INTRODUCTION
Political observers often make the point that, when politicians put together coalitions capable of enacting new legislation or funding new initiatives, they practise ‘the art of the possible’ (Butler Reference Butler1971). In conservation policymaking, politicians build these winning coalitions most frequently when their proposals build on processes already under way in a society (Kaimowitz Reference Kaimowitz2002). For example, political representatives from smallholder districts decide to support payments for ecosystem services (PES) policies (Alix-Garcia et al. Reference Alix-Garcia, de Janvry and Sadoulet2005) when the regrowth occurring on abandoned agricultural lands makes smallholders in their districts eligible for payments for the increments in ecosystem services provided by the abandoned lands. In this dynamic, politicians support conservation when the proposed reforms build on pre-existing land use trends that their constituents condone or support.
When this dynamic unfolds in tropical places, the reform-facilitating trends often have their origins in earlier processes of deforestation that inscribed social orders and use patterns onto tropical landscapes. When these deforesting social orders begin to disintegrate, they create political opportunities for preserving or restoring forests. To recognize and exploit these opportunities, observers need to understand the drivers of tropical deforestation during earlier periods. With these hypothesized links between past deforestation processes and contemporary conservation policies in mind, this paper characterizes in a broad interpretive way the changing historical dynamics of tropical deforestation during the 20th and 21st centuries. It begins by discussing the relative absence of historical work on tropical deforestation. Then it identifies the pivotal historical events that have separated the modern eras of tropical deforestation. Subsequent sections of the paper focus on the dynamics that characterized deforestation during each era: poverty traps in the colonial area, new land settlement schemes after independence, and finally, agri-business expansion during the neo-liberal area. The paper concludes with a section that links each of these deforestation processes to contemporary conservation initiatives whose design acknowledges and tries to benefit from situations created by the three deforestation dynamics.
THE HISTORICAL GEOGRAPHY OF TROPICAL DEFORESTATION
European geographers have long been astute observers of historical changes in forest cover. Mather and Needle (Reference Mather and Needle1998) and Mather (Reference Mather1992) conducted a series of studies of forest cover change in Europe during the 18th, 19th, and 20th centuries, and used this work to argue that a transition in forest cover trends from net deforestation to net reforestation occurred during the later stages of the industrial revolution in Europe. Williams (Reference Williams2005), another European geographer, set a standard for historical narratives of forest cover change. He discussed deforestation throughout the world, but his most detailed narratives depicted the dynamics of deforestation in European landscapes since the middle ages. The most pronounced of recent trends in forest cover have occurred in the tropics, but these trends did not receive detailed, case by case attention in the works of Mather, Williams, or other prominent historical geographers (Parsons Reference Parsons1969; Denevan Reference Denevan2001), so relatively little is known about the historical trajectory and dynamics of tropical forest cover change.
Social scientists also contributed to this knowledge deficit. Economists, sociologists and political scientists produced hundreds of analyses of cross-national variations in deforestation rates over recent, five- or ten-year periods of time (Angelsen & Kaimowitz Reference Angelsen and Kaimowitz1999). While useful for identifying the attributes of countries with high and low deforestation rates, these studies have not discussed changes over longer time periods in the historical contexts within which tropical deforestation has occurred. Rudel (Reference Rudel2007) distinguished historically between two complexes of drivers, one state led and more prevalent before 1990, and the other corporate led and dominant after 1990.
The present study extends this historical analysis back into the colonial era. By including colonial era deforestation, it is possible to assess the much debated relationship between impoverished peoples and deforestation (Sunderlin et al. Reference Sunderlin, Dewi, Puntodewo, Müller, Angelsen and Epprecht2008). Of greater import, the historical identification of three distinct complexes of drivers represents the first step in the argument outlined above, that deforestation's dynamics across a set of places during different historical periods sets in motion a series of landscape changes that sometimes can be exploited for conservation gains during a later period.
This argument outlines three ideal typical patterns of deforestation. These patterns feature combinations of proximate and underlying agents of deforestation, both the people who clear the land and the political-economic structures that encourage forest destruction. Each pattern is an ideal type. It represents a hypothetical pattern of forces, nowhere present in its entirety, which recurs across cases of landscape change during a particular historical period (Weber Reference Weber1949). Each ideal typical pattern should simplify and clarify the common elements that characterized deforestation during a particular historical period. At the same time, no one case of deforestation fully embodies the ideal type for that historical period.
The links between deforestation processes and conservation policies also have an ideal typical quality. They too fail to characterize any one case completely. For example, poverty traps imposed on smallholders by the latifundia–minifundia complex in Latin America never entrapped smallholders to the degree that occurred in the colonial societies of East Africa, because the impoverished upland populations in Latin America could, with considerable effort, migrate on foot to sparsely populated lowland regions with arable land (Carr Reference Carr2009). At the very least, the idealized links hypothesized here between deforestation dynamics and conservation opportunities should provide a heuristic for outlining conservation strategies for other times and other places.
The empirical plausibility of these macro-scale characterizations of deforestation's dynamics derives from the presence of underlying global-scale political-economic drivers during defined historical periods. The colonization of Africa, Asia and Latin America by the European powers between the 16th and 20th centuries represents one underlying driver. The creation of newly independent states after decolonization in the 19th and 20th centuries represents a second underlying driver. The globalization of markets for agricultural products from the 16th to the 21st centuries represents a third underlying driver. The salience of these drivers shifts over time. The colonial regimes were pervasive, with allowances for regional differences, until 1960. Newly-independent states predominated in Africa and South-east Asia during the 1960–1990 period. The globalization of markets for agricultural commodities, present at least since the 16th century, grew tremendously in scale after 1960.
These historical changes in the larger political-economic order account in large part for the shifts in the dynamics that have driven deforestation during the past century. Colonial regimes, which ended with decolonization during the two decades following World War II, confined indigenous peoples to marginal lands and, in so doing, created natural resource degrading poverty traps (McPeak & Barrett Reference McPeak and Barrett2001) in which poor peoples had such limited livelihood choices that they frequently had to deforest or further degrade already heavily disturbed forests. During the 1960s, the leaders of newly independent states trying to survive amidst the Cold War, worried about the integrity of their frontiers and tried to secure them by launching new land settlement schemes that provided land in remote regions to loyal settlers (Rudel & Horowitz Reference Rudel and Horowitz1993). In effect these programmes created ‘living frontiers’. The roads to these settlements increased access to urban markets and promoted the deforestation in these heretofore forested regions (Smith Reference Smith1982). After 1980, state-led efforts to clear land declined in frequency, while agribusinesses, always present during earlier eras, expanded and became the primary driver of deforestation in two (South America and South-east Asia) of the three continental scale tropical forests (Rudel Reference Rudel2007). The widespread adoption of neo-liberal economic tenets in developing world policymaking during this period promoted international trade which, as it grew, spurred additional land clearing by large-scale oil palm planters and cattle ranchers (DeFries et al. Reference DeFries, Rudel, Uriarte and Hansen2010; Aide et al. Reference Aide, Clark and Grau2013; Gasparri et al. Reference Gasparri, Grau and Gutierrez Angonese2013). During each of these periods proximate and underlying drivers differed from one another (Geist & Lambin Reference Geist and Lambin2002). Tenant farmers or large landowners cleared the land. They are the proximate agents of change. States and international trade flows are the underlying forces that compelled or induced the forest clearing. Several well documented cases help illustrate the spatial patterns of land clearing and their proximate as well as their underlying drivers in the historical geography of tropical deforestation (Table 1).
Table 1 Historical patterns of tropical deforestation during the 20th and 21st centuries.
The historical composition of these drivers has shifted over the past century (Fig. 1). The historical changes in the amounts of deforestation attributable to the different sets of drivers stem from rough estimates of the frequency of their occurrence in historical accounts of land-cover change during particular periods in the 20th and 21st centuries. A 2002 meta-analysis of the accumulated tropical deforestation studies (Rudel Reference Rudel2005) served as the source for these estimates.
Figure 1 The three eras of tropical deforestation.
Indigenous peoples and poverty traps during the colonial era
From the early 16th century until the mid-20th century, European colonial regimes took possession of distant lands and subjugated their inhabitants in an attempt to extract maximum advantages from these territories. The colonies provided land for European settlers, raw materials for European factories, and agricultural products for European consumers. Colonial authorities issued decrees prohibiting native peoples from engaging in shifting cultivation, as in Madagascar (Jarosz Reference Jarosz1993), requiring them to plant cotton, as in the Congo (Likaka Reference Likaka1997), and ordering them to work in mines, as in Bolivia (Frank Reference Frank1967). Colonial officials almost always allocated lands to people according to their standing in the imposed social order. Europeans commandeered the prime agricultural lands and relegated indigenous peoples to the marginal lands.
Disease shaped this process in the Americas. Within a few decades of the Spaniards’ arrival in Middle America, epidemic diseases carried by the new arrivals had decimated the Aztec population. The surviving indigenous populations clustered in the more mountainous, agriculturally less well-endowed lands of southern Mexico, which the Spanish were less likely to visit or occupy. These places became ‘regions of refuge’ for Amerindians (Aguirre-Beltran Reference Aguirre Beltran1979). Patron–client arrangements emerged at smaller geographic scales, creating a latifundia–minifundia complex (Stavenhagen Reference Stavenhagen1981) in which agrarian elites owned large expanses of land (latifundia), often in the well watered bottom lands of valleys, while less advantaged, poorer campesinos worked small rain-fed plots of land on hillsides (minifundia). A usufruct arrangement tied the peasants to the landlords. The elites owned the hillside land and allocated it to peasant families in return for the families’ labour on the latifundia’s bottom lands for several days each week. A wave of agrarian reforms in the early 1960s eliminated these feudal arrangements, but the overall latifundia–minifundia complexes persisted in many places despite the reforms (Grindle Reference Grindle1986). The expansion in cultivated areas in the 20th century extended the regions of refuge pattern. Only hillside and mountainous locations contained unclaimed lands that the growing numbers of smallholder households could occupy and farm (Kammerbauer & Ardon Reference Kammerbauer and Ardon1999). For example, smallholders in Costa Rica cleared forests in topographically-accentuated terrain in the central Cordillera in the 1970s because they could not acquire arable lands elsewhere in the country (Nygren Reference Nygren2000).
A somewhat similar dynamic unfolded in East Africa during the first half of the 20th century. Favoured elites confined large populations of indigenous peoples to marginal lands. For example, colonial officials in Kenya reserved the prime agricultural land on the well-watered high plateau in the Machakos district east of Nairobi for white colonists from the UK, and relegated Africans to lower elevation semi-arid lands further east, where they quickly degraded the meagre forest resources (Tiffen et al. Reference Tiffen, Mortimore and Gichuki1994, pp. 3–13). Across the border in present day Tanzania, a similar dynamic unfolded. European colonists occupied the well-watered lower slopes of Mount Meru, and, in so doing, pushed African smallholders onto more arid lands at lower elevations (Neumann Reference Neumann1998).
The clustering of poor people on marginal lands sometimes occurred in never colonized places. Ethiopia largely avoided colonial rule, with the exception of five years between 1936 and 1941, but, as in colonial settings, large populations of smallholders ended up farming marginal lands, gradually denuding the landscape through shorter fallows and clearing the few remaining forested lands far from roads (Getahun et al. Reference Getahun, Rompaey, Turnhout and Poesen2013). A similar pattern of natural resource degradation and deforestation, driven by the spread of smallholder agriculture, occurred in the accentuated topography of central Honduras before 1975 (Kammerbauer & Ardon Reference Kammerbauer and Ardon1999).
Smallholders in a region of refuge, a latifundia–minifundia complex, or on a reserve created by colonial officials are all at risk of getting caught up in a natural resource degrading poverty trap in which they destroy forests on marginal lands in order to meet the immediate needs of their families. The yields from these marginal lands are often low, and with continued use they deteriorate further, so cultivators in these places get caught up in vicious cycles in which they must destroy forests and degrade land that provides them with their livelihoods. Their poverty during an early period encourages the further degradation of natural resources that later on reinforces their poverty. They become trapped.
Certain patterns of land-cover change mark places where poverty traps drive deforestation. Patches of remnant forests get eliminated in lowland settings (Kammerbauer & Ardon Reference Kammerbauer and Ardon1999), and scrub growth spreads over extensive upland areas (McElwee Reference McElwee2009), so landscapes take on a disturbed or denuded appearance. Active deforestation fronts occur far from roads, in remote intermountain valleys or high on the slopes of ridges whose lower slopes have already been cleared for agriculture (Getahun et al. Reference Getahun, Rompaey, Turnhout and Poesen2013). In these circumstances poverty causes deforestation, but impoverished smallholders are only the proximate agents of land clearing. The underlying cause for deforestation, in most cases, has been a highly unequal social order, sometimes abetted by population growth (Blaikie & Brookfield Reference Blaikie and Brookfield1987). The social inequalities stem, most frequently, from colonial regimes that allocated the most agriculturally well-endowed lands to colonial elites and consigned smallholders to marginal lands where, to subsist, they have to clear forests from steeply sloped lands. While poverty trap driven deforestation continues to characterize some of the world's poorest societies, macro-sociological events like decolonization, growing urban labour markets, and increasing flows of migrants from rural to urban places have reduced the numbers of situations marked by this forest-destroying dynamic during the past forty years.
Colonists and new land settlement schemes during the 1960s and 1970s
Decolonization amidst a Cold War between a democratic capitalist block led by the USA and an authoritarian socialist block led by the Soviet Union changed both the proximate and underlying drivers of deforestation. For two decades, beginning in the 1960s new land settlement schemes, launched by national governments with encouragement from the USA and its allies, led to the deforestation of large areas of forest in Latin America and South-east Asia. The impetus for these programmes came from elites made insecure by the fledgling status of their states in a world torn by Cold War rivalries. Every newly-created state in South-east Asia faced some sort of rural insurrection during the 1950s and the 1960s. Usually, the rebel groups had an ethnic identity, operated in remote usually-forested regions, and espoused an ideology at odds with the ruling regime. The leftist ideological tilt of the rebel groups appealed to rulers in the communist bloc countries, and they sometimes provided material support to the rebels (Uhlig Reference Uhlig, Manshard and Morgan1988). In Latin America, the geo-political threat took a slightly different form. The states were old, having won their independence from colonial powers in the 19th century, but, like the new states in South-east Asia, they were also ‘weak’, meaning that they had only a minimal presence in remote rural regions. The police, if they were present, did not provide consistent protection. Local people did not participate in national politics, and could not be mobilized by the state to pay taxes or perform other civic duties (Migdal Reference Migdal1988).
When Fidel Castro's rural-based guerrillas succeeded in toppling the Batista regime in Cuba and establishing a leftist, and subsequently a communist regime in 1959–1960, it gave heart to impoverished peoples in Latin America, and the number of rural insurrections increased (Wickham-Crowley Reference Wickham-Crowley1992). In response, the governing elites in many countries, with support from American officials, moved to secure the remote rural regions in their countries through new land settlement schemes, sometimes referred to as colonization programmes. Modelled on the 19th century homesteading acts in the USA and Canada, these programmes were usually paired with agrarian reform efforts and publicized as a means by which the regime would correct the historical injustices of skewed land distributions that had concentrated landownership among small numbers of favoured families (Thiesenhusen Reference Thiesenhusen1995). The colonization programmes would provide landless peasants with small farms, 5–100 ha in extent, in a remote rural area. In addition the regime would plan villages and build roads into these regions, so as to insure the settlers’ access to urban markets.
Of most importance to the elites, the participants in the colonization schemes would come from loyal groups of people who had lived in more densely settled regions of a country. The participants often evinced a certain level of patriotism. In Ecuador, the colonists who settled near a disputed international boundary (with Peru) would in moments of excitement proclaim that ‘hacemos patria!’ (‘we are nation building’) (Salazar Reference Salazar1986). Oftentimes, the communities formed under state auspices would become growth poles, attracting spontaneous, as well as sponsored migrants from other regions of a country. For every migrant from Java to the outer islands in Indonesia sponsored by the state's transmigration programme, there were five spontaneous migrants from Java to the outer islands (Holden & Huoslef Reference Holden, Huoslef and Sandbukt1995). In this manner, sponsored settlements reduced forest cover both directly and, by encouraging more migration, indirectly.
In this instance, as with the poverty trap dynamic, poorer people were the proximate agents of deforestation. A large number of young landless persons, new to the remote areas, were willing to invest in the hard work of felling old growth tropical forests and cultivating the recently-cleared lands. While smallholders were the proximate causes of the deforestation, state support for forest destruction was the crucial underlying factor that drove this dynamic. In contrast to deforestation in the poverty trap dynamic, where land clearing occurred in patches and along the edges of isolated remnant forests far from roads, land clearing in new land settlement schemes occurred primarily along roads. To facilitate the marketing of agricultural products, colonists would clear land on those portions of their parcels closest to the roads, creating corridors of cleared land along the roads (Smith Reference Smith1982). In satellite images the deforested land looked like a ‘fishbone in the forest’. Corridors of cleared land on both sides of farm to market roads ended in another corridor of cleared land along a trunk road which led to distant urban centres. If the state did not build roads, farmers would not clear land because it became too difficult for them to transport agricultural products from their farms to markets. As suggested by these clearing patterns, state-led new lands deforestation differed from poverty trap deforestation in that growing consumer demand from urban areas and even overseas played a role in supporting the high prices that spurred agricultural expansion.
The larger political economic conditions that supported state-led deforestation collapsed during the 1980s. Government officials in tropical settings confronted debt crises (Cohen Reference Cohen, Blanchard and Fischer1992) in which lenders demanded that governments curb expenditures in order to receive additional loans from the International Monetary Fund (IMF). The high cost of new land settlement schemes per participant made them targets for budget cutting in this context. Programmes to open up remote forested regions for settlement were vulnerable to budget cuts for several other reasons. The insurgences had disappeared or declined in strength in all but a few places in Latin America and South-east Asia (Wickham-Crowley Reference Wickham-Crowley1994; Abuza Reference Abuza, Clad, McDonald and Vaughn2011), so the geopolitical rationale for the continued colonization of remote areas had lost its force. Growth in rural to urban migration had swelled urban populations, and, in accord with this change, political parties on the left gradually de-emphasized rural focused agrarian reform and colonization issues in favour of a more urban-oriented agenda focused on subsidies for basic goods like food and fuel. For these reasons state expenditures to open up new areas for smallholder settlement in rain forest regions declined substantially after 1985. These projects did, however, leave their mark in the form of populist landscapes inhabited by family farmers in places like Rondonia in the Brazilian Amazon and Kalimantan in Indonesia.
Large-scale agricultural enterprises, long distance trade, and contemporary deforestation
Growth in long distance, overseas trade has driven tropical deforestation for centuries. As early as the 16th century, the Dutch and Spanish colonists destroyed forests and established sugarcane plantations on coastal lands in the West Indies (Walters & Hansen Reference Walters and Hansen2013). Subsequent technological advances in transportation and refrigeration created further waves of expansion in tropical plantation agriculture. For example, the rapid growth in banana plantations in Central America during the early 20th century occurred after merchants figured out how to time the harvest of the fruit so that it would ripen as it was sold in North American cities (Soluri Reference Soluri2005).
The proportion of forests lost to corporate-led deforestation increased substantially during the last two decades of the 20th century. With growth in the purchasing power of middle classes in the developing world and continued demand for agricultural products from long affluent European and North American consumers, the overall demand for natural resources from the tropical biome grew substantially during the last decades of the 20th century. Large landowners, the proximate agents of deforestation in this pattern, moved quickly to meet the demand. Neo-liberal regimes and merchants who make markets provided important underlying supports for this pattern of land clearing. To comply with the conditions of debt relief agreements with the IMF, politicians in the developing world accepted the Washington consensus about the economic efficacy of neo-liberal policies and removed barriers to international trade (Babb Reference Babb2012). Entrepreneurs moved quickly to profit from these economic opportunities in the Amazon basin and insular South-east Asia. They took out loans from banks to log large tracks of forest and then converted these lands into oil palm plantations, cattle ranches and soybean farms (Kaimowitz & Smith Reference Kaimowitz, Smith, Angelsen and Kaimowitz2001; Defries et al. Reference DeFries, Herold, Macedo and Shimabukuro2013; Ferreira Filho et al. Reference Ferreira Filho, Bento and Horridge2014). Smallholders also participated in these processes of conversion. The loggers would take only the most valuable species of trees, leaving most trees standing, and degrading the remaining flora and fauna in the forests. Prospective farmers would then take advantage of the access to forested land provided by the new logging roads and set up labour-intensive small-scale agricultural enterprises like pepper farms along the roads. These follow-on farmers were particularly common in the outer islands of Indonesia during the 1980s (Vayda & Sahur Reference Vayda and Sahur1985).
When corporate enterprises became the primary proximate drivers of deforestation, the tempo of land clearing began to fluctuate with financial cycles. Although the magnitude of the spike in Brazilian deforestation rates in 1995 remains open to question, the cause for the increase appears to be related to a dramatic change in financial conditions in Brazil. During a period of accelerating inflation from 1991 to 1994 the prices for beef and for land in the Brazilian Amazon rose rapidly. Under pressure from the IMF, Brazil then adopted a series of austerity measures, and replaced the cruzeiro with a new currency, the real. With the introduction of the new currency and associated anti-inflation measures, interest rates plummeted from approximately 20% to 2% between 1994 and 1995. The low interest rates spurred ranchers and other agribusiness owners to take out loans to expand their operations which in turn led to the sharp increases in deforestation (Fearnside Reference Fearnside2005). Changing financial conditions also played a role in the decline in Brazilian deforestation rates after 2004. The robust Brazilian economy strengthened the value of the real which in turn placed Brazilian exports like beef and soybeans at a competitive disadvantage in world markets. The decline in the volume of agricultural exports in turn reduced the incentives for growers to convert more forests into fields. To be sure, expanded efforts to enforce the laws against deforesting more than 20% of a tract of land reduced forest losses, but so did the appreciating currency (Richards et al. Reference Richards, Myers, Swinton and Walker2012; Assuncao et al. Reference Assuncao, Gandour, Rocha and Rocha2013; Boucher et al. Reference Boucher, Roquemore and Fitzhugh2013). With corporate-led deforestation becoming more prevalent, the dynamics of deforestation had become ‘financialized’.
States continued to facilitate the agricultural expansion during this period through expenditures on infrastructure like penetration roads, paved roads and port facilities. To promote their interests, large landowners formed associations of growers like APROSOJA (the Soybean and Corn Growers Association of Mato Grosso) to lobby for state benefits like the Avanca Brasil programme of road improvements. According to the lobbyists, these expenditures would improve the country's position in international trade. For example, the Avanca Brasil programme of infrastructure improvements planned to pave a road from southern soybean-growing regions to ports on the Amazon River in order to improve the Brazilian growers’ access to the overseas trade in soybeans (Laurance et al. Reference Laurance, Cochrane, Bergen, Fearnside, Delamonica, Barber, D’Angelo and Fernandes2001). While the public sector improved the main roads, large landowners and logging companies built extensive networks of feeder roads after 1990 (Perz et al. Reference Perz, Overdevest, Caldas, Walker and Arima2007).
This change in the proximate drivers of deforestation registered in satellite images of active deforestation fronts. The corporate frontiers featured large blocks of cleared land, consisting of fields, often adjacent to one another, that large-scale farmers worked with machinery that they transferred from field to field (Morton et al. Reference Morton, DeFries, Shimabukuro, Anderson, Arai, del Bon Espiritu-Santo, Frietas and Morisette2006; Barona et al. Reference Barona, Ramankutty, Hyman and Coomes2010). This block pattern of deforestation contrasts with the small-scale clearings in remote places of the colonial era and the fishbone patterns of cleared land along state-constructed roads during the immediate post-independence era.
This corporate-led pattern of deforestation appeared most evident in two countries, Brazil and Indonesia, with large forests and numerous agri-businesses. The scale of the forest destruction in these two places distinguishes them from other nations experiencing forest losses. Brazil and Indonesia together accounted for 60.6% of the world's old growth tropical forest losses between 2000 and 2005 (Hansen et al. Reference Hansen, Stehman and Potapov2008). The large scale of the agricultural expansion and the rapid pace of forest destruction in the two countries stemmed in part from the corporate-led drivers of the process. These agri-businesses benefited from economies of scale and borrowed big sums of money to do things on a large scale. For these reasons they destroyed extensive amounts of forest. Not surprisingly, the volume of agricultural exports and the growth in the size of a country's urban populations proved to be the best predictors of national deforestation rates during this corporate led period of deforestation (DeFries et al. Reference DeFries, Rudel, Uriarte and Hansen2010).
Of the three deforestation processes discussed here, the corporate-led dynamic accounts for most recent forest losses in Asia and Latin America, but it does not explain recent patterns of deforestation in sub-Saharan Africa. Logging firms have been very active in the Guinean–Congolian forests of West and Central Africa, but smallholders remain the most active agents of deforestation in sub-Saharan Africa (Ernst et al. Reference Ernst, Mayaux, Verhegghen, Bodart, Christophe and Defourny2013), and rates of deforestation vary dramatically from wetter to drier biomes. In the wet Central African region mineral extraction has promoted urbanization and discouraged agricultural expansion everywhere, except in peri-urban zones around cities. In more arid zones deforestation has been more widespread (Rudel Reference Rudel2013).
As suggested by the recent dynamics in African deforestation, the association of a complex of proximate and underlying drivers with discrete places and historical periods can be taken too far. For example, poverty trap induced deforestation by poor peoples did not suddenly disappear with the end of colonial regimes in the 1960s. It persisted among some of the world's poorest people in densely-populated upland regions, like highland Ethiopia (Getahun et al. Reference Getahun, Rompaey, Turnhout and Poesen2013). Similarly, the first new land settlement schemes began before independence, with the Dutch in Indonesia (Uhlig Reference Uhlig, Manshard and Morgan1988) and the British in Zambia (Moore & Vaughan Reference Moore and Vaughan1994). Some of these programmes, notably in Indonesia, Malaysia and Brazil, persisted into the 1990s. Finally, large-scale agricultural enterprises and logging operations that either degrade or destroy tropical forests began in many instances during the late 19th century (Hill Reference Hill1963; Soluri Reference Soluri2005). Corporate-led deforestation became the predominant pattern of land clearing only during the last decades of the 20th century. Different patterns of deforestation co-exist during the same historical period, but their prevalence shifts between historical periods with changes in the larger political economy (Fig. 1).
THREE DEFORESTATION DYNAMICS, THREE CONSERVATION OPPORTUNITIES
Over the past two decades, ecologists have become increasingly aware of the biophysical legacies that episodes of land clearing leave in the soils and landscapes of deforested places (Foster et al. Reference Foster, Swanson, Aber, Burke, Brokaw, Tilman and Knapp2003). Geographers have noted how landscapes from earlier eras created cultural legacies that, at least in Europe, have contributed to recent conservation initiatives (Antrop Reference Antrop2005). A similar line of analysis, elaborated here, traces how the deforestation dynamics of earlier eras, through their influence on newly-created agricultural landscapes, generate opportunities for conservation at a later date. The following paragraphs sketch out the link between each one of the three deforestation dynamics and subsequent conservation opportunities.
Natural resource degrading poverty traps, rural to urban migrants, and forest restoration in the uplands
The forest-destroying dynamic associated with poverty traps during the colonial era confined poor people to marginal lands, frequently in upland settings where over time they destroyed the remnant forests. This dynamic became less common during the second half of the 20th century, when young people began to leave upland regions to search for work, primarily in cities. The first well-documented examples of rural to urban migration from tropical uplands followed by reforestation occurred in Puerto Rico, beginning in the 1950s (Rudel et al. Reference Rudel, Perez-Lugo and Zichal2000a ). Sometimes people relocated within the same locale. In Asian settings farmers refocused their agricultural production on lowlands and abandoned highland agriculture after new roads improved access to urban markets from roadside fields in the lowlands (Sikor & Truong Reference Sikor and Truong2002; McElwee Reference McElwee2009). The out-migration of young people from these upland settings where they had practised labour-intensive agriculture led predictably to the regrowth of trees on these lands. Elevation began to predict land-cover change in Latin America. Reforestation began to occur in the uplands, and deforestation continued to occur in the lowlands (Aide & Grau Reference Aide and Grau2004; Redo et al. Reference Redo, Grau, Aide and Clark2012).
The on-going expansion of secondary forests in upland settings may sequester enough carbon to make smallholders with titles to their lands eligible for payments under a PES mechanism like REDD+ (reducing emissions from deforestation and degradation). The payments could in turn induce additional reforestation on these lands. In the push-pull dynamic that has accounted for most rural to urban migration (Gibson & Gurmu Reference Gibson and Gurmu2012), the creation of natural resource degrading poverty traps by colonial regimes provided the ‘push’ that spurred massive out-migrations of poor people from these places when urbanization created more prosperous livelihoods elsewhere. Regrowth in various forms took over the abandoned lands and began, at least in small ways, to enhance ecosystem services. The prospect of payments for these environmental services stimulated widespread interest in Mexico's first PES programme among smallholders (Alix-Garcia et al. Reference Alix-Garcia, de Janvry and Sadoulet2005). The emigrants’ economic success in the cities may even contribute to the success of upland conservation programmes in the places that they left. The alternative incomes from urban employment have facilitated participation in PES programmes in the highlands of Ecuador by making it easier for small landowners to bear the costs of titling their lands (Bremer et al. Reference Bremer, Farley and Lopez-Carr2014).
The spread of conservation agriculture in old colonization zones
While poor rural peoples abandoned the deforested upland settings to escape poverty traps, the beneficiaries of state-sponsored new land settlement schemes usually tried to maintain a presence on their lowland farms which, despite declines in yields, remained valuable assets. Their longer-term commitment to the land inclined them towards the ecology of small-scale sustainable agriculture (Netting Reference Netting1993). Over time these smallholders became better acquainted with their lands and the trees in their forests. In theory, the growing knowledge should lead to new and more selective use of the natural resources on farms (Mather & Needle Reference Mather and Needle1998). The growing familiarity with both the economic possibilities and the environmental benefits from mixed uses on their lands has led to environmental conservation routines and the creation of subsidiary streams of income. For example, small-scale cattle ranchers (< 50 ha) in the Ecuadorian Amazon have over the past 15 years allowed spontaneously-generating commercially-valuable trees like the Ecuadorian laurel (Cordia alliodora) to grow in their pastures. The use of shade-resistant pasture grasses like gramalote (Axonopus scoparius) has allowed smallholders to pasture cattle and grow trees on the same tracts of land. After 15 years of tree growth, smallholders have been able to sell the wood and, in so doing, add a secondary stream of income to their household budgets. In addition, the litter from the trees provides organic material for the soils and, if enough carbon is sequestered from the growth of the trees, they too could become part of a REDD+ payments scheme. In brief, the state-led deforestation schemes of the 1960s and 1970s laid the foundations for the emergence of smallholder initiated, conservation oriented cattle ranching 30 years later (A. Lerner, unpublished data 2014).
Corporate-led deforestation, restrictions on agricultural credit, and the slowing of Brazilian deforestation
Because large enterprises make intensive use of credit to facilitate the conduct of business, the emergence of large enterprises as the pre-eminent drivers of deforestation after 1990 made banks a possible point of intervention in attempts to control deforestation. Short-term fluctuations in economic conditions began to affect the pace of deforestation in both South America and South-east Asia (Richards et al. Reference Richards, Myers, Swinton and Walker2012). When interest rates climbed and economies slowed down, large landowners would curtail plans to convert more forests into fields and deforestation rates would decline. Similarly, when interest rates declined and economic growth accelerated, landowners would take out loans and clear more land for cattle, so the pace of deforestation would increase. Deforestation rates also began to track short-term fluctuations in the price of beef in Brazil. Farmers would respond to rising prices for beef in one year by expanding the extent of their cattle pastures the next year (Fearnside Reference Fearnside2005).
To curb illegal deforestation under these conditions, Brazilian authorities implemented an enforcement scheme in 2005 that relied on credit restrictions and satellite imagery taken every two weeks. The imagery when combined with cadastres of the region could identify landowners who were out of compliance with Brazil's rules about the extent of cleared land on a farm. If more than a certain number of farms in a region were out of compliance with the deforestation law, then the state would suspend agricultural credit to all farmers in the municipality. This measure got the attention of large landowners because they relied on annual loans to finance their farming operations (Assunacao et al. Reference Assuncao, Gandour, Rocha and Rocha2013; Boucher et al. Reference Boucher, Roquemore and Fitzhugh2013). By making the ban on credit applicable to all farmers in a district, this penalty created peer pressure for compliance with the law. By 2010, deforestation in Brazil had declined to approximately 25% of the peak rate in 2004. The appreciation of the Brazilian currency most likely played a role in the decline of deforestation in Brazil, but so did the restrictions on credit. The financialization of deforestation processes had created conditions in which conservation could be effectively pursued through financial measures.
In sum, each of the three historically distinct drivers of deforestation links to a contemporary conservation opportunity (Table 2).
Table 2 Deforestation legacies, conservation opportunities: early processes of deforestation set in motion landscape changes that offer opportunities for conservation during later periods.
DISCUSSION
Ideal types (Tables 1–2 and Fig. 1) represent syntheses of field research, and thus they come with some intellectual risks. One involves differential coverage in the literature on land-use changes in different tropical regions. There are, for example, more accounts of deforestation in the Amazon basin than there are of similar processes in the Congo River basin (Rudel et al. Reference Rudel, Flesher, Bates, Baptista and Holmgren2000b ), so interpretations of historical patterns of tropical land cover change may rely more on the more exhaustively reported events in the Amazon basin. Inevitably, analysts, myself included, have more field experience in some settings than in others, and these experiences tend to shape judgments about the salience of particular historical narratives. While these considerations complicate syntheses of field research, the overall value of these syntheses needs to be assessed against the intellectual gains that flow from a generalized analysis that ties historically predominant processes of deforestation to subsequent strategies for conserving and restoring biodiversity.
Additional value could conceivably be extracted from this synthetic work by building models of the generalized processes of deforestation depicted here. Poverty traps in regimes that advantage colonial elites can be modelled. Similarly, the changing policy positions of states from the smallholder assistance phases of the 1960s to the neo-liberal regimes of the 1990s could be portrayed in models that project changes in agricultural area and corresponding pressures to convert forest land into fields. These models would explore the possibility that the historical dynamics discussed above might generate over time some land cover changes that have not been anticipated in the analyses here.
The link between earlier deforestation processes and later conservation policies varies with contingencies in political processes. When the earlier processes of deforestation set in motion land-use changes like the spontaneous regeneration of forest cover on fields abandoned by rural out-migrants, they create political opportunities for conservation advocates (MacAdam Reference MacAdam1982). For example, as in Mexico (Alix-Garcia et al. Reference Alix-Garcia, de Janvry and Sadoulet2005), smallholders who have scaled back their agricultural pursuits may be receptive to government or non-profit initiatives that would pay for the environmental services provided by the reforesting lands. Activists may not, for extraneous reasons, be ready to exploit these political opportunities, so the link between deforestation processes and later conservation policies is in no sense mechanistic. It depends on local political contingencies.
Effective forest conservation policies acknowledge and build upon variations from place to place in land-cover change. For example, REDD was originally conceived of as a programme for preventing the further destruction of old growth forests (Holloway & Giandomenico Reference Holloway and Giandomenico2009). The payments for avoided deforestation would go primarily to large landowners and indigenous groups with titles to large blocks of undisturbed forest. These stipulations make REDD seem like an appropriate policy for countering large-scale corporate-led deforestation in regions with large blocks of rain forest at risk. REDD would not address deforestation dynamics on the populist frontiers established by new land settlement schemes where smallholders have already cleared a high proportion of the old growth forests from their lands. The recently introduced REDD+ would compensate landowners for the carbon sequestering on their lands through regrowth on abandoned agricultural lands. It represents what proponents call a ‘whole landscape’ approach to reducing carbon emissions by promoting carbon sequestration in predominantly agricultural landscapes as well as in largely forested landscapes (DeFries & Rosenzweig Reference DeFries and Rosenzweig2010). ‘Whole landscape’ initiatives would apply to belts of smallholder peri-urban agriculture in sub-Saharan Africa, as well as to other lands characterized during earlier historical periods by entrapped rural poor and new land settlement schemes (ASB-ICRAF [Alternatives to Slash and Burn-International Center for Agro-Forestry] 2013). In this and other instances, effective forest policies would seem to begin with an appreciation of historical differences in land clearing processes across regions.
CONCLUSION
A masterful study of the destruction of Brazil's Atlantic Forest ended with these words: ‘As usual in the history of the human relation to the natural environment, effective countermeasures began to be taken only after an immense, undeniable disaster had occurred’ (Dean Reference Dean1995, p. 347). This essay argues similarly, that in the debris of a disaster, the seeds for regeneration can be found. While Dean (Reference Dean1995) certainly was correct to underline the importance of the magnitude of the disaster for the recovery that follows, this essay argues for a more organizational or structural link between environmental decline and eventual recovery. The human organization that destroyed tropical forests unwinds at a later date in response to a changing historical context (Table 2). The children of impoverished smallholders who cleared upland forests move to cities and relieve some of the human pressure on upland landscapes. The children of smallholder cattle ranchers in old colonization zones allow a modicum of reforestation to occur on their lands as they try to diversify their sources of income by growing and selling trees from silvo-pastoral landscapes (A. Lerner, unpublished data 2014). Finally, the heads of large enterprises deforest with their eyes on ‘the bottom line’ of a financial sheet. They become cognizant of the environmental damages from their operations and curtail them only when states threaten to alter their bottom lines, by denying them credit for agricultural operations. In this instance, and in those involving poverty traps and new land settlement schemes, effective conservation strategies begin with a detailed appreciation of the dynamics of deforestation, either past or present. In this sense, the dynamics of deforestation would appear to contribute over time to the emergence of conservation opportunities.
ACKNOWLEDGEMENTS
I would like to thank Heidi Hausermann and Pamela McElwee for valuable comments on an earlier draft of this manuscript. Mike Siegel helped with the figure. Princeton University's STEM programme provided an audience who asked some searching questions that helped me formulate the arguments presented here.
