Niche differences, not fitness differences, explain predicted coexistence across ecological groups

Understanding the drivers of species coexistence is essential in ecology. Niche and fitness differences (i.e. how species limit themselves compared to others and species' differences in competitive ability, respectively) permit studying the consequences of species interactions. Yet, the multitude of methods to compute niche and fitness differences hampers cross-community comparisons. Such shortcoming leaves a gap in our understanding of the natural drivers of species coexistence and whether niche or/and fitness differences capture them. Here, we standardised niche and fitness differences across 953 species pairs to investigate species coexistence across ecological groups and methodological settings (experimental setup, natural co-occurrence, population model used and growth method). Using data gathered from 29 empirical papers, we asked whether large niche differences, small fitness differences or both explain predicted coexistence. Moreover, we performed an automated clustering algorithm to understand whether different underlying mechanisms drive species interactions. Finally, we tested whether any ecological or/and methodological settings drive these clusters. Species pairs predicted to coexist have larger niche differences but not smaller fitness differences than species pairs predicted not to coexist. Also, species pairs group into two clear clusters along the niche difference axis: those predicted to coexist and those that are not. Surprisingly, ecological or methodological settings do not drive these clusters. Synthesis. Overall, our results show that species coexistence is mainly influenced by mechanisms acting on niche differences, highlighting the importance of sustaining mechanisms that promote niche differences to maintain species coexistence. In addition, our results provide evidence that communities predicted to coexist differ from those that are not in ways that transcend their ecological grouping.