Individual And Stand Mortality Model For Larix Gmelinii Subsp. Principis-rupprechtii. By Using Forest And Climatic Variables

A complex of endogenous and exogenous factors acting as induction agents lead to tree death during forest succession.It is an important content of forest ecosystem succession,forest growth process simulation and prediction.The effect of climate on tree mortality was identified as an important exogenous factor.Thus,it is very important to develop the forest mortality model to accurately reflect the forest survival in different stages,to accurately understand and grasp the dynamic change law of forest ecosystem under different climatic conditions,and to guide the sustainable forest management and adaptive management under the background of climate change.Larix(Larix gmelinii subsp.principis-rupprechtii.)is the main afforestation tree species in northern China,and therefore it dominates the forest ecosystems of the region.Larch forests are usually characterized by a large amount of biomass and high primary productivity.Larch forests can play critical roles in the regional carbon storage and carbon cycling.This study used the Larix natural forest as the research object.We established 102 temporary sample plots(TSPs)in state-owned Larix forests in Shanxi Province,China to collect mortality data: 67 in the Guandi Mountain Forest and 35 in the Boqiang Forest.Due to the hierarchically structured-data,spatially correlated observations,climatic effect on forest mortality,and the great difference of tree mortality in different regions.Based on the principle of modern statistical model,we developed the individual tree mortality models,diameter class mortality models,and stand mortality models.The main results are as follows:(1)It is more suitable to describe the mortality data of nested structure(sample plot /block)by building different scales models and adding random effect.Logistic model was the best fitting model for individual tree / diameter class model,and hurdle model was the best fitting model for stand model.(2)The difference of tree mortality in this study area was large,which had the greatest relationship with SMD.With the increase of DT,the tree mortality decreased;With the increase of MAP,the number of mortality increased.(3)Two smallest sample diameter classes are recommended for estimating the random effects and model localization when both the measurement cost and potential errors of the predictions need to be balanced.