Urbanization, agricultural intensification, and environmental services: A spatial analysis

Urbanization is a global trend. Currently, about half of the world's population lives in cities. According to the latest calculations, this figure is expected to grow to 68 percent (about 6.8 billion people) by 2050. However, urbanization rates vary greatly from country to country and from continent to continent. Asia and Africa are the main hotspots of future urbanization, with India, China, and Nigeria alone accounting for 35 percent of urban population growth by 2050. At the same time, however, 90 percent of the world's rural population lives on these continents. Accordingly, there will be a major shift from rural to urban populations in many developing and emerging countries in the coming decades. Since the rural population in large parts of Africa and Asia is made up of smallholders, most of whom belong to the lower income brackets, it is particularly important to examine how future urbanization trends will affect their food and income security. This refers to potential positive economic consequences for smallholders, such as better access to modern technologies or alternative sources of income in the urban labor market, as well as to negative effects, such as the decline of necessary ecosystem services through intensive agriculture or soil sealing. Before this context, the three studies of this dissertation investigate both urbanization effects on agricultural management systems and the effects of agricultural intensification on ecosystem services. This is done using the example of the megacity Bangalore in southern India. Bangalore represents an increasingly important type of urbanization, so-called megacities. These are urban agglomerations with more than 10 million inhabitants and extremely high urbanization rates. According to the official census, Bangalore's population doubled between 2001 and 2011. Since these cities will become increasingly common in the future and are constantly expanding into rural areas, it is particularly important to study the effects of such a city on agricultural systems. The empirical analysis of this dissertation draws from two data sets based on household surveys and scientific experiments in two research transects along the rural-urban gradient of Bangalore. Socioeconomic data from 1,275 households, interviewed from December 2016 to May 2017, are available. Furthermore, in February and March 2018 131 households were surveyed again about their agricultural management and pan trap experiments were carried out in one of their fields to determine pollinator diversity. Since it can be assumed that urbanization effects increase in complexity the faster and uncontrolled urban growth occurs, the methodological focus of this dissertation is on spatial modeling with approaches from both spatial statistics and spatial econometrics. The first study of the dissertation mainly aims at the conceptual representation of complex effects of polycentric urbanization dynamics on agricultural systems. In addition, effects on agricultural intensification through potential income diversification due to improved access to urban labor markets are investigated. Previous studies model urbanization effects on agricultural systems with one-dimensional variables such as distance to the nearest city under the assumption that these variables are proportional to transport costs. These represent how well households are connected to markets. The one-dimensional variables are also based on the assumption that urbanization effects are monocentric (i.e. resulting from only one city) and concentric (i.e. the same gradient regardless of the direction). In a theoretical model in which we define the opportunity costs of agricultural intensification in the form of non-agricultural income, we can show that these assumptions of previous studies can be problematic. Empirically, we therefore use Structured Additive Regression (STAR) models in which we estimate the effect of household and village location (GPS coordinates) in the rural-urban interface on the number of modern inputs used (indicator of agricultural intensification) as two-dimensional Penalized (P-) splines. Using this method, we show that the spatially explicit estimation of urbanization effects has advantages over the urbanization indicators used by other studies (e.g. distance to the nearest city). Thus, the influence of streets and satellite towns is highlighted. In addition, urbanization effects vary greatly between the two research transects. We therefore deduce that urbanization effects on agricultural intensification are nonlinear and nonconcentric. In the second study we estimate urbanization effects and the effects of changing weather patterns on the decision of agricultural households to adopt borewell technology. For the empirical analysis we use so-called duration models. These are a dynamic model family and the development is defined by the probability that an event will occur in a given period of time (e.g. the installation of a well). Urbanization effects are again estimated with P-Splines and the effects of changing weather patterns with annual data on pre-monsoon, monsoon and annual precipitation. Results show that the probability of drilling a well increases with proximity to Bangalore, but also to satellite towns. In addition, a decrease in annual precipitation increases the likelihood that households will drill a well. In addition, an increase in monsoon rains seems to increase the well adoption rate. As a result, both the urbanization of the region and weather changes lead to an increased uptake of groundwater lifting technology. The last study examines the effect of the use of chemical fertilizers and pesticides on the abundance and species richness of wild bees in the research transects. Wild bees are an important pollinator group in the region and a decline of these animals as a consequence of more intensive agriculture would ultimately have a negative impact on the productivity and living standards of smallholders throughout the region. Therefore, it is important to understand how agricultural management decisions affect the pollinator population. One problem of such a research objective is the different scaling of agricultural decisions at plot level and the dispersal range of bee populations at landscape level. Since the latter is significantly larger (up to 6 km) than the plots of smallholders (on average 2 acres), this can easily lead to spatial correlation and spillover among observations at plot level. In the literature this problem is generally avoided by looking at dependencies at the landscape level. However, this leaves only little room for recommendations for action, since land use is to be understood as an aggregate of many individual decisions at plot level. By using spatial Durbin models, econometric models with spatial lags in dependent and independent variables, we can control for spatial correlation and spillover. Results show that bee populations in the Bangalore region have a movement radius of about 4 kilometers and that there are negative externalities to bee abundance on neighboring plots, especially in pesticide use. In addition, the long-term intensive management of plots also results in a decline in both abundance and bee abundance.