Rural-urban connectivity and agricultural land management across the Global South

Highlights

• Simple analytical framework to assess urbanization processes affecting rural areas.

• Spatial patterns in rural-urban connectivity are explored across the Global South.

• Propensity score matching is used to test for rural-urban connectivity effects.

• Rural-urban connectivity positively associated with irrigation in most regions.

• High market access areas have 27–40% greater mean nitrogen fertilizer rates.

Abstract

Research on how urbanization affects rural agriculture has typically focused on loss of farmland due to urban expansion. However, more distal pathways that could link urbanization to rural agriculture, including enhanced connectivity through rural-urban migration and market access, remain poorly understood. Here, we assess whether greater rural-urban connectivity is associated with changes in agricultural land management across the Global South. Such associations are complex, and thus difficult to measure at this scale. We therefore take a two-step approach to investigate these relationships. First, using a multivariate clustering approach, we define a series of rural-urban connectivity typologies from existing spatial data on land use, demographics, rural market access, and rural population change (as a proxy for outmigration). We examine the variation in key agricultural outcome variables (mean cereal crop yields, % of attainable yields met, and cropping frequency) within the typologies, which shows that greater overall connectivity (market access and population change) is associated with higher cereal yields, yield attainment, and cropping frequency. Second, building on these clustering results, we develop hypotheses about the relationship between rural-urban connectivity and agricultural land use intensity. We then use propensity score matching to test these hypotheses by comparing locations with similar sociodemographic and land use characteristics. When controlling for gross domestic product (GDP) per capita, agricultural land, and population density, rural locations with relatively high market access, negative population change, and greater built-up area have significantly higher mean nitrogen application rates, irrigated areas, and cereal yields across the Global South. Results vary by region, but greater rural-urban connectivity and urban extents are generally associated with higher overall agricultural inputs and yields, particularly in Asia. However, we find little support for a relationship between connectivity and either % attainable yields met or field size. Our findings stress the need to better understand the mechanisms that link urbanization processes and agricultural management at different spatiotemporal scales.

Introduction

Urbanization, defined as changes in demographic composition and expansion of built-up areas, is one of the most important drivers of land-use change globally (Schneider et al., 2015). Urban expansion has contributed to the loss of substantial amounts of productive farmland in many regions (Bren d'Amour, 2016). However, urbanization is also linked to a myriad of physical, economic, or social processes occurring in rural areas (Jedwab et al., 2014). For example, a relationship between urbanization and diet trends toward more energy-dense processed foods and animal products could create feedbacks to rural agricultural production systems (Satterthwaite et al., 2010, Seto and Ramankutty, 2016). With increasing urbanization and food demand globally—especially in cities of the Global South—understanding the magnitude and variability of contemporary rural-urban interactions could provide insight on future land-use dynamics.

A growing body of research considers how urbanization impacts the agricultural systems that may supply food and other resources for cities. Rural and urban places coexist in a continuum, linked by multiple types of spatial and sectoral connections (Tacoli, 2004, Seto et al., 2012). Flows of people, goods, and information link urban populations to rural landscapes (e.g., Seto et al., 2012, Güneralp et al., 2013, Djurfeldt, 2015), which we refer to as rural-urban connectivity. Markets link rural activities to cities and larger regions via trade, enabling rural agricultural production in one location to benefit other distant places (Verburg et al., 2011). Market accessibility metrics can therefore be useful for examining spatial relationships between rural agricultural production systems and cities, as markets can affect decisions about agricultural investment and production (Sloat et al., 2018). Farmers close to urban markets can more easily purchase agricultural inputs, access services such as credit and insurance, and trade their products, which can lead to both increased agricultural productivity and agricultural specialization (Masters et al., 2013).

Outmigration from rural to urban areas can also link urbanization to agrarian change through the various connections between individuals’ places of origin and destination (Tacoli et al., 2015). Urban centers often offer more diverse employment, educational, and cultural opportunities compared to rural areas, effectively extracting labor from agricultural sectors. Resulting changes in land-to-labor ratios can affect farm characteristics, including farm size, use of inputs, and commercialization of farms in general (ISCP (Independent Science and Partnership Council) 2013, ISU (International Sustainability Unit) 2015). The remittances received by rural households from outmigration of family members may lead to greater investments in agricultural land in terms of labor and other inputs (Tacoli, 2004, Ospina et al., 2018), though some studies have found mixed effects of outmigration on agricultural outcomes (Gray and Bilsborrow, 2014, Ochieng et al., 2016). Conversely, remittances could lead to livelihood diversification, allowing households to move away from agricultural livelihoods altogether (Ellis, 1998).

Several local- and national-scale studies examine the mechanisms linking urbanization to agricultural land practices (e.g., Masters et al., 2013, Onwuchekwa and Ukandu, 2015, Jiang et al., 2013, Ospina et al., 2018). For example, Jiang et al. (2013) use a panel econometric model to assess linkages between urban expansion onto agricultural lands and changes in agricultural land use intensity in China. However, systematic analysis of rural-urban linkages is rarer in regional- and global-scale research (Seto et al., 2012). Global land systems classifications provide a conceptual starting point by highlighting underlying spatial patterns, such as regions where high rates of urbanization coincide with different forms of agriculture (van Asselen and Verburg, 2012, Václavík et al., 2013). Land system classifications increasingly incorporate different socioeconomic and agro-environmental factors, such as per capita incomes and land management characteristics (Ellis and Ramankutty, 2008, Letourneau et al., 2012). However, to our knowledge, connectivity variables, such as rural-urban migration and market accessibility, have not yet been integrated in land systems classifications at continental or global scales. Integrating rural-urban connectivity into land systems classification could help to better understand the drivers and feedbacks between urbanization and rural agricultural land use.

As a starting point for empirical analysis of potential relationships between urbanization and agricultural land use and management, we devised an analytical framework that combines the concept of rural-urban connectivity with key themes from the literature on global land systems (Fig. 1). Our framework depicts potential social and economic interactions between urban and rural areas embedded within a larger relationship between rural land system change and agricultural land management. We apply this framework to examine the extent to which urbanization processes may be associated with different agricultural inputs and outcomes across the Global South. Specifically, we focus on the role of rural-urban connectivity (i.e., rural market accessibility and rural outmigration) as a potential mechanism that could influence rural agriculture, distinguishing between the more proximate effects of urban land extent and the more distal effects of market access and migration. Agricultural land use intensity can be measured in multiple ways, including agricultural outputs per unit land per unit time (e.g., crop yield), cropping frequency, or inputs per unit area (e.g., fertilizer use) (Turner and Doolittle, 1978, Jiang et al., 2013, Erb et al., 2013). Accordingly, we adopt several indicators of agricultural inputs and outcomes, which are relevant to characterizing agricultural management and land use intensity.

Our aim in this study is to explore whether rural-urban connectivity, in terms of market access and human migration, is related to agricultural inputs (e.g., fertilizer application rates) and outcomes (e.g., crop yields) across the Global South. We use a two-step approach to test this. First, we conduct an exploratory analysis with multivariate clustering to examine how rural-urban connectivity relates spatially to several different agricultural and urbanization variables. Multivariate clustering facilitates exploration of complex relationships among multiple variables at the same time. By examining the spatial patterns in the resulting clusters and how these compare to key agricultural outcome variables (crop yields, % yield attainment, and cropping frequency), we draw three specific, directional hypotheses about the relationship between rural-urban connectivity and agricultural land use intensity. In a second step, we use an inferential statistical approach with propensity score matching techniques to test these hypotheses. Matching facilitates comparison among locations that differ in the connectivity measures by controlling for potential confounding factors, such as GDP per capita (sensu Jiang et al., 2013). In the matching analysis, rural-urban connectivity is compared across binary control/treatment groups (e.g., high vs. low market access) and each agricultural outcome/response variable (e.g., crop yields, % yield attainment, and cropping frequency) is considered in a separate model. Our findings from both approaches provide some evidence that urbanization processes may be associated with certain types of changes in nearby agricultural landscapes, stressing the need to consider what implications projected future urban growth could have for agricultural land use and management.