Global change and the intensification of agriculture in the tropics

zAbstract

Bridging understanding of local environmental change with regional and global patterns of land-use and land-cover change (LUCC) remains a key goal and challenge for our understanding of global environmental change. This meta-analysis attempts to bridge local and regional scales of LUCC by demonstrating the ways in which previously published case studies can be compared and used for a broader regional synthesis in the tropics. In addition to providing results from a meta-analysis, this paper suggests ways to make future case studies more widely comparable.

Introduction

Human modification of the surface of the Earth is so substantial that some claim that the entire terrestrial surface of the Earth is altered by human action (Meyer and Turner, 1994; Redman, 1999). Indeed, our impact is so profound that we may have entered what Crutzen and Stoermer (2001) term “the anthropocene”. While uncertainty surrounds the exact ramifications of these changes, the central role of human activity in altering the planet's biophysical and biogeochemical processes is unquestioned by all but the staunchest skeptics. Alongside the combustion of fossil fuels and other industrial processes involving the release of gases to the atmosphere, the alteration of land cover—the outermost covering of the terrestrial surface, including vegetated as well as barren and frozen surfaces—ranks among the most important human impacts on the planet (Raven, 2002; Steffen et al., 2004; Turner et al., 1990).

A multi-year, multi-investigator effort, the international Land-Use and Land-Cover Change (LUCC) Project studies the alteration of the surface of the Earth. This paper furthers the research agenda LUCC¹ by describing the comparative analysis of 91 case studies of agricultural change in Latin America, Sub-Saharan Africa, and South and Southeast Asia. At the outset of research we hoped that by analyzing a wide range of peer-reviewed case studies from the tropics we would find trends that described the pathways to intensification in agriculture. Such an analysis promises to be important to advancing agricultural change theory, and to contribute to the construction of more realistic models of global land dynamics. The article begins with a description of the methods employed, and proceeds to describe the range of outcomes analyzed, and the causal and mediating conditions associated with those outcomes. It concludes with a discussion of the implications for land change science (Rindfuss et al., 2004), and suggests recommendations for the conduct of future case studies of agricultural change. This study constitutes a companion to Geist and Lambin, 2001, Geist and Lambin, 2002, Geist and Lambin, 2004.

Understanding the ways that land-cover changes through altered land use—the manner in which the biophysical attributes of the land are manipulated—has long been a question in geography, agricultural economics, archaeology, and related disciplines (Butzer, 1982; Redman, 1999; Turner, 2002). LUCC studies are increasingly important as global change science recognizes that human and biogeochemical cycles are closely linked (Intergovernmental Panel on Climate Change (IPCC), 2001). Agricultural change is a crucial part of the Earth system by virtue of the sheer magnitude of agricultural land use. Ramankutty and Foley (1999; Ramankutty et al., 2002) estimated that almost 20% of the potentially vegetated surface of the Earth (about 18 million km²) was under agricultural use as of 1992. While a doubling of agricultural production was accomplished during the past 35 years, another will likely be required over the next four or five decades to meet increasing human demands for food and fiber before population growth levels off with current agro-technologies (Alexandratos, 1995; Tilman, 1999).

Rather than more land under cultivation, the increase in food production will come from intensified agriculture. Agricultural intensification is a biologically important process because it involves the alteration of plants and animals from dependence on states and flows not managed by people to those managed by people. That is, increased inputs per unit area or time supplant natural processes of nutrient and biological regeneration. Conservation of biological diversity, hybridization of local seed varieties, and non-economic species eradication may all occur in the process of intensification (Mittermeier et al., 1998).

The hydrosphere is significantly affected by agricultural intensification. The diversion of streams and impoundment of surface flow alter hydrological systems to the degree that they can influence stream channel morphology, downstream vegetation and water supply, and evapotranspiration (Graf, 2001; L’vovich and White, 1990). Soil conditions also experience alteration under intensifying agricultural systems. Erosion, nutrient depletion, and carbon loss may be engendered by intensification (Altieri, 1995). Finally, agricultural intensification is often characterized by the application of agrochemicals and powered by petroleum-driven machines, with implications for the cycles of key biogeochemical elements, including the carbon, nitrogen, and phosphorous cycles. In addition to the surface aspects of land alteration, increasing evidence links Earth surface change with climate change (Watson et al., 2001). Focusing on land-cover dynamics, different processes (deforestation, desertification, urbanization, agricultural intensification) have very different implications for the structure and functioning of local ecosystems and operate quite differently depending on local social, economic, and political contexts.

Earth observation technology provides information on land-cover dynamics with increasing precision, accuracy, and frequency, but the ability to discriminate between types of land cover over broad areas is still quite rudimentary and subject to a suite of error sources (Duckham et al., 2003). Understanding the nature of these dynamics—describing the underlying land-use practices, and explaining the factors that give rise to them—poses a crucial, challenge to the advancement of global change science. Only by understanding the factors that shape land use can reasonable policies be set to influence future changes.

It is very tempting to associate the increasing human impact on the Earth over the past several centuries with the dramatic increase in human population to arrive at a simple causal mechanism, wherein population equals impact. This simple relationship (i.e., population=environmental impact) is modified to incorporate both the affluence (A) of the population (P) in question and the technology (T) with which people affect those impacts (I). The so-called IPAT or ImPACT identity (Ehrlich et al., 1993; Kates et al., 2001; Waggoner and Ausubel, 2002) offers an elegant and seemingly comprehensive identification of “driving forces” that apply at broad scales, but both sides of the equation become increasingly unsatisfactory as the scale of analysis narrows. At the regional and local levels, studies repeatedly demonstrate that a suite of variables—institutional, economic, biophysical, and cultural—are as important as the universal concepts embodied in IPAT. Recent attempts at operationalizing the variables in question do not lessen the difficulty of answering the fundamental human question: “What is, and what ought to be our relationship to the natural world?” (Kates, 1987, p. 532).

Ideally, these foundational human–environment questions could be addressed through “natural” experiments. These experiments would study a number of places in a standardized fashion, holding certain factors constant and observing the effect of variations in other key factors. Indeed, some of the most valuable studies of land-cover change processes have taken this approach (Kasperson et al., 1995; National Research Council (NRC), 1999; Turner et al., 1990). But this approach is costly and slow moving and, therefore, generally involves a small number of places. In addition, the important academic questions of the period in which the studies were commissioned feature prominently while neglecting sufficient assessment of other potentially important variables. In the commissioned studies noted, e.g., the importance of gender dynamics or ethnicity was not addressed—deemed important in later studies (Rocheleau et al., 1996)—while demography and market variables received excellent treatment. Gender dynamics are highlighted to demonstrate that different academic generations value at times distinct explanatory variables.

Other explanatory frameworks employ rich theoretical backgrounds such as common property, market, or policy factors that seem to highlight the causes of change at particular scales and at points in space and history. When scales of analysis change, however, these explanations prove inadequate, leading some scholars to revert to idiographic conclusions about their particular study. That is, the nuanced manner in which change occurs may lead to multiple studies that embrace radically different explanations of agricultural change, vexing attempts to develop broad conclusions. In this study, we hoped that comparing cases across theoretical and geographic ranges would reduce the nearly infinite causes of change to a handful of factors and that we could reduce, and identify necessary and sufficient conditions for change. As we demonstrate in this paper, the major variables supplied to the global change community as factors of change do play an important role in what researchers deem as important.

Given the difficulty of commissioning studies, other approaches to developing regional analyses are being undertaken. An approach implemented by various participants in global change research involves the examination of the corpus of case studies published by anthropologists, economists, geographers, political scientists, and other social scientists interested in land-use practices (Geist and Lambin, 2001; McConnell and Keys, 2005; Rudel, 2005; Angelsen and Kaimowitz, 1999; Rudel and Roper, 1996; Moran, 1995). This coordinated effort at the comparative, meta-analysis of case studies yields new insights and improved understanding of land-use and land-cover dynamics in the tropics, where important change occurs. Recent meta-analyses of tropical deforestation, e.g., elucidated a range of variables that operate along with, and sometimes in place of, population growth, to shape the patterns and rates of this key land-cover change (Angelsen and Kaimowitz, 1999; Geist and Lambin, 2001; Rudel and Roper, 1996). Both Rudel and Roper (1996) and Geist and Lambin (2001) noted that the causes of deforestation proved complex as the factors varied by case study.