Ecological genetics of Bromus tectorum : I. A hierarchical analysis of phenotypic variation.

An understanding of how genetic differentiation and phenotypic plasticity may interact to promote the spread of an introduced species requires information on the hierarchical distribution of genetic variation within the species in its new range. For example, a lack of genetic variation within marginal populations of an introduced species may slow its rate of spread into new habitats. In a glasshouse study, we examined the phenotypic variation among populations, among families, within families, and the homogeneity of variances within families for morphological and life history characters of an alien, self-pollinating annual grass, Bromus tectorum. The populations of B. tectorum studied were collected from both large, central populations in steppe habitats and small, peripheral populations in forest sites along a broad temperature-moisture gradient in western North America. Most variation in average flowering time was due to differences among seed source populations; among family variation contributed less than 20% to the total variance for flowering time. Populations from arid steppe habitats were the earliest flowering while the population from the most mesic forest habitat was last to flower. In contrast, the within-family variance was a major contributor to the total variance for plant dry weight, seed number per plant, total seed weight, and individual seed weight. The amount of total variation explained by among-family differences ranged between 18% for average seed weight to 30% for total plant dry weight. There was no consistent difference in within-population genetic variability between large, central populations in steppe habitats and smaller, potentially more isolated populations in forest habitats. Significant heterogeneity in within-family variance in some of the source populations suggest that families differ in the capacity for phenotypic response to environmental variation. Considered independently from source population, there was no consistent trend linking a particular trait to increased heterogeneity of within-family variances.