There are many papers in Early View here, check them out. I wanted to highlight just a few of them that I personally thought were interesting.
- Marko Spasojevic and Suding (here) examine community assembly mechanisms via functional diversity patterns. How communities are assembled is a very classic question in ecology. Papers on this topic often test the relative importance of competitive forces vs. environmental filtering forces. They test the relative importance of not just filtering and competition, but also equalizing fitness processes (functional redundancy towards an optimal trait; Chesson 2000) and facilitation. They find evidence for filtering, equalizing processes, competition, and facilitation. This is an important study in contributing to expanding the discussion of mechanisms for community assembly other than filtering and competition.
- Paul Selmants et al. (here) examine the effects of realistic species loss from those of random species loss on invasion resistance at the plant community level. Many studies have used an approach of looking at random species losses and how that influences X or Y processes. Selmants et al. found that plots with realistic species losses were more resistant to invasion than those with random species losses. They posit that the realistic species loss scenario created vacant niche space more quickly than the random species loss scenario.
- Zhang et al. (here) conducted a meta-analysis of 54 studies to look at the relationship between diversity (evenness, species richness, and traits) and productivity (this brings to mind a review on evenness effects on ecosystem properties by Hillebrand et al. in 2008). Evenness was the strongest predictor of productivity – an equitable distribution of abundance among species in a forest leads to more productive forests. They do admit that testing mechanisms is beyond their paper, and pose a few ideas for mechanisms. I was particularly intrigued by their methods. I have never used boosted regression trees (BRT) for meta-analysis. However, they did use R, and a package for R called gbm (check it out here) – so I might check it out myself. Apparently, these BRT models can handle missing values in predictor variables – very odd, and interesting, indeed.