A site analysis and classification system for Eucalyptus grandis on the Zululand coastal plain.
A site analysis for Eucalyptus grandis planted on the Zululand coastal plain was carried out. Data from the permanent sample plot program from Mondi Forests was used to derive meaningful site quality relationships. Although water availability to trees is clearly identified as the single most important factor in forest land use management in South Africa, the matrix of other site factors such as soil, climate, genetic advancement and environmental constraints make timber plantations operationally complex and fascinating for research. The correlation between environmental parameters influencing tree growth and the yield obtained from a stand of trees is researched in this study. Growth models in the form of mathematical relationships are developed to enable the forest manager to predict tree growth from easily attainable input variables such as age, diameter or clay content of soils. The Chapman-Richards model was used to define the basic sigmoidal height growth curve over age for a given site. A site index model developed through a non-linear modelling process was constructed from permanent sample plot data. This model proved to be different from the site index model developed for a larger data set of the same physiographic zone. A site quality prediction model estimating site index at reference age five, from soil attributes was constructed. Soil morphology and grid referenced climatic data were found to be of limited value for the prediction of site index, but organic content in the top-soil and clay content in the sub-soil proved to be valuable predictors of site growth potential. For further site analysis studies, soil and climate variables will have to be measured on-site as opposed to using computer simulated figures. A site classification exercise was carried out by using the statistical technique known as clustering. Clusters were derived for the study area making use of clay content and mean annual precipitation (MAP) as input variables to separate the study area geographically, into meaningful structures on the basis of similarity. Significant clusters were derived using Ward's technique, which proposed three distinct site classification units for the study area. Site index for each of the site classification units was modelled and it was proved that the models predicted significantly different height - age relationships for each unit. From this site classification exercise it is shown that the variance in height growth within each site classification unit (SCU) is sufficiently small for each unit to be regarded as an independent site, uniform in its attributes of soil, climate, topography, water and nutrients status. A methodology for site classification is proposed from this study.