In plant ecology, research has focused on variation in traits at the scale of individuals or larger. This is especially true for chemical defenses. Several studies, however, have revealed how chemical defenses can vary greatly at very small scales, such as within-leaves, and that this variation can be important for herbivores. It is unknown how common variation is at these smaller scales and how it compares with variation at the scales that are traditionally examined. In this study, we measured seed chemical defense of Texas bluebonnet (Lupinus texensis) individuals at 8 sites in central Texas. We also compared seed mortality due to predation by a Lycaenid caterpillar and attack by fungus to determine if higher levels of chemical defense correlated with increased seed cohort survival. We also quantified variation in chemical defense at smaller scales, such as within pods and within branches of the same individuals. For a subset of individuals, variation at these scales was compared with variation at the individual and site levels. These across-scale comparisons have not often been made for plant defensive chemistry, yet are crucial for revealing the relative amount of variation in plant traits that has been overlooked.
We did not find a correlation, at the individual level, between concentration of defensive chemicals and seed mortality rate due to either enemy. In preliminary analyses, we did, however, find a linear relationship between chemical defense and seed position within a pod. Caterpillar seed-predators seem to selectively choose seeds from within a developing cohort and we also found a suggestive trend between caterpillar attack and seed position. Lastly, comparisons of variation in chemical defense across scales reveal that variation at the smaller scales (within pods, amongst pods, and amongst branches of the same individual) is large and comparable to that at the larger scales (amongst individuals and amongst sites). These results suggest that within-individual variation can be quite substantial and in fact comparable to variation at the scales that are traditionally used for measurements. Our results also imply that this small-scale variation in seed traits may be meaningful for selection by a seed-predator, however, more work is needed to resolve this relationship.