Publications

Quantifying relationships between governance, agriculture, and nature: empirical-statistical-and pattern-oriented modeling

Mandemaker, M.

Summary

Quantifying relationships between governance, agriculture, and nature: empirical-statistical- and pattern-oriented modeling

Abstract

An improved understanding of complex processes of both socio-political and economic governance may help to abate negative effects of increasing pressures of rising global food demand and agricultural land use on nature. Therefore, in this thesis, relationships between governance, agriculture, and nature were quantified, and it was investigated to what extent governance can be meaningfully included into land-use modeling. As outlined in the general introduction of this thesis, two main approaches to the inclusion of governance were distinguished: a semi-positivistic empirical-statistical approach and a positivistic process-based approach. Therefore, this thesis consists of two parts: one that includes governance through empirical-statistical modeling, and one that includes governance through process-based modeling.

In the first part of this thesis (chapters 2–3), relationships between quantified perceptions of state-centric governance, arable agriculture, and nature were identified through semi-positivistic empirical-statistical approaches. To investigate whether state-centric governance directly drives—and if so, how—expansion and/or intensification of arable agriculture at the global level (chapter 2), empirical-statistical relationships between agricultural production dynamics and state-centric governance indicators were identified for 173 countries between 1975 and 2007. Four groups of countries were distinguished: those with both area and yield increases (“growth” countries); those with increasing yields but decreasing area (“intensifying” countries); those with decreasing yields but a growing area (“expansion” countries); and those with both declines in yields and area (“decline” countries). Differences between these four groups were analyzed, and also governance-production relationships within these groups. The analysis of governance-production relationships within the four groups suggests that countries with a lower quality of governance are more inclined to achieve production increases by expanding agricultural area rather than increasing yields. Moreover, as quality of governance becomes higher, this orientation of countries towards production tends to flip from expansion towards intensification. Overall, higher quality of state-centric governance is more likely to drive the intensification of arable agriculture, while lower quality of state-centric governance is more likely to drive the practicing of cheaper forms of arable agriculture, requiring less financial investment to keep up with the growing demand for food.

To investigate whether state-centric governance directly drives—and if so, how—dominant spatiotemporal processes of fragmentation and expansion of nature, empirical-statistical relationships between overall quality of state-centric governance and such processes were identified through cross-national comparison for 20 European countries for the period 1990–2006 (chapter 3). Land-cover change trends were detected by comparing CORINE land-cover maps from the years 1990 and 2006. Dominant spatiotemporal processes of change in nature were characterized by capturing past developments in two indicators: overall change in nature area and change in average patch size of nature. This resulted in four different dominant spatiotemporal processes of structural change: “Formation of larger-than-average patches of nature and/or Expansion of nature and/or Connection of nature”; “Formation of smaller-than-average patches of nature”; “Removal of smaller-than-average patches of nature”; and “Removal of larger-than-average patches of nature and/or Shrinkage of nature and/or Dissection of nature”. The majority of countries expanded their total nature, although this is likely not to have been the result of deliberate nature restoration alone but also of land abandonment or afforestation for production purposes, which do not necessarily lead to positive development in terms of biodiversity. For the period 1990–2006, overall quality of governance was found to be positively related to expansion of nature, and negatively to increasing patch size of nature. Generally, lower scores of overall quality of state-centric governance are more likely to drive the processes of removal, shrinkage, and dissection of nature through changes in nature area. Higher scores of overall quality of state-centric governance are more likely to drive the processes of formation, expansion, and connection of nature through changes in both nature area and patch size. Overall, it appears that quality of state-centric governance indeed drives the deliberate development of nature for the purposes of restoration and conservation, halting fragmentation and the ongoing decline of biodiversity in Europe.

On the one hand, the combined results of chapters 2–3 may imply that increased inclination toward expansion of agriculture could be commensurate with increased removal, shrinkage, and dissection of nature (i.e., negative change in nature area). On the other hand, the combined results of chapters 2–3 may imply that increased inclination toward intensification of agriculture could be commensurate with increased formation, expansion, and connection of nature (i.e., positive change in nature area). Overall, this would be consistent with the fact that lower scores of quality of state-centric governance are more likely to drive the practicing of cheaper forms of arable agriculture, requiring less financial investment to keep up with growing demands for food. In particular, expansion at the expense of nature, which is common in developing countries where nature is often abundant yet poorly protected (chapter 2).

In the second part of this thesis (chapters 4–5), a positivistic process-based approach was applied to include governance into land-use modeling. To investigate to what extent found the semi-positivistic empirical-statistical relationships could be simulated, a spatially explicit pattern-oriented individual-based land-use-transition model was constructed (chapter 4). Given that conceptualizations of decision-making and other mechanisms controlling behavioral and environmental constraints approach those of a real-world study area sufficiently close, and that land-use change data of this study area permit identification of the degree and composition of systematic influence on aggregate land-use patterns, meaningful comparison of empirical and simulated spatial statistics is possible. Furthermore, it is possible to approximate real-world land-use patterns with simulated ones, by minimizing differences between real-world- and simulated spatial statistics. These differences could be minimized by applying Pattern-Oriented Modeling (POM), iteratively comparing real-world- and simulated land-use patterns, and making systematic adjustments to explore which mechanisms and settings deliver optimal results. It was concluded that the empirical-statistical result that occurrence of agricultural intensification is likely to be more commensurate with higher quality of governance (chapter 2), can be reproduced as an emergent property of aggregated individual behavior of farmers constrained by state-centric governance (by the constructed model).

To investigate under which conditions nature could be protected from agriculture in an optimal way, individual farmers were subjected to different state-centric governance settings in the constructed model (chapter 5). One state-centric governance setting was assumed to be determined by the quality of the investment climate, reflected by the quality of the state-centric governance dimension “Regulatory quality”. Another was assumed to be determined by the level of entrepreneurial awareness, reflected by the quality of the state-centric governance dimension “Government effectiveness”. Together, these assumptions could be combined into four state-centric governance settings: “Low Entrepreneurial Awareness and Low Quality of Investment Climate” (I); “Low Entrepreneurial Awareness and High Quality of Investment Climate” (II); “High Entrepreneurial Awareness and Low Quality of Investment Climate” (III); and “High Entrepreneurial Awareness and High Quality of Investment Climate” (IV). Furthermore, farmers were subjected to two different top-down policy instruments in each state-centric governance setting: subsidies and fines. The most effective- and second-most effective protection of nature was generated by applying fines in state-centric governance settings III and IV, respectively. It was concluded that the empirical-statistical result that better protection of nature is likely to be commensurate with higher quality of governance (chapter 3), can be reproduced as aggregate emergent properties of individual behavior of farmers in different policy scenarios (by the constructed model) (chapter 4). Furthermore, that abstract process-based simulation (pattern-oriented modeling) of spatially explicit agricultural land-use systems allows for identification of different and verifiable governance conditions and policy strategies regarding protection of nature, and for identification of the optimal verifiable governance conditions and policy strategy for protection of nature and the economic interests of farmers in agricultural land-use systems.

It was concluded that governance can be meaningfully included into land-use modeling in both semi-positivistic (i.e., empirical-statistical) and positivistic (i.e., process-based) ways. However, different combinations of empirical specificity and degree of abstraction, i.e., different models, describe the same real-world complexity differently. Therefore, a conceptual representation of land-use models characterized by empirical specificity and degree of abstraction was proposed (chapter 6). According to this representation, there are three main classes of land-use models. There are those models that are more empirically specific than abstract (), indicating increasingly empirical-statistical models. Furthermore, there are those models that are more abstract than empirically specific (), indicating models that are increasingly process-based. It was observed that if a model of a real-world land-use system could optimize the empirical specificity with which this model can be validated and the degree of generality to which this system can be understood, then its predictive power might be maximized. That is, the class of models for which empirical specificity is equal to the degree of abstraction () might represent a class of models for which predictive power is potentially maximal. However, the research in this thesis suggests there is still a considerable gap between the used empirical-statistical models () and the constructed pattern-oriented model (). Therefore, it is recommended that research efforts be devoted to the development of models that further reduce the gap between empirical specificity and degree of abstraction, that is (. Moreover, to apply and validate such models successfully, large amounts of data would be required that are currently unavailable. This is why even more research efforts should be devoted to the development of high-resolution monitoring systems of land use, to allow for larger and better databases.