Figure Legends
Figure 1. Ellipses depicting isotopic niches of grassland herbivores in nine protected areas in the South African grassland. Blues are browsers, greens are grazers, and reds are intermediate-feeders. See Table 1 for key to protected area abbreviations.
Figure 2. Comparison of observed isotopic niche overlaps with expected values derived from null models. Each symbol represents a) the average, and b) the standard deviation, for each assemblage – white circles are results when only grazer species were considered in the analysis. Error bars depict 95% confidence intervals over 102 iterations of the null models, and the diagonal represents the expected value if not different from null.
Figure 3. Coexistence criteria for grassland herbivores based on isotopic niche overlaps (O ), with fitness equivalences depicted byR max ratios (where R maxwas estimated using Cole’s equation). The shaded region in a) and b) represents the region in which stable coexistence is possible, demarcated by 1/O (upper border) and O (lower border; note the log10 scale of the y -axes). In c) and d), the relative strength of stabilization due to niche partitioning is expressed as the difference between minimum permissible (O *) and observed O values, relative to O *. Values > 0 imply over-stabilization, and values below 0 (the dashed horizontal) fall outside the coexistence area in a and b. BR and GR (white symbols) indicate expected exclusion of browser or grazer species, respectively.
Figure 4. Invasion growth rates (IGRs), relative toR max, for each population derived from simulated multispecies competition models, parameterized by observed niche overlaps (representing competition co-efficients) and census data. Results are compared with models without niche overlap (complete partitioning, with resultant IGRs approachingR max), and with complete overlap (zero stabilization, with negative IGRs, i.e. below the dashed horizontal, for most populations).