Apiolaza, L.A., Gilmour, A.R. and Garrick, D.J. 2000. Canadian Journal of Forest Research 30: 645–654. (PDF paper 133 KB).

Apiolaza, L.A. and Garrick, D.J. 2001. Canadian Journal of Forest Research 31: 654–662. (PDF paper 106 KB).

Apiolaza, L.A., Gilmour, A.R. and Garrick, D.J. 2000. Canadian Journal of Forest Research 30: 645–654.

Breeding objectives for three silvicultural regimes of radiata pine

Apiolaza, L.A. and Garrick, D.J. 2001. Canadian Journal of Forest Research 31: 654–662.

A generic vertically integrated firm, comprising a production forest, a sawmill, and a pulp mill was modelled under three silvicultural regimes: direct to pulp, intermediate (includes production thinning), and intensive (includes production thinnings and pruning). The harvest age traits included in the breeding objective were total volume (m³/ha) and average wood density (kg/m³). Economic values for each trait were calculated as the difference in discounted profit for a unit marginal increase of volume or density, and expressed as relative weights to facilitate comparisons between the objectives. The methodology was applied to a Chilean case study using representative economic and production circumstances. The breeding objectives so derived were 1vol + 2.4den for pulp, 1vol + 1.1den for intermediate, and 1vol + 1.2den for the intensive regime, where vol and den are the breeding values for volume and density, respectively. The firm was profitable under all regimes. Genetic correlations between the objectives for each regime were higher than 0.9, indicating that a single breeding strategy with objective 1vol + 1.5den could be adopted, with almost no loss of genetic gain relative to selecting for a particular silvicultural regime.

Canadian Journal of Forest Research

The Canadian Journal of Forest Research is my favourite ‘generic’ forestry journal. Below there are some abstracts of papers of mine published in this journal.

Abstracts

Variance modelling of longitudinal data from progeny tests

Apiolaza, L.A., Gilmour, A.R. and Garrick, D.J. 2000. Canadian Journal of Forest Research 30: 645–654.

We estimate variance components for height (m) of radiata pine trees measured at 10 unequally spaced ages in an open-pollinated progeny using a mixed model. We compare several forms for the additive genetic covariance matrix (G₀). These include structures based on autoregressive, banded correlations and random regressions models as well as the full unstructured form. We use the full unstructured form for the residual matrix (R₀) and an autoregressive form for the block and plot strata matrices. The best model for G₀ considering the likelihood value and number of parameters is the autoregressive correlation form with age specific variances. It gives a genetic correlation between successive measures of 0.987^lag, with lag expressed in years. Heritability under this model increases with age from 0.078 (age 1) to 0.206 (age 7) and the declines to 0.099 at age 15. The heritability pattern with banded correlations is similar to that with autoregressive correlations. The heritabilities under the full unstructured form are slightly higher. The covariance structure implicit in the random regressions model appears quite different and is considered unsatisfactory. Putting structure into G₀ smoothes the estimates of genetic parameters, facilitating model fitting and convergence of the likelihood maximisation algorithm. We conclude that fitting a structured matrix, such as that based on autoregressive errors, which reflects the relationship present in repeated measures, will often do better than the full unstructured matrix for longitudinal data.