Social context modulates digestive efficiency in greylag geese (Anser anser)

In group-living animals, social context is known to modulate physiology, behaviour and reproductive output as well as foraging and nutritional strategies. Here we investigate the digestive efficiency of 38 individuals belonging to different social categories of a semi-feral and individually marked flock of greylag geese (Anser anser). During 9 consecutive days in winter 2017, when the ground was fully covered with snow (i.e. no grass or other natural forage available) and the accessible food was standardized, 184 individual droppings were collected and analysed to estimate the apparent digestibility of organic matter (ADOM). Lignin was used as an indigestible internal marker in the food and droppings. The digestive efficiency was higher in pairs with offspring as compared to pairs without offspring or unpaired birds. Furthermore, individuals with high ADOM were more likely to breed successfully in the following season than those with low ADOM. Our findings demonstrate that social status modulates digestive efficiency, probably via a chain of physiological mechanisms including a dampened stress response in individuals enjoying stable social relationships with and social support by their family members (i.e. their own pair-partner and offspring). Our findings underline the importance of the social network in modulating physiology, such as digestive efficiency, and ultimately reproductive success.

Most goose species live in large flocks with a complex social structure, as there is evidence for extended family bonds and female-centred clan formation 29,30 . In general, the presence of offspring plays a major role in social relationships among individuals: families dominate pairs without goslings in aggressive encounters, and pairs tend to win over single birds [31][32][33] . Greylag geese (Anser anser) are long-lived (10 years and longer 34,35 ) and lifelong monogamy is the rule (i.e. males and females associate year-round 36 ). Outside the breeding season, greylag geese are highly gregarious with strong family bonds; fledged goslings remain with their parents until the next breeding season 36 . Within flocks, agonistic interactions between pairs, families and clans are common. Such agonistic and supportive social contexts strongly modulate both the hypothalamic-pituitary and the sympathetic-adrenergic stress responses 37,38 . The effect of social support depends on family size because, in greylag geese with offspring, the excretion of immune-reactive corticosterone metabolites decreases with the number of offspring 17,38 . Furthermore, social and environmental factors interact with the immune system, as manifest in individual haematology: haematocrit and differential leucocyte counts are contingent with a suite of individual (i.e. sex, age), social (i.e. pair-bond status, parental experience) and environmental factors (i.e. season) 39,40 . Finally, emotional social support, enjoyed by individuals in long-term social bonds, contributes to reducing long-term glucocorticoid levels and thereby may help avoid gastrointestinal diseases related to chronic stress 41 .
We suggest that, via such mechanisms, social context affects the digestive efficiency in the highly social greylag geese, which were a valuable model species for recent social complexity research 42 . In the present study, we investigated the potential link between digestion efficiency and social status in 38 individuals belonging to different social categories (i.e. paired individuals with and without goslings, unpaired ones and juveniles) of a semi-feral and individually marked flock of greylag geese in winter.
The study was conducted during 9 consecutive days in winter 2017, when the ground was fully covered with snow, i.e. no grassland accessible. This enabled us to standardize the quality of food available to the geese. The apparent digestibility of organic matter (ADOM) was assessed from individual droppings by using lignin as an internal marker in food and droppings.
Because of the complex relationships between social context and physiological parameters, we expected paired birds with offspring (i.e. socially well embedded and enjoying social support) 43 to show a better digestive efficiency than unpaired individuals. Furthermore, we considered ADOM as an indicator of the physiological condition of an individual 44 at the end of the winter, when the breeding season starts. According to the condition-dependent model, individuals should adjust their reproductive decisions as a function of their body condition 45 . Therefore, we considered ADOM to indirectly modulate the reproductive performance in the following breeding season.

Results
During the study the birds were fed with standardised pellets with 95.6% dry matter content (DM). Ash and lignin were 7.5% and 5.3%, respectively. The individual droppings collected had an average analytical DM (±SD) of 95.0 ± 1.0%, and a content of 29.3 ± 13.5% and 9.5 ± 2.9% on DM basis for ash and lignin content, respectively. The calculations for lignin and ash were made on DM basis, i.e. on the dry content and not on the total sample (see the methods section below).
Model-averaged results identified social category as the strongest determinant of apparent digestibility of organic matter (ADOM). Paired individuals with offspring showed a significantly higher digestive efficiency than unpaired individuals or paired ones without offspring (Fig. 1). Age was the least important parameter, and sex had no influence on ADOM.
With respect to the interplay between ADOM in winter and the reproductive success in the following breeding season, hatched goslings was the most influential variable on ADOM. Individuals with hatched offspring in the next season showed a greater ADOM value at the beginning of the mating season than geese without (Fig. 2). Having a nest per se and/or fledged goslings had no influence on ADOM.
Top-ranked models are presented in Table 1, model averaged coefficients in Table 2.

Discussion
Our main findings indicate that social status modulates digestive efficiency, probably by dampening the stress response in socially well-embedded individuals, i.e. paired birds accompanied by their offspring. These results suggest fine tuning mechanisms between social context and physiological parameters as expressed by the percentage of apparent digested organic matter (ADOM). Paired individuals with offspring, irrespective of sex, showed a higher digestive efficiency than paired birds without offspring or unpaired ones. The seasonal patterns of corticosterone in greylag geese discussed by Kotrschal et al. 46 support our findings: early in the mating season, unpaired males excrete more immune-reactive corticosterone metabolites than paired ones. We interpret this as being related to the social situation within the flock because family members enjoy social support and the benefit of lower corticosterone levels compared to lone individuals when the mating season begins 30 . In geese, long-lasting family bonds are known to provide advantages for both the offspring and the parents, as was shown by the high efficiency of digestion in paired individuals with offspring. Juveniles, for instance, benefit in terms of body condition: juvenile barnacle geese (Branta leucopsis) within family units were disturbed less during feeding than those that had already left the family unit. Furthermore, Black et al. 47 showed that 10-month-old juvenile barnacle geese in family bonds were in better body condition after migration than those without family bonds. From the parents' perspective, the accompanying offspring actively contributed to detecting predators and to gaining and defending foraging space. This enabled the parents to spend more time foraging and ultimately enhanced the chances of reproductive success in the immediate future 47,48 . We therefore suggest that a high digestive efficiency linked to social status has a positive impact on the reproductive success of the following season. compensated for a lower energy intake per time by longer foraging bouts 49 . In our study, however, the provided food was standardised and available ad libitum, so that individuals of all social categories were able to forage to satiation. Juvenile greylag geese generally reach sexual maturity between 1 and 2 years of age. In our study, all focal juveniles were two years old and none of them accompanied the parents, i.e. was involved in a secondary family 30 . Nevertheless, they had a high ADOM, which might reflect different metabolic traits linked with age. However, our analysis indicates age per se to be the least important factor affecting ADOM (Table 2). This suggested that other factors are involved, for example metabolic differences linked to attaining sexual maturity and/or to the transition to adulthood, as shown by other studies 50,51 .
Furthermore, ADOM is apparently related to body condition, which is relevant for the forthcoming reproductive performance; focal individuals with hatched offspring in the breeding season following the data collection period showed higher values of ADOM than those failing to hatch goslings. Our study was conducted in February, a few weeks before the beginning of the laying period. ADOM therefore seems to be a meaningful value  for an individual's capability of facing an energetically demanding period such as the egg laying one. However, the correlative characteristic of our approach calls for further experimental study to corroborate such insights. Additionally, a low digestive efficiency might be compensated by a larger amount of food eaten, which would probably extend foraging time. Thereby, subordinates with low digestive efficiency might be forced to positions within the group that are more exposed to predators 52 , requiring them to make a trade-off between foraging and vigilance behaviour. Saito 53 claims that in some cases the mere presence of a dominant individual may cause a reduced energetic intake in a subordinate (socially mediated interference) 7 .
We suggest that long-term social bonds go beyond dampening corticosterone levels and benefitting both haematological and immunological parameters 17,18,54 to also affect further stress-related parameters such as digestive efficiency. In fact, socially well-embedded individuals benefit from better nutrient absorption than individuals lacking the supportive presence of social allies within the flock. Similar patterns have been shown for humans and non-human primates 55 . Such benefits might be mediated by increased levels of mesotocin, the analogue of mammalian oxytocin in birds, which facilitates the release of simpathethic-controlled gastrointestinal hormones and weight gain 56,57 . In mammals, oxytocin plays a major role in regulating social behaviour and positive interactions, thereby facilitating bonding or attachment by increasing social contact between individuals 58-61 . Babygirija et al. 59 , for instance, showed that social attachment allows rats to overcome daily stressful events and improve the impaired gastric motor function by up-regulating central oxytocin expression.
Our study provides new insight into the effects of complex social life on behavioural physiology of a model bird species and suggests further experimental research to understand the relationship between ADOM and life history traits.

Material and Methods
Study area and focal animals. The study area is located at 550 m above sea level in the valley of the Alm River in the northern part of the Austrian Alps (47°48′E, 13°56′N). The non-migratory flock of greylag geese was introduced by K. Lorenz and co-workers in 1973 62 . The birds are unrestrained and suffer natural predation with losses of up to 10% of the flock per year 35 . The geese are individually marked with coloured leg rings and are habituated to the close presence of humans 37,38 . Individual life-history data have been collected since 1973, which provides reliable information about an individual's social relationships within the flock (i.e. paired or not) as well as information on reproductive performance (i.e. having a nest or not; having hatched/fledged goslings or not). During the period of data collection, the flock totalled 167 individuals. Focal birds were 38 individuals (20 males and 18 females) belonging to different social categories within the flock, i.e. paired with and without offspring, unpaired individuals and juveniles (5 males and 3 to 5 females per category, Table 3). Age ranged from 2 to 24 years (mean age ± SD = 9.86 ± 6.43).

Data collection.
Individual droppings were collected during 9 consecutive days in winter 2017, from 14 to 22 February. During this period the snowpack was completely closed (i.e. no grassland was available), which ensured standardisation of the foraging situation: the birds fed exclusively on the provided pellets, which were available ad libitum. To our knowledge, there is no evidence for diurnal variation of digestive efficiency. Therefore, samples were collected twice a day, starting approximately one hour after feeding of the birds and lasting for 2.5 hours, i.e. between 0900-1130 and 1600-1800.  Droppings were collected immediately after defecation in 1 l plastic bags -one sample per bag -and frozen at −20 °C within 1 h until further analysis. A total number of 184 individual droppings were sampled from the focal birds (mean number of droppings per individual ± SD = 4.8 ± 0.4). After data collection was completed, samples were dried in an oven for 24 h at 55 °C and milled through a 1 mm screen (Pulverisette; Fritsch, Idar-Oberstein, Germany). Both the provided pellets and the collected droppings were analysed for (i) dry matter (DM) content by heating at 105 °C for 3 h (method 930.15 63 ), (ii) ash by incineration at 550 °C for 2 h (method 942.05 63 ) and (iii) acid detergent lignin 64 . Lignin was used as an internal marker to calculate the apparent digestibility of organic matter (ADOM) according to the following formula 65 : apparent digestibility (%) = [(1 − lignin in pellets) (g)/ lignin in droppings (g)] × 100.
Statistical analysis. Data were processed by using R version 3.2.5 66 and the additional packages 'nlme' 67 for linear mixed effects models (LME), and 'MuMIn' (Multi-Model Inference 68 ); for information-theoretic model selection and model averaging based on information criteria.
The distribution of the residuals was analysed with Shapiro-Wilk tests and visual inspection of Q-Q-Plots. To assess different effects on ADOM (response variable), we used LMEs(type-III sum of squares) with either (1) social category (paired with or without offspring, unpaired and juvenile), sex and age (years), (2) reproductive success of the following season (nest: yes/no; hatched goslings: yes/no) or (3) fledged goslings of the following breeding season (yes/no; sub dataset: including only pairs with at least one hatched offspring) as fixed factors. Bird identities were included as random factors in all models to control for the repeated sampling from individual birds. Three different regression models were used because fixed factors with multiple levels were included in the first model. To avoid overfitting and successfully calculate the models, a second model with the same dataset was used. The third model contained a sub-dataset of the original one for the parameter "fledged goslings".
To select the best models, an information-theoretic approach was used to calculate all possible candidate models 69 . All full models were compared to their corresponding null model. We ranked them according to their second-order form of Akaike's information criterion (AICc to account for small sample sizes 70 ), and selected the top-ranked models with a ΔAICc ≤ 2. The models were averaged to create model-averaged coefficients 69 . Those models in which the variation of the explanatory variables was not better explained by the full model than the null model (i.e. both models ΔAICc ≤ 2) are not presented in Table 2.

Data Accessibility
Data are provided in the electronic supplementary material appendix S3.