Insect pollinators of crops and wild plants are under threat globally and their decline or loss could have profound economic and environmental consequences. Here, we argue that multiple anthropogenic pressures – including land‐use intensification, climate change, and the spread of alien species and diseases – are primarily responsible for insect‐pollinator declines. We show that a complex interplay between pressures (eg lack of food sources, diseases, and pesticides) and biological processes (eg species dispersal and interactions) at a range of scales (from genes to ecosystems) underpins the general decline in insect‐pollinator populations. Interdisciplinary research on the nature and impacts of these interactions will be needed if human food security and ecosystem function are to be preserved. We highlight key areas that require research focus and outline some practical steps to alleviate the pressures on pollinators and the pollination services they deliver to wild and crop plants.

Bees Get That Caffeine "Buzz": Caffeine improves memory in humans, millions of whom find that their daily dose enhances clarity, focus, and alertness. The human relationship with caffeine is relatively recent, however, and thus its impact on our brains is likely a by-product of its true ecological role. Caffeine occurs naturally in the floral nectar of Coffea and Citrus plants. Wright et al. (p. 1202 ; see the Perspective by Chittka and Peng ) found that caffeine presented at naturalistic levels significantly improved the ability of bees to remember and locate a learned floral scent and potentiated the responses of neurons involved in olfactory learning and memory.

Floral chemistry mediates plant interactions with pollinators, pathogens, and herbivores, with major consequences for fitness of both plants and flower visitors. The outcome of such interactions often depends on compound dose and chemical context. However, chemical diversity and intraspecific variation of nectar and pollen secondary chemistry are known for very few species, precluding general statements about their composition. We analyzed methanol extracts of flowers, nectar, and pollen from 31 cultivated and wild plant species, including multiple sites and cultivars, by liquid‐chromatography–mass‐spectrometry. To depict the chemical niche of each tissue type, we analyzed differences in nectar and pollen chemical richness, absolute and proportional concentrations, and intraspecific variability. We hypothesized that pollen would have higher concentrations and more compounds than nectar, consistent with Optimal Defense Theory and pollen's importance as a male gamete. To investigate chemical correlations across and within tissues, which could reflect physiological constraints, we quantified chemical overlap between conspecific nectar and pollen, and phenotypic integration of individual compounds within tissue types. Nectar and pollen were chemically differentiated both across and within species. Of 102 compounds identified, most occurred in only one species. Machine‐learning algorithms assigned samples to the correct species and tissue type with 98.6% accuracy. Consistent with our hypothesis, pollen had 23.8‐ to 235‐fold higher secondary chemical concentrations and 63% higher chemical richness than nectar. The most common secondary compound classes were flavonoids, alkaloids, terpenoids, and phenolics (primarily phenylpropanoids including chlorogenic acid). The most common specific compound types were quercetin and kaempferol glycosides, known to mediate biotic and abiotic effects. Pollens were distinguished from nectar by high concentrations of hydroxycinnamoyl‐spermidine conjugates, which affect plant development, abiotic stress tolerance, and herbivore resistance. Although chemistry was qualitatively consistent within species and tissue types, concentrations varied across cultivars and sites, which could influence pollination, herbivory, and disease in wild and agricultural plants. Analyses of multivariate trait space showed greater overlap across sites and cultivars in nectar than pollen chemistry; this overlap reflected greater within‐site and within‐cultivar variability of nectar. Our analyses suggest different ecological roles of nectar and pollen mediated by chemical concentration, composition, and variability.

In the fight against arthropod crop pests using plant secondary metabolites, most research has focussed on the identification of bioactive molecules. Several hundred candidate plant species and compounds are now known to have pesticidal properties against a range of arthropod pest species. Despite this growing body of research, few natural products are commercialized for pest management whilst on-farm use of existing botanically-based pesticides remains a small, but growing, component of crop protection practice. Uptake of natural pesticides is at least partly constrained by limited data on the trade-offs of their use on farm. The research presented here assessed the potential trade-offs of using pesticidal plant extracts on legume crop yields and the regulating ecosystem services of natural pests enemies. The application of six established pesticidal plants (Bidens pilosa, Lantana camara, Lippia javanica, Tephrosia vogelii, Tithonia diversifolia, and Vernonia amygdalina) were compared to positive and negative controls for their impact on yields of bean (Phaseolus vulgaris), cowpea (Vigna unguiculata), and pigeon pea (Cajanus cajan) crops and the abundance of key indicator pest and predatory arthropod species. Analysis of field trials showed that pesticidal plant treatments often resulted in crop yields that were comparable to the use of a synthetic pesticide (lambda-cyhalothrin). The best-performing plant species were T. vogelii, T. diversifolia, and L. javanica. The abundance of pests was very low when using the synthetic pesticide, whilst the plant extracts generally had a higher number of pests than the synthetic but lower numbers than observed on the negative controls. Beneficial arthropod numbers were low with synthetic treated crops, whereas the pesticidal plant treatments appeared to have little effect on beneficials when compared to the negative controls. The outcomes of this research suggest that using extracts of pesticidal plants to control pests can be as effective as synthetic insecticides in terms of crop yields while tritrophic effects were reduced, conserving the non-target arthropods that provide important ecosystem services such as pollination and pest regulation. Thus managing crop pests using plant secondary metabolites can be more easily integrated in to agro-ecologically sustainable crop production systems.