Cover Image

PAPERBACK
$48.00



View/Hide Left Panel

1
The State of Knowledge

Intellectual debate about the relationships between human population dynamics and natural resources goes back at least 200 years (Malthus, 1798, 1803; Lloyd, 1833). The most intense focus was on human demands on land because of the simple Malthusian argument that population growth would eventually outstrip the productive capacity of lands. Analysis of population–environment relationships became broader in the second half of the twentieth century, when recognition became widespread that human activity posed major environmental threats not only through land use, which among other effects can degrade the food-producing capacity of lands, but also through pollution resulting from industrial activities that supported economic growth. Among the landmarks in this broadening of focus were the arguments raised in the early 1970s by Ehrlich and colleagues (Ehrlich and Ehrlich, 1970; Ehrlich and Holdren, 1971) and in the “limits to growth” models of Meadows and colleagues (1972).

This newer thinking was broader not only in the range of environmental effects linked to population growth but also in its attention to the relationships among population growth and other factors, such as economic growth, technological development, and change in human institutions. A famous formulation of these relationships is the I = PAT or Kaya identity, which defined environmental impact as the product of total population (P); economic output per capita, or affluence (A); and all other human activities per unit of output (symbolized by T, for technology) (Holdren and Ehrlich, 1974). This identity became both controversial and potentially useful analytically when used as a model of the social forces that cause environmental degradation, or what later came to be called driving



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 7
Population, Land Use, and Environment: Research Directions 1 The State of Knowledge Intellectual debate about the relationships between human population dynamics and natural resources goes back at least 200 years (Malthus, 1798, 1803; Lloyd, 1833). The most intense focus was on human demands on land because of the simple Malthusian argument that population growth would eventually outstrip the productive capacity of lands. Analysis of population–environment relationships became broader in the second half of the twentieth century, when recognition became widespread that human activity posed major environmental threats not only through land use, which among other effects can degrade the food-producing capacity of lands, but also through pollution resulting from industrial activities that supported economic growth. Among the landmarks in this broadening of focus were the arguments raised in the early 1970s by Ehrlich and colleagues (Ehrlich and Ehrlich, 1970; Ehrlich and Holdren, 1971) and in the “limits to growth” models of Meadows and colleagues (1972). This newer thinking was broader not only in the range of environmental effects linked to population growth but also in its attention to the relationships among population growth and other factors, such as economic growth, technological development, and change in human institutions. A famous formulation of these relationships is the I = PAT or Kaya identity, which defined environmental impact as the product of total population (P); economic output per capita, or affluence (A); and all other human activities per unit of output (symbolized by T, for technology) (Holdren and Ehrlich, 1974). This identity became both controversial and potentially useful analytically when used as a model of the social forces that cause environmental degradation, or what later came to be called driving

OCR for page 7
Population, Land Use, and Environment: Research Directions forces of environmental change (National Research Council, 1990, 1992; for reviews of work using the equation analytically, see Dietz and Rosa, 1994; Chertow, 2001). Debates about the relative importance of population vis-à-vis other driving forces, about whether the effects of population growth are always negative, and about whether effects are uniform across settings were common in the 1970s and 1980s, but they have since largely receded from the scientific literature. Researchers’ interests have focused increasingly on understanding the driving forces, including not only population, affluence, and technology, but also human values, social institutions, public policies, and more; their effects and interactions; the mechanisms by which they affect environmental outcomes; and feedbacks from environmental conditions to human activity. Just as demographic driving forces may lead to environmental outcomes in various ways, environmental conditions and changes may also influence population size, structure, and change. By the early 1990s, human influences on the natural environment were understood to occur through two main processes: change in land cover and land use (including use of waters) and “industrial metabolism,” that is, the transformation of materials and energy for industrial production and economic consumption (National Research Council, 1988, 1990). Population and other driving forces operate through these processes to generate specific outcomes that act as proximate causes of environmental change (National Research Council, 1992). Proximate causes tied to land use and land cover change include conversion of forests into agricultural lands, of farm fields and pastures to urban uses, and of wetlands to agricultural or urban uses, all of which transform habitats for nonhuman species and alter biogeochemical cycles. Industrial metabolism generates other proximate causes, including releases of phosphates and heavy metals into waterways and of nitrogen, sulfur, and carbon oxides into the atmosphere. Accordingly, research on population-environment relationships has been dominated by two rather distinct streams of work: one focusing on effects mediated by changes in land use or land cover, and another focusing on effects mediated by materials and energy transformations. Efforts to integrate both kinds of human influence have been far less common (e.g., Lutz, 1994; Curran et al., 2002; York, Rosa, and Dietz, 2003; Rosa, York, and Dietz, 2003). SCOPE OF THIS BOOK The limited resources available for this study have led us to select only part of this large field for coverage. This volume focuses on research in which change in land use or land cover is a key mediator of human–environment interactions, in which demographic variables figure prominently among the driving forces investigated, and in which efforts are made

OCR for page 7
Population, Land Use, and Environment: Research Directions to investigate the causal mechanisms by which human population changes affect land use and environmental outcomes. Clearly, this is only part of a large and complex picture in which institutions, public policies, market conditions, technological changes, and other factors also figure. In the conceptual framework developed for the Millennium Ecosystem Assessment, for instance, changes in land use and land cover figure among the “direct drivers” affecting ecosystem services, demographic factors among the “indirect drivers” (Alcamo, Bennett, and the Millennium Ecosystem Assessment Project, 2003). The chain of causality leading from population to land use to environmental effect that is of primary interest in this volume is embedded in the Millennium Ecosystem Assessment framework, although the main focus of that assessment is on ecosystems services and human well-being rather than any particular strand of interconnected effects. Population–land use–environment relationships have great environmental significance, both locally and globally. Human activity has changed the face of the Earth, particularly in the past few hundred years, with major environmental consequences. About 50 percent of the Earth’s land surface has been transformed by direct human action, mainly for farming, pasturing, and forestry, and also for industry, urban development, and transport (Turner et al., 1990; Steffen et al., 2004). More than half of all accessible fresh water is used directly or indirectly by humankind, and ancient and often nonrenewable underground water resources are being depleted rapidly in many areas (Gleick, 1999, 2003). Human use of land and fresh water has transformed global precipitation regimes and altered ecosystems worldwide, reducing the diversity of the world’s biota and affecting the overall ability of the biosphere to sustain life (Steffen et al., 2004; Vitousek et al., 1997a). Intensification and diversification of land use, together with advances in technology, have also led to rapid changes in the global cycles of carbon, nitrogen, and other critical elements (Melillo, Field, and Moldan, 2002). More nitrogen is now fixed synthetically and applied as fertilizers in agriculture than is fixed naturally in all terrestrial ecosystems, resulting in fundamental ecological changes in lakes and rivers and in leaching of other nutrients from soils (Vitousek et al., 1997b). Land use change, mainly through its effects on the global carbon and nitrogen cycles, is also responsible for a significant proportion of the phenomenon of global climatic change (National Research Council, 1992). Demographic factors, including population growth, density, fertility, mortality, and the age and sex composition of households, are known to be important influences on land use and land cover change. Human migration, including shifts from rural to urban areas, movements between countries for economic or political reasons, and large-scale planned resettlements, as in Amazônia and Indonesia, also significantly affect land cover and land use. Although overall population numbers are sometimes strongly related

OCR for page 7
Population, Land Use, and Environment: Research Directions to land cover changes, such as deforestation (Allen and Barnes, 1985), recent research shows that this overall relationship depends on many factors, including land settlement policies and market forces (Geist and Lambin, 2002), cultural and institutional factors, and characteristics of the biophysical environment itself. In other words, the impacts of demographic factors need to be understood in the context of other drivers of land use and land cover change. Moreover, some of these factors affecting land use and the environment also influence demographic variables. For example, land tenure can affect fertility at the household level, with more secure tenure (all else being equal) resulting in lower fertility rates (Moran, 1993; Bilsborrow, 1994). Thus, the relationships among demographic changes, changes in land use and management, and the states, properties, and functions of environmental systems are complex (Alcamo et al., 2003; Turner et al., 2003a). They are also matters of scientific and practical interest (e.g., Kates et al., 2001; National Research Council, 1999a, 1999b). The land use branch of population-environment research has received more attention in recent years than the industrial metabolism branch (Pebley 1998).1 Over the past decade, catalyzed in part by an international land use/land cover research program (Turner et al., 1995; Lambin et al., 1999) and organized programs of research support from the National Institute of Child Health and Human Development (NICHD), the John D. and Catherine T. MacArthur Foundation, the William and Flora Hewlett Foundation, the National Aeronautic and Space Administration (NASA), and some other sources, significant progress has been made in understanding the land use connection between population and environment. Site-based studies, in particular, have flourished over the past decade. These studies, as well as the broader field, have set the foundation for even more progress to be made. The connections between population, land use, and environment have been identified as an area in which significant research gains are expected over the next decade (National Research Council, 1999a, 2002). Much of the recent research devoted to identifying causal mechanisms and processes has examined changes in population characteristics, land uses, and environmental conditions over time at specific sites. This volume 1   One reason for limited research attention to processes linking demographic and other human variables to environmental consequences through production and consumption of industrial products has been the absence of targeted funding for this research in the United States. Industrial metabolism has become a focus of interest in the engineering field (see, e.g., Williams, Larson, and Ross, 1987; National Academy of Engineering, 1994; Graedel and Allenby, 1995; National Research Council, 2004b), but much less has been done to integrate social scientific analysis (however, see, e.g., National Research Council, 1984, 1997; Grossman and Krueger, 1995; York, Rosa, and Dietz, 2003; Rosa, York, and Dietz, 2003).

OCR for page 7
Population, Land Use, and Environment: Research Directions focuses heavily, although not exclusively, on these site-based studies, especially those integrating multiple methodologies and perspectives. Research that fully incorporates population and land use and environment must cross disciplines—indeed, it must cross the social and natural sciences. Focusing on a specific site is one approach to achieving this integration. In addition, site-specific studies have been a particular emphasis of research supported by the population-environment programs of the sponsors of this study. For these reasons, we have looked closely at site-specific studies and also other studies that have tried to clarify the mechanisms linking demographic variables, land use change, and environmental outcomes. The notion of demographic factors as driving forces suggests one-way causation, and indeed the primary emphasis in many studies has been on the effects of human activities on environmental variables, including research undertaken as part of the Intergovernmental Panel on Climate Change (Fischer and O’Neill, Chapter 3) and the Millennium Ecosystem Assessment (Alcamo et al., 2003). Feedbacks from environmental variables to demographic and land use change are also important, although less studied. The panel organized a workshop in Irvine, California, on January 14-15, 2004, at which we discussed the insights from research on population–land use–environment interactions and the challenges of conducting this kind of research with researchers in the field. We invited representatives of several research programs that included both natural scientists and social scientists to prepare papers summarizing the progress of their programs and the challenges they confronted in doing their research. Revised versions of the papers appear as appendixes to this volume. We selected researchers who would provide a variety of intellectual perspectives; diversity in focus by world region, level of economic development, and rural or urban setting; and representation of individuals whose initial research questions were primarily environmental and others whose initial questions were primarily demographic. Most of the research discussed at the workshop was site-focused, but also discussed global-level modeling issues to enhance discussions of issues of scale, modeling, and causal analysis. This volume takes stock of the progress that has been made in understanding population–land use–environment linkages and causal mechanisms to see what has been learned, to identify gaps and problems that remain, and to develop a set of recommendations for future research. It is worth noting that we have emphasized scientific criteria for making our recommendations rather than criteria of practical importance. Thus, we have recommended research directions that would strengthen knowledge of the ways that demographic and land use variables affect environmental conditions, but we have not offered judgments on which environmental conditions are most worthy of this kind of analysis or which areas of research are most likely to yield results of practical importance in the near term. Our

OCR for page 7
Population, Land Use, and Environment: Research Directions review and recommendations complement other efforts, such as the Millennium Ecosystem Assessment, for which the goal is to inform the decisions of policy makers as they might affect ecosystems (Alcamo et al., 2003:12). This chapter provides a synthetic overview of progress in population–land use-environment research over the past decade, showing the current state of knowledge, the areas of rapid development, and the major challenges for research. Chapter 2 presents our conclusions and recommendations for further development of research on the population–land use–environment relationships that are our focus. The main audiences for our recommendations are researchers working in the field and the sponsors of their research. We have not made recommendations for research on other questions of human–environment interaction. In particular, we have not offered recommendations for research on population–environment relationships that are mediated by industrial production or consumption, despite the obvious importance of the topic, nor for research on the connections between land use and industrial metabolism as they affect population–environment relationships. AREAS OF RESEARCH PROGRESS A useful reference point for the present work is a volume published a little more than a decade ago, Population and Land Use in Developing Countries (National Research Council, 1993), which summarized the state of science at that time. It addressed one major question: What are the effects of population growth on land use change? It focused on one aspect of human demography (population growth) and concentrated on a subset of land use changes associated with expansion of agriculture. The potential environmental consequences of land use change, a strong motivator for the workshop that produced the volume, were addressed by some of the participants. At the end of the workshop, the participants identified two key areas in which research was needed (Jolly and Torrey, 1993). One was to develop better and more detailed data, over time, on population and land use variables. The other was for case studies that clearly analyze the roles of the various factors, such as property rights institutions, market conditions, and soil and climate characteristics, that condition population–land use relationships. It was expected that such studies could explain why population growth does not have uniform effects on the land and help develop a causal understanding of population–land use relationships. Research since then has made progress in these and other directions (e.g., O’Neill, MacKellar, and Lutz, 2001; Lutz, Prskawetz, and Sanderson, 2002). In the process, it has opened up additional areas of inquiry.

OCR for page 7
Population, Land Use, and Environment: Research Directions Unpacking Concepts A major area of progress over the past decade has been in adding nuance to earlier arguments about population–environment relationships. Some of this research is “unpacking” or decomposing broad concepts of population, land use, and environment into more refined elements and thereby producing deeper understandings. Knowing that population size, density, or growth is associated with land use change and environmental change is just the beginning of analysis (e.g., Fischer and O’Neill, Chapter 3). For example, population growth might be caused by natural increase, a surplus of births over deaths, or migration. The former was a major focus of research for many years, in part due to high fertility in many parts of the world combined with dramatic decreases in death rates after World War II. Indeed, much of the early work was at the national level, where natural increase dominates as a component of population growth (e.g., Allen and Barnes, 1985). More recent work has focused subnationally, where the role of migration in population change is likely to be particularly pronounced. Researchers investigated a series of hypotheses regarding effects of fertility on land use in rural areas dependent on subsistence agriculture: that larger families require more food in order to subsist when children are young, have more labor with which to cultivate land when children are older but still living in the home household, and require more land when children grow up and need a place of their own to farm (i.e., the developmental cycle of domestic groups; see Goody, 1958, 1976). Households may adjust to increasing family size and subsistence needs in a variety of ways (Bilsborrow, 1987; Davis, 1963), including enlarging the area under cultivation (e.g., Cruz, 1996; Mortimore, 1993; Paulson, 1994; Pichon, 1993, 1997; Pichon and Bilsborrow, 1999; Umezaki et al., 2000) or putting pressure on lands held in common (e.g., Axinn, Barbar, and Biddlecom, 2002; Foster et al., 2003, Shivakoti et al., 1999). They may also adjust by intensifying land use. Societies dependent on swidden agriculture may increase frequency of cultivation, thereby reducing fallow time. Studies found a correlation between fertility-driven high rates of population growth and shortened periods of fallow (Saikia, 1998; Umezaki et al., 2000). Increasing frequency of cultivation was of particular interest to Chayanov (1966) and to Boserup (1965), who first developed a theory of intensification. Intensification may occur in many ways, including multiple cropping; use of fertilizers, pesticides, herbicides, and high-productivity seeds; and improved irrigation. A notable feature of recent research is increased analytic attention to migration as a component of population growth and change. Migrations can have a substantial impact on land use (United Nations Population

OCR for page 7
Population, Land Use, and Environment: Research Directions Fund, 2001). These effects arise not only from relatively permanent migrations to frontiers (e.g., Gutmann et al., Chapter 4; Moran, Brondizio, and VanWey, Chapter 5; Walsh et al., Chapter 6) or to urbanizing areas (Redman, Chapter 7; Seto, Chapter 8), but also from shorter term and more temporary movements. Examples of the latter include “floating workers” in China (Seto, Chapter 8), seasonal and circular migrants in Thailand (Walsh et al., Chapter 6), tourists (Liu et al., Chapter 9), and commuters. There is also the issue of possible environmental effects of policy-driven migrations, often aimed at replacing indigenous peoples, as in forced migrations or population dilutions of such peoples in North America in the nineteenth century and in the Soviet Union, Tibet, and elsewhere in the twentieth. These policy interventions tend to replace adapted indigenous land use systems with ones imported from other ecological zones. On long time scales, migration can have substantial impacts through land use on the environment (Gutmann et al., Chapter 4). Recent studies are showing how a more nuanced treatment of migration gives better understanding than simple analyses of population numbers. Research has also expanded to include population effects on coastal and marine as well as terrestrial ecosystems. Demographic changes are increasingly affecting the environment through human migration to coastal areas. During the past 50 years, there has been a dramatic increase in human migration to coastal zones, related urbanization of many coastal areas, increased linkages between coastal and inland populations, and a transformation in the use and management of coastal and marine resources. A recent set of studies completed under the John D. and Catherine T. MacArthur Foundation Program on Population, Consumption, and Environment focused on these ecosystems (Ambio Vol. 31, Number 4, 2002). Migration as a driving factor has been a major focus of this research, which has also addressed governance institutions for such common pool resources as productive estuaries, as well as policy interventions, for example to promote shrimp aquaculture (McCay and Acheson, 1987; Ostrom, 1990; National Research Council, 2002; Matson et al., Chapter 10; Seto, Chapter 8). Another shift in research focus is increasing attention to households. Changes in fertility, mortality, marriage, and migration have produced smaller households in many parts of the world. The detailed impact of household and family dynamics on land use and the environment is increasingly well understood (Galvin et al., 2002; Moran et al., Chapter 5; Walsh et al., Chapter 6; Liu et al., Chapter 9). In some contexts, growth in the number of households is at least as important as growth in the number of persons (MacKellar et al., 1995; Liu et al., 2003). There is some evidence from various sites that the number of households and their composition in terms of family cycle or life course significantly influences land use and

OCR for page 7
Population, Land Use, and Environment: Research Directions environmental change (Moran et al., Chapter 5; Walsh et al., Chapter 6). The more complete the understanding of households and their dynamics, the better one will understand and be able to predict the associated change in land use and the environment. Unpacking the concept of land use has also been productive. Land use and land cover are typically organized into a small number of major classes (e.g., urban or rural; forest, cropland, desert, tundra, water; permeable or impermeable surface). Finer distinctions can be very informative, however. For example, uses of space may differ greatly, even in seemingly uniform urban or rural settings (Redman, Chapter 7). Areas classified as uniformly urban in fact have widely differing intensities of population and land use within them (Grove et al., 2004; Pickett et al., 2001; Redman, Chapter 7; Weeks, Larson, and Fugate, Chapter 11). A given area revealed in remote-sensing data as containing little or no permeable surface might consist mainly of low-density uses, such as single-family or small multifamily dwellings, or it might be covered by high-density housing or nonresidential office, manufacturing, service, or commercial facilities. Each of these land uses has its own signature in terms of population composition and environmental impacts. Similarly, an area that appears as uniformly rural may include dispersed or clustered settlement; also, it may have large areas of undifferentiated habitat or be divided up into a multitude of tiny patches. These differences can be closely tied to both population and environmental variables (Kaufman and Marsh, 1997). For instance, tree cover may be primary forest or secondary forest (Moran and Ostrom, 2005), with different implications for biodiversity loss. The variation in land uses within major classes suggests the importance of developing similarly nuanced classification systems for remote imaging based on efforts to identify spectral signatures for land use types that correspond to systematic variations observed on the ground. In much of the early research on population and land use, little attention was given to environmental outcomes beyond those involving land cover. Recent research is beginning to specify environmental variables in much more detail. For example, in this volume, Matson and colleagues (Chapter 10) describe a complex study of two interconnected ecosystems in the Yaqui Valley (in Sonora State, Mexico), one an inland wheat-growing region and the other a coastal area devoted to shrimp aquaculture. The project is particularly strong in documenting the effects of land use change on the environment, for instance, the consequences of agricultural intensification in the form of increasing fertilizer inputs for the functioning of both the surface water systems draining the valley and for the coastal ecosystems of the Sea of Cortez. The study makes clear that population growth is both cause and consequence of land use change, but the population–land use connection is not developed in detail.

OCR for page 7
Population, Land Use, and Environment: Research Directions The Yaqui Valley study also illustrates the added value of decomposing broadly defined population, land use, and environment variables into more refined ones. This study was able to identify fertilizer runoff as a key mechanism driving environmental impacts. For anyone concerned with alleviating these impacts, the study suggests that it might be more productive to address the issue of fertilizer application directly than to focus on birth rates, human settlement patterns, or even rates of land conversion to agriculture. Such conclusions can be only tentative, though, in the absence of research that considers the roles of all these factors together. At present, research that completely bridges population, land use, and environment is rare. Most studies have focused either on connections between population and land use, with environment relatively deemphasized, or on connections between land use and environment, with population relatively deemphasized. This is true of all of the studies included in this volume. Site-Based Studies Researchers have responded to the call in Population and Land Use in Developing Countries (National Research Council, 1993) for better and more detailed data by conducting detailed and sustained studies at many sites around the world. As the authorized chapters in this volume illustrate, such studies have been a major source of scientific progress over the past decade. They have carefully collected and analyzed multilevel, longitudinal, and spatially explicit data on population, land use, and environmental processes for specific localities, large and small (Galvin et al., 2002; Gutmann et al., Chapter 4; Matson et al., Chapter 10; Moran et al., 1994, 2003, Chapter 5; Walsh et al., Chapter 6; Redman, Chapter 7; Thornton et al., 2003; Turner, Geoghegan, and Foster, 2004). Over time, as the breadth, depth, and detail of these studies has grown, the study sites have come to represent “microcosms” (Matson et al., Chapter 10) and “laboratories” (Entwisle et al., 1998) for the interdisciplinary study of social and environmental change. The addition of new data over time as projects have extended and expanded their purviews has greatly enhanced the value of the overall data sets. Although site-specific studies have become much more common, the studies are uneven in terms of the world regions, types of population and land use changes, and institutional and environmental contexts they cover. In terms of regional representation, the studies represent Asian and Latin American settings better than African and Middle Eastern ones, although of course there are exceptions (e.g., Weeks et al., Chapter 11; Tiffen, Mortimore, and Gichuki, 1994; Turner, Hyden, and Kates, 1993; Geist and Lambin, 2002; Laney, 2002; Galvin et al., 2002). Also, although the bal-

OCR for page 7
Population, Land Use, and Environment: Research Directions ance is shifting, it remains the case that rural settings are better represented than urban and urbanizing ones, shifts from one type of land use to another more frequently dealt with than changes within types, and terrestrial ecosystems more than coastal and marine ones. Scale Site-based studies vary considerably in their scale. The concept of scale can be confusing because it has at least two distinct meanings. In one meaning, scale is a study’s coverage or extent. A study of the U.S. Great Plains over more than a century (e.g., Gutmann et al., Chapter 4) has a broader spatial and temporal scale in this sense than a study of metropolitan Phoenix over 30 years (Redman, Chapter 7). In another meaning, scale corresponds to the unit of analysis or level of organization. In this meaning, studies that analyze data on countries or major subnational political divisions (e.g., Fischer and O’Neill, Chapter 3; Gutmann et al., Chapter 4) work at a larger scale than studies that use data on individuals or households (e.g., Moran, et al., Chapter 5; Walsh et al., Chapter 6). Often, these two meanings of scale go together. Thus, studies at local levels frequently involve research on small decision-making units, such as individuals or households, and small land units, such as plots. Regional and national studies are often concerned with larger units, such as counties. Studies of individuals or households typically cover relatively short temporal scales, often 5 or 10 years, at most 20 or 30 years. Studies of counties, states, or provinces may cover a century or more (Gutmann et al., Chapter 4). Increasingly, studies are departing from the norm in which coverage matches units of analysis. For instance, there are studies of individuals and households that are national in scope (Foster, Chapter 12). There are studies of pixels covering fractions of a square kilometer (km2) that track seasonal and annual variation in land cover over long periods of time (Walsh et al., Chapter 6; Redman, Chapter 7) and of villages that extend back at annual time steps 50 years or more (Entwisle et al., 2004). The studies in this volume indicate that current research represents a broad range in terms of spatial coverage and level of organization. They range from intensive studies of small sites such as Nang Rong district in Thailand (1,300 km2), the Wolong Nature Reserve in China (2,000 km2), and the development areas along the road in Altimira in the Brazilian Amazon (3,800 km2); to larger ecologically defined areas such as the Yaqui Valley in Mexico (23,500 km2 in the irrigated area), the Red River Delta in Vietnam (16,000 km2), and the Pearl River Delta in China (26,000 km2); to large metropolitan areas (Cairo, Phoenix); to very large ecologically defined areas (the U.S. Great Plains); to countries (India, China); to the entire world. As studies of broader coverage begin to use data at finer levels of

OCR for page 7
Population, Land Use, and Environment: Research Directions cover change in the Yaqui Valley, including the impact of climate (e.g., a recent prolonged drought), the consequences of an earlier white fly infestation, and the implications of policy change in Mexico (e.g., establishment of the ejido system of land reform under the Mexican constitution) and internationally (the North American Free Trade Agreement). Social institutions, particularly institutions for governing human–environment interactions, have long been recognized as important mediating factors between individual human behavior and larger scale environmental systems. Their mediating role is partly dependent on the size and dynamics of the human populations using particular environmental systems (e.g., Blaikie and Brookfield, 1987; Wade, 1988; Ostrom, 1990; Baland and Platteau, 1996; National Research Council, 2002). Issues of scale dependence and cross-scale interaction have been described in other work on human–environment interactions (e.g., Vayda, 1983; Palloni, 1994; Gibson, Ostrom, and Ahn, 1998; Young, 2002; Alcamo et al., 2003). Research on population–land use–environment relationships provides many good test beds for understanding these issues. When key aspects of context change, especially when they change dramatically, there are opportunities to study their impact in a site-based study. When they do not change, however, it is not possible to understand such contextual effects through studies conducted in a single site. Comparisons across studies can begin to reveal them, if the data are sufficiently comparable (examples of comparative analysis can be found in Moran and Ostrom, 2005). Linking Social and Environmental Data Research on population–land use–environment relationships depends on integration of social and environmental features at the appropriate social, spatial, and temporal scales, and equally important, appropriate linkages of processes that cross sectoral components of the coupled human–environment system. Such integration has been difficult, although progress has been significant during the past decade (National Research Council, 1998; Fox et al., 2003; Moran and Ostrom, 2005). Much progress has been made possible through the use of GIS tools, in which data on population, land use, and environment can be connected by coding them spatially. However, there is no definitive one-to-one link among population units, land use units, and relevant environmental units. Over the past decade, analysts have made progress linking population and land units, as well as land units and environmental units, in some contexts. For example, households might be linked to field plots in areas where household impacts on land use are mainly through the plots they farm (Moran et al., Chapter 5; Walsh et al., Chapter 6). Villages can be

OCR for page 7
Population, Land Use, and Environment: Research Directions readily linked to village territories (Walsh et al., Chapter 6; Foster, Chapter 12), as can larger administrative units, such as counties be linked to their territories (Fischer and O’Neill, Chapter 3; Gutmann et al., Chapter 4), although an administratively defined territory may or may not correspond to the land owned, used, or linked in some other way to the individuals and households that live in those units. With effort, pixels can be linked to landscapes, watersheds, or ecosystems, and to political boundaries (Redman, Chapter 7; Seto, Chapter 8; Matson et al., Chapter 10). There is no straightforward cross-step all the way across the population–land use–environment nexus, however. One difficulty arises when multiple units in each domain are relevant to multiple units in other domains, such as when a household includes members who work nearby lands and other members who have temporarily migrated to find work. The challenge of linking people to land and to environmental conditions is particularly difficult when the populations responsible for changes in land and environment are linked to them through global markets, as with hardwoods, cash crops such as coffee, and the products of aquaculture. It is possible to measure environmental changes at the place where they are visible, as in ecological indicator systems, or to attribute them to the places where consumer or industrial demand acts to bring them about, as is done with indicators like the “ecological footprint” (Wackernagel and Rees, 1996; Wackernagel et al., 2002; Parris and Kates, 2003). So far there is no reliable way to do both at once: to locate the environmental footprint of one country or city on the map of another country. Certain land–environment links are similarly difficult to make when land use changes result in nonlinear responses affecting feedbacks though ecosystems or global environmental systems, such as changes in ecosystem state following fire or flood that affect methane emissions, or community or biodiversity changes that alter pathways of carbon sequestration. Data Collection Available data sources do not always provide information at the appropriate resolution for addressing researchers’ questions. For example, decadal censuses provide valuable sources of information about population size and structure, but they are limited in important ways relevant to understanding population–land use–environment interactions. One problem is the decadal level of temporal resolution: important population changes occur over much shorter intervals, as pointed out in several papers in this volume (e.g., Seto, Chapter 8; Matson et al., Chapter 10; Weeks et al., Chapter 11). Decadal censuses do a particularly poor job tracking in- and out-migration, focusing instead on net migration (the difference between in-flow and out-flow). Information on net migration is useful, but low

OCR for page 7
Population, Land Use, and Environment: Research Directions levels of net migration may hide substantial rates of turnover: “For every stream, there is a counterstream” is a truism of migration research. Furthermore, censuses are not good sources of information about seasonal, circular, and other forms of short-term migration, nor do they generally include questions about commuting or other forms of local mobility. Because of these problems, researchers are designing and collecting their own data in the form of multilevel, longitudinal, and spatially explicit surveys designed to address an evolving set of specific questions. Surveys have the capability of measuring change at multiple levels of observation, over short as well as longer periods of time. There are also challenges in measuring land use change and environmental outcomes. For instance, there are fairly well-established methods for measuring land cover from remote observations at the level of major classifications (e.g., forest, agricultural, urban) (Skole et al., 1994; Loveland et al., 1991), but the capability to measure finer differences in land cover (e.g., levels of intensity of agriculture, degrees of urbanization) is still under development. Weeks and colleagues (Chapter 11) propose a satellite-based approach to measuring neighborhood-level variation in urban land cover that combines information about the composition and spatial configuration of pixels within census tracts to describe an urban gradient in Cairo, Egypt. Seto (Chapter 8) describes the inability to find a single model that could classify land cover adequately for several adjacent Landsat scenes in the Red River Delta of Vietnam, as well as the need to analyze each scene individually. Yet through use of intensive field studies, it has been possible to distinguish between stages of secondary succession in the Brazilian Amazon (Moran et al., 1994; Moran and Brondizio, 1998). Another challenge is to make full use of the historical detail in time series available in remote images, which typically have much higher temporal resolution than censuses, land surveys, or social surveys. By analyzing data from multiple sequential images, much can potentially be learned about social and environmental dynamics following a perturbation to the system. Contributors to this volume describe two possible approaches. One is the application of life course principles (Elder, 1998; Rindfuss, 1991) to an analysis of pixel or plot trajectories (McCracken et al., 1999; Walsh et al., Chapter 6; Redman, Chapter 7). The other is the application of econometric methods designed for the analysis of time series of multidimensional social data to a time series of classified images (Seto, Chapter 8). Both approaches incorporate social science methods into the analysis of satellite-based observations, thus illustrating a potential benefit of cross-fertilization of fields. Historical analysis is also possible by quantifying land surface changes, although it is difficult to link changes in an ecosystem to associated biogeochemical and biodiversity characteristics.

OCR for page 7
Population, Land Use, and Environment: Research Directions Comparability of Data Researchers on population–land use–environment relationships typically investigate questions that flow from their interests and the attributes of their sites. They have usually found their own ways to measure the variables of interest because coherent global databases on land use or on the demographic and environmental variables of interest to the research groups do not exist at the relevant levels of analysis. Thus, ensuring comparability of data on the same variables from different sites is an important challenge in developing general knowledge from multiple site-based studies (for an example of an effort to meet these challenges, see National Research Council, 2001b). In principle, such comparable data might come from space platforms, but these data must be analyzed in different ways to address different research questions and to fit the important land use and land cover changes under way in specific contexts. Moreover, the proper interpretation of spectral information in terms of phenomena on the ground is not the same everywhere on Earth (e.g., Seto, Chapter 8; Campbell, 2002). Researchers have been creative in developing their own databases. The difficulty of this work, and also its value, is illustrated in the papers included in this volume. This strategy of data development has advantages and limitations. It has yielded considerable knowledge about population–land use–environment processes at a number of sites around the world and, as projects have evolved over time, the breadth, depth, and detail of the data have evolved as well. However, what has been learned may not be comparable across sites. The issues of contingency and cross-scale interaction have already been noted. Population–land use–environment relationships depend on levels of technology, formal and informal institutions, local, national, and international markets, policies, and the natural environment. These relationships are likely to vary from site to site. In addition, there is no assurance that the same variable, or the same relationship between variables, has precisely the same meaning from one research site to the next. Indeed, samples may be very differently constructed. Some studies begin with land (e.g., Moran et al., 1994, 2003, Chapter 5; Messina and Walsh, 2001), whereas others begin with administrative or social units, such as villages or households (Walsh et al., Chapter 6; Foster, Chapter 12). Starting point matters. If one draws a sample of land units, for example field plots, except under unusual circumstances the owners of those plots will yield a biased sample of households associated with those plots. Landless households will be excluded, as will absentee landlords. Samples based on land units can be generalized to populations of land units but not necessarily to populations of households or other social units. The reverse is also true. If one draws a sample of, for example, villages, land associated with

OCR for page 7
Population, Land Use, and Environment: Research Directions these villages is likely to be a biased sample of all land. Thus, an estimate of forested land based only on land located in and around villages would probably be inaccurate. Places where people have chosen not to settle would be excluded. Methods for sampling land and people simultaneously are under development (e.g., Moran et al., Chapter 5), but until these are in widespread use, it will be important to keep starting point in mind when generalizing the results of any particular study. This fact creates serious problems for building a knowledge base that holds across sites and for clarifying the role of contextual conditions as explanations of observed population–land use–environment relationships. It also makes difficult any effort to address questions of scale dependence and cross-scale interaction. The growing number of detailed and sustained site-based studies represents an unparalleled collection of resources for better understanding population processes, land use change, and environmental determinants and consequences. The value of these resources will not be maximized if only a small core of investigators use the data. One effort to share data between the social and biological sciences appears in the framework adopted by the Long-Term Ecological Research (LTER) Network for core social data (Redman et al., 2004), which prescribes six general areas of social and demographic data to parallel the five core areas of ecological data that the network has long espoused. Another effort, used in the Baltimore Ecosystem Study LTER project, developed several categories of data for characterizing the physical, biological, and socioeconomic components of an ecosystem occupied, built, or managed by people. Important features of this framework include attention to institutional structures, different kinds of capital and infrastructure, and the capacity for self-reflection and learning (Grove and Burch, 1997). In addition, data distribution clearinghouses now exist, such as the Consortium for International Earth Science Information Network and the Inter-University Consortium for Political and Social Research, where scientists can deposit their data and other scientists can access them. If data are to be shared, however, great care must be taken to protect the confidentiality of information about human subjects, who may be at risk. Great scientific value derives from the ability to link social survey data to spatial and environmental coverages in a geographic information system, but socio-spatial links reveal respondents’ local communities, and sometimes their households, of residence (Rindfuss et al., 2002). TOWARD CAUSAL UNDERSTANDING The earlier volume Population and Land Use in Developing Countries (National Research Council, 1993) identified the need to conduct case

OCR for page 7
Population, Land Use, and Environment: Research Directions studies that examine population–land use relationships in relation to markets, property rights regimes, and other factors that may affect these relationships. This approach is a useful way to build the needed causal understanding. Since then, three major kinds of research have been advancing the field in the direction of the needed causal understanding: detailed descriptive studies of change at specific sites, regression-based statistical analyses of change processes, and studies that emphasize mathematical modeling. Research has also begun to examine scale dependencies, cross-scale interactions, thresholds, and feedbacks. These complexities are at the heart of the methodological challenges researchers face.2 Descriptive Studies Careful documentation and descriptive analysis have played a key role in the development of the field over the past decade. Studies that break down population, land use, and environment into their more specific component factors have yielded better evidence of the mechanisms that connect change in each of the three subjects with the others, as well as better understanding of the complex interactions between them (see Cadwallader, 1988). For example, Moran and his colleagues were able to distinguish between the land use practices of households in different settlement cohorts in the Altamira site on the Lower Xingu Basin in the Brazilian Amazon. They linked the developmental stages of households to particular land use strategies: young households focused on production for food needs quickly convert forest for the cultivation of staple crops; as households accumulate capital and labor, they expand into other activities, such as cattle ranching and cash crops; as children become adults, households begin to shift toward productive uses that require less labor. These stages are visible in distinctive trajectories of deforestation and land use that the researchers call the colonist footprint (Brondizio et al., 2002; McCracken et al., 1999; Moran et al., Chapter 5). These observations are an important starting point for explanation, which then would need to consider potential age, period, and cohort effects, various combinations of which could account for the observed patterns. Descriptive research has the capability to expose, or unpack, the assumptions and theoretical structures that researchers in each specialty use and has promoted the integration across disciplines necessary to understand the three-way linkage among population, land use, and environment (Pickett et al., 1999). 2   A useful review of methodological issues in determining the causes of land use change (Rindfuss et al., 2004) appeared in print as this book was being completed.

OCR for page 7
Population, Land Use, and Environment: Research Directions Regression-Based Statistical Analyses Significant progress has been made over the past decade by using statistical techniques of regression analysis on data sets that integrate satellite-based measures and survey data or that integrate population data into spatial data sets. Almost every chapter contributed to this volume describes an analysis that incorporates both satellite and survey or census-based measurement. None of the studies published in the 1993 volume did this. Many of the findings reported in this volume are based on statistical regression analysis. For instance, Foster (Chapter 12) describes a study of forest cover in India. Observing a recent increase in the amount of forest cover in the territories around rural villages, Foster and colleagues (2003) evaluated competing explanations for this trend using a regression approach that related changes in agricultural productivity, population, and wages to changes in forest cover. The analysis led the authors to conclude that the growth in forest around villages was largely driven by demand for wood and paper at the national level coupled with trade barriers that discouraged the importation of these products. This is an example of a cross-scale interaction, in which economic demand and policies enacted at the national level affect relationships within and between villages. Studies based on quantitative statistical analyses have so far stopped short of fully considering population-environment interactions as a complex system, and particularly the feedbacks from environment to population. Virtually always, population has been treated as exogenous to land use change and environmental impact. Yet there is little reason to believe that this assumption is valid. For instance, numerous studies have examined the impact of frontier living on fertility (e.g., Easterlin, 1976; VanLandingham and Hirschman, 2001) and the size of landholdings on number of children born (Genus, 2002; Cain, 1985; Maglad, 1994; Stokes et al., 1986). There is also quite a bit of research on relationships between the environmental characteristics of areas and migration into and out of those areas (e.g., Cruz, 1996; Pichon, 1993, 1997; Amacher et al., 1998; DeWalt and Stonich, 1999), although little of it addresses the effects of environmental factors and degradation on out-migration from origin areas Although not specifically the focus of any of the papers contributed to this volume, population size, structure, and change are consequences as well as causes of land use and environmental attributes. This feedback is incorporated into emerging conceptual models of coupled human-natural systems (e.g., Pickett et al., 1994; Redman et al., 2004; Turner et al., 2003a, 2003b; Parker et al., 2003). The identification problem it poses is a major challenge to the proper specification of statistical models and especially causal inferences based on these models. Identification problems are not unique to statistical analyses (King et al., 1994), as illustrated by the challenge of separating

OCR for page 7
Population, Land Use, and Environment: Research Directions age-period-cohort effects in interpreting trajectories of deforestation and land use described above. Modeling Studies Mathematical modeling has been used in population–environment research at least as far back as the world systems model proposed in The Limits to Growth (Meadows et al., 1972). Renewed interest in mathematical modeling of population–environment relationships has been spurred in part by the difficulty of reconciling the complexity of population, land use, and environment relationships, even in single-site studies, with the simplifications necessary for statistical modeling. Fischer and O’Neill (Chapter 3) discuss the role of population in current global models of land use change and environment, with particular reference to the models used by the Intergovernmental Panel on Climate Change to produce scenarios of greenhouse gas emissions and those used to produce scenarios of changes in future ecosystem goods and services for the Millennium Ecosystem Assessment. The authors conclude that although the global models use a variety of approaches to simulating land use change, population generally serves as a simple scale factor affecting demand for agricultural products on the consumption side and as a proxy for labor force on the production side. In short, the treatment of population in these models is much less sophisticated than the state of the art in population–environment research, as illustrated by the site-specific studies in this volume. Fischer and O’Neill present a modeling effort at the country level that might serve as a bridge between local studies and global models. Interest in other approaches to modeling complex relationships among population, land use, and environment is also beginning to emerge in the literature. Messina and Walsh (2001) have used a cellular automata approach to model land cover change in the Ecuadorian Amazon in a spatially explicit way. Cellular automata models are composed of a regular grid of cells each in a finite state that are iteratively updated in discrete time steps according to a set of transition rules (Walsh et al., Chapter 6). The models provide a formal framework for investigating the behavior of complex, extended systems. The rules that drive the Walsh and Messina cellular automata model are derived from formal theories of growth and change and the results of empirical analyses. The models are now being developed to explore deforestation and the extensification of agriculture in relation to human settlement, the extension of roads, and other changes under way in Nang Rong, Thailand (Walsh et al., Chapter 6). Some researchers are beginning to use agent-based modeling approaches (e.g., Sanders et al., 1997; Parker et al., 2002; Walker, 2003). Agent-based models examine the characteristics and activities of individual agents as

OCR for page 7
Population, Land Use, and Environment: Research Directions they interact and change over time as they adapt (or not) to their environment and learn (or not) from experience. These models are well suited to the study of complexity in land use change (Berry et al., 2002; Parker et al., 2002). For example, Liu and his colleagues have developed an agent-based model that simulates interactions among population, land use, and panda habitat at multiple levels in the Wolong Nature Research in Sichuan Province, China (Liu et al., Chapter 9; An et al., 2003). They have used the model to project the demographic and ecological consequences of policy scenarios in a spatially explicit manner. Individuals and households are the agents in these models. As with any modeling attempt, of course, the projections are only as good as the parameter estimates on which they are based. Neither cellular automata nor agent-based models test causal relationships directly. Rather, the models are a vehicle to work out the implications of complex interrelationships among population, land use, and environment specified in theory. Interestingly, none of the models just described yet includes feedbacks from land use and the environment to population size, structure, or change, although the potential clearly exists for doing this. PROGRESS IN MULTIDISCIPLINARY AND INTERDISCIPLINARY RESEARCH Research on population–land use–environment relationships over the past decade has opened up new research directions, suggested new hypotheses, and produced interesting results. Much of this research has necessarily been multidisciplinary or interdisciplinary, because the disciplines that study human population dynamics, land use change, and environmental change have not usually communicated much with each other. Projects funded by NICHD, other government agencies, and private foundations, focusing mainly on population–land use interactions, have made good progress in eliciting collaboration across the social sciences and between the social sciences and some natural science disciplines, notably those involved in interpreting remote observations of land use and land cover change. Other studies, focused mainly on connections between land use and environment, have generated productive collaborations linking ecologists and social scientists (Gutmann et al., 2004; Moran and Ostrom, 2005). These collaborations across disciplines have taken a variety of forms. One involves incorporating the methods and materials from one discipline into another. For instance, Foster (Chapter 12) describes a project that uses a satellite-based measure of greenness (the normalized difference vegetation index) in an analysis of property rights institutions and their consequences for forest cover that is fundamentally rooted in the discipline of economics.

OCR for page 7
Population, Land Use, and Environment: Research Directions Another approach creates a multidisciplinary team with a coordinated approach to an interdisciplinary question and a division of labor that highlights the unique contribution of specific disciplines. Fischer and O’Neill (Chapter 3) describe such an approach to the modeling of land use and land cover change in China. Mathematical modeling provides a common language that can enable researchers from different disciplines to communicate to each other certain aspects of their expertise that are critical to the complex system they are studying, even though they do not share disciplinary paradigms. Still another approach is to fuse two or more disciplines to create an interdisciplinary framework and analysis. Sometimes a single researcher will become trained in more than one discipline, and the synthesizing is at the level of the individual, but more often, interdisciplinarity is achieved through collaboration. Many of the teams contributing to this volume have attempted this kind of synthesis of theory, approach, data, and interpretation. Many barriers stand in the way of mixing disciplines in research on population, land use, and environment. Differences in the cultures of separate disciplines are one such barrier. The differences may be as fundamental as vocabulary. When social scientists talk about populations, they mean human populations. When ecologists talk about populations, they are equally likely to be referring to plant or animal populations. When social scientists refer to surveys, they mean surveys of individuals, households, or some other socially defined unit. For ecologists, surveys may connote what land surveyors do. Many other examples could be given. Unless special efforts are made to develop common meaning, not just common language, differences in vocabulary can lead to miscommunication and create misunderstandings (Bohm, 1996). Discipline-specific standards with respect to data, methods of analysis, and interpretation are also a challenge. To be credible, interdisciplinary research may have to meet standards in all of the disciplines involved. As all who have tried it have discovered, interdisciplinary research takes extra time and patience. As Matson noted in comments made during the workshop, the researchers as well as the research need to be integrated. Mixing disciplines in research has benefits, too, including an interesting cross-fertilization among members of some of the teams, with demographers now concerned with the resolution of digital elevation models and environmental scientists concerned about issues involving human subjects. A discussion of some lessons learned from a large interdisciplinary research center’s effort in this area can be found in Moran and Ostrom (2005: Chapter 14). Interdisciplinary research is inherently difficult (National Research Council, 2004a), but it is made more so by the balkanized structure of the university, reward systems, and peer networks. Faculty generally hold

OCR for page 7
Population, Land Use, and Environment: Research Directions appointments in departments, and students generally get Ph.D.s in a single discipline. In making judgments about whom to hire, promote, and grant tenure, there may not be an appreciation for the additional time it takes to do top-quality interdisciplinary research, or for the fact that multiple players are involved and, as a result, most publications are multiauthored. At least in social science departments, coauthors of multiauthored works may not be given credit commensurate with their contributions when it comes to promotion and tenure decisions. Publication may also pose problems. The editors of the most prestigious journals in individual disciplines may not be interested in, or competent to review, interdisciplinary research. These difficulties pose particular problems for scientists in their early careers who work in interdisciplinary teams. Efforts must be made to overcome these institutional and cultural barriers. Finally, these challenges of interdisciplinarity raise questions about how best to train students to contribute to population–environment research. Although there are hundreds of undergraduate programs across the country with names like environmental studies and environmental science, they vary widely in terms of the disciplines included and in rigor. Graduate programs vary less in rigor, but the opportunities for interdisciplinary work by graduate students vary considerably. Systematic in-depth and rigorous training in the full range of social and environmental sciences relevant to population–land use–environment research is comparatively rare at the graduate level, compared with training in multiple disciplines in either the social or natural sciences.