Alloparenting Hypothesis Statement

Parenting behavior among adult chimpanzees

Figure 2(a) shows the occurrence rates of parenting behavior of focal adults toward Daiya, the male twin. Sango had the highest parenting behavior rate of all adult chimpanzees and was engaged in all four kinds of parenting behaviors (walking together, infant carriage, grooming, and physical contact). Sango was the only adult who was observed carrying Daiya. Robin, the father of Daiya and one of Sango's affiliated adults, was also engaged in parenting, particularly in physical contact with Daiya. The other four female adults cared for Daiya at a lower rate than did Sango and Robin.

Figure 2(b) shows the occurrence rate of parenting behavior of focal adults toward Sakura, the female twin. Unlike in the case of Daiya, Cherry, who was one of Sango's unaffiliated adults, was engaged in parenting Sakura for as much time as did Sango. Additionally, Sango, Cherry, and Koyuki (an adult affiliated to Sango) were engaged in all four kinds of parenting behavior. Moreover, besides Sango, three adult females (Koyuki, Cherry, and Chelsea) were found carrying Sakura.

The effect of proximity between focal adults and Sango, and focal adults and twins on parenting behavior

To determine the effects of the social relationships of the twins with Sango on the parental behavior by focal adults other than Sango, we conducted correlation analyses on proximity rate between twins and focal adults, Sango and focal adults, and the occurrence rate of alloparental behavior by focal adults.

There was no significant correlation between the occurrence rate of parental behavior toward Daiya by focal adults and the proximity rate between Sango and the focal adults. However, there was a positive correlation between the occurrence rates of parental behavior toward Daiya by focal adults and proximity between Daiya and the focal adults (Table 1). The same was true for Sakura: There was no significant correlation between the occurrence rate of parental behavior toward Sakura by focal adults and the proximity rate between Sango and the focal adults. However, there was a positive correlation between the occurrence rates of parental behavior toward Sakura by focal adults and proximity between Sakura and the focal adults (Table 2).

The effect of proximity between Sango and the twins on proximity between focal adults and twins

Table 3 shows the cross-tabulation of the number of scans revealing the relations between whether Daiya and his mother Sango were within the reach of each other and whether each focal adult and Daiya were simultaneously within the reach of each other. In this analysis, we counted the total scans conducted when the adult chimpanzee was in the outside enclosure, and categorized the scans into four conditions that occurred in a combination of whether Daiya and his mother, Sango, were within each other's reach, and whether each focal adult and Daiya were simultaneously within each other's reach. For example, the total number of scans conducted when Robin was in the outside enclosure was 382, and for 32 of these scans, both Sango and Robin were within reach of Daiya, for 114 scans Sango was within reach but Robin was out of reach of Daiya, for 62 scans Sango was out of reach but Robin was within reach of Daiya, and for 174 scans both Sango and Robin were out of reach of Daiya.

Koyuki and Daiya were within the reach of each other at a higher rate when Sango and Daiya were also within the reach of each other than when they were out of reach (χ2(1) = 6.12, p = 0.013). In contrast, Chelsea and Daiya were within the reach of each other at a higher rate when Daiya and Sango were not within the reach of each other (χ2(1) = 16.39, p < 0.00006).

Table 4 shows the cross-tabulation of the number of scans revealing the relations between whether Sakura and her mother were within the reach of each other and whether each focal adult and Sakura were simultaneously within the reach of each other. All of Sango's affiliated adults (Robin and Koyuki) and Sakura were within the reach of each other at a higher rate when Sango and Sakura were also within the reach of each other (Robin: χ2(1) = 8.52, p < 0.004; Koyuki: χ2(1) = 4.31, p < 0.038). In contrast, two of Sango's non-affiliated adults (Cherry and Chelsea) and Sakura were within the reach of each other at a higher rate when Sakura and Sango were not within the reach of each other (Cherry: χ2(1) = 22.36, p < 0.000002; Chelsea: χ2(1) = 11.95, p < 0.001).

Communication between focal adults and twins before infant carriage

By using the 1-min fixed-interval point sampling method, we recorded infant carriage at 50 sampling points. We observed communications between focal adults and infants before four of these 50 sampling points. All four communications were between Sakura and focal adults. Two were adult-initiated communications between Koyuki and Sakura, and Sango and Sakura. The other two were infant-initiated communications between Cherry and Sakura, and Chelsea and Sakura. Both Cherry and Chelsea were Sango's non-affiliated adults. The details are as follows.

Adult-initiated communication between Koyuki and Sakura

At 15:32 h on July 23, 2011, Sakura and Koyuki were in the proximity to Sango. When Koyuki started to move away, Sakura gently touched Koyuki. Thereafter, Koyuki extended her right hand to Sakura with her palm facing up (Figure 3a). Next, Sakura approached Koyuki and climbed onto Koyuki's back (Figure 3b).

Adult-initiated communication between Sango and Sakura

At 13:22 h on August 18, 2011, Sango began to move under the tower, and Daiya and Sakura followed. When Sango reached the lower step, Sango directed her gaze to Sakura and stretched her hand toward Sakura (Figure 4a). Thereafter, Sakura fell into Sango's arms and climbed onto Sango's back (Figure 4b). Daiya went to Sango's and Sakura's side.

Infant-initiated communication between Cherry and Sakura

At 11:58 h on February 5, 2012, Cherry and Sakura were resting about 2 m apart. After Sakura approached the wall and kicked it, she approached Cherry and roughly touched Cherry's face with her left hand (Figure 5a). Thereafter, Cherry shifted her posture and moved (Figure 5b), and Sakura climbed on Cherry's back (Figure 5c).

Infant-initiated communication between Chelsea and Sakura

At 0: 30 h on February 5, 2012, Chelsea, Sakura, and Cherry were resting in the cave with about 1 m of distance among them. Sakura approached Chelsea, touched Chelsea's head, and dragged Chelsea outside the cave (Figure 6a, b). Thereafter, Chelsea moved, and Sakura climbed on Chelsea's back (Figure 6c).

Proximity between Sango and the twins

To determine whether Sango was equally close to each of her twins, we compared the numbers of proximity and non-proximity scans with Sango and each of Daiya and Sakura by using a chi-squared test. Daiya was close to Sango at a significantly higher rate than was Sakura (Figure 7; χ2(1) = 5.59, n = 764, p < 0.018).

Artificial colonies

The study was carried out in 2007 and 2009 in artificially established colonies of the feral pigeon in a building in NW Poland, to which individuals were transported from four feral pigeon breeding colonies in Słupsk (NW Poland, 54° 28′N, 17° 10′E) in 2005. Only fledglings formed the artificial colonies, which were completed over several months each time. The overall number of introduced birds in all experiments was 116. All pigeons were individually colour-ringed, and to reduce the variability in breeding performance associated with age42, all birds used in the experiments were at least 2 years old.

We noted the dates of same-sex pair formation and their persistence during the breeding period. All observations of bird behaviour were performed from a hide and lasted 3–7 hours per day. The sex of the pigeons was determined according to the characteristic mating behaviours of the sexes at an older age43. These sexual behaviours are highly sex-specific, e.g. only males bow-coo, females always initiate heteropreening, courtship feeding, driving and female guarding are always performed by a male43. Also, copulation is initiated by a male mounting a female43. Every individual was reliably sexed according to these types of behaviour. We considered all attempts at “kissing”, copulation and establishing a nest territory during which mating behaviour towards a same-sex individual was observed as same-sex sexual behaviour of males and females. We did not observe any atypical behaviour of females in f-f pairs. Females from f-f pairs behaved in the same way as females from f-m pairs. However, females from f-f pairs copulated with other males (in this way eggs were fertilised). There was also one clear difference between f-f and f-m pairs: in f-f pairs there was no driving/mate guarding because no male was present. The experimental removal of males did not change the “female-like” mating behaviour of females.

The major breeding strategy of the feral pigeon is to hatch as many broods as possible. Each brood always consists of two eggs, so increasing the number of broods in one breeding season is an adaption to increase breeding output44,45. As chicks become normothermic, the adults begin to prepare for the next brood. The overlapping broods require the cooperation of two parents; for 80% of the time, the new clutch is incubated by females, while males feed the fledglings from the previous brood. After the new brood hatches, both parents feed their offspring.

Experiment creating a female-biased sex ratio

The experiment was conducted between February 2007 and February 2009. In February 2007, we removed 18 paired males and all young unpaired males (the offspring from 2006) from the first artificial colony. The removed individuals were housed in a separate room and given water and food ad libitum.

At the end, the structure of the colony was 20 f-m pairs (total 106 broods), five f-f pairs (total 19 broods) and 14 single females (total 26 broods; note that for single breeding females, we added two pairs [two broods - four hatchling histories] from 2009 in which the experimental conditions were the same as described above). The hatching success (measured as the number of hatched eggs) and fledglings per brood was obtained for 106 f-m broods, 19 f-f broods and 26 single females’ broods. For further analyses we selected the histories of 18 broods (one brood per pair; two broods from two pairs were excluded because the relevant data became lost) from 20 f-m pairs. Thus, the sample sizes for the incubation time were 18 f-m broods, 13 broods of f-f pairs (six broods excluded because females abandoned their nest) and 12 single females’ broods (14 broods excluded because females abandoned their nests). In the case of sample sizes for brood overlap (the number of days from the hatching of chicks to the next egg laying event) we had 18 f-m broods and 13 f-f broods (six broods excluded because females abandoned their nests); single females’ broods were not analysed because females laid eggs only 3 times while they were feeding their previous brood.

To obtain the mass gain curve of hatchlings, we analysed only the successfully hatching broods. Hence, we analysed 18 hatchling (9 females and 9 males) histories from five same-sex f-f pairs, 36 hatchling histories (19 males and 17 females) from 18 opposite-sex pairs and 14 hatchling histories (6 males and 7 females, 1 chick not sexed) from 14 single breeding females. The sex distribution of offspring was evenly spread across the groups (χ2 = 0.173, df = 2, p = 0.917). The masses of the nestlings were recorded every day from hatching until fledging (33 days) using a PESOLA spring scale. Opposite-sex pigeon pairs always have two eggs, so if females in f-f pairs laid more than two eggs, we randomly removed excess eggs from the nest; this was the case for 17 broods. The removal of excess eggs was necessary because pigeons can only raise two chicks43. Before the experiment started we had measured the weight of females without any knowledge of their future pair associations; thus, during the experiment we had 10 females in f-f pairs and 10 single females to test whether the larger weight of females is associated with f-f pairing.

Breeding parameters such as the number of broods per season and the overlap between broods (days) were analysed using general linear models. For incubation time (days), we used a general linear mixed model, and for the number of hatchlings and fledglings per brood, we used generalised linear mixed models with a Poisson error distribution. For mixed models, we set the identity of the pair (id) as a random effect. The weight of single females against f-f paired females was tested using a general linear model. All statistics were performed using the R statistical software46 and the additional “lme4” package47. The explanatory variables were tested using the drop1() function. The assumptions of all models were checked graphically, but we did not find any violation of either homoscedasticity or normality of residuals.

To fit the model to mass gain of pigeon chicks, we searched for the best structure of the mass growth curve. To obtain the most parsimonious time structure we used fractional polynomials48. This method is based on the fit of regression models that have m terms of the form tp, where p is defined as the exponents selected from the small set integer and non-integer values (S = {−2, −1, −0.5, 0, 0.5, 1, 2, 3}). The linear predictor is defined by

where t is a time covariate (days) and M is the order of covariate t. For each term, the power pm is selected from the restricted set S. Note that according to48, when M = 2 and p1 = p2, the linear predictor (1) is β0 + β1tp1 + β2tp1 log(t) and when p = 0, the predictor is log (t). We fitted all possible models from M = 1 to M = 3. All performed models are presented in Appendix 1. The “best” model was chosen using Akaike’s Information Criterion (AIC, the lower the value, the better fitted the model). This procedure penalises the addition of new parameters, so it helps to avoid over-fitting49. The computational process of fractional polynomials was done using the package “CorrMixed”50 in the R program46.

Subsequently, the selected structure of the time covariate was used as a covariate in a GLMM (general linear mixed model with normal error structure). To compare different parent structures, we allowed this factor (three levels: f-m, f-f, and single) to interact with the polynomial time covariate. We used a mixed model because the measurements of chicks were sequential data taken every day until fledging (33 days). We checked for different random effect structures. In the most general structure, we added hatchling id, pair id or nest id and the time of measurement as random effects. We found the random slope of time for each pair (time/id) and the random intercept of nest id with the lowest AIC value. We verified the model graphically to check for model assumptions and did not find any violations.

The parameter estimations and estimated means are given with standard errors (SE). Multiple comparisons were carried out using Tukey’s test.

First experiment creating a male-biased sex ratio

In 2008, we established the second artificial colony. At the beginning of the breeding season (in February) we removed 20 sexually mature paired females (with and without breeding experience). The removed individuals were housed in a separate room and were given water and food ad libitum. The male-biased colony structure consisted of 20 opposite-sex pairs and 25 single unpaired males. In this experiment, we created conditions for egg adoption. If two males were observed in a nest we added one egg and observed the outcome, i.e. whether the egg was adopted and incubated or abandoned.

Second experiment creating a male-biased sex ratio

The experiment was planned after field observations which revealed that single males may show mating behaviour towards fledglings. The experiment was performed on the third artificial colony in 2008. We widowed 15 males (by removing the females and placing them in another room) just before the chicks fledged. This was performed under conditions of a male-skewed sex ratio. To facilitate the father-son pairs, the lighter of the two orphaned chicks was removed. Earlier studies showed that young males are heavier than young females42,51; these removed chicks must have been females. There are significant differences between sexes in postembryonal development; after the fifth day of life males become heavier than females. This difference increases in time and males are 25 grams heavier than females on average before fledgling. In broods consisting of two female offspring, this method may not be reliable. However, as the young birds had been ringed, the final identification of sex was confirmed as the birds grew up, based on their adult sex-specific behaviour52.

Data availability

The authors declare that data supporting the findings of this study are available within the paper and the supplementary information files.

Ethical note

This study complied with Polish regulations regarding the ethical treatment of research subjects in the experimental period 2007–2010, and the colony was established with the approval of the relevant authorities, the University Dean (“Zezwolenie Dziekana Wydziału Mat-Przyr Akademii Pomorskiej na prowadzenie badań na gołębiu miejskim 2/2010”) and with approval of the Ministry of the Environment (“Zmienność parametrów hematologicznych i biochemicznych u gołębia miejskiego (Columba livia f. urbana) w różnych okresach lęgu”, approval 20.07.2007, decision number: DLOPiK - op/ogiz-4200/III-21/3706/07/jr).


Leave a Reply

Your email address will not be published. Required fields are marked *