Local and Landscape Impacts on Ecological Networks and Ecosystem Services

Local and landscape features of urban agroecosystems also affect interactions such as herbivory, predation and parasitism, and pollination. In this section, we outline interactions between plants and arthropods with a specific focus on pest control and pollination and also examine local and landscape drivers of ecological networks, given their importance in predicting urban agroecosystem function. Networks depict ecological interactions between species across adjacent trophic levels (Bascompte et al. 2003, Ings et al. 2009) and network metrics used to quantify interactions can predict how ecosystem services respond to perturbations (Bluthgen 2010). Ecological networks can increase our understanding of how management and community characteristics relate to ecosystem services (Perrings et al. 2010, Bohan et al. 2013), but to our knowledge, only two studies have explicitly linked changes in natural enemy- herbivore or plant-pollinator networks to ecosystem services in urban agroecosystems (Geslin et al. 2013, Theodorou et al. 2017, discussed below). Analysis of networks in agroecosystems is thus an important analytical tool (Woodward and Bohan 2013, Tylianakis and Binzer 2014) but underexplored in the urban agroecosystem context. Trait-function relationships may strongly impact the outcome of ecosystem services (Poisot et al. 2013, Peralta et al. 2014) but these relationships are poorly understood, especially for urban systems (Lin et al. 2015).

Natural Enemy and Herbivore Interactions and Networks

Differences in local and landscape features of urban agroecosystems mediate ecological interactions and likely impact ecological network structure with critical implications for biological pest control. In urban agroecosystems, local-scale factors such as floral richness enhance natural enemies and pest suppression services (Bennett and Gratton 2012, Gardiner et al. 2014, Burks and Philpott 2017). However, predator exclusion experiments in urban agroecosystems suggest that there may be discrepancies in factors that promote different natural enemy groups at different scales. For example, pest control levels increase with local vegetation complexity but not with landscape diversity (Philpott and Bichier 2017). Further, management factors and their influence may differ regionally. Prey removal rates in predator exclusion experiments were generally higher in California, USA than in Chiapas, Mexico, and the factors that influenced the removal rates of different kinds of prey (aphids, moth eggs, and moth larvae) at times deviated between the regions (Morales et al. 2018). These differences may be due to differences in regional filters or differences in the particular urban landscape context (e.g.. Aronson et al. 2016). For instance, shifts along an urban to rural gradient can impact natural enemy-herbivore interactions in urban green spaces, and the relative strength of bottom-up and top-down regulatory forces. In the Central Arizona Phoenix area, bird exclusion led to higher abundances of insect herbivores in urban areas but not rural desert areas, and birds in urban areas experienced lower predation risk thereby exerting higher top-down control of arthropods (Faeth et al. 2005). Turrini et al. (2016) ran predator exclusion experiments with syrphid larvae, vetch aphids, and bean plants and found differences between urban and rural areas. In particular, predator control of aphids was stronger in rural sites, whereas aphid impacts on plant biomass were stronger in urban sites. Thus, urbanization negatively affected both available plant biomass and top-down control of aphids.

Although several studies examine how ecological networks involving natural enemies and herbivores shift with rural agroecosystem management, this is relatively unexplored in urban agroecosystems. Past research on antagonistic networks of herbivores and their natural enemies has largely focused on understanding network structure and network dynamics (Pascual and Dunne 2006), with less work on investigating responses to global change forces as has been done for mutualistic networks (see Pollinator and Plant Interactions and Networks below). Specifically, recent reviews of antagonistic networks indicate that antagonistic network structure is often modular, with distinct modules, or compartments, of interacting species making up the larger network. Interestingly, studies of antagonistic network dynamics have indicated that antagonistic interactions are quite variable across time points and that environmental factors and landscape gradients can drive high rates of interaction turnover (Tylianakis and Morris 2017). Host-parasitoid networks (for bee and wasp parasitoids) in tropical agricultural landscapes strongly respond to shifts in land use, even though species richness and evenness do not (Tylianakis et al. 2007). Contrary to expectations, given similar species richness in modified and unmodified habitats, food web interaction evenness declined, the ratio of parasitoids to hosts increased, and the specialization of the most abundant parasitoid increased in more simple agroecosystems compared with complex agroecosystems in the landscape (Tylianakis et al. 2007). Along a regional gradient of land-use intensity, tropical host-parasitoid networks are more homogenous in their interaction composition in more simplified, deforested habitat types compared to forested habitat types (Laliberte and Tylianakis 2010). However, changes in network structure associated with land-use may not always have obvious effects on ecosystem services. In aphid-parasitoid-hyperparasitoid networks, agricultural intensification increased food web complexity and temporal variability, while also decreasing parasitism rates (an ecosystem service), and increasing hyperparasitism rates (an ecosystem disservice) (Gagic et al. 2012). Studies on natural enemy-herbivore networks in urban agroecosystems (Philpott et al. 2020) reveal that interaction richness (assumed to boost pest control) increases with floral richness and declines with landscape- level agricultural cover whereas trophic complementarity (assumed to negatively impact pest control) increases with agricultural cover. But a more detailed examination of the specific impacts of shifts in ecological networks on pest suppression is warranted. By examining more about the details of networks, the number of links between natural enemy and herbivore taxa, and shifts in shared herbivores for different natural enemies, we may learn more about why loss of certain species along local or landscape management gradients may disproportionately affect pest control function within the urban agroecosystem.

Pollinator and Plant Interactions and Networks

Plant-pollinator network observation has long existed (e.g., Darwin 1862), but more rigorous quantification and utilization of networks has recently bloomed, especially across natural and rural agricultural landscapes (reviewed in Bascompte and Jordano 2007, Heleno et al. 2014). Within these landscapes, pollinator network ecology has focused on three main themes: 1) network structure. 2) network dynamics, and 3) network response to global change (Valdovinos 2019). Past studies on network structure reveal that most mutualistic networks tend to have a similar distribution or number of links (Jordano et al. 2003), that networks are often nested and asymmetric, where interactions involving specialists are subsets of interactions involving generalists (Bascompte et al. 2003, Vazquez et al. 2007), and that networks are frequently modular with certain groups of interacting species (Vazquez et al. 2007. Valdovinos 2019). In terms of network dynamics, plant-pollinator networks are highly plastic (CaraDonna et al. 2017, Ponisio et al. 2017), with frequent partner switching (Kaiser-Bunbury et al. 2010). Studies on network response to global change is grow ing and often focuses on simulating species loss by removing pollinator or plant species (Memmott et al. 2004) or simulating climate change (Memmott et al. 2007), habitat loss (Fortuna and Bascompte 2006), or species invasion (Valdovinos 2019), and then measuring impacts on network metrics. These reviews broadly reveal that highly nested networks are more resilient to extinction events and that specialized pollinators (with narrow diet breadths) are particularly prone to extinction under global change scenarios (Valdovinos 2019), but relatively few studies have examined whether these general patterns exist in urban systems.

One study conducted across urban green spaces in Bangkok, Thailand found that the most common urban pollinators were bees and that urban plant-pollinator networks exhibited high pollinator gen- eralism (Stewart et al. 2018). Moreover, the most abundant pollinators were also the most generalist; the authors propose this feature as particularly beneficial for persistence in an urban landscape where the floral environment is highly managed and frequently altered. In non-garden urban green spaces,

Mukherjee et al. (2018) and Maruyama et al. (2019) found similar patterns of higher generalization in more urbanized areas for butterfly-plant networks in Kolkata, India and for hummingbird-plant networks in southeastern Brazil, respectively; however, Mukherjee et al. (2018) also found greater nestedness with increasing urbanization levels. Similarly, a study conducted in the UK found that urban sites had lower overall network specialization but that pollinators exhibited both greater generality (foraging on a greater number of plant species) and also greater specialization (visited a lower proportion of available plant species) compared to pollinators in agricultural and natural sites (Baldock et al. 2015).

A handful of studies have explicitly examined plant-pollinator interactions within urban agricultural sites, and these studies have found contrasting patterns with respect to levels of robustness and specialization. In a massive study comparing pollinator communities and plant-pollinator networks across a wide range of urban land use types. Baldock and colleagues found that residential and allotment gardens served as pollinator network ‘hotspots’ via high pollinator diversity and dispersal into other urban sites, which increased city-scale network robustness (resilience to secondary extinctions following an initial loss of species) (Baldock et al. 2019). In other words, urban gardens may play a particularly critical role in the persistence of plant-pollinator networks across distinct urban land uses. Other urban garden network studies have focused primarily on patterns of specialization within urban gardens themselves (not at city-scales) and have found distinct patterns. For example, in an urban garden system in Chicago, USA, Lowenstein and colleagues found that garden pollinators were quite generalist, with a substantial fraction of the plant community experiencing limited pollinator visitation and a smaller subset of plants experiencing very high levels of visitation (Lowenstein et al. 2019). In contrast, in a study comparing urban residential gardens with semi-natural areas in Montreal, Canada, Martins (2017) found that gardens supported more specialized bees that visited a smaller fraction of the plant community compared with semi-natural sites. Overall, these studies suggest that urban gardens play a critical role in supporting city-level plant-pollinator robustness; but that within the focal urban garden systems itself, it is possible that both specialized and generalized pollinator groups dominate, depending on the local, landscape, and biogeographic context. More work examining network dynamics or potential response of networks to global change explicitly within urban systems is needed.

While the utilization of plant-pollinator network studies in urban areas has clearly been growing, networks are rarely linked to pollination function (Kaiser-Bunbury and Bluthgen 2015). To our knowledge, only two studies have investigated the explicit links between plant-pollinator networks and pollination services in urban systems. The first study evaluated ‘open’ and ‘tubular’ experimental flowers in different landscapes, and found that the number of interactions performed by flower-visitors decreased in urban habitat compared to semi-natural and agricultural habitat, and specifically the number of Syrphidae and solitary bees visits were lower in urban habitat, leading to lower reproductive success of the ‘open’ flower functional group (Geslin et al. 2013). In the second study connecting urban plant-pollinator networks with pollination functions, the authors found that flower visitation metrics (e.g.. linkage density) increased with urban land cover, flower visitors were more generalized in urban compared to agricultural areas, and bees were more specialized in urban areas; despite these patterns, no network metrics were predictive of plant reproductive success (Theodorou et al. 2017). Instead, urban cover, bee richness, and flying insect abundance better predicted plant reproductive success (Theodorou et al. 2017). Overall, these studies indicate an urgent need to understand the impacts of plant-pollinator networks on plant reproduction and to determine the importance of network metrics relative to other biotic and abiotic factors within urban systems.

< Prev   CONTENTS   Source   Next >