Table 2 Testing the stress gradient hypothesis for mixed species groups (MSG) of African savanna herbivores.
From: Refining the stress gradient hypothesis for mixed species groups of African mammals
Hypothesis | Prediction | Test | |
|---|---|---|---|
Quantitative aspects of MSGs | |||
H1 | Predation risk mediates MSG occurrence | Greater likelihood for MSG occurrence inside the lion home range (LHR) vs. outside of the LHR | Logistic regression |
H2 | Primary productivity mediates MSG occurrence | Since primary productivity is essential for herbivores, we expected a strong signal of this variable. However, because stress perceptions associated with primary productivity (indicated by NDVI) varies by species (i.e. small bodied species generally perceive tall grass [which could be indicated by high NDVI conditions] as stressful due to low nutrient concentration and high predation risk, whereas large bodied species may perceive low NDVI as stressful due to their food quantity requirements), a prediction on the general direction of its effect is not possible | Logistic regression |
H3 | Human-induced stress mediates MSG occurrence | Lower likelihood for MSG occurrence in the core area of the ranch vs. the edge area of the ranch | Logistic regression |
H4 | Habitat structure mediates MSG occurrence | No specific prediction; will depend on habitat preferences of each species | Logistic regression |
Structural and compositional aspects of MSGs | |||
H5 | If the frequency of MSGs is mediated by environmental variables (H1–H4), we would also expect the structure of animal networks to differ across environmental conditions | Under environmental conditions favoring the frequency of MSGs (H1–H4), we would expect (1) that species generally increase the strength of associations with heterospecifics (increase in Node strength), (2) that the variability of species associations varies (changes in Y-measure), and that (3) the transitivity of associations increases, i.e. the tendency of species to form highly connected subsets increases (increase in Clustering coefficient) | Welch’s t-test on aggregated values of subsampled networks |
H6 | We expected dyadic associations to emerge under stressful conditions | Under stressful conditions we expected to observe additional dyadic associations to emerge, which cannot be explained by chance alone | Randomization test |
H7 | Prey preferences of lions and food preferences of herbivores mediate the composition of MSGs | Species with similar susceptibility to lion predation are more likely to be found together in MSGs in order to reduce predation risk; species with similar food preferences are more likely to occur together in MSGs because these species may provide information on resource availability | Randomization test |
