Introduction
Niche partitioning, where species reduce the effects of competition by exploiting different components of the resource (or other niche dimension), is a fundamental criterion for species coexistence in classical ecological theory. Here, each species positions itself along a niche axis to minimize overlap with competitors (MacArthur & Levins 1967; May 1973; Pianka 1973). In highly diverse communities, however, species numbers can outnumber available niches by orders of magnitude (Hutchinson 1961; Hubbell 2001). When a species in such a diverse community exploits a different niche to reduce overlap with one competitor, it can inadvertently increase overlap with another. If this happens, species may organize themselves in clusters along a niche axis (Scheffer & van Nes 2006; Vergnon et al. 2012), leading to high variance in overlaps between species pairs (see supplementary Fig. S2 and accompanying text). Species with high niche overlaps, such as those in the same niche cluster or guild, can coexist if they have similar competitive abilities, reducing rates of exclusion (Chesson 2000; Adleret al. 2007; Carmel et al. 2017). For stable coexistence, niche partitioning must therefore only be sufficient to overcome fitness differences between competing species (Letten et al. 2017).
The tug-of-war between maximizing niche partitioning and minimizing competitive differences is encapsulated by modern coexistence theory (MCT: Chesson 2000; Adler et al. 2007; Barabás et al.2018). In the MCT framework, species with strongly overlapping niches require greater similarities in competitive ability than when niches are distinct. Here, competitive ability is defined as ecological fitness, i.e. long-term population growth rates in the absence of competition. The key condition for coexistence is that competing species all have positive invasion growth rates, meaning they can recover from low numbers even while other species exist at resident densities (Chesson 2000; Adler et al. 2007; Letten et al. 2017; Broekmanet al. 2019; Grainger et al. 2019). This invasibility criterion requires the following inequality to be satisfied
\(O_{\text{ij}}<\frac{f_{j}}{f_{i}}<\frac{1}{O_{\text{ij}}}\)(Equation 1)
where O is niche overlap and f is fitness of speciesi and j , respectively. Equation 1 arises from the interplay between intra- and interspecific competition effects in multispecies models of population dynamics (Chesson 2000; Broekmanet al. 2019). Thus, according to MCT, species can coexist by having small fitness differences (i.e. an equalizing mechanism), by reducing niche overlap to the point that intraspecific competition effects are stronger than interspecific effects, and so species regulate their own populations densities more than they do each other (i.e. a stabilizing mechanism), or by a combination of both.
Some communities show an apparent excess of stabilization, where the amount of niche partitioning is more than is necessary to overcome fitness differences (Adler et al. 2010). Fitness equivalence, on the other hand, is typical for species within a taxon characterized by a particular life history (Clauss et al. 2020). MCT potentially integrates these respective emphases on stabilizing and equalizing processes, but so far empirical application has lagged behind theoretical developments (Ellner et al. 2019). This is due partly to difficulties in obtaining the necessary demographic, life history, and ecological parameters. Empirical studies of MCT depend on either invasibility experiments or model parameterization studies (HilleRisLambers et al. 2012; Broekman et al. 2019). For studies of natural systems, the latter requires long population time-series to estimate interaction co-efficients (Adler 2013), and to capture multigenerational dynamics in the case of long-lived organisms. Data requirements are even greater for species-rich communities because the number of pairwise interactions scales quadratically with the number of species (Broekman et al. 2019) .
Few communities illustrate the phenomenon of niche partitioning as well as large mammal herbivores (hereafter herbivores). Numerous studies have revealed differences in regulation by predators (Sinclair et al.2003; Hopcraft et al. 2012), demographic influences on population dynamics (Owen-Smith 2006), or, most often, in resource use (Arsenault & Owen-Smith 2008; Kleynhans et al. 2011; Kartzinel et al. 2015; Pansu et al. 2019). They are an exceptionally diverse group, given their large body sizes (ranging from ~5 to 5 000 kg): in Africa alone, there are more than 100 species, with local communities including as many as 30 species all depending exclusively on plants (Gordon & Prins 2019). Moreover, herbivores are among the most well-studied animal groups, hence ecological and behavioural data are available for many free-ranging populations. Resource niches are readily quantified by analyzing stable carbon and nitrogen isotope compositions of faeces and body tissues, which are consistent with isotopic compositions of diet sources, and hence reflect animal trophic niches (Newsome et al. 2007). In subtropical savannas and grasslands, herbivore 13C/12C compositions reflect variations in C3 (browse) and C4(grass) biomass consumption (Tieszen et al. 1979; Cerling & Harris 1999; Codron et al. 2005). Herbivore15N/14N compositions are determined by interactions between diet quality, climate, and physiology (Ambrose 1991; Robbins et al. 2005; Codron & Codron 2009). Comparative fitness and competitive abilities of herbivore species can be obtained from life history characteristics, especially those related to rates of reproduction and population growth (Clauss et al.2020), and from census data routinely collected by wildlife management.
In this study, we combine data on isotopic niche partitioning and density-independent reproductive rates to quantify the stabilizing and equalizing effects on coexistence of multiple herbivore communities. Our objectives were to determine i) the prevalence and type of isotopic niche partitioning; ii) whether partitioning is sufficient to overcome fitness differences and explain stable coexistence in a MCT framework; and iii) the contribution of niche partitioning to species’ invasion growth rates in diverse communities. For objective (i), we first partitioned the variation in isotopic niches attributable to geographic, intraspecific and interspecific differences, and then compared patterns of isotopic niche overlap with results from null models to describe non-random niche structures We then evaluated conditions for coexistence (objective ii) by comparing pairwise niche overlaps to species’ fitness inequalities. Lastly, we parameterised simulation models of multispecies population dynamics using fitness and niche overlap estimates, which allowed us to quantify species’ invasion growth rates after adding and removing stabilizing effects (objective iii). Combined, these steps provide an integrated evaluation of the role of niche partitioning in maintaining diversity in herbivore communities.