Experimental macroecological approach tests the influence of biotic interactions, species richness, and abundance as determinants of the species abundance distribution

Background/Question/Methods

An emerging conceptual framework proposes that  macroecological patterns are influenced predominately by constraints imposed by system-level properties (e.g., total abundance, species richness).  This framework suggests that macroecological patterns will be robust to many shifts in species composition, but also that they will be responsive to major changes in biotic interactions that influence system-level constraints. If system-level properties are the main determinants of macroecological patterns, then we expect these patterns to be robust to the identity of particular players in the system unless those players have traits that fundamentally alter that ecosystem. Here, we ask: Do changes in species composition, in response to experimental manipulation, influence the shape of the species-abundance distribution (SAD), or does SAD shape only change when S and N also change? We compiled data from over 297 experimental communities comprising widely varying taxa (e.g., zooplankton, arachnid, herpetofauna), experimental manipulations (e.g., caged exclosures, habitat manipulation, land management), across all continents except Antarctica. Experimental data is rarely used with a macroecological approach and is a potentially powerful way to investigate the determinants of macroecological patterns. 

Results/Conclusions

Data was analyzed from 297 communities, resulting in 200 control-experiment comparisons. SADs were characterized by examining their fit to the log-series vs. the log-normal frequency distributions and by using the maximum-likelihood estimation for poisson log-normal (97% of SADs fit the log-normal distribution).  This estimation results in two values describing SAD shape: mean (μ, which describes mean abundance) and standard deviation (σ, which describes evenness).  Although many sites had large changes in compositional similarity due to the experimental manipulations (95% Bray-Curtis>0.2), SADs are robust to changes induced by experimental manipulations, including changes in species composition, and changes in S and/or N. Most SADs did not differ in the paired comparisons (chi-squared test, 97% p>0.05). Many sites experienced no change in S and N (1:1 line fit; S, r²=0.79; N, r²=0.76). Paired controls and experiments exhibited little deviation in μ (1:1 line fit, r²=0.85), but significant deviation in σ (1:1 line fit, r2=0.59). However, differences in σ did not appear to be related to percent change in S or N (S, r²<0.01; N, r²<0.005). Our results suggest that the species abundance distribution is robust to many manipulations and that further study is needed to understand determinants of the pattern.