[Simulation and uncertainty analysis of natural regeneration for pine-oak forests.]

The complexity and uncertainty of forest regeneration is crucial for predicting forest ecosystem dynamics. A natural regeneration model of pine-oak forests in Qinling Mountains was constructed with competition, climate and topography factors using Bayesian statistics and global sensitivity analysis (GSA). The alternative models were based on Poisson, negative binomial (NB), zero-inflated Poisson (ZIP), and zero-inflated negative binomial (ZINB) models. According to the uncertainty of model parameter transfer, the analysis results were quantified, and the dominant factors of small probability events affecting forest regeneration were explained. The results showed that the ZINB model was the best one in the simulation of Pinus tabuliformis and Quercus aliena var. acuteserrata. Stand basal area, light interception, slope location and minimum temperature during growing season were the most critical factors affecting natural regeneration of P. tabuliformis, while stand basal area, cosine of aspect interacted with the natural logarithm of elevation, annual mean temperature, and precipitation of the warmest quarter were the most critical factors for Q. aliena var. acuteserrata. The contributions of various factors to the predictive uncertainty were: competition factor (25%) < climate factor (29%) < topography factor (46%) for the simulation of P. tabuliformis regeneration, and climate factor (12%) < competition factor (24%) < topography factor (64%) for the simulation of Q. aliena var. acuteserrata regeneration. The natural regeneration quantity of P. tabuliformis was positively correlated with mean annual temperature and minimum precipitation during growing season, and negatively correlated with the mean temperature in the driest quarter. The natural regeneration quantity of Q. aliena var. acuteserrata was positively correlated with mean annual temperature, minimum precipitation during growing season, precipitation of the warmest quarter, and negatively correlated with mean temperature of the driest quarter. The ZINB model based on Bayesian methods could effectively quantify the major factors driving forest regeneration and interpret the uncertainty propagated from parameters, which was useful for predicting forest regeneration.