In the framework of phenotypic plasticity, tolerance to browsing can be operationally defined as a norm of reaction comparing plant performance in undamaged and damaged conditions. Genetic variation in tolerance is then indicated by heterogeneity in the slopes of norms of reaction from a population. We investigated field gentian (Gentianella campestris) tolerance to damage in the framework of phenotypic plasticity using a sample of maternal lines from natural populations grown under common garden conditions and randomly split into either a control or an artificial clipping treatment. We found a diversity of tolerance norms of reaction at both the population and family level: the impacts of clipping ranged from poor tolerance (negative slope) to overcompensation (positive slope). We detected heterogeneity in tolerance norms of reaction in four populations. Similarly, we found a variety of plastic architectural responses to clipping and genetic variation in these responses in several populations. Overall, we found that the most tolerant populations were late flowering and also exhibit the greatest plastic increases in node (meristem) production in response to damage. We studied damage-imposed natural selection on plasticity in plant architecture in 10 of the sampled populations. In general, there was strong positive direct selection on final number of nodes for both control and clipped plants. However, the total selection on nodes (direct + indirect selection) within each treatment category depended heavily on the frequency of damage and cross-treatment genetic correlations in node production. In some cases, strong correlated responses to selection across the damage treatment led to total selection against nodes in the more rare environment. This could ultimately lead to the evolution of maladaptive phenotypes in one or both of the treatment categories. These results suggest that tolerance and a variety of architectural responses to damage may evolve by both direct and indirect responses to natural selection. While the present study demonstrates the potential importance of cross-treatment genetic correlations in directing the evolution of tolerance traits, such as branch or node production, we did not find any strong evidence of genetic trade-offs in candidate tolerance traits between undamaged and damaged conditions.
