Additionally, a few recent studies have shown that pesticide application can have negative impacts on native bee species abundance and richness , and declines occur at both local and landscape scales and are more likely to occur after multiple applications . However, very little is currently known about the complexities of pesticide impacts on wild pollinators and further research is required to understand the influence of specific pesticide application techniques and chemicals on pollinator life history and behavior in the field.Our understanding of the effects of climate on pollination services is weak, in part because determining individual plant and pollinator responses to climate is a vast task, and scaling those responses up to the community level is challenging, given the complexity of such interactions. Pollinator visitation to plants is a community-level phenomenond merely knowing the species present does not necessarily indicate how they will interact to provide pollination services. In particular,behavioral changes in pollinator foraging may play an important role in determining the effects of climate on pollination services. Climate may cause changes in the quality, quantity, and timing of floral rewards in space, and pollinator behavioral responses to those changes are difficult to predict. Furthermore, plastic flower bucket pollinator behavior is layered on top of changes in pollinator timing and abundance in space.

We may begin to piece together potential effects of climate on pollination by considering evidence for each of these pieces separately. To determine the sensitivity and vulnerability of pollination services to climate, we will first review what is known about how plants and pollinators themselves are being affected. and Hegland et al. . Although the phenologies of some insect pollinators and many plant species have been shown to be affected by warming, few studies have found temporal mismatches in plant– pollinator interactions. Honey bees, bumble bees, and some butterflies are active earlier with warmer temperatures and many plant species are flowering earlier with warmer temperatures, as well . The majority of reported phenological changes for plants and pollinators show advanced activity, yet some show delayed phenology , illustrating the need for detailed, and possibly species specific, investigations of different systems before generalizations can be made. In addition, variability in early-season phenologies of plants, and to a lesser degree, butterflies, is increasing . When flowering phenologies were experimentally advanced and plants were placed in an unmanipulated community , few mismatches were observed . By resampling and comparing plant–pollinator phenologies and interactions with historical phenologies and interactions, it is possible to test the degree to which interactions are being disrupted by phenological changes. In a comparison spanning 120 years, few mismatches in plant–pollinator interactions directly attributable to phenological change were found .

These empirical investigations, though limited in number, suggest that temporal mismatches may not play as large a role in decoupling plant–pollinator interactions as previously indicated by models . Despite the lack of evidence for widespread temporal mismatches, given the current degree of changes in climatic conditions in some regions, there are likely some cases in which mismatches have led to declines or extirpations of plant and/or pollinator populations. Demographic consequences of such temporal mismatches, however, are not well understood. Thomson showed strong direct effects of frost events and snowmelt patterns on the reproduction of glacier lilies. Temporal mismatches of glacier lily flowering and the activity of their pollinators may also limit seed set, though this has not been investigated explicitly. Burkle et al. have observed the extirpation of half of the bee species historically present in mid-western US forests, though the cause of this bee decline, involving many specialist bees, may not be solely due to temporal mismatches . Effects of climate on pollinator nesting habits and reproduction are largely unknown and have been identified as an important gap in our understanding . Miller-Rushing et al. outlined a way to better understand the demographic effects of phenological mismatches, including key questions and approaches using experiments, models, and long-term data.In addition to temporal mismatches influencing pollination services due to phenological changes in plant and pollinator activity, spatial mismatches may also be as, or more, important.

Spatial mismatches between plants and pollinators may occur due to range shifts resulting from climate changes, or due to indirect climate change effects such as habitat destruction and fragmentation. For example, under different climate scenarios, it is possible that humans may use land differently, by altering existing urbanization, development, and/or crop planting patterns, and thus potentially destroying and fragmenting existing habitat. Second, climate can determine the distribution of many plant species, and thus, climate changes can result in spatial shifts in plant populations. Much of the empirical support for range shifts has been performed along elevational gradients in the alpine, where the upward movement of species with warming is relatively straightforward to detect. Plant distributions along elevational gradients can change rapidly with climate, and have been documented as shifting upward in elevation an average of 22 m per decade in the southwestern United States and 29 m per decade in western Europe . Other studies focused specifically on flowering forbs and their range shifts find similar changes in elevation . In a cross-study comparison, Parmesan and Yohe found many species, including alpine herbs, to be shifting poleward at an average rate of 6.1 km per decade. In some systems, involving narrowly distributed endemic plants, range shifts may not be possible and declines in species richness have been observed . At the other end of the spectrum, some plant species, such as invaders, may adapt and evolve to occupy new environmental conditions associated with climate change, expanding into novel ranges . However, there is also evidence for lack of plant response to recent warming in the alpine, with clones of long-lived plants remaining in the same locations over thousands of years . Sherrer and Korner argue that high microclimatic variation in the alpine would allow plants to ‘escape’ larger scale changes in climatic conditions without moving more than a few meters. Determining which plant species would shift their ranges due to warming and the direction and magnitude of their responses remains challenging. Few studies have documented range shifts in pollinator species, primarily because knowledge of historic and current distributions are lacking. There are initiatives to map historic pollinator distributions based on museum specimens for comparison to current ranges . In a few studies, butterfly range shifts have been documented to be moving poleward due to climate changes . Changes in pollinator species distributions are more easily attributed to habitat loss or fragmentation directly , though the ultimate causes of habitat change may be related to climate shifts and associated changes in human land use. At a local scale of kilometers, loss of historically occurring plant–pollinator interactions can be due to spatial uncoupling in which plants become separated from their pollinating partners in isolated forest fragments. A relatively new method for understanding plant and pollinator interactions at the community level is via the construction of interaction networks . In these networks, single lines connecting plants and pollinators indicate the existence of interactions between two species,makingit easier to visualize and analyze changes in these interactions. Examined as a whole, the pollinator network can also convey critical information regarding the structure and function of community-level interactions . Pollinator network analysis has generally shown that, within sites, flower buckets wholesale there is substantial interannual variation in what pollinators interact with what plants, but despite this variation, pollination events still take place, potentially indicating resilience to environmental change.

Interestingly, despite natural variation in who interacts with whom, structural properties of community plant– pollinator networks remain fairly constant over time . For this reason, it is possible to use network metrics, like nestedness and connectance to indicate, over and above this natural variation, the health and stability of interactions and potential directional change over time due to anthropogenic environmental changes. Highly nested networks are fairly robust to environmental change ; thus declines in the nestedness of a network over time can serve as a barometer, indicating loss of functional resilience in plant–pollinator interactions. Pollination networks can be utilized to understand community-level impacts of climate change via two main approaches: evaluation of networks after simulated biodiversity loss or phenological mismatch and comparison of networks after actual habitat restoration. We discuss both in the sections below, respectively.Simulation studies use existing plant–pollinator networks to simulate the effects of environmental change, for example, they help us to understand what happens when we delete species or advance the phenologies of species, as might be predicted by climate warming. For the most part, these simulation studies have found that pollination networks are more robust to the removal of random species than to the selective removal of highly linked species . Many studies have found that the removal of the most-linked pollinator species results in a relatively linear decline of plant species, explained by the nested interaction network topology and a redundancy in pollinator links per plant . In another simulation study, Devoto and colleagues removed species from the network that they predicted would exhibit spatial range shifts due to climate-change induced increased precipitation, and found that removal of these species had fairly minimal impacts on the plant–pollinator network. In another simulation study by Kaiser-Bunbury et al. , the authors allowed species in the network to switch to those that they have known potential to interact with, with the interesting finding that networks could be stabilized following species loss if remaining species could indeed ‘rewire’ to form new interactions. While results from these simulation studies suggest that pollinator networks may be more robust to species loss and range shift than other ecological networks , it is important to keep in mind that plants and animals may be differentially impacted by biodiversity loss, given that pollinators represent higher trophic levels, which are inherently more sensitive to habitat disturbance . Because pollinators are dependent on plants for food, it is not surprising that removal of the strongest interacting species in the network can have the strongest negative impact on simulated animal extinction rates . Just a few pollinator network studies have examined the impact of climate on pollinator networks by modeling plant and pollinator phenology . In these studies, Memmott and colleagues use existing predictions of temperature changes to calculate potential phenological shifts in plant and pollinator emergence and senescence, if those predictions are accurate. In their 2007 study, Memmott and colleagues examined simulated networks where the onset of flowering and flight activity of all plants and pollinators would be advanced by 1, 2, and 3 weeks. Across all three scenarios, they found that 17–50% of all pollinator species, especially those with smaller and more specialized diet breadths, faced reduced floral resources and potential extinction. Similar to the findings of other network studies, the predicted impacts on pollinators were much greater than for plants, though plants still experienced a 50% reduction in pollination activity, likely leading to reduced reproduction and eventual population decline . In a following study, Memmott and colleagues asked specifically how the sowing of wildflowers changes the impact of simulated climate on the pollinator network. Their experimental planting and simulated climate-impacted network showed that by sowing plant species that bloomed at the beginning and end of the flowering season, the overall period of nectar resource availability could be extended for pollinators. However, it should be noted that these simulation studies did not allow for behavioral flexibility, which might allow for new interactions to develop between plants and pollinators within the system. Another potential interaction between climate and pollination systems may be the introduction of non-native species into the pollination network , which may occur if nonnative species become invasive under those environmental change conditions . Depending on whether non-native plants interact negatively or positively with native plants, their invasion could stabilize or destabilize pollinator networks. Examination of a plant and pollinator network in central United States has shown that non-native plants are far more common than non-native pollinators . In this study, non-native plants interacted with significantly fewer pollinator species than native plants, but were still relatively well connected in the network. Likewise, in another simulation study specifically examining the removal of non-native species, the pollination network’s structural integrity was diminished by non-native species loss, due to the high levels of connectivity between alien and native species within the pollinator network . Thus, if non-native plant species increase overall floral resource availability and duration, they may positively impact pollinator populations in the short term.