Development of regeneration and recruitment functions for Continuous Cover Forestry (CCF)

Natural regeneration and recruitment of seedings into continuous cover forest (CCF) stands are critical for the development and establishment of permanent irregular structure forests. Development of robust predictors of recruitment and regeneration in CCF stands is important for both forest managers and forest modelers. The current growth model developed for uneven aged forests in Ireland, CARBWARE, does not currently have a regeneration or recruitment function. Therefore, data collected from Sitka spruce dominated (CCF) stands in Ireland and the UK and only data from Sitka spruce was used to develop models to predict regeneration and recruitment of seedings. A range of model formulations and statistical approaches, including traditional logistic regression, linear and non-linear regression and machine learning methods, particularly support vector regression (SVR), where used to predict the probability and amount of regeneration and recruitment of seedings into CCF stands. The results demonstrate that the SVR model is an attractive alternative method to predict the recruitment probabilities and amounts in comparison to previously developed linear/nonlinear models. However, we show that all stand based model estimations of regeneration was not robust enough for application to CCF management. The SVR model simulations are broadly consistent with the current ecological understanding of recruitment in Sitka spruce dominated CCF stands in the UK and Ireland. • Predictive models developed for Sitka spruce regeneration and recruitment in CCFs. • ML models outperformed regression in predicting recruitment probability and amount. • Crown competition and DBH variation significantly influence trees recruitment.