Picky eating as a means for coexistence

One of the most fundamental questions of ecology is, "How can species coexist?" A large body of ecological theory has predicted that co-existing species with shared, potentially limiting resources must differ in some way to reduce competition (1–3). Key to being different from sympatric species is the ability to partition resources (4, 5), slicing the environment into slivers of “pie” so each population within an assemblage can have their own piece. Prevailing paradigms of how species partition resources include spatiotemporal segregation (6), partitioning habitat (7), partitioning phenotypic traits of the resources, such as height strata (8) or monocots versus eudicots (9), and even through morphological differences between the competitors themselves (10). Pansu et al. (11) offer a complementary perspective that bridges the gap from the small scale of nutrient partitioning to the broad scale of partitioning time and space: partitioning plant species. They postulate that divergent use of plant species underlies observed niche differences and can promote stable coexistence, and therefore structures trophic networks.

For herbivores to coexist via partitioning plant species, each population must consume only a portion of the resource taxa that are consumed by the overall herbivore assemblage at any given space or time. Using an impressive dataset of fecal samples collected from 30 herbivore species across 10 savanna sites, Pansu et al. (11) used DNA metabarcoding to identify plant taxa within the diets of herbivores. They detected 893 plant taxa from 124 families, but the median population’s diet comprised only 31 taxa. Given that populations within assemblages did indeed differ in their diets, the next requirement is that dietary dissimilarity must be greater between sympatric species than within. At all of the sites sampled this condition was met. The diet of each focal herbivore was compositionally distinct from the diets of most, if not all, other sampled herbivore species at each site. These findings support that herbivores are eating only a portion of the plant taxa available to them, reflecting the constraint of their fundamental niche. Pansu et al. (11) sum the implications of this finding into one key phrase: “whatever their cause, their effect can only be to relax interspecific competition and intensify intraspecific competition relative to the scenario in which all herbivore species eat the same plant taxa.”

With the two basic tenets of partitioning resources confirmed, Pansu et al. (11) moved on to get a better understanding of what drives the strength of plant taxa differentiation. Phenotypic dissimilarity between resources, or the herbivores themselves, represents axes in which herbivores are classically hypothesized to partition resources. Some species are adapted to browse on the tallest vegetation, whereas others are constrained to consuming only low-lying plants (8). Generalist herbivores are adapted to consume a variety of plants of varying quality, whereas others may specialize on high-quality forage only (12). At a broad level, dietary discrepancy between sympatric herbivores is suspected to be weaker between grazers than between browsers because monocots (grasses) are less diverse than eudicots, thus offering less opportunity for partitioning (9). The herbivore community sampled by Pansu et al. (11) included both ruminants and nonruminants and spanned 5 to 5,000 kg across the browser–grazer spectrum. The dietary dissimilarity of plant taxa consumed by these herbivores was greater when body mass difference was greater, and was greater between pairs of nonruminants than between pairs of ruminants. Diet dissimilarity was particularly high for sympatric browsers relative to sympatric grazers, though even the strictest grazers and browsers consumed different plant species than their sympatrics. These results suggest that phenotypic traits of the herbivores themselves, as well as the plants they are eating, support partitioning of resources to the level of plant taxa, and not just according to broad strata. Pansu et al. (11) also demonstrate that the grazer-dominated assemblages exhibited higher niche overlap, lower modularity, and higher nestedness. Taken together, these results suggest that partitioning plant taxa can indeed structure communities.

While phenotypic characteristics can impact diet dissimilarity between sympatric species by defining the fundamental or realized niche, the environmental context can also act to strengthen or relax competition between sympatric species (13). Pansu et al. (11) tested if the strength of plant partitioning depended on the competitive environment. The strongest partitioning of plant taxa occurred in areas with lower rainfall, when food is putatively most limiting, suggesting that when competitive constraints are relaxed, plant partitioning is weaker. Using a quasi-experiment, they found that in Gorongosa, where herbivore densities were low as a result of recent population declines, herbivores partitioned plant taxa to a lesser degree than in other sites where the herbivores are dynamically stable at higher densities. Partitioning plant taxa therefore offers an intriguing framework for investigating how intrinsic and extrinsic contexts combine to shape communities (14).

Our title hints that herbivores are picky eaters as a means to coexist with sympatric species. Superficially, it may appear self-evident that these herbivores are picky eaters, but pickiness implies a choice, and demonstrating choice requires a metric of availability. If plants were not equally available to sympatric herbivores over time and space, the biological inference of the observed plant taxa partitioning may be diminished, because actually choosing different plants may not have occurred. By restricting sampling in time and space, Pansu et al. (11) did the best they could to account for plant availability. Yet, demonstrating selectivity of plants would require a measure of their availability to each herbivore species. Gathering such evidence would be a monumental task, because availability does not only depend on synchronizing measures of plant abundance in time and with herbivore presence, but also on accounting for the structure of the forage in relation to the morphology of the herbivores in question. The authors have quantified selectivity in a more restricted setting where it becomes logistically practical (15) and have acknowledged the challenge of measuring availability across larger areas to identify generalities, as well as some opportunities that lie ahead (see section S2 in supporting information of ref. 11).

By confirming four key predictions, Pansu et al. (11) find support for partitioning plant taxa as a means to support coexistence (Fig. 1). They give the example of giraffe (Giraffa Camelopardalis), kudu (Tragelaphus strepsiceros), and dik-dik (Madoqua cf guentheri) of Laikipia, which have highly disparate body sizes but all consume the shrub Senegalia brevispica. The shrub makes up a dominant proportion of the herbivores’ diets, though likely stratified by browsing height. However, they add that the remaining diets of the three herbivores are partitioned by plant species. Partitioning plant taxa may indeed reflect an all-encompassing multidimensional space in which species can partition their resources, but potentially lacks a mechanistic basis. The authors found support for the prediction that sympatric herbivores in areas with relaxed competition had lower diet dissimilarity, but the importance of the environmental context was marginally lower relative to phenotypic predictors. This finding implies that additional biotic interactions may have a role in shaping realized diets—a role we suggest should not be overlooked in many systems. For example, in the boreal North American large-mammal community, woodland caribou (Rangifer tarandus) and moose (Alces alces), two of the most widely distributed sympatric large herbivores, partition their resources. Caribou select lowland peat complexes and areas of abundant lichen biomass, whereas moose select uplands with high browse availability (16). While moose and caribou show the pattern of partitioning plant taxa (17), the process leading to the pattern is more likely through “apparent competition” (18, 19), where partitioning arises from the predator-avoidance strategy of woodland caribou, rather than completion per se (17). Given that resource partitioning is thought to maintain species diversity (20), understanding the mechanisms behind resource partitioning is key to predicting changes in species diversity and community structure with future global change (21).

Fig. 1.

Four key findings of Pansu et al. (11), who hypothesize that herbivores partition plant taxa, which contributes to stable coexistence.