The aboveground biomass for the five species, segregated by overstory density, height class, and diameter size class is shown in Table 1. Foliage was separated from other live 0-0.63 cm diameter fuels because of its different chemical composition affecting combustion, and a higher probability of detachment from the other classes of fuel after crown scorching by fire. Foliar biomass averaged about one-third of total live aboveground biomass, with higher proportions in small trees and lower proportions in the 2-3 m height class. Attached dead fuels on these live trees usually comprised less than 10 percent of total aboveground biomass. There were only small quantities of live fuels in the greater-than-7.6 cm diameter size class and small amounts of dead fuels exceeding 0.64 cm diameter. Average physical dimensions, stand basal area, and age of sample trees are shown in Table 2.
Analyses of Variance
The three way complete analysis of variance for foliar biomass indicated that the main effects were all significant, as well as the species overstory density interaction. Orthogonal contrasts defined in advance of analysis showed that white fir foliar biomass exceeded that of red fir, and the two fir species foliar biomass exceeded that of the two pines. There were no differences between the two pines or between red fir and mountain hemlock. Trees grown in open stands had significantly more foliar biomass than trees grown under dense canopies. As expected, taller trees had more foliar biomass than shorter ones. The significance of the species-overstory interaction suggests that the foliar biomass of some species reacts differently to a change in stand density. Foliar biomass of the pines and firs increased with decreasing stand basal area. Foliar biomass of mountain hemlock was relatively stable over a wide range of stand basal area.
The sum of all live fuel classes showed significant biomass differences between density and height classes, but not between species; no interactions were significant. The total of all fuel classes had significant biomass differences for all three main effects. Orthogonal contrasts on the species main effect indicated that the total aboveground biomass of the two fir species combined exceeded that of the two pine species combined.
The analyses of variance indicated that aboveground biomass for a given species was affected by height and the subjective “open” and “dense” overstory density class. Other work has suggested that groundline diameter is a useful biomass predictor (Edwards and McNab, 1979). Diameter at breast height was not used here because roughly half of the trees were below that height. The independent variables chosen for biomass regressions were natural logarithms of: tree height times 10 (nearest 0.1 m); groundline diameter times 10 (nearest 0.1 cm); the square of groundline diameter times 10 multiplied by the height times 10; and basal area of stand plus one in m2/ha. The dependent biomass variables were also transformed to natural logarithms after 1 g was added to each value to avoid zeros in the data. The five biomass categories for each species were:
(live foliage + l)
(all live fuels between 0-0.63 cm + 1)
(live fuels between 0.64-2.53 cm + 1)
(live fuels between 2.54-20.32 cm + 1)
(total aboveground biomass + 1)
Regressions were individually calculated for “dense” and “open” overstory classes, and for combined overstory classes when basal area was added as a second independent variable. Segregation by overstory class was not done for mountain hemlock because of the analysis of variance results. Equations for each fuel class are presented in Tables 3-7. The single independent variable with the best goodness-of-fit is displayed, along with the multiple regression where basal area has been added. As with most regressions, the use of very low values of the independent variable may give unrealistic values of biomass.
Foliar biomass (Table 3) is best predicted by groundline diameter. Mountain hemlock is the only species where height consistently provided a better fit to the data. The close relationship between foliar biomass and basal diameter is consistent with the findings of Grier and Waring (1974); they found a high correlation between foliage mass of conifers and sapwood basal area. For the trees sampled here, most of the wood is sapwood, so that groundline diameter can be considered an index of sapwood diameter.
When foliar biomass is combined with other live fuels in the 0-0.63 cm diameter size class (Table 4), groundline diameter is again the best predicrences tion in the majority of cases; only mountain hemlock is better predicted by height. The volumetric variable, diameter2 times height, is the best predictor in 7 of the 20 equations.
The live fuel biomass in the 0.64-2.53 cm diameter class Is best predicted by volume or height (Table 5). These fuels are larger branch or upper stem components, so that height of the tree should be a good predictor of the biomass of this size class.
The biomass of the 2.54-20.32 cm diameter size class is best predicted by groundline diameter (Table 6). There were some zero values in these categories for the smaller trees, so that goodness-of-fit values were considerably lower than for the other fuel classes. For the larger trees, the basal section is included in this class, so groundline diameter is expected to be the best predictor of the largest sizes of biomass. Total aboveground biomass is best predicted by an index of volume (Table 7). Only for mountain hemlock is height a best predictor. In all cases, goodness-of-fit exceeded 95 percent; less than 5 percent of variance was unexplained by the equation.
Application of the best-fit equation will provide the most precise estimation of biomass, but for many purposes the procedure will be cumbersome, requiring measurement of height, basal diameter, or both. Because height (or height class) is most easily measured, a listing of biomass equations using height as the independent variable is provided in Table 8.
The goodness-of-fit by species for open and dense overstory group equations tended to encompass the fit obtained by pooling the data within species and using stand basal area as a second independent variable. For some species, there was little difference between the separate and pooled R2 values. Where a wide range in goodness-of-fit existed between equations for open-grown and dense understory-grown trees, the pooled data tended to average the fit with the addition of a point estimate of basal area. For example, the foliar biomass R2 for dense understory grown lodgepole pine was 0.70; using a pooled data set and stand basal area increased the R2 to 0.85. Conversely, using the pooled equation for open grown lodgepole decreased R2 from 0.95 to 0.85.
Similar regression techniques were employed to predict age of trees from physical dimensions. Age data were not transformed, and basal area was used untransformed and with a natural logarithmi transformation. Using thirteen different groups of tree ages segregated by species and overstory density (Table 9), the same combinations of independent variables used for biomass estimation were regressed on the age data.
The fit of the equation is generally much lower than for the biomass equations. The range of R2 for the several equations calculated for each combination of species and density (Table 7) was 0.20-0.96; only four groups had R2 exceeding 0.80. Equations for the best set of variables for each group are shown on the right side of the table.
It is apparent that the age of small understory trees in this study can be only generally estimated from commonly measured physical dimensions. Trees in the 0-3 m height range growing under dense canopies were more precisely aged than open grown trees, which is surprising given the stagnation and erratic growth often associated with understory trees in dense forests. However, the equations are of marginal utility even for local application, and extrapolation to other areas is not recommended.