Despite the central role of microorganisms in biogeochemistry, process models rarely explicitly account for variation in communities. Here, we use statistical models to address a fundamental question in ecosystem ecology: do we need to better understand microbial communities to accurately predict ecosystem function? Nitrogen (N) cycle process rates and associated gene abundances were measured in tropical rainforest soil samples collected in May (early wet season) and October (late wet season). We used stepwise linear regressions to examine the explanatory power of edaphic factors and functional gene relative abundances alone and in combination for N-cycle processes, using both our full dataset and seasonal subsets of the data. In our full dataset, no models using gene abundance data explained more variation in process rates than models based on edaphic factors alone, and models that contained both edaphic factors and community data did not explain significantly more variation in process rates than edaphic factor models. However, when seasonal datasets were examined separately, microbial predictors enhanced the explanatory power of edaphic predictors on dissimilatory nitrate reduction to ammonium and N2O efflux rates during October. Because there was little variation in the explanatory power of microbial predictors alone between seasonal datasets, our results suggest that environmental factors we did not measure may be more important in structuring communities and regulating processes in October than in May. Thus, temporal dynamics are key to understanding the relationships between edaphic factors, microbial communities and ecosystem function in this system. The simple statistical method presented here can accommodate a variety of data types and should help prioritize what forms of data may be most useful in ecosystem model development.