Existing literature on the effects of urbanization on species occurrence, abundance, and diversity often relies on urban-rural gradient studies . These studies generally find that increased urbanization decreases the diversity of organisms . Confirming these findings are an abundance of patch-matrix literature suggesting that the quality of the habitat patch itself, its size, and the composition of the matrix surrounding it are determining factors for species occurrence in fragmented landscapes . Specific to UA, higher imperviousness surrounding urban farms has been related to decreased parasitoid abundance and richness , decreased predator abundance and richness , and even decreased predation on sentinel prey . To better understand PH richness and abundance in urban farms and associated biological control services, we conducted an in-situ survey at urban community farms in the East Bay of the San Francisco Bay Area, USA. Eleven farms participated in 2018 and ten farms in 2019. Farms were asked to participate in research based on two factors: 1. farm size, to ensure a comparative sample of small, medium, and large farms, and 2. high or low levels of surrounding impervious surface per the National Landscape Cover Database . Landscape factors and APM practices of farms were measured. APM practices included area of non-crop usage , area of production, crop plant abundance , crop richness, floral richness,plastic plant pot sizes and percent of farm surface with complex ground covers including mulch and leaf litter. Landscape factors included percent of impervious surface at 200-, 500-, and 1000-meter radii. Sampling iterations occurred from May to mid-October each year. On-farm non-crop area was defined as a not actively managed area of the farm occupied by non-crop flora. Farm size in m2 was calculated through Google Earth Pro and ground-proofed during on-farm spatial measurements. Brassica abundance was determined by counting all brassicas on the farm when sampling occurred.
Crop plant richness was determined by eight meter transects measured perpendicular to garden beds three times during the growing season. Different cultivars of the same species were counted separately when measuring crop richness. Floral richness was surveyed three times per growing season by completing a comprehensive count of each flowering plant at each survey site. Randomized 4m2 quadrats were used to estimate percent of and type of cover . Ground cover quadrats were measured across crop and non-crop areas. Percent of surrounding impervious surface for each farm was measured using the NLCD at 8m resolution .Collection of PH was accomplished by using an insect vacuum on Brassica oleracea cultivars, including broccoli, kale, collards, and tree collards. Each sampled plant was randomly selected and was only sampled if it was standing free of other herbaceous cover and flowering plants. A total of nine plants of each cultivar present were sampled per visit. Vacuum sampling occurred monthly from May to October. Vacuuming of each plant lasted for five seconds. For this work, we assume that sampled wasps were performing foraging or host-seeking behaviors on selected plants . Each sample was frozen until processed by extracting all PH and identifying them to the lowest taxonomic level possible per previous literature . PH identification was accomplished using Hymenoptera of the World . Chalcidoidea were identified with the Annotated keys to the Genera of Nearctic Chalcidoidea , and Braconidae using the Manual of the New World Genera of the Family Braconidae .Cabbage aphids, Brevicoryne brassicae were visually identified and abundance was assessed by doing a total count on three random leaves on nine brassicas per cultivar, including counts of apterous, alate, and parasitized aphids. Aphid abundance counts were performed monthly from May to October on non-vacuum sampling days to reduce PH disturbance. Parasitism rates were calculated as number of parasitized aphids divided by number of total aphids on each leaf.
Generalized linear mixed models were constructed using the MASS R package to explore whether APM practices or landscape factors affected PH abundance on common brassicas. Each response variable: All PH, PH super family, family, and subfamily abundance, overall site PH diversity, and rates of aphid parasitism were modeled with both local and landscape factors. Local factors include the percent of mulch ground cover, floral and crop richness, production, and non-crop area. Landscape factors include percent impervious surface at 200, 500, and 1000m radii, and farm size. Seasonal factors included both year and season and were assessed as categorical variables: early-season , mid-season , and late-season . The fitdistrplus package in R was used to find appropriate distributions for modeling . A negative binomial or Poisson distribution with a log link function was selected as appropriate given the zeroinflation of the count data. Models were fitted with the glmer.nb or glmer function in R package MASS . Preliminary models with all measured local and landscape factors were constructed for each response variable. Explanatory variables of low importance for all response variables were excluded from subsequent models. Final models were assessed for fit using the Akaike Information Criterion and diagnosed for over or under-dispersion by comparing observed residuals with expected residuals using the DHARMa package in R. Poorly fitted models were excluded from the results . Partial regression plots for final models were developed using the “effects” package in R and are reported in Results . The slope of the line in these plots represents the association between a single explanatory variable and a response variable accounting for the effects of each other variable within the fitted model.To test the local and landscape effects on the enemies hypothesis vis-a-vis APM on populations of PH in urban agroecosystems, we collected data from twelve urban farms in the San Francisco Bay Area over a period of two growing seasons. Participating farms were selected to represent a continuum of size, spatial composition, and surrounding imperviousness.
Non-crop area was a significant predictor for all PH, cynipoid, and braconid wasps. Effects of APM practices were varied, but increased crop richness and mulch coverage were associated with increased abundance of all Chalcidoidea, including the Aphelinidae. Increases in crop richness also showed an increase in parasitism rates of aphids on brassica crop plants. Unexpectedly, Floral richness showed a negative relationship to the abundance of all PH, as well as chalcids, and all Braconidae. All PH showed a significant decline in abundance during the late season of 2019. All measures of impervious surface surrounding urban farms had no effect on PH abundance or aphid parasitism on the urban farms. Landscape effects to arthropod mediated ES continue to have mixed results and this research supports previous findings in urban agriculture which show both negative and positive effects to natural enemy abundance and diversity . Non-crop areas identified in this research are difficult to identify explicitly as either managed or unmanaged and existed on a spectrum that was often difficult to quantify in interviews or through survey work. However, these areas most frequently had been improved with flowering perennials or annuals, medicinal or “native” flora, and farmers typically stated the purpose as providing a resource for wildlife or beneficial insects. Previous research supports farmer efforts. Structural diversity has been found to elicit positive responses with regard to diversity and abundance of predators and PH in previous UA studies . These areas may provide critical over-wintering habitat in annual cropping systems,blueberry plant container additional hosts or prey, shelter, floral nectar resources for nectarivorous insects . Our findings suggest that these non-crop areas have the potential to influence agroecosystem function in UA, and may be an important part of APM practices, even in highly fragmented landscapes. Moreover, floral richness had little effect on PH abundance, or parasitism of aphids, signaling that increase in PH abundance were not due to floral nectar within these non-crop areas. Another mechanism that may be of importance are the spatial composition of the agroecosystem. Our research did not take into account the overall distribution of non-crop area within the farm, which may have failed to account for spatial heterogeneity that has been found to illicit positive and negative biological control responses in agroecosystems . Future research on urban farms should account not only for the proportion of non-crop areas, but also spatial heterogeneity to further explore these effects. Overall, APM practices, such as increased mulch coverage and crop plant richness were important predictors of PH abundance, and increased aphid parasitism rates. The connection between mulch, complex ground covers, and increased abundance and diversity of parasitic wasps has been previously observed in urban agroecosystems , a variety of natural habitats, and rural agroecosystems . It is unlikely that mulch would provide a direct resource for PH, but PH may benefit from mulch as a potential overwintering habitat or it may provide habitat for potential hosts. Many of the collected PH were parasitoids of dipteran larvae; these larvae are herbivorous but complete part of their life cycle in soils. I suggest that the overall biodiversity of urban farms with increased mulch coverage may create a bottom-up trophic cascade that increases overall soil arthropod diversity benefiting PH populations. Floral richness had a negative effect on PH abundance in all models. Floral richness was chosen as an explanatory variable as it has previously been found to increase PH abundance in UA . The vast majority of PH are nectarivorous, and this additional nectar resource has been suggested frequently as a strategy for increasing populations, potentially leading to increased parasitism . However, conflicting data raises questions about this on farm manipulation and whether PH seek hosts in the same area they feed, or they disperse to increase fecundity . A large proportion of our overall sample of PH were cynipoids, potentially from the genus Alloxysta, known hyperparasitoids of both dominant primary aphid parasitoids in our sample, Aphidiinae, and Aphelinidae . These reductions in primary aphid parasitoid populations may be due to direct or indirect negative effects from this hyperparasitoid that also feeds on floral nectar . In urban agroecosystems, floral provisioning as a habitat manipulation may be complicated by the inherent fragmentation and quality of the urban matrix.
For floral resources to be an effective APM practice, this resource must be limited. Potential concentrations of alternate off-farm floral resources may complicate this affect. While this research expanded upon previous findings and can be of utility for urban agroecosystem management, many questions remain. Firstly, the effects of hyperparasitism on biological control in UA. Our third most collected taxon was Cynipoidea, many of which are often hyperparasitoids of aphid parasitizing wasps . Given that these cynipoids were collected from plant foliage in close proximity to many primary aphid parasitoids, there is some anecdotal evidence that these cynipoids were engaging in host-seeking behavior. If some of the measured on-farm management practices, such as increased non-crop areas also increase abundance of Cynipoidea, this could result in decreased biological control services. In this case, floral provisioning may potentially be acting as an ecosystem disservice . Unfortunately, we were unable to collect parasitized aphids and rear any hyperparasitoids during this research, but these findings suggest that hyperparasitism in fragmented UA landscapes may be a mechanism affecting APM strategies in UA. Crop plant richness positively affected the abundance of all Chalcidoidea and the subfamily Aphelinidae. Crop richness was also a predictor of greater parasitism rates of cabbage aphids on sampled brassica. Similar findings in rural and urban agroecosystems, including increased PH abundance and biological control services in relation to increased crop diversity have been previously documented . Given that intercropping is commonly practiced in UA, these results validate the efficacy of the practice, and offer an opportunity to investigate the extent of the effect in future research efforts. 4.3 Seasonal factors Seasonal effects on PH abundance were mixed, but many affects were measured in the second year of our sampling. Of note, in 2019, we had fewer sampling events as one farm was unable to participate in our study, but more PH were collected in that year despite the smaller sampling pool. Rates of aphid parasitism were significantly decreased between mid- and late season in 2019. It is unknown what drove these effects, but notable that such a significant difference could occur between sampling seasons. Future research efforts should consider seasonal differences and weather when drawing conclusions about on-farm or landscape factors to PH abundance or diversity or associated biological control services.