Estimating biomass of white spruce seedlings with vertical photo imagery

Estimation of individual tree seedling biomass is required in a variety of forest management and research applications such as assessment of net prima ry productivity and carbon sequestration potential of forest stands, unders tory forest fuel inventories, and development of silvicultural guidelines t o promote the growth of desired tree species. Photo imagery is a promising non-destructive method for estimating the aboveground biomass of tree seedl ings. This method was tested using naturally regenerated white spruce (Pice a glauca (Moench) Voss) seedlings growing in the understory of a mixed coni fer shelterwood in central Ontario. In the fall of 1997, 45 seedlings were sampled from plots exposed to one of three mechanical release treatments (e arly spring release, mid summer release, and no release (control)) in 1994. Each seedling was photographed in the field to measure the vertical projec ted area (silhouette area) of the aboveground portion of the seedling. Seed lings were harvested, basal diameter and total height measured, and biomass (dry mass) of foliage, branches, main stem and total aboveground plant tis sue determined. Regression analysis revealed a strong relationship between both silhouette area and basal diameter, and seedling biomass. Coefficients of determination for regression equations using silhouette area were equal to 0.892, 0.918, 0.926, and 0.937 for the main stem, branches, foliage, an d total aboveground biomass, respectively. Respective coefficients of deter mination for regression equations using basal diameter were 0.960, 0.945, 0 .953, and 0.977. Silhouette area-based equations for total aboveground and foliar biomass differed significantly (P < 0.005) among release treatments. No significant differences among treatments were observed between silhouet te area-based equations for biomass of branches and main stem (P > 0.05), o r between basal diameter-biomass (allometric) equations for all components (P > 0.1). The method was then tested by validating the biomass equations u sing an independent data set from 35 white spruce seedlings from the same s ite and cohort, but exposed to different treatments and microenvironmental conditions. For each seedling, biomass components were predicted using silh ouette area-based and allometric equations, and a relative error of predict ion calculated. The mean relative error for silhouette area-based predictio ns varied among biomass components from -20.25% to -3.21%, with standard de viation of the error ranging from 23.04% to 33.44%. The mean relative error for allometric equations ranged from -2.46% to -21.75%, with standard devi ations of 23.34% to 32.61%. These results suggest that: (1) photo imagery c an be used as an alternative to more traditional allometric methods of biom ass estimation, and (2) general (developed for a broad range of growing con ditions) equations derived by either method are preferable to those specifi cally calibrated for a given growing environment.