Deforestation Drivers: Population, Migration, and Tropical Land Use

The tiny global minority residing in rural frontier areas—how long people remain there; the timing, magnitude, and characteristics of their consumption; and their demographic transitions—promises a vast impact on future tropical deforestation. It is here, not in cities and not in long-settled rural areas, that fertility and native population growth are extremely high, rural migration is dynamic, and land cover change remains extraordinarily expansive per capita. According to recent UN projections,¹ the vast majority, indeed very likely all, of the world's net population growth over the next several decades will occur in the world's poorest cities.

In a paper published in 2009², the first author argued for the importance of rural migration in tropical deforestation in Latin America. This paper updates and builds on this work, following the rapid urbanization and globalization of agricultural production during recent years and in light of more advanced satellite imaging demonstrating this shift. We contend the highest population growth rates and old growth forest conversion will continue to occur in remote rural environments, with notable implications for land cover change research and policy³ specifically, and for coupled human–environmental systems more broadly.⁴

Despite the inordinate ecological imprint of remote rural populations, what will happen with the population of distant rural communities remains one of the biggest mysteries in demographic and land change research. While data on these populations are scarce,⁵ the handful of studies examining fertility in remote frontier environments in the Amazon⁶ and Central America⁷ suggests extremely high fertility rates exceeding seven births per woman in some locations, even where in some instances it has declined precipitously in rapidly urbanizing former frontiers.⁸ The timing, pace, and magnitude of the demographic transition among this group, for which we have patchy data coverage at best, will assert an outsized impact on forest transitions, whether directly, as small farmers, or indirectly, following migration to urban areas where the impact will be more removed in space and time vis-á-vis consumption. It remains an open question as to whether the deforestation impact will increase through an earlier transition—more urban dwellers consuming meat but fewer people overall (i.e., increasingly through teleconnected impacts⁹)—or from a delayed transition—more people overall residing on the planet and more people for a longer period living as subsistence farmers with a direct impact spatially and temporally on tropical forests. In a delayed transition scenario, population pressure on the land from high fertility will be further exacerbated by continued out-migration to newly opened agricultural frontier in other locations. Evidence from Ecuador¹⁰ and Guatemala¹¹ suggest that these unusually rapidly growing populations produce a disproportionate number of frontier migrants; subsequent-generation frontier farm children are much more likely to follow the example of their parents and move to another frontier location than are other rural populations. Thus, while the bulk of migration today is characterized by rural-to-urban or urban-to-urban flows, the small fraction of farmland-seeking rural-to-rural migrants has an immediate, enduring, dynamic, and disproportionately large impact on global deforestation, particularly within and adjacent to protected areas.¹²

We argue that forest frontier colonization by smallholder farmer migrants may continue to be the main proximate cause of deforestation—especially of old-growth forests of high biodiversity and ecological integrity—exceeding the amount of intact forest conversion caused by the more popular culprits, commercial logging and capital-intensive industrial agricultural operations. Because there are no regional-scale data on the proportion of Latin American migration that is rural-to-rural or on the amount of deforestation caused by these migrants, we present several points and provide examples of migrant-induced deforestation in Guatemala to describe how smallholder migrant farmers may represent the largest proximate driver of deforestation in Latin America, and possibly worldwide. This argument is not new; it is a claim posited by researchers with some assertion as recently as the 1990s.¹³,¹⁴ But the position has lost traction in recent years with the dramatic increase in large-scale global commodity-driven farming operations.

The following paragraphs summarize and expand upon the first author's argument¹⁵, with updated references and selected illustrative examples. The remaining portion of the paper introduces the importance of spatial and temporal scale, the use of appropriate demographic variables, and discusses potential limitations and caveats to recent conclusions derived from remote sensing analysis. With Latin America over 80% urban following rapid rural-urban migration since the mid 20th century, migration recently has turned to secondary and tertiary cities internally or to international destinations.¹⁶ Yet despite this unprecedented coupled urban-land use transition, which undoubtedly represents a large and growing portion of global land cover change, we argue that small farms continue to play a major role in rural land change in general, but more particularly in old-growth forest conversion to low-intensity agriculture. Within the developing world a large portion of the remaining forests are located in legally protected areas, areas with a large portion of the globe's old-growth forests while a much smaller portion of forests authorized for clearing located near already established farming communities. Paradoxically, forests located in unprotected areas may be no better conserved than legally protected areas. A study analyzing 163 forests in 13 countries did not find a statistically significant difference in forest conditions, as measured by vegetation density, between legally protected forests and forests that are not legally protected. Conversely, given the scarcity of forest dependent agricultural resources and fuel wood in high population density areas great care is often taken to protect these resources¹⁷. Legally protected forest reserves do not share the same communal incentive for conservation. study analyzing 163 forests in 13 countries did not find a statistically significant difference in forest conditions, as measured by vegetation density, between legally protected forests and forests that are not legally protected.¹⁸ In contrast, efforts are often taken to preserve communal woodlands, given the scarcity of forest dependent agricultural resources and fuel wood in high population density areas²⁰, In communities that share a forest reserve there is incentive for members of the community to observe and regulate communal resource use. In legally protected forest reserves, there is no such incentive. As an example, rural migrants to Guatemala Maya Biosphere Reserve invoke national law allowing farm households to occupy idle lands to justify settlement within this United Nations Educational, Scientific, and Cultural Organization (UNESCO)-supported conservation area. Park guards are reluctant to apply national conservation law when confronting subsistence farmers who will fight, in some instances literally to the death, to maintain their family's subsistence plot.²⁰ Similar dynamics have occurred in neighboring countries.²¹

On the Guatemala frontier, as elsewhere, agriculture is expansive rather than intensive because of several characteristics: cheap available family labor, scarce capital, low technology, high cost of transportation, and insecure land tenure. Further, as soon as migrant colonists arrive on the frontier they often clear land in order to establish their claim and begin agricultural operations. As lands become consolidated and converted to large scale commercial agriculture, small farmer colonists are displaced further into the external frontier regions where they initiate a new round of deforestation on previously untouched forest. In this process, as illustrated in Figures 1 and 2, small farmers are displaced from the internal frontier—the border between cleared agricultural land and forest within a farm—to the external frontier—the border between agricultural land and forest outside the last row of farms adjacent to a forested area.²² As deforestation continues within internal frontier regions there is less land available for farm expansion, increasing land scarcity, and placing pressure on farmers to increase farming intensity Land overexploitation can leave the soil exhausted, promoting further displacement and migration of farmers to the external frontier forests. Land abundance also encourages settlement in the external frontier in the first place. Migrant colonist farmers will not gladly settle in already established high-population-density farming regions featuring land unavailability and high land costs.

Therefore, focusing on capital-intensive industrial agriculture as the immediate driver of land cover change may belie the fact that the deforestation of old growth tropical forest was originally caused by small scale farmer migrant families.²³ How much initial deforestation is caused by small farmer colonization relative to corporate farming and ranching remains unknown but is likely to represent a disproportionate and significant amount of old growth deforestation of high conservation priority.²⁴

Overall, the proportion of deforestation worldwide that is proximately caused by frontier colonist farmers may even be increasing in some critical conservation corridors. In the past few decades in many rural frontier areas, the rate of deforestation has increased while the rate of population growth has declined, indicating the amount of forest clearing on a per-farmer basis has increased.²⁵ This suggests that even as rural areas depopulate, farmers and cattle ranchers in these low population density areas may be increasingly responsible for deforestation.²⁶

Lastly, forest impact per capita is further enlarged among migrant farm households when, as is often the case, they clear land in new settlement frontiers several times during their life course. This is exemplified in ongoing research in Guatemala where we have observed multiple generations of households migrating sequentially from areas of low land availability to settle in conservation areas, where forest conversion to agriculture changes conservation landscapes to agricultural landscapes.²⁷ Over time, land consolidation and fragmentation, coupled with the exhaustion of forest resources in situ, catalyze subsequent migration by the same household or by second-generation households.²⁸ The pattern is unsurprising and well documented. A (semi)subsistence household, with various options available in response to resource scarcity in a current location, might logically decide to migrate to a place where land is available de facto (if not de jure) as characterized by many Latin American protected areas. Since the impact of a small farm household on forest conversion is located in several different places over time, it is essential that the political, demographic, social, economic, and ecological mechanisms behind rural–rural migration are well studied before policy may be developed to influence this rural–rural migration pathway in ways that conserve the remaining tropical forests and support rural development.

In analyzing the multiple temporal and spatial processes by which deforestation unfolds, it is useful to differentiate internal frontiers from external frontiers.²⁹ Internal frontier refers to the forest that remains on established farms within settled areas. Changes that occur within the internal frontier can be analyzed by studying the ways farmers utilize their land. On the other hand, the external frontier refers to the area of untouched old growth forest that exists beyond the existing areas of farmer settlement. It is here on the external frontier that the majority of future deforestation will take place. Therefore, it is only through researching migration to the external frontier that we can better understand the potential for deforestation in currently unsettled areas. Migration to the external frontier will remain a critical driver of forest conversion in Latin America as the region harbors twice the tropical forest extent of any other world region.

In conducting studies to quantify the impact that small-scale migrant farmers have on deforestation, it is important to use the appropriate demographic variable and geographic scale. Using national-scale data, a recent study correlated forest loss with export industrial agriculture and urban population growth, while finding no correlation with rural population growth.³⁰ However, if rural population growth data is used at the national scale, the demographic situation at the local level on the external frontier may be inappropriately reflected. Rural to rural migration will not emerge in national scale data on rural population change since the migrants settle the external frontier from other rural areas rather than from urbanized areas. Therefore, the use of national scale rural population growth as the variable to characterize the deforestation impacts of rural populations may underrepresent their true impact.³¹ In order to accurately assess the impact of rural-to-rural migration and small-scale farmers on forest loss, it is essential to use migration data, or at the very least to use a measure of rural population growth, but only at a sufficiently small scale to reveal in-migration to the forested external frontier regions.

When using remote sensing to quantify forest loss, it remains imperative that the importance of scale and spatial relationships of deforestation drivers be considered. The two drivers discussed thus far in this article, export industrial agriculture clearings and small-scale farmer clearings, have vastly different geographic characteristics, including spatial distribution, clearing size, nearest neighbor distance, shape, and percent of canopy remaining within a clearing. Export industrial agriculture is typically characterized by large rectangular clearings that are spaced closely together and are almost completely denuded of trees (see Figures 1 and 2). On the other hand, small-scale farmers' clearings on the external frontier are often of small size, irregularly shaped, spaced relatively far apart from each other, and still retain a sizeable amount of trees within the farm plot (see Figures 1 and 2).

If all of these distinct geographic properties are not accounted for when designing a remote sensing study, deforestation and its drivers may be misrepresented. As an illustration, a study by Hansen et al.³² used 500-meter MODIS data to identify areas of deforestation. Within these areas a random sample of blocks 18.5 km by 18.5 km of 30-meter Landsat imagery was used to derive rates of deforestation that could be extrapolated to the regions of deforestation as a whole. Accordingly, deforestation from kilometer-sized export agriculture farms would have been easily detectable, but deforestation from small-scale rural migrant farmers, whose incompletely denuded, spaced-apart plots are smaller than the 500-meter pixel size, would have been concealed. Thus, deforestation may be more severe than studies indicate in regions where small-scale frontier farmers remain key drivers of forest loss, further stressing the importance of avoiding the ecological fallacy by matching the scale of the methodology to the scale of the phenomena of research interest. Biomes are also critical in understanding deforestation drivers. In biomes of low rainfed water availability, where agriculture exists, it will tend to be dominated by capital and water intensive large farms and ranches. Where rain is abundant, rain-fed agriculture and small scale household farms may be more likely drivers of land change A recent study examining deforestation by distinct biomes corroborates a migration–poverty pathway of continued deforestation of moist tropical forests using satellite imagery from 2001 to 2010 for Central America.³³ The research indicates that The Human Development Index was associated with moist tropical forest loss with Nicaragua and Guatemala, the two poorest nations in the region, experiencing particularly rapid deforestation of moist forests –much of it in areas of small farm household colonization-accompanied by more modest but significant recovery of conifer and dry forests – often areas of out-migration.³⁴ These findings corroborate a Kuznetz curve association of U-shaped deforestation impact among small farm households at one end of the development spectrum, and wealthier urban or peri-urban households on the other.³⁵

Rural migrant farmers will continue to be a dominant driver of the deforestation of primary tropical forests despite the recent expansion of export-oriented large-scale industrial agriculture in the tropical regions of Latin America and elsewhere in the tropics. For the reasons outlined, rural migrant farmers primarily contribute to deforestation along the external frontier, whereas the expansion of industrial agriculture will contribute to further deforestation—but much more to the conversion among non-forest land classes—within the internal frontier. If studies are to accurately assess the extent and drivers of deforestation, and if policies are to be effective at reducing deforestation, it is essential that remotely sensed data be appropriately interpreted with a nuanced appreciation of distinct land use signatures inscribed in space, scale, and shape. It will also remain imperative to continue to collect data on the ground among small farm households regarding their migration history, demographics, socioeconomic characteristics, and land use practices. More research and policy initiative may be usefully directed at the minority population residing in forest frontiers and those with high probabilities of migrating to such areas. The most remote rural dwellers, those for whom we have the least data, will continue to exert an effect heavily disproportionate to their size on tropical deforestation and on the future global population.³⁶

1. United Nations, Department of Economic and Social Affairs, Population Division. “World Population

Prospects: The 2010 Revision, Volume I: Comprehensive Tables,” ST/ESA/SER.A/313 (Geneva: United Nations, 2011).

2. D.L. Carr, Population and Deforestation: Why Rural Migration Matters, Progress in Human Geography 33, no. 3 (2009): 355–378.

3. B. L. Turner, E. F. Lambin, et al., The Emergence of Land Change Science for Global Environmental Change and Sustainability, Proceedings of the National Academy of Sciences USA 104, no. 52 (2007): 20666–20671.

4. L. An and D. López-Carr, Modeling Coupled Human-Natural Systems: Research Directions, Ecological Modeling 229 (2012): 1–4.

5. See, for example, D. L. Balk, U. Deichmann, et al., Determining Global Population Distribution: Methods, Applications and Data, Advances in Parasitology, 62 (2006): 119–156.

6. J. Bremmer, R. Bilsborrow, et al., Fertility Beyond the Frontier: Indigenous Women, Fertility, and Reproductive Practices in the Ecuadorian Amazon, Population & Environment 30, no. 3 (2009): 93–113.

7. E. Sutherland, D. L. Carr, et al., Fertility and the Environment in a Natural Resource Dependent Economy: Evidence From Petén, Guatemala, Población y Salud en Mesoamérica 2, no. 1 (2004): 1-12.

8. D. L. Carr, W. K. Pan, and R. E. Bilsborrow, Declining Fertility on the Frontier: The Ecuadorian Amazon, Population and Environment 28, no. 1 (2006): 17–39.

9. See, for example, P. Meyfroidt, T. K. Rudel, et al., Forest Transitions, Trade, and the Global Displacement of Land Use, Proceedings of the National Academy of Sciences USA 107, no. 49 (2010): 20917–20922.

10. A. Barbieri, D. L. Carr, and R. E. Bilsborrow, Migration Within the Frontier: The Second Generation Colonization in the Ecuadorian Amazon, Population Research and Policy Review 28, no. 3 (2009): 291–320.

11. L. Suter and D.L. López-Carr, Beyond the Middle of Nowhere: Out-Migration From the Agricultural Frontier in the Sierra del Lacandón National Park, Guatemala, Proceedings of the Latin American Studies 2012 International Congress (2012).

12. Carr, note 2 above.

13. R. A. Houghton, Land-Use Change and Tropical Forests, BioScience May, 44 (1994): 305–331;

14. N. Myers, “Tropical Deforestation Rates and Patterns. In Brown, K. and D. Pearce, Eds. (1994). The Causes of Tropical Deforestation: The economic and statistical analysis of factors giving rise to the loss of the tropical forests. London, University College London Press Ltd.

15. D.L. Carr, Population and Deforestation: Why Rural Migration Matters, Progress in Human Geography 33, no. 3 (2009): 355–378.

16. United Nations et al., note 1 above.

17. E. Kalipeni, Demographic Responses to Environmental Pressure in Malawi, Population and Environment 17, no. 4 (1999); and B. J. Wallace, How Many Trees Does It Take to Cook a Pot of Rice? Fuelwood and Tree Consumption in Four Phillippine Communities, Human Organization 54, no. 25 (1995): 182–186.

18. T. M. Hayes, Parks, People, and Forest Protection: An Institutional Assessment of the Effectiveness of Protected Areas, World Development 34, no. 12 (2006): 2064–2075.

19. E. Kalipeni, Demographic Responses to Environmental Pressure in Malawi, Population and Environment 17, no. 4 (1999); and B. J. Wallace, How Many Trees Does It Take to Cook a Pot of Rice? Fuelwood and Tree Consumption in Four Phillippine Communities, Human Organization 54, no. 25 (1995): 182–186.

20. D.L. Carr, A Tale of Two Roads: Land Tenure, Poverty, and Politics on the Guatemalan Frontier, Geoforum 37, no. 1 (2006): 94–103.

21. J. Ericson, M. Freudenberger, et al., Population Dynamics, Migration, and the Future of the Calakmul Biosphere Reserve (Washington, DC: American Association for the Advancement of Science [AAAS] Program on Population and Sustainable Development [PSD], 1999), p. 39; P. H. Herlihy, ‘Wildlands’ Conservation in Central America During the 1980s: A Geographical Perspective, Conference of Latin American Geographers 17/18 (1990): 31–43; and J. H. Smith, Land-Cover Assessment of Conservation and Buffer Zones in the BOSAWAS Natural Resource Reserve of Nicaragua, Environmental Management 31, no. 2 (2003): 252–262.

22. W. F. Laurance, Switch to Corn Promotes Amazon Deforestation, Science 318, no. 5857 (2007): 1721.

23. D. C. Morton, R. S. DeFries, et al., Cropland Expansion Changes Deforestation Dynamics in the Southern Brazilian Amazon, Proceedings of the National Academy of Sciences USA 103, no. 39 (2006): 14637–14641.

24. D. L. Carr, Farm Households and Land Use in a Core Conservation Zone of the Maya Biosphere Reserve, Guatemala, Human Ecology 36, no. 2 (2008): 231–248.

25. D. López-Carr, J. Davis, M. Jankowska, L. Grant, A. C. López-Carr, and M. Clark, Space Versus Place in Complex Human–Natural Systems: Spatial and Multi-Level Models of Tropical Land Use and Cover Change (LUCC) in Guatemala, Ecological Modeling 229 (2012): 64–75.

26. P. M. Fearnside, Will Urbanization Cause Deforested Areas to be Abandoned in Brazilian Amazonia?, Environmental Conservation 35, no. 3 (2008): 197–199.

27. Suter and López-Carr, note 11 above.

28. Carr, notes 2 and 24 above, and Barbieri et al., note 10 above.

29. Carr, note 2 above.

30. R. S. DeFries, T. Rudel, et al., Deforestation Driven by Urban Population Growth and Agricultural Trade in the Twenty-First Century, Nature Geoscience 3, no. 3 (2010): 178–181.

31. Ibid.

32. M. C. Hansen et al., Humid Tropical Forest Clearing From 2000 to 2005 Quantified by Using Multitemporal and Multiresolution Remotely Sensed Data, Proceedings of the National Academy of Sciences USA 105 (2008): 9439–9444.

33. T. M. Aide, M. Clark, R. Grau, D. López-Carr, D. Redo, M. Bonilla, M. Levy, The Deforestation and Reforestation of Latin America and the Caribbean (2001–2010). Biotropica (2012, in press); and D. Redo, “Asymmetric Forest Transition Driven by the Interaction of Socioeconomic Development and Environmental Heterogeneity in Central America, Proceedings of the National Academy of Sciences USA 109, no. 23 (2012): 8839–8844.

34. Ibid.

35. R. R. Chowdhury and E. F. Moran, Turning the Curve: A critical Review of Kuznets Approaches, Applied Geography 32, no. 1 (2012): 3–11.

36. David López-Carr thanks T. M. Whitmore, R. E. Bilsborrow, and B. L. Turner II for their insights over the years in helping form some of these ideas. He also recognizes the following support for this research: National Institutes of Health Career Development Award, K01 (HD049008); National Science Foundation Dynamics of Coupled Natural and Human Systems (CNH) (EF 0709627): National Science Foundation Geography and Regional Science Grant (BCS-0525592); and a University of California, Santa Barbara Faculty Career Development Award.

David López-Carr is a Professor in the Department of Geography at the University of California, Santa Barbara and Director of Latin American and Iberian Studies. He researches population, health, and environment interactions in the developing world.

Jason Burgdorfer is a graduate student in the Department of Geography at The George Washington University. He has research interests in human-environment relations, health geography, and applied geospatial techniques.