Shifts in host-parasitoid networks across community garden management and urban landscape gradients

Biological pest control relies on interactions between herbivores and their natural enemies. Maintaining this ecosystem service requires considering herbivore and natural enemy interactions and their response to anthropogenic change at multiple scales. In this study, we use ecological networks to quantify the network structure of interactions between herbivorous insects and their parasitoids. We examine how herbivore host abundance, parasitism rates, and shifts in network structure relate to changes in local habitat management and landscape context. We sampled herbivores and parasitoids in Brassica oleracea plants at 22 urban gardens on the California central coast. At each site, we measured local management characteristics (e.g. vegetation, ground cover, canopy cover) and quantified surrounding landscape composition (e.g. urban, natural, open, and agricultural cover). For the eight sites with large enough networks, we calculated three network structure metrics (interaction richness, vulnerability, and functional complementarity). We then used generalized linear and mixed models to examine relationships between herbivore host abundance, parasitism rates, garden management and landscape characteristics, and network metrics. We found that both local management and landscape composition influenced parasitism, while only local factors affected host abundance and network structure. Higher network interaction richness was marginally associated with enhanced parasitism rates for two host species and lower parasitism rates for one host species. Our results suggest that local garden management decisions may shift the structure of host-parasitoid networks, which may subsequently affect host parasitism rates, but outcomes for biological pest control will likely vary across host species.

Methods
The following datasets describe garden site characteristics, host insect abundance, host insect parasitism, and host-parasitoid network metrics sampled in urban community gardens on the California central coast in 2017 and 2018. We used generalized linear and mixed models to examine relationships between host abundance, host parasitism, garden site characteristics, and network metrics. We include a brief overview of the methods used to collect these datasets below. For more detailed methods, see the corresponding article describing this study published in Ecosphere. Details about each dataset are found in the README file.

Garden site characteristics: At each garden site, we sampled vegetation and ground cover each month within a 20 x 20 m plot at the center of each garden. We also measured the size of each garden site and determined the ecoregion each site was located within. Additionally, we characterized the landscape surrounding gardens using the 2016 National Land Cover Database (NCLD, 30 m resolution) to measure the percent land cover that was natural, open, urban, and agricultural within 1 and 2 km of garden sites. We also calculated landscape diversity using a modified Shannon diversity index.

Host insect abundance and parasitism: Host insects were sampled using visual surveys of Brassica oleracea plants targeting aphids in 2017 and non-aphid herbivores (cabbage whites, diamondback moths, beet armyworms, cabbage loopers, harlequin bugs, and whiteflies) in 2018. We identified all herbivore hosts to morphospecies and recorded their abundance. Parasitized aphids (mummies) and all non-aphid herbivores were then collected and reared under lab conditions for two weeks. We recorded the number of herbivores that remained healthy (not parasitized) and the number of herbivores from which parasitoids emerged (parasitized), and we used this as a basis for calculating parasitism. We calculated four different measures of parasitism for each of three focal host species (Aleyrodes proletella, Brevicoryne brassicae, and Pieres rapae): 1) presence or absence of parasitism per site (site-level parasitism), 2) proportion of sampled B. oleracea plants with parasitized hosts (plant-level parasitism), 3) presence or absence of parasitism per host individual (individual-level parasitism), and 4) proportion of parasitized hosts (population-level parasitism).

Host–parasitoid network metrics: We identified all eclosed parasitoids to morphospecies and noted the herbivore host species from which they emerged. We used these observed parasitism interactions to construct host-parasitoid networks for each site. We calculated four network metrics relevant to parasitism: interaction diversity (the number of unique interaction links in a network), vulnerability (the average number of parasitoid species interacting with each host species), functional complementarity (the level of resource partitioning among parasitoid species), and nestedness.

Datasets included:

1) HostAbundance_PopulationParasitism.csv

Garden site characteristics (garden size, ecoregion, vegetation cover, and ground cover in 2017, and 2018, and averaged across both years)
Surrounding landscape characteristics (percent urban and agricultural land cover within 1 km of garden sites)
Host abundance and population-level parasitism measures for three focal species
Host-parasitoid network metrics

2) SiteParasitism.csv

Garden site characteristics (garden size, ecoregion, vegetation cover, and ground cover in 2017, and 2018, and averaged across both years)
Surrounding landscape characteristics (percent urban and agricultural land cover within 1 km of garden sites)
Host abundance and site-level parasitism measures for three focal species
Host-parasitoid network metrics

3) PlantParasitism.csv

Garden site characteristics (garden size, ecoregion, vegetation cover, and ground cover in 2017, and 2018, and averaged across both years)
Surrounding landscape characteristics (percent urban and agricultural land cover within 1 km of garden sites)
Host abundance and plant-level parasitism measures for three focal species
Host-parasitoid network metrics

4) IndividualParasitism_ALPR.csv

Garden site characteristics (garden size, ecoregion, vegetation cover, and ground cover in 2018)
Surrounding landscape characteristics (percent urban and agricultural land cover within 1 km of garden sites)
Host abundance and individual-level parasitism measures for whiteflies (A. proletella)
Host-parasitoid network metrics

5) IndividualParasitism_BRBR.csv

Garden site characteristics (garden size, ecoregion, vegetation cover, and ground cover in 2017)
Surrounding landscape characteristics (percent urban and agricultural land cover within 1 km of garden sites)
Host abundance and individual-level parasitism measures for cabbage aphids (B. brassicae)
Host-parasitoid network metrics

6) IndividualParasitism_PIRA.csv

Garden site characteristics (garden size, ecoregion, vegetation cover, and ground cover in 2018)
Surrounding landscape characteristics (percent urban and agricultural land cover within 1 km of garden sites)
Host abundance and individual-level parasitism measures for cabbage whites (P. rapae)
Host-parasitoid network metrics

7) NetworkMetrics_ParasitismSummary.csv

Garden site characteristics (garden size, ecoregion, vegetation cover, and ground cover averaged across 2017 and 2018) for a subset of 8 sites with networks large enough to quantify network metrics
Surrounding landscape characteristics (percent urban and agricultural land cover within 1 km of garden sites)
Host abundance and all parasitism measures (site-level, plant-level, individual-level, population-level) for three focal species
Host-parasitoid network metrics