The main focus of community ecology is to study communities that consist of mutiple, potentially interacting, species. One obvious and popular way of quantifying community structure is the notion of species richness. However, reducing the information present in a abundance or presence/absence data into a single data point (species richness or its derivatives) could potentially result in decreased power to detect or even completely wrong conclusions about the actual structuring forces in a community of interest. The main question for this study was thus to investigate what aspects of the species richness concept could be used as meaningful properties of a community, and what aspects should be replaced by different concepts and analysis strategies. I hypothesized that the data reduction to local species richness estimates would result in a decreased power to detect important structuring forces compared to multivariate techniques. I tested this using 158 data sets that combined information on species abundances or presence/absences, environmental variables, and spatial locations. I also hypothesized that the data reduction to regional species richness estimates would measure aspects related to dispersal, and capture important information not measured through species abundance or presence/absence data. I tested this by reviewing a series of experiments that manipulated both local and regional processes in freshwater communities. 
Results/Conclusions

I found that reducing the information to local species richness resulted in a lack of power to detect important processes: spatial patterns went undetected in 46 % of the data sets where species abundances showed important spatial patterns, environmental patterns went undetected in 32 % of data sets. However, regional species richness was related to dispersal-associated variables in all the experiments I reviewed. Combined, these results indicate that local species richness should be approached with caution by community ecologists, while the study of regional species richness could potentially provide a lot of information about dispersal characteristics in a system. Since beta diversity is derived from the local and regional species richness estimates, the value of this concept is also questionable because of the limitations of the local species richness concept. Finally, embracing the complexity provided by species abundance data through powerful multivariate techniques will result in important insights into empirical communities.