APPENDIX
Body size and species coexistence in consumer-resource interactions: A comparison of two alternative theoretical frameworks
Sumanta Bagchi1,2,3 and Mark E. Ritchie1
1Department of Biology, SyracuseUniversity, 107 College Place, Syracuse, NY – 13244, USA.
2Nature Conservation Foundation, 3076/5, IV-Cross, GokulamPark. Mysore, Karnataka – 570002, India.
3Correspondence: S. Bagchi, Dept. of Ecosystem Science and Management, TexasA&MUniversity, 325 AnimalIndustriesBuilding, College Station, TX – 77843, USA. Tel: +1 979 8450283, Fax: +1 979 8456430. E-mail:
Contents:
a)Fig. S1: Diagram showing co-variation between diet diversity (Shannon’s H’ index) and body weight of species in different assemblages listed in Table 1.
b) Table S1: Results of a general linear model to test for statistical significance of the co-variation between diet diversity and species’ body weight from studies listed in Table 1.
Information-theoretic indices, such as Shannon’s H’ can summarize the information on whether a species’ diet is composed largely of a single item, or an even mix of several items. Low values on H’, or diet diversity, would indicate that the species’ diet is dominated by a few items. Alternatively, a high value of H’ would indicate that the species’ diet consists of an even mix of several items. So, our model prediction can be statistically tested against empirical data by evaluating co-variation between species’ diet diversity and body size. To test the predictionof variation in diet-diversity with body size, we reviewed data on diets from several multi-species guilds: grasshoppers, large mammalian herbivores, fishes, and mammalian carnivores. From the information categorized into suitable functional groups, we calculated Shannon’s H’ of diet for grasshoppers in Montana (Belovsky 1986); mammalian herbivores in Serengeti (Hansen et al. 1985), Montana (Belovsky 1986), and Tibetan plateau (Harris and Miller 1995); fishes from Amazonian floodplains (de Merona and Rankin-de Merona 2004) and Lake Malawi (Reinthal 1990); and African carnivores (Ray and Sunquist 2001) (summarized Table 1). The observed correlation (Spearman’s ρ) between diet-diversity and consumer body size was negative in each case (Fig. S1). The overall significance of this trend was tested with a general linear modelwhere guild type was included as a categorical variable. The relationship between diet diversity (Shannon’s H’) and body size (weight in kg, log-transformed to meet statistical assumptions) was significantly negative, although the slope of the relationship differed among guilds (Fig. S1, Table S1). The interaction term in this statistical model suggests that slope differs between guilds, but the overall correlation is negative. Since information on body weights of Lake Malawi fishes was not available, those data were not included in the statistical analysis. Notably, no correlation was found in a guild of domestic livestock that has been historically assembled by humans (Fig. S1).
Figure S1: Negative relationship (Spearman’s rank correlation) between diet diversity (Shannon’s H’ index) and body size of consumers in different guilds from the studies listed in Table 1. Statistical relationships are summarized in Table S1.
Table S1. Results of general linear model to evaluate co-variation between diet diversity and body size. Data are from the studies listed in Table 1.
Effect / df / SS / MS / F / PGuild type / 5 / 12.00 / 2.40 / 42.99 / < 0.0001
Body weight / 1 / 0.63 / 0.63 / 11.25 / 0.002
Interaction / 5 / 0.73 / 0.14 / 2.61 / 0.04
Error / 35 / 1.96 / 0.05