Quantifying aboveground biomass for common shrubs in northeastern California using nonlinear mixed effect models

Abstract

Quantifying shrub biomass can assist in natural resource management decision making. Nonlinear mixed effect models (NMEM) were developed to predict total aboveground biomass as well as biomass in leaves, 1-h, 10-h, and 100 or more hour fuel classes for seven species of shrubs common to the northeastern California. Using crown area as a predictor, an allometric (power) model was used as a base model. Coefficients varied by species, component, and by a nested combination of these random effects.The results showed that NMEM that used shrub species as random effect performed better than nonlinear fixed effect models in estimating total and component biomass in shrub species used in this study. Additionally, when fixed effect models were fitted by species, not all regression parameters were statistically significant at 0.05 level of significance. NMEM were able to account for within species variation very well. The largest variation was observed in total biomass while the smallest variation was observed in the biomass in 100 or more hour fuel class. The mean prediction bias and root mean square prediction errors for total shrub biomass was 0.0409 kg and 0.9249 kg respectively. While there were differences between the fixed effects models and mixed effects models, the mixed effects models would be preferred to the fixed effect models for future studies involving total biomass prediction for similar shrub species and regions.