Estimating biomass of white spruce seedlings with vertical photo imagery

Estimation of individual tree seedling biomass isrequired in a variety of forest management andresearch applications such as assessment of netprimary productivity and carbon sequestrationpotential of forest stands, understory forest fuelinventories, and development of silviculturalguidelines to promote the growth of desired treespecies. Photo imagery is a promising non-destructivemethod for estimating the aboveground biomass of treeseedlings. This method was tested using naturallyregenerated white spruce (Picea glauca (Moench)Voss) seedlings growing in the understory of a mixedconifer shelterwood in central Ontario. In the fall of1997, 45 seedlings were sampled from plots exposed toone of three mechanical release treatments (earlyspring release, mid summer release, and no release(control)) in 1994. Each seedling was photographed inthe field to measure the vertical projected area(silhouette area) of the aboveground portion of theseedling. Seedlings were harvested, basal diameter andtotal height measured, and biomass (dry mass) offoliage, branches, main stem and total abovegroundplant tissue determined. Regression analysis revealeda strong relationship between both silhouette area andbasal diameter, and seedling biomass. Coefficients ofdetermination for regression equations usingsilhouette area were equal to 0.892, 0.918, 0.926, and0.937 for the main stem, branches, foliage, and totalaboveground biomass, respectively. Respectivecoefficients of determination for regression equationsusing basal diameter were 0.960, 0.945, 0.953, and0.977. Silhouette area-based equations for totalaboveground and foliar biomass differed significantly(P < 0.005) among release treatments. Nosignificant differences among treatments were observedbetween silhouette area-based equations for biomass ofbranches and main stem (P > 0.05), or betweenbasal diameter-biomass (allometric) equations for allcomponents (P > 0.1). The method was thentested by validating the biomass equations using anindependent data set from 35 white spruce seedlingsfrom the same site and cohort, but exposed todifferent treatments and microenvironmentalconditions. For each seedling, biomass components werepredicted using silhouette area-based and allometricequations, and a relative error of predictioncalculated. The mean relative error for silhouettearea-based predictions varied among biomass componentsfrom –20.25% to –3.21%, with standard deviation ofthe error ranging from 23.04% to 33.44%. The meanrelative error for allometric equations ranged from–2.46% to –21.75%, with standard deviations of23.34% to 32.61%. These results suggest that: (1)photo imagery can be used as an alternative to moretraditional allometric methods of biomass estimation,and (2) general (developed for a broad range ofgrowing conditions) equations derived by either methodare preferable to those specifically calibrated for agiven growing environment.