Recommended Research Directions
Over the past decade, considerable progress has been made in understanding the environmental effects of demographic change that are mediated by changes in land use and land cover. Much of this progress has been made through studies that have followed population–land use–environment interactions at particular sites over many years. The ability to conduct such studies has been promoted by the existence of targeted funding for such research and by the development of international networks of researchers, aided in part by the organization of the international Land Use/ Land Cover Change Project, and its successor, the proposed Global Land Project, under the auspices of the International Human Dimensions Programme on the Human Dimensions of Global Change and the International Geosphere-Biosphere Programme. Methodologically, research has been greatly advanced by the availability and use of observations from space, particularly from Landsat, that were little used to study land use change a decade ago. The rapidity of change in the use of remote observations can be seen by comparing research reported in the appendixes to this volume with the research reported in Population and Land Use in Developing Countries (National Research Council, 1993) more than a decade ago.
Site-based studies have yielded substantive advances in understanding, with potential implications for policy. They have shown how the environmental effects of population change depend not only on increases in population numbers but also on other demographic changes, including migration and changes in household size. For example, research based in Wolong, China, indicates that policies that relocate young people out of the panda reserve are not only more socially acceptable than policies that relocate
other groups, but also more ecologically effective and economically efficient (Liu et al., Chapter 9). Site-based studies have shown that environmental effects are seen most clearly when broad land use and environmental categories, such as the urban-rural distinction, are unpacked into more focused distinctions. For example, the distinction between primary forest and secondary forest is crucial to understanding the role of migration, land use practices, deforestation, and rates of regrowth in the Brazilian Amazon (Moran et al., Chapter 5). Site-based studies have shown that population–land use–environment relationships are scale dependent, that is, that the relationships that are evident at one level of spatial, social, or temporal analysis are not necessarily found when analysis is conducted at another level. Such findings caution against naïve generalization across scales and have opened the question of how interactions occurring at one scale affect or embed interactions at other scales.
Site-based studies have also led to methodological advances, such as in the design of multithematic longitudinal data sets, the development of methods to link population units to land units and land units to environmental effects, and innovative approaches to modeling. They have also led to improved methods of measuring land use variables by remote observation—a necessary building block for research that can compare larger numbers of sites.
The site-based approach has also clarified the key challenges for future research. One is to move from descriptive to causal understanding of population–land use–environment relationships. A major challenge is to improve comparability of data and analyses across sites in order to build generalizable knowledge. There is also the challenge of developing interdisciplinary collaboration across the wide range of sciences relevant to this topic and developing the capacity for this interdisciplinary research in the next generation of scientists.
After consideration of research developments over the past decade and the state of knowledge on population–land use–environment relationships, the panel makes the following recommendations for the continued development of this field.
1. Research should be increasingly coordinated to promote creation of a body of integrated knowledge. Knowledge about the relationships among demographic, land use, and environmental variables has been substantially enriched in recent years by empirical studies using new multithematic, multilevel, longitudinal, and spatially explicit data sets to describe and analyze these relationships at specific sites around the world. Of course, with respect to the development of general knowledge, there are inherent limitations of such site-specific studies, as noted in Chapter 1. Such general knowledge is critical for anticipating possible scale dependencies and cross-
scale interactions, aggregating appropriately from site-specific knowledge, developing forecasts for specific sites, and understanding the likely trajectory of these processes elsewhere. Several actions can help move the field in directions conducive to building general knowledge.
a. Organizations that support population–land use–environment research should work with researchers to develop minimum reporting standards for data collected and analyzed in site-based studies. Reporting standards should be developed to describe the data used in these studies to facilitate comparison among them.
Site-specific studies will continue to contribute important knowledge. Their value can be greatly increased, however, by efforts to establish and increase data comparability across sites. When comparing results from two or more sites, one explanation for any differences among them is in the design and analysis of the data on which the results are based. This explanation cannot be evaluated if the information about the design and analysis of the data is incomplete. To obtain generalizable results from existing studies that will contribute to a deep understanding of fundamental aspects of population–land use–environment relationships, it is therefore critical to establish minimum standards with respect to collecting and reporting data. Even though we do not judge it appropriate at this time to standardize measurement techniques, the need to assess the reliability and generalizability of results presumes that studies are explicit about the properties of the data they are using and how key variables are measured. Survey data provide one example. Researchers using such data in population–environment studies should, at a minimum, report on sample design, size, and response rate. The American Association of Public Opinion Research (2004), for example, has published a set of standard definitions for survey response rates that could serve as a useful guide. New surveys should use sample selection and fieldwork procedures that conform to high standards that maximize comparability and generalizability.
b. Individual projects should provide an inventory of important contextual variables for their study sites. Results may differ among two or more sites because of differences among places in technologies used; formal and informal institutions; local, national, and international markets; policies; and the natural environment. For example, differences in property rights (e.g., regarding common land), in the existence and spatial scope of product markets, and in the institutions and regulations governing use of resources can have important effects on how population interacts with land and water use. Specification of the differences and similarities in these characteristics across localities is fundamental to interpreting and generalizing from research findings, developing an understanding of contextual effects and cross-scale interactions, and providing insights into possibly efficacious policy interventions.
c. Efforts should be made to coordinate definitions of variables measured and research designs chosen in different site-specific studies to enhance the creation of a body of integrated knowledge in the field. So far efforts to produce generalizations have been limited by the incompatibility of variables and research designs used in the individual studies. While some of these differences are necessary because of site-specific conditions or varying research questions, others may be due to a lack of communication and coordination among study groups. Such communication might be accomplished through funder-sponsored conferences, international research projects such as the Global Land Project project, or at other meetings of the research community.
d. Two substantive areas of research—regions of new settlement (“frontier”) areas and regions of rapid urban (including suburban) development—are ripe for producing integrated, general knowledge. These areas have been the focus of multiple recent studies of similar phenomena in different regions and thus hold promise for building knowledge of generic processes of population–land use–environment interaction. Organizations that support population–land use–environment research should encourage efforts that examine the differences and similarities across sites of these kinds, with attention to contextual differences and similarities. Collaboration and organized communication (e.g., research workshops) involving researchers across projects and sites who are looking at similar phenomena in different places may facilitate in important ways the creation of a body of integrated knowledge about population–land use–environment relationships.
2. Research should continue to decompose or unpack the complex, general phenomena of population, land use, and environment and examine causal relationships involving their more specific component factors. Recent research has demonstrated the importance of insights that can come when broad variables, such as population growth and land cover, are further differentiated. For instance, the effects of population growth depend on whether growth is due to natural increase or migration, and if migration, the relative size of streams into and out of a place, whether the migration is temporary or permanent, and on the characteristics of the migrants. Population effects may depend in important ways on changing numbers of households by size and on the age structures of the members of these households over time and on temporary migrations that are not often measured, such as of commuters, tourists, seasonal workers, and illegal migrants. Land cover categories observable with current remote sensing techniques often contain highly heterogeneous land uses, especially in urban areas. Disarticulated analyses of such general factors will continue to yield better understanding of the mechanisms and feedbacks that connect
population, land use, and environment. It will also help clarify the assumptions and theoretical structures that researchers in each specialty use and thus facilitate interdisciplinary communication and integration.
Population–environment research on land use should include research with a substantive focus on water. Human land use inevitably affects and is affected by ground and surface waters and habitats defined by waters. Population–environment relationships mediated by water are particularly important because of the importance of water as a resource and a substrate and carrier for nutrients and pollutants, and because most of the world’s urbanization and much intensification of food production are occurring in coastal and riparian regions. Although these areas constitute relatively small percentages of the Earth’s surface, they are very important to marine productivity, storm and flood impacts, and transformation of wetland habitats. Even in inland and upland areas, watershed function is highly influenced by human population and land use change. Infiltration and ground water recharge, runoff and erosion, riparian zone desiccation, and altered flood regime are among the important environmental processes that link with population and land use.
3. Research should investigate the dynamic interactions involving population and land use and environmental variables. Until now, most of the studies addressing the relationships of demographic, land use, and environmental issues have focused mainly on population and land use or land use and environment. Few have fully coupled all three classes of variables. Although continued research on pairs of these elements remains useful, it is important to expand research efforts to connect all three elements and forge appropriate scientific linkages. Fully integrated studies that incorporate aspects of population, land use, and environment are needed to better understand how human activities are altering the Earth’s system and how these activities are affected in turn by environmental changes.
In developing these integrated studies, researchers should be attentive, as noted above, to the specific demographic, land use, and environmental factors involved and to contextual factors that may influence these relationships, including social institutions and geographic location, among others. They should be attentive, as noted below, to the ways in which population–land use–environment relationships may vary depending on the spatial, temporal, and social units used for analysis and to the mechanisms and causal processes involved. The units of analysis and regional scale of a study should typically reflect the scales at which the processes of concern operate and at which decisions are made. Studies should also be explicit in describing the dynamics of change over time in all three elements and in their interactions.
4. Research should increasingly explore scale dependencies and cross-scale interactions. Population–land use–environment relationships that appear strong at one spatial, temporal, or social scale of analysis sometimes weaken or disappear when analyzed at other scales. Moreover, phenomena are linked across scales: larger scales may set the context or limitations for relationships at smaller scales, and relationships at smaller scales may aggregate to larger scales in surprising ways that indicate emergent properties. These general issues of human-environment interaction can be studied productively in the context of population–land use–environment relationships. Adding considerations of scale do not necessarily make research problems more difficult or complex. Often, thinking carefully about the spatial, temporal, and social scales of processes allows investigators to choose a scale of analysis that simplifies the problem.
a. Researchers focused on population–environment relationships should pay explicit attention to spatial, temporal, and social scale in framing their studies and offering explanations. They should be explicit about the scales at which they are working. Future research will advance understanding more effectively by focusing more explicitly on defining the scales of each study and placing studies in relationship to each other on spatial, temporal, and social scales. Researchers should ensure that their data are spatially and temporally explicit with regard to level of analysis and frequency of data collection, and thus amenable to scaling up or down.
Research funders should use their influence to develop gridded approaches that can place local studies within spatial matrices and to encourage levels of temporal resolution (including of remotely sensed images) adequate to understanding dynamic processes. The field is now ready to respond to the challenge of using the historical detail available from time series of remote images, as indicated by some of the contributions to this volume that employ such methods as analysis of pixel or plot trajectories (Walsh et al., Chapter 6; Redman, Chapter 7) and econometric methods for time-series analysis (Seto, Chapter 8). A gridded approach can help shape the samples of research sites and measurement times to bear an understandable relationship to the larger spatial, temporal, and social scales to which site-specific data will be aggregated.
b. Researchers should be encouraged to address explicitly the extent to which the population–land use–environment relationships they study vary by scale of analysis, how these scale dependencies may vary by place or time, and how relationships at one scale may influence those at another scale. There are many strategies for examining these scale dependence and scale interaction issues, including the incorporation of multiple scales of analysis within single studies; the linking of research activities that overlap on one dimension (e.g., same place) but differ in scale of analysis; and the use of dynamic modeling approaches, such as cellular automata, agent-
based models, artificial intelligence approaches, and nonlinear dynamics. The dynamic modeling approaches may be helpful in examining temporal and social scale interactions, such as the conditions under which complex systems go into periods of rapid restructuring and the ways in which phenomena at small spatial or social scales may aggregate in nonlinear ways.
5. Organizations that support population–land use–environment research should support and encourage continued development of linked data sets that include information about population, land use, and environmental variables and that are spatially explicit, multilevel, and longitudinal. The objectives of building integrated knowledge; linking population, land use, and environment; and understanding scale dependencies and cross-scale linkages will all be greatly advanced by the development of data sets that allow for comparisons across variables, research sites, and scales and for analyses of feedbacks from environmental to demographic variables. Investment should be made both in continuing existing linked longitudinal data sets and in developing similar data sets at new sites in under-studied regions and in places that offer unique research opportunities.
Development of linked data sets will also help identify gaps in research by showing unevenness in data availability as a function of world regions; types of demographic and land use change; institutional and environmental contexts; and the availability of integrative studies that explicitly link population, land use, and environment. Such data sets will also help identify the depth of knowledge available for addressing particular important topics, such as urban expansion, “frontier” and coastal development, and critical environmental issues (e.g., desertification, carbon sinks, loss of biodiversity, invasive species expansion).
a. Continued investment is warranted in developing methods for data and process integration. For example, there are numerous challenges in developing geographic information systems (GIS) that can effectively integrate across time space for integrated studies of population–land use–environment relationships. Even with better measurement, problems will remain because there is no one-to-one link between population units, land use units, and environmental units. Moreover, multiple units in any location are relevant to multiple units in other places. The complexity of these linkages poses a significant methodological challenge for researchers and GIS developers.
b. Guidelines must be developed for use of linked data sets and for making data available to researchers beyond the original research team. A major issue is the tension between the value of broadly available multilevel, spatially explicit data and the confidentiality of individuals, households, or villages that might be identified from the data. Given the recommended investments in new and existing data sets, it is important that maximum
value be obtained. However, when researchers such as the contributors to this volume follow standard social science procedures, participants receive promises of confidentiality. Identifiers that reveal the name or spatial location of a village or dwelling unit pose the risk that information about individuals can be seen in or deduced from a data set, thus violating the confidentiality promise. The problem increases with data sets that are multilevel, longitudinal, and spatially explicit and that are potentially useful not only to researchers but also to others who might use the information in ways not desired by the people the data describe. Solutions to this problem are urgently needed.
6. Increased effort should be devoted to modeling and quantifying causal relationships among population, land use, and environment using a variety of approaches, as well as to analyzing uncertainties in models of these complex systems. Population–land use–environment relationships are embedded in a larger coupled human-natural system. Mathematical models provide a way to recognize this while working out the particular mechanisms involved in these relationships. They provide a means to address dynamic relationships among population, land use, and environment, including endogeneities and feedbacks that are not readily uncovered by analyses of observational data alone. Mathematical models have considerable value for structuring discussion across disciplines, identifying key questions that require empirical research, and providing forecasts for policy analysis.
7. A research effort should be made to identify more effective mechanisms to facilitate interdisciplinary research. Scientists who conduct interdisciplinary population–environment research report special difficulties not usually encountered in disciplinary research. Some of these difficulties impede collaboration, even among senior scientists. These include the use of different words for essentially similar concepts and of the same words to mean different things in different disciplines, the presence of bodies of tacit knowledge that remain hidden during most conversations, disagreements over the best way to tackle interesting research questions, invidious distinctions between “hard” and “soft” science, and the challenge of administering large and sometimes spatially and temporally dispersed teams. Senior scientists report that such difficulties result in slower progress toward research results and increased difficulty in publishing interdisciplinary work. In addition, this research is too often discounted by disciplinary peers. Such problems may pose significant barriers to career progress for junior scientists in a number of disciplines.
Such barriers to interdisciplinary research are widely noted, both among population–land use–environment researchers and more broadly among
academic researchers. Little systematic knowledge exists, however, about which barriers are most significant for particular lines of research or about which interventions are most effective in reducing impediments to the progress of research or the training of future researchers.
Organizations that support population–environment research should support a systematic assessment of which approaches seem to result in the best and most rigorously trained young scholars in this field and the most productive research collaborations. This assessment should be based on systematic analysis of experience from population-environment research and related interdisciplinary fields. It should address such questions as these: Does strong interdisciplinary training at the undergraduate or graduate levels make young scholars more effective at interdisciplinary collaboration? Are experienced graduate or postdoctoral mentors necessary to motivate and train interdisciplinary collaborators? Are graduate students and junior faculty who participate in interdisciplinary projects hindered in their careers, or is the perception that interdisciplinary participation is risky partially or entirely mythical? What differentiates effective from ineffective leaders of interdisciplinary teams? How do effective interdisciplinary teams recruit and reward participants? Do certain types of administrative structures, such as interdisciplinary research institutes, foster successful interdisciplinary research? What styles of communication and administration favor successful operation of dispersed multiinstitution and international collaborations?