Abstract:

Aims: Species coexistence mechanisms and species abundance distribution patterns at different time and spatial scales have been and may continue to be key issues in community ecology, whereas species abundance distribution is one of the most powerful tools for understanding the mechanisms generating community structures and biodiversity patterns. Studying species abundance distribution patterns in different habitats of forest communities can help to reveal the mechanisms that drive the formation and maintenance of biodiversity. This study took desert riparian forest communities as the research object, and analyzed the difference of species abundance distribution patterns in different habitats, in order to understand the dominant ecological processes of species abundance distribution and reveal assembly mechanism of desert riparian forest communities in extreme arid areas.
Methods: We surveyed two 4-ha desert riparian forests with contrasting inundation regimes and soil properties. We then employed four statistical models (i.e., log-series model, log-normal model, Poisson lognormal distribution model and Weibull distribution model), two niche-based models (i.e., niche preemption model, broken stick model) and two neutrally-based models (i.e., metacommunity zero-sum multinomial distribution model, Volkov model) to fit observed patterns of species abundance distributions for the two communities. We used the K-S test and Akaike information criterion (AIC) to select the best fitting model.
Results: Our results showed that: (1) The species composition and species abundance differed significantly between the two desert riparian forest communities. In the less inundated and drier habitat, the species abundance distribution tended to have less interspecific variation, in addition to lower species diversity, abundance, forest cover, and fewer common species. (2) The statistical models and neutrally-based models performed better than the niche models in fitting species abundance distributions. The metacommunity zero-sum multinomial distribution model was the best-fitting model for the arid sandy habitat far from the river bank. The log-normal model and the Poisson lognormal model were the best-fitting models for the floodplain, whereas the neutral model (i.e., Volkov model) and the two aforementioned statistical models showed similar performance.
Conclusions: We therefore infer that a neutral process played a leading role in the assembly of the desert riparian forest communities. Since the species abundance distribution could be result of joint effect of multiple processes, we can not rule out the potential role of a niche-based process.

Key words: desert riparian forest, habitat, species abundance distribution, model fitting, ecological process, community assembly mechanism