Plant toxin levels in nectar vary spatially across native and introduced populations

Secondary compounds in nectar can function as toxic chemical defences against floral antagonists, but may also mediate plant-pollinator interactions. Despite their ecological importance, few studies have investigated patterns of spatial variation in toxic nectar compounds in plant species, and none outside their native range. Grayanotoxin I (GTX I) occurs in nectar of invasive Rhododendron ponticum where it is toxic to honeybees and some solitary bee species. We examined (i) geographic variation in the composition of nectar GTX I, as well as GTX III (which is not toxic to these species), in the native and introduced range of R.ponticum, (ii) how their expression is structured at patch and landscape scales within ranges, and (iii) whether climatic and environmental factors underpin spatial patterns. While both GTXs varied within ranges, variation in GTX I, but not GTX III, was detected between ranges. GTX I expression was thus markedly lower or (in 18% of cases) absent from nectar in introduced plants. Spatial autocorrelation was apparent at both patch and landscape scales and in part related to heat load interception by plants (a function of latitude, aspect and slope). As expression of nectar GTXs was generally robust to environmental variation, and aggregated in space, this trait has the potential to be spatially discriminated by consumers. Given the specificity of change to GTX I, and its differential toxicity to some bee species, we conclude that its expression was likely to have been influenced during invasion by interaction with herbivores/consumers, either via pollinator-mediated selection or enemy release from floral antagonists.Synthesis. As the first demonstration of large-scale geographic variation and spatial structure in toxic nectar compounds, this work deepens our understanding of the chemical ecology of floral interactions in native and introduced species. Spatially explicit studies of nectar secondary compounds are thus required to show how the extent and structure of spatial variation may affect floral ecology. Future development of invasion theory should incorporate a holistic view of plant defence, beyond antagonistic interactions, which integrates the consequences of chemically defended mutualist rewards.