Introduction

Increased parasite transmission is considered a major cost of group living (Alexander 1974). Recently, it has been suggested that certain aspects of group living confer benefits that reduce the costs of parasite infection (Ezenwa et al. 2016). Such benefits include social immunisation, microbially mediated resistance and socially mediated tolerance (Ezenwa et al. 2016). This means that under certain conditions, group living could be advantageous rather than costly in terms of parasite infection.

Initially, group size has been hypothesised to be positively associated with parasite infection risk (Coté and Poulin 1995; Loehle 1995), whereas recent research hints towards the possibility of group size being a weak predictor of parasite infection risk (Rifkin et al. 2012) or negatively associated with parasite intensities (Patterson and Ruckstuhl 2013). Instead, behaviour of animals in social groups, for example frequency of social interactions and connectedness, affects infection rates among group members (Rimbach et al. 2015; Balasubramaniam et al. 2019; Habig et al. 2019). Next to social interactions (Duboscq et al. 2016; Romano et al. 2016; VanderWaal et al. 2016), physiological processes that allow for increased parasite replication or survival in the host can significantly affect an individual’s susceptibility to parasites. For example social behaviour can significantly affect an individual’s physiological stress response (Wascher et al. 2009; Wittig et al. 2015), which can affect the immune system and make individuals more susceptible to parasite infections (Apanius 1998; Akinyi et al. 2019). Otherwise, parasitic infections affect an individual’s ability to engage in social behaviour (Sheridan et al. 1994; von Holst 1998; Hanley and Stamps 2002; DeVries et al. 2003; Klein 2003; Lopes et al. 2016). Behavioural effects on an individual’s physiology and immune system range from relatively short term, as in the effect on disease susceptibility (McEwen et al. 1997), to long term, as in the effects on reproductive outcome (Buchholz 1995; Marzal et al. 2005; Hillegass et al. 2010) and have serious impacts on host longevity (Rousset et al. 1996; Archie et al. 2014).

The social environment can either facilitate or inhibit susceptibility and exposure to parasitism. Adverse effects of the social environment on health and susceptibility to parasites may be caused, for example by increased competition and aggressive behaviour (Azpiroz et al. 2003; Hawley et al. 2006; Chester et al. 2010). In meerkats, receiving but not initiating aggressive interactions was positively correlated with increased susceptibility of tuberculosis infection (Drewe 2010). Social status also affects an individual’s risk to be infected with parasites. A recent meta-analysis shows dominant individuals facing a higher parasite risk compared to subordinate individuals. This effect is mediated by social system (linear versus egalitarian hierarchies) and mating effort (Habig et al. 2018). In baboons (Papio cynocephalus), high-ranking males were less likely to become ill, and they recovered more quickly than low-ranking males (Archie et al. 2012).

Affiliative social interactions have been found to be associated with both increased and reduced risk of parasite infection. In rhesus macaques, Macaca mulatta, allo-grooming mediated transmission of Escherichia coli and central individuals in the social network could be considered ‘super-spreaders’ (Balasubramaniam et al. 2019). In Cape ground squirrels, Xerus inauris, increased durations of allo-grooming were associated with lower counts of ectoparasites (Hillegass et al. 2008). In northern bald ibis (Geronticus eremita), initiating more affiliative interactions was associated with lower excretion of parasite products (Puehringer-Sturmayr et al. 2018), potentially by having a stress-reducing effect, which in turn might positively affect an individual’s immune system (Sachser et al. 1998; Frigerio et al. 2003; Stöwe et al. 2008; Young et al. 2014; Müller-Klein et al. 2019). In a captive population of cooperatively breeding carrion crows, individuals with strong affiliative relationships excreted less samples containing coccidian oocysts (Wascher et al. 2019).

In the present study, I investigate associations between social behaviour and gastrointestinal parasite burden in the carrion crow, Corvus corone. Corvids express high variability in their social organisation depending on life history and ecological factors. Within species, the social organisation might vary between different life history stages, seasons or populations. For example in most European populations, carrion crows form socially monogamous pairs during the breeding season and large flocks during the rest of the year (Meide 1984; Glutz von Blotzheim 1985), whereas in northern Spain, crows live in stable social groups of up to nine individuals, consisting of the breeding pair and retained offspring as well as immigrants, which are in most cases male individuals (Baglione et al. 2003). Group living in carrion crows is linked to a higher level of cooperation, for example in nestling provisioning (Baglione et al. 2005). Within their social groups, corvids establish valuable social relationships, characterised by spatial proximity, high levels of tolerance, relatively low frequencies of aggressive interactions, and relatively high frequencies of affiliative behaviours (Fraser and Bugnyar 2010; Heinrich 2011). Within such valuable relationships, individuals support one another in agonistic encounters (Emery et al. 2007; Fraser and Bugnyar 2012) and share information and resources (Bugnyar et al. 2001; de Kort et al. 2003). Monogamous pair-bonds in corvids are characterised by complex social interactions that demand a high level of cooperation, coordination and affiliation between the paired individuals, which may explain, in turn, the evolution of advanced cognitive skills (‘relationship intelligence’: Emery et al. 2007; Wascher et al. 2018). Corvids usually establish linear dominance hierarchies (Izawa and Watanabe 2008). In a previous study on a cooperatively breeding population of captive carrion crows, individuals with strong affiliative bonds excreted a smaller proportion of samples containing coccidian oocysts as did individuals living in larger groups (Wascher et al. 2019).

Here, I investigate the association between aspects of the social environment of non-cooperatively breeding captive carrion crows (strength of affiliative relationships, dominance hierarchy and group structure) and fecal egg count. Fecal egg counts provide a reliable estimate of parasite infection rates (Seivwright et al. 2004; Daş et al. 2011). Further, I compare the results of the present study with the previously published data in Wascher et al. (2019). I expect a positive social environment, such as engaging in strong affiliative relationships, to reduce excretion of parasite eggs and oocysts. In the previous study, individual position in the dominance hierarchy was not correlated with eggs and oocyst excretion patterns and I expect to confirm these results. Lastly, I investigate the association between group size (i.e. pairs or trios versus flocks of eight or nine individuals) and parasite product excretion. Parasite transmission may be expected to be facilitated by an increase in group size; however, in line with Wascher et al. (2019), I expect to find a negative association between group size and the number of parasite eggs and oocysts excreted by individuals.

Methods

Study subjects and ethics statement

I collected the data for this study in four phases, between 2008 and 2010 and between 2012 and 2015 from a population of captive carrion crows housed in large outdoor aviaries at the Konrad Lorenz research station (KLF), Grünau, Upper Austria (47° 51′ 02″ N, 13° 57′ 44″ E). I observed 21 individuals (10 males and 11 females) kept in different group formations, such as groups of eight or nine individuals and pairs or trios. Due to the long-term character of the present study, individual birds were opportunistically moved between different compartments and group compositions due to age, reproduction, group expansion or the death of individual birds (Table 1). Aviaries were approximately 20–45 m2 and were equipped with wooden perches, natural vegetation and rocks. At the start of the study, groups were kept in the large aviary (45 m2) and then consecutively moved into smaller compartments when separated into pairs and trio. In January 2012, all birds were moved into same-sized compartments (20 m2). An enriched diet consisting of fruit, vegetables, bread, meat and milk products was provided on a daily basis. Water was available ad libitum for both drinking and bathing. This study complied with Austrian and local government guidelines. Individuals remain captivity housed in the Cumberland game park and the KLF (under the license AT00009917), before and after completion of the present study. Data collection was entirely non-invasive, and no further animal experimental license was required.

Table 1 List of all focal individuals and information about population, sex (female (F), male (M)), age (year of hatching; if not known, individuals are classified as adult), group (group composition: family, flock, pair, trio), phases of data taking during which the individual was recorded (G08: March to May, November to December 2008; G09: September to December 2009; G10: January to June 2010; G12: January to July 2012; G15: May to July 2015), number of droppings and behavioural focal observations collected, maximum composite sociality index (CSI) and Elo-rating (corrected for observational effort)
Full size table

Behavioural data

I conducted a total of 899 individual focal observations. Each observation lasted 5 min, and I recorded all occurring behaviours. For this study, I focused on the frequencies of agonistic behaviour (threat, chase flight and fight) and affiliative behaviours (allopreen and contact sit). I recorded the identity, role (initiator/receiver) of interacting individuals and the outcome of the agonistic interaction (winner/loser), with the loser of an agonistic interaction defined as the individual that retreated.

Composite sociality index

For each phase of data collection, I calculated a composite sociality index (CSI) for each crow dyad within a group according to Silk et al. (2010). I included two affiliative behaviours, namely contact sitting and allopreening, to calculate the CSI. Following Archie et al. (2014), I corrected for a different observational effort between individuals by regressing CSI against the number of observations per individual. The residuals from the regression presented the ‘corrected CSI’ (will be referred to as CSI from here onwards). The higher the CSI of a dyad, as compared with the frequency of the affiliative interactions observed within that dyad’s group, the stronger the affiliative bond between the two individuals in that dyad. For statistical analysis, I used the maximum CSI among all dyads for each individual, which reflected their strongest affiliative relationship within the group. For descriptive purposes, I classified dyads as ‘bonded’ when they displayed a higher CSI than the average of the entire sample and lower rates of aggression than the average of the entire group.

Elo-rating

I calculated the relative success levels of individuals in agonistic encounters as an Elo-rating in the R package ‘aniDom’ (version 0.1.4; Sánchez-Tójar et al. 2018). Elo-rating allows to track dynamic changes in rank over the different phases of data collection. Each individual was rated based on the outcome of each discrete interaction (winner/loser) and the (predicted) probability of that outcome occurring (Neumann et al. 2011). Similar to above, I calculated a ‘corrected Elo-rating’ to correct for a different observational effort between individuals by regressing Elo-rating against the number of observations per individual. The residuals from the regression presented the ‘corrected Elo-rating’ (will be referred to as Elo-rating from here onwards).

Parasitological examination

During the entire study period, I collected a total of 559 individual droppings directly after defecation (for a detailed overview, see Table 1). I determined individual gastrointestinal parasite load from droppings. From 2008 until November 2011, I used a modified version of the flotation method (Schnieder et al. 2006). I suspended the fresh droppings (0.1 g) in a 2-ml collection tube with 1 ml saturated saline. I shook collection tubes for 10 s and afterwards centrifuged for 5 min at 3000 rpm. After centrifugation, I filled the collection tubes with saline solution and positioned a cover slip (18 × 18 mm) onto the tube. The high density of the saline solution causes the parasite eggs and oocysts to float up and be caught on the cover slip (Carta and Carta 2000). After 10 min, I moved the cover slip onto an object slide and identified and counted the parasite eggs and oocysts (by size and shape). From December 2011 onwards, I used a McMaster counting chamber. I weighted the entire dropping, then diluted with 3 ml saturated NaCl solution per 0.1 g of dropping and mixed thoroughly. Afterwards, I poured the solution into both compartments of the McMaster counting chambers. After a 10-min resting period, I counted the number of parasite eggs and oocysts in each compartment and calculated the number of parasite products per 1 ml of dropping.

I used a compound microscope with 100-fold and 400-fold amplification for parasite examination to identify coccidian oocysts, several nematode species (Capillaria sp., Ascarida sp., Syngamus sp. and Heterakis sp., Trichostrongylus tenius) and cestodes to a varying degree (Table 2). I used presence versus absence of parasite eggs and oocysts for further analysis, which allowed a direct comparison between the two applied methods of dropping examination (flotation versus McMaster). As only nine samples contained cestodes, I conducted no further statistical analysis on this parasite group.

Table 2 Number of samples containing and not containing different parasite products
Full size table

Data analysis

All data was analysed by CAFW. During behavioural data collection, the author was blind to the outcome of parasitological examination and vice versa. I analysed factors affecting the proportion of droppings containing coccidian oocysts and nematode eggs in crows using the glmer function in R (version 3.5.3; R Core Team 2019) in the lme4 package (version 1.1-19; Bates et al. 2015). In two models, the number of samples containing nematode eggs or coccidian oocysts for each period of data collection was the response term. I calculated GLMMs with binomial error distribution and a two-vector response variable comprising the number of infected and non-infected samples for each individual in each phase. I employed various model diagnostics to confirm model validity (visual inspection of the distribution of residuals, Q–Q plots, residuals plotted against fitted values), none of which suggested violation of the model’s assumptions. To assess multicollinearity between fixed factors, I calculated variance inflation factors (VIFs) using the vif function in the package car (version 3.0-6; Fox and Weisberg 2011). VIFs for all models were below 1.6, indicating that there was no issue with multicollinearity (Zuur et al. 2009). Strength of affiliative relationships (CSI value), social structure (pair/trio or group), number of related individuals, sex and Elo-rating were included as explanatory variables. For each model, I fitted individual identity as a random term to control for the potential dependence associated with multiple samples from the same individuals. The statistical significance level was adjusted to p ≤ 0.025 following Bonferroni, to account for multiple testing of coccidia oocysts and nematode eggs. In addition to the main analysis, I compared the results based on the present dataset (collected from a non-cooperatively breeding population of captive carrion crows) with those based on a previously published data (from a cooperatively breeding population) (Wascher et al. 2019). Data collection and analysis (for example calculation of CSI and Elo-rating scores) were comparable in both studies. In order to investigate potential differences in social structure between the two populations, I compared the number of social relationships applying a GLMM with Poisson error distribution, as well as CSI value and Elo-rating applying two general linear models (GLMs) with Gaussian error distribution. Population was included as explanatory variable, individual identity was fitted as a random term in each model and I calculated models in the lme4 package.

Data availability statement

The datasets generated and/or analysed during the current study are available in the Open Science Framework repository, https://osf.io/3ytpf/.

Results

Social relationships

I observed 30 bonded dyads (out of 213 dyads in total), of which 19 were male-female dyads (five between related individuals and 14 between unrelated individuals). Six dyads were male-male dyads (all between unrelated individuals) and five female-female dyads (all between unrelated individuals). On average (± SD), males and females had 2.125 (± 1.642) and 1.625 (± 1.505) bonded partners, respectively. The mean CSI (± SD) between bonded dyads was 2.467 (± 2.165) for male-female, 2.258 (± 1.744) for female-female and 3.911 (± 2.062) for male-male bonds. Neither the number of social bonds, the CSI value or Elo-rating differed between the cooperatively breeding and non-cooperatively breeding populations of crows (number of social bonds: estimate ± SE = − 0.157 ± 0.363, z = − 0.433, p = 0.664; CSI: estimate ± SE = − 0.154 ± 0.748, z = − 0.206, p = 0.836; Elo-rating: estimate ± SE = − 9.733 ± 54.373, z = − 0.179, p = 0.857).

Occurrence of coccidian oocysts and nematode eggs

Two hundred forty-three samples from 18 individuals contained coccidian oocysts (43% compared to 31% in the Spanish population of cooperatively breeding crows). Crows kept in groups of eight or nine individuals excreted less samples containing coccidian oocysts compared to crows kept in pairs or trios (estimate ± SE = 0.986 ± 0.276, z = 3.564, p < 0.001, Fig. 1). Higher-ranking individuals excreted fewer samples containing coccidian oocysts than lower ranking individuals (estimate ± SE = − 0 ± 0, z = − 2.167, p = 0.03, Fig. 2). However, the number of samples containing coccidian oocysts was not related to the number of related individuals in the group, CSI (after Bonferroni correction) and sex (Table 3 (coccidia)).

Fig. 1
figure 1

Percentage of samples containing coccidian oocysts in carrion crow droppings in relation to the group structure. Box plots show the median and the interquartile range from the 25th to the 75th percentiles. The upper whisker is the maximum value of the data that is within 1.5 times the interquartile range over the 75th percentile. The lower whisker is the minimum value of the data that is within 1.5 times the interquartile range under the 25th percentile

Full size image
Fig. 2
figure 2

Percentage of samples containing coccidian oocysts in carrion crow droppings in relation to individual Elo-ratings

Full size image
Table 3 Results of the generalised mixed linear model investigating factors affecting patterns of coccidian oocyst and nematode egg excretion
Full size table

Overall, 172 samples, collected from 14 individuals contained nematode eggs (31% compared to 9% in the Spanish population of cooperatively breeding crows). None of the factors investigated significantly affected excretion patterns of nematode eggs (Table 3 (nematodes)).

Discussion

In this study, individuals from a population of non-cooperatively breeding carrion crows excreted less samples containing coccidian oocysts when kept in larger groups (8 or 9 individuals) compared to those individuals kept in smaller groups (2 or 3 individuals). Further, lower-ranking individuals excreted more samples containing parasite oocysts compared to higher-ranking individuals. A similar association between group-size and parasite excretion patterns was found in a population of cooperatively breeding carrion crows (Wascher et al. 2019). However, other patterns in the present study did not replicate previous findings. In a group of cooperatively breeding crows, the proportion of samples containing coccidian oocysts was negatively correlated with the strength of affiliative social relationships, but dominance hierarchy did not affect parasite excretion patterns (Wascher et al. 2019).

The present study suggests that variation in the social system within a species can result in differences in parasite product excretion patterns. Here, in a population of non-cooperatively breeding carrion crows, subordinate individuals excreted more samples containing coccidian oocysts. In contrast, no such association was found in a population of cooperatively breeding carrion crows (Wascher et al. 2019). In a way this is surprising, as dominance is a mechanism to suppress reproduction in subordinate individuals in some cooperatively breeding mammalian species (Creel et al. 1992; Young et al. 2006) and I would have expected an effect of rank position on parasite excretion patterns in the cooperatively breeding population of carrion crows, but not in the non-cooperatively breeding group. On the other hand, avian cooperatively breeding societies have often been described as highly tolerant, especially towards related individuals (Baglione et al. 2003; Dickinson et al. 2009), which could explain the present results. The relationship between social rank, glucocorticoid excretion and immune function is a complex one and can depend on factors like mating system, costs of rank acquisition and maintenance, or social stability (Goymann and Wingfield 2004; Cavigelli and Chaudhry 2012; Beehner and Bergman 2017; Habig et al. 2018). Further, the relationship between glucocorticoids and parasite burden might not be linear, and only high levels of physiological stress might influence parasite infection (Romeo et al. 2020). In two recent meta-analyses on several vertebrate taxa, dominant individuals exhibited higher parasite burden compared to subordinates, especially in linear versus egalitarian hierarchies and in mating systems where dominance rank predicts mating effort (Habig and Archie 2015; Habig et al. 2018). This is in contrast to the results of the present study. Carrion crows also can be considered to form linear and stable hierarchies (Chiarati et al. 2010), with dominant individuals gaining priority of access to resources such as food (Chiarati et al. 2011). In Siberian hamsters, Phodopus sungorus, social defeat affects immune function (Chester et al. 2010). Similar to the results in the present study, subordinate individuals display a higher parasite burden in a variety of species (Guenons, Cercopithecus mitis: Foerster et al. 2015; mice, Mus musculus: Rau 1983; plain zebra, Equus quagga: Fugazzola and Stancampiano 2012). Higher levels of parasite excretion in subordinate individuals might reflect higher exposure to psychosocial stressors compared to dominant individuals (Levy et al. 2020).

I did not find an effect of sex on parasite excretion patterns. Differences in parasite burden between the sexes are mostly mediated by endocrine-immune interactions (Zuk and McKean 1996; Klein 2004). For example high levels of testosterone are expected to be positively correlated with high parasite burden (Hudman et al. 2000; Decristophoris et al. 2007). Related to this, males are usually expected to have higher levels of endoparasites compared to females. However, this pattern does not hold in all species. Similar to the present study in carrion crows, a relationship between sex and parasite excretion was missing, for example in red-fronted lemurs, Eulemur fulvus rufus (Clough et al. 2010), brown mouse lemur, Microcebus rufus (Rafalinirina et al. 2019) and four species of songbirds (Granthon and Williams 2017). Data for the present study was collected throughout the year in non-reproductively active birds, hence sex-steroid levels are expected to be generally low (Soma 2006; Hau et al. 2008), and therefore both sexes might have been similarly susceptible to parasite infection in the studied population.

Increased exposure to parasites and disease transmission is considered one of the major disadvantages of group living. Group size is usually positively related to parasite burden, which is seen as a major selective force in group-living animals (Coté and Poulin 1995; Rifkin et al. 2012; Patterson and Ruckstuhl 2013). Similar to a previous study in captive cooperatively breeding carrion crows, individuals in the present study excreted less samples containing coccidian oocysts when living in larger groups. A limitation of the present study is the fact that I only used two categories of group size (2–3 and 8–9 individuals). Individuals were grouped together in these categories opportunistically. Juvenile crows in the wild form ‘non-breeder flocks’ (Meide 1984; Glutz von Blotzheim 1985), usually larger than 8–9 individuals; however, group sizes at the study site needed to be kept small in order for them to be manageable. It had to be considered that groups size and parasite load are associated in a non-linear manner, which could not be assessed in the present study (Hopkins et al. 2020). Infection by parasites in group living animals can be caused by different mechanisms. For example parasites can be transmitted directly during social interactions, hence individuals engaging more socially are at greater risk of infection. Parasite infection can also be related to physiological processes, which allow parasites increased replication or survival in the host, for example a compromised immune system, mediated by the physiological stress response (Muehlenbein and Watts 2010). Unfortunately, in the present study, I have not measured individual stress levels and therefore cannot verify an association between parasite excretion patterns and levels of stress. The relationships between the physiological stress response, the immune system, and susceptibility to parasites are quite complex. A recent meta-analysis of 110 records from 65 studies in mammalian hosts from experimental and observational studies generally indicated a positive relationship between glucocorticoids and parasite burden (Defolie et al. 2020); however, overall results about the relationship are complex. Future studies investigating the direct link between social behaviour, physiological stress response and parasite excretion patterns would be desirable.

In contrast to Wascher et al. (2019), strengths of affiliative social relationships were not associated with the proportion of samples containing coccidian oocysts. Neither the strength nor the number of affiliative relationships differed significantly between the subject populations in these studies. Therefore, it does not seem that populations show a general behavioural difference in affiliative relationships within groups. In cooperatively breeding crows, individuals with strong affiliative relationships excreted less samples containing coccidian oocysts (Wascher et al. 2019), which could reflect a stress reducing and immune system enhancing function of affiliative social relationships. Strong relationships between pair and partners, between kin and also between non-paired and unrelated individuals play a significant role in the social life of corvids and have been hypothesised to be an important driver in the evolution of cognition (Emery et al. 2007). I expected to find, but was unable to confirm in the present study, that the nature of an individual’s social relationships would affect their health and physiology. This is especially interesting as dominance rank was negatively correlated with the proportion of samples excreted containing coccidian oocysts. Aggressive social interactions have previously been shown to be among the most potent stressors (Wascher et al. 2008) and position in the hierarchy is related to stress (Creel 2001) and immunodepression (Barnard et al. 1998) in a number of species. Social relationships have been suggested to mitigate the effects of aggression onto the physiological stress response (Sachser et al. 1998) and therefore I would have expected to find an effect of strength of affiliative relationships on parasite product excretion in carrion crows.

In the present study, the excretion of coccidian oocysts but not excretion of nematode eggs was found to be related to social factors. These results replicated those of Wascher et al. (2019) and may be related to differences between these parasite species in their lifecycles. Coccidian oocysts have a shorter prepatent period compared to nematode eggs (Edgar 1955; French and Zachary 1994). This could potentially make them more sensitive to short-term changes in stress levels and immune system. This is supported by a study in carrion crows, showing coccidian oocysts, but not nematode eggs to significantly increase in the first week after a major stressor (Spreafico et al. 2012). In graylag geese, excretion of coccidian oocysts, but not nematode eggs are significantly increased in the first week after social isolation (Ludwig et al. 2017).

In summary, more droppings containing coccidian oocysts were excreted by individuals kept in pairs and trios compared to groups of eight or nine individuals. Strength of affiliative social relationships did not correlate with parasite product excretion; however, subordinate individuals excreted more samples containing coccidian oocysts compared to dominant individuals. The present results illustrate differences in the social system in carrion crows also resulting in different associations between social environment and parasite product excretion.