Discussion

The results of this study further quantify acquisition and mobility challenges of food-insecure Americans during the COVID-19 pandemic, with important differences by urbanicity levels. We extend the literature on how food insecurity is shaped by socio-economic, spatial and infrastructure inequalities, which has important implications for integrated solutions to promote food security from urban to rural areas.

Our findings on the acquisition and barriers to food in urban and rural areas are consistent with previous literature. In rural areas, food-insecure individuals commonly reported food acquisition from supercentres (e.g., Walmart), reinforcing prior reports of the declining presence of local and independently owned grocery stores in rural towns in favour of the consolidation of large grocers on which rural residents rely^{(Reference Cho and Volpe11,Reference Palmer, Bonanno and Clancy12)} . Importantly, fewer food stores also mean greater distances between stores in rural areas, suggesting that for some food-insecure individuals, limited access to a personal vehicle or public transportation is likely to exacerbate food insecurity^{(Reference Piontak and Schulman3)}. Relative to urban areas, locally sourced food was also less common in rural areas. Consistent with prior studies, our findings suggest that the growth for local food (e.g., farmers markets) in urban areas outpaces those in rural areas^{(Reference Low, Adalja and Beaulieu13)}. To address this gap, food advocates in the federal government, private sector and civil society have increased access to local food in the past decade by investing in produce incentives at farmers markets and supermarkets, the acceptance of the Supplemental Nutrition Assistance Program’s Electronic Benefits Transfer cards at farmers markets, and transportation infrastructure through new distribution systems, like mobile markets, that deliver locally sourced food to families with limited transportation^{(Reference Whittaker and Hodgson14,15)} . This research points to opportunities for future work to explore solutions that could further strengthen food economies and address the underlying economic and infrastructural challenges of rural food insecurity.

Change in employment and limited availability of culturally preferred foods especially constrained food-insecure individuals in urban areas. These challenges underscore a need to expand understanding of food access towards one that is inclusive, enables greater agency and leverages income generation opportunities in various sectors of community food systems. Across the country, a growing number of farmers are providing culturally preferred foods to ethnically diverse populations in their respective city and surrounding region. One member organisation of Buffalo’s Seeding Resilience is Providence Farm Collective where Somali Bantu refugees resettled in Buffalo grow their heritage foods on land that is 20 miles outside the city^{(Reference Dewey16)}. Meanwhile, a network of Freedom Gardens is expanding backyard and front yard food production among Black, Indigenous and People of Colour in the City of Buffalo^{(Reference Raja17)}. In the Minneapolis-St. Paul area, Hmong farmers supply culturally preferred foods like long beans, bitter melon and amaranth greens in farmers markets serving the Twin Cities region^{(Reference Holt18)}. At the retail level, a food landscape study in a low-income urban neighbourhood of San Diego, with a diverse immigrant population, found that out of eighty-two food stores nearly half were ethnic markets providing access to fresh, affordable and culturally preferred foods^{(Reference Joassart-Marcelli, Rossiter and Bosco19)}.

Although our research generated new and important findings on food insecurity experiences across urbanicity, there are some limitations. The Harris Poll was not a probability-based panel and may be subject to selection bias towards individuals with internet access; however, its recruitment was national in scope and applied survey sampling weights to generate nationally representative estimates. The panel consisted of adults who were invited and agreed to participate online, which may have resulted in selection bias towards people with internet access and undercounting of some groups (e.g., individuals with lower income or living in rural areas); however, Harris Poll adjusted for respondents’ likelihood to be online using a propensity score weight. Responses were self-reported^{(Reference Bottino, Rhodes and Kreatsoulas20)}; however, our food insecurity measure was based on the previously validated 18-item US Department of Agriculture Household Food Security Survey, which limits the possibility of social desirability bias^{(Reference Hager, Quigg and Black10)}. Reports of locally sourced food and acquisition from home or community gardens may be underreported as data were collected early in the growing season. We performed a cross-sectional study and could not assess causal relationships between the pandemic and food-related experiences.