Resumen
We analyzed variations in three tree properties: tissue density, carbon fraction, and carbon density within bole tissues of nine Californian conifer species. Model performance for all three tree properties was significantly improved with the addition of covariates related to crown characteristics and position within the tree. This suggests that biomass and carbon mass estimates that rely on fixed wood density and carbon fraction may be inaccurate across tree sizes. We found a significant negative relationship between tissue density and carbon fraction within tree bole tissues, indicating that multiplying biomass by an average carbon fraction to obtain carbon mass is likely to lead to inaccurate estimates. Measured carbon fractions in tree tissues deviated from the widely used 0.5 value from a low of 1.4% to a high of 17.6%. Carbon fraction model parameters indicate the potential for an additional deviation from this 0.5 value of up to 2.7% due to the interaction between relative height and wood density. Applying measured carbon fractions to whole bole biomasses resulted in carbon mass estimates as much as 10.6% greater than estimates derived using the 0.5 value. We also found a significant, though modest, improvement in carbon fraction model estimates by assigning trees to groups based on tree bark characteristics.