|Abstract or Summary
- Index selection provides an efficient means of conducting selection on multiple traits by combining information on economic value,
heritability, and the genetic and phenotypic correlations between traits to improve overall merit. The use of this method in forestry has been hampered by the lack of knowledge of the relative importance of
individual traits in determining tree value. Data recently made available from a lumber recovery study made it possible to a) estimate relative economic weights for individual tree traits that are important in determining lumber value of coastal Douglas-fir (Pseudotsuga menziesii (Mirb.) France var. menziesii) under two grading systems (visual and machine-stress-rated (MSR)) and b) test the implications of
alternative weights under various multi-trait selection indices. Economic weights were derived by stepwise regression of total tree value, based on either visual or MSR grading of recovered lumber, on a
variety of traits measured on 164 young-growth (ages 36-66 years) Douglas-fir trees in western Washington and Oregon. Two traits were found to have a significant influence on tree value under visual grading, stem volume and branch diameter, with relative economic weights 0.06 dm3 and -5.22 cm, respectively. In addition to volume and branch diameter, wood density also significantly influenced tree value under
MSR grading (relative economic weights 0.06 dm3, -6.69 cm, and 0.06 kg/rn3, respectively), where lumber strength is measured directly. Selection indices and expected responses of volume, branch diameter, and wood density to index selection were examined utilizing progeny test data from 20-year old trees of 85 open-pollinated families replicated on three planting sites. The derived selection indices were quite different for the two lumber grading schemes, as were the expected
responses in individual traits. Under visual grading, expected response in volume is large, but because wood density was not included in the index, a large negative (unfavorable) response in wood density is also expected. Using MSR grading, a positive response is expected in both volume and wood density, but the expected response in volume is less than when visual grading is used to determine economic weights. Little response in branch diameter is expected even though branch diameter was included as a trait in both indices. This is due to the low variability of branch diameter and its adverse genetic correlation with volume.
Changes in genetic and phenotypic parameter estimates (tested by using only a portion of the progeny test data) resulted in different
index coefficients and changes in expected responses of individual traits. These results emphasize the necessity of evaluating the implications of using a particular index prior to applying it to a
breeding population. In one case, a desired gain index, with desired relative responses for volume and wood density equaling the MSR-based economic weights, was successful in predicting positive responses in both volume and wood density where a Smith-Hazel index had resulted in a negative response in volume. Thus, this type of restricted selection index provides a means of increasing overall merit while dictating the
allowable changes in each trait.