Are larger primates less faunivorous? Consumption of arthropods by Amazonian primates does not fulfil the Jarman-Bell and Kay models

Terrestrial arthropod groups, including insects, spiders, and millipedes, represent an important food resource for primates. However, species consumed and patterns and rates of arthropod-related feeding for most frugivorous primates are still poorly known. We examined stomach contents of 178 Amazonian primate specimens of nine genera and three families obtained from community-based collections voluntarily donated by subsistence hunters. Based on our results, we assessed whether consumption patterns followed the expected negative relationship between arthropod ingestion and body size as postulated in the Jarman-Bell and Kay models. We identified 12 consumed arthropod taxa, including insects (beetles, grasshoppers, ants, flies, caterpillars, praying mantises, and others), spiders and millipedes. Medium and large-bodied primates consumed a greater diversity (measured by Simpson’s diversity index) and richness of arthropods, but differed in terms of composition of taxa consumed. Cacajao, Sapajus and Cebus consumed proportionally more Orthoptera and Coleoptera compared to the other primate genera analyzed. We did not find significant correlations between richness and diversity of arthropods consumed and primate body mass. There was a slight tendency for the decrease in the relative content of arthropods in the diet with increased body mass in medium and large primates, which does not provide full support for the Jarman-Bell and Kay models. The study of arthropod consumption by arboreal primates in the wild remains challenging. Our study suggests that arthropods supply essential nutrients for frugivorous primates, and provides an alternative method to analyse faunal consumption patterns in primates.


INTRODUCTION
To meet their daily energy, protein, and micronutrient requirements, frugivorous primates in tropical forests consume a variety of non-fruit diet items, including leaves and/or animal matter (Janson and Chapman 1999;Chapman et al. 2012). The animal component of the diets of these primates is mostly formed by terrestrial arthropods (i.e., ground-and tree-living arthropods). In forest habitats, such arthropods are widely available, though patchily distributed on the ground and in vegetation, often highly mobile or found embedded in fruits and galls or within wood (Stork and Blackburn 1993;Santos-Barnett et al. 2022). For many primates, arthropods represent an important food resource (Redford 1987;Rothman et al. 2014). Because of their generally high digestibility, elevated content of lipids, proteins, vitamins, and minerals (Raubenheimer and Rothman 2013;Rothman et al. 2014), and their high energy yield per unit mass, terrestrial arthropods may be a significant source of nutrients for frugivorous primates (DeFoliart 1995;Lambert 1998). However, the frequency and importance of arthropod consumption for most frugivorous primates remains poorly known (Rowe et al. 2021).
The three families of platyrrhines (Atelidae, Cebidae and Pitheciidae) comprise arboreal species found in the tropical regions of Mexico, Central and South America (Wang et al. 2019). The tree-living habit of these primates hinders the collection of information on arthropod consumption, which is often via direct observations, due to the height of the canopies in which these monkeys normally move and feed (Milton and Nessimian 1984;Pickett et al. 2012;Nielsen et al. 2018). Given how swiftly arthropods can be ingested, and the often opportunistic nature of such encounters, these limitations may considerably underestimate the diversity and relative content of arthropods consumed (Rowe et al. 2021).
According to the Jarman-Bell principle (Gaulin 1979), arthropod consumption is generally ecologically related to the nutritional requirements in mammals, and generally inversely correlated with the species' body size. Large animals therefore consume large amounts of lower quality food (e.g., leaves) due to their lower energy requirement per body mass, while smaller species tend to consume mainly high-quality items (e.g., arthropods). For primates, Kay's Threshold hypothesis predicts that significant consumption of animal matter will be typical in primates < 350 g, and consumption of leaves will predominantly occur in primates > 500 g (Kay 1984). Since most frugivorous primates have intermediate body sizes (Gaulin 1979;Kay 1984;Hawes and Peres 2014), leaves may be predicted to be the main complementary food item.
Here, we analysed stomach samples of nine Amazonian primate genera obtained from subsistence hunters who voluntarily donated these organs to our study. By directly examining the contents of ingested food material, it was possible to accurately assess and compare the arthropod richness, diversity, and relative content consumed. This method provided a useful snapshot of foods eaten shortly before collection (Milton and Nessimian 1984), despite the rapid gastrointestinal transit of foods, particularly in frugivorous primates (from 2.5 to 7.2 hours of gastric residence time, Chen et al. 2008). Using the data gathered, we also determined whether our observations fit the theoretical predictions established in the Jarman-Bell and Kay models.
All primate stomachs were opportunistically collected during normal hunting activities of the communities' inhabitants. Subsistence hunting was not focused on primates, Table 1. Number of specimens (N) and mean adult body mass of Amazonian primate taxa analysed for stomach contents. Arthropod frequency of occurrence = percentage of stomachs that contained arthropods; total content mass = mean (± standard deviation) of stomach content biomass; arthropod content mass = mean (± SD) of arthropod stomach content mass; relative content of arthropods = mean (± SD) of the arthropod content mass relative to total stomach content mass; frequency of occurrence of arthropod taxon = percentage of stomachs that contained each arthropod taxon; λ = Simpson' diversity index. ACTA AMAZONICA instead they form part of an extensive list of consumed regional mammals. No animals were killed other than those harvested as part of local hunters' usual activities. Samples were identified by a code that included the primate genus and date of hunting, and kept in containers with 4% formaldehyde, following Mayor et al. (2017). In ASDR and MSDR, hunters also recorded the animals' body mass (in kg) using a 20 kg Pesola scale (accuracy = ± 0.3%).
The research protocol was approved by the Chico Mendes Institute for Biodiversity Conservation in Brazil (license SISBIO # 29092-1) and by the Research Ethics Committee for Experimentation in Wildlife at Dirección General de Flora y Fauna Silvestre in Peru (license # 0229-2011-DGFFS-DGEFFS). The samples collected in Peru were deposited at the Instituto Veterinario de Investigaciones Tropicales y de Altura of Universidad Nacional Mayor de San Marcos (Peru) and those collected in MSDR and ASDR were deposited in the Scientific Collection of Mastozoology of the Instituto de Desenvolvimento Sustentável Mamirauá (Brazil).
Stomachs were opened at their greatest curvature. The contents were washed over granulometric sieves with different mesh sizes (4.0, 0.6 and 0.075 mm) to retain whole arthropods and their fragments, and the retained material was then dried in an oven at 60 °C for up to 6 h (following Torres et al. 2022). Using an Olympus stereoscopic microscope, we isolated all whole and fragmented arthropods found in the sieved contents at 10 to 40X magnification. We weighed the mass of the whole stomach content, and the arthropod content was weighed on a Shimadzu analytical balance (precision: 0.0001 g). Arthropods were identified by the authors of this study with the help of experienced entomologists (see acknowledgements) up to order level and, in a few cases, to family level.

Data analysis
We conducted all primate analyses at genus level by pooling congeneric species into the same taxon. We calculated the relative content of arthropods in each stomach by dividing the arthropod content mass by the total mass in the contents of each stomach, and the frequency of occurrence of each arthropod taxon for each primate genus.
We estimated the diversity of arthropods consumed by each primate genus using the Simpson diversity index (λ), following the equation: λ = 1 -Σ(P i )²; where P i is the frequency of each arthropod taxon in relation to the total richness consumed by each primate genus. When λ values approach 1 (one) they indicate high diversity, and when they approach 0 (zero) they indicate low diversity (Simpson 1949). We produced richness rarefaction curves with 95% confidence intervals using EstimateS 9.1.0 (Colwell and Elsensohn 2014), based on a resampling with 100 permutations of our samples, aiming to reduce the effect of undersampling of some primate genera. We also extrapolated the sample size of each primate genus based on the sample size of the most sampled taxon (Alouatta, n = 48) in order to estimate the stabilization of the rarefaction curves and predict the richness of arthropods for each primate genus if sample sizes were equal.
We used a multidimensional analysis (principal coordinates analysis, PCoA), to assess the differences in terms of composition of arthropod taxa (i.e., presence/absence of each arthropod taxa in each primate stomach) consumed by primate genera based on the binomial distance index, using R 3.5.2 (R Core Team 2019). The binomial distance deals with binary presence/absence data and can handle varying sample sizes. As a complementary approach, we performed an analysis of similarity (ANOSIM) to obtain the statistical significance of dissimilarities in arthropod composition between the primate genera obtained by the PCoA.
We performed Pearson and Spearman correlations (according to the normality of the data tested by the Shapiro-Wilk test) to assess the relationship between primate body mass and the contribution of arthropods to their diets in terms of relative content consumed, the Simpson diversity index and estimated richness with extrapolated sample sizes. We tested these correlations in two scenarios. In the first scenario, we considered the nine primate genera evaluated, and in the second, we kept only medium-and large-sized primates and excluded genera with body weights < 500 g (i.e., Leontocebus and Saguinus), since these could mask the relationships or bias the analysis due to small sample size. We used the mean primate body mass recorded by hunters in the ASDR and MSDR, and from literature information for YMR (Table 1).

RESULTS
We found terrestrial arthropods in the stomachs of all studied primate genera. Overall, we recorded the presence of arthropods in 73% (130/178) of the stomachs, with arthropod frequency of occurrence ranging from 33.3 to 100% among primate genera (Table 1). Mean ingested arthropod content varied from 0.02 -0.9 g among primate genera, and the relative arthropod content averaged from 0.00005 to 0.1 of the total stomach content mass (Table 1). Arthropod content proportions were higher in Lagothrix, Cacajao, Cebus and Sapajus (Table 1), and lower in Ateles and Leontocebus, though sample sizes for the latter were lowest (Table 1).
We were able to identify 94.6% of the arthropods consumed, classified into 12 taxonomic groups. The most frequent arthropod groups were insects (present in 90% of the stomachs with arthropods), followed by spiders (Arachnida, Araneae, 3.7%) and myriapods (0.8%). Ants (Hymenoptera, Formicidae) accounted for 36.8% of all records, followed by beetles (Coleoptera, 19.4%) and grasshoppers (Orthoptera, 13.6%). The frequency of other arthropods ranged from 0.4% for earwigs (Dermaptera) to 5% for caterpillars (Lepidoptera larvae) and wasps (Hymenoptera, Vespidae). VOL. 52(3) 2022: 208 -217 ACTA AMAZONICA In terms of arthropod richness, rarefaction curves indicated that our sample size was appropriate for estimating the arthropod richness consumed by most primate taxa, except for Lagothrix (n = 30) and Leontocebus (n = 3) (Figure 1). The highest richness was observed in Lagothrix, Cacajao and Sapajus (Figure 1), with at least nine arthropod taxonomic groups ingested (Table 1); beetles and ants represented 40-65% of all records, followed by grasshoppers with 10-61%. On the other hand, only ants were found in the three Leontocebus specimens, the genus with the lowest richness and a null Simpson diversity index value. Simpson diversity index values of 0.8 indicated the highest arthropod diversity was consumed by Sapajus, Cebus, Lagothrix, Saguinus and Cacajao, while the lowest index was obtained for Ateles, followed by Pithecia and Alouatta, which tended to have moderate diversity indices ( Table 1).
The two PCoA axes explained 76% of the distribution of individuals in the ordination space based on the composition of arthropods in their diet (axis 1: 47%, axis 2: 29%). The composition of arthropods consumed by the different platyrrhines were clearly differentiated (ANOSIM R = 0.21, p < 0.01; Table 2), as data from Cacajao, Cebus, and Sapajus were concentrated on the right side of the first axis ( Figure  2). Cacajao, Cebus and, to some extent, Sapajus are clearly separated from the other genera in their consumption of spiders, bugs, wasps, caterpillars, and, most importantly, beetles and grasshoppers. Cebus also differed in consuming less ants than the other genera.
We did not find any significant association between primate body mass and arthropod consumption in any of the scenarios analyzed (Figure 3). In scenario 1, which included all nine primate genera, the correlation between the relative content of arthropods showed a tendincy to a negative relationship, while the other variables showed a slight tendincy to a positive relationship. In scenario 2, which included only medium and large-sized primates, all variables tended to a negative relationship, as expected in the theoretical models. However, none of the correlations strongly supported the predictions of the Kay and Jarman-Bell models.

DISCUSSION
Platyrrhines in general have frugivorous diets which are supplemented by other foods such as leaves, but also arthropods (Chapman et al. 2012;Hawes and Peres 2014). Here, we have shown that arthropods are consumed frequently, but not in large volumes. We found a wide range  ACTA AMAZONICA of arthropods consumed by the platyrrhine genera studied, with differences of up to 2,000 times in the relative content of arthropods consumed among the genera. Although terrestrial arthropod consumption was highest in the Cebidae and Pitheciidae, it was also important for Atelidae, especially for the large-bodied Lagothrix. Our findings support direct field observations of intentional consumption of arthropods being typical of cebines Cebus and Sapajus (e.g., Fragaszy et al. 2004;Oliveira et al. 2014), the callitrichines Leontocebus and Saguinus (e.g., Peres 1993), the pitheciids Pithecia and Cacajao (e.g., Ayres 1989;Heymann and Bartecki 1990;Barnett et al. 2013Barnett et al. , 2017, and the ateline Lagothrix (e.g., Stevenson et al. 1994). Although the literature provides some reports of intentional consumption of arthropods by Ateles (Link 2003;Santos-Barnett et al. 2022), our small sample size for this genus did not allow us to confirm whether the consumption of arthropods in this genus is intentional or not. On the other hand, arthropod frequency and richness in the stomach contents of the more folivorous platyrrhines (e.g., Alouatta, Crockett 1998) may indicate that arthropod consumption can be intentional in these primates. Further studies are needed to clarify this issue and to understand the nutritional implications of this ingestion.
The arthropods found most frequently in the stomach contents were insects from the orders Coleoptera, Hymenoptera and Orthoptera, also reported in other studies as the main taxa consumed by primates (McGrew 2014;Rothman et al. 2014;Rowe et al. 2021). These same taxa represent the PCoA loadings with the greatest effects on the separation of groups in terms of the composition of arthropod taxa consumed. While the frequent consumption of Coleoptera and Orthoptera clustered Cacajao, Cebus and Sapajus, the higher occurrence of Formicidae (Hymenoptera) clustered the other genera. These differences suggest differential strategies among coexisting Amazonian primate taxa to reduce food niche overlap (e.g., Lyke et al. 2019) and to seek and acquire prey.
Ants (Hymenoptera, Formicidae) were eaten by all primate genera evaluated, corroborating the recent finding that myrmecovory is widespread among 18 of the 22 platyrrhine genera (Ferreira et al. 2021). However, the higher frequency of ants in the diet of the genera furthest to the left-hand side of the PCoA ordination space, particularly Alouatta and Pithecia, may point to a more opportunistic or accidental consumption of these arthropods. Ants are individually small, but they are extremely abundant and gregarious social insects that can be eaten in large quantities during a single feeding event (Redford 1987; McGrew 2014), or even accidentally eaten with fruits or leaves (e.g., Ayres and Nessimian 1982). On the other hand, the clear separation of Cacajao, Cebus and Sapajus due to the proportionally higher frequency of Coleoptera and Orthoptera in their diet suggests more specifically developed visual and capture capacities, as these insects are fast in flying and jumping (Rothman et al. 2014), and are usually large and very active (Ozanne and Bell 2003).
Other arthropods may have only been consumed opportunistically or accidentally, so explaining why they appeared at very low frequencies in the primate stomachs analysed. Flies (Diptera) and fig wasps (Hymenoptera, Vespidae) may be accidentally consumed along with fruits (see Redford et al. 1984;McGrew 2014;Lyke et al. 2019). Some fly families cause infestations in ripe fruits (e.g., Tephritidae and Lonchaeidae, Souza-Filho 2019) or insect-induced leaf galls (Redford et al. 1984;McGrew 2014), and fig wasps' life cycle is mutualistic with figs (Ficus spp., Moraceae) (e.g., Weiblen 2004). On the other hand, primates may prefer seeds/fruits infested with arthropods from which they can obtain a mixture of energy and protein (Redford et al. 1984), as recorded for Cacajao ouakary (Barnett et al. 2017), Ateles chamek and A. marginatus É. Geoffroy, 1809 (Santos- Barnett et al. 2022). Cacajao ouakary also consumes insects embedded in rotting wood, tearing through dead wood with their canines and extracting insects with their teeth, a behavior pattern very similar to removing seeds from hard-shelled fruits (Barnett et al. 2013).
Our results did not statistically support the predictions of the Jarman-Bell and Kay models, either in terms of proportion of stomach content, diversity or richness of arthropods consumed by Amazonian primates in the two scenarios analyzed. Observational studies on congeneric species of small primates (< 500 g, Saguinus and Leontocebus) show that, contrary to Jarman-Bell and Kay predictions, these primates consume less arthropods than medium and large-sized primates, naturally deviating from the models. Furthermore, despite their small body size, these primates have a large caecum and colon, similar to folivorous species (Ferrari et al. 1993), suggesting a lower dependence on faunivory. Even so, arthropods comprise between 5.7 and 12.4% of the diet of Leontocebus fuscicollis Spix, 1823 and Saguinus mystax Spix, 1823, respectively (Peres 1993). The greater arthropod richness and diversity that we found for Saguinus specimens agreed with the greater levels of Leontocebus faunivory reported by Peres (1993). The changes in the coefficients of determination of correlations when these genera were excluded are compatible with the abovementioned evidence for lower arthropod consumption in Leontocebus and Saguinus.
Predictions of models based solely on body mass do not match with the dietary needs and morphological adaptations in platyrrhines, although there is a tendency for an inverse relationship in the relative content of arthropod consumed, providing tentative support for the models for medium-and large-bodied primates. It is likely that these relationships are neither simple nor linear, and other factors are involved in the evolutionary process determining the dietary requirements of primate species. Beyond body mass and morphophysiological digestive adaptations, cognitive skills and the availability of food resources are interrelated factors that are likely to be important in shaping feeding strategies (Hartwig et al. 2011;Raubenheimer and Rothman 2013). All these complementary factors need further investigation.
One limitation of our study was that stomach contents may underestimate the relative content and richness of ingested arthropods, as these are the undigested remains of chitinous arthropod structures that allows taxonomic identification only up to order or family level (e.g., Ozanne and Bell 2003). Soft-bodied forms, such as caterpillars and Figure 3. Pearson correlation between the mean body mass of Amazonian primate genera and three arthropod consumption variables for two scenarios: 1) including all nine genera (small-, medium-and large-sized), and 2) including seven genera (medium-and large-sized). A -mean relative content of arthropod in stomach contents (Scenario 1: r s = -0.2333, p = 0.55; Scenario 2: r s = -0.5357, p = 0.22); B -Simpson diversity index (Scenario 1: r s =0.3051, p = 0.42; Scenario 2: r s = -0.0371, p = 0.93); C -estimated richness of consumed arthropods (Scenario 1: r = -0.0438, R² = 0.2%, p = 0.93; Scenario 2: r = -0.2455, R² = 6%, p = 0.59). This figure is in color in the electronic version.
other larvae, as well as aphids, may also be underestimated because they have fewer chitinous body parts. Therefore, the use of stomach content of arthropods as an estimator used in our study likely introduced a bias towards forms with heavily sclerotized exoskeletons in detriment of lightly sclerotized arthropods.
Another limitating factor in the study is the opportunistic sampling by local subsistence hunters, which does not allow to control the number of sampled individuals. In the scenario in which our study was carried out, primate hunting by itself is a seasonal event, which occurs with greater intensity in periods of flooding, when fishing yield (the main source of animal protein in the participant communities) decreases (Endo et al. 2016). Thus, our samples are not distributed in a way that allows to control seasonal differences in primate diet due to availability of food resources or the effects of environmental variables. Seasonality effects should be explored in future research, to better understand the consumption of arthropods by frugivorous primates.
Finally, the participation of hunters by donating samples that originate from their usual subsistence activities, allows communities to contribute to research without encouraging any further killing of animals. As such, the use of donated samples can be a useful alternative to study several ecological aspects of wild animals (Mayor et al. 2017;Jesus et al. 2022), including the diet of hard-to-observe frugivorous arboreal primates.

CONCLUSIONS
We described the patterns of consumption of arthropods in nine frugivorous Amazonian primate genera, as well as the diversity of arthropods in the diet of these species. The frequency, relative content, richness, diversity, and composition of arthropod taxa consumed showed that arthropods are important complementary food items that may allow these species to meet essential nutritional requirements. In this sense, even arthropods consumed accidentally can provide nutritional benefits. Arthropods may also complement the diet when fruit resources are scarce or absent. The study of arthropod consumption by frugivorous arboreal primates remains a challenge, but the use of alternative sampling methods, such as the community-based sampling employed here, in combination with direct observations of primate foraging behavior in the field, have the potential to generate well-grounded knowledge on the feeding ecology of primates.