Ecosystem services are essential for human well-being, especially in urban areas where 60% of the global population will live by 2030. While urban habitats have the potential to support biodiversity and ecosystem services, few studies have quantified the impact of local and landscape management across a diverse suite of services. We leverage five years of data (>5,000 observations) across a network of urban gardens to determine the drivers of biodiversity and ecosystem service trade-offs and synergies. We found multiple synergies and few trade-offs, contrasting previous assumptions that food production is at odds with biodiversity. Furthermore, we show that landscape-level natural habitat cover interacts with local management to mediate services provided by mobile animals, specifically pest control and pollination. By quantifying the factors that support a diverse suite of ecosystem services, we highlight the critical role of garden management and urban planning for optimizing biodiversity and human benefit.

Study system

Between 2013 and 2017, we collected data three to five times from May to September within 28 urban community gardens on the central coast of California, USA (see Egerer, Arel, et al., 2017; Philpott & Bichier, 2017)(Fig S1). During each garden visit, we counted the number and species of trees and shrubs within a 20 m x 20 m plot and measured herbaceous plant species richness, flower abundance, and ground cover composition (percent bare soil, rocks, grass, mulch) within four (2013–2015) or eight (2016–2017) randomly placed 1m x 1m quadrats in the plot (see Philpott & Bichier, 2017).

Landscape composition

We quantified landscape composition within 2 km of each garden using data from the 2011 National Land Cover Database (NLCD, 30 m resolution)(Homer et al., 2015). The 2-km buffer size is inclusive of typical foraging distances for our focal animal taxa (arthropods and birds) and the time period is closest to the average sampling year (as in Cohen et al., 2021; Philpott & Bichier, 2017). We calculated the percent of natural landscape cover (comprised of deciduous forest, evergreen forest, mixed forest, shrub/scrub, and grassland/herbaceous), which ranged between 0–61.2%, with spatial statistics tools (ArcGIS v. 10.1)(ESRI, Redlands, California, USA).

Ecosystem Services

In the same gardens, we measured seven ESs: pest control, pollination, climate regulation, carbon sequestration, water storage, food production, and human well-being (Table S1). We evaluated four pest control metrics (ES-Pest-control) – three measuring prey removal (see Philpott & Bichier, 2017) and one measuring parasitism (see Egerer, Liere, et al., 2018). We evaluated six pollination metrics (ES-Pollination) – two measuring open pollination success (see Cohen et al., 2021), three measuring fruit set (the proportion of flowers developing into fruits), and one measuring conspecific pollen deposition. We evaluated one climate mitigation metric (ES-Climate) – variation in daytime garden temperatures (see Lin et al., 2018), one carbon sequestration metric (ES-Carbon) - soil organic matter (see Egerer, Ossola, et al., 2018), and one water conservation metric (ES-Water) - soil water holding capacity (see Egerer, Ossola, et al., 2018). We included five food production metrics (ES-Food) – four measuring weight or volume of focal crops (see Cohen et al., 2021) and one assessing gardener-reported food production. We evaluated four human well-being metrics (ES-Well-being) that describe gardener perceptions of well-being derived from gardener survey data (see Egerer, Philpott, et al., 2018)(see Supporting Information and Table S1 for details).

Mobile Ecosystem Service Providers

We evaluated biodiversity levels for putative ecosystem service providers (ESPs), specifically mobile animals that a) consume or have non-consumptive negative effects on herbivores (ESP-Natural-enemy) or b) visit flowers to collect pollen and nectar (ESP-Pollinator). We evaluated 14 metrics for natural enemies sampled with a variety of traps (pitfalls, sticky traps), visual surveys, and point counts (Table S1). We quantified abundance and species richness for ants (Hymenoptera: Formicidae), carabids (Coleoptera: Carabidae), spiders (Arachnida: Araneae) (see Egerer, Arel, et al., 2017; Otoshi et al., 2015; Philpott et al., 2019), ladybeetles (Coleoptera: Coccinellidae)(see Egerer, Bichier, et al., 2017), and insectivore birds (Animalia: Aves)(see Mayorga et al., 2020). We quantified abundance and family richness for parasitoids (Hymenoptera: Parasitica)(see Burks & Philpott, 2017) and all arthropods (see Philpott et al., 2020). We evaluated 4 metrics for pollinators sampled with netting, pan-trapping, and non-lethal observation (Table S1). We quantified abundance and richness of bees (Hymenoptera: Apidae) (see Plascencia & Philpott, 2017; Quistberg et al., 2016) and pollinators (Hymenoptera, Diptera, Coleoptera, and Lepidoptera)(see Cohen et al., 2021). For all 18 metrics, we extrapolated richness using the Chao1 index (Chao, 1987), which accounts for uneven sampling and undersampling, using the vegan package in R (Oksanen, 2013).

The data can be opened by any software capable of opening spreadsheets.