Comparison Of Models For Estimating Stem Surface Area Of Coniferous Trees Grown In Old-Growth Natural Forests

Stem surface area (S) plays an important role in the eco-physiological processes of trees or forests such as stem respiration, self-thinning mortality, and rainfall interception. As the direct measurement of S is time-consuming and labor-intensive, models for predicting S from commonly measured tree attributes have been developed for coniferous trees grown in plantations. However, there have been no models for trees grown in natural forests. In this study, we compared regression models for estimating S using 122 sample trees of eight coniferous species felled in old-growth natural forests in Kiso district, Nagano prefecture, central Japan. The relationship of S to the product of diameter at breast height and tree height (DH) could be expressed as S = 1.924DH (R (2) = 0.996), independent of the species. The estimated slope coefficient of the regression of the natural forests was close to that of plantations reported in a previous study. These findings indicated the generality and wide applicability of the model. By contrast, the estimated slope coefficient of the regression between S and basal area (G) varied with the species, and the slope was gentler in the natural forests than the plantations. Monte Carlo simulation revealed that only 20 sample trees were necessary to estimate regression coefficient for the relationship between S and DH, whereas more than 60 trees were needed if G was used as predictor. In conclusion, the regression model between S and DH is useful when predicting S of various coniferous trees grown in both natural forests and plantations.