Hand and mouth use during food acquisition under naturalistic conditions
We characterized encounter with packaged food and urbanization for 25 videos of bonnet macaque (BM), 5 videos of Japanese macaque (JM) and 10 videos of vervet monkey (VM) which resulted in 174, 282 and 130 instances of FA and FP, respectively. As a result of paucity of video records, all categories of encountering packaged food and urbanization were not adequately obtained for all taxa. Classification of food items considered in the analyses and their relative representation is shown in Figure S1. In addition, comparison of hand/mouth use in FA between urbanization conditions with regard to individual food type show higher hand use in more urbanized groups than less urbanized groups (BM: fruit; JM-leaf and seed; Fig. S1), except for processed food in BM where hand use was equivalent between moderately urbanized and highly urbanized groups.
Inter-rater agreement based on a two-way random effects model (ICC(2,2), see Koo & Li, 2016) was over 0.86 (N=50) for each variable (age, foraging, attached, embedded and anatomical manipulator). We found near identical pattern of hand use during FA in every primate species we analyzed, i.e., higher hand use on higher encounter with packaged food and higher degree of urbanization. The dataset in bonnet macaques (BM) was devoid of any record of mouth use in the High categories of encounter with packaged food and urbanization and had the following trend in mouth use across both these variables, encounter with packaged food - High (P=0/75) < Moderate (P=18/73) and degree of urbanization - High (P=0/46) < Moderate (P=18/98) (Fig. 2). The illustration of Pearson’s standardized residuals for bonnet monkeys in Figure 2 affirms the hand/mouth bias across High and Moderate categories in encounter of packaged food and urbanization. Assessment of FA in the Japanese monkey (JM) and vervet monkey (VM) showed a similar trend across encounter of packaged food (JM: High (P=0/93) < Moderate (P=15/51) < Low (P=73/127), Cohen’sd Moderate-Low=0.508; VM: High (P=4/49) < Low (PL=11/33), Cohen’sd High-Low=0.527) and urbanization (JM: Moderate (P=15/144) < Low (P=73/127); VM: Moderate (P=4/50) < Low (PL=11/33)) (Fig. S2 and Fig. 2). The alternate statistical approach, which modeled FA in the vervet monkey as a negative binomial distribution within a binary logistic regression did not find any influence of age, foraging style, food attachment and food embeddedness but extent of encounter with packaged food (Log likelihood(N=82)=-34.9; FA: Mouth (ref. ); Encounter with packaged food: High (ref. ); βLow=-1.727±0.639SE, p=0.007; Table S1). The probability of FA using hands by vervet monkeys of the low encounter category decreased by a factor of 0.18 relative to vervet monkeys in the high encounter category (Table S1). Because the covariates of FA, extent of encounter with packaged foods and urbanization were correlated (Contingency coefficient=0.71), only one of the variables was used in the model.
Processing of embedded food typically involved peeling fruits/flowers/nuts and removing inedible portions of fruits/leaves, while extraction involved tearing garbage bags using hands, mouth or both (hand-mouth) in tandem. The use of hand during hand-mouth assisted extraction/processing also included situations where the hand has a passive role of supporting food as well as ones where the hand is actively involved in orienting/manipulating food (e.g., grabbing by hand while tearing with mouth) while in the mouth. Due to low sample sizes (NBM=20, NJM=11, NVM=47) and insubstantial statistical representation across encounter with packaged food and urbanization, we could not analyze hand/mouth use during food extraction by any species.