**1. Introduction**

African savannas are home to the greatest extant concentration of wild large herbivores, many of which are facing extinctions that have far-reaching consequences for whole ecosystems, and plant communities in particular [1–3]. Although large herbivores can range widely, many exhibit habitat preferences that reflect different needs to access water, avoid predators, or forage on palatable plants [4–7]. Elucidating spatial relationships between plants and large herbivores is paramount to understanding and managing change in savanna ecosystems.

A common expectation in biodiversity science is that the diversity of plants and animals is positively correlated [8,9]. The latitudinal diversity gradient and the geographic mosaic of 'interaction biodiversity' are thought to be potential outcomes of this positive association [10]. Positive associations in plant and consumer diversity can emerge from similar responses to abiotic gradients [9], disturbance [11], and biotic interactions [12].

**Citation:** Freeman, P.T.; Ang'ila, R.O.; Kimuyu, D.; Musili, P.M.; Kenfack, D.; Lokeny Etelej, P.; Magid, M.; Gill, B.A.; Kartzinel, T.R. Gradients in the Diversity of Plants and Large Herbivores Revealed with DNA Barcoding in a Semi-Arid African Savanna. *Diversity* **2022**, *14*, 219. https://doi.org/10.3390/d14030219

Academic Editors: Luc Legal and W. John Kress

Received: 31 January 2022 Accepted: 11 March 2022 Published: 17 March 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

For example, resource gradients can generate overlapping gradients in plant and animal diversity across biomes (e.g., tundra vs. tropical rainforest) as well as within landscapes (e.g., mesic vs. xeric microhabitats) [13,14]. Whether positive biodiversity associations across trophic levels enable predictions about landscape-level linkages between the local diversity of plants and the activities of highly mobile herbivores that eat them is an open question, although recent dietary DNA-metabarcoding studies sugges<sup>t</sup> the diversity of plants and large herbivores is more tightly linked than currently appreciated [15].

Analyses of local plant and animal communities can be used to test key hypotheses about how shared habitat associations may develop. The Resource Specialization Hypothesis posits that locally diverse plants support locally diverse consumers by providing a broader array of resources that can accommodate the dietary niches of more consumer species [16,17]. In contrast, the More Individuals Hypothesis posits that more diverse plant communities are often more productive, thereby increasing both the abundance of individual consumers and hence the number of consumer species [16,18]. If a positive correlation between the diversity of plants and large mammalian herbivores exists, it may arise from their generalist feeding strategies and requirements—consistent with the More Individuals Hypothesis. However, comparing these hypotheses to gain understanding about whether plant and animal diversities are locally linked requires consideration of what it means to be a 'specialist' and whether the evolutionary relationships of species reflect traits that determine where they occur [10,12,19]. Strong inferences can be made about how spatial distributions are structured by incorporating information on species' ecological and evolutionary similarities [20]. A relatively small number of species can represent a broad diversity of evolutionary lineages, and vice versa [21]. Within communities of large herbivores, for example, some species may consume relatively few plant taxa (taxonomic specialists) from many plant lineages (phylogenetic generalists) whereas others may consume many taxa from few lineages [19]. Thus, even 'generalized' large herbivores may exhibit feeding or habitat preferences that establish colocalized 'hotspots' of plant and animal diversity—consistent with the Resource Specialization Hypothesis and contrary to the More Individuals Hypothesis.

We combined a plant-DNA-barcode phylogeny, a comprehensive map of woody plants, and camera-trap data to evaluate whether the abundance and diversity of plants and large herbivores are correlated across the ForestGEO plot at Mpala Research Centre, Kenya. The Mpala ForestGEO plot spans soil and topographic gradients, making it possible to evaluate local covariation in the taxonomic and phylogenetic diversity of woody plants and large herbivores at scales of tens to hundreds of meters while controlling for the influence of large-scale biogeographic differences in species pools [22]. We used these data to test three hypotheses: (*i*) the diversity of woody plants and large herbivores is significantly different between habitats, such that habitats with high woody-plant density also have high taxonomic and phylogenetic diversity of both plants and animals; (*ii*) local herbivore diversity is strongly and positively correlated with local plant diversity, both within and between habitats that support different baseline resource availabilities (i.e., tree density) and diversities (i.e., taxonomic and phylogenetic); and (*iii*) the taxonomic and phylogenetic compositions of plant and herbivore communities are spatially congruent, consistent with a linkage between the diversity of plants and large mammals.
