The role of dietary competition in the origination and early diversification of North American euprimates

Abstract

The conditions under which early euprimates (adapids and omomyids) originated and evolved is an area of longstanding debate. The leading hypotheses of euprimate origins promulgate diet as a core component of the early evolution of this group, despite the role of dietary competition in euprimate originations never being tested directly. This study compared three competition models (non-competition, competitive displacement, competitive coexistence) with observed patterns of dietary niche overlap, reconstructed from three-dimensional molar morphology, at the time of the euprimate radiation in North America (at the Paleocene–Eocene boundary). Overlap of reconstructed multidimensional dietary niches between euprimates and members of their guild were analysed using a modified MANOVA to establish the nature of the competitive environment surrounding euprimate origins in North America (an immigration event). Results indicated that adapids entered the mammalian guild in the absence of competition, suggesting dietary adaptations that were unique within the community. Conversely, omomyids experienced strong, but transitory, competition with nyctitheriids, suggesting that omomyids possessed the ability to out-compete this group. These results show that adapids and omomyids experienced different competitive scenarios upon their arrival (origination) in North America and confirm the significance of diet (and dietary adaptations) in euprimate origination and early diversification in mammalian communities.

1. Introduction

Elucidating the adaptive and competitive conditions resulting in the origin and diversification of the first unequivocal euprimate (primate of modern aspect) fossils at the Paleocene–Eocene boundary (ca 56 Ma) is crucial for understanding the course of evolution of the entire euprimate clade, including humans. The two leading euprimate origins hypotheses, the ‘grasping hypothesis' [1,2] and the ‘visual predation hypothesis' [3,4], assert that ‘key innovations' involved in food acquisition (e.g. convergent orbits or grasping hands) were at the root of the initial euprimate radiation—that is, the euprimate dietary niche was a primary driver of their origination. Because key innovations are defined as novel traits that are adaptive [5–8], these hypotheses assume that euprimates first evolved in one of two scenarios: through the exploitation of an open dietary niche (absent competition) or through the exclusion of non-euprimate dietary competitors (strong competition). However, the role that diet played in establishing euprimates as successful members of early mammalian communities has not been explicitly addressed. On the other hand, if dietary competition between euprimates and non-euprimates was insubstantial (weak competition), diet could not have been a driving force in early euprimate evolution.

The Paleocene–Eocene boundary also marks the point of the Paleocene–Eocene Thermal Maximum (PETM), one of the most dramatic peaks in global temperatures in the whole of the Cenozoic [9,10], and a correlation between this climatic event and mammalian taxonomic turnover is well-supported (e.g. [11–14]). The PETM coincides with the extinction or major decline of groups ecologically similar to euprimates, including carpolestids and plesiadapids [15–18]. Other euprimate ecological vicars (e.g. microsyopids, paromomyids, peradectids and rodents), however, persisted through this transition [16,18,19]. Shortly after their immigration to North America, euprimates greatly diversified, indicating an ‘invasion radiation’ of this clade [20–23]. As a result of the dramatic nature of Paleocene–Eocene climatic change and the coincidence of euprimate origination and diversification with the decline of only some likely euprimate dietary competitors, the competitive environment into which these North American euprimates arrived is not clear.

The purpose of this study is to characterize the dietary competitive environment in which euprimates evolved subsequent to their arrival in North America as a means of directly addressing the catalyst for euprimate origination (immigration) and diversification on this continent. Although diet is only one component of an organism's ecological niche (other key components include activity pattern, body mass, substrate use and locomotor repertoire, for example), the dietary niche was evaluated here given its potential underlying significance in early euprimate evolution. Furthermore, diet is likely the most limiting resource among the niche components consistently reconstructed in the mammalian fossil record and thus is arguably the most relevant in the evaluation of competitive interactions [24].

Competitive environments are defined by species interactions, and many models of interaction (including ‘non-interactions') at the macroevolutionary level have been described (e.g. [25–28]). For an invasion radiation, such as the origination of euprimates in North America, first appearing in the Gulf Coastal Plain of Mississippi near the base of the PETM, only three main types of competitive interaction are possible [29–31]. Because competition is defined by niche overlap [32], these models are characterized by the pattern of dietary niche separation between invasive (in this case, euprimate) and incumbent (euprimate potential competitor) species (figure 1). In this study, euprimate potential competitors are defined as sympatric small-bodied, at least partly arboreal, insectivorous–frugivorous mammals (herein, euprimates and their competitors are referred to as the ‘euprimate competitive guild').

Figure 1.

Models of competitive interaction evaluated in this study. (a) Non-competition. Before euprimate origination (Cf2–3), the future euprimate dietary niche in Wa0 is either (i) unoccupied (non-replacement) or (ii) occupied by one or more non-euprimates that will go extinct before euprimates arrive (post-extinction replacement). Upon and after euprimate origination (Wa0, Wa1–2), euprimates are the only taxon that occupies their niches. (b) Competitive displacement. Before euprimate origination (Cf2–3), one or more non-euprimates occupy the future euprimate dietary niche in Wa0. Upon euprimate origination (Wa0), euprimates and non-euprimate(s) share a dietary niche. After euprimate origination (Wa1–2), euprimates and non-euprimate(s) either (i) continue to occupy the same niche, but the double-wedge pattern of inverse abundance is observed or (ii) the niches of euprimates and non-euprimate(s) diverge. (c) Competitive coexistence. Before euprimate origination (Cf2–3), one or more non-euprimates occupy the future euprimate dietary niche in Wa0. Upon and after euprimate origination (Wa0, Wa1–2), euprimates and non-euprimate(s) share a dietary niche in the absence of the double-wedge pattern. Note that the patterns for adapid and omomyid euprimates may differ. Cf, Clarkforkian; Wa, Wasatchian. (Image credit: Bogdan Bocianowski, Mateus Zica (modified by T. Michael Keesey) and Nobu Tamura (modified by T. Michael Keesey). See https://creativecommons.org/licences/by-sa/3.0/legalcode. Images were resized and coloured.) (Online version in colour.)

The first model, non-competition, refers to the absence of incumbent species, which, if present at the point of origination, would be in direct competition with the invasive species (figure 1a). As a result, the invasive species exploits an ‘empty niche' or ‘open ecospace.’ This scenario can take two forms: non-replacement (‘expansion radiation' [28]) occurs when an invasive species enters a niche that had been consistently unoccupied within the community (figure 1a(i)); post-extinction replacement [26] (variably referred to as ‘opportunistic replacement' [15] or ‘incumbent replacement' [32]) is similar to the non-replacement model except that the empty niche is newly available due to the extinction of ecologically similar incumbent species just prior to the invasive species' arrival (figure 1a(ii)).

The second model, competitive displacement [15] (‘competitive replacement' [6], ‘taxonomic displacement' [17,33]), refers to strong competition between species (figure 1b). Interspecific competition is defined as a mutually negative interaction between species due to the presence of a limiting resource [29]. The most common criterion for the identification of competition in the fossil record is the demonstration of the ‘double-wedge pattern' of abundance, as a result of the inverse relationship between abundance profiles of competing species (e.g. between the invasive and incumbent species) [6,34] (figure 1b(i)). Recently, however, more sophisticated mathematical models have been developed that incorporate rates of speciation and extinction in conjunction with diversity dependence and dynamics (e.g. birth–death models) of fossil taxa that allow more refined signals of competition to be detected (e.g. [35–37]). It is also possible that competition may result in evolutionary niche divergence or ‘character displacement', in which the trait morphologies of species diverge in response to competition. In this scenario, temporal morphological change (in this study, molar shape change over time) will occur in the invasive or incumbent species (or both) such that niche overlap decreases via niche divergence (figure 1b(ii)). Thus, competition will be reduced and may eventually cease over time [38–40].

The third model is competitive coexistence [28] (‘diffuse competition' [41]) in which the invasive and incumbent species occupy the same niche, but neither the double-wedge pattern nor niche divergence is observed; thus, competition is weak (figure 1c). Competitive coexistence has been documented in extant communities and has been ascribed to partial niche separation, the presence of only intermittent competition such that neither species is permanently affected, or sustained low-intensity competition [42–45].

The greater Bighorn Basin in northern Wyoming is characterized by an uncommon diversity and abundance of early Paleogene mammalian fossils, including euprimates, and is one of only a handful of such fossil deposits to span the Paleocene–Eocene boundary. To best reconstruct true competitive guilds, the fossil sample comprises specimens collected from a single geological formation (Willwood Formation) within the Bighorn Basin. Although it is thought that euprimate origins predate this community and occurred on another continent [46–48], the Bighorn Basin is arguably the earliest paleocommunity in which the competitive environment of euprimate (both adapid and omomyid) immigration into a North American mammalian community can be assessed, given its large sample size and high diversity of early Paleogene mammals. It is noted that another North American Paleogene mammalian community, the Red Hot fauna of Mississippi, is coincident with the PETM, and includes the earliest known euprimates (omomyids) in North America [30,31]. The Mississippi fauna does not markedly pre- or post-date the euprimate radiation of the Bighorn Basin community, and while significant for understanding euprimate origins in, and dispersal to, North America, it does not provide a sufficiently broad taxonomic context from which to evaluate the hypotheses of the role of competition in euprimate immigration and diversification outlined here.

While communities and guilds cannot be known with absolute certainty in the fossil record, the restriction of the units of analysis in this study to a single geological formation at a single site (a proxy for sympatry) and to narrow time intervals (a proxy for synchronism) minimizes the effects of time- and geographic-averaging, while simultaneously maintaining adequate sample sizes necessary to test the hypotheses herein. This sample was divided into three time intervals defined by sub-NALMAs (North American Land Mammal Ages) of the Clarkforkian (Cf) and Wasatchian (Wa) NALMAs, spanning the time period of euprimate origination (56.1–55.1 Ma): Cf2–3 (prior to euprimate origination), Wa0 (the point of euprimate origination) and Wa1–2 (after euprimate origination) [49,50].

Dietary niches were reconstructed using landmark data of microCT scans of first (M₁) and second (M₂) mandibular molars (figure 2) [51]. These measures are closely associated with dietary regimes and accurately reconstruct dietary competitors in an extant mammal community [51]. Dietary niche overlap among taxa was evaluated using a modified MANOVA of principal component (PC) scores that can identify overlap between N-dimensional niches, as they have been defined and evaluated in previous ecological research (e.g. [51–54]). This analysis does not require knowledge of the nature of the dietary niche (i.e. the actual diet) of each group but only whether dietary niches overlap in multidimensional space, which is particularly advantageous in the study of fossil taxa with no extant analogues. The dietary niche of each taxon (defined temporally within each of the three time periods) is defined as a multidimensional ‘hypervolume,' or the region of the overall dietary niche space occupied by the taxon across the total number of morphological axes of interest.

Figure 2.

Three-dimensional landmarks used in this study, shown on a microCT rendering of a Cantius M₂ in (a) occlusal, (b) buccal and (c) lingual views. Landmarks associated with the principal cusp tips are the largest spheres (shown in red in the online colour version); semi-landmarks are the smallest spheres and are arranged along major occlusal crests (light blue); cervical landmarks, from which cusp heights were estimated, are shown in (b) and (c) along the cervical margin (green spheres); and cusp edges, from which cusp sharpness (angle) was measured, are small spheres positioned on either side of the principal cusps in (b) and (c) (dark blue). A single landmark associated with the deepest point in the talonid basin is also indicated (yellow sphere). The 20 occlusal outline landmarks and three landmarks placed on the occlusal plane are not shown. Landmarks and measurements are defined in [51]. Cusp abbreviations are as follows: pacd, paraconid; prcd, protoconid; hycd, hypoconid; etcd, entoconid; mtcd, metaconid. (Online version in colour.)

2. Material and methods

(a). Fossil sample

The fossil sample comprised 370 mandibular molar specimens from 16 families, representing 8 mammalian orders from the euprimate competitive guild (sympatric small-bodied, at least partly arboreal, insectivorous–frugivorous mammals) (see electronic supplementary material, Table S1). Species included were those with presumed ecological niches most similar to euprimates, and thus those whose possible engagement in dietary competition with euprimates was most likely. All unworn-to-minimally worn available M₂ specimens were included in this study (supplemented by M₁ and M_X (molar position unknown, excluding most distal molar) specimens where necessary), as this molar is most often used as a proxy for diet in the mammalian taxa examined here (e.g. [55–57]). Specimens were each designated to a sub-NALMA based on the stratigraphy defined in [58]. For this reason, Wa1 and Wa2 faunas were combined into a single group (Wa1–2) to coincide with the stratigraphic correlations outlined in that source [58]. Owing to the scarcity of Cf3 specimens in the sample collections, Cf2 and Cf3 taxa were consolidated into a single Clarkforkian (Cf2–3) temporal group.

(b). Morphometric data collection

Specimens were first moulded and then cast using an epoxy resin, and these casts were scanned at a resolution (voxel size) of 27.4 µm using an Inveon microCT scanner (60 kV, 300 µA, 5000 ms) at the University of Arizona Cancer Center. Three-dimensional surface reconstruction of molar crowns from the microCT scans and landmark placements were performed in Amira 5.2.0. A total of eight linear, areal and angular measurements were calculated from the three-dimensional landmark data: total crest length, mean trigonid cusp height, mean trigonid cusp angle, mean talonid cusp height, mean talonid cusp angle, trigonid–talonid relief (postvallid wall height), talonid basin area, and talonid basin depth (figure 2; electronic supplementary material, Table S1). All measurements were scaled to molar size using a ratio of the measure to the two-dimensional molar occlusal area (calculated using semi-landmarks placed around the molar in occlusal view). The full complement of landmarks and semi-landmarks collected, and the subsets of these landmarks that comprise each morphometric measure, are defined in [51].

(c). Analysis of niche overlap

A principal component analysis (PCA) was first performed on all individual specimens for all three time intervals combined in order to reconstruct the overall dietary niche space for this community. In this analysis, all PCs cumulatively accounting for 95% of the variation were included.

Dietary niche overlap was evaluated using a modified pairwise non-parametric multiple analysis of variance (MANOVA; derived from [59]) and is conceptually similar to a between-groups PCA [60]. However, the application here might best be described as multivariate pairwise comparisons using PC scores. This analysis was performed at the family level (with the exception of erinaceomorphans, due to small sample sizes for this suborder); thus, all specimens identified only to the ordinal level were not included. Furthermore, this analysis requires a minimum of three specimens per family per time period, resulting in the analysis of a subset of the total dataset (N = 353) (electronic supplementary material, Table S2). The PCA of this subset resulted in very similar values for all eigenvectors and eigenvalues when compared to the PCA of the larger sample.

The pairwise MANOVA creates an F-statistic calculated from sums of squares (SS) of distances among specimen coordinates in multidimensional PC space. Variance between groups (SS_B) is the sum of squared distances between the coordinates of each specimen and the centroid of the entire sample, and variance within groups (SS_W) is the sum of squared distances between the coordinates of each specimen within a taxonomic group and the centroid of that group (electronic supplementary material, Table S3). A high F-statistic value, therefore, indicates that the variance between groups is greater than the variance within groups, signifying group separation. As such, the null hypothesis of this analysis is that ‘groups occupy statistically similar positions in the multivariate PC space, the ecological interpretation of which is the presence of niche overlap,' a requisite of one or more models of competition [51]. Rejection of the null hypothesis demonstrates the absence of overlap between dietary niches.

(d). Testing for phylogenetic signal

To determine if the molar morphological measures, the basis of the analysis of niche overlap, exhibit phylogenetic patterning, a multivariate generalized K-statistic (K_mult) and associated p-value were calculated [61] (electronic supplementary material, figures S1 and S2). This statistic is similar to Blomberg's K-statistic [62], which tests the strength of the phylogenetic signal in single trait analyses, but has been altered to accommodate multivariate data. This analysis was performed in R v. 3.4.1 following [61] and was conducted at the family level (with the exception of erinaceomorphans; see above) for those specimens included in the MANOVAs to mirror the sample composition of these analyses. The results of this analysis produced a K_mult value of 0.375 (p = 0.63), indicating a lack of significant phylogenetic structure in the data. Thus, the measures used here were assumed to characterize functional differences among molar morphologies, indicating they are sufficient proxies to reconstruct dietary niches of the taxa in this study.

3. Results and discussion

An examination of the eigenvectors indicates that the first six principal components cumulatively contribute to approximately 94% of the variation, and thus, the values of PC1–PC6 were used in the subsequent pairwise MANOVAs. Combined, the eigenvectors of PC1 and PC2 indicate a morphological gradation from the top left (low values of PC1, high values of PC2) to the bottom right (high values of PC1, low values of PC2) quadrants of a bivariate plot of these PCs. In other words, taxa with tall, sharp cusps, small basins, short crest lengths and high trigonid–talonid relief (e.g. peradectids and palaeoryctids) are located in the top left quadrant, whereas taxa with low, bulbous cusps, large basins, long crest lengths and low trigonid–talonid relief (e.g. rodents), are positioned in the bottom right quadrant of this bivariate PC space. Those taxa located in the central area of the plot indicate more generalized molar morphologies and include euprimates, plagiomenids and some plesiadapiforms (e.g. microsyopids). The addition of PC3 results in an oblique three-dimensional gradation from more bunodont to less bunodont dentitions (see figure 3; electronic supplementary material, Table S4).

Figure 3.

Three-dimensional plot of the first three principal components of the PCA of all specimens. Only specimens from the first time interval, Cf2–3 (before euprimate origination), are shown. PC axes were rotated for the clearest view of the later euprimate niches. Three-dimensional convex hull polygons circumscribe individual taxa and are used to illustrate the dietary niche spaces of these groups. (Online version in colour.)

Overall, the results of the MANOVAs indicate that first, the euprimate dietary niche of adapids and omomyids are distinct from each other (p = 0.004, p = 0.001 for Wa0 and Wa1–2, respectively) (electronic supplementary material, Table S5), consistent with previous dietary reconstructions of these taxa (e.g. [23,63–65]). This distinction in ecological niche is further supported by differences in the reconstructions of activity pattern, body size and substrate use between adapids and omomyids [64]. Second, within the euprimate competitive guild, there was minimal dietary niche overlap between euprimates and non-euprimates. Figures 3–5 are PC plots, illustrating the overall dietary niche space of the euprimate competitive guild, for each of the three time periods. It is critical to note that these figures were not the basis for the pairwise MANOVA analyses, and importantly, illustrate only three of the six PCs used to determine the presence or absence of statistically significant niche overlap between taxa (electronic supplementary material, figures S3 and S4). Although convenient illustrative tools, graphical representations of multivariate niche spaces including fewer dimensions than the total number considered in the full analysis can be misleading, as they do not accommodate variation or separation along these additional, and potentially ecologically important, axes. In both time periods in which they are represented, euprimates exhibit near-zero values on all PCs, indicating a molar morphology ‘intermediate' to the bunodont morphology of rodents, at one extreme, and the sharp, high cusp morphology of peradectids and palaeoryctids, on the other (figure 3). As such, the dietary niches of euprimates inhabit a central location within the three-dimensional plots, surrounded by the niches of other taxa, obscuring their unique (for the most part) position within the guild-wide dietary niche space.

Figure 4.

Three-dimensional plot of the first three principal components of the PCA of all specimens, showing only specimens from the second time interval, Wa0 (euprimate origination). (Image credit: Bogdan Bocianowski, https://creativecommons.org/licences/by-sa/3.0/legalcode.) (Online version in colour.)

Figure 5.

Three-dimensional plot of the first three principal components of the PCA of all specimens, showing only specimens from the last time interval, Wa1–2 (initial euprimate diversification). (Image credit: Bogdan Bocianowski, https://creativecommons.org/licences/by-sa/3.0/legalcode.) (Online version in colour.)

Across the three time periods examined, the adapid dietary niche does not significantly overlap with that of any other taxon, demonstrating the uniqueness of this niche within the overall competitive guild. Thus, adapids entered the Bighorn Basin community in the absence of competition (non-replacement) (electronic supplementary material, Figure S5). The niche of Bighorn Basin omomyids, on the other hand, does overlap with two other taxonomic groups. Before the euprimate origination, Cf2–3 erinaceomorphans occupied a dietary niche that was then taken up by omomyids upon their arrival into the community (figure 3). It should be noted that the pairwise MANOVA comparison between Wa0 omomyids and Cf2–3 nyctitheriids also approaches non-significance (p = 0.046), indicating possible niche overlap between these taxa (and thus near-occupation of the Wa0 omomyid dietary niche by Cf2–3 nyctitheriids). However, once they appear in Wa0, omomyids do not occupy the same dietary niche as erinaceomorphans (nor do they overlap later during Wa1–2) but instead exhibit niche overlap with nyctitheriids (p = 0.090) (figure 4). This pattern is consistent with the presence of competition between omomyids and nyctitheriids upon omomyid immigration (figure 1). In Wa1–2, when omomyids diversify into several genera, niche overlap between omomyids and nyctitheriids disappears (figure 5). Furthermore, the niche overlap between Wa0 and Wa1–2 omomyids (p = 0.115) indicates that the omomyid dietary niche remained relatively stable and occupied the same position within the overall guild-wide niche space during this time. This is evidence for a divergence of the nyctitheriid niche away from the omomyid niche between the Wa0 and Wa1–2 time periods. Thus, although the dietary niches of Clarkforkian nyctitheriids and Wa0 euprimates do not overlap, this pattern of initial niche overlap between euprimates and nyctitheriids at the time of euprimate arrival (i.e. Wa0) and subsequent niche divergence in the successive time interval is consistent with the presence of competitive displacement between these two groups (electronic supplementary material, Figure S5).

It is plausible, however, that changes in the abiotic environment were responsible for these niche patterns rather than competitive interaction. The end of the Wa0 sub-NALMA is associated with the termination of the Carbon Isotope Excursion such that mean annual precipitation increased and mean annual temperature decreased across the Wa0–Wa1–2 boundary [9,18,50,66–68]. Thus, rather than strong competition, the initial divergence in nyctitheriid and omomyid niches may have been the result of a nyctitheriid response to a shift in climate associated with the end of the PETM, either directly or indirectly (e.g. through slight substrate use or activity pattern changes) related to the dietary niche. Alternatively, the fact that the niche of only Wa0 (rather than later) nyctitheriids overlaps that of euprimates might indicate that the occupation of the euprimate niche by nyctitheriids in Wa0 was the consequence of the warmer, drier climate present during that specific sub-NALMA, i.e. the PETM. This same time period has also been associated with molar morphological change, specifically size, in other early Paleogene North American mammals [69–73]; although see [31], demonstrating the effects that this climatic event likely had on mammalian biology. Similarly, although no clear change in erinaceomorphan diversity at the Paleocene–Eocene boundary is evident [14], it is also possible that competition with non-euprimate taxa, an increase in predator diversity, or climatic change in the earliest Wasatchian caused a shift in the erinaceomorphan dietary niche.

In addition, the erinaceomorphans considered here are typically reconstructed as predominantly terrestrial, saltatorial mammals (forest-floor dwellers) [74,75], and although nyctitheriids are considered arboreal, their likely scansorial locomotion (in contrast to the grasp-climbing of euprimates) may have contributed to a reduction in dietary niche overlap with euprimates [76,77]. Thus, it is possible that a difference in substrate use greatly minimized, if not precluded, instances of shared food resources used by euprimates and nyctitheriids. Furthermore, differences in the anterior dentitions or activity patterns of these groups may have further differentiated their dietary regimes. Consequently, even if climatic change was not responsible for the shift in the nyctitheriid niche after Wa0, dietary competition with euprimates may yet have been absent. Further exploration into euprimate origination into North America, including those earliest euprimates (omomyids) and their broader faunal community from Mississippi [30,31], may result in slightly different patterns of competition between euprimates and their potential competitors.

Finally, it should be noted that time-averaging, the variable lengths of time represented by each time interval (Cf2–3: 0.30 Ma, Wa0: 0.05 Ma, Wa1–2: 0.55 Ma) and collection bias may have an effect on the results presented here. Specifically, these issues have the potential to directly influence sample composition within each time period. First, due to the incompleteness of the fossil record and lack of available specimens, not all taxa present during certain time periods (e.g. microsyopids in Cf2–3, plagiomenids in Wa0 and Wa1–2) were included in the analyses. Moreover, the reconstructed niches are necessarily not perfect representations (in size or shape) of the true dietary niches of each taxon due to missing specimens as the result of the fossilization process. Second, although competitive interactions occur at the level of the population (whose best approximation in the fossil record is the species), species–species comparisons were not possible in the fossil sample due to small sample sizes. As a result, the patterns observed herein potentially (1) veil competitive interactions within higher taxa and (2) conflate competitive interactions among species within higher taxa due to species' combined inclusion in a single niche. Third, the unequal time bins and specimen availability per time bin (which are not directly linked in this study) have the possibility to enlarge (in the case of relatively more specimens) or reduce (in the case of relatively fewer specimens) the sizes (and the likelihood of overlap) of the reconstructed dietary niches. Furthermore, this leads to the potential inclusion of species with slightly different diets or that were not truly synchronistic within the same dietary niche. Similarly, the time scale at which the analysis was conducted could obscure competitive interactions that occurred at the boundaries between time periods, thereby concealing synchrony. For example, it is possible that erinaceomorphans and omomyids occupied the same dietary niche near the Wa0 boundary, causing the shift of the erinaceomorphan niche away from that of omomyids. Given the constraints of the fossil record and the competitive models outlined in figure 1, however, the data cannot currently support this conclusion. As all available M₂ specimens were included in this study, collection bias may also misrepresent the number of specimens per taxon (and thus the resulting dietary niche) when compared to true community abundances.

Only increased numbers of specimens can alleviate these issues, but given the relative rarity of certain groups within North American Eocene fossil assemblages, it may not be possible to substantially increase the specimen numbers for each taxon within the euprimate competitive guild. This also makes it unlikely that shorter, more refined temporal units can be used, as the temporal duration of intervals used in these analyses is also dependent on sample size. However, different classifications of time (e.g. equal time bins, the sub-NALMA revision of [78]) may affect observed patterns of niche overlap and thus the identification of taxa that may have engaged in competitive interactions during this time.

Regardless of the impact of factors outside of competition, it is significant that the dietary niches of both adapids and omomyids remained stable from the time of their origination to Wa1–2 (indicated by niche overlap between Wa0 and Wa1–2 adapids and omomyids, respectively; see electronic supplementary material, Table S5). This means that changes (if they occurred) in the abiotic or non-competitive biotic environments did not impact the molar morphology, and thus the reconstructed dietary niches, of euprimates during this time. As discussed above for erinaceomorphans and nyctitheriids, stable niche occupation across the time periods examined was not the case for other taxa.

4. Conclusion

The dietary niches of euprimates appear differentiated from those of most other members of the euprimate competitive guild within the Bighorn Basin community. Adapids occupied a unique niche within this guild, suggesting that these euprimates were able to take advantage of dietary resources unavailable to other taxa. Although niche overlap was observed between nyctitheriids and omomyids upon euprimates' arrival into the community, this overlap was short-lived, as the dietary niches of these two groups were distinct in the subsequent time interval. Furthermore, omomyids remained in this shared dietary niche, indicating the presence of competitive displacement between omomyids and nyctitheriids at the beginning of the Eocene. This pattern suggests that omomyids may have possessed adaptations that allowed them to exploit dietary resources at the expense of other taxa. The absence of competition in adapids and the presence of strong competition in omomyids are direct evidence of the fundamental role that dietary niche played in the establishment, and continued success through diversification, of euprimates as members of this broader North American mammalian community.

Data accessibility

The datasets and statistical analytical code supporting this article have been uploaded as part of the electronic supplementary material.

Competing interests

The authors have no competing interests.

Authors' contributions

L.K.S. conceived of and designed the study, carried out all data collection and statistical analysis and drafted the manuscript; G.T.S. participated in the conception and design of the study, coordinated the study and helped draft and finalize the manuscript. Both authors gave final approval for publication.

Funding

This research was funded by NSF-BCS 1155997 and the Institute of Human Origins.