We just had a paper published that explores the use of nonlinear beta regression for estimating litter decomposition rates in single-pool models. Our initial hypothesis was that beta regression would be better suited to litter decomposition rate estimation because of the bounded nature of proportional litter mass loss data, and because of reduced variance near bounds. However, we found that nonlinear normal regression was actually quite good, despite its assumptions being violated. The beta model, however, should still be useful. Download the paper on the publications page |

news >