Aim: Understanding how climate conditions influence plant–pollinator interactions at the global scale is crucial to understand how pollinator communities and ecosystem function respond to environmental change. Here, we investigate whether climate drives differences in network roles of the main insect pollinator orders: Diptera, Coleoptera, Lepidoptera and Hymenoptera.

Location: Global.

Time period: 1968–2020.

Major taxa studied: Diptera, Coleoptera, Lepidoptera and Hymenoptera.

Methods: We collated plant–pollinator networks from 26 countries and territories across the five main Köppen–Geiger climate zones. In total, we compiled data from 101 networks that included >1500 plant species from 167 families and >2800 pollinator species from 163 families. We assessed differences in the composition of plant–pollinator interactions among climate zones using a permutational ANOVA. We calculated standard network metrics for pollinator taxonomic groups and used Bayesian generalized mixed models to test whether climate zone influenced the proportion of pollinator network links and the level of pollinator generalism.

Results: We found that climate is a strong driver of compositional dissimilarities between plant–pollinator interactions. Relative to other taxa, bees and flies made up the greatest proportion of network links across climate zones. When network size was accounted for, bees were the most generalist pollinator group in the tropics, whereas non-bee Hymenoptera were the most generalist in arid zones, and syrphid flies were the most generalist in polar networks.

Main conclusions: We provide empirical evidence at the global scale that climate strongly influences the roles of different pollinator taxa within networks. Importantly, non-bee taxa, particularly flies, play central network roles across most climate zones, despite often being overlooked in pollination research and conservation. Our results identify the need for greater understanding of how global environmental change affects plant–pollinator interactions.

Priority effects occur when the order of species arrival affects subsequent ecological processes. The order that pollinator species visit flowers may affect pollination through a priority effect, whereby the first visitor reduces or modifies the contribution of subsequent visits. We observed floral visitation to blueberry flowers from honeybees, stingless bees or a mixture of both species and investigated how (i) initial visits differed in duration to later visits; and (ii) how visit sequences from different pollinator taxa influenced fruit weight. Stingless bees visited blueberry flowers for significantly longer than honeybees and maintained their floral visit duration, irrespective of the number of preceding visits. In contrast, honeybee visit duration declined significantly with an increasing number of preceding visits. Fruit weight was positively associated with longer floral visit duration by honeybees but not from stingless bee or mixed species visitation. Fruit from mixed species visits were heavier overall than single species visits, because of a strong priority effect. An initial visit by a stingless bee fully pollinated the flower, limiting the pollination contribution of future visitors. However, after an initial honeybee visit, flowers were not fully pollinated and additional visitation had an additive effect upon fruit weight. Blueberries from flowers visited first by stingless bees were 60% heavier than those visited first by honeybees when total floral visitation was short (∼1 min). However, when total visitation time was long (∼ 8 min), blueberry fruit were 24% heavier when initial visits were from honeybees. Our findings highlight that the initial floral visit can have a disproportionate effect on pollination outcomes. Considering priority effects alongside traditional measures of pollinator effectiveness will provide a greater mechanistic understanding of how pollinator communities influence plant reproductive success.

For most food crops the identity and efficiency of pollinators across key growing regions remains a significant knowledge gap that needs to be addressed before we can develop crop-specific approaches for pollination service delivery. Here, we conducted a systematic literature review and meta-analysis on watermelon (Citrullus lanatus (Thunb. Matsum. & Nakai)), a globally important fruit crop, to identify the floral visitors and their efficiency across different growing regions. We found that 265 insect species visit watermelon flowers (including 5 orders, 18 families and 75 genera) across 17 countries and 6 continents. Bees and flies were the most abundant flower visitors overall, but show distinct regional differences. Honey bees were the majority visitor in 53% of growing regions (range: 0 - 94%), whilst wild bee species were more abundant in 42% of regions (range: 3.4 - 100%). Honey bees and other bees were equally effective at depositing pollen on stigmas, but varied in effectiveness for fruit set and seed set. Pollination data from global studies appear to be limited for the largest-scale watermelon producers, namely: China, Turkey, and India, with the majority (56%) of data available from North America. This synthesis identified four key themes for improving pollination in watermelon: increasing honey bee densities on crops where local polices and environmental conditions are suitable; introducing other managed pollinators; identifying key wild pollinator taxa to encourage within crops; and improving local and landscape management practices to support pollinators.

Improving the conservation and management of wild and managed insect pollinators is needed to increase the sustainability of global food security. In particular, refining crop-pollination management requires assessment of ecological processes occurring within pollinator communities and their interaction with plant traits. This thesis addresses three integral components of crop pollination: pollination in protected cropping environments, intra- and inter-specific processes central to plant-pollinator interactions in blueberry and the need for predictive models of pollinator body size to improve functional trait ~ pollination relationships.

I conducted a systematic review and meta-analysis of the pollinator efficiency of managed insects for food crops grown in protecting cropping environments. I identified the pollinator fauna in three blueberry species (Vaccinium spp.) and undertook pollination limitation and single and multiple visit fruit set and weight experiments to ascertain each blueberry species’ pollinator dependence and pollinator species’ efficiency. I examined the influence of priority effects in species visitation order (either honeybees:Apis mellifera or native stingless bee: Tetragonula carbonaria) on fruit weight outcomes in blueberry (V. corymbosum) and raspberry (Rubus idaeus). I then assessed the colony dynamics of wild T. carbonaria populations foraging on blueberry flowers in relation to their floral visitation rates. Finally, using a global dataset of bee and hoverfly morphological measurements, I built predictive models to estimate body weight in these two important pollinator groups.

Eusocial bees were the most commonly used managed pollinators in protected crops and both eusocial and solitary bees were similarly efficient in terms of fruit set, fruit weight and seed set. Significant issues facing pollination in protected cropping environments include environmental fluctuations and a general lack of knowledge in regards to pollinator health and longevity and stocking rates, and the efficiency and management of non-bee taxa. Australian grown blueberries are characterised by a pollinator community dominated by three bee species: Apis mellifera, Tetragonula carbonaria and Bombus terrestris. These pollinators were similar in their efficiency, yet their efficiency was dependent upon each blueberry species’ degree of pollen self-incompatibility. In mixed-species visitation sequences, species visit order priority effects were important in predicting fruit weight outcomes for blueberry but not raspberry. Blueberry flowers visited first by stingless bees resulted in larger fruit than those first visited by honeybees, however this effect diminished in longer floral visitation sequences. The number of wild T. carbonaria colonies found to forage in a given blueberry field ranged from 80 to more than 200 and floral visitation rates were correlated with intraspecific colony diversity. Models of pollinator body size as a function of the intertegular distance were highly predictive when taking into account species taxonomy, sexual dimorphism and biogeography.

Management of both wild and managed pollinators is highly context-specific and requires consideration of ecological processes at the individual (i.e., functional traits), population (i.e., intraspecific diversity) and community (i.e., interspecific diversity and interactions) level. Greater consideration of plant reproductive requirements is required within crop-complexes such as blueberry, which differ in their reliance upon cross-pollination for fruit production and improved yields. A more holistic understanding of ecological processes central to pollination services across all levels of biological organisation will improve pollination management practices and help ensure global food security.

Intra-seasonal variation in abiotic and biotic conditions can have profound consequences for pollinator community compositions and foraging movement, with flow-on effects upon pollination services. Yet, few studies have related such variations to pollination services in crop systems. In a cultivated highbush blueberry system with two primary pollinators — the managed European honey bee and a wild stingless bee species — we investigated how pollinator abundances, bee foraging behaviour, and con- and heterospecific stigmatic pollen loads changed over early, mid, and late blueberry blooming. Both con- and heterospecific stigmatic pollen loads declined following early bloom. This shift was associated with a decline in the abundance of stingless bees, whereas the abundance of honey bees only declined during late bloom. Simultaneously, honey bees were more likely to forage for blueberry pollen, and stigmatic pollen loads, relative to bee abundance, increased during late bloom. Although mixed pollen loads were common on pollinator bodies, especially on pollen foraging honey bees, heterospecific pollen deposition on blueberry stigmas was low. Given the similar effectiveness of honey bees and stingless bees as pollinators of blueberries, we contend that the observed seasonal variation in pollen deposition is likely caused by the decline in stingless bee abundances, as honey bees were not able to fully compensate for the loss of stingless bees during late bloom. Greater consideration of seasonal heterogeneity of pollinator abundance and behaviour, as part of pollination management plans, may aid in ensuring high pollination services throughout the entirety of crop bloom.