Artigos

INTRODUCTION
The promotion of shorter food sales circuits is considered a crucial global agenda, as shortening food chains can contribute to multiple objectives outlined in the United Nations Agenda 2030 for Sustainable Development. Specifically, this aids in advancing Responsible Consumption and Production (Sustainable Development Goal 12) 1. Farmers' markets play a pivotal role in these shorter marketing circuits for food by establishing direct relationships between producers and consumers, thereby ensuring the economic and social sustainability of small-scale farmers. They integrate food and nutritional security systems in numerous Brazilian cities and regions, representing a sustainable strategy for urban food supply 2.
Traditionally, the Brazilian population buys fresh or minimally processed foods, such as fruits and vegetables (FV), in farmers' markets 3. A study carried out in Porto Alegre found that farmers’ markets are one of the main places to buy food among people who have a healthy eating pattern, mainly composed of FV 4. This suggests that these markets serve as indicators of healthier food environments, as their presence nearby can encourage the purchase of foods with higher nutritional value, often at lower prices compared to other traditional food establishments.5
Physically, farmers' markets are temporary retail spaces occupying streets and public areas on specific days and times 6, their locations are defined by food supply decentralization policies to ensure accessibility for the entire city's population 7. However, previous studies in Brazil have found that farmers' markets are predominantly situated in central city regions and near higher-income populations 8-10. This tendency may contribute to lower FV consumption among socially disadvantaged populations. Several social determinants, including income, education, race/skin color 11, along with environmental factors like the availability of establishments selling these products and urban spatial accessibility, influence FV consumption. Hence, the barriers faced by individuals from lower social positions in accessing farmers' markets can perpetuate a pattern of social health inequalities in Brazilian cities.
Most of the studies that studied farmers' market accessibility in Brazil adopt container approaches to measure spatial accessibility, assessing the co-location of individuals and establishments, that is, the belonging of both to the same geographic area (neighborhoods, census tracts or buffers around the dwellings). An alternative, more realistic approach that has become the standard is the use of accessibility indicators that effectively calculate the possibility of people physically reaching certain destinations (whether they are places of supply, employment, study, or leisure), located in any part of the city, using any actual mode of transport (walking, bicycle, bus and car) and moving through the streets of the city’s street network 12. The study aims to describe the level of physical accessibility to farmers' markets in Porto Alegre and its associations with income and race/skin color of inhabitants.
METHODS
Study design and population
This study employs an ecological approach with a focus on spatial analysis, conducted using secondary data from Porto Alegre, RS, Brazil. In Porto Alegre, the municipal government has overseen the organization of farmers' markets since the 1950s, with the enactment of Municipal Decree n. 474 that regulated their operation and assigned specific locations for the markets. These locations were determined based on criteria such as neighborhood demographic density and proximity to wholesale centers. Subsequent legislations have been implemented over the years to regulate various aspects of these markets, as well as other forms of street commerce. However, these measures have not significantly altered their quantity or locations 13.
Unit of analysis
Based on data from the 2010 Brazilian Census, the Access to Opportunities project, spearheaded by IPEA (Instituto de Pesquisa Econômica Avançada), reconfigured the information initially gathered and consolidated in census tracts. This was achieved by aggregating the data into a spatial grid composed of hexagonal cells with a 357-meter diagonal. This grid system is known as "H3 – Hexagonal hierarchical geospatial indexing system," developed and made accessible by the Uber company (Available at: https://h3geo.org/).
This homogeneous zoning and grouping system eliminates the issue known as the "Modifiable Areal Unit Problem" (MAUP), which arises when using spatial units that lack uniformity in shape and size and exhibit considerable variability in physical elements, as is the case with Brazilian census tracts 14. Particularly for an analysis primarily focused on measuring actual distances and travel times, such heterogeneity can introduce biases in calculations. Hence, adopting an MAUP-free system, such as the hexagonal grid, is justified. In Porto Alegre, the grid consists of 6,114 hexagonal cells, of which, for this study, only those with at least one inhabitant in 2010 were taken into account, resulting in a sample of 3,232 cells (52.8%).
Sociodemographic data
The data processed by IPEA provided information on the total population, the count and proportion of white, black, brown, indigenous, or yellow residents, as well as the average per capita household income. Race/skin color and income variables were transformed into binary categories. In the first instance, every cell with over 25.5% of the population (representing the ? 3rd quartile of the 2010 distribution) made up of ethnic-racial minorities (black, brown, indigenous, and yellow) was categorized as 'high presence' of these population groups, while the rest were classified as 'low presence.' In the second case, every hexagon with an average income value below BRL 474.35 (representing the ? 1st quartile) was labeled as 'lower income', while the remaining cells were designated as 'higher income'.
Data on food markets
Data regarding the type, location, days, and operating hours of the markets were gathered from the Porto Alegre City Hall webpage in August 2022 in the form of a geospatial file. This file contains addresses and geographic coordinates, enabling precise location within the city space. Each market's location was correlated with the grid, facilitating the creation of an origins-and-destinations matrix (with cells containing population as origins and cells housing markets as destinations).
The three types of farmers' markets existing in the city – (I) conventional farmers' markets, recognized as 'model markets' ['feiras modelo']; (II) farmers' markets; and (III) ecological farmers' markets, exclusively vending organic and agroecological foods – were consolidated into a single category. This consolidation was made considering that they all predominantly sell fresh foods, especially fruits and vegetables (FV).
Additionally, the location data for other types of food establishments in Porto Alegre for 2020 were examined for comparative analysis. These data, obtained from the State Treasury Department of Rio Grande do Sul, were categorized according to the nature of the products sold, following the criteria outlined in the 2018 Technical Mapping of Food Deserts prepared by CAISAN (Interministerial Chamber for Food and Nutrition Security). The categories included: i) exclusively or primarily unprocessed or minimally processed foods; ii) mixed products, typically incorporating on-site food services (such as bakeries and restaurants); iii) exclusively or primarily ultra-processed foods (e.g., confectionery, snack bars, fast-food outlets); and iv) grocery stores and supermarkets, evaluated separately based on the variety of products offered.
Analyses
The analysis of the distribution of food retail and farmers' markets in geographic space started with the NNI – Nearest Neighborhood Index - which computes the average distances between each point and the nearest point and divides this number by the expected average distance in a hypothetical random distribution with the same number of points covering the same area. The closer this value is to 0, the more concentrated the distribution is 15.
To assess the physical accessibility of the population to markets (destination), the Minimum Travel Time (MTT) indicator was used. This is the minimum travel time (in minutes) from each cell with population (origin) by each mode of transport to the closest cell with markets available 16.
The IPEA script package r5r, designed for the R software, was utilized to compute the route between each origin-destination pair. This calculation considered Porto Alegre's actual road network and incorporated four distinct modes of transportation: walking, cycling, public transport (bus), and car. In the case of walking and cycling, the algorithm seeks the shortest path between the origin and the destination along any accessible road. However, for the car, it factors in driving routes, traffic lights, and any other limitations present in the city of Porto Alegre's spatial road network model. (Available in: https://dadosabertos.poa.br/)
For bus trips, the code uses the same network as automobiles but adds the itineraries of the lines, the location of the bus stops and the frequency of the vehicles, available in the GTFS – General Transit Feed Specification (Available in: https://gtfs.org/) – from Porto Alegre, in 2019, (POA Dados Abertos https://dadosabertos.poa.br/). The calculation considers the path (generally traveled on foot) between the centroid of the cell of origin and the nearest bus stop, the route of the bus line between this and the bus stop closest to the market(s), as well as the final path (on foot) between both. Consequently, the total travel time by bus is the combined duration of walking times, waiting times, and the actual bus journey itself.
The code specifications have incorporated off-peak hours to calculate travel times by car and bus. This decision is based on the observation that motorized trips for purchasing food at farmers' markets typically occur between 9:00 am and 12:00 pm, as well as between 2:00 pm and 6:00 pm, deliberately avoiding the periods of 7-9:00 am, 12-1:00 pm, and 6-8:00 pm. These timeframes were considered non-peak hours for travel estimation17.
From the iteration of the search for routes and the calculation of the respective travel times between each point of origin and each of the markets for each of the travel modes, we classified the resulting times into 4 intervals: up to 10 minutes, 10 to 20 minutes, 20 to 30 minutes, and over 30 minutes.
Differences in accessibility between modes of transport and between cells with higher and lower income and with higher and lower presence of ethnic-racial minority residents were described by absolute and relative frequency., and their association was tested through Pearson’s chi-square for heterogeneity and linear trend, adopting a significance level of 5%. The data were treated statistically in the R software 4.2.1 and GeoDa 1.20.
RESULTS
Population and farmers' markets descriptives
Out of the 3,232 cells containing a population, 808 (25%) were categorized as lower income, while 812 (25.12%) were identified as having a 'high presence' of ethnic-racial minorities. Among the latter group, 425 cells (52.3%) were also classified as lower income, suggesting an expected association between race/skin color and income in Porto Alegre. These proportions mirror the figures discovered within the census tract grid (the original unit of analysis in the 2010 Census): 25% in the former case and 25.12% in the latter, affirming the accuracy of the data interpolation to the hexagonal grid created by IPEA.
Figure 1 illustrates that out of the 44 total markets in the city (comprising 6 farmers' markets, 6 agroecological farmers’ markets, and 32 of the model farmers’ market type), 59% of them are situated in the central zone or along primary avenues, which have been encompassed within the group of 'centralities' in Porto Alegre. Table 1 demonstrates that this percentage is the smallest among all categories of food establishments in the city. Additionally, the market concentration level (NNI = 0.91) is also the lowest in comparison to other types of food outlets.
Figure 2 shows the spatial distribution of the markets overlaid on the distribution of binary cells indicating higher/lower income (2a) and race/skin color (2b). Among the 44 markets, 5 (11%) are positioned in cells with a high concentration of ethnic-racial minorities, and none are situated in lower-income cells. Consequently, there are no markets located in areas simultaneously characterized by a predominance of ethnic-racial minorities and lower income.
Accessibility
Table 2 displays accessibility for each mode of transportation, accompanied by the combined population residing in those specific cells. Upon closer examination of the extreme categories, it is evident that 21.1% of the population can reach a farmers' market on foot within 10 minutes, 26.7% by bus, and 58% by bicycle, while 88.8% can do so by car. Conversely, 26.5% of the city's total population resides in areas more than 30 minutes on foot from the nearest market, with 14.3% requiring more than 30 minutes by bicycle, 13.8% by bus, and only 0.4% by car.

Table 3 reveals that only 2.7% of the population residing in cells classified as lower income can reach the nearest market on foot in less than 10 minutes, 36.4% by bicycle, 3.9% by bus, and 69.3% by car. Conversely, 54.2% of the lower-income population needs more than 30 minutes to reach the nearest market on foot, with 32.3% by bicycle, 37.6% by bus, and only 1.2% by car. In contrast, only 21.6% of the higher-income population takes more than 30 minutes on foot, 11.2% by bicycle, 9.5% by bus, and a minimal 0.3% by car. Regarding race/skin color, 11.2% of the population in cells with a high minority presence can access a market in less than 10 minutes by walking, 46.2% by bicycle, 15.2% by bus, and 78.4% by car. In areas with a low minority presence, 26.2% can reach a farmers' market on foot within this time frame, 64% by bicycle, 32.5% by bus, and an overwhelming 94.1% by car.
On the other hand, 43.8% of the population in areas with a high minority presence needs to walk more than 30 minutes to the nearest market, with 23.9% by bicycle, 25.9% by bus, and just over zero by car. In other locations, 17.7% walk more than 30 minutes, 9.3% by bicycle, 7.6% by bus, and 0.6% by car. All these differences are statistically significant in according to the chi-square tests for heterogeneity with p <= 0.001.
DISCUSSION
The farmers' markets in Porto Alegre exhibit a comparatively lower level of geographic concentration compared to other food stores, indicating a relatively balanced distribution across the city. This reflects the reality that their placement isn't solely guided by market principles as with typical businesses. Instead, their positioning is influenced by public supply policies aimed at bringing producers closer to consumers citywide. However, approximately a quarter of the markets are concentrated in the central area of the city, indicating some level of concentration. In Porto Alegre, whose primary centrality (“centro expandido”) is situated to the west of the geometric center, there appears to be an imbalance in the distribution.
This observation aligns with prior research that highlights how areas outside the central zones of cities typically lack farmers' markets, potentially hindering a large portion of the population's access to these resources9-11. Our accessibility analysis reinforces this, indicating that while markets are located in areas with significant residential presence, only about a quarter of the population can reach a market within a 10-minute walking or bus journey. In stark contrast, nearly 90% of the population can reach a market within the same timeframe using a car.
Our findings highlight advantages in terms of market accessibility for high-income areas, which is consistent with existing literature. Previous studies conducted in Brazil have indicated that regions with higher incomes tend to have a greater presence of establishments selling fresh and minimally processed foods6,9-11 Additionally, our results regarding locations with a predominantly white population align with Brazilian studies, such as the work by Honório et al. (2021), which examined racial disparities and concluded that areas with fewer options for healthy foods were more populated by Black, Brown, or Indigenous individuals 18 Similar studies in the United States have suggested that neighborhoods with a higher concentration of Black populations tend to have lower availability of healthy food options 19.
The most economically disadvantaged segments of minority ethnic-racial groups can only achieve a level of accessibility to markets comparable to the white and high-income groups by relying on a car for transportation. This underscores the historical advantages afforded to individual, motorized transportation in shaping Brazilian cities, both in terms of establishing activity locations and directing investments in the transportation system. This situation perpetuates disparities and underscores the privilege attached to motorized transportation, contributing to uneven access to essential resources and services, like access to food markets, among different socioeconomic and racial groups 20.
Despite these disparities, the bicycle has emerged as a viable alternative, facilitating short trips to markets for over half of the population. This underscores its effectiveness as an urban transportation mode 19. However, in the Brazilian context, the use of bicycles remains highly limited, constituting only about 1% of trips in cities with over one million inhabitants in 2018 20. This low adoption is due not only to limited coverage and poor quality of cycling infrastructure in most cities but also to behavioral resistance from other road users. Additionally, cyclists encounter various forms of prejudice, further hindering the widespread acceptance and usage of bicycles as a primary means of transport in urban areas 21, 22, 23.
Due to the challenge of accessing farmers' markets within a short timeframe, individuals from socially disadvantaged backgrounds often resort to purchasing fresh food from the nearest and most accessible places, typically grocery stores and supermarkets. However, in these establishments, prices are generally higher, and the quality of products may be lower. This situation may lead to the purchase of smaller quantities of fresh foods and a prioritization of ultra-processed alternatives 24. This trend aligns with the prevailing reality in Brazil and many Global South countries, where the spatial organization of cities and the prevailing mobility systems tend to accentuate spatial inequalities, contributing to social inequities in food access, nutrition, and overall health 25, 26.
Social inequities represent unjust variations between different groups within a population. These disparities encompass differences based on various factors including income, education, gender, access to opportunities, and socioeconomic status, among others. Additionally, racial inequities specifically pertain to differences between groups attributed to race or skin color, encompassing the essence of racism. This encapsulates unequal treatment, opportunities, or advantages and disadvantages experienced by distinct racial or ethnic groups within society 27. Moreover, the challenges faced by this population in accessing FV can be seen as a manifestation of structural racism. Structural racism is defined as a type of racial discrimination perpetuated in society through customs, practices, and laws, becoming ingrained in people's daily lives 28. In this context, structural racism molds individuals' environments and opportunities, imposing additional barriers for black, brown, and indigenous communities in accessing healthy food options, potentially resulting in adverse health outcomes.
Strengths and limitations
The primary strength of this study lies in its pioneering nature, as it delves into the research on food environments in Brazil by bridging health sciences with urban sciences. It employs a more realistic methodology to evaluate intra-urban accessibility, comparing various population strata beyond traditional proximity analyses. Unlike most cited studies employing a 'container' approach to assess accessibility, our study, to the best of our knowledge, is the first to consider real accessibility to street markets by different modes of transport using the physical road network of the city.
The results, however, should be interpreted while considering some limitations: (1) its ecological approach makes causal inference impossible at the individual level, but it can generate hypotheses and indicate paths for other study designs; (2) in this sense, it was necessary to work by measuring the accessibility of locations and, only then, computing the resident population, which leads to differences between the percentage of cells and population in each category in each mode that reflect the fact that the more central hexagons have more inhabitants than the peripheral ones; (3) the use of data from the 2010 Brazilian Census, already quite outdated, can mask changes in socio-spatial distributions.
In conclusion, our study highlighted the presence of social and ethnic-racial inequities in accessing farmers’ markets in Porto Alegre. Moving forward, it's important to conduct similar studies employing comparable methodologies in other Brazilian capitals. This approach will help assess the consistency of these findings in different urban contexts, contributing to a more comprehensive understanding of the unequal dynamics and patterns of accessibility to farmers’ markets across different regions in Brazil. Knowing these patterns allows for the improvement of healthy food supply policies with a focus on combating social inequities.
Acknowledgments
This study was supported by the National Council of Technological and Scientific Development. RC received CNPq Research Productivity Grant.

REFERENCES
1. Belletti G, Marescotti A. Short food supply chains for promoting local food on local markets. United Nations Industrial Development Organization. Food and Agriculture Organization (FAO) 2020. Available from: https://hub.unido.org/node/2879
2. Schwingshackl L, Schwedhelm C, Hoffmann G, Lampousi AM, Knüppel S, Iqbal K, Bechthold A. Schlesinger S, Boeing H. Food groups and risk of all-cause mortality: a systematic review and meta-analysis of prospective studies. Am J Clin Nutr 2017; Jun;105(6):1462-1473.
3. Câmara Interministerial de Segurança Alimentar e Nutricional (CAISAN). Mapeamento dos desertos alimentares no Brasil, Brasília: CAISAN; 2018. Available from: https://aplicacoes.mds.gov.br/sagirmps/noticias/arquivos/files/Estudo_tecnico_mapeamento_desertos_alimentares.pdf
4. Cunha CML, Canuto R, Rosa PBZ, Longarai LS, Schuch I. Association between dietary patterns and socioeconomic factors and food environment in a city in the South of Brazil. Cien Saude Colet 2022; Feb;27(2):687-700.
5. Costa BVL, Oliveira CDL, Lopes ACS. Ambiente alimentar de frutas e hortaliças no território do Programa da Academia da Saúde. Cad. Saúde Pública 2015; v. 31 (159-S169).
6. Cruz MS, Schneider, S. Feiras alimentares e mercados territoriais a estrutura e o funcionamento das instituições de ordenamento das trocas locais. Raízes: Revista De Ciências Sociais E Econômicas 2022; 42(1), 93–113.
7. Schneider S, Salvate N, Cassol A. Nested Markets, Food Networks, and New Pathways for Rural Development in Brazil. Agriculture 2016; 6(4):61.
8. Jaime PC, Duran AC, Sarti FM, Lock K. Investigating environmental determinants of diet, physical activity, and overweight among adults in Sao Paulo, Brazil. J Urban Heal 2011; 88(3):567-581
9. Pessoa MC, Mendes LL, Gomes CS, Martins PA. Food environment and fruit and vegetable intake in an urban population: A multilevel analysis. BMC Public Health 2015; 15, 1012.
10. Silva ADDCE, Silva ARD, Hofelmann DA. Distribuição espacial dos equipamentos públicos para comercialização de frutas, legumes e verduras em Curitiba, Paraná, Brasil. Cien Saude Colet 2021; Aug;26(8):3111-3121.
11. Canuto R, Fanton M, Lira PIC. Iniquidades sociais no consumo alimentar no Brasil: uma revisão crítica dos inquéritos nacionais. Cien Saude Colet 2019; Sep 9;24(9):3193-3212.
12. Instituto de Pesquisa Econômica Aplicada (IPEA). Projeto Acesso a Oportunidades, 2019. Available from: https://www.ipea.gov.br/acessooportunidades/sobre/
13. Porto Alegre. Secretaria Municipal da Produção, Indústria e Comércio. Resolução 01/2015, 2015. Disciplina o funcionamento das feiras ambulantes de hortigranjeiros do Município de Porto Alegre, revoga as Resoluções nº 02/2010 e nº 03/2011, e dá outras providências. Diário Oficial de Porto Alegre, 2015.
14. Openshaw, S. The Modifiable Areal Unit Problem. Geo Books, 1983.
15. Clark PJ, Evans, FC. Distance to Nearest Neighbor as a Measure of Spatial Relationships in Populations. Ecology 1954; 35(4).
16. Levinson D, King, Transport Access Manual: A Guide for Measuring Connection between People and Places, 2020.
17. Pereira RHM, Saraiva M, Herszenhut, D, Braga CKV, Conway, MW. r5r: Rapid Realistic Routing on Multimodal Transport Networks with R5R in R. Transport Findings, 2021.
18. Honório OS, Pessoa MC, Gratão LHA, Rocha LL, Castro IRR, Canella DS, Horta PM, Mendes LL. Social inequalities in the surrounding areas of food deserts and food swamps in a Brazilian metropolis. Int J Equity Health 2021; Jul 21;20(1):168.
19. Bower KM, Thorpe RJ Jr, Rohde C, Gaskin DJ. The intersection of neighborhood racial segregation, poverty, and urbanicity and its impact on food store availability in the United States. Prev Med 2014; Jan; 58:33-9.
20. Pereira RHM, Braga CKV, Serra S, Nadalin, V. (2019). Desigualdades socioespaciais de acesso a oportunidades nas cidades brasileiras, 2019. Available from:. http://repositorio.ipea.gov.br/handle/11058/9586
21. Goel R, Goodman A, Aldred R, Nakamura R, Tatah L, Garcia LMT . Cycling behaviour in 17 countries across 6 continents: levels of cycling, who cycles, for what purpose, and how far? Transport Reviews 2022; 42:1,
22. Associação Nacional de Transportes Públicos. Sistema de Informações da Mobilidade Urbana - SIMOB/ANTP - Relatório Geral 2018, 2020. Available from: http://files.antp.org.br/simob/sistema-de-informacoes-da-mobilidade--simob--2018.pdf.
23. Rojas-Rueda D. Health Impacts of Urban Bicycling in Mexico. Int J Environ Res Public Health 2021; Feb 26;18(5):2300.
24. Garcia L, Pearce, Abbas A, Mok A, Strain T, Ali , Crippa A, Dempsey PC, Golubic R, Kelly P, Laird Y, McNamara E, Moore S, Sa TH. de, Smith, AD, Wijndaele K, Woodcock J, Brage S. Non-occupational physical activity and risk of cardiovascular disease, cancer and mortality outcomes: A dose–response meta-analysis of large prospective studies. Br J Sports Med 2023;57:979-989.
25. Almeida IJ, Garcez A, Backes V, Cunha CML, Schuch I, Canuto R. Association between the community food environment and dietary patterns in residents of areas of different socio-economic levels of a southern capital city in Brazil. Br J Nutr 2022; Jul 11:1-9.
26. Darmom N, Drewnowski A. Contribution of food prices and diet cost to socioeconomic disparities in diet quality and health: a systematic review and analysis. Nutrition Reviews 2015; v. 73, n. 10, (p. 643-660).
27. Devakumar D, Selvarajah S, Abubakar I, Kim SS, McKee M, Sabharwal NS, Sain A, Shannon G, White AIR & Achiume, ET. Racism, xenophobia, discrimination, and the determination of health. Lancet 2022; Dec 10;400(10368):2097-2108.
28. Krieger, N. Measures of Racism, Sexism, Heterosexism, and Gender Binarism for Health Equity Research: From Structural Injustice to Embodied Harm—An Ecosocial Analysis. Annual Review of Public Health 2020; 41:1, 37-62.