World climate patterns in grassland and savanna and their relation to growing seasons

Abstract

The climate at eleven IBP savanna or grassland study sites from five continents are described and principal components analysis is used to compare them. A multivariate linear discriminant function based on mean monthly precipitation, mean monthly temperature, latitude and altitude, is used to predict the length of the growing season at each site. At most sites, the actual and predicted start and end of the growing season agreed closely. It is concluded that growing season on a world-wide basis may be predicted fairly reliably from a small number of abiotic variables by means of a multivariate discriminant function.