In order to manage the forest plantations, yield tables play a vital role by providing the necessary growth information against the age for different site types. Although the complex growth models have become popular recently, still yield tables are very widely used for prediction of growth and yield.
Alstonia macrophylla Wall ex G. Don commonly known as hawari nuga was an introduced species to Sri Lanka from Malaysia. Although the timber quality is not valuable as that of most of the commercial timber species in Sri Lanka, the Forest Department and the private sector have established a considerable amount of plantations of this species in the low country wet zone due to its fast growth rate and high adaptability to different site types. In addition to that, A. macrophylla is growing as individuals and small blocks in homegardens, barren lands and edges of natural forests. However, at present, a yield table or any other method is not available for this species to predict its growth for different site types.
Therefore the present study was conducted with the objectives of classifying the site quality of A. macrophylla plantations and to construct yield tables for those site classes. For this reason, 23 even-aged plantations and 29 blocks (non-plantations) were selected from Galle, Kalutara, Matara and Ratnapura districts. The selected plantations varied in age from 5 to 22 years and the age could not be identified for the non-plantations due to the unavailability of records or due to the growth of trees of different ages. The data collected were the dbh, total height, timber height, crown dimensions and the necessary data for the calculation of stem volume by using Newton’s formula.
In order to classify the sites, a top height related index was used and it was possible to identify two different site types for A. macrophylla. However, there were no adequate amount of plantations for the site class I and therefore a yield table could not be constructed for that site although all the necessary growth models were constructed in this study.
In order to construct the yield table, first a height-age relationship was built to predict the total height at any age. Then the tree dbh was modelled against total height so that the dbh of any age can indirectly be predicted by using the dbh-height model. Using dbh and total height another model was constructed to predict total stem volume and in addition to that, a merchantable volume prediction model was built from dbh and total height. In addition to the above models used for the construction of yield tables, a separate simple volume prediction model which is independent from the site quality was built for the trees growing as non-plantations.
Other than the untransformed variables, four transformations were used in this study for the selected candidate variables in order to identify the best models with the highest accuracy. Data were fitted to both linear and non-linear forms and the preliminary selection of the suitable models were done based on the R2 value. In order to identify the best models among the selected ones, both qualitative (standard residual distribution, fitted line plots) and quantitative (average model bias, mean absolute difference and modelling efficiency) tests were utilised. For the finally selected models the biological reality was also tested. As a rule, it was expected build the models of same structure to predict the similar variable for all site types. A validation test was also conducted with independent data for the finally selected models to identify the suitability for the use in the field.
The selected models to construct the yield table were very low in bias. However, certain models did not indicate a very high R2. The reason may be the high variability of the data in each age. Finally a yield table was prepared for the site class II with the final harvest at 20 years after planting. The recommended initial spacing was 2.5x2.5 m and also two thinnings were recommended in between.
In addition to the above, the site quality of A. macrophylla growing in the selected four districts was further classified using GIS in order to prepare a site quality map.