Intra-specific variation in mortality of even-aged Cryptomeria japonica (L. f.) D. Don. forests can be explained using relationships among long-term stand characteristics

Abstract Context Understanding tree mortality is critical for sustainable forest management. Long-term tree mortality may differ depending on the stand development process and can be influenced by forest management regimes. Logistic regression is widely used to explain tree mortality based on site productivity, age, size, and competition. However, the explanatory variables are interrelated. Thus, we attempted to explain long-term Japanese cedar tree mortality by considering interrelated variables. Aims The aim of this study was to elucidate the direct and indirect effects of site productivity, age, individual size, and competition on the long-term mortality of Japanese cedars. Methods Data were collected from 5130 even-aged Japanese cedar trees over approximately 50 years. We compared each variable between dead and living trees. We then constructed a mortality model using a conventional logistic approach and selected the best model for the stepwise methods. Finally, we applied a piecewise structural equation model (SEM) to identify these variables’ direct and indirect effects. We compared the conventional logistic model and piecewise SEM models and discussed the advantage of applying the SEM models. Results Annual mortality was approximately 4% in the most fertile stands, increasing gradually with decreasing site fertility. Dead tree size and competition status differed according to age and site productivity. Competition, individual size, and stand density were selected for the best logistic model (area under the curve (AUC) = 0.74, Brier score = 0.042), whereas age and site productivity were not ( p > 0.05). The piecewise SEM results showed that age and site productivity indirectly affected tree mortality through individual size and stand density (Fisher’s C = 4.569, p = 0.102). Conclusion Long-term Japanese cedar tree mortality can be explained by individual size and competition as direct influencing factors and age and site productivity as indirect influencing factors. This indicated that hidden factors cannot be explained using the conventional logistic approach. Further studies are required to explore the potential factors contributing to tree mortality thoroughly.