Still, we do not have any replication for the different levels of connectance and so it is impossible to draw any conclusions regarding this structural property. We expect that, following many plant-pollinator network and food web extinction simulations, higher connectance would lead to greater robustness . However, the addition of negative interactions could complicate the interactions, making highly connected interaction networks vulnerable if positive interactions were lost, leaving only negative interactors . Future work exploring the role of both nestedness and connectance in networks that incorporate a range of negative interactions could help improve understanding of the extent to which these commonly used structural metrics predict robustness in networks.In all three networks, removing pollinator species from generalist to specialist first has a larger impact on the robustness of the network than with the order of specialist to generalist removals. This pattern has been noted in previous studies though ours is the first study to examine the role of extinction order after incorporating negative interactions in the networks. Memmott et al. noted from their study that while robustness was impacted when species were removed from generalist to specialist, the effect was not as dramatic as expected or as reported in food web studies when the most linked interactors are removed . In our case, adding negative interactions in to the network made the network less robust, blueberry pot size resulting in patterns more like those of food web studies where removal of the most linked species causes a collapse to low richness .
Future studies should focus on exploring the impact of higher proportions of negative as well as networks of varying properties to determine how and why networks are most vulnerable and to understand if order of extinction matters only in specific cases. Extinction cascade length, or, the number of additional species losses triggered by initial pollinator extinctions is a property unique to simulations that incorporate negative interactions in to the network. As far as we know this is the first study to incorporate negative interactions in to plant pollinator networks in an effort to assess network robustness to pollinator extinctions. This makes it impossible to compare the role of extinction cascades on network robustness to other studies examining the same question. Still, our results indicate that extinction cascades play an important role in our simulations but that networks behave idiosyncratically. Understanding if variation in total cascade length is biologically meaningful and whether total cascade length helps us understand how vulnerable a network is to extinctions will require further exploration. The focus should be to pinpoint properties that make networks more vulnerable to cascades and the point at which networks are most vulnerable . Our exploration into the role of negative interactions in extinction simulations lead us to question the expectation that plant-pollinators interactions will be robust to extinctions. If by chance we lose the “good” pollinators from communities leaving less effective or detrimental pollinators then we may see a series of additional extinctions or population declines. These results certainly warrant future explorations in to the role of variation in interaction quality and the potential to impact communities through extinction cascades.
While not exhaustive, this study is an important exercise in assessing how one key assumption in plant-pollinator simulation models may overestimate network robustness. There are other studies that are making an effort to incorporate more biologically realistic interactions into network studies , though not through the addition of negative interactions. Vieira and Neto expand on traditional extinction models by allowing for variation in how much one species relies on the other and variation in the assumption species can persist even if only a minor, weakly interacting partner is present. They find that the relaxing the first assumption leads to a decrease in robustness and the second assumption leads to an increase in robustness. Moving beyond the expectation that all interactions between plants and their pollinators are strictly positive will help us to explore not only the robustness of networks to extinctions, but also has the potential to help us further understand how and why mutualisms are maintained when faced with exploitation . Campbell et al. is the only other study to incorporate negative interactions into network simulations. They do so by identifying interactions as either mutually beneficial or beneficial for one species and detrimental to the other. While they do not directly assess how these antagonistic interactions impact network robustness when faced with extinctions, they do reveal the “critical species” in their simulations—those species that cause significant community collapse when removed—are species that tend to have asymmetric interaction direction. Their results suggest that when these critical species are lost, the network is left with an abundance of negative interactions in their absence, which can lead to further extinctions and often collapse. An important next step in assessing the robustness of plant-pollinator networks to extinctions would be to build networks that incorporatevariation in negative interactions and complete similar extinction simulations.
We expect that a more continuous interaction space will lead to similar results that we found in this study, with just a slower rate in loss of robustness due to variation in interaction strength . In recent years there has been a substantial effort to incorporate realistic dynamics in to ecological network models. These efforts have revealed that the incorporation of community dynamics such as population sizes and interaction strengths and topological dynamics such as re-wiring can offer a more realistic assessment of the robustness of networks faced with extinctions. While the inclusion of population and topological dynamics were beyond the scope of this study, we acknowledge that such dynamics could potentially add to our understanding of how negative interactions impact network robustness. For example, incorporating migration into our models could make the networks more robust to extinctions if the species moving in were positive interactors. However, one could imagine a scenario where a less effective pollinator such as a honey bee moves in after an extinction event of a more effective pollinator . In this scenario, a plant species would be rescued from the loss of interaction but could experience reproductive loss with the new, less effective, pollinator. Valdovinos et al find that allowing for adaptive foraging , leads to enhanced network robustness against species extinctions. This is due to an increase in the floral resources extracted by the specialist pollinators as well as an increase in the rates of visitation to specialist plants in the network. Allowing for adaptive foraging in our simulations would likely make the networks more robust to extinctions , although, this robustness would likely vary depending on extinction order . Finally, allowing for re-wiring to a new interaction partner after experiencing either extinction or fluctuations in partner availability has been shown to lead to more robust networks . Still, if plants re-wire to a less effective pollinator they may suffer reproductive consequences. Future studies exploring such dynamics could substantially add to our understanding of how incorporating negative interactions in to plant pollinator network models impact network robustness when faced with species loss. Habitat loss and degradation are leading to global pollinator losses . Given the central importance of pollination in food production and the maintenance of biodiversity and ecosystem function, it is imperative that we come to a predictive understanding of how pollinator losses will affect plant-pollinator systems. Evidence suggests that losing even a few pollinators can have a strong negative effect on the plants that rely on pollination for reproduction . In the absence of empirical studies, network-based simulations left researchers hopeful that plant communities would be robust to pollinator extinctions Our study is a step in showing that these simulation models may overestimate network robustness by assuming that all interactions in the networks are positive. By incorporating more realistic representations of the interactions that take place in a plant-pollinator community, plant raspberry in container then we are more likely to identify properties of networks that determine robustness to extinctions, helping direct conservation efforts.
Future studies that improve on predictive models that will allow us to anticipate likely changes in pollination services, and help us designing strategies to maximize ecosystem resilience will be essential to preserving pollination services.Global production of pollinator dependent crops has increased by 300% in the past 50 years . At the same time, managed honey bee populations are declining due to a complex of factors including novel diseases, pesticides and habitat change . Pollinator deficiencies may precipitate significant yield reductions and increased food prices, ultimately jeopardizing food security . Unmanaged bees are highly effective pollinators of a variety of crops and act as insurance against loss of pollination function due to honey bee deficits . While proximity to natural habitat increases populations of such alternate pollinators , intensive agricultural landscapes often contain little remnant habitat. As a result, re-diversification of agricultural areas has been proposed as a means of bolstering pollination services from these alternate pollinators . Diversification of agricultural landscapes can take place at many scales, including within fields , along field edges , or bordering landscape features . One benefit of field edge techniques is that they create habitat without sacrificing arable land , and comprise a large portion of non-cropped area in farming regions globally . Farm bill conservation programs in the United States and agri-environmental schemes in the European Union prioritize on-farm habitat creation projects that target pollinators, providing incentives through cost-share programs . Despite the prominence of these programs, there is little information as to the effectiveness of field-margin diversification techniques, and specifically, whether they can bolster pollinator services and affect yields to the same levels documented in patches of natural habitats while simultaneously conserving pollinator species . One common field edge diversification technique, hedgerow restoration , has been found to increase pollinator richness within field edges and up to 100 m into nearby crop fields . Additionally, hedgerows show potential for increasing pollination function within adjacent fields. Using sentinel canola plants, Morandin, Long and Kremen found that wild bees enhanced seed set, once the contribution from managed honey bees was accounted for. However, the canola plants provided a highly attractive resource within an unattractive crop matrix of processing tomato, which provides few nectar rewards and requires buzz-pollination to release pollen stores. These conditions are not reflective of the field conditions created by monoculture plantings of pollinator-dependent crops, which generate hundreds of thousands of synchronous, though short-lived, blooms within a single field . Pulses of highly attractive floral resources can create dilution effects, drawing species away from adjacent seminatural habitat and reducing pollination services there . Yet in spite of the attractiveness of MFC fields, wild bee abundance and richness has been found to be higher in habitats, including hedgerows, in closer proximity to MFC fields . The effects of MFCs may be species-specific, with some exhibiting higher preference for MFCs over other resources . Specialist pollinators, such as the squash bee , seek out fields of their host plant, cultivated squash, in the landscape . While the influence of MFCs on pollinator populations and services has been well-studied, whether the presence of field-scale restorations can augment pollinator populations and pollination services within MFC fields remains an open question . We examine the ability of hedgerows to enhance pollination services in a simplified agricultural landscape when adjacent to a mass-flowering, pollinator-dependent crop, cultivated sunflower . We ask whether the identity of the pollinator species found within hedgerows during the crop bloom period is the same as those found within adjacent sunflower fields. Then, using an independent data set, we determine whether the most abundant wild sunflower visitors, sunflower specialist bees, also utilize hedgerow plantings in our study landscape. We also determine whether hedgerow presence affects wild bee abundance and richness in sunflower fields, and if this, in turn, translates into increased sunflower seed set.Field sites were located in Yolo County, an intensively-farmed agricultural region of California’s Central Valley that contains a mixture of conventionally managed row and orchard crops. The majority of natural and semi-natural habitat in the county is concentrated around the borders of agricultural lands and not embedded within them . We sampled 18 sunflower fields between June and July . Half of the fields were adjacent to bare or weedy edges , and half were adjacent to hedgerows . Sites were paired based on the timing of the sunflower bloom, the sunflower variety , and landscape context.