Urban agriculture in walkable neighborhoods bore fruit for health and food system resilience during the COVID-19 pandemic

More than half of the global population lives in urban areas, elevating urban food issues to a critical challenge. Urban agriculture is gaining attention for its multi-dimensional benefits to sustainability and resilience, including food security, climate risk mitigation, biodiversity, social capital, and health and well-being. In parallel, walkable neighborhood concepts (e.g., the 15-minute city) emphasize access to amenities within walking distance to reduce emissions and promote health. The COVID-19 pandemic highlighted vulnerabilities in food systems and intensified interest in both urban agriculture and neighborhood-scale accessibility. Yet, research often focuses on retailer access (food deserts) and provides limited quantitative evidence on local food production, food system resilience, and differential health effects of various green spaces. This study addresses these gaps by empirically testing the associations between access to local food through urban agriculture and subjective well-being, physical activity, and food security concerns during COVID-19, using Tokyo as a case where intraurban agriculture and walkable access to fresh food are common. Specific objectives were: (1) to examine associations of three local food access types (allotment farms, home gardens, farm stands) with subjective well-being, physical activity, and food security concerns during the state of emergency and in the future; (2) to compare local food access with other urban green spaces and with other food purchasing sites; and (3) to identify characteristics of people accessing local food during the pandemic, focusing on work style and income.

Prior studies document urban agriculture’s contributions to sustainability and resilience, including reduced carbon footprints and potential shifts to greener diets, but integration with neighborhood-level access has been underexplored. Food desert literature emphasizes retail access, especially in low-income areas, often neglecting local production and offering limited quantitative evidence. Emerging work highlights food system resilience as the capacity to adapt to disruptions; during COVID-19, many argued for urban agriculture’s role, though empirical investigations of resilience outcomes were scarce. Studies in North America examined gardener characteristics and gardening’s perceived values. Urban green space research links nature exposure to cognitive, mental, and physical health and well-being, but few studies differentiate health impacts among types of green spaces. This study builds on these threads by quantitatively comparing urban agriculture modalities to parks/greenways and retailers on health and food security outcomes.

Design and setting: Cross-sectional online survey of adults (≥20 years) residing in Tokyo, Japan, conducted June 4–8, 2020, just after the first national state of emergency. Sample and recruitment: 4,126 respondents via Macromill, Inc.; quotas ensured at least 1,000 experienced users of allotment farms and/or farm stands and balanced distributions across age groups and residential areas (Tokyo special wards and Tama suburban cities). After excluding missing data (income, neighborhood, incomplete IPAQ), the analytical sample was 3,135 participants. Context: Tokyo’s intraurban areas (urbanization promotion areas) include widespread small-scale farmlands enabling walkable/bikeable access to local foods. Measures: (1) Access variables—frequency during the emergency to allotment farms, home gardens (including terraces/balconies) with a separate identification of those growing food, farm stands (direct-to-consumer sales), supermarkets, convenience stores, co-op deliveries, and other urban green spaces (small parks, large parks, greenways), each on a five-point scale. Regular users were defined as accessing at least once or twice a week. (2) Health outcomes—Subjective well-being measured by S-WHO-5-J; dichotomized into poor vs. fine per WHO-5 criteria. Physical activity measured by IPAQ short form; total weekly MET-minutes computed and dichotomized as active vs. not active based on Ministry of Health, Labor and Welfare thresholds (18–64 years: 1380 MET-min/week; ≥65 years: 600 MET-min/week). (3) Food security concerns—Anxiety about availability of fresh food (vegetables, fruits) rated on a five-point Likert scale for during the emergency and for a hypothetical future logistics disruption. Sociodemographic covariates: gender, age, household income, family structure (living alone; living with young child), employment status (not working; working from home; working at an office—with emphasis on working from home due to pandemic), housing type (detached house vs. apartment), and neighborhood environment. Spatial analysis: Percentage of farmland within respondents’ neighborhood units (Cho-cho-moku) computed using ArcGIS Pro 2.9; presence/absence of farmland included as an attribute. Statistical analysis: Binomial logistic regressions assessed associations of local food access with subjective well-being (Model 1a) and physical activity (Model 2a), and of other green spaces with these outcomes (Models 1b, 2b). Comparative models contrasted each local food type against the green space most strongly associated with each outcome: allotment farms, home gardens, and farm stands versus large parks for well-being (Models 1c–1e) and versus greenways for physical activity (Models 2c–2e). Ordinal logistic regressions assessed associations of local food access with food security concerns during the emergency (Model 3a) and in the future (Model 4a), and of other purchasing sites (Models 3b, 4b). Comparative models contrasted each local food type versus supermarkets (during emergency, Models 3c–3e) and versus convenience stores (future, Models 4c–4e). Additional binomial logistic regressions identified characteristics of users of allotment farms (Model 5a), home gardens (Model 5b), and farm stands (Model 5c). All models adjusted for sociodemographic covariates listed above. Diagnostics and model selection: Pearson correlations and VIFs assessed multicollinearity. All-subsets models were ranked by AIC; models with ΔAIC < 2 were retained. Model averaging provided averaged parameter estimates and standard errors. Odds ratios (ORs) and 95% confidence intervals (CIs) were reported; significance at p < 0.05. Software: R 4.0.2 with MASS, MuMIn, and Tidyverse. Ethics: Approved by The University of Tokyo Graduate School of Engineering Ethical Committee (KE20-8); informed consent obtained online; anonymous and confidential.

- Subjective well-being: Users of allotment farms and home gardens had significantly higher well-being than non-users. In head-to-head comparisons, allotment farms were more strongly associated with well-being (OR 2.06; 95% CI 1.41–3.00; p < 0.001) than large parks (OR 1.44; 95% CI 1.16–1.79; p = 0.001). Home gardens also showed a positive association (OR 1.59; 95% CI 1.30–1.95; p < 0.001). - Physical activity: All three local food access types were associated with higher physical activity—allotment farms (OR 1.61; 95% CI 1.15–2.26; p < 0.01) > farm stands (OR 1.48; 95% CI 1.17–1.88; p < 0.01) > home gardens (OR 1.22; 95% CI 1.01–1.48; p < 0.05). Among green spaces, greenways had the strongest association (OR 3.36; 95% CI 2.73–4.14; p < 0.001), exceeding all local food types in direct comparison. - Food security concerns: Users of allotment farms and farm stands reported significantly lower anxiety about fresh food availability both during the state of emergency and for the future. For future anxiety, allotment farm users had particularly high ORs indicating lower anxiety (OR 2.92; 95% CI 2.08–4.10; p < 0.001). Home gardeners had lower anxiety during the emergency (OR 1.22; 95% CI 1.01–1.48; p = 0.04), but not significantly for the future. Compared to other purchasing sites, allotment farms and farm stands had higher ORs than supermarkets (during emergency) and higher ORs than convenience stores (future) for reduced anxiety. Supermarket use was associated with lower anxiety during the emergency; convenience store use was associated with lower future anxiety; co-op deliveries showed no significant differences. - User characteristics: Common positive attributes for all three local food access types were not living alone and working from home. Specific attributes: home gardens—women, older adults, and living in a detached house; farm stands—living in a detached house and in an area with farmland; no additional specific attributes for allotment farms. Working from home showed strong associations: allotment farms (OR 3.02; 95% CI 2.04–4.49; p < 0.001), home gardens (OR 1.59; 95% CI 1.25–2.01; p < 0.001), farm stands (OR 1.81; 95% CI 1.41–2.32; p < 0.001). Low household income was not significantly associated with any local food access type. - Overall: Urban agriculture within walkable neighborhoods was linked to improved well-being and physical activity and to reduced food security concerns, with effect magnitudes varying by type (strongest and most comprehensive for allotment farms).

The study addresses the research question by demonstrating that access to local food via urban agriculture in walkable neighborhoods is associated with better subjective well-being and higher physical activity, and with reduced food security concerns during a systemic shock (COVID-19) and for anticipated future disruptions. The comparative analyses suggest that self-cultivation (allotment farms, home gardens) may confer stronger well-being benefits than conventional urban parks, and that direct-to-consumer access (farm stands) and self-cultivation can mitigate food security anxiety more than traditional retailers. These findings support the integration of diverse urban agriculture modalities into neighborhood-scale planning to enhance health and food system resilience. Heterogeneous benefits across modalities suggest tailoring urban agriculture options to different populations: allotment farms for comprehensive benefits but requiring more effort; home gardens accessible for older adults and women; farm stands providing low-barrier access for those lacking time/space. The strong association with working from home indicates that flexible work styles can catalyze engagement with local food systems, potentially transforming underused urban spaces into productive landscapes. Equity implications are notable: access to local food was not income-dependent in Tokyo, likely due to mixed-income neighborhoods and affordability of farm-stand produce and municipal allotment rents, aligning with equitable resilience goals. The integration within walkable neighborhoods also aligns with decarbonization aims, potentially reducing food-miles and promoting active travel, thus offering a multi-benefit planning strategy.

This study provides quantitative evidence that integrating urban agriculture into walkable neighborhoods contributes to public health (subjective well-being and physical activity) and strengthens food system resilience by reducing anxiety over fresh food availability during crises and into the future. Allotment farms showed the broadest and strongest associations, with home gardens and farm stands offering distinct, complementary benefits. Policy implications include promoting a diversity of urban agriculture spaces (allotments, home gardens, farm stands) within 15-minute neighborhoods; leveraging flexible work trends to engage residents; repurposing underused urban lands and rooftops; and conserving agricultural landscapes in expanding cities. Future research should employ longitudinal designs to assess post-pandemic persistence of effects, incorporate objective health measures, and further disentangle causal mechanisms and differences among green space types.

Key limitations include timing and seasonality (survey conducted June 4–8, 2020, just after the first state of emergency, during planting/growing with harvest just beginning), which may influence findings; reliance on subjective measures for health and well-being due to pandemic constraints (objective measures might yield different results); and the cross-sectional design, which precludes causal inference. Longitudinal studies are needed to assess durability of observed associations beyond the emergency period.