact with the model

^¶Wing venting is defined as rapid flapping or fluttering of the wings while perched

Figure 2.

Examples of the range of brown-headed nuthatch aggression scores in experimental trials. (A) The individual remained in the pines for the duration of the trial and received an aggression score of one; (B) the individual remained greater than one model length away from the model, landed on the snag less than six times, and received an aggression score of three; and (C) the individual made physical contact with the model and received an aggression score of six.

Following the 10-minute exposure period, the observer removed the model and returned to the blind and conducted post-exposure observations to confirm that adults returned to the nest. After 10 minutes of post-exposure observation, the observer approached the nest to remove the camera and the cloth.

Determining Breeders’ Cooperative Status in Year After Trial

To determine the breeders’ cooperative status the year following the trial, we located the marked individuals and monitored the nests they were tending the following year. Cooperative status in the following year was based again on the presence or absence of helpers. We only assessed cross-year data for breeding males because we did not locate many breeding females in the following year, probably due to mortality or dispersal.

All field protocols were approved by the animal care and use committees at TTRS (protocol VE2011–01-15–17) and Florida State University (ACUC 1207 and ACUC 1505). Our controls of sampling bias were in accordance with the STRANGE framework (Webster & Rutz, 2020). Individuals tested for this experiment were breeding male and female adults from a wild population. We tested individuals across our population site and tested for effects of concurrent studies (minimizing sampling bias associated with the following STRANGE categories: “Social Background,” “Rearing History,” Genetic Make-up”). We used multiple methods to capture individuals (e.g., as nestlings or with mist-nests as adults), which helped to minimize sampling bias (STRANGE category: “Trappability and Self-Selection”). Individuals were tested at the same stage in the breeding season (minimizing biases associated with STRANGE category: “Natural Changes in Responsiveness”), exposure to human observers and equipment was minimized, and individuals had time to habituate (minimizing biases associated with the following STRANGE categories: “Acclimation and Habituation” and “Trappability and Self-Selection”). Individuals had no experience with the models prior to the trials (minimizing biases due to the STRANGE category: “Experience”). Though individuals may have differed in their prior experience with the real nest predators, we attempted to control for this by sampling individuals across the study area.

Video Scoring

Video recordings of each trial were scored using SONY Movie Studio V.11 (Sony Creative Software, Inc.) coupled with the synchronized voice recordings. We analyzed 27 trials (N=7 nests 2015, N=20 nests 2016) that met the following conditions: breeder identity and group composition were known and individuals tending the nest were clearly observed during the exposure period. Of these 27 trials, 17 breeding males were observed in the following year, with known group composition, for which we tested how breeder aggression related to cooperative group formation in the year after they were tested for aggression. Analyzed trials represent a subset of the total trials conducted (24 in 2015 and 38 in 2016) as many trials did not satisfy the inclusion criteria.

We tallied the aggressive behaviors performed by each individual during the exposure period of the trial and assigned each individual a maximum aggression score (Table 1). Video recordings were scored by two trained researchers. To assess inter-rater reliability between the two reviewers, both researchers scored the responses of the breeding males and breeding females in eight trials. Observer reliability was assessed using weighted Cohen’s kappa (Hallgren, 2012) and we confirmed high reliability (Kappa=0.96, z=4.49, p<0.001, N=16 individuals).

Statistical Analyses

All statistical analyses were conducted in R version 4.0.2 (R Core Team, 2014) using the lme4 (Bates et al., 2014), car, and stats packages (R Core Team, 2014). We report means ± one standard deviation unless stated otherwise. We used a nonparametric Wilcoxon test to assess variation in breeding male and breeding female aggression scores because the scores were not normally distributed (Shapiro-Wilk’s Test W=0.74, p<0.001, N=54 individual scores) and did not fit a Poisson distribution. Breeding males and breeding females were included once, with the exception of two females that appeared twice because they had paired with a different male each year. There was no effect of removing those nests, therefore we kept them in the analyses. We tested whether aggression scores for the male and female of a breeding pair were correlated with one another to determine if more aggressive males were paired with more aggressive females. We calculated the correlation using the non-parametric Kendall’s test, which tests for an association between paired samples and works well with small sample sizes.

To determine whether the current cooperative status of the group related to the aggression score of the breeders, we ran a binomial generalized linear model (GLM; logit-link function) with breeders’ cooperative status as the response variable. We included male aggression score and female aggression score as the fixed effects and each nesting pair was included once. We detected no interaction between these two fixed effects (p=0.99). We tested for multicollinearity among predictor variables and found they were not strongly correlated (VIF = 1.03). We also tested whether breeders’ aggression scores were related to brood size at the time of trial (a measure of parental investment) using a non-parametric Spearman correlation.

We used a binomial GLM (logit-link function) to determine whether a breeding male’s aggression score related to its cooperative status the following breeding season. The male’s aggression score and current cooperative status were included as fixed effects and the response variable was the male’s cooperative status in the subsequent breeding season. We found no evidence of multicollinearity between predictor variables (VIF = 1.06).

A concurrent study attempted to manipulate the adult sex ratio in large plots established in the study area (Cox et al., 2019). We tested for but found no effects of these experimental manipulations on aggression scores or the relationship between aggression and cooperative group formation (Supplemental Material). Therefore, in the results section we report the results from analyses that used the full dataset and did not include plot as a fixed effect. We report the results from the restricted dataset, which tested for an effect of experimental plot, in the supplementary files.

Results

Male aggression scores averaged 4.22 (±2.01 max=6, min=1, N=27) and females aggression scores averaged 3.81 (±2.40, max=6, min=1, N=27). Aggression scores did not differ for males and females (Wilcoxon V=59, p=0.42, N=54 individuals from 27 nests). Of the 27 pairs assessed, eight breeding pairs included both a highly aggressive male and female (i.e. physical contact with the model, aggression score = 6), six pairs included highly aggressive females (aggression score = 6) and less aggressive males (aggression score <6), five pairs contained a highly aggressive male paired and a less aggressive female, and eight pairs included less aggressive males and females (aggression score <6). Aggression scores of paired males and females were not correlated (Figure 3; Kendall’s tau=0.16, Z=0.94, p=0.34, N= 27 pairs).

Figure 3.

Male and female breeders’ aggression scores were not significantly correlated with one another (Kendall’s tau=0.16, z=0.94, p=0.34, N=54 individuals at 27 nests). Black circles indicate individual males and females and black lines connect breeding pairs. Points and lines are jittered to show individuals with equal aggression scores.

Of the 27 nests where breeder aggression was assessed, eight were cooperative and 19 were pair-only. Breeders’ mean aggression scores were 4.81±1.94 at cooperative nests and 3.68±2.24 and at pair-only nests. Breeders’ current cooperative status in the year they were tested for aggression was unrelated to the breeding male’s aggression score (Table 2, Figure 4a) or the breeding female’s aggression score (Table 2, Figure 4b). Of the eight trials at cooperative nests, helpers responded to the heterospecific model during four trials with an average aggression score of 4.17±2.04 (range 2–6). The aggression scores of breeders at nests where helpers participated was 5.12±1.64 and the aggression scores of breeders at nests where helpers did not participate was 4.50±2.27. Female aggression scores positively correlated with brood size at time of trial (Spearman correlation rho=0.50, p<0.01, N=27 female scores), but there was no correlation for males (Spearman correlation rho=0.15, p=0.45, N=27 male scores).

Table 2.

Results of binomial generalized linear models (GLM; logit-link function) assessing whether breeders’ aggression scores related to (a) their cooperative status in the year of the aggression trial (N=27 breeding pairs) and (b) whether breeding males’ aggression scores related to their cooperative status in the year after the aggression trial (N=17 breeding males).

Response Variable	Fixed Effects	Estimate + S.E.	Z Value	P Value
(a) Cooperative status of breeders in year of aggression triala	Breeding Male Aggression Score	0.37±0.27	1.38	0.17
	Breeding Female Aggression Score	0.15±0.20	0.76	0.45
(b) Cooperative status of male breeders in year after aggression trialb	Breeding Male Aggression Score	−0.59±0.40	−1.47	0.14
	Cooperative Status in Year of Aggression Trial (Cooperative vs. Pair-Only)	1.68±1.41	1.20	0.23

Variables were tested for multicollinearity and the predictor variables were not strongly correlated (VIF_a = 1.03, VIF_b = 1.06).

Figure 4.

The partial effect of experimentally measured aggression was unrelated to the probability of having a helper (having a helper indicated by one on the y-axis) at the nest in the test year for (A) breeding males and (B) breeding females. Aggression scores were defined as categorical scores ranging from one (least aggressive) to six (most aggressive). Gray lines indicate upper and lower 95% confidence intervals. Circles represent individual aggression scores and were jittered so individuals with the same aggression scores could be shown (N = 27 male and 27 female aggression scores). The model included the effects of male aggression score and female aggression score on the probability of having a cooperative helper at the nest in the test year.

Whether a breeding male recruited a helper in the following breeding season was unrelated to their aggression score (Table 2, Figure 5) or to their cooperative status in the year aggression scores were assessed (Table 2).

Figure 5.

The partial effect of experimentally measured aggression was unrelated to the probability of breeding males having a helper (having a helper indicated by one on the y-axis) at the nest in the year following testing. Aggression scores were scored categorically on a scale from one (least aggressive) to six (most aggressive). Gray lines indicate upper and lower 95% confidence intervals. Circles represent individual aggression scores and were jittered to show individuals with the same aggression scores (N = 17 breeding males). The model also included the effect of cooperative status in the year they participated in the aggression trial.

Discussion

Predation is one of the most important determinants of nesting success among passerines (e.g., Martin, 1995) and breeders vary in their aggressive response to this threat (e.g., Cain et al., 2011). In this experiment, we used a model that posed a greater threat to offspring than to adults and in this situation more aggressive individuals should be better able to defend their nest compared to less aggressive individuals. Nest defense is an important benefit helpers provide (Teunissen et al. 2020a) and how breeders respond to nest predators could affect the extent to which they benefit from the presence of helpers and whether they form cooperative groups. Contrary to our expectation, breeders’ aggression scores were not related to their current cooperative status and there was no effect of breeding male aggression on future helper recruitment. Furthermore, we found no evidence that the aggressive behaviors of breeding males and breeding females differed from one another and no evidence that aggression scores of breeding partners were correlated. These results suggest that aggressive responses to a nest predator were not closely related to the formation of cooperative groups in the study population.

We considered the hypothesis that breeder aggression is related to the current and future recruitment of helpers. First, we predicted that less aggressive breeders would be more likely to form cooperative groups in the year they were tested and for breeding males, the following year, which might occur if less aggressive breeders benefit from helpers contributing to nest defense (Kazama & Watanuki, 2010). We expected more aggressive individuals would be less likely to form cooperative groups because these more aggressive individuals may have an advantage (Krebs, 1978) and defend unassisted. We also considered whether more aggressive breeders were more likely to form cooperative groups, which could occur if more aggressive breeders benefited from helper provisioning as a result of the lower levels of care aggressive breeders often provide in bi-parental species (e.g., males Mutzel et al., 2013; females e.g., Cain & Ketterson, 2013). We did not find support for either prediction. When compared to less aggressive male and female breeders, more aggressive breeders were equally likely to be in cooperative groups in the year they were tested. Furthermore, more aggressive males were not more or less likely to form cooperative groups the following year. Collectively, these data suggest that breeder aggression is unrelated to cooperative group formation.

Among cooperative breeders, nest defense is an additional way helpers can assist breeders (Teunissen et al., 2020a). Helpers did assist in nest defense at half of the cooperative nests assessed (N=8 cooperative groups) and displayed levels of aggression similar to breeders. Whether helpers participated in nest defense appeared to be unrelated to breeders’ aggression scores; the aggression scores of breeders at cooperative nests where helpers participated did not differ greatly from the aggression scores of breeders at cooperative nests where helpers did not participate. In other species, helpers tend to respond to nest threats less often than adults (Teunissen et al. 2020b) and their responses may depend on the type of threat (e.g., threat is adult predator vs. offspring predator Arnold 2000; Teunissen et al. 2020b). The predator used in this experiment posed a significant threat to nestlings, not adults, which may help to explain why helpers did not respond as frequently as breeders.

Although we detected no relationship between breeder aggression and cooperative group formation, it is possible that both less aggressive breeders and highly aggressive breeders form cooperative groups, and potentially for different reasons, thus neutralizing the overall effect. Although we did not explicitly test this, we did not observe helpers participating during nest defense more frequently when assisting less aggressive breeders and previous work in this population did not detect differences in breeders’ provisioning rates among cooperative and pair-only groups (Cusick et al. 2018). The benefits helpers provide are primarily associated with offspring care: chicks raised in cooperative groups received more food and weighed more (Cusick et al. 2018). Brown-headed nuthatches are cavity nesters that often confront both nest predators and other cavity nester species that attempt to usurp excavated cavities (e.g., eastern bluebird (Sialia sialis), white-breasted nuthatch (Sitta carolinensis), tufted titmouse (Baeolophus bicolor), and Carolina chickadee (Poecile carolinensis; JACox and JACusick pers. observation). Nest defense provided by helpers represents another benefit breeders receive. The aggressive response of breeding groups could be linked to improved fledging success and suggests that identifying how group aggression affects nesting success could help identify additional helper benefits. Additional studies are needed in both facultative and obligate cooperatively breeding species to understand the generalizability of the relationship between breeder aggression and cooperative group formation and potential benefits of group defense.

We assessed breeder aggression in response to a heterospecific nest predator, but it is possible that how aggressive individuals are towards a conspecific could affect cooperative group formation. Aggressiveness is one of the most well-documented behavioral types (Sih et al., 2004; Sih et al., 2010); individuals display consistency in their level of aggression across contexts (e.g., Duckworth, 2006; Hardman & Dalesman, 2018; Rangel-Patiño et al., 2018). For example, in western bluebirds, aggression displayed by male and female breeders towards a known nest competitor (tree swallow, Tachycineta bicolor) did not differ significantly compared to their aggressive response to a conspecific (Duckworth, 2006). Male aggression scores were also highly correlated in the contexts of nest defense and male-male competition (Duckworth, 2006).

More aggressive males, which may be expected to exclude potential helpers, may be more likely to form cooperative groups with previous offspring due to nepotism (Dickinson et al., 2009; Nelson-Flower & Ridley, 2016;). This could occur if more aggressive males were more likely to fledge offspring and recruit them as helpers, or if aggressive breeders were more tolerant of related helpers (e.g., Chiarati et al., 2011). We find this unlikely because, although helpers are typically offspring of the breeders they assist, non-kin helpers regularly occur in our population (Haas et al., 2010; Han et al., 2015; Cox et al., 2019, Cusick, 2019). Approximately 25% of helpers assist breeders to whom they were not related (Han et al., 2015) and in some years during experimental manipulations helpers assisted breeders that did not raise them (Cusick, 2019). Although we did not test whether helpers were genetic relatives of the breeders they assisted in this study, we observed that both highly aggressive and less aggressive breeders were assisted by helpers they had raised. However, most of these helpers were cross-fostered and therefore not genetic offspring (Cox et al. 2019). In addition, we found that some breeders were assisted by helpers they did not raise. These data suggest that nepotism alone does not underpin the formation of cooperative groups for more aggressive breeders in this species.

Whether individuals vary in cooperative behavior during aggression, and not just their aggressive response alone, may indicate whether they are cooperative in other contexts (e.g., cooperative group formation). Although it was beyond the scope of this study to quantify whether individuals cooperatively attacked the model, during some aggression trials lone individuals attacked the model, in other trials individuals took turns attacking the model, and in some trials all members of the group attacked the model simultaneously. Considering whether individuals displayed consistency in their cooperative tendency across different contexts (e.g., nest defense and offspring care) could be another measure of the variation associated with cooperative breeding behavior. Additionally, variation in cooperative group formation may be driven by intrinsic differences among potential helpers rather than intrinsic qualities of breeders. In western bluebirds, less aggressive offspring were more likely to stay and help their parents, while more aggressive offspring dispersed and reproduced (Duckworth et al., 2015; Potticary & Duckworth, 2018). Helpers may also differ in their cooperative tendency (Le Vin et al., 2011) or how much they contribute to different tasks (Arnold et al., 2005; Teunissen et al., 2020a), which could affect their recruitment or contribution within breeding groups. Future work should consider whether potential helpers vary in their cooperative tendencies at early stages and whether variation in other behaviors affects later decisions to cooperate.

There were no sex-based differences in breeders’ aggression scores in response to a nest predator. We have also observed the death of a female floater (Cox & Cusick, 2018), which typically occurs as a result of female-female aggression in other birds (e.g., Gowaty & Wagner, 1998). An absence of sex-based differences in aggression has also been observed in other species (e.g. Greenberg & Gradwohl, 1983; Chek & Robertson, 1991; Holtmann et al., 2019), while other studies have found female aggression to be a better predictor of successful nest acquisition and defense than male aggression (e.g. Rosvall, 2008). Context may influence the degree of aggression breeding males and females exhibit. Breeding females’ aggression scores were correlated with brood size at time of trial, but we did not observe this relationship for breeding males. Female aggression may also relate to the value of the offspring present (e.g., Trivers, 1972; Wiklund, 1990; Shew et al., 2016). Additionally, breeders may have differed in their previous experience with live nest predators, which could also influence their responses to the model, although we tried to minimize this by testing adults across the study population. Furthermore, male and female aggression scores were not correlated among breeding pairs, suggesting the absence of non-random mating based on aggressive tendencies, as observed in other species (e.g. Rosvall, 2010; Fedy & Stutchbury, 2005). In our study population, identifying how individuals vary in aggressive tendency will be an important next step for understanding variation in group interactions and nest defense.

Individuals varied in their aggressive responses to a heterospecific nest predator, but contrary to expectations, these differences did not relate to cooperative group formation among breeders. Our results suggest the expression of cooperative and aggressive tendencies may not represent a behavioral conflict. Future work should consider whether individuals cooperate during aggressive interactions and whether variation in cooperation during aggressive encounters predicts variation in cooperation in other behavioral contexts.