Abstract:

The herbaceous layer is an important component of forest ecosystems and plays an important role in maintaining forest biodiversity. To understand the mechanisms shaping the forest herb community patterns over multiple growing seasons, we used herbaceous data collected in a 25 ha broad-leaved Korean pine (Pinus koraiensis) mixed forest plot in Changbai Mountain, Northeast China and fitted species abundance distributions (SADs) using different models. We used both pure statistical models including log-normal, log-series, and mechanistic models, including two niche models (broken-stick and niche preemption) and two neutral models (metacommunity zero-sum multinomial distribution and Volkov model). Further, we applied the AIC and Kolmogorov-Smirnov tests to compare the goodness-of-fit of these models. Our results showed: (1) The observed SADs of the herb layer varied by season. While there were similar proportions of rare and common species in spring, there were more species with moderate abundances in summer and more rare species in autumn. (2) The best-fitting models of SADs were similar in different seasons. In our analyses, the log-series model was the best pure statistical model across the three seasons. For the mechanistic models, neutral models performed better at explaining patterns of SADs than niche models. The metacommunity zero-sum multinomial distribution model was the best model in spring and summer and the Volkov model was the best one in autumn. This indicates that stochastic processes may play a dominant role in maintaining the herb species abundance distributions. Our study showed that although the SAD patterns varied over growing seasons for the herb layer in the broad-leaved Korean pine mixed forest, the underlying mechanisms governing these patterns are similar and neutral models always perform better than niche models in fitting the SADs.

Key words: forest herb layer, niche process, neutral process, statistical model, model fitting