资源环境科技发展态势分析平台

Most studies comparing the predictive performance of distance-independent and -dependent competition metrics have been conducted in even-aged, single species stands. In addition, past studies have generally not considered more sophisticated distance-dependent competition metrics such as open sky-view or light interception indices. Using data from fully stem mapped inventory plots (n = 260) established in structurally complex, mixed-species experimental forests (n = 5) in the Northeastern US, we evaluated prediction performance of one- and two-sided distance-dependent as well as more conventional distance-independent competition metrics in both annualized diameter increment (n = 1735) and probability of survival models (n = 1953) developed across 26 contrasting softwood and hardwood species. Prediction accuracy of non-spatial and spatial models was further assessed by classifying observations by species shade-tolerance, relative tree height, time since harvest, and tree clumping. Our analyses revealed no general superiority of distance-dependent competition metrics over their distance-independent counterparts for either diameter increment or survival, which was particularly true for the latter variable. Distance-dependent metrics tended to outperform distance-independent ones for observations from shade-tolerant species, trees of lower crown classes, and immediately after harvest. However, results on the comparative analysis of the predictive power of the different types of competition metrics under varying stand conditions and situations were often inconsistent and not always conclusive. In addition, model complexity, i.e. number of non-spatial explanatory variables in the base model, strongly affected the performance of distance-dependent competition metrics. Our findings thus appear to support the critical assumption that distance-independent competition metrics are sufficient for most operational growth and yield applications, even in managed, naturally-regenerated and species rich forests.