Forest volume growth and yield are often viewed as functions of site quality, age, and some measure of stand density, as well as interactions among these variables. Stand density, in turn, is taken to be a function of site quality, age, and an initial measure of stand density. Site quality, expressed by site index, depends on the development of dominant height in relation to age (Clutter et al. 1983). It is apparent that forest growth and yield modellers deal with an interdependent system with underlying growth processes. Each equation in such a system describes a different relationship among a different set of the variables in the system, but all relationships are assumed to hold simultaneously. However, in most cases, the interdependence inherent in stand dynamics is ignored in model fitting. It is customary in forest procedures to obtain parameter estimates for each model separately, within a model system (Borders and Bailey 1986).
Growth models are of limited use on their own, and require ancillary data to provide useful information. With suitable inventory and other resource data, growth models provide a reliable way to examine silvicultural and harvesting options and to determine the sustainable timber yield for different areas and management strategies (Vanclay 1994). Growth models may also have a broader role in regional forest management and in the formulation of forest policy.
Sommer and Dow (1978) suggested there is often an evolutionary trend in the development of growth and yield data:
a) production statements;Few tropical and subtropical countries have sufficient plantation experience with most exotics to reach the stage of yield regulation tables. Consequently their data falls mostly into the first two categories. Some countries have produced provisional or tentative yield tables of one kind or another but only those with long experience and a large database are in a position to construct comprehensive yield-regulation tables or management manuals (Sommer and Dow 1978). However, advanced computer simulation models are available for some species (see below).
b) production series;
c) provisional yield tables;
d) normal yield tables;
e) empirical yield tables;
f) variable density yield tables;
g) yield regulation and management tables;
h) growth simulation models
The greatest obstacle to developing advanced simulation models for many tropical forests is that suitable data is not available. Often yields are estimated by extrapolating models developed elsewhere or using generalised information such as is presented in this report.
Forest growth models fall into a number of groups, each of which is characterized by a central design paradigm, and each of which is most suitable for a particular situation. The selection of a modelling paradigm is driven mainly by currently or potentially available data, and the objectives for which the model is to be constructed (Alder 1995). Vanclay (1994) noted that there is no best method for forest growth models.
A completed model must undergo a process of validation or testing. This is both an ongoing and terminal activity. Ongoing validation is an integral part of the development of a model, whilst terminal validation seeks to define the limits of predictive capability (Alder 1995).
Simulation modelling has been motivated by the financial and policy structures of plantation forestry with close industrial link. This electronic method permits the construction of growth models by iterative solution. Such models can be manipulated to provide appropriate technical solutions for optimal management objectives (Sommer and Dow 1978).
A simultaneous growth and yield model for P. elliotti plantations was developed by Pienaar and Harrison (1989) using Correlated Curve Trend (CCT) data from the coastal regions of Zululand. The model includes equations for predicting height, basal area, total stem volume, marketable volume and tree survival, on a stand level (von Gadow and Bredenkamp 1992). A range of similar models, including looking at how growth, silviculture and stem quality impact on end-use, have been developed in New Zealand for P. radiata (Burdon and Miller 1992). Other growth models for this species have been developed in Chile and Australia.
The “Langepan” model is a well-documented growth model for Eucalyptus grandis stands, based on CCT spacing trial on an excellent growing site near KwaMbonambi in Zululand. The Langepan model dramatically illustrates the effect of planting spacing on diameter and height. Langepan model predictions for mean tree heights and volumes are possible for use in different climatic zones (von Gadow and Bredenkamp 1992).
