In natura, metal hyperaccumulating plants usually have foliar metal concentrations 100 times higher than non-accumulating, neighbouring plants. The evolutionary significance of metal hyperaccumulation by plants is still unclear. However, the defence hypothesis, which suggests that metals concentrated in leaves of hyperaccumulators act as “elemental” defences against herbivores, has attracted most attention.

The toxicity of metals to herbivores is the base of the defence hypothesis. This toxicity has clearly been shown in herbivore growth or survival experiments using artificial diets, hyperaccumulating plants, or normal plants with high metal concentrations. However, a simple demonstration of toxicity or deterrence is not sufficient to infer that a trait functions primarily as a defence. For instance, high chloride concentrations in halophytes, though resulting in delayed insect larval growth, are not assumed to protect plants against herbivores.

Results/Conclusions

From our experience with the model Zn hyperaccumulator Thlaspi caerulescens (Brassicaceae), and in line with general theories of plant defence against herbivores, we here propose to discuss four main points that should still be investigated to prove that metals are involved in the defence of hyperaccumulating plants.

First, in contrast with studies on secondary plant compounds, very few field experiments were conducted with hyperaccumulators in natural conditions. The importance of such evidence will be discussed in relation with the influence of the environment on the metal accumulation phenotype. Secondly, due to the particular nature of “elemental” defences, plants with contrasting foliar metal concentrations can be obtained either in growing plants on low- and high-metal soil treatments, or in using plants with different genetic capacities of metal accumulation. Most published studies used treatment-induced variations of metal accumulation, and we will discuss the possible consequences of the “metal-treatment” effect. Thirdly, natural selection mainly acts on genetic variations existing within populations, and these variations must not be completely obscured by environmental effects (i.e. heritability must be significant). Such evidence available from herbivory studies with hyperaccumulators will be discussed. Finally, we will review the few studies that assessed the relative importance of secondary compounds vs metals in the defence of hyperaccumulators.