The Making of Platyrrhine Semifolivores: Models for the Evolution of Folivory in Primates

Body Size

Mammalian foliage eaters, including primates, are expected to attain a relatively larger body mass than their frugivorous or insectivorous relatives (see Eisenberg, 1978, 1981; Harvey, Clutton-Brock and Mace, 1980; Kay and Covert, 1984; Harvey and Clutton-Brock, 1985; Garber, 1987; Fa and Purvis, 1997; Lambert, 1998). Among primates, the examples of relatively larger-sized indriids, gorillas and siamangs are cases in point. As a group, atelines are larger than other platyrrhines; the smallest species of Howlers are at least 25% heavier than the largest living pitheciine (see Fleagle, 1999). But as Strier (1992) emphasized, there is a wide gap in size between Alouatta and Brachyteles that is filled by intervening frugivores, Ateles and Lagothrix (Fig. 1). Furthermore, while Alouatta appears to be more specialized as a folivore than Brachyteles in a number of ways, it is considerably smaller in body mass. Interspecific comparisons also indicate body size is quite labile among the ecologically flexible, widely distributed Alouatta, where local conditions may exert directly selective pressures to maintain an ancestral size, increase, or decrease it. Overall, the body size ranges demonstrated by atelines suggest a more complicated picture of size evolution than one expects based on the folivory model as proposed.

The roughly 5–10 kg weight range exhibited by Alouatta and Brachyteles is also eclipsed by the largest extinct platyrrhines known thus far. Improved calculations based on a robust sampling of platyrrhines suggest the alouattin Protopithecus weighed over 20 kg (Halenar, 2011a). Caipora, an atelin, is also in that size class (Cartelle and Hartwig, 1996). Both of these animals have teeth that are clearly nonfolivorous (Rosenberger et al., in press). This, too, highlights the point that the two most typically folivorous platyrrhines are certainly not the largest in body mass. The 5-k weight figure, approximating the lowest body weight for leaf-eating platyrrhines, may also not represent a functional threshold. About one third of the diet of the ∼1 kg Aotus reportedly includes leaves (Fernandez-Duque, 2011), which is about three times the fraction taken in by Ateles and Lagothrix (DiFiore et al., 2011).

Relative Brain Size

Mammalian foliage-eaters or grazers have relatively smaller brains than frugivores (e.g., Jerison, 1973; Harvey, Clutton-Brock and Mace, 1980; Eisenberg, 1981; Barton, 1996), or mammals tending to search for foods such as fruits, seeds or vertebrate prey (MacNab and Eisenberg, 1989). As a primary example among primates, leaf-eating colobines have relative brain sizes that are estimated as 35% smaller than the more omnivorous cercopithecines and the apes (Martin, 1984). In their primate wide study, Harvey et al. (1987) noted that none of the 6 of 19 subfamilies they examined that had folivorous species ranked in the top ten groups in measures of brain size. It is well established that Alouatta also has a relatively small brain (e.g., Schultz, 1941; Hershkovitz, 1977; Harvey, Clutton-Brock and Mace, 1980; Eisenberg, 1981; Stephan et al., 1981; Harvey et al., 1987), relative to overall skull size and body weight, and particularly among the platyrrhines (Hartwig et al., 2011). Its encephalization quotient is comparable with the values found in leaf-eating colobine monkeys (e.g., Aiello and Wheeler, 1995).

Data on brain volume and body mass of Brachyteles are scant (see Hershkovitz, 1977), but it appears that relative brain size in Brachyteles is not reduced when endocranial volume is regressed against skull length which, of course, must also be partially correlated with braincase size. Brachyteles falls comfortably among atelins which exhibit an elevated regression line relative to three alouattin genera for whom measurements are available, including the subfossil skulls of Paralouatta and Protopithecus (Rosenberger et al., in press). This contrast is also evident when regressing endocranial volume against tooth size (the summed area of three lower molars) (Fig. 2). When two- and three-molared callitrichines are included in the analysis, essentially the same picture emerges but the r² value rises to 0.87. Alouatta and Brachyteles are separated convincingly in relative brain size when molars are used as a body size proxy.

Gut Size and Differentiation

Measurements of gut size and differentiation in Brachyteles may not exist. Osman Hill (1962) says the shape of the stomach resembles Ateles, and he (p. 322) emphasizes that it is “relatively enormous.” In summarizing Hill's findings, Bauchop (1978) noted the anatomical evidence merited speculation that Brachyteles has gastric modifications for folivory. In addition to stomach size and morphology, he noted an enlarged parotid gland and an extremely large caecum “resembling a second stomach.” Muriqui have very large pot bellies, like Ateles, which genus also has an enlarged caecum and a sacculated colon (Hill, 1962).

Quantitative studies of the relationship between guts and diet by Chivers and Hladik (1980), however, do not place convincingly Alouatta among the folivores, although the evidence is equivocal. In their principle diagnostic ratio relating to the potential for foregut fermentation, the Coefficient of Gut Differentiation (surface area of stomach + caecum + colon/ surface area of small intestine), the Alouatta samples differ. A. seniculus falls squarely in their frugivore category, in the same bracket as Ateles and Lagothrix, as well as Aotus and Saguinus. A. palliata has higher values, falling in the bottom tier of their folivore group along with several colobines and cercopithecines. This category also overlaps the frugivorous taxa with more differentiated guts. Using the Chivers and Hladik data and several additional sources to expand the sample of New World monkeys (Fig. 3), plotting the coefficient against body length demonstrates little difference among the atelines, although A. palliata is quite distinct. These data confirm the similarity in proportions between Saguinus, Aotus, and Alouatta seniculus and also demonstrate comparably differentiated guts in Callicebus and Pithecia.

Passage Rates

As reviewed recently by Lambert (1998), the passage rates of digesta are slower in folivores than in frugivores, which allow the former more time to ferment, extract and absorb nutrients from hard-to-process leaves. Here, there is good data on a range of platyrrhine genera (Fig. 4) via projects pioneered by Milton (1984), an effort (done on captive animals) that surely deserves to be replicated to verify and perhaps adjust some measurements that proved difficult due to the physical condition of the individuals at time of testing (see also Edwards and Ullrey, 1999). However, they demonstrate that Alouatta have slow passage rates while Brachyteles exhibits a more rapid transit time, as does the highly frugivorous Ateles. Norconk et al. (2002) confirm that the passage rates of Pithecia, also with a well-differentiated gut, are also notably slow in comparison to other NWM.

Dental Morphology

The incisors and molars of primates in general, and platyrrhines in particular, have received much attention in assessments of functional dental morphology and diet (e.g., Hylander, 1975; Kay, 1975; Rosenberger and Kinzey, 1976; Kay and Hylander 1978; Eaglen, 1984; Kay and Covert, 1984; Rosenberger, 1992; Anthony and Kay, 1993; Cooke, 2011). Two of the prominent finds are that folivores have relatively small anterior teeth, presumably because of the unique handling requirements of a flexible, two-dimensional “mat” of material where involvement of the lips may be useful (Lucas, 2004), as well as large shearing molars. This combination is well illustrated by plotting incisor size and shearing length quotients derived from the residuals of bivariate regressions which relate each of the variables to body size proxies (Anthony and Kay, 1993). Among the three-molared platyrrhines, Alouatta and Brachyteles stand apart, particularly on the y-axis which represents upper relative incisor width (Fig. 5). The separation of the semifolivores in molar shearing potential is less marked, as Alouatta overlaps with Aotus. Cooke (2011) reached a similar finding in assessing the three-dimensional morphology of platyrrhine second molars, wherein some multivariate components associated folivores with the insectivorous-frugivorous Saimiri. Among the semifolivorous species measured by Anthony and Kay, there is also a more complicated association between these variables than expected given the biomechanical explanations behind these features. For example, the Muriqui, which appears to be less committed to a leafy diet, exhibits a higher value for molar shearing potential than any of the three Howler species in the sample.

It is not clear if this counterintuitive result stems from a faulty model or actually reflects adaptive compromise. One might argue the latter; Brachyteles compensates for having a gut less specialized for chemical processing by evolving a molar battery better designed for shearing. Perhaps feeding experiments can be developed to test this hypothesis. Alternatively, the shearing potential measure may be less robust than desired. It may be a “noisy” measure at low taxonomic levels.

Activity Budgets and Positional Behavior

Minimizing energy expenditure by maintaining a low metabolic rate is logically associated with nutritionally limited food sources. Thus McNab (1978) demonstrated that herbivory or folivory is highly correlated with hypometabolism across mammals. Kurland and Pearson (1986), however, argued that this is not clearly the case for primates, as other nondietary factors may be at play. Later, McNab (1986) noted that the basal metabolic rate of the leaf-eating Colobus is not unexpectedly low, perhaps because it exhibits a high level of activity. On the other hand, experienced fieldworkers see it differently: Oates (1994:100) says Black-and-white colobus are “…noted for their inactivity.” Either way, reducing energy consumption by any means would be advantageous if energy input is limited. Activity budgets measured indirectly in the wild support this notion. In a broad survey, Milton and May (1976) showed that primate folivores have smaller home ranges than frugivores or omnivores. This is borne out by more closely examining African colobine species (Oates, 1994), where the more folivorous forms (Black, and Black-and-white colobus) tend to have smaller ranges than more eclectic feeders (Red colobus). Among the NWM, there is a marked contrast in use of time and space between Alouatta and the frugivorous atelins (Table 2). By far, Howlers spend less time eating/foraging, moving, and traveling each day. In time spent eating/foraging and moving, Brachyteles is intermediate between Alouatta and Lagothrix and Ateles, but the contrast with Howlers is less marked. In measures of day range, Alouatta has exceptionally low values as well.

As a corollary to overall activity, summaries of platyrrhine locomotion make it clear that while detailed behavioral information on Brachyteles is lacking, field work indicates the Muriqui is a much more agile, speedy, creative, and flexible locomotor than Alouatta, a genus notable for its frequently lethargic pace (e.g., Youlatos and Meldrum, 2011). This is borne out by anatomical studies. For example, (Halenar, 2011b) showed that the three-dimensional morphology of the Howler elbow joint was distinguished clearly from Brachyteles, which overlapped with Spider monkeys (Fig. 6).

Morphology, Behavior, and Platyrrhine Semifolivory

Comparative empirical studies across the primates and other mammals have noted a pattern of anatomical and behavioral features associated with a leafy or herbaceous diet. Core features that can be assessed in leaf-eating platyrrhines include, body size, brain size, incisor size, molar size and morphology, gastrointestinal morphology, and ranging patterns and locomotor profiles which relate to energy expenditure.

Alouatta, long considered the most folivorous platyrrhine, exhibits the expected adaptations in teeth and brain but differs from the most folivorous catarrhines, the colobines, in lacking foregut fermentation. To distinguish the Alouatta pattern of leaf-eating, several have characterized Howlers as if they are only partially adapted to folivory, “behavioral folivores” in Milton's (1978) formulation, requiring the animal's to be very judicious in selecting leafy foods. To emphasize the more complete picture of the Howler diet and the potential for multiple sources of selection, compound terms have also been used, such as frugivore-folivore (Rosenberger, 1992; Cooke, 2011). Since then, the leaf-eating habits of Brachyteles have also become well known. It, too, differs from the colobine model, and both genera appear to have evolved folivory independently. Thus, while following Milton's lead we prefer to regard these platyrrhines as “semifolivores.” Other terms may be more apt, but the concept of semifolivory may also be widely applicable to include other primates that have evolved only part of the amalgam of characters already generalized as foliage-eating adaptations. One point it is intended to suggest is that primates may benefit ecologically in different ways by evolving a leaf-eating habit. Adaptations to acquire and process food may be paramount, but they do not evolve in a vacuum. A broader notion of folivory may also lead to informative reconstructions of the conditions and pathways under which leaf-eating evolved in different groups, as seen below. The study of primate dietary adaptation has over the years witnessed a similar need to redefine or modify the concept of frugivory, viz. seed-predation and sclerocarpic frugivory (e.g., Kinzey, 1992; Rosenberger, 1992).

The phenotype of platyrrhine semifolivores appears to accord well with the inference that mammalian foliage-eaters tend to be larger than their nearest relatives. However, this generalization only applies when comparing Alouatta and Brachyteles with modern forms, which is a necessarily skewed sample for it lacks historical context. It also does not fit the pattern of body size distributions among atelines. In fact, within the speciose genus Alouatta, there is a large range of sizes. As pointed out by Strier (1992), some forms are considerably smaller than the obligate frugivores Ateles and Lagothrix. There is a possibility that the morphotype of the Alouatta-Stirtonia-Paralouatta clade was reduced in body size from a larger alouattin ancestor. This prospect relates to Protopithecus, which is roughly twice the size of other large alouattins while also being the most primitive in both its craniodental and postcranial morphology (Hartwig and Cartelle, 1996; Halenar, 2011b). Indeed, once a balance of adaptations for folivory was established in the Alouatta lineage, it might have been beneficial to reduce energy consumption by lowering absolute body size up to a point, so long as the advantages of allometrically large teeth and guts were maintained.

Regarding relative brain size, an equally if not more complex picture emerges when assessing the platyrrhine semifolivores. Alouatta meets the expectations of the generalized model in having a relatively small brain, but Brachyteles does not. It is fair to attribute the Howler condition to selection for a smaller energy-hungry cerebrum as an accommodation to a nutritionally limiting diet. In this hypothesis, the Alouatta brain could be seen as de-encephalized, derivedly reduced in connection with folivory. In its specifics, this is consistent with Aiello and Wheeler's (1995) Expensive Tissue Hypothesis as they point out: there is an inverse relationship between gut size and brain size in primates and Howlers are their prime example wherein guts are derivedly large and brains are correspondingly small.

However, the selective agency of a folivorous diet may be only part of the story. Two additional NWM genera have relatively small brains, Callicebus and Aotus (see Hartwig et al., 2011). There are several possible explanations for this. As Dunbar has argued most recently (e.g., Harvey et al., 1980; 1998), monogamous primates have relatively smaller brains than species exhibiting larger social groups. But it may also be that diet plays a selective role here as well. For, like Howlers, Titi and Owl monkeys have rather well differentiated guts, suggesting a digestive adaptation to a nonpredaceous diet. As neither Aotus nor Callicebus have an elevated preference for leaf-eating in the manner of Alouatta or Brachyteles, another dietary constituent may be involved, perhaps seed coats (testa), which have also been implicated as a source of hard-to-digest secondary compounds among colobine monkeys including lignins, condensed tannins and toxic phenolics, (e.g., Chivers, 1994; Kay and Davies, 1994). Indeed, Waterman and Kool (1994) note that seeds may have higher concentrations of toxins than leaves.

In the context of Callicebus and Aotus, we cannot discount the possibility that at least a portion of the smallness of Alouatta's brain reflects a dietary preadaptation, a retention or predisposition stemming from a prior adaptation to a frugivorous habit that involved seed-eating (see below) or a liberal, facultative mixture of seeds and leaves as staple protein sources. Maisels et al. (1994) have shown that two sympatric species of folivorous Colobus in Zaire consume large quantities of both leaves (61%, 50%) and seeds of legumes (33%, 27%), attesting to the likelihood that local preferences reflect satisfactory adaptive trade-offs between these two foods made possible by gut specializations relevant to processing both.

The two clades of atelid NWMs have evolved seed- and leaf-eating specialists within the context of a broadly frugivorous, nonpredaceous framework. Outside the atelines, seeds account for about 60% of the diet of pitheciins, the highly specialized seed-predators (e.g., Kinzey, 1992; Rosenberger, 1992; Norconk, 2011). Thirty-five percent of the diet of the smallest indisputable atelid, Callicebus, which is nearly a fifth the body mass of small Alouatta, includes seeds and leaves. Even the comparably small, nocturnal Aotus is reported to feed on 32% leaves (Fernandez-Duque, 2011), although we caution that the evidence here is seriously limited. This points to a broad capability among the least dentally derived pitheciines to digest seeds and leaves together in combination, much as colobines apparently do. Thus the smallness of Aotus and Callicebus brains may also be a manifestly primitive character state of atelids, but still linked to diet.

The alouattin fossil record (Rosenberger et al., in press) is consistent with this view and introduces yet another factor to consider. The more primitive alouattins, Protopithecus and Paralouatta, are both relatively small-brained, and they lack the reduced incisors and shearing molars of Alouatta and Stirtonia, the fossil most closely related to living Howlers. This means the evolution of relatively small brains preceded the evolution of a highly folivorous dental complex in the clade. The explanation for relatively small brains in these apparently frugivorous monkeys is an open question. The fossil material is insufficient to test for sexual monomorphism, so small group size (e.g., Barton, 1996; Dunbar, 1998) cannot be invoked with credibility. Even if the single skeleton of Protopithecus is a large-canined male, as it appears to be, small brains in Alouatta are associated with the highest levels of sexual dimorphism exhibited in NWM (e.g., Plavcan and Kay, 1988), and their mating systems center around one-male and multimale groups, depending on the species (e.g., DiFiore et al., 2011). A notable size difference between the two known distal humeri attributed to Protopithecus has also been interpreted tentatively as evidence of size disparities within the species (Halenar, 2011a), which could be an effect of sex dimorphism. Equally tenuous, though meriting consideration, is the supposition that small brain size in Protopithecus is (developmentally) connected with or constrained by a morphological pattern relating to the evolution of the howling mechanism (see Hartwig et al., 2011), which seems to have been an important factor in their cranial design (Rosenberger et al., in press). Extending the Aiello and Wheeler (1995) model, we also cannot rule out the possibility that one or both of these large-bodied, more basal, bunodont alouattins already had relatively enlarged guts as a precocious, expensive-tissue adaptation to selection for seed-eating, which would benefit from a similar digestive strategy.

Irrespective of these uncertainties, it is clear Brachyteles does not have a reduced relative brain size, so de-encephalization is not a prerequisite to evolving or maintaining a semifolivorous feeding regime. And this goes to a larger point. A multiplicity of selective factors must be satisfied when producing either large or small brains, or any shift from the ancestral condition, depending on the particulars of a species' overall ecological strategy. In the case of atelines, Dunbar (1998) may be correct that a relatively large brain is part of the selective package necessary to maintain a large, complex social group in Brachyteles, which may involve dozens of individuals (e.g., DiFiore et al., 2011). Barton (1996) has also suggested that brain size is positively correlated with frugivory, while promoting the idea that multiple ecology-based selective factors are involved in selecting for the size of the brain and its components. If so, the status of Brachyteles is consistent with a frugivorous ancestry as well as large group size. But absent a comparably powerful social vector in Alouatta, where groups tend to be less than one third as large (see DiFiore et al., 2011), there may not have been an overriding selective advantage to increase the size of a primitive, relatively small, relatively inexpensive brain. Thus the particular way in which semifolivory emerges as an adaptive package in any clade is influenced by factors that are not explicitly trophic. The advanced brachiating-like locomotor systems of Brachyteles, possibly an extension of a heritage shared with Ateles (but see Halenar, 2011b), was evidently maintained by selection even as the lineage was shifting towards more foliage-eating, thus constraining the Muriqui not to evolve a more sedentary, small-group lifestyle resembling Alouatta.

Similarly, information on guts and passage rates do not strongly segregate the semifolivores from other platyrrhines, nor does it align them with the colobine pattern. This is an important point, for it suggests a continuity of digestive adaptations may be shared by all atelids, as noted, though exceptionally elaborated in Alouatta. This has implications for interpreting ateline history as well as for understanding the process and pattern behind the origins of folivory in other leaf-eating primates. Chivers (1994), Kay and Davies (1994), and Lambert (1998) proposed that seed-eating might be an intermediate step in the evolution of folivory from a frugivorous ancestry. This would be consistent with the findings reported here. While the teeth of the most advanced NWM seed-eaters, pitheciins, are fully the opposite of what one might suppose as morphologic precursors to an Alouatta or Brachyteles dentition, their guts and passage rates depart from the patterns exhibited by the more insectivorous and predaceous cebids (e.g., Fooden, 1964; Chivers and Hladik, 1980; Lambert, 1998). They suggest special adaptations for hindgut fermentation, as in the semifolivores. The genus Callicebus may thus be an important atelid model for a preadaptive, morphotypic feeding pattern, as its feeding preference combines seeds and leaves in larger proportions than in any other atelid—and its dentition falls into neither the seed- nor leaf-eating structural paradigms (see Rosenberger, 1992; Cooke, 2011).

One of the least equivocal morphological findings recognized in this study is that incisor and molar teeth are superbly sensitive to selection for harvesting and masticating leaves. As different as they are in lifestyle and disposition, relative brain size, locomotor behavior, ranging patterns and activity rhythms, Alouatta and Brachyteles have biomechanically similar teeth. The combination of small incisors and large, crested molars, at a minimum, appear to be a necessary and effective dental predictor of a leaf-eating specialization. Because the gut of Brachyteles is insufficiently known, its physiology cannot be securely factored into this equation. However, in light of the role seed-eating appears to play in selecting for gut adaptations, the possible evolutionary linkage between these two strategies, and the evidence that nonfolivorous alouattins had attained very large body size, it is tempting to extend this point by inferring that the digestive system as a whole is also probably highly sensitive to the demands of feeding on leaves and may precede the evolution of other anatomical systems as folivory becomes full blown.

Ecology of Platyrrhine Semifolivory

Alouatta and Brachyteles seem to prefer fruits but take leaves when necessary. Some have also stressed shifting frequencies in the uptake of young leaves and fruit at different sites and times (see DiFiore et al., 2011). While this notion might evoke a proximate causal explanation of leaves as a “fallback” food, that is not necessarily the most valuable perspective for investigating the deeper evolutionary reasons. Efficient leaf-eating may simply be an integrated feature of overall dietary and ecological strategies in both forms, possibly for different reasons and expedited in different ways.

Nothing is known about the ecological conditions under which semifolivory evolved in alouattins. Given the enormous geographical range of Alouatta and their ecological flexibility, it has been suggested that Howlers evolved as pioneers (e.g., Eisenberg, 1981; Rosenberger et al., 2009), which implies a formative preference for living in marginal neotropical habitats. This also implies that the syndrome arose where there was a limited range of ecological competitors and a persistently low supply of easy-to-eat fruits. However, if the Stirtonia-Alouatta lineage, the only segment of the alouattin clade that appears to be fully committed to leaves [although Stirtonia appears to have less shearing-enhanced lower molars than two species of Howlers (Cooke, 2011)], arose in a lush lowland habitat, being able to eat the most abundant forest product available would likely minimize ecological overlap and competition with the large, resident frugivore guild of Amazonian platyrrhines, particularly as it requires far less daily travel, which has its own intrinsic energetic benefits. In other words, semifolivory in Alouatta may also have evolved partly in response to an abundance of coexisting frugivores. The likelihoods of these two alternative hypotheses may not be distinguishable at this time. But neither seems to require a temporally-based fallback selectional model.

As an Atlantic Coastal Forest (ACF) endemic, it must be assumed that Brachyteles, in contrast, evolved its ecological adaptations in situ (Rosenberger et al., 2009). This region is less rich biotically than the Amazonian lowlands. Consequently, it now supports a maximum of five to six sympatric primates, including Alouatta, that is, half as many species as one finds in many lowland forests. Additionally, the ACF supports two semifolivores living together, unlike Amazonia, where Alouatta is not partnered with any other folivorous primate genus. This implies low fruit productivity is a large factor in selecting for folivory in the largest NWMs inhabiting the region, even though Brachyteles would appear to be predisposed by heritage and habitus to prefer fruits. It apparently does in primary forest habitats when they are available, eating 59% fruit and 32% leaves (de Carvalho et al., 2004), almost reversing the proportions seen elsewhere (Table 1).

One would have to assume that the presence of Alouatta in the ACF is simply a local ecological derivation of its larger geographical situation. Brachyteles, on the other hand, is best seen as evolving semifolivory as an overprint upon a highly frugivorous plan that is more comparable to Ateles and Lagothrix than to Alouatta. If its larger incisors, relative to Alouatta, are taken as an indicator, they may signal that Brachyteles is still more tied to fruits. Thus, given the more intense seasonality of the ACF, semifolivory in Brachyteles may be more of a fallback strategy geared to the regularity of intensely lean periods rather than the selective benefits of niche separation among an aggregate of syntopic primate frugivores, which are absent from the ACF. Its ecological separation from Alouatta is already made possible by differences in locomotor skills and ranging habits. But the fallback metaphor should not be taken too far. We know too little about the nutritional requirements of large-bodied atelines (see Felton et al., 2009), or the nutritional potential of ACF trees, to presume there is a temporal phenological rhythm that selects for leaf-eating in Brachyteles. Ingesting leaves in relatively high proportions may be a fundamental ecological strategy, more related to space than time.

Although it may be appropriate to see the dietary strategies of Alouatta and Brachyteles as a joint preference for leaves over fruit if and when the latter is unavailable, this does not detract from the notion that Alouatta and Brachyteles are each specialized leaf-eaters of a certain kind. That Brachyteles parallels Alouatta only in some important respects while neither exhibits the full complement of colobine-like folivorous adaptations is an indication that there are multiple ways of being folivorous and more than one selective regime behind the phenomenon. In using the term “semifolivore,” we also wish to draw attention to this. Howlers and Muriqui are not distinguished from colobines by an “incomplete,” anatomical commitment to leaf-eating, as if there is a single scale with which folivory is measured and a single adaptive strategy for its evolution in different lineages. The platyrrhines have simply done folivory differently. As we argue, heritage and ecology may have provided alternative historical factors determining or guiding the dietary potentials of both genera in shifting away from their apparently frugivorous ancestors. This raises several issues and questions.

There is some evidence that the fruit yield of tropical forests of Africa and South America (Ganzhorn et al., 2009) are not equivalent in terms of potential nutrients for the arboreal primates. New World fruits may provide a richer source of protein than their Old World counterparts, especially on Madagascar. Ganzhorn et al. propose this as an explanation for the greater taxonomic abundance (and anatomical diversity) of platyrrhine frugivores, which exceeds both catarrhines and strepsirhines. Across the hemisphere in Africa, a lower protein concentration in fruits may also help explain the evolution of two prominent alternate strategies. In the face of enhanced critical resource competition, there may have been: (1) greater selective pressure among arborealists to process leaves in full by evolving elaborate gastric specializations that actually compromise their capacity to digest fleshy fruit (see Kay and Davies, 1994) and (2) a stronger evolutionary impetus to forsake the trees altogether and develop terrestrial lineages within, and ultimately outside, the rain forests. As a corollary to the latter point, platyrrhines may have been less prone to terrestriality if the richness of fruits permitted more extensive resource partitioning within their comparatively smaller body size bracket.

Synthesis: The Making of Platyrrhine Semifolivores

If so, the platyrrhine situation, where semifolivory appears to have evolved twice via parallelism, may be the consequences of a specific shared heritage and a deeper evolutionary explanation. African leaf-eating colobines may be represented by the same number of genera, two (Colobus and Procolobus), but these are apparently sister-taxa that differentiated through a single speciation event. Alouatta and Brachyteles represent two genera that evolved semifolivory within different ateline subclades, the alouattins and atelins. Why? How? How is it that this happened twice among platyrrhines?

Perhaps because the atelid clade within which atelines differentiated may be fundamentally preadapted to evolve folivory, whether or not New World forests are capable of supporting a large frugivorous cohort. Atelines and pitheciines may have shared a common ancestor that was adapted to digesting seed coats and leaves, at least facultatively. While the most dentally derived pitheciines are the seed predators Pithecia, Chiropotes, and Cacajao (e.g., Kinzey, 1992; Rosenberger 1992; Norconk, 2011), even the more primitive and basal genus Callicebus eats more seeds and leaves than any nonpitheciines (see Norconk, 2011) or nonateline (see DiFiore et al., 2011).

Seed coats can have a high lignin content, like leaves, as well as toxins and tannins and other secondary compounds (e.g., Davies and Oates, 1994; Kigel and Galili, 1995; Dixon and Sumner, 2003). Lignin is a chemical compound that is a major constituent of wood, which gives a sense of how indigestible and tough lignified material can be. Thus seeds present some of the same digestive challenges as leaves, in addition to their own unique requirements for harvesting and mastication. Species eating seeds may benefit from comparable digestive adaptations, including guts that are more differentiated than is typical among other frugivores and insectivores, to allow fermentation and detoxification. As noted, this and the high percentage of seed-eating in Colobus (see Fashing, 2011), led Kay and Davies (1994), Chivers (1994), and Lambert (1998) to suggest seed-eating as an intermediate dietary step between frugivory and folivory among cercopithecoids. The same may hold true for platyrrhines.

If ancestral atelids were prone to feeding on seeds and evolved guts to accommodate this pattern, atelines would likely have retained this condition in their last common ancestor. This would mean they were originally preadapted for a shift toward leaves, which in turn would have increased the likelihood of parallelism. A seed-eating capacity may, in fact, be more than latent among atelines, as Lagothrix and Ateles ingest seeds. Although dietary tabulations appear to deemphasize seed-eating in atelins by comparison with pitheciines, Ateles and Lagothrix are known to be prodigious and effective dispersers of seeds (Stevenson, 2000; Link and DiFiore, 2006). This implies a gut subject to and capable of degrading the inherent secondary compounds of seed coats. Their large pot-bellies, which resemble the distended gut of Brachyteles in outward appearance, are evidently capable of retaining a large volume of seeds. And, although the diet of Ateles may involve only a fraction of the leaves eaten by an Alouatta or Brachyteles (e.g., DiFiore et al., 2011), leaf-eating is a regular activity even when fruits are available (8%; Stevenson et al., 2000). This adds to the notion of a continuum between seed- and leaf-eating among the atelids.

A second, hardly separable adaptive feature that might predispose atelines to leaf-eating is sheer body mass. As discussed above, their comparatively large body size corresponds with a volumetrically enlarged gut, and all the trappings associated with that to benefit processing leaves, such as a slower passage rate and more usable space for cultivating microbial symbionts.

There is some cladistic and paleontological evidence supporting this scenario of a seed-to-leaf shift in the differentiation of the Alouatta lineage. As noted, there are three fossil genera that offer anatomical information pertaining to phylogeny and adaptation of alouattins (Rosenberger et al., in press); a fourth possible alouattin (Solimoea) is a single tooth with consistent but more limited clues. Stirtonia is the one most closely related to Howlers. It is comparable to Alouatta in body size and its dentition is reasonably well known. The design of its molars conforms closely to Howlers and shearing quotient measures indicate a diet “nearly as folivorous” (Anthony and Kay, 1993:356) as Alouatta. Cooke's (2011) geometric morphometric study of lowers molars finds that Stirtonia plots just outside the range of two Howler species. Neither Paralouatta nor Protopithecus, in contrast, have molars designed for shearing and the latter is known to have very wide, large incisors as well, both marks of frugivores. In her classification, Cooke's identifies Paralaouatta as a folivore/frugivore, in contrast with the folivores Alouatta and Brachyteles. Paralouatta is roughly the size of a large Alouatta and Protopithecus is much larger (Halenar , 2011b). As both these genera occupy more basal positions on the alouattin cladogram than Stirtonia, in conjunction with the out-group evidence from pitheciines, they indicate the last common ancestor of alouattins was not an obligate folivore. Nor was it a semifolivore in the sense of Alouatta or Brachyteles, for none of the common dental prerequisites were present. But it may well have been a seed-eating frugivore, possibly even quite capable, facultatively, of digesting leaves in significant proportions because of an allometrically capacious gut.

A Common Scenario: Models For the Evolution of Leaf-Eating in Primates

The evolutionary pathway proposed for the development of semifolivory in platyrrhines presents intriguing parallels with cercopithecoids. As is well known, the two subfamily clades, cercopithecines and colobines, differ fundamentally in diet, as the colobines have evolved chemical and mechanical adaptations appropriate for ingesting leaves (e.g., Davies and Oates, 1994). It is also quite clear that the colobine dental plan is derived (e.g., Szalay and Delson, 1979), that the high-cusped, sharp-crested molar teeth were modified from a blunter design. But underlying this is the cercopithecoid's fundamental dental specialization of bilophodonty, still retained in both groups and requiring only simple modifications to achieve the colobine condition. The biomechanical explanation for bilophodonty has been clarified by modeling and experimentation, and it is evident that this unique morphology is especially favorable to the breakdown of seeds (see Lucas and Teaford, 1994). In other words, seed-eating, or the potential for seed-eating, must have been present as a preadaptation in the colobine morphotype, setting the stage for a dietary shift towards folivory.

Similar dietary transitions may have occurred among the semifolivorous platyrrhines although the hard anatomical evidence is less compelling. There is no manifest dental morphological link between the derived dentition of Alouatta and the ateline morphotype, or Brachyteles and the atelin morphotype, to suggest biomechanical derivation from a seed-eating complex. Their hypothetical molar morphologies do not present a definitive, diet-specific morphology. The clues here are more circumstantial: prevalence of dedicated seed-eating in close relatives, the possibility that gut specializations to enable seed- and leaf-eating being widespread among atelids, and the relatively large body size of atelines, especially extinct alouattin forms which are probably more primitive dentally than the modern genera. If this scenario proves useful, it would be interesting to test for its generality by considering folivory in strepsirhines as well. Do they also show indications of seed-eating antecedents?

We suggest their evolutionary pathway may have taken a different course. Strepsirhines, including the crown group and its basal members, the adapiforms, may have been folivorous from the start. There are two lines of evidence supporting this supposition. The earliest fossil strepsirhines come from the Eocene epoch (e.g., Szalay and Delson, 1979; Fleagle, 1999; Gebo, 2002), although even at that time they were taxonomically diversified and nothing secure is known of their antecedents, that is, what taxonomic group and morphological bauplan provided their foundation. Generally, these early adapids were medium- to large-bodied primates with large crested molar teeth. Some genera, such as the North American Notharctus (e.g., Gebo, 2002) and the African Afradapis (Seiffert et al., 2009), show a stunning morphological and functional correspondence with living leaf-eaters such as Alouatta (Fig. 7) and the living strepsirhines Propithecus and Indri. Like living strepsirhines (see Rosenberger and Szalay, 1980; Eaglen, 1986), their anterior teeth also correspond with modern folivores in being relatively small and morphologically reduced. In the advanced European adapines, the compact, spatulate incisor-canine battery is shaped like croppers.

The extant strepsirhines are dietarily very diverse (see Fleagle, 1999), but one can argue that they are primitively folivorous. Several lines of evidence can be cited to support this hypothesis. Morphologically, if the lemuriforms, as opposed to the more predatory lorisiforms, are good models of the dental morphotype (e.g., Szalay and Delson, 1979), there is continuity in the morphology of molar teeth shared with adapiforms. There is even continuity in the morphology and inferred functional capability of the upper anterior teeth (e.g., Rosenberger, 2010). Vigorous harvesting roles, as expected in fruit-eaters, do not correspond with the reduced incisor morphology of lemuriforms and lorisiforms, especially when a toothcomb is present in the lower teeth. Additionally, there is evidence that some modern strepsirhines, including both lemuriforms and lorisiforms, are hypometabolic (Kurland and Pearson, 1987), which could reflect a primitively folivorous diet as well, since hypometablism is broadly associated with leafy and herbaceous diets among mammals (e.g., McNab, 1986).

Thus the strepsirhines may have been leaf-eaters from a very early point in their career. If their anterior tooth complex was reduced from a more enlarged battery (Rosenberger and Szalay, 1980), this would imply a dietary shift away from the incisal harvesting of fruits and towards leaves. Fruits would have been a more likely preadaptive, preferred food in the ancestors of strepsirhines than a diet emphasizing insects for it is more difficult to posit a strongly predaceous antecedent, as insectivory is associated with a small, fast gut, the opposite of expectations for folivores (Chivers and Hladik, 1980). An analogous, comparable evolutionary constraint may have been at work among platyrrhines, where the predaceous cebids have apparently not produced any folivores while fruit-eating atelids have done so at least twice. The largest cebid known, Acrecebus, was the size of Lagothrix (Kay and Cozzuol, 2006), that is, within the range of NWM semifolivores. Although it is known only from a single upper molar, that tooth is shaped much like the highly bunodont, nonfolivorous crown of its close relative, Cebus, and shows no signs of a shearing design.

Plesiadapiforms: Early Origins of the Seed-to-Leaf Pathway?

The best fossils available for shedding light on the preadapiform morphology are the controversial plesiadapiforms. The coexisting tarsiiforms (see Szalay, 1976; Gunnell and Rose, 2002), the only other primates pertinent to the time of adapidforms, are not relevant since there is growing evidence many were quite specialized ecologically as vertical clinging and leaping, nocturnal predators occupying a broadly construed “tarsier adaptive zone” (Rosenberger, 2011b). They are the smallest-bodied adaptive radiation of euprimates, generally falling well below 500g in weight (see Fleagle, 1999), and must be regarded as an assemblage occupying the insectivorous-frugivorous spectrum—if they were not nearly as predatory as modern Tarsius (e.g., Gursky, 2007). So, what did the taxonomically and anatomically diversified plesiadapiforms eat? This is a difficult matter for which there may be no modern consensus, and a radiation as taxonomically diversified as this is not expected to occupy a single, narrow feeding niche. Insects, fruits, and seeds were likely to have been important in the smaller forms, with their comparatively bunodont molars (Szalay, 1968; Block et al., 2007). Leaves may have been emphasized in the larger forms, including some of the Plesiadapidae (Gingerich, 1976; Bloch et al., 2007; Boyer et al., 2009), which may have been roughly 1–3 kg in body weight (Fleagle, 1999). But the most prominent and puzzling dental characteristics of plesiadapiforms are their moderately large to enormous incisors. The more robust varieties of the plesiadapiform incisor complex would suggest vigorous harvesting behaviors, but gleaning, probing, and picking activities would also have been part of the repertoire where the incisors were proportionately smaller and more delicate and more horizontally oriented (Rosenberger, 2010).

Definition of the plesiadapiforms' preferred food targets depends on clarifying both parts of the organism-environment equation, one dealing with plants and the other with plesiadapiform evolutionary morphology, as well as assumptions derived from current adaptational models. Wing and Tiffney (1987) point out that the consequences of the K/T boundary event would have “destroyed” the existing ecological interactions between angiosperms and herbivores. Nevertheless, angiosperms are assumed to have been the major vegetable food source for plesiadapiforms at the time, even though woody flowering plants were not yet in their heyday during the late Cretaceous and Paleocene. Then, they were producing smaller seeds and fruits without significant fleshy coverings, no larger than 1 mm³ to 10 mm³ in size (e.g., Tiffney, 2004; Moles et al., 2005). Many would have been abiotically dispersed dry fruits living in the shrub layer. But there was a large upward shift in seed size from the late Cretaceous to the beginning of the Eocene (Wing and Boucher, 1998), when the coevolutionary relationships with mammalian (and avian) dispersers matured and the angiosperm's developed larger, attractive, fleshy, sugary fruits, and gaudy flowers. Seed size enlarged (Wing and Tiffney, 1987; Tiffney, 2004; Eriksson, 2008), as did the primates. Eriksson et al. (2000) estimates an increase in seed size of 2–3 orders of magnitude by the early Eocene. It therefore stands to reason that the dietary profile of the arboreal plesiadapiforms, as more archaic omnivorous frugivores, would have been invested in the relatively small seeds of the early angiosperm configuration, which the animals would have harvested with their probe-like incisors (Rosenberger, 2010). If some of their targets involved mechanically well-protected seeds, the face-feeding plesiadapiforms would have had to rely quite heavily on incisors for securing food. But without the benefits of large size, we can also assume that the seed coats of pre-Eocene angiosperms would have evolved nonmechanical protectants, allelochemicals that are typically concentrated in seeds and immature fruit (see Kigel and Gali, 1995).

On the plesiadapiform side, since it is evident that folivory was very probably not an option for smaller forms and nutritionally poor wood would have posed even more severe mechanical and chemical disincentives, angiosperm seeds would have been the most abundant edible nonprey material available to them. Morphologically, the essence of this argument is based in the contrasts plesiadapiforms exhibit relative to euprimates in their new Eocene ecological context. The plesiadapiforms were small, nocturnal, smell-oriented, without a fine sense of sight, lacking touch-sensitive nailed fingertips on Rays II–V able to finely discriminate texture, probably wanting the integrated hand-eye coordination that comes with visual acuity and requires high-resolution neural wiring, and lacking the long-limbed, flexible locomotor skeleton that implies limited, finely controlled, free flowing locomotion in the trees (see Bloch et al., 2007).

Collecting small pendant fruits and their seeds would have required no sophisticated handling. The advanced hand-eye coordination and stable arboreal sitting postures that allow dexterous two-handed manipulation and feeding in modern primates—also keys to unlocking large encased seeds like legumes—became efficient only later. As did hands-free, below-branch pedal hanging, which relies on powerful pedal grasping? These capacities may have emerged along with stereoscopy, in concert with the modern euprimate positional behavior system and arboreal balancing mechanisms, all of which would have been associated with the reorganization of the locomotor skeleton of the euprimate morphotype (Dagosto, 1986). This would have coincided with the development of larger seeds and an alteration in vegetation structure, namely the establishment of closed canopy forests in the Eocene (e.g., Eriksson et al., 2000; Tiffney, 2004; Moles et al., 2005). Only then would there be a selective benefit accruing to species able to travel extensively and efficiently through the treetops, foraging for clumped but patchily distributed fruits. If the plesiadapiforms were indeed focused on small seeds, which would have provided lesser nutritional reward per unit and thus require bulky collections, the common occurrence among them of large diastemata behind the incisors may be an indication some also evolved squirrel-like cheek pouches, useful for storing seeds while foraging.

What this speculative scenario implies is that seed-eating—not monolithically, but as a dietetically critical fruit component complimenting insects and in some case leaves—many have been at the root of the primate radiation. If the successful plesiadapiform radiation included consumers of both seeds and leaves, as the dental morphology suggests, it points to an underlying capacity that may have been preadaptive to dietary shifts within a broad zone of digestive tolerance. The original primate digestive system may have adapted early on to the requirements of chemically digesting seeds and nonfleshy fruit, which overlap physiologically with the needs of a leaf-eater. The pathway of seeds-to-leaves may have been set in place early in primate evolution, with the Eocene adapiforms perhaps being the only higher taxon to make a wholesale shift as a basal condition of their adaptive radiation. The anthropoids would later also have been able to make the shift via preadapation but, being diurnal and without much competition in a new adaptive zone, they first moved to broaden their dietary choices by exploiting a new variety of abundant, diversified “low hanging fruit.” If the earliest pre-Eocene plesiadapiform primates (and other vertebrates) developed an adaptive preference for small seeds, as argued here, the modern angiosperms may have evolved a counter strategy to enhance their reproductive success by enveloping their now larger seeds in more interesting, energy rich fruit pulps, thus saving their disseminules from destruction while also enhancing their dispersal potential (see Mack, 2005). Later, as seen in the radiations of New and Old World anthropoids, as niche differentiation became a selective priority in ever more complicated ecological webs, this same seed-to-leaf pathway would produce the semifolivorous platyrrhines and the folivorous colobines as upward shifts in body size made seeds—often in balanced combination with leaves—an efficient option for obtaining protein and other nutrients.