Needle litterfall prediction models for even-aged natural shortleaf pine ( Pinus echinata Mill.) stands

Abstract

Needle litterfall was collected over seven growing seasons in two even-aged natural shortleaf pine ( Pinus echinata Mill.) stands in southeast Oklahoma. The data were used to develop a model quantifying the relationship of stand and site characteristics and weather variables to annual needlefall. This model is of use to forest managers interested in forecasting future fuel loads, or producing alternative forest products. Because of their proximity to the western edge of the natural range of shortleaf pine, long-term monitoring of these stands may yield information about the effects of climate change on litterfall. Spring temperatures during the growing season in which the needles were produced, the amount of needlefall that occurred two seasons previous, and site index were statistically significant explanatory variables in the model. The model explained 63 percent of the variation in the data. A comprehensive set of statistical misspecification tests employed to verify the assumptions of linear regression models indicated the hypotheses of normality, linear functional form, homoskedasticity, parameter stability, and independence could not be rejected. Finally, because historical needlefall data may not exist for many stands, a three-stage least-squares model was developed to forecast needlefall quantities.