Savanna browse production. 2: Prediction

Abstract

Browsing herbivores are generally highly selective feeders, selecting for new shoots and young plant material. Determinations of browser carrying capacity are then greatly inflated when based on available browse biomass estimates. Our study was initiated with the aim of modelling browse production rates of key savanna tree species in the northern Zululand region of KwaZulu-Natal, with the assumption that browse production equates to utilised browse biomass. Predictive models for the production of browse, at different game feeding levels, on deciduous, semi-deciduous and evergreen trees were developed using multivariate adaptive regression spline functions. The most significant predictors of the production of browse were measurable plant dimensions, namely available canopy volume, stem diameter and foliage density, and climatic variables, namely maximum and minimum daily temperatures and cumulative daily rainfall for each quarter. Comparing the browse production figures of consecutive years supplies land managers with valuable insight whether the browse component of the system is under stress, is maintaining, or is encroaching, as well as the impact of browsers on the system.

Keywords:

• browser carrying capacity
• growing season
• multivariate adaptive regression splines
• predictive models