Plant diversity aboveground can exert top-down pressure on herbivores by attracting predatory insects, while organic soil amendments rich in beneficial microbes can limit herbivores from the bottom up by enhancing plant defensive chemistry. Aboveground and belowground forces always operate simultaneously to shape herbivore pressure, but understanding how they interact is a longstanding and persistent challenge. Here, we examine how organic composts mediate effects of plant diversity across trophic levels, using zucchini plants (Cucurbita pepo) as a study system. Over two field seasons, we manipulated vermicompost treatments in 18 experiments in school gardens that varied in surrounding plant and floral resource diversity, and measured responses of insect herbivores and their natural enemies.  Vermicompost strengthened a positive relationship between flower richness and foliar-feeding omnivores, suggesting that robust reservoirs of omnivores at flower-rich sites mounted stronger responses to compost-treated host plants. Predators increased with flower richness, but were not affected by vermicompost. Net outcomes of vermicompost and plant diversity were neutral for herbivores.

Synthesis and Applications: Altogether, our results reveal that bottom-up factors protecting plants are modified by their environmental context, and may more effectively attract natural enemies in landscapes with diverse floral resources. Therefore, we recommend augmentation of biodiversity aboveground (i.e. floral resources) together with biodiversity belowground (organic soil amendments) to strengthen crop protection.

Experimental treatments

In August 2021 and May 2022, experiments were established at nine school garden sites in Athens, GA, USA (Appendix Table S1). In 2022, three of the original school garden sites were dropped from the study due to teacher turnover and other factors beyond our control, and were replaced by three new sites. Therefore, experiments were repeated in six sites across both years, with a total of 12 unique sites across the two-year study. Within each year, all sites were separated by a minimum distance of 0.8 km (Appendix Figure S1).

Each site consisted of two adjacent raised garden beds (1.2 m × 2.4 m), with eight zucchini seedlings placed 0.6 m apart in two rows. One bed at each site was randomly selected for vermicompost treatments while the other served as a control. Four sites had pre-existing garden beds, while new beds were established at the remaining five sites. New beds were filled with organic raised bed planting mix (Kellogg Garden Organics, Carson, CA). In 2022, at sites where beds were already established, we homogenized the top 30 cm of soil in both beds by moving half the soil from one bed to the other with a tarp and mixing thoroughly. Dry organic poultry fertilizer (30 mL Symphony® 5-4-23, Seven Springs Farm Supply, per label instructions) was mixed into planting holes for both control and vermicompost beds, while an additional 120 mL vermicompost (castings produced by Eisenia fetida fed horse manure, juice pulp, and vegetable waste; Earthen Organics Worm Farm®, 1-0-0) was added to the holes in the beds assigned to receive vermicompost. Because vermicompost had a 1% nitrogen concentration, the amendment slightly augmented plant-available nitrogen relative to the control, but both control and vermicompost treatments included sufficient nitrogen for vigorous plant growth. Background nitrogen concentrations in soils are noted in supplemental data. The control and experimental beds received equal amounts of water every 24 h (each plant was watered for 10 seconds) until the plants established, after which they were watered every 48-72 h.

Insect sampling

Three weeks after plants established, weekly visual surveys of herbivores and natural enemies were initiated and continued for four weeks in 2021, and three weeks in 2022. Many plants in our 2022 study were killed by squash vine borers at four weeks of sampling (Melittia cucurbitae). Herbivores most common to our system included aphids, squash bugs (Anasa tristis), leaf/plant hoppers (Hemiptera: Cicadellidae & Flatidae), and fleahoppers (Hemiptera: Miridae). Omnivores preying on herbivores in our system included minute pirate bugs (Orius insidiosus) and big-eyed bugs (Geocoris spp.), and the most common predators were spiders (Araneae) and ladybeetles (Coleoptera: Coccinellidae). While spiders and ladybeetles occasionally feed on nectar and pollen, they were classified separately from ‘foliar-feeding omnivores’ because they do not feed on foliar tissue and are less responsive to changes in plant quality (Blubaugh 2023). On each plant, six leaves and petioles were carefully inspected on both sides (two new, two old, two intermediate), and all insects observed were counted, identified to family where possible on-site, and to order otherwise (for spiders). Insects in open flowers were recorded as well. Surveys terminated after three weeks in 2022 as nearly all plants from three sites were killed by squash vine borers. In 2021, four plants from each bed were randomly sampled on each sample date, and in 2022, all eight plants were sampled with the addition of more team members. Plants were excluded from surveys if they were severely damaged by deer browsing or by squash vine borers (damaged/dead plants from excluded from each site are noted in supplementary data).

The timing of our two experiments ensured wide coverage of the communities of pests that attack cucurbit crops. In 2021, we captured late-season pest communities, which were dominated by squash bugs and aphids. In 2022, experiments were performed early in the growing season, where the most common pests were cucumber beetles, squash vine borers and aphids.

Plant Diversity Survey

To quantify plant diversity in the surrounding area, we gridded a 450 m² area surrounding the raised beds, and recorded all crop and non-crop plant families observed. Our survey area was similar in scale to a similar study in urban agroecosystems (Philpott & Bichier 2017), and was sufficient to characterize most crop and non-crop plant diversity at all our sites. Schoolyard landscapes were primarily defined by turfgrass or pavement beyond the 12 m survey radius that typically included the entirety of all garden areas at each site. We estimated plant richness at the family level in both years of the experiment, and in 2022, we also counted inflorescences of all actively flowering plants, and estimated richness of flowering plants at the family level. Some garden sites included intentionally managed pollinator gardens, and others did not, creating a gradient in diversity of available flowers. We did not quantify floral traits or evaluate the quality of floral resources (i.e. nectar/pollen) offered by each plant.

Crop performance and leaf quality

To evaluate the effects of our treatments on the nutritional quality of foliar tissue, we collected the third unfurled leaf from the apical meristem from three randomly selected plants in each bed. Leaf tissue samples were processed at the Agricultural and Environmental Service Laboratory at the University of Georgia to determine percent total nitrogen concentrations with an Elementar Vario Max Total Combustion Analyzer (Langenselbold, Germany). We also recorded numbers of leaves, flowers and fruits (<30 cm long), to estimate effects of our treatments on plant performance and crop yield. Ripe fruits were harvested after visual surveys completed.