Reduced chemical defence in ant-plants? A critical re-evaluation of a widely accepted hypothesis

Since its original formulation by Janzen in 1966, the hypothesis that obligate ant-plants (myrmecophytes) defended effectively against herbivores by resident mutualistic ants have reduced their direct, chemical defence has been widely adopted. We tested this hypothesis by quantifying three classes of phenolic compounds (hydrolysable tannins, flavonoids, and condensed tannins) spectrophotometrically in the foliage of 20 ant-plant and non-ant-plant species of the three unrelated genera Leonardoxa, Macaranga and Acacia (and three other closely related Mimosoideae from the genera Leucaena, Mimosa and Prosopis). We further determined biological activities of leaf extracts of the mimosoid species against fungal spore germination (as measure of pathogen resistance), seed germination (as measure of allelopathic activity), and caterpillar growth (as measure of anti-herbivore defence). Condensed tannin content in three of four populations of the non-myrmecophytic Leonardoxa was significantly higher than in populations of the myrmecophyte. In contrast, we observed no consistent differences between ant-plants and non-antplants in the Mimosoideae and in the genus Macaranga, though contents of phenolic compounds varied strongly among different species in each of these two plant groups. Similarly, among the investigated Mimosoideae, biological activity against spore or seed germination and caterpillar growth varied considerably but showed no clear relation with the existence of an obligate mutualism with ants. Our results did not support the hypothesis of ‘trade-offs’ between indirect, biotic and direct, chemical defence in ant-plants. A critical re-evaluation of the published data suggests that support for this hypothesis is more tenuous than is usually believed. The general and well-established phenomenon that myrmecophytes are subject to severe attack by herbivores when deprived of their ants still lacks an explanation. It remains to be studied whether the trade-off hypothesis holds true only for specific compounds (such as chitinases and amides whose cost may be the direct negative effects on plants’ ant mutualists), or whether the pattern of dramatically reduced direct defence of ant-plants is caused by classes of defensive compounds not yet studied.