<p>Anthocyanin pigments are synthesised in the leaves of many plants, however the adaptive significance of these pigments is not entirely understood. It has been postulated that their red colours may function as visual signals through coevolution between herbivorous insects and their host tree species, though the hypothesis lacks solid empirical evidence. I investigated the leaf signalling hypothesis using Pseudowintera colorata, focusing on five areas: 1) I exploited the natural polymorphism in leaf colour of P. colorata to test the predictions that (i) bright leaf colour is a reliable signal of a plant’s defensive commitment; (ii) insects in the field avoid trees that are brightly coloured; and (iii) the trees with the brightest leaves will have higher fitness. Relative to green leaves, redder foliage contained higher concentrations of polygodial, a sesquiterpene dialdehyde known to have strong antifeedant properties, and incurred less insect feeding damage. Redder trees hosted fewer Ctenopseustis spp. leafroller larvae than neighbouring matched green trees. Contrary to the predictions of the leaf signalling hypothesis, there was no difference in any of the measured fitness parameters between red and green trees, indicating that the leaf colour polymorphism in P. colorata is stable. 2) Many insects are sensitive to volatile organic compounds (VOCs), however the role of VOCs in plant-herbivore signalling has not been investigated. I analysed VOCs released from undamaged, herbivore- and mechanically-damaged red and green leaves of P. colorata, and the olfactory preferences of brownheaded leafroller (C. obliquana) larvae. While the VOC profiles of browsed and unbrowsed leaves were statistically distinguishable, the VOC profiles released from intact, herbivore-, and mechanically-damaged P. colorata leaves did not reliably identify leaf colour. Moreover, naïve and experienced C. obliquana larvae displayed no preference for the volatiles from mechanically-damaged red or green leaves. Therefore, I concluded that VOC compounds are not likely to play a large role in mediating insect herbivore-plant interactions in P. colorata. 3) Studies of leaf signalling rarely consider the influence of the light-absorbing properties of non-green pigments upon photosynthesis. I compared the photosynthetic and photoinhibitory responses of red and green leaves from matched, neighbouring pairs of P. colorata of contrasting colour. Redder P. colorata leaves in the field had a lower maximum photosynthetic assimilation rate than matched green leaves from neighbouring trees. However, I was unable to detect any measurable advantage in terms of photoprotection in the red P. colorata leaves as indicated by chlorophyll fluorescence profiles. My results indicate that the presence of anthocyanin pigments within non-senescing leaves may impose a slight photosynthetic cost to the plant. 4) I used literature searches, field surveys and laboratory bioassays to identify which invertebrate herbivores are most likely to participate in leaf-signalling interactions with P. colorata. Feeding preference bioassays showed that brownheaded leafrollers (C. obliquana and C. herana) and Auckland tree weta (Hemideina thoracica) preferentially consumed leaf material from green than red P. colorata leaves. Results from these bioassays, combined with my field surveys suggest that Ctenopseustis spp. leafroller larvae are the most likely coevolution partners for P. colorata. 5) There is a well-established link between nitrogen deficiency and leaf reddening. Additionally, leaf nutrients can influence foraging behaviour and performance of insect herbivores. I measured N and C contents of leaves from neighbouring matched pairs of red and green P. colorata. There were no significant differences in the amounts of, or ratio between, N and C between matched red and green leaves. This result indicates that differences in colour and herbivory among P. colorata leaves are not attributable to differences in leaf nutrients. Taken together, my results suggest that foliar anthocyanins in P. colorata do function as visual signals, however their effect on herbivory is small. Additionally, interindividual variation in non-senescing leaf colour in P. colorata may be stable due to a trade off between signalling and photosynthesis. Discussions of leaf signalling need to follow the examples of other fields studying the interactions between plants and insects and move from overly simple models to those that incorporate more of the complexity that is observed in the natural world.</p>