Development of tree recruitment models for 10 species groups in the sugar maple-dominated mixed forests of eastern Canada

Individual tree recruitment is an important element needed to understand stand dynamics, as it influences both stand composition and productivity. Forest growth simulators usually include recruitment models. The quality of recruitment predictions can have long-term impacts on estimations of forest growth, ecosystem health and the commercial utility of managed forests. The main objective of this study was to develop a recruitment model for commercial-size trees (i.e., trees with a diameter at breast height > 9 cm) of 10 species groups using different dendrometric and environmental variables. The resulting model will be included in a growth simulator used to support forest management planning. We hypothesized that accounting for sapling density as a covariate would improve the recruitment model's predictive performance. Using empirical data from periodically measured permanent sample plots (1982–2019) located throughout the managed mixed hardwood forests of Quebec, we constructed models with and without sapling-related covariates and compared them on the basis of cross-validation model performance statistics. Our results show that including sapling density significantly improved model performance. From this, we conclude that adding sapling density as a covariate can significantly improve a recruitment model's predictive power for eastern mixed hardwood forests.