Home Sociology Understanding Society and Natural Resources
Different Perspectives and Research Approaches
To obtain a full understanding of the causes and consequences of land change a complementary use has to be made of different research approaches. These can be classified as the narrative, the empirical and the modeling approaches (Lambin et al. 2003). The results of the narrative and empirical approach are often used as input to the modeling approach that aims at formalizing the identified relations in a structured framework.
The narrative approach seeks depth of understanding through historical detail and interpretation. It tells the land change story, providing an empirical and interpretative baseline by which to assess the validity and accuracy of the other visions. It is especially beneficial in identifying stochastic and random events that significantly affect land change but might be missed in approaches employing less expansive time horizons or temporal sampling procedures (Briassoulis 2000). The narrative approach is mostly valid at the level of individual actors and one of the challenges of the approach is to link it with the features of land change that occur at more aggregate levels of analysis. This has given rise to efforts to better link 'people and pixels' through georeferencing narrative research and efforts to link the narrative approach to empirical approaches using geographical data (Liverman and Cuesta 2008; Rindfuss et al. 2003; Rindfuss and Stern 1998). By linking household data to the spatial units of land managed by those households, it becomes possible to relate household characteristics to the actual land management applied in the field.
The empirical approach builds on the narrative approach but takes a more quantitative perspective by identifying significant relations and pattern in the collected data while testing hypothesis that are either based on the narrative research approach or through deductive reasoning (Pfaff and Sanchez-Azofeifa 2004). Such empirical analysis can take place at various levels of spatial and temporal aggregation, ranging from the analysis of household survey data (Overmars and Verburg 2005) or the analysis of spatial units, i.e. pixels or polygons, organized in geographic data layers (Chomitz and Gray 1996; Veldkamp et al. 2001) to the analysis of time series of country-level statistics (Rudel et al. 2009). A major drawback of the empirical quantification of relations between land use and its supposed drivers is the induced uncertainty with respect to the causality of the supposed relations. The danger lies in leaping directly from the exploratory stage, or even from statistical tests based on descriptive models, to conclusions about causes (James and McCulloch 1990). Besides, most causal explanations are valid at the scale of study, mostly the individual actor of land change, and therefore subject to upscaling problems. This asks for validation of the causality of empirically derived relations. A combination of the narrative perspective with the empirical perspective can help to test the validity of the empirical relations. An example of such a combined approach is a study of Overmars in the Philippines (Overmars and Verburg 2005). Overmars used an approach that evaluates the results of statistical models based on geographic data by a household-level analysis of decision making.
The modeling approach uses theoretical, assumed or empirical relations to construct a model that allows the exploration of land change dynamics across historic (observed) or future time periods. Models especially allow the analysis of 'what-if' questions through acting as an artificial laboratory for conducting controlled experiments which are very difficult to establish in the real world. Similarly to the empirical perspective, land change models are aimed at a wide variation of different spatial and temporal scales. Local agent-based models mostly represent individual actors within a community or small region (Matthews et al. 2007) while spatial models often are applied at the regional level, simulating the changes in land use of land units or pixels. Land use is also an explicit part of larger scale models operating at the global level, ranging from global equilibrium models of the world economy (Hertel et al. 2010) to integrated assessment models of global environmental change (Thomson et al. 2010). The following section will describe the way in which human-environment interactions are addressed in land change models in more detail.
From the above it is clear that both the different research approaches and the different spatial scales of analysis are able to provide complementary insights. However, the linking of the approaches across the different scales may not be straightforward. Coleman (1990) developed a framework that describes the interaction between micro and macro levels for social systems. The same framework can also be applied to land change models. Land change assessments made at the regional level, using remote sensing and geographic data, are often explained by specifying a micro-level mechanism. Figure 8.2, based on the work of Coleman (1990), depicts the relations between the macro and micro levels. Macro-level analyses (pathway A) of land use are normally based on empirical techniques, e.g. the analysis of spatial patterns of land use derived from remote sensing. Pathway B explains the underlying processes from which the different land use patterns have emerged, e.g. the individual decisions in response to the (changing) socio-economic and physical context. Aggregated, these individual decisions lead to changes in land use pattern that can be analyzed in the more macro-scale analysis. This aggregation may not be straightforward due to non-linear relationships causing the 'ecological fallacy' or 'modifiable area unit problem' (Easterling 1997; Marceau and Hay 1999).
Fig. 8.2 Illustration of the relations between macro and micro-level analysis of land change (Based on Coleman 1990)
These terms relate to the bias that is introduced when non-linear relations at individual level are applied to aggregate data. Also, interactions between agents,
e.g. leading to collective behavior, as well as the role of institutions and other 'collective' agents lead to aggregate results that deviate from the sum of individual decisions (Gibson et al. 2000; Liu et al. 2007). Tools have been developed to analyze the role of processes across multiple scales, e.g. multi-level statistics (Neumann et al. 2011; Overmars and Verburg 2006; Pan and Bilsborrow 2005) and agent-based models, that model the emergence of patterns from individual decision making (Parker et al. 2008). Still, the importance of scalar dynamics in analyzing human-environment interactions is still frequently overlooked.
|< Prev||CONTENTS||Next >|