Results of structural equation model (SEM) and path analyses of the direct and indirect effects of the QTL. The focal plant traits were combined into a latent factor, and the same process was done for the neighbour plant. Two types of SEMs were constructed: an identified pleiotropy model, where the locus affected the expression of only the focal plant traits directly, and an unidentified pleiotropy model where the locus had a direct impact on (e.g. a path to) the neighbour plant traits. The effects from the identified and unidentified pleiotropy models indicate standardized regression weights (or path coefficients) indicating the relative amount of variation explained by models. Comparing the effects in the identified pleiotropy model to those in the unidentified pleiotropy model gives a measure of how much of the functional basis of the phenotypic interaction is described by the traits in the model, and in this case shows that the phenotypic ‘indentified pleiotropy model’ is a much better description of the interaction. See figure 4 for a diagram of this model.