Introduction

One of the main objectives of ecology is to understand the coexistence patterns of species and identify the mechanisms regulating the assembly of biological communities (^{Llorente-Bousquets and Morrone 2003}). In this context, several hypotheses have been proposed regarding the relative importance of deterministic and stochastic processes in community assembly (^{Schöener and Haken 1986}; ^{Hubbell 2001}) which varies depending on the spatial and temporal scales of measurement (^{Gavilanez and Stevens 2013}; ^{Plasencia-Vázquez et al. 2014}; Stevens and Gavilanez 2015; ^{Aguirre et al. 2016}). Deterministic hypotheses propose that community composition is determined by niche differentiation according to the principles of competitive exclusion. This hypothesis prioritizes deterministic biotic interactions or abiotic filtering mediated by niche conservatism (^{Weiher et al. 2011}). Environmental filtering (stress tolerance) proposes that the similarity of species within a given community increases due to abiotic restrictions (^{Cornwell et al. 2006}). On the other hand, ecological differentiation (niche partitioning, limitation of similarities) proposes that ecological interactions prevent similarities between coexisting species (^{MacArthur and Levins 1967}; ^{Chesson 2000}). On the other hand, stochastic models consider processes such as dispersal limitation and demographic drift, which produce assemblage patterns that can explain spatial autocorrelation in the presence of species, regardless of environmental variables. Particularly, dispersal limitation proposes that the presence of species in a community is limited by their ability to reach the site (^{Hurtt and Pacala 1995}; ^{Beaudrot and Marshall 2011}).

Recently, studies focused on the multiple dimensions of diversity have been developed (^{Webb et al. 2002}; ^{Petchey and Gaston 2006}; ^{Cadotte et al. 2011}; ^{Srivastava et al. 2012}) to better understand the mechanisms underlying local community assembly, as well as distribution and diversity patterns at broader scales (^{Jarzyna and Jetz 2016}; ^{Brum et al. 2017}). Additionally, approaches that directly consider the effect of species on ecosystems, such as functional diversity, have been developed (^{Tilman et al. 1997}; ^{Gómez-Ortiz and Moreno 2017}). Likewise, new strategies to evaluate the evolutionary relationships of species through their phylogeny have been proposed (^{Webb et al. 2002}). These new approaches for assessing diversity, such as functional and phylogenetic diversity, aim for a comprehensive quantification of biodiversity (^{Rosenzweig 1995}; ^{Cadotte et al. 2011}; ^{Rattis et al. 2018}). However, few studies have assessed diversity using these approaches simultaneously (^{Weinstein et al. 2014}; ^{Stevens and Gavilanez 2015}; ^{Brum et al. 2017}).

Studies on mammals and the multiple dimensions of biodiversity seek to understand the processes involving these vertebrates within ecosystems. Several of these studies consider characteristics such as body size, relating them to the functions provided by mammals within their natural habitats (^{Smith and Lyons 2011}). ^{Safi et al. (2011}) suggested that phylogenetic diversity and species richness increase in relation to mean annual temperature, while functional diversity decreases along with a higher seasonality. ^{González-Maya et al. (2016}) reported that functional diversity in mammal communities within the Neotropics decreases with the degradation of ecosystems and the loss of threatened species. On the other hand, ^{Oliveira et al. (2016}) found that species richness and functional diversity are decoupled in various regions of the world, and that species richness is closely correlated with environmental conditions while functional diversity depends mainly on non-equilibrium factors, including the evolutionary time to overcome the conserved niche. According to this analysis, species-rich regions (especially the Neotropics) could have many species that may be functionally redundant.

Primates are one of the most seriously threatened animal groups in tropical areas, mainly due to habitat loss, deforestation, and fragmentation (^{Stevenson 2016}; ^{Brum et al. 2017}; ^{Roncancio et al. 2010}; ^{Bueno et al. 2013}; ^{Rattis et al. 2018}). They play central ecological roles in ecosystems as dispersers, pollinators, predators, and prey. Additionally, they are part of the diet of various native cultures in the region (^{Cueva 2005}; ^{de la Torre 2010}; ^{de la Montaña 2013}). In Ecuador, primates have been studied in aspects such as conservation status, demography, diversity, diet, distribution, and survival in forest patches under anthropic pressure (^{Lizcano et al. 2016}; ^{Cervera et al. 2017}). Although these studies are an important contribution to the knowledge of primates, they have favored a one-dimensional perspective of diversity (i. e., taxonomic diversity) without considering their evolutionary history and ecological function (^{Cisneros et al. 2014}; ^{Brum et al. 2017}).

The present study focuses on characterizing in multiple dimensions of diversity Ecuadorian primate communities inhabiting different ecosystems of Ecuador, and evaluating the influence of environmental, structural, and spatial factors as possible assembly mechanisms of these communities.

Materials and methods

The characterization of primate communities of Ecuador was conducted through a systematic survey of literature, using databases such as Scopus, Google Scholar, and ISI Web of Science, using the following keywords (in English and Spanish): “primate community + Ecuador”, “primate diversity + Ecuador”, “primates + Ecuador”. We also reviewed theses and unpublished reports issued between 1989 and 2017. Studies that met our selection criteria were used to ensure data comparability (Table 1). Primate community composition (incidence) for the selected study sites were obtained from the papers. Spatial coordinates were projected in UTMs and later converted to WGS 84. This procedure allowed for the spatial reference to be compatible with the raster files containing altitude data and type of ecosystem (^{MAE 2013}).

Taxonomic diversity was characterized using presence/absence data for each study site. Functional diversity was estimated based on morphological, ecological, and behavioral data of the recorded species based on the information available in PanTHERIA (^{Jones et al. 2009}) and All the World’s Primates (^{Rowe and Myers 2016}) databases. We included variables related to body weight, body size, home range, and population density, which are related to how individuals interact with each other and the environment (^{Lefcheck et al. 2015}). In addition, niche breadth of each species was estimated based on the number of ecosystems they inhabit in Ecuador, which was determined using species range maps and a layer with information on the ecosystems of mainland Ecuador (^{MAE 2013}), usingQGIS version 2.10 (^{QGIS Development Team2015}).

Information regarding primate species diet was obtained from the database published in the database All the World’s Primates by ^{Rowe and Myers (2016}). We also conducted a thorough literature search regarding diet of each of the species reported. Based on this information, the following functional characteristics were determined:

Trophic breadth: Maximum number of food categories used by a species, with 13 being the highest number. For this category we grouped species in three levels: low (between 1 and 4 categories), medium (between 5 and 9 categories), and high (between 10 and 13 categories).

Percentage of fruit in the diet: Percentage of fruit in the total food consumed was calculated based on the food records reported in the All the World’s Primates database (^{Rowe and Myers 2016}).

Trophic guilds: Trophic guilds used in this study were adapted from those proposed by ^{Benchimol and Peres (2014}). Five trophic guilds were defined: 1 = Folivore-facultative frugivore: species that consume leaves and some fruits according to availability; 2 = Frugivore-folivore: species that feed mainly on fruits and leaflets; 3 = Frugivore-insectivore: species that feed mainly on fruits, insects, and sometimes leaflets; 4 = Granivore-frugivore-insectivore: species with a wide food range, mainly seeds, fruits, and insects according to their availability; 5 = Insectivore-frugivore-gummivore: species that mainly consume insects, fruits, bark, and exudates.

We calculated the Gower index using functional characteristics to build a distance matrix. This matrix was used to estimate the functional diversity indexes FD, FDISP, MPD_FD, and MNTD_FD, which characterize the diversity and dispersion of species in the functional space (Table 2).

Finally, phylogenetic diversity was characterized using the phylogeny by ^{Kuhn et al. (2011}), updating the nomenclature to ^{Tirira et al. (2020}). The phylogenetic diversity indexes PD, PSC, MPD_PD, and MNTD_PD (Table 2) were calculated based on metrics by ^{Webb et al. (2002}) and ^{Helmus et al. (2007}).

Similarity between communities was evaluated via cluster analysis, which also served for comparing the diversity between the resulting groups (functional and phylogenetic). Gower distance was used for functional diversity and divergence times, in millions of years, for phylogenetic diversity. This analysis was performed to assess whether different functional and phylogenetic groups of primates could be identified. All analyses were performed in R.

To determine the influence of different assembling mechanisms on the variability of the taxonomic, functional, and phylogenetic structure of primate communities, three groups of predictor variables were defined (environmental/environmental filtering - X1, spatial/ dispersal limitation - X2, and structural/competition - X3). These variables are key to diversity and composition patterns of mammal communities, including Neotropical primates (^{Plasencia-Vázquez et al. 2014}; ^{Aguirre et al. 2016}; ^{Gavilanez and Stevens 2013}). Lastly, a variance partitioning analysis was applied (^{Borcard et al. 1992}; ^{Legendre and Legendre 1998}; ^{Legendre and Gallagher 2001}) to discriminate the extent to which the variables contribute to the variation in the taxonomic, functional, and phylogenetic dimensions of primate community structure and whether they do so in isolation or synergy.

Environmental data were obtained from the BioClim database using a 30s (~1 km²) spatial resolution (^{Hijmans et al. 2005}) using the coordinates of each locality using QGIS (^{QGIS Development Team 2015}). A principal component analysis (PCA) of 19 bioclimatic variables was performed to obtain a subset of orthogonal axes (^{Legendre and Legendre 1998}). Based on this analysis, six representative environmental variables (that represented more than 90% of variability in environmental data) were selected to evaluate their influence on community structure (Table S4). The influence of spatial processes associated with dispersal limitation (^{Beaudrot and Marshall 2011}) was assessed with a matrix of Euclidean distances between the identified communities. Forest structure elements, particularly canopy height (^{Oliveira and Scheffers 2019}), are variables related to the availability of resources and niches (^{Gouveia et al. 2014}), therefore associated with competition. Canopy height data were obtained from the layers created by ^{Simard et al. (2011}), which resulted from the use of a “LIDAR” device. This information for each community identified was obtained by overlapping the corresponding raster layer.

Statistical analyses were performed in R version 4.1.1 (^{R Core Team 2017}) using the packages Vegan (^{Oksanen et al. 2018}), FD (^{Laliberté et al. 2014}), picante (^{Kembel et al. 2010}), and spatstat (^{Baddeley and Turner 2005}).

Results

Of the 192 studies reviewed, 46 were conducted in Ecuador, and 14 primate communities that met the established requirements were selected. Four of these communities were distributed in the coastal region and ten in the amazon region. The total number of species recorded was 17, representing 80 % of the diversity of primates in Ecuador. The community with the highest richness was located in the surroundings of the Kiwcha settlements in the northern region of the Yasuní National Park (Amazon region), with 12 species. In contrast, communities with lowest richness were in western part of Ecuador, near the coast, Jama Coaque and Pacoche, with two species each (Table 3). The 14 communities covered nine ecosystems, three in the coastal region and six in the amazonn (Figure 1).

A marked variation was found in the functional attributes (Table S2). Average weight for the species registered was 3,088.4 ± 2,807.4 g (range: 123.94 to 9,067.9 g). Average size (head and body) was 387.6 ± 118.1 (154.6 to 576.3) mm. Of the recorded species, Cebus aequatorialis was found in the largest variety of ecosystems. Furthermore, we observed variations between communities in functional characteristics related to diet. The primate community with the highest number of trophic guilds was Kichwa, with five guilds. The most common guild was granivore-frugivore-insectivore, with 13 species, while the least common was frugivore-insectivore, with two species. Most of the recorded species had a narrow trophic breadth. The “high” trophic breadth was the least represented category, absent in six communities. The community near the Kichwa settlements had the highest number of fruit-eating species in their diet (See Table S3).

Again, the community with the highest diversity in all dimensions was the Kichwa community in the Amazon (Table 4), which showed a wide range of coexisting functional groups and evolutionary lineages. On the other hand, some coastal communities showed low functional diversity values, although functional diversity indexes such as MPD_FD, were relatively high since the species that compose them differ functionally. Despite having an intermediate richness, the community of the San Miguel River showed the greatest functional dispersion (FDISP; mean distance of each species to the centroid of the composition), with a high MPD_FD value. The primate community in Pompeya Sur had the lowest functional dispersion and was composed of functionally similar species (low MNTD_FD values).

Figure 1 Ecuador map indicating the 14 primate communities considered in the study. 

The 17 species identified in the 14 communities were grouped into four primate families (Figure 2). The Kichwa community showed the greatest phylogenetic diversity (Table 4), with a PD value of 163.2. The MPD_PD index, representing the mean phylogenetic distance between species pairs, was higher for Pacoche and Jama-Coaque communities since the species in them belong to different and phylogenetically distant families (MPD_PD = 40.5). By contrast, the primate community inhabiting the foothills of Kutukú had the most phylogenetically related species (MNTD_PD = 0.19). The Phylogenetic Species Clustering (PSC) index indicated that the communities with the phylogenetically closest species were Rio Cayapas and Tesoro Escondido, which are geographically close in the northwest of the country, within the equatorial Chocó.

Five functional groups were identified (Figure 2). Species of the family Atelidae were clustered into two functional groups. Ateles fusciceps, Alouatta palliata, and A. seniculus were more closely related in terms of body weight, body size, and trophic breadth, while the group of Ateles belzebuth and Lagothrix lagothricha shared the same trophic breadth and guild. The representatives of the family Pitheciidae formed two functional groups. The species of the genus Pithecia were functionally similar to Cebus aequatorialis in terms of body weight and body size, and shared almost the same trophic guild. The titi monkeys of the genera Cheracebus and Plecturocebus were functionally related to Sapajus macrocephalus and Saimiri cassiquiarensis, sharing the same trophic guild and a similar home range. The species of the family Callithrichidae formed a single functional group with a similar home range, trophic guild, size, and weight. The phylogenetic clustering showed that pitheciids and atelids are the oldest families in the study area.

Figure 2 Contrast of phylogenetic and functional diversity between groups. Squares mark groups; lines indicate the distribution based on functional traits. 

As for taxonomic diversity, both environmental (X1) and spatial (X2) variables separately explained the highest variation (X1 = 28 % and X2 = 24 %, respectively) in the taxonomic composition of the communities. On the other hand, structural variables (X3) only accounted for 1 % of the variation. Functional diversity, environmental variables (X1), and forest structure (X3) were associated with a greater variation in the functional diversity of communities (25 %). Finally, the cluster that included the three predictor variables explained 25 % of the variation in phylogenetic diversity (Figure 3).

Discussion

The taxonomic diversity recorded in the present study is consistent with the one reported by ^{Sampaio et al. (2018}) in communities of the southern Amazon, Purus state, Brazil, reflecting the high diversity of mammals that characterizes the western Amazon (^{Voss and Emmons 1996}). This great diversity has been related to the large rivers that limit species dispersal (^{Ayres and Clutton-Brock 1992}; ^{Van Roosmalen et al. 2002}). It has also been reported that the high diversity of primate species in the Amazon region is associated with high fruit production levels (^{Stevenson 2016}; ^{Camaratta et al. 2017}) and structural complexity that creates microhabitats due to the different orography in the region (^{Homeier et al. 2010}).

Communities of the western region show a low diversity (S = 4) and are represented by endemic, and highly threatened species such as Ateles fusciceps and Cebus capucinus, which inhabit the easternmost section of the tropical Andes hotspot in the Chocó area. These areas, and the primate communities that inhabit them, are subject to environmental, biotic, and anthropic pressures that influence at the local (behavior) and macro (distribution) levels, affecting their composition, diversity patterns, and roles in the ecosystems (^{Kamilar and Beaudrot 2018}; ^{Kaisin et al. 2020}).

Figure 3 Representation of the variance partitioning analysis with a Venn diagram for the three dimensions of biodiversity: X1 = environmental, X2 = spatial (dispersal), X3 = structural (canopy strata). 

At the functional level, the variety of guilds (n = 5) and broad trophic niche of the species were important, mainly in Amazonian communities. Multiple species presented complementary functional traits that are important in the functioning of ecosystems (^{Pereira-Bengoa et al. 2010}; ^{Córdova-Tapia and Zambrano 2015}). The most common trophic breadth category was low (1 to 4 food types in the diet), indicating that most registered species have a level of specialization in their diet, which can make species sensitive to forest conversion (^{Cervera et al. 2017}). On the other hand, species with broad trophic niche (e. g., Sapajus macrocephalus) were recorded to include between 10 to 14 food types in their diet. These were common in Amazonian regions where resource availability may be higher. In some cases, when a generalist species becomes locally extinct, its ecological role may be assumed by another species (^{Galetti et al. 1994}; ^{Stoner et al. 2003}; ^{Link et al. 2006}; ^{Gómez-Posada 2012}).

Coastal communities comprise the same trophic guilds (facultative folivore - frugivore and granivore - frugivore - insectivore), indicating lower interspecific competition levels associated with resource availability. Differences in the diet of coexisting species (howler, capuchin, and spider monkeys) have been attributed to historical competition events that led to divergent dietary choices or foraging techniques (^{Fleming 1979}; ^{Arcos et al. 2013}; ^{Cervera et al. 2015}). However, it is worth highlighting that all coastal species include at least a low proportion of fruit in their diet, contributing to the ecological role of this functional trait (seed dispersers) in these ecosystems. Therefore, these species, and their disappearance can have a long-term impact on western tropical ecosystems, which are highly disturbed in Ecuador (^{Urbina 2010}).

Although functional characteristics of species suggest how they interact with each other and with the environment (^{Cadotte et al. 2011}; ^{Meachen and Roberts 2014}; ^{Gómez-Ortiz and Moreno 2017}), it is necessary to analyze the other dimensions of diversity. The comparison between the phylogenetic and functional clustering of primate species in the communities analyzed in this study showed that relationships between species are defined by the way in which they use the resources, creating cohesive functional groups that reflect an important phylogenetic dispersion, as in the case of Cebidae and Pithecidae. However, callitrichids had a conserved trophic niche, because they are very similar in body size, trophic niche breadth, and trophic guild. These species use the same resources, potentially reducing their coexistence; this is confirmed by analyzing the distribution maps of the species (^{IUCN 2016}), which show no overlap. Furthermore, the atelids formed two subgroups with different functional characteristics.

Communities with high taxonomic diversity, such as those in the lower Amazon, showed patterns of phylogenetic overdispersion (high MNTD_PD and MPD_PD and low PSC values). This illustrates the coexistence of species representative of ancient (pitheciids and atelids) and recent taxa (cebids and callitrichids), as well as a high functional diversity (high FD and FDISP), indicating that the resources available for use by primate species are diverse (^{Cooper et al. 2008}; ^{Kamilar and Guidi 2010}).

For the phylogenetic dimension, the best predictor of community structure was structural variability associated with strata diversity, which may be related to a high environmental heterogeneity and niche partitioning among different primate species in a community. Structural variation can foster the coexistence of species with similar requirements and functions, contributing to highly diverse communities, such as those reported in the Amazonian region (^{Arcos et al. 2013}; ^{Gómez-Ortiz and Moreno 2017}).

^{Kamilar et al. (2015}) suggested that zones with climatic stability favor a higher speciation rate. This could be reflected in the communities inhabiting the lower Amazon, which show high phylogenetic diversity. By contrast, the structure of communities within the dry seasonal forests of the Coast, where diversity is lower, seems to be governed by processes related to limited dispersal due to the Andes Mountain range barrier (^{Beaudrot and Marshall 2011}). However, these ecosystems may harbor higher endemism in some groups, including vertebrates (^{Olguín-Monroy et al. 2013}).

Our results suggest that both deterministic (environment and habitat structure) and stochastic processes (dispersal) play central roles in the structuring of equatorial primate communities (^{Cadotte et al. 2009}; ^{Flynn et al. 2011}). Part of the variation not explained in this study could be addressed by considering interspecific interactions, spatial scale, and seasonality (^{Belmaker and Jetz 2013}; ^{Stevens and Gavilanez 2015}; ^{Weinstein et al. 2017}).

Regardless of other factors, predictions considering the spatial dimension were the most important to explain taxonomic diversity. These results are supported by ^{Beaudrot and Marshall (2011}), who state that dispersal limitation is the primary mechanism in structuring primate communities. Neutral processes (^{Hubbell 2001}) related to spatial factors were important for the taxonomic and phylogenetic dimensions of biodiversity. Our findings show that the distribution of closely related species in communities may be controlled by stochastic factors, such as random speciation, extinction, and ecological drift (^{Pavoine and Bonsall 2011}).

There is an urgent need to understand community diversity patterns and their assembling mechanisms from a perspective encompassing beyond the taxonomic dimension. Our study highlights the complementarity of the information provided by different dimensions of biodiversity. Therefore, diversity should be assessed in a multidimensional way to better understand the mechanisms responsible for the establishment and persistence of communities and their ecological functions in ecosystems. Our findings support the importance of conducting diversity analyses on a spatial scale broader than local communities to make inferences on the ecological processes that influence the assembling and persistence of diversity, particularly in highly diverse communities such as those of Neotropical primates in Ecuador. This study shows that a varied resource availability (structure) could partly define the composition of these communities by reducing competition between species. Finally, our results provide valuable information to develop conservation strategies for Ecuadorian primates, as the roles of spatial processes and environmental and structural variables, and their association with the multiple dimensions of biodiversity, should be considered to set priority areas of conservation in a better way and ensure their maintenance over time. In this way, the environmental issues currently facing these communities and ecosystems can be comprehensively addressed.