Introduction

Animals living in anthropogenic environments face both costs and benefits associated with such environments. While the costs are diverse, e.g., increased aggression both from conspecifics (Southwick et al. 1976) and humans (McCarthy et al. 2009), chronic stress (Marechal et al. 2011), culling or translocation (Berman et al. 2007; Malaivijitnond and Hamada 2008), the benefits are mainly limited to increased access to anthropogenic food resources (hereafter called AFR) (Oro et al. 2013). Compared to natural food resources, AFR are frequently higher in calories, and many (if not most) AFR have highly predictable locations and timing of access. Such predictable foods may increase foraging efficiency, and ultimately individual fitness of animals relying on AFR (Oro et al. 2013). A significant lower intake of natural food in provisioned animals indicates that AFR are often preferred over natural food resources (Barbary macaques, Macaca silvanus: Maibeche et al. 2015; rhesus macaques, Macaca mulatta: Sengupta and Radhakrishna, 2018; long-tailed macaques, Macaca fascicularis: Sha and Hanya 2013). Possibly from the benefits of foraging on AFR, non-human primate (hereafter called primate) groups living in an anthropogenic environment experience decreased interbirth intervals, lower infant mortality, as well as increased group sizes and population densities (Altmann and Muruthi 1988; Jaman and Huffman 2013; Kurita et al. 2008; Warren et al. 2011). Access to AFR is further associated with changes in activity budgets, diet, and social interactions (Berman and Li 2002; El Alami et al. 2012; Ilham et al. 2018; Kaburu et al. 2019a, b; Marty et al. 2019; McKinney 2011). Any of these aspects may influence animals’ fitness (Marty et al. 2019). While the degree, consistency, amount, and type of AFR can vary greatly across populations and situations, a substantial number of primate groups do have some access to such food resources, potentially affecting their behavior and fitness. Since access to AFR may be directly linked to an individual’s fitness, knowledge of those attributes that influence inter-individual differences in access to AFR is crucial to understand the selection pressures on individuals living in an anthropogenic environment.

While more and more primate groups get access to some AFR, access is not equally distributed across group members, but is likely dependent on the ability of some individuals to monopolize these food resources. The monopolization potential of AFR is expected to follow a gradient from professionally provisioned AFR that is confined to small spaces, to more widely distributed AFR (e.g., Ram et al. 2003). Professional provisioning occurs in situations where humans, often from official positions in governmental, educational, or conservation organizations, provision primates consistently and systematically, many times to manage such populations for tourism (e.g., Tibetan macaques, Macaca thibetana, of Mt. Huangshan, China: Berman et al. 2007) or research purposes (e.g., rhesus macaques of Cayo Santiago: Hernandez-Pacheco et al. 2016). The individual monopolization potential further decreases with an increased temporal and spatial unpredictability of AFR. If the occurrence of AFR is less predictable in space and time (e.g., food is provided throughout the day from several people at different locations), individuals with a high monopolization potential will need time to get to the AFR before they can prevent other individuals from accessing them. In such situations, typically described as opportunistic provisioning (e.g., food provisioning by visitors and local people), a lower predictability and thus monopolization potential can be assumed (Ram et al. 2003). However, compared to natural food, all AFR are generally considered to be more clumped and more predictable in their spatial and temporal distribution (Becker and Hall 2014; Koganezawa and Imaki 1999; Saj et al. 1999; Strum 1994; Warren et al. 2011). In summary, the distribution, size, and predictability of AFR may all affect to what degree certain individuals can monopolize them.

Previous studies investigating access to AFR in primates have mainly focused on professionally provisioned groups. In general, these studies found a consistent pattern where the social position in a group predicted access to AFR. High-ranking female Japanese macaques (Macaca fuscata), Barbary macaques (M. sylvanus), and Yellow baboons (Papio cynocephalus), for example, were found to feed longer on AFR and stay longer at the feeding site than low-ranking females (Altmann and Muruthi 1988; Fa 1985; Kurita 2007; Mori 1995; Soumah and Yokota 1991). In contrast, Ram et al. (2003) did not find rank-related access to AFR in opportunistically provisioned female bonnet macaques (Macaca radiata). The authors speculated that the spatial and temporal variation in provisioning decreased predictability of opportunistically provided food leads to a lower monopolization potential, thus rank would not predict access to such AFR. Finally, while patterns in access to AFR have been described in both professionally and opportunistically provisioned populations, most studies focused on female primates (Altmann and Muruthi 1988; Mori 1995; Soumah and Yokota 1991), whereas males where either ignored or did not show a rank difference in time spent at the feeding site (Fa 1985; Sapolsky and Share 2004).

In this study, we aim to investigate access to AFR in opportunistically provisioned groups of three primate species (long-tailed macaques; Macaca fascicularis, rhesus macaques; Macaca mulatta, and bonnet macaques; Macaca radiata). None of the study groups were professionally provisioned, i.e., received food provided by the government or other organization(s). Instead, they all had access to opportunistic food provided by visitors and local people. Despite being in between professionally provisioned and wild groups regarding the monopolization potential, we predict that access to AFR in opportunistically provisioned groups is biased towards males and high-ranking individuals since AFR are generally expected to be more predictable and clumped in comparison to natural food.

Methods

We observed adult individuals in nine groups of primates ranging from temperate areas in northern India, to tropical environments in southern India and Malaysia (Table 1). In the northern Indian city of Shimla (31.05 N, 77.1 E), we observed four groups of rhesus macaques from June 2016 to February 2018 (for details on the study site see Kaburu et al. 2019a). From July 2017 until May 2018, we studied two groups of bonnet macaques in the Thenmala Dam and Ecotourism Recreational Area (8.90 N, 77.10 E) located at the outskirts of the small town of Thenmala within the state of Kerala in southern India (Balasubramaniam et al., under review). In Malaysia, we observed four groups of long-tailed macaques in Kuala Lumpur, from September 2016 until February 2018 (for details on the study site see Marty et al. 2019). All groups were living in anthropogenic environments where visitors and tourists provided AFR to the monkeys. AFR in Malaysia mainly consisted of sweet and salty snacks (e.g., ice cream/chips) and occasionally fruits. Apart from AFR, the long-tailed macaques in Malaysia mainly consumed leaves, flowers, seeds, and insects from the natural vegetation (personal observations) similar to forested groups (Yeager 1996). The two sites differed considerably. While Batu Caves (groups MF3&4) (3.23 N, 101.7 E) is a tourist attraction with around 4000 visitors a day where the shared interface is limited to flights of stairs and the temple, Templer Park (groups MF1&2) (3.29 N, 101.6 E) has around 150 visitors a day and contains multiple paths where people interact with the monkeys (Marty et al. 2019). Among the rhesus macaques in Shimla, AFR was largely made up of sugar pellets called prasād, although occasionally macaques were fed with both sweet and salty snacks such as biscuits, chocolate, and chips (Kaburu et al. 2019a, b). In this site, macaques also had access to a forested area where they could feed on natural food sources, especially leaves and flowers (Kaburu et al. 2019a, b). For the group MM1, the shared interface consisted of a road with the surrounding buildings whereas groups MM2, 3, and 4 mainly interacted with people at the temple area (Kaburu et al. 2019a, b). In Thenmala, bonnet macaques were provisioned, albeit less frequently compared to Shimla and Malaysia, with sweet or salty snacks (e.g., chips, ice cream), spicy rice and wheat-based items (e.g., biryani, chapathis), and fruits (e.g., bananas). They also foraged on environmentally available foods, both anthropogenic (garbage, fruiting trees planted in gardens) and natural (insects, leaves, and fruits from forest patches within their home range). The shared interfaces for both groups (MR 1 & 2) consisted of roads, parking areas and garden areas (Balasubramaniam et al., under review). While the degree of provisioning from visitors differed between sites, none of the groups received any professionally provisioned food. Thus, the amount of AFR varied between days and locations within a group's home range.

Table 1 Demography of all study groups with number of males and females and the total observation time
Full size table

We used the proportion of AFR in the diet as a proxy for access to such food resources, on the premise that individuals prefer AFR food over natural food. This is indicated by a significantly lower intake of natural food in provisioned animals (Maibeche et al. 2015; Sengupta and Radhakrishna 2018; Sha and Hanya 2013). AFR were defined as all food items that humans brought into or cultivated in the study area (Kaburu et al. 2019a, b). We recorded the general activity (including feeding on either natural or AFR) through instantaneous sampling every 2 min within 10-min focal observations. Individuals with less than 30 data points (i.e., 2-min instantaneous records) were excluded from further analyses. In addition, only records where the observer unambiguously identified the food item were included. These observations accounted for more than 80% of all feeding observations. Interobserver reliability between the observers was all above 0.85 as assessed by Cohen’s kappa (Martin and Bateson, 1993). To calculate dominance ranks, we recorded all dyadic displacements (approach/leave interactions), submissions, and aggressive interactions between individuals with a clear winner/loser outcome during focal sampling and ad libitum sampling. For both males and females, a separate hierarchy was calculated using the package Perc in R (Fujii et al. 2016). A linearity and steepness test in R (R Development Core Team 2009) using the package “Steepness” (Leiva and de Vries, 2011) revealed that all dominance hierarchies were significantly steep and linear. We further standardized ordinal ranks to account for group size and created a rank index ranging between zero and one, indicating the bottom- and top-ranking macaque, respectively (see Kaburu et al. 2019b).

We used a generalized linear mixed model (GLMM) to determine individual differences in the proportion of anthropogenic food intake. The outcome variable was the total number of instantaneous records in which the individual was observed feeding on AFR. Sex and rank were the predictor variables. Species was included as a control variable due to potential inter-species differences in macaques’ propensities to interact with humans. In addition, group was included in the model as a random effect. The total number of instantaneous records with a clearly defined food object was used as an offset in the model. We analyzed the data in R (R Development Core Team 2009) using the ‘lme4’ package (Bates and Maechler 2011). We used a negative binomial model, as the initial Poisson GLMM indicated an over-dispersion of the data. The predictors did not reveal auto-correlation during model diagnostics (using variance inflation factors).

Results

We recorded 25,371 instantaneous records from 319 individuals across the nine study groups. The null model was significantly different from the full model (Chi2 = 42.70, df = 7, P < 0.001) and the predictors explained most of the observed variance in the outcome variable (conditional effect size R2 = 0.81).

Both sex and rank significantly predicted the proportion of anthropogenic food intake. Specifically, males consumed anthropogenic food significantly more often than females did (Table 2, Fig. 1). Similarly, higher-ranking individuals were more often observed feeding on anthropogenic food than were lower-ranking individuals (Table 2). Long-tailed macaques on average consumed more anthropogenic food than the two other species.

Table 2 Results from the GLMM investigating the predictors for the proportion of anthropogenic food intake
Full size table
Fig. 1
figure 1

Proportion of anthropogenic food intake recorded during 2-min instantaneous records. Solid horizontal lines represent the medians; box height corresponds to the interquartile range; whiskers indicate the maximum and minimum values; dots represent outliers. MF = Macaca fascicularis, MR = Macaca radiata, MM = Macaca mulatta

Full size image

Discussion

Our study shows that the benefits of living in an anthropogenic environment are unequally distributed across group members in urban-dwelling macaques. Specifically, males and high-ranking individuals consume more AFR than other individuals do, indicating greater access to such food resources for these classes of individuals.

We suggest that the main determinant of the observed pattern is the higher monopolization potential of some individuals. In all macaques, males generally outrank females due to a moderate-to-strong sexual dimorphism in body mass and a strong-to-extreme dimorphism in canine size (Plavcan 2001). Within their own hierarchy, both males and females with a higher rank were able to access AFR more frequently, most likely due to their ability to successfully displace lower-ranking individuals from preferred food sources. Our observation is in line with a previous experimental study showing that dominant individuals do monopolize AFR (Kaplan et al. 2011). Similarly, our results support previous studies on rank differences in access to AFR for female Japanese macaques under professionally provisioned conditions (e.g., Soumah and Yokota 1991; Mori 1995). However, they are contradictory to a study on bonnet macaques living in a comparable environment with opportunistic provisioning from visitors (Ram et al. 2003), which did not find a rank difference in the frequency of feeding on AFR compared to natural food. These differences in results are likely due to the low sample size in Ram et al. (2003)’s study compared to our study, or due to differences in the spread or amount of the provisioned AFR.

Rank and sex differences in access to food are not restricted to AFR but are expected to apply to natural food resources as well. High-ranking individuals in general are expected to have priority of access to preferred food resources in both urban and forested environments (e.g., Fedigan 1983). In females, high-ranking individuals are known to grow faster, mature earlier, and show shorter interbirth intervals than low-ranking individuals, and all these factors increase their fitness (reviewed in Harcourt 1987). In males, access to food resources determines body mass, which is often positively associated with rank in the wild (e.g., Dittus 1977; Dixson et al. 1993; Fragaszy et al. 2016; Marty et al. 2017). Access to preferred, high caloric food resources due to high rank can therefore be expected to help preserve a certain body mass, enabling males to maintain a high rank.

Even though the evolutionary implications of access to natural and AFR are similar, the potential impact on an individual’s health and fitness differs due to certain unique characteristics of AFR. Anthropogenic food is generally high in calories (McLennan and Ganzhorn 2017; Riley et al. 2013), and is likely to be more abundant, clumped, and more predictable in their spatial and temporal distribution than natural food (Becker and Hall 2014; Koganezawa and Imaki 1999; Saj et al. 1999; Strum 1994; Warren et al. 2011). As a result, it is likely that the monopolization potential of AFR exceeds that of natural food resources. This may specifically be the case in situations where humans provide spatially and temporally predictable AFR (professional provisioning/garbage sites). Yet, our results indicate that the monopolization potential is still high even if the predictability is lowered through opportunistic provisioning. In general, access to AFR may further increase within-group competition over food resources (e.g., Southwick et al. 1976), increase selective advantages for high-ranking individuals regarding their reproductive success, and may positively affect high-ranking male tenure. However, whether, or to what degree, low-ranking individuals with restricted access to AFR benefit from an anthropogenic environment compared to individuals living in forested groups, remains to be investigated.

Similar to the benefits of access to AFR, the costs associated with such resources may also not be equally distributed across individuals but rather accumulate in individuals with increased access to such food resources. Indeed, Sugiyama et al. (2014) found more congenital malformations related to access to AFR in male Japanese macaques compared to females. In a similar case in olive baboons (Papio anubis), diseases such as tuberculosis were found to spread primarily in males that were feeding on AFR, ultimately leading to the death of all males eating such food (Sapolsky and Share 2004). Females were left unaffected as they were not feeding on these infected sources of AFR (Sapolsky and Share 2004). Species and sex-specific differences in reproductive patterns may also bias the trade-off between costs and benefits of access to AFR. In species where males have a very short reproductive period [high reproductive skew and short tenure, e.g., Crested macaques, Macaca nigra; (Marty et al. 2017)], short-term benefits associated with AFR (e.g., increased body mass) might outweigh the long-term costs, as these costs are likely to manifest later in life after the benefits of increased fitness have been received. Alternatively, for males in species with a relatively longer reproductive period and more generally for females, these long-term costs may affect their fitness more substantially. Yet, a generally shorter interbirth interval, lower infant mortality, and higher population densities in many provisioned populations indicate that the benefits still outweigh the cost of access to AFR (Altmann and Muruthi 1988; Jaman and Huffman 2013; Kurita et al. 2008; Sugiyama and Ohsawa 1982; Warren et al. 2011).

While patterns of access to AFR were consistent for males and high-ranking individuals within and across groups, we observed marked differences in the total proportion of AFR in the diet between groups. These differences, however, can be explained by differences in exposure to sources of AFR (visitors and locals). Among the long-tailed macaque populations we studied, for example, two groups lived in an area with more than 4000 visitors per day, whereas the other two groups lived in a recreational park with around 150 visitors a day (Marty et al. 2019). Thus, the population with a higher exposure to visitors has higher proportions of AFR in their diet. This result shows that access to AFR is not dependent on the quantity of the AFR provided but rather shows the same pattern across different levels of opportunistic provisioning.

Our results highlight the importance of investigating inter-individual differences in cost–benefit assessments of living in an urban environment, in addition to group-level differences. Selection pressures specifically might be elevated in urban environments as a consequence of increased competition over AFR. However, more comparative and individual-based data are needed to reveal whether and how these differences in selection pressures differentially impact the long-term behavior, health, and fitness of individuals living in urban and forested groups.