ObjectiveTree recruitment is the basis to ensure forest long-term maintenance, and the recruitment model can predict the development of forest and quantify the future health and productivity of forest ecosystem.

MethodAbout 295 permanent sample plots were established across the natural range of Quercus mongolica in the Jilin Province of northeastern China in 1995. Stand factor, site factor, and climate factor were selected to construct recruitment model of Quercus mongolica. The basic forms of model include Poisson distribution and negative binomial distribution. The zero-inflated model was added to these basic models because of the existence of a large number of zero values in the sample plots. The sample plot’s random effect was taken into account in order to solve the problem of nested and longitudinal data in the model. Finally, the validation data were used to verify the fitness of model.

ResultStand arithmetic mean diameter and the number of trees per hectare were the most important factors, and both were negatively correlated with the probability and quantity of tree recruitment. Both site and climate factors had no significant effect on tree recruitment. The accuracy of the negative binomial distribution model was higher than that of the Poisson distribution due to the over-dispersion of the data. After considering sample plot’s random effect, all the models obviously improved the simulation accuracy of the model except for the standard negative binomial distribution model. The simulation effect of the negative binomial distribution model was the best when considering random effect and zero-inflated model.

ConclusionIn order to ensure the occurrence of tree recruitment, it is very important to determine science management and initial planting density in forest management.