1. The optimal defence theory (ODT) of chemical defence provides a predictive framework for the distribution of anti-herbivore defences in plants. One of its predictions is that chemical defences will be allocated within a plant as a function of tissue value, where value is correlated with the cost of having that tissue removed. While many studies have examined intra-plant variation in defence chemistry, these results have rarely been compiled quantitatively to assess whether defence allocation is consistent with the prediction of ODT that more valuable tissues should be more defended than less valuable tissues.
2. We performed a formal meta-analysis of published and unpublished studies to examine the predictive utility of ODT. Specifically, we examined whether defence chemicals occur at higher concentrations in flowers versus leaves and in younger leaves compared to older leaves, under the assumption that younger leaves are more valuable than older leaves. We also examined whether the expansion status of younger leaves, nodal position of the leaves, growing conditions and chemical class of defence compounds affected the mean effect sizes.
3. We found that tissues with higher assumed value had significantly higher concentrations of defence chemicals than tissues with lower value. In particular, we found that younger leaves had higher concentrations of defence chemicals than older leaves, consistent with the predictions of ODT. The magnitude of this difference was higher in the younger leaf/older leaf comparison than in the flower/leaf comparison, with no evidence that flowers were more defended than leaves. The overall results were not affected by chemical class, young leaf expansion status, growing conditions or leaf position on the plant.
4. Synthesis. We found general agreement between the predictions of ODT and the intraplant distribution of chemical defence and conclude it is a useful model. The effect size varied depending on the tissue compared. Explicit measures of tissue value, in particular as it relates to relative fitness, are required to further validate the predictive utility and general applicability of ODT.