The corporate food regime uses this “commodity fetishism” to hide the socio-natural relations of productions that might dissuade consumers and inhibit endless capital accumulation . Work in economic geography on global commodity chains and circuits has sought to “lift the veil” on the secret lives of commodities in the apparel and agri-food industries . This work seeks to unravel the ‘geographical knowledges’ or ‘political ecological imaginaries’ that people possess about settings, biographies, and origins that obscure the material and social relations of production and consumption . Revelations on these material and social relations of production have stimulated a rise in ethical consumerism , although a double fetish can still occur when products are valorized by virtue of their alterity . The commodity circuits approach has undoubtedly been fruitful in research on the geographies of food . Stemming from critiques of commodity chain analysis stating that it was too focused on production, the circuit approach accounts for the role of consumption and culture in the lives of commodities, examining “how culinary culture is constituted through commodity meanings and practices as they circulate and are reconstructed across systems or networks from one site to another” . The idea of circuit connects productively to actor-network theory, the final leg of this theoretical framework. This relational approach, sees the world as a collection of heterogenous assemblages made up of hybrid human actors and non-human actants,snap clamp the latter describing anything that is a source of action, but lacks the motivation we typically associate with human actors , such as soil, water, and organic certification. Their networks are constantly being negotiated through processes of translation in which actants are enrolled in network assemblages .

This theory is useful for examining urban food networks which are composed of “interconnected networks of farmers and gardeners, government agencies, supportive organizations, foundations, and investors, as well as the natural environment and the policies and programs that affect city’s food and environmental systems” . For instance, Goodman argues that different forms of agriculture represent competing collectives that “must foil efforts by competing collectives to translate and enroll their constituent entities” . Actors may shift and enroll in competing networks as a reaction to changing circumstance or events. Similarly, new actants will emerge and become part of networks as new policies, spaces, products, technologies, and stories emerge. This concept connects to arguments that see different forms of urban agriculture, like soilless and soil-based, as part of difference networks .One of the strengths of actor-network theory is its ability to bring together the material and discursive to consider how narratives are produced by different actors in ways that reflect and shape practices . Actor-network theory has also been particularly fruitful in understanding social movements , with ramifications for the study of alternative food movements such as urban agriculture . However, it has been the source of scholarly critique, particularly around how it treats power relations. For example, Ginn and Demeritt argue the theory “merely describes rather than also critiquing persistent inequalities… remain[ing] complicit in reproducing relations of inequality” . This conceptualizes power as de-centralized within networks, potentially ignoring the disproportional power particular actors have to persuade, and thus enroll, others, including non-human actants, into their network.

Instead, researchers call for a weaker actor-network theory, that recognizes that “agents, while social, natural, and relational, vary greatly in their powers to influence others; that power, while dispersed, can be directed by some more than others” .The three theories connect productively to questions of justice. Inspired by Marxism and political-economy, urban political ecology examines the uneven power relations and inequities – such as exploitative land and labor practices – that underlie commodities like urban agriculture in the context of capitalism. Commodity circuit analysis and actor-network theory come together to unveil these inequities by “following” urban agriculture across its local commodity circuit and examining the networks of actors and non-human actants that scaffold the material and symbolic lives of urban agriculture commodities. In combining these three perspectives, my theoretical framework allows me to examine the race- and class based power relations that are embodied in various urban agriculture networks and draw conclusions about justice. This research uses mixed methodology – a subcategory of research methods that allows data from multiple sources to be integrated to create comprehensive, empirical accounts of phenomena . Specifically, this research uses a quantitative preliminary design, which consists of using quantitative observations as a starting point to inform a broader qualitative study. This design may be used to accomplish two goals: 1) to explain quantitative results with qualitative research and 2) to use quantitative data to inform sampling choices for subsequent qualitative analysis . This research capitalizes on both of these strengths. Here, quantitative analysis, specifically topic modelling, is the starting point for understanding how people at various urban agriculture sites think and talk about justice and is subsequently elaborated using a qualitative method called multi-locale ethnographic analysis. This in-depth form of analysis allows us to further understand the nuanced, everyday experiences of justice that exist throughout urban agriculture commodity circuits.

The prerequisite topic modelling also informs sampling, particularly the urban agriculture sites chosen as case studies for ethnography, and is supported by exploratory spatial data analysis. Combining quantitative and qualitative methods allows me to thoughtfully answer the questions at the heart of my dissertation with rigor that would arguably be limited by choosing a single method. Multiple forms of data were required for this research. First, an exhaustive list of urban agriculture production sites and regional organizations in San Diego County was compiled in Excel with respective geographic locations and website addresses. For sites and organizations with websites , textual data on all web pages were pulled and compiled in .txt files. The abstracts of scientific literature on urban agriculture were also downloaded using the Web of Science database and compiled into a .txt file to aid in topic modelling of website textual data. To analyze the socio-economic landscapes in the county – population, economic characteristics , race, ethnicity, and immigration, housing, and businesses data – I use tract-level data from the 2013-2017 American Community Survey 5- Year Estimates, of the United States Census. Multi-locale ethnographic analysis required ethnographic fieldwork. This fieldwork included two years of extensive participant observation at multiple sites in the local urban agriculture networks of three case sites chosen through analysis of textual data from websites and geographic data from exploratory spatial data analysis. Thirty-four semi-structured interviews with actors in these networks including city planners, growers, technical experts, funders, nonprofit leaders, farmers’ market vendors, chefs, consumers –were also performed. The interviews were approximately an hour in length and covered institutional histories, actors’ personal motivations for participating in urban agriculture, their growing practices, their perceptions of the local food environment, and the struggles and barriers they perceive to urban agriculture.

Finally, secondary data, particularly newspaper and magazine articles, were pulled to supplement the primary data collected at these sites. Ethnographic data collection across multiple sites is consistent with the multi-locale ethnographic approach which moves away from the convention of examining a single site as an isolated container of examining social relations .Textual data from websites and scientific literature were analyzed using topic modelling – a common content analysis research method . The main objective of topic modelling is to condense textual, verbal, or visual messages into concepts or categories that can be used to describe phenomena and build conceptual maps . Topic modeling has been used to analyze various texts,growing blueberries including journal articles , blogs , and more recently, Twitter data . Here, this technique was used to discern how members of soilless and soil-based urban agriculture organizations think and talk about food justice, specifically in online discourse. First, the abstracts from scientific literature on urban agriculture were condensed and modelled to create a reference for a second model of the textual content of local urban agriculture organization and business websites. This subsequent model was used to build discursive maps that unveiled the connections between online discourse, affiliation, and growing method . Discursive maps also provided a visual representation of the discursive relationships between urban agriculture production sites in San Diego County which aided in choosing case sites for ethnographic analysis. Socio-economic maps were also used to choose the case sites. These maps were created using ESRI Geographic Information Systems software. They combined the geographic locations of urban agriculture sites with the Census tract level data on socio-economic landscapes of San Diego County. Using these geographic maps, the discursive maps, and basic information on site characteristics such as growing method and affiliation, three case sites were chosen for further analysis: Coastal Roots Farm, Mt. Hope Community Garden, and Solutions Farms. Stakeholders at all of these food growing sites expressed concerns for social justice issues like poverty, racial oppression, and homelessness, respectively. Coastal Roots Farm and Mt. Hope Community Garden were chosen because they, interestingly, the sites shared discursive concerns – they are on top of one another on the discursive map – and use soil-based growing practices, but are located in neighborhoods with incredibly disparate socio-economic circumstances. Their discursive similarities and geographic differences, and seemingly different approaches to justice, make these sites interesting for further analysis. Solutions Farm, an aquaponic social enterprise model, is another interesting site because it is discursively and geographically distant from the other two sites, and uses a soilless growing method to tackles social concerns.

With the case sites chosen, targeted exploratory spatial data analysis was performed to examine and understand the socio-spatial landscapes and place characteristics that influence their actor-networks. ESDA has become a popular method in social science research including human geography because it enables researchers to visualize and explore socio-economic data and identify spatial patterns . This analysis enriched our discussion of interview and participant observation data. These data were coded using secure, online coding software called Dedoose. Both a priori and emergent coding schemes were used to understand justice narratives and practices described and embedded in these data. The a priori coding scheme was designed to encompass direct acknowledgments of justice including food security, food justice, social justice, and food sovereignty narratives and practices, as well as the barriers network members perceive to urban agriculture. Emergent coding was focused on more nuanced expressions of justice illustrated in everyday events that reveal the power dynamics and struggles that influence justice. Combining and analyzing ethnographic data collected at multiple sites within the actor-networks spanning the three distinct commodity circuits was necessary for examining the “people, connections, associations, and relationships across space” that influence justice narratives and practices. These methods – topic modelling, exploratory data analysis, and ethnographic multi-locale analysis – were integrated to deepen understandings of justice in San Diego County. Together, they allowed us to move from the publicly-available information displayed on growers’ websites to a more detailed examination of the nuanced, everyday experiences that unfold across the actor-networks that scaffold their entire commodity circuits . This dissertation is divided into three papers related specifically to the three research questions outlined in the introduction of this chapter. They are tailored to each question individually, but build upon one another productively to reveal the broader connection between food justice and multiple forms of urban agriculture in San Diego County. In thefirst paper, “Untangling method and motivation in urban agriculture: moving beyond a politics of technology,” a collaborative effort with Tim Schempp, M.S. and Dr. Andre Skupin, we begin the process of examining urban agriculture in San Diego County by investigating the themes underlying urban agriculture actors’ online presence, specifically website content. Themes, first identified through topic modelling of scholarly literature on urban agriculture, include location, food security, community gardening, social movements, food access, climate change, and innovation, among others. We use a novel, computer mediated method that combines natural language processing, dimensionality reduction, and data visualization to create discursive maps of the themes that urban agriculture organizations and businesses in San Diego County use to represent themselves. The discursive maps allow us to examine the associations between content themes present on websites and factors including growing method and affiliation .