Neutral biodiversity despite strong ecological trait selection of N2-fixing trees across the Amazon Basin

Background/Question/Methods

Linking biodiversity and ecosystem function is critical but difficult because 1) biodiversity and functional traits may be governed by very different factors, making it challenging to understand and predict how they interact; and, 2) analysis of biodiversity in simplified or artificial systems may be straightforward but is substantially more problematic at scales of biomes or sub-continents.  Here we focus on a trait with major biogeochemical consequence – symbiotic N₂ fixation in trees – and examine how it interacts with properties of biodiversity across the highly diverse forests of the Amazon Basin.  We used data on >48,000 trees and >70 large-scale tropical forest plots from the RAINFOR network.  Our analysis considers first the distribution of the fixation trait as a component of total tree biodiversity (N₂ fixers as well as non-fixers), and second, the functional connection between the diversity and abundance of specific tree taxa, soil nutrients, and the landscape-level disturbance regime.  Specifically we evaluate two alternative hypotheses: (1), that fixers are locally selected and that this selection causes biodiversity to vary at the local scale; and (2), that both fixation and the diversity of fixers are governed at larger scales, possibly at the scale of the entire Amazon basin.

Results/Conclusions

Our diversity analysis indicates that, across all 38 genera of N₂ fixers, the fixation trait displays little to no clear structure across the Amazon basin.  Fixers were consistently 10% of total species irrespective of variation in any local factor such as soil nutrients, location, disturbance regimes, or other environmental conditions.  This result suggests a neutral (or stochastic) model for local selection of fixer biodiversity.  However, closer inspection revealed that across all sites 65% of N₂-fixing trees belonged to only 4 genera, and that these genera were substantially overabundant relative to neutral model expectations.  Further analysis of these genera of “super fixers” revealed the existence of substantial local selection across both soil phosphorus and nitrogen gradients.  For phosphorus, the pattern indicated a break-point pattern in which fixer abundance was rare at low phosphorus, high at medium phosphorus, and low again at high phosphorus.  For nitrogen, however, the pattern was consistent with a mechanism in which fixers supply nitrogen to Amazon forests (rather than the alternative – that fixer abundance responds to low nitrogen conditions).