Geographical variation of floral scents in generalist entomophilous species with variable pollinator communities

Abstract

Abstract Floral scents are considered major attractants for pollinator species and scent composition is likely to be under pollinator‐mediated selection. Relatively, few studies have investigated geographical variation in floral scents in plant species with generalist pollination, and little is known about the mechanism behind this variation, often interpreted as the result of local adaptation to pollinators without considering phenotypic plasticity. One possible way to investigate this question is to compare floral bouquets both in natural populations and in populations reared under controlled conditions. We compared the identity and proportions of volatile organic compounds (VOCs), collected on plants both in the field and in the greenhouse, emitted by four generalist plant species found in six calcareous grasslands distributed along an environmental and diversity gradient. We recorded pollinator visits in the field to document geographical variation of pollinator communities associated with each plant species. We then analysed the relation between VOCs and pollinator variation using a redundancy analysis (RDA). We found that VOC composition differed among species and among populations, but also between field and greenhouse. The RDA showed a significant effect of the variation in pollinator communities associated with the variation in the VOCs emission, even after removing potential confounding effects, suggesting some stability in the compounds attractive to a given pollinator species. However, only one species showed significant geographical differences among populations tested in both conditions, suggesting local genetic adaptation. Another species showed geographical variation in VOC proportion in the field, but these results were not confirmed by the greenhouse extractions, suggesting an effect of phenotypic plasticity. Our findings suggest that geographical variation of floral scent may, in some cases, be genetically based and possibly explained by pollinator‐mediated selection, even in generalist pollinated species, which represent the large majority of pollination systems. However, our results also showed significant phenotypic plasticity, suggesting that studies of floral volatiles must be conducted in natural populations to obtain a representative picture of the signal that can be potentially perceived by pollinators. A free Plain Language Summary can be found within the Supporting Information of this article.