Multispecies Allometric Models Predict Grass Biomass in Semidesert Rangeland

Abstract

Multispecies allometric models to predict grass biomass may increase field study efficiency by eliminating the need for species- specific data. We used field measurements during two growing seasons to develop single-species and multispecies regression models predicting the current year’s aboveground biomass for eight common cespitose grass species. Simple and stepwise regression analyses were based on natural log expressions of biomass, basal diameter, and height, and a dummy variable expression of grazing history. Basal diameter had the strongest relationship with biomass among single-species (adjusted R2 = 0.80 to 0.91) and multispecies (adjusted R2 = 0.85) models. Regression slopes (b) for diameter among single-species (b = 1.01 to 1.49) and the multispecies (b = 1.25) models suggests that biomass will double when diameter increases <75%. Height and grazing history added little predictive value when diameter was already in the model. When applied to actual populations, biomass estimates from multispecies models were within 3-29% of estimates from the single-species models. Although the multispecies biomass-size relationship was robust across the cespitose life-form, users should be cautious about applying our equations to different locations, plant sizes, and population size-structures.