Growth Model Evaluation: South-wide Loblolly Pine Seed Source Study
A growth model useful to the forest geneticists should have the following five P's: (1) perfection of fit to the data presented, (2) predictability of growth potentials, (3) possibility of inference based on parameter estimates, (4) power of discrimination among seed sources, and (5) persistency of regression coefficients over time. To illustrate, the function of ln(HT) = BO + B1/AGE + B2 ln(1 + 1/AGE) was fitted to the height growth of 15 seedlots of loblolly pine in a south-wide study. The degree of determination was 0.999 at least. The five-year and the ten-year projections were low by 3% and 5% respectively. The regression coefficients BO, B1 and B2 were highly significant among seedlots. When the three growth periods were compared, the coefficient of variation for the regression coefficients was less than 5%.
Download this file:
Download this file — PDF document, 397KbDetails
Author(s): Fan H. Kung
Publication: Tree Improvement and Genetics - Southern Forest Tree Improvement Conference - 1985