Discussion

Urbanization is a prominent form of land use change, but we do not have a clear picture of how urban development influences interspecific interactions, including those that occur among plants and pollinators. This understanding is important not only for wild plant populations but also for the pollination services that generate food production for humans (Klein et al. 2007, Ollerton 2017). Across a network of urban farms spanning a gradient of urbanization in the city of Chicago, we found that pollen limitation increases with increasing urbanization for one of our two study species. In S. lycoperscium , we found that both ecologically and agriculturally important measures of pollination success were consistently limited by pollination services. We observed pollen limitation across all sites, but critically, the magnitude of pollen limitation was greater at more urban sites with greater surrounding impervious surface. In contrast, for C. pepo , we found little evidence of pollen limitation, although seed set was pollen-limited at our most urban site. Consistent with our hypothesis, impervious surface increased the magnitude of pollen limitation of fruit set, however, this was a marginal and weak effect that we did not find for seed set. Our results suggest that the increase in impervious surface associated with urbanization can have negative consequences for the pollination and reproduction of plants, but the extent to which this occurs should depend on the plant species and its pollination ecology.
We tested for pollen limitation in two species, providing insight to how susceptible a specialized versus generalized plant species may be to increased pollen limitation in response to urbanization. Solanum lycopersicum , which relies on a subset of bees with specialized foraging behavior (buzz pollination), exhibited nearly ubiquitous pollen limitation that was greater in magnitude at sites in more heavily urbanized areas. Cucurbita pepo , which can be pollinated by the majority of bees as well as other insect pollinators, displayed much more variable fruit and seed production and little evidence of pollen limitation. Our findings are consistent with broad trends between pollen limitation and land use change, where plants in urban habitats are generally more pollen-limited than those in natural or managed landscapes, and those that are specialized on a particular pollinator group may be especially vulnerable (Bennett et al. 2020). However, urbanization likely has varying effects between species and study systems, and these patterns in pollen limitation have only been observed when broadly comparing urban to non-urban plants between different sites (Bennett et al. 2020), as opposed to the same species in urban vs. non-urban sites or across an urbanization gradient. Studies that investigated the effects of urbanization at the species level have focused on reproduction in wild plant species and thus far reported mixed results. Species have exhibited similar reproduction between urban and rural sites (Verboven et al. 2012, Barker and Sargent 2020), greater pollen limitation in urban sites (Hou et al. 2019, Carper et al. 2022), or less pollen limitation in urban sites, although this was in an invasive species (Parker 1997). While our results are consistent with the idea that specialized plant species are more susceptible to pollen limitation in urban habitats, we only investigated one species each for generalized and specialized pollination systems. Research incorporating multiple specialized and generalized species within the same system is needed to understand whether and how this pattern persists more broadly.
Patterns in pollination services should also be assessed across a gradient of urbanization rather than placing sites into discrete urban or non-urban categories. Even within our heavily urbanized sites (impervious surface > 60%), we observed substantial variation in pollen limitation and visitation, suggesting that responses to urbanization are complex and may depend not only on whether a habitat is urban, but the degree of urbanization as well as other factors beyond impervious surface. Local habitat characteristics such as the amount of available green space, gardening practices used in surrounding areas, and floral and nesting resources from home gardens and other plantings are likely to vary considerably across the urban landscape and could play an important role in shaping pollinator communities. Additionally, environmental factors such as heat, light, and soil moisture can affect tomato reproduction by limiting pollen production, pollen viability, and fruit production (Dingley et al. 2022). Because we were working within a set of established and functioning urban farms in different areas of Chicago, we could not control for small-scale variation in conditions such as heat and sunlight, or neighborhood influences like the surrounding floral landscape and nearby pesticide use. Studies that have examined pollinator communities across urbanization gradients have been able to identify local habitat factors that may interact with or offset effects of landscape-level urbanization alone, such as floral resources, pollinator foraging regime, habitat patch size, and season (Williams and Winfree 2013, Choate et al. 2018, Rivkin et al. 2020, Birdshire et al. 2020, Gruver and CaraDonna 2021). Indeed, in both species, reproductive success was variable across sites in the supplemental pollination treatment where plants should be receiving sufficient pollen, and was particularly low at a site with an intermediate level of impervious surface (66.7%), providing evidence that other factors we did not examine are affecting reproduction. Our results highlight the need for more studies addressing the impacts of urbanization along urbanization gradients rather than landscape categories, and incorporating local habitat characteristics, to tease apart the relationships among urbanization, pollination services, and plant reproduction, and the mechanisms driving them (Wenzel et al. 2020).
If crops are pollen-limited in response to urbanization as we have shown for sun gold tomatoes, then urban farms, especially those in heavily urbanized areas, are likely underproducing. Understanding the drivers of pollination services and pollen limitation in urban areas is necessary to improve crop production at urban farms and support the reproduction of wild and managed plant populations in or near urban habitats. Pollinator visitation rates to S. lycopersicum flowers decreased with impervious surface, suggesting that reduced visitation rates are part of the explanation for the increase in pollen limitation at more urban sites. Pollen limitation is a consequence of insufficient pollen receipt, which could be the result of a deficit in either the quantity or quality of pollen received. Pollen limitation could increase with urbanization as a result of low pollinator visitation, as decreases in pollinator abundance and richness have been observed in response to urbanization across a range of systems and taxa (Ahrné et al. 2009, Geslin et al. 2013, Fortel et al. 2014, Choate et al. 2018, Cardoso and Gonçalves 2018, Pereira et al. 2020). These declines are likely due to a combination of factors such as habitat fragmentation and loss of floral or nesting resources and could substantially reduce visitation to plants in these areas. Declining pollinator populations in response to urbanization might explain the contrasting patterns we found between our focal species. The lower visitation rates and greater pollen limitation seen in S. lycopersicum , which relies on bees capable of sonication, may reflect a reduced abundance of these pollinators.Cucurbita pepo , which can be pollinated by many bees as well as other insect pollinators, may have a greater buffer against declining pollinator populations as it can utilize a much broader range of visitors.
On the other hand, the pollen limitation and visitation patterns we observed may be related to more nuanced changes in plant-pollinator interactions rather than overall pollinator abundance. Specifically, changes in pollinator community composition or foraging behavior could also be limiting plant reproduction by decreasing the quality of pollination services provided to plants. There is support for the idea that urban and suburban areas can serve as refuges to pollinators, with neutral or even positive impacts of urbanization on pollinator groups like bees (Carper et al. 2014, Hall et al. 2017, Theodorou et al. 2017), especially when compared to primarily agricultural landscapes (Wenzel et al. 2020). In Chicago specifically, studies have found diverse wild bee communities that support pollination services to plants, where bee richness was most related to floral diversity rather than impervious surface (Lowenstein et al. 2015, Gruver and CaraDonna 2021). However, the pollinator community differed in that highly urban areas housed more non-native bee species and fewer native species than less urban areas (Gruver and CaraDonna 2020, 2021).
The increase in pollen limitation and lower visitation rates that we observed in S. lycopersicum in response to urbanization may therefore reflect shifts in pollinator community composition. Pollination of S. lycopersicum was carried out primarily by solitary bees we could not identify in the field, followed by bumblebees (Bombus spp.). Bombus visitation was not related to impervious surface, suggesting that their abundance or pollination behavior is not strongly driving pollen limitation, despite being one of the most studied and commonly used group of pollinators for supplemental pollination in S. lycopersicum (Cooley and Vallejo-Marín 2021). Visitation by other, smaller bees decreased marginally with urbanization. The decreased pollinator visitation and greater pollen limitation we found in S. lycopersicum at more urban sites are therefore likely driven by these other bee species. Although their buzz pollination behavior and efficiency is less studied, smaller and often solitary bees such as Lasioglossum , Melissodes , andAugochlora spp., are capable of buzz pollination and were frequently observed sonicating our plants and more broadly across Chicago (Cardinal et al. 2018, Rosi-Denadai et al. 2020, Gruver and CaraDonna 2021). In an area where bee abundance was greater in urban than non-urban sites, pollen limitation was still greater in urban areas for two of the three species studied, and this was linked to lower receipt of conspecific pollen for plants in the urban sites (Carper et al. 2014, 2022). These results coincide with patterns of bees in urban landscapes visiting a greater number of plant species overall, but a lower proportion of available plant species, likely due to greater plant richness (Baldock et al. 2015, Theodorou et al. 2017). Pollinator visit and search durations have also been shown to increase in urban areas, which could decrease overall visitation rates (Andrieu et al. 2009). Pollination services may also be changing in response to less studied abiotic characteristics of urbanization such as pesticide use, the urban heat island effect, or anthropogenic air pollution (Wenzel et al. 2020, Bloom et al. 2021, Descamps et al. 2021), which has been found to reduce floral visits by pollinators by as much as 83–90% (Ryalls et al. 2022).
Taken together, our results suggest that pollination deficits can increase with urbanization, and there are likely multiple interacting factors underlying this relationship. Understanding the drivers behind decreased pollination services in urban areas can help us develop best practices for improving pollination services in the future. We demonstrate here that between two plant species, the more specialized was more pollen-limited and received lower pollinator visitation in response to increasing urbanization when compared to the generalized species. Strategies for encouraging pollinator visitation to crops to increase yield could focus on providing habitat and planting resources that support a range of pollinators, particularly those that visit both specialist and generalist plant species, although the utility of these strategies needs to be empirically tested with data. Further research that considers multiple species across urbanization gradients and longer time scales, with additional local and landscape level factors, is necessary to understand the mechanisms behind the observed pollination deficits that increase with urbanization.