| Forest genetic experiments have traditionally relied on simple pedigrees and model assumptions regarding the distribution of genetic and environmental effects. In this thesis, measured quantitative traits, collected environmental data and pedigrees reconstructed using DNA markers were used to test the underlying assumptions of these simple genetic models, in populations of red spruce ( Picea rubens) in Nova Scotia, Canada. In turn, an alternative selection strategy is proposed using the reconstructed pedigrees. First, open pollinated and polycross mating systems are two simple pedigree structures, where the progenies are assumed to be related maternally, as half-sibs. Complete, complex pedigrees were reconstructed using likelihood based paternity exclusion, to assess the degree of differential male reproductive success. Mating was highly skewed across, but not within half-sib families. Second, an individual tree, mixed model was used to assess the impact that skewed mating had on quantitative genetic parameters, in open pollinated and polycross progeny tests. Genetic response to selection increased markedly, but the increased response was attributed to identifying the male source of variation, not due to the correction of skewed male mating. Third, relatedness within natural, mature tree populations is often assumed to be present, as the result of past limited seed dispersal. Pedigrees in two natural stands were reconstructed with DNA markers using a group likelihood approach and were fit into an individual tree, mixed, random regression model to describe incremental growth over time. Pedigree structure was too weak to estimate genetic effects, but removing fixed environmental effects allowed more parsimonious comparisons among trees; important for plus tree selection within stand. Finally, the distribution of within site environmental effects, often assumed to be homogeneous, was explored in progeny tests. Electrical conductivity was mapped at a fine spatial scale, to test its use as a surrogate environmental variable driving tree growth. Electrical conductivity displayed patchy spatial structure much stronger than patchy growth structure, but the two were not correlated. In conclusion, genetic evaluation could be shifted from progeny testing to populations where the relatedness is known to be strong, provided the genetic gain achieved compensates for the cost of pedigree reconstruction. |
