Relative abundance of grazing and browsing herbivores is not a direct reflection of vegetation structure: Implications for hominin paleoenvironmental reconstruction.

Abstract

The diet of fossil herbivores inferred from enamel stable carbon isotopes is often used to make paleoenvironmental reconstructions. While many studies have focused on using environmental indicator taxa to make paleoenvironmental reconstructions, community-based approaches are considered to provide a more complete picture of paleolandscapes. These studies assume that the diet and relative abundance of herbivores are related to the areal extent of different vegetation types on the landscape. Here, we quantitatively test this assumption in 16 modern ecosystems in eastern and southern Africa with a wide range of woody vegetation cover. We conducted a landscape-level spatial analysis of vegetation patterns using a published land cover data set and computed landscape metrics. We compiled data on relative abundance and diet of herbivores inferred from carbon isotope studies for all large herbivores in these ecosystems. We found that despite differences in the total areal extent of different vegetation types, numerous sizable patches of each vegetation type are available in most ecosystems. However, despite variation across the ecosystems examined, grazers are typically the most abundant herbivores even in sites that have a higher proportion of forest and shrub cover. This indicates that the diet and relative abundance of herbivores is not a simple reflection of the total areal extent of vegetation types available on the landscape. The higher proportion of grazers observed in these ecosystems is a result of multiple factors including habitat heterogeneity, differences in biomass turnover rate between grasses and woody vegetation, resource partitioning, and the advantages of group living in open environments. Comparison of diet and relative abundance of herbivores in modern ecosystems to fossil herbivore assemblages shows that very different vegetation regimes can support similar herbivore assemblages. This study has significant implications for paleolandscape reconstructions and cautions against a simplistic wooded vs. grassland paleoenvironmental interpretations based on fossil herbivore assemblages.