Philip M. Fearnside. Ph. 0 . ISSN 0034-7108 ln~tituto Naci orial de Pesquisa1 Ja Amazuni a UN PA) ~.:;.' Caixa Postal 4 7 8 :; 69011 M" naus. Ama2GM• ··;, EJR~. Slt... :~3~~.tA • · '-.\ -~.. :~ ...... , REVISTA BRASILEIRA DE B10LOGIA

( '

I1

MCMXVI

ACADEMIA BRASILEIRA DE CIENCIAS 1997 • :R~ista; ·:ai:~s.:Ue.:tra; :de ~OlCXJia 57 (41: 531-549.

LIMITING FACTORS FOR DEVELOPMENT OF AGRICULTURE AND RANCHING IN BRAZILIAN AMAZONIA*

PHILIP M. FEARNSIDE** Instituto Nacional de Pesquisas da Amazonia (INPA), C.P. 478, 69011-970 Manaus, Arnazonas, Brazil (With 3 figures)

ABSTRACT Limiting factors restrict both the intensification of agriculture and ranching uses and the scale to which these land uses can be expanded. The expression of limiting factors on development is mediated through human planning: the perception of. limits to agricultural yields and of the severity and probability of environmental impacts can lead to decisions to limit agricultural expansion. Limits to intensification of agriculture include agronomic limits on per-hectare yields, technological limits and research, and cultural limits. Limits to expansion of agricul­ tural areas include physical resource limits such as phosphate deposits, limits of social values, institutional limits (including the credibility of institutions), limits on human habitation (such as health), and limits to environmental risks. Limits from considerations in political and mili­ tary spheres often override "rational" decisions based on land capability and environmental consequences. This kind of "interference," however, can cause a variety of impacts that, if properly evaluated, would likely make the net result of such development projects a negative one for Brazil's national interests. Key words: carrying capacity, limiting factors, Amazonia, deforestation, cattle ranching, eco­ nomic development.

RESUMO Fatores limitantes ao desenvolvimento da Agricultura e Pecuaria na Amazonia brasileira

Os fatores limitantes restringem a intensifica~iio do uso da agropecuaria e a escala no qual esse uso da terra pode ser expandido. A expressiio dos fatores limitantes sobre o desen­ volvimento e mediada atraves do planejamento humano: a percep~ao dos Iimites para a pro­ du~iio agricola e da severidade e probabilidade dos impactos ambientais pode levar as decis6es que venham Iimitar a expansiio agrfcola. Os limites para a intensifica~ao da agricul-

Received July 9, 1997 Accepted July 30, 1997 Distributed November 14, 1997 •Invited paper. •*Foreign Member of Academia Brasileira de Ciencias. E-mail: [email protected] Tel: 55 (92) 642-3300 (Ext. 1822) I Fax: 55 (92) 236-3822

Rev. Brasil. Biol., 57 (4): 531-549 532 PHILIP M. FEARNSIDE

tura incluem limites agronomicos sobre a prodm;ao por hectare, limites tecnol6gicos e de pes­ quisa, e limites culturais. Os limites para a expansao de areas agricolas incluem limites de re­ cursos fisicos tal como as reservas de fosfato, limites de valores sociais, limites institucionais (incluindo a credibilidade das institui<;6es), limites de habita<;iio humana (tal como saude), e limites aos riscos ambientais. Os limites das considera<;6es nas esferas politicas e militares frequentemente superam as decis6es e as consequencias ambientais. Esse tipo de "interferen­ cia", entretanto, pode causar uma gama de impactos que, se devidamente avaliados, provavel­ mente tomaria o resultado final de· tais projetos de desenvolvimento negativo para os interesses nacionais do Brasil. Palavras-chave: capacidade de suporte humano, fatores limitantes, Amazonia, desmatamento, agropecuiiria, desenvolvimento economico.

INTRODUCTION: Among the types of limiting factors affect­ LIMITING FACTORS ing plant growth in general, including agriculture in Amazonia, are soil nutrients, water, light, carb­ Kinds of limiting factors on dioxide, and attacks of pests and diseases. In Justus Liebig (1840) observed that plants re­ agricultural systems, limiting factors can be quire certain chemical substances from soil and pushed back (up to a point) by outside inputs of that they cannot grow unless a minimal quantity energy, fertilizers and other items that depend on of each is present. The nutrient in short supply availability of capital. therefore limits growth of the plant - a principle The use of agricultural inputs from outside that guides theory in ecology and the practice of makes explicit definition of limits on these inputs applying fertilizers in agriculture to this day. This (and consequently on human carrying capacity) "law of the minimum" was expanded by Black­ necessary if consideration of limiting factors is to man (1905) to include the limiting effects of the be meaningful. Outside inputs of energy, fertiliz­ maximum, and by Shelford (1911) to include the ers and initial capital can circumvent local lacks. modifying effect of the tolerance of organisms At the extreme, one could even have agriculture on the moon if importing everything were permit­ when responding to minimum or maximum limits. ted. Practical limits in importing inputs are likely Shelford's "law of toleration" completed the con­ to be determined by markets for the agricultural cept of limiting factors as generally applied today. products exported from the system and on the fi­ Most use of limiting factors focuses on the nit_e nature of available stocks of inputs in source effects of a single factor, and, indeed, the reduc­ areas. These limits are critical because of Ama­ tion in complexity that this approach allows is one zonia's vast area: 5 x 106 km2 in Brazil's "Legal of its greatest strengths in making sense of an oth­ Amazon" administrative region (Fig. 1). erwise confusing array of possible influences. Limiting factors on Amazonian development However, single factors rarely limit organisms or can be grouped into a series of classes, of which populations in the real world. Hubbell argued agronomic limitations represent only one. Also strongly against the "spate of single-factor answers important are institutional limitations, limits im­ in the last 20 years" (1973: 95). He suggests in­ posed by social values, and by social mechanisms stead that "several factors may act simultaneously, or customs, human living conditions, and by the conceivably equally" in limiting populations at macroeconomic context of development. Salati et any instant in time. Not only can production be al. (nd) have examined the question of limiting co-limited by several factors, but synergisms factors for sustainable development in Amazonia, among factors can result in greater production in­ grouping limitations into those affecting biologi­ creases when certain combinations of factors are cal productivity and those in the institutional supplied together. The best example is water and sphere. These authors point out the need for policy nitrogen limitation of rangeland production in the makers to understand that nature poses limits to Sahel region of Africa (see review by Hall, 1990). agricultural production, and emphasize the poten-

Rev. Brasil. Biol., 57 (4): 531-549 AGRICULTURE AND RANCHING IN BRAZILIAN AMAZONIA 533

MATO GROSSO

~ CALHA NORTE 00888® PICO DA NEBLINA N. P. 0 IOO 300 SOO km !we - ... ~~~~j POLONOROESTE

Fig. 1-Brazil's Legal Amazon region. tial for improving efficiency by making better use impediments imposed by social values and of indigenous knowledge and by institutional macroeconomic factors. streamlining through participation of organiza­ tions acting at different levels, ranging from the Expression of limiting factors micro (grassroots) to the macro (public policy) level. Mediation of Planning Limits on development act in two ways: lim­ Assessing limiting factors for development its on per-area production imposed by restrictions is fundamentally different, in some ways, from as­ on the intensification of use on any given hectare sessing limiting factors for a population of organ­ of land, and restrictions on the area to which the isms in a natural ecosystem or in a farmer's field. activity can expand, as dictated, for example, by Development limits are mediated by human intel­ the area "available." Some of the factors are inter­ ligence, through the mechanisms of planning (in­ changeable between the two types of restrictions, cluding zoning) and a variety of regulatory such as fertilizer, labor, and capital between inten­ hurdles (including environmental impact assess­ sifying production per hectare and expanding the ments). If poor soil will lead to failed crops, or if number of hectares. Some factors are not inter­ clearing will lead to soil erosion and land degrada­ changeable, as they do not involve allocation be­ tion, then perception of these limits can lead to de­ tween competing options; examples include cisions not to cut forest for agriculture. Rather

Rev. Brasil. Biol., 57 (4): 531-549 534 PHILIP M. FEARNSIDE

than simply allowing settlers to encounter the garis), cowpeas (Vigna sinensis), bitter and sweet physical limits for themselves, an additional layer manioc (Manihot esculenta), black pepper (Piper of information, value judgments and decision­ nigrum) and cacao (Theobroma cacao). The domi­ making is interposed between the physical limita­ nance of pH tends to exaggerate its implied role in tions and the decisions on development. What, for maintaining crop yields over time. The pH may be example, are the acceptable risks of crop failures? kept at a reasonable level by frequent burning of What are the acceptable risks of major environ­ agricultural fields, but other limits, such as lack of mental catastrophes? These and a wide variety of organic matter and the various cations associated other considerations will determine the limiting with it, are likely to appear within a few years factors applicable to development of agriculture (Feamside, 1986a). Rice (Oryza saliva) is limited and ranching in Amazonia. by pH in combination with organic matter, alumi­ num ions and total phosphorus (Feamside, 1986a). Effect of Variability In the case of cattle pasture, available phosphorus Variability of any factor, rather than only limits production (Serrao & Falesi, 1977; Feam­ mean levels, is an essential aspect of limitation. side, 1979a). The yields of all annual crops de­ Soil phosphorus, rainfall, money or other necessi­ pend not only on soil fertility but also on planting ties vary greatly in space and time. The ability to density, interplanted crops, and attacks of pests survive shortfalls of a given factor depends, in and diseases. part, on such mechanisms as a monetary cushion (i.e., cash, or an equivalent reserve such as cattle, Technological limits and research that can be called upon when needed). In tradi­ The temptation is always strong to believe tional agricultural societies, cash and storage of that research will remove virtually all limits to de­ value and products are rare, but forms of insur­ velopment, and nowhere are such flights of imagi­ ance are provided by sharing among members of nation freer to run wild than in Amazonia. kin or community groups, and by what Allan Because of the region's vast area, assumptions of (1965) has termed the "natural surplus of subsis­ huge areas of intensive agriculture in Amazonia tence agriculture." This mechanism involves are the most important factor underlying fantastic planting more area of each crop than would be global conclusions on the earth's carrying capacity needed for the mean yield to supply subsistence for humans. This may be assumed to be a part of needs, thereby providing a cushion against short­ then-US president Ronald Reagan's belief that falls. "farm studies" had shown that the world could support 28 billion people if other countries could LIMITS TO INTENSIFlCATION be persuaded to cultivate their tillable land at the intensity found in the United States (Holden, Agronomic limits on per-hectare yield 1980: 989). Liebig's "law of the minimum" is applied in It is easy for planners to convince them­ agriculture in the form known as the "linear re­ selves that crop yields can increase indefinitely, sponse and plateau model." Crop yield response and that they can increase at ever increasing rates. below a sufficient or "check" level of each nutri­ For example, when the Brazilian Enterprise for ent is assumed to be linear, while no response oc­ Agriculture and Cattle Ranching Research (EM­ curs above this threshold. This model has been BRAPA) was founded in 1974, a series of projec­ shown to be adequate for practical purposes for tions of per-hectare yields were made for different agronomic work on tropical soils, with only mar­ crops assuming linear, logarithmic and exponen­ ginally better results being obtained using quad­ tial growth in yield (Paez & Dutra, 1974). Re­ ratic functions (Waugh et al., 1975). The linear cently, Gallopin and Winograd (1995: 27) arrived response and plateau model has been applied to at a rosy conclusion regarding the prospects for a crops used by colonists on Brazil's Transamazon "sustainable scenario" by assuming that per-hec­ Highway (Feamside, 1986a). Because of acid soil tare yields of crops will increase exponentially at in most of the area (and in most of Amazonia), pH 1.5-2.0%/year (see Feamside, 1996a). The idea is most often the limiting soil character, for exam­ that exponential growth is an option is misleading, ple, for maize (Zea mays), beans (Phaseo/us vu/- and the notion that one can select it as if picking

Rev. Brasil. Biol., 57 (4): 531-549 AGRICULTURE AND RANCHING IN BRAZILIAN AMAZONIA 535 something off a shelf is even more dangerous. In should the Yurimaguas technology be widely im­ reality, Brazilian per-hectare yields have tradition­ plemented. While these services have been pro­ ally been almost constant, increases in the total vided free of charge (i.e., as a subsidy) by NCSU harvest coming instead from expansion of areas in the case of the farmers collaborating with the under cultivation (Paiva et al., 1976: 62-68). Yurimaguas experiment station, either farmers, The best-known attempt to push back the taxpayers or consumers in Amazonian countries limits of soil restraints on sustaining agricultural would have to bear these expenses should the sys­ production through time is the project to develop tem be expanded. When Sanchez et al. (1982) continuous cultivation undertaken by North Caro­ published the first results of the experiment, the lina State University (NCSU), in conjunction with eight-year-old experimental plots required - in ad­ Peruvian institutions, at Yurimaguas, Peru dition to N, P and K - replacement of five other (Sanchez et al., 1982; Nicholaides et al., 1985; see nutrients: magnesium, copper, zinc, boron and Feamside, 1987, 1988; Walker et al., 1987). Soil molybdenum. Three years later, two more nutri­ depletion is a fundamen,tal problem that becomes ents had been added to the list: sulfur and magne­ increasingly expensive and problematic to correct sium (D. Bandy, personal communication, 1985). as time proceeds under continuous cultivation. All The research group complained about the diffi­ nutrients removed in harvested crops or lost culty of obtaining adequate purity in the soil sam­ through erosion, leaching, and other processes can ples and sufficient precision in the laboratory be replaced in the form of fertilizers. The cost of analyses: with micronutrients, a difference of only replacing them must include not only the substan-. a few parts per billion can have a large impact on tial expense of purchasing fertilizers and trans­ crop yields. The difficulty of obtaining such preci­ porting them to the site, but also the expense of sion would undoubtedly be much greater for farm­ identifying which elements are missing, and in ers handicapped by geographical isolation, lack of what amounts, for each field, and communicating education, and a tenuous link to laboratory facili­ this information to the farmer in time to allow cor­ ties through a chain of often poorly trained and rection of the deficiencies before yields are af­ poorly motivated extension personnel. Any error fected. The principal macronutrients (nitrogen, or delay in calculating the correct mix of fertiliz­ phosphorus and potassium), together with lime, ers can cause losses in yields. Sanchez et al. account for most of the expense of purchase and (1982: 824) admit: "In the complete treatment, transport. Sanchez et al. (1982: 825) state that the fertilizers and lime were added according to rec­ quantities of fertilizer needed to supply these ele­ ommendations based on soil analysis. During the ments are similar to those used by farmers in the second or third year, however, yields began to de­ southeastern part of the United States. While this cline rapidly. Soil analysis identified two possible fact might seem to imply that agriculture could be factors ... lime, and ... magnesium." If yields can as profitable in the Amazon as in the Carolinas, it suffer from misassessment of nutrient needs in an should be remembered that long distances make experimental plot closely monitored by a highly the cost of fertilizer much higher and prices re­ qualified team of research agronomists, one would ceived for crops are much lower than is the case expect such crop yield declines to be much more elsewhere. frequent in the fields of Amazonian farmers. Fail­ Correction of micronutrient depletion, while ures would be higher because of the uncertain requiring only small weights of imported fertil­ supply of inputs and information about which in­ izer, would add substantially to the cost and risk to puts are needed. farmers practicing the system. Nutrients must be Separate information is needed for each field balanced to avoid detrimental synergisms. The in order to make the system work properly. system requires analysis of soil and plant samples Sanchez et al. (1982: 824) state that "the timing of after each crop in order to calculate the proper nu­ the appearance of soil fertility limitations and the trient mix for fertilization. It is awesome to imag­ intensity of their expression varied among the [3 ine the expansion of laboratory and extension test] fields, even though they were near each services that would be necessary to handle the other, were on the same soil mapping unit, and millions of samples that would be generated had the same vegetation before clearing. The in-

Rev. Brasil. Biol., 57 (4): 531-549 536 PHILIP M. FEARNSIDE

tensity of the fire during clearing is considered a in point is the decision of the Cotrijui cooperative factor contributing to this variability." The diffi­ in 1986 to give up on a proposed silkworm-raising culty of getting results in time for appropriate cor- initiative in Mato Grosso because, according to rective measures when only a few experimental officials of the cooperative, settlers in the area fields are involved should give some indication of were considered to have insufficient education to the magnitude of problems that would be faced by master the technology. farmers in trying to extract such results from gov­ ernment-run soils laboratories and their associated LIMITS TO EXPANSION bureaucracies. A parallel situation may be found in the difficulties of farmers on Brazil's Transama­ Physical resource limits zon Highway in obtaining the release of credit dis­ bursements at the appropriate time in the Aside from limits of space, physical re­ agricultural year (Moran, 1981). sources, such as phosphate deposits, restrict the areas to which different land uses can expand. Cultural limits Land-use decisions embody value judgments as to The idea that Amazonia might one day re­ what use should be made of different pieces of semble the valleys of the Ganges or the Yangtze, land. When decisions are made at a regional scale, with dense human populations supporting them­ as in the economic-ecological zoning now under­ selves on irrigated rice, points both to the impor­ way in Brazil, the scale of the areas falling into tance of cultural limitations and to the great different uses results in emergent properties that leeway that exists between physical limits and re­ are not thought of at a micro level. A formula for straints on human populations. Amazonia is not making these decisions is not easy to arrive at in a likely to undergo such a transformation in the way that guarantees "wisdom" in balancing the foreseeable future if only because the population different roles that each location might play in the that inhabits the region would have to undergo ecosystem and in society. The tendency of zoning radical cultural changes to make the backbreaking methodology up to now, for example, has been to routine of transplanting rice attractive. It is un­ evaluate the different restrictions on agricultural likely that Amazonian agriculture will change choices imposed by soil quality, topography, rain­ through transfusions of immigrants from other fall, and other physical factors. The most fertile places where more intensive methods are already sites are then allocated to the most intense uses, part of the cultural tradition. The Japanese immi­ while those with little agricultural potential are grants at Tome-Ai;u provide a good example zoned as forest reserves. (Fearnside, 1980a; Subler & Uhl, 1990), as do The need for caution in designating areas for Gauchos brought to the Transamazon Highway agriculture is illustrated by the question of agricul­ from the extreme south of Brazil by INCRA with tural expansion in Acre. Preliminary zoning maps the mistaken expectation that they would serve as produced by EMBRAPA indicate large areas for a model for colonists from other parts of the coun­ agriculture, including the western two-thirds of try (Moran, 1981). Acre (Brazil, EMBRAPA, 1988). A preliminary Cultural change leading to intensification or zoning by the Brazilian Institute for Geography disintensification of agriculture has been the sub­ ·and Statistics (IBGE) differs from that of EM­ ject of much discussion among social scientists BRAPA on its recommendations for most of Ama­ (reviewed by Brookfield, 1972). Boserup (1965: zonia, but agrees that western Acre should be used 62-63) pointed out the tendency of people who for agriculture (Regis, 1989). This area's soils are migrate from densely to sparsely populated areas believed to be better than those of most of Ama­ to abandon the intense methods formerly em­ zonia, at least at a very general scale. Acre is also ployed in favor of extensive methods. This has the best area for establishing extractive reserves, often occurred in Amazonia, for example, among both because of relatively high densities of forest colonists who arrived on the Transamazon High­ trees that produce valuable products and because way from other parts of Brazil in the 1970s. of better social organization among the forest's Educational limitations restrict the types of human inhabitants (Allegretti, 1990; Fearnside, agricultural options that can be implanted. A case 1989).

Rev. Brasil. Biol., 57 (4): 531-549 AGRICULTURE AND RANCHING IN BRAZILIAN AMAZONIA 537

Land-use decisions based on permitting the source. The temptation is strong to view Ama­ maximum intensity that physical conditions will zonia as a potential cornucopia capable of solving allow can quickly pass limits in other spheres population and land distribution problems; the when individual zoning allocations are considered limits of applying the intensive agriculture sug­ together. Zoning is more than the sum of its parts. gested make this a cruel illusion (see Fearnside, Misleading results can be expected from zoning 1990a). These limits are best illustrated by the in­ methods that fail to include provisions for limita­ viability of applying to any significant part of tions of various types on the area that can be allo­ Amazonia the Yurimaguas technology for continu­ cated for any given land use. One may examine ous cultivation discussed previously. each cell in a grid in a geographical information One of the factors leading to the high carry­ system (GIS), comparing the soil, rainfall, etc., ing capacity values the study ascribed to Ama­ with the demands of a given crop, and conclude zonia is the assumption that land quality in that each individual cell can be allocated to the uncultivated areas is equal to that in already culti­ use in question, and yet arrive at ..a global conclu­ vated ones. The study goes so far as to claim that sion that is patently unrealistic. This, for example, "there is evidence that the productivity of the re­ is the explanation of the conclusion that Brazil serves may be higher, but, for the sake of simplic­ could support seven billion people, reached by a ity, it is assumed that the potential productivity of study conducted by the Food and Agriculture Or­ the unused land is the same as that of the land un­ ganization of the United Nations (FAO), in col­ der cultivation" (FAO, 1984: 43). Unfortunately, laboration with the United Nations Fund for as is true in most parts of the planet, the best land Population Activities (UNFPA) and the Interna­ is brought into cultivation first, with land quality tional Institute for Applied Systems Analysis (11- progressively declining in new settlement areas ASA) (FAQ, 1980, 1981, 1984; Higgins et al., until only very marginal lands remain. In Ron­ 1982). donia, for example, 42% of the land in coloniza­ It is worthwhile examining the FAO/UN­ tion projects settled in the 1970s was classified by FPNIIASA study in some detail, as the illusion a government soil survey as "good for agriculture embodied in it that Amazonia can be turned into a with low or medium inputs;" for projects started in major breadbasket - an idea that long predates the the first half of the 1980s, 15% of the land was so classed, while for planned areas the amount is a FAO/UNFPNIIASA study - is a persistent and minuscule 0.13% (Feamside, 1986b). pernicious one in Brazilian planning for the re­ gion. The study's results contain numerous glar­ The FAO/UNFPNIIASA study's implied ing inconsistencies with reality, indicating that recommendation that developing countries should such efforts need to be based on more ground encourage migration into tropical lowlands from truth. The Brazilian Amazon is all mapped in the more highly populated areas at higher altitudes FAO/UNFPNIIASA study as capable of support­ and/or latitudes (e.g., FAO, 1984: 21) is likely to ing between one-half and one person per hectare prove an environmental catastrophe, as similar at the present low-input level of technology, and programs already have in such countries as Co­ between five and ten people per hectare with high lombia, Ecuador, Bolivia and Indonesia. The sug­ inputs (fertilizers, mechanization and an optimal gestion that "fragmented" land holdings must be mix of rain-fed crops). These calculations lead to grouped into "consolidated" properties as part of the conclusion that Brazil could support an incred­ the transition to high-input agriculture (FAO, ible 7.1 billion people, were high-level inputs ap­ 1981: 16) would play havoc with the social func­ plied (Higgins et al., 1982: 104). The implied tion of many tropical settlement programs. The possibility of converting the region to high-input conclusions on human carrying capacity are af­ mechanized agriculture runs up against limits of fected by the study's lack of consideration of eq­ resource availability to supply the inputs. Ama­ uity in the distribution of food produced, in zonia has virtually no deposits of phosphates; addition to the active anti-equity bias of the land transporting them is expensive and, when the vast tenure system the report implicitly recommends. extent of Amazonia is considered, quickly enters The necessity of phosphate fertilizers makes into conflict with the absolute limits of this re- prospects poor for maintaining the most common

Rev. Brasil. Biol., 57 (4): 531-549 538 PHILIP M. FEARNSIDE

land use in deforested areas - cattle pasture - over substantial portion of Amazonia to fertilized pas­ large areas in Amazonia. In the early 1970s when ture would greatly hasten the day when stocks of the fiscal incentives program for Amazonian pas­ phosphate are exhausted in Brazil and in the tures was rapidly expanding, the agency that is world. Brazil would be wise to ponder carefully now EMBRAPA maintained that pasture im­ whether its remaining stocks of this limited re­ proved the soil (Falesi, 1974, 1976). Unfortu­ source should be allocated to Amazonian pastures. nately, available phosphorus declines sharply from Large expanses of pasture can be expected the peak caused by ash from initial burning of the to be subject to disease and insect outbreaks in the forest; after ten years, levels of this critical ele­ same way as other large monocultures. Switching ment are at least as low as those under virgin for­ the grass varieties planted can counter such prob­ est and far below the amounts required by pasture lems to some extent, but the cost and frequency of grasses (Feamside, 1980b; Hecht, 1981, 1983). In such changes can be expected to in­ 1977 EMBRAPA changed its position that pasture crease. Brachiaria decumbens (braquiaria), a pas­ improves the soil, recommending instead that pro­ ture grass formerly common on the ductivity be maintained by applying annually 50 Belem-Brasilia Highway, was devastated in the kg/ha of phosphorus, equivalent to about 300 early 1970s by outbreaks of the homopteran kg/ha of superphosphate (Serrao & Falesi, 1977; known as cigarrinha (Deois incompleta Ceropi­ Serrao et al., 1979). The much greater productiv­ dae ). Guinea grass or coloniiio (Panicum maxi­ ity of pasture when fertilized with phosphate is mum) became a favorite in the area, and its obvious (Koster et al., 1977). The problems are performance was described by EMBRAPA as the cost of supplying phosphate and the absolute limits to minable stocks of this mineral. A report on Brazil's phosphate deposits pub­ lished by the Ministry of Mines and Energy indi­ cates that only one small deposit exists in Amazonia, located on the Atlantic coast near the border of Para and Maranhiio (de Lima, 1976) ,~ (" ' , R ' """~ : ICEARA,GRAN E (Fig. 2). In addition to the deposit's small size, it PAR A ;r,'. «--'?' :, \ (-,DQ.j\IO~TE has the disadvantage of being made up of alumi­ / ; ~'?" ,,,, /--', ~PA,~AIB€-_ '; ''-; /~Aui /PEiNAMBlJcct num compounds that render its agricultural use j ~I ,,.1, ; '-, ...... ,. _. ... ------/' ~ \J ("_...... ,/ \'~,LA)J AS suboptimal, but not impossible if new technolo­ I ..?- ,-, ___ , (SERGlPE gies were developed for fertilizer manufacture I J I \ MATO \ J....o ~' B A H I A (dos Santos, 1981: 178). An additional deposit has GROSSO ,~l I I \ .... , been reported on the Rio Maecuru, near Monte I GOIAS ,,.\_,/ v- r' DF f ... _, Alegre, Para, but estimation of its size is incom­ ,/ ci ( M I N A S ~--, plete (Beisiegel & de Souza, 1986). Almost all of .-f I I < ----- '4 A ,' GER A IS ,,, \... ,. .... --a, I ' Brazil's phosphates are in Minas Gerais, a site MATO '-,,_,,-- A JES IRITO very distant from most of Amazonia. Brazil as a GROSSO ,>--_:-,,,---, A /iSANTO whole is not blessed with a particularly large stock DO /SAO ',_ ,''· I ' ' of phosphate - the United States, for example, has SUL A (,_,/lflO D JANEIR:HOSPHATE deposits about 20 times larger (de Lima, 1976). :~:~~\~.r~ On a global scale most phosphates are located in sliNTA ___ _ A DEPOSITS Africa (Sheldon, 1982). Continuation of post­ ----,CATA INA World War II trends in phosphate use would ex­ I 0 ',_ O DO 300 !DJ km GRANDE ? .... M haust the world's stocks by the middle of the next DO century (Smith et al., 1972; United States, CEO & SUL Department of State, 1980). Although simple ex­ trapolation of these trends is questionable because Fig. 2 - Phosphate mines and deposits in Brazil (from de of limits to continued human population increase Lima, 1976). The deposit in the Maecuru River valley is from at past rates (Wells, 1976), the conversion of a Beisiegel & de Souza, 1986.

Rev. Brasil. Biol., 57 (4): 531-549 AGRICULTURE AND RANCHING IN BRAZILIAN AMAZONIA 539

"magnificent" (Falesi, 1974). Yield declines later perennial crops, such as cacao) that might other­ became apparent as available phosphorus deple­ wise be desirable choices from the standpoints of tion and invasion of weeds proceeded. Weed inva­ sustainability and environmental impact. sion in Panicum maximum is facilitated by the Market limits, reflected in falling cacao bunchy growth habit of this species, which leaves prices since 1977, make the advantages of cacao bare spaces between the tussocks of grass and by (e.g., Alvim, 1981) unlikely to continue for long poor germination of seeds produced by the grass even in the small portion of Amazonia that is pres­ in the field. By the 1980s, cigarrinha had adapted ently devoted to this land use, let alone in other ar­ to Panicum maximum as well, but not yet at the eas that might be zoned for expansion of cacao devastating levels reached in Brachiaria decum­ plantations. In Rondonia, the primary cacao-grow­ bens. In the late 1970s EMBRAPA began recom­ ing area in Brazilian Amazonia, the World Bank's mending creeping signal grass or braquiaria da POLONOROESTE project (a regional develop­ Amazonia (Brachiaria humidicola ). This species ment project that included paving the BR-364 was at first tolerant of cigarrinha attack, but the insects have become increasingly well adapted to Highway), had cacao as the mainstay of its agri­ this species. EMBRAPA now recommends brizan­ cultural program. The Bank's proposal for tiio (Brachiaria brizantha) and andropogon (An­ POLONOROESTE, written before the project was dropogon gayanus) pasture grasses. The continual launched, projected the decline in cacao prices changing of species and fertilizer recommenda­ that, in fact, has since occurred as predicted tions does not change the basic characteristics of (IBRD, 1981). pasture that make its sustainability doubtful. The sustainability of pasture, as well as its Limits of social values social and environmental impacts, are closely tied Henry Walter Bates (1863), the great nine­ to the potential size of these areas. A small area of teenth-century naturalist who mistakenly believed pasture can be maintained on imported nutrient in­ that Amazonian soils were fertile, marveled that in puts while a large one cannot. A small area would Amazonia a [white] farm owner could make an cause climatic impacts that are within the capabili­ acceptable living with less than dozen slaves. So­ ties of natural systems to correct or absorb, cial values have obviously changed since the time whereas a large area would at some point trigger of Bates to exclude production systems based on processes that lead these equilibria to degenerate slavery, or at least the type of slavery that was (Feamside, 1985, Salati & Vose, 1984). The most abolished in Brazil in 1888. The limits of social worrisome characteristic of pasture is that there is acceptability are not static. The "slavery" of char­ no immediate limit to thwart its continued expan­ coal workers in the Grande Carajas area, for ex­ sion. Unlike annual and especially perennial ample, represents a system that sooner or later crops, market limits for the system's products are must come to an end on the grounds of social jus­ unlikely to halt its expansion: the demand for beef tice. Denunciations of slavery in Brazil before the is tremendous and would be even greater if more International Labor Court in Geneva in 1994 pro­ meat were to become available. The availability of voked a continuing scandal involving Brazil's labor also does not restrain pasture as it does other charcoal industry (Sutton, 1994; Pachauski, 1994; crops because of low labor demands of extensive systems used in Amazonia (Feamside, 1980c, Pamplona & Rodrigues, 1995). The question of 1986a). Pasture's dominance among land-use whether other socially questionable practices choices allows a small human population to have should be accepted as givens has profound impli­ maximum impact on deforestation (Davidson, cations for what limits production. Changes in 1987: 8; Feamside, 1983a). Amazonia's land tenure situation and practice of Input limitations set strict bounds on the ex­ land speculation would greatly affect the direction pansion of all fertilizer-demanding agricultural of development. The amount of land "available" systems, including agroforestry systems (Feam­ for intensive development also depends on side, 1995a). Markets for the products would re­ whether it is regarded as acceptable to obliterate strict the expansion of many land uses (especially indigenous cultures.

Rev. Brasil. Biol., 57 (4): 531-549 540 PHILIP M. FEARNSIDE

Institutional limits wise development choices are made. Components Institutional limits to development take of the system include the requirement since 1986 many different forms. One example is the institu­ that proposed major developments be examined tional limitation imposed by extension and other through an Environmental Impact Report (RIMA) support services for agriculture. These include in­ and a Study of Environmental Impacts (EIA). Un­ frastructural limits, such as the impracticality of fortunately, many of the mechanisms that make up soil laboratories being able to attend the demand this system have failed repeatedly in specific in­ that would be needed to extend the Yurimaguas stances to fulfill their role in guaranteeing envi­ technology to wide areas. They also include cul­ ronmental protection. Strengthening the regulatory tural limits, such as the refusal of extension agents system would be necessary to relax the limiting to visit farms beyond reach of their vehicles. The effect that lack of credibility places on multilateral cultural barriers to communication between exten­ bank financing. sion agents and small farmers in Amazonia are An example is provided by the improvement sufficient to negate almost all effect these agents of the BR-429 highway in Rondonia. This road, might have (see Moran, 1981; Fearnside, 1986a). connecting Presidente Medici with Costa Financing, ranging from bank loans for Marques, opens the Guapore River Valley to set­ small farmers up to major international loans for tlement. Virtually none of the land opened by the development projects, represents a limiting factor road is suitable for agriculture with low or me­ for many kinds of land use. The limited nature of dium inputs, and settlement could only be ex­ funds available for allocation through financial in­ pected to bring severe agricultural, social and stitutions obviously poses a limit to the level of fi­ environmental problems (Fearnside, 1986b). Be­ nancing. This is not the only kind of limitation, cause all Rond6nia, including the Guapore Valley, however. is part of the area of influence of the World Bank­ In the case of financing for agriculture, bu­ financed POLONOROESTE project, the Brazilian reaucratic restrictions on loans commonly rule out government had committed itself to protecting the financing for the poorest layers of society. Land ti­ environment in all of this area when it accepted fi­ tles are normally required by banks financing agri­ nancing for POLONOROESTE (which rebuilt the culture, and the poorer farmers are least likely to BR-364 highway linking Cuiaba with Porto Velho have these. Various kinds of guarantees demanded and provided associated development funding). for "high risk" initiatives can also stand in the When the plan to improve the BR-429 road for way. An example is the demand (relaxed in 1995) all-weather traffic became a public controversy, by the German government for guarantees in case the Governor of Rondonia made a solemn public of misuse of funds by the various cooperatives statement before World Bank representatives that and other nongovernmental organizations to be the BR-429 would not be improved (witnessed by funded through the demonstration projects (PD/A) this author, 16 Sept. 1987). A few months later the sector of the Pilot Program to Conserve the Bra­ improvement had taken place nevertheless. zilian Rain Forest. Aside from discrepancies between discourse One institutional limit that affects major in­ and action, environmental protection may be an ternational development loans is that of institu­ impossibility even when action is taken. No mat­ tional credibility. This acts as a limiting factor ter how sincerely government authorities may when loans are not approved (or not contem­ swear that deforestation will be avoided in any plated) due to lack of trust in the commitment of given development project, much of the clearing implementing institutions to respect environ­ process remains outside government control. For mental limits. Such lack of trust can have a firm example, a road through a reserve to open up an basis in fact, and bears some examination. area on the other side may appear as a good idea Brazil, like many other countries, has a regu­ when listening to the solemn guarantees of gov­ latory system governing evaluation and approval ernment agencies that no settlement will be al­ of proposed development projects. This system is lowed along the road, but developments in composed of a series of decrees, laws and consti­ practice are apt to be different once the road is tutional provisions. and is designed to assure that built. The best example is the road through the

Rev. Brasil. Biol., 57 (4): 531-549 AGRICULTURE AND RANCHING IN BRAZILIAN AMAZONIA 541

Cuyabena Reserve in Ecuador for opening oil sion, for example, is a straightforward constraint. fields initially developed by Texaco. Allocations may affect nearby areas as well, as How to control deforestation is one of the when sm,all farmers are placed in close proximity most important and most difficult problems for to natural forest reserves. Invasion of reserve areas any development policy. In the Brazilian Amazon often follows, making use of the road infrastruc­ a tradition prevails that is over 400 years old of ture implanted for the small farmers. Examples in­ complete disregard for any law designed to pre­ clude the Sete de Setembro Indigenous Area near serve the flora and fauna of the area (Sternberg, Cacoal and the Uru-Eu-Wau-Wau Indigenous Area 1973). In all of Brazil the law is perceived as near Ouro Preto do Oeste (both in Rondonia). In­ something which is only to be applied to "ene­ terposition of silviculture has been suggested as an mies," and which can always be avoided by means effective barrier where this land use is viable of the omnipresent jeito (Rosenn, 1971). A tradi­ (Fearnside, 1983a; Davidson, 1987: 8). Small tion exists dating from colonial times of maintain­ farmers are not alone in their impact on nearby ing thousands of laws technically in force, and forest areas: burning of cattle pastures on large only applying a few of them. These problems ranches can lead to fires entering adjacent forest, make it difficult to formulate effective laws to especially if it has been selectively logged (Uhl & control deforestation, and it is not reasonable to Buschbacher, 1985). expect that this context of the problem will change Limiting factors imposed by the need to in the near future. meet environmental quality standards should be interpreted in terms of conceptualization of the Limits on human habitation landscape as a mosaic of different areas where dif­ Human health problems, particularly ma­ ferent criteria apply, including requirements for laria, have kept wide areas of Amazonia from be­ environmental quality (Odum, 1969; Eden, 1978; ing densely settled over the past five centuries. In Fearnside, 1979b). This vision fits well with the small farmer settlement areas today, malaria has a current paradigm of buffer zones around conserva­ significant impact on agricultural production (and tion units, with different levels of biodiversity and on the dependability of that production) by inca­ other aspects of the original ecosystem maintained pacitating key family members at the time of year in concentric circles around the core areas (Sayer, when their labor is needed for specific agricultural 1991). tasks. The effect of human health on labor supply The most obvious limiting factor for expan­ and ability to perform agricultural tasks has been quantified for Transamazon Highway colonists sion of agriculture and ranching in Amazonia is (Fearnside, 1978, 1986a) using data on the season­ the area of forest that must be maintained intact. ality of hospital admissions in Altamira (Smith, The different forms of land use imply environ­ 1976). mental impacts (with distinct levels of impact de­ pending on whether the land use proves to be Land conflicts represent a limit on agricul­ sustainable). The impact of converting forest to tural activities in many parts of the region. In ar­ another land use depends not only on the patch of eas such as the Bica do Papagaio region in land for which conversion is being considered, but northern Tocantins, which is notorious for blood­ also on what has been done with the remainder of shed over conflicting land claims, some types of development would be difficult. Even in the ab­ the region. As the cumulative area cleared in­ sence of land conflicts, rapid turnover in settler creases, the danger increases that each additional populations adds an aspect of instability to land hectare of clearing will lead to unacceptable im­ uses that require maintaining a consistent manage­ pacts. For example, the risk of species extinctions ment system over a long period. increases greatly as the remaining areas of natural forest dwindle. The role of Amazonian forest in Limits to risk the region's water cycle also implies increasing risk with the scale of deforestation: when rainfall Environmental Risks reductions caused by losses of forest evapotran­ Environmental impacts limit the allocation spiration are added to the natural variability that of land to many uses. Susceptibility to soil ero- characterizes rainfall in the region, the resulting

Rev. Brasil. Biol., 57 (4): 531-549 542 PHILIP M. FEARNSIDE droughts would cross biological thresholds lead­ is given in Figure 3B. Unlike the relation of ac­ ing to major impacts (Feamside, 1995b). These ceptable risk to magnitude of impact (Fig. 3A), thresholds include the drought tolerance of indi­ the relation of risk to deforestation (Fig. 3B) is a vidual tree species and the increased probability of scientific question; construction of a curve of this fire being able to propagate itself in standing for­ type based on field studies should be a high prior­ est. Fire entry into standing forest in Brazilian ity. As deforested area increases, the probability Amazonia already occurs in areas disturbed by increases of a major perturbation such as a severe logging (Uhl & Buschbacher, 1985; Uhl & Kauff­ drought that exceeds the tolerance of many tree man, 1990). During the El Nifio drought of species adapted to a relatively stable climate. 1982/1983, approximately 45,000 km2 of tropical The maximum permissible amount of defor­ forest on the island of Borneo burned when fires escaped from shifting cultivators' fields. Of the estation can be calculated from graphs A and B of 35,000 km2 of this area in the Indonesian province Fig. 3. Starting with the size of the impact that of East Kalimantan, at least 8,000 km2 was pri­ would be provoked by perturbation from defores­ mary forest, while 12,000 km2 was selectively tation, one can determine the corresponding maxi­ logged forest (Malingreau et al., 1985). In Ama­ mum acceptable level of risk to society from Fig. zonia, "mega-El Nifio" events have caused wide­ 3A. One can then determine from Fig. 3B the per­ spread conflagr.ations in the forest four times over centage of forest that could be cut and still stay the past 2000 years (Meggers, 1994). The effect of within bounds of this acceptable level of risk. large-scale deforestation is to tum relatively rare events like these into something that could recur at much more frequent intervals. How these dan­ gers are incorporated into land-use decisions greatly influences the carrying capacity of the re­ gion for humans. If one assumes that the entire re­ gion could be converted to agricultural use without unacceptable consequences, then the car­ rying capacity one would calculate would be much higher than if one assumes that enough for­ est must remain intact to keep the risk of environ­ mental catastrophes within defined limits. MAGNITOOE OF IMR\CT Figure 3A shows the relationship between the magnitude of an impact and the maximum probability of the impact occurring that society is willing to accept. Small impacts, such as failure of 8) a given crop, may be acceptable even if they occur every year (probability of 1.0), but society should I insist on there being only a very tiny probability 15 of a major catastrophe, such as a year dry enough I I I to allow fire to destroy large areas of standing I I I tropical forest. This is analogous to precautions I MAXMUM against explosions or other major accidents in­ io!-----lr_PE_RMl_SSIBL __ E --~100(%) volving nuclear power plants: only infinitesimally small risks are acceptable to society. The accept­ DEFORESTATION (%) ability of risk to society (Fig. 3A) is not a scien­ Fig. 3 - A) Relationship of the level of acceptable risk to the tific question, but rather a moral and political one magnitude of the impact. Given that some of the consequences of deforestation are grave, the maximum acceptable probability that needs to be debated and decided in a demo­ of these impacts occurring should be low (Source: Fearnside, cratic manner. 1993b). B) The maximum percentage of deforestation permissible as determined from the maximum acceptable risk A rough sketch of the possible relationship (this probability is determined from part A) (Source: Feamside, of environmental risk to expansion of cleared area 1993b).

Rev. Brasil. Biol., 57 (4): 531-549 AGRICULTURE AND RANCHING IN BRAZILIAN AMAZONIA 543

Agricultural Risks ing system may interfere, for example, for military Development choices create risks not only or geopolitical reasons, to override this kind of ra­ for those participating in development activities tionality. The limits to development in portions of and those living in the immediate vicinity of the the region that are subject to this kind of alterna­ projects, but also for third parties who may be tive rationality are different from those that might thousands of kilometers away. For example, when be deduced in other areas. Government fiats can­ cacao was promoted in Rondonia as the agricul­ not, however, override natural laws, such as the tural mainstay of the World Bank-financed relationships between soil fertility and agricultural POLONOROESTE project, the risk was substan­ yield. They can alter the outcome of events, how­ tially increased of witches' broom disease (caused ever, by supplying fertilizers and other inputs as by the fungus Crinipellis perniciosa) spreading to subsidies and by inducing farmers to commit Brazil's principal cacao-growing region in former themselves to farming certain areas through prom­ Atlantic forest areas of the state of Bahia. With di­ ises of land titles and various forms of assistance. rect bus service connecting Rond~:mia with Bahia, Military development has a potentially im­ it was only a matter of time before the fungus portant role in promoting projects that will lead to jumped this geographical gap. The disease, in fact, deforestation in frontier areas. It has also ob­ entered Bahia in 1988, only seven years after the structed (but not completely impeded) demarca­ POLONOROESTE project began, and devastated tion of indigenous areas in a 150 km-wide strip cacao there over the course of the succeeding along the borders, thereby facilitating the entry of seven years. Undoubtedly, the loss to Brazil's gold miners (garimpeiros), loggers, squatters, and economy represented by destruction of cacao in others in these areas (de Oliveira Filho, 1990). Bahia surpassed all economic gains from agricul­ Military plans for development are made in­ ture in the POLONOROESTE area. dependent of the zoning exercise, and also of the Health Risks normal procedures for obtaining environmental clearance. This independence is not granted in the Another example is the risk posed by oncho­ decrees that mandate zoning and environmental cerciasis, or African river blindness, a disease that impact reports (RIMAs ), but military develop­ in South America is limited to the frontier be­ ments are, in fact, executed with complete free­ tween Roraima and Venezuela. The disease, dom. According to IBAMA staff, for example, in thought to have been introduced to the area by 1990 the army built a road cutting through the missionaries who had worked in Africa, is spread Pico da Neblina National Park without even con­ by black flies (Simulium spp.) that occur over sulting IBAMA (the agency responsible for na­ wide areas of Brazil. Isolated cases have already tional parks), let alone passing through the appeared in other parts of the country, such as environmental review procedures required for Goias, in gold miners returning from Roraima highway construction. (Gerais & Ribeiro, 1986). If military plans go for­ ward to maintain garrisons in the area and to open Settlement of the area along Brazil's interna­ these frontier areas to colonization for geopolitical tional borders has been an objective of the coun­ reasons, the risk would be substantially increased try's military for many decades. The best example of major outbreaks of river blindness throughout of the danger of allowing military considerations Brazil. to determine the location of settlement projects is the Sidney Girao Colonization Project, which was LIMITS OF POLITICAL placed on Rondonia's border with Bolivia for stra­ AND MILITARY INTERFERENCE tegic reasons in the early 1970s (Mueller, 1980). The poor soil in the area resulted in the lots being Rational planning may be based on percep­ abandoned so rapidly that the government was un­ tion of limits, leading to decisions, for example, able to fill the project until all of the other settle­ not to encourage agricultural settlement in areas ment areas in Rondonia were overflowing with where low soil fertility and other aspects of agri­ land-seeking migrants. The failure of the project cultural site quality make predicted yields low. has been officially recognized as being due to the However, other sectors of Brazil's decision-mak- poor soil (Valverde et al., 1979).

Rev. Brasil. Biol., 57 (4): 531-549 544 PHILIP M. FEARNSIDE

Construction of the Northern Perimeter and settlements do not appear in the plans that Road (BR-210), to open access to lands near Bra­ have been made public to date. Although individ­ zil's borders with Colombia, Venezuela and the ual military proposals rise and fade with some Guianas, began 1973, but was halted in 1974. Dis­ regularity, pressure from military leaders for set­ illusionment with the poor soils revealed by publi­ tlement along Brazil's Amazonian frontiers re­ cation of RADAMBRASIL maps for the area mains an unchanged feature of the political contributed to this decision (Foresta, 1991: 28). landscape in which development planning takes The Calha Norte Project to construct mili­ place. tary bases and/or airstrips at 16 locations along Brazil's northern borders, was announced in 1986, GUIDELINES FOR RESEARCH and implementation proceeded without any envi- AND ACTION . ronmental impact study - even though the entire project came into being after RIMAs had become When confronted with the existence of a mandatory. The area affected by bases such as limiting factor, the normal reaction on the part of these is potentially much larger than the surround­ both decision-makers and researchers is to con­ ·ings of the military installations. Although not in­ centrate effort on finding a way to overcome the dicated in the project bqdget, the plan called for limitation. The question of whether or not the construction of highways and the promotion of limit should be pushed back in the first place is settlements. The statement of motives (exposi~iio normally not even considered. However, this basic de motivos) proposing the project to then-Presi­ question must be answered before any effort to dent Jose Sarney stated clearly that "it is funda­ overcome a limitation can make sense. mental that government action also contemplate Once a decision has been reached that limits increasing the road infrastructure ... and the in­ must be pushed back to some extent, then it is crease of colonization in that border region" necessary to obtain information on the full range (Setubal et al., 1986: 3). Once the roads are built, of factors that limit attainment of the defined de­ squatters and speculators can be expected to enter velopment objectives. Information is needed on to cut the forest independent of any government the unit costs and scaling effects of confronting policies, as has already happened repeatedly in each limitation. The simple reflex that all limiting other parts of the region (Fearnside & Ferreira, factors must be beaten back is both wasteful and 1984). No part of the Calha Norte Project area is unwise. Once relevant information is organized shown on RADAMBRASIL maps as being suit­ and interpreted, the effectiveness, cost, and the so­ able for agriculture (Brazil, Projeto RADAM­ cial and environmental side effects of attacking BRASIL, 1974-1977, Vols. 6, 8, 9, 11, 14). the different limits can be compared. In 1991, plans were announced for the Which limits should be respected and which Trans-frontier Highway, an 8000-km road which, should be fought against depends on what the lim­ for strategic reasons, would parallel all of Brazil's its are. For example, most would agree that limits international borders in Amazonia (Joma/ do such as obeying Brazil's labor laws must be re­ Brasil, 12 August 1991). The project withered spected, while lack of education and problems of when funds were not approved by the National inefficiency and corruption in institutions must be Congress. fought against. Agreement is less easy on other In 1993, Brazil's top military leaders drafted kinds of limits, such as those that involve alloca­ a list of nine demands, with which then-President tion of public resources or the redistribution of ltamar Franco concurred (Folha de Sao Paulo, 11 land holdings. A common reaction is to view as August 1993). The demands included the resump­ givens the current highly unequal distribution of tion of colonization in Amazonia and the "review" land tenure and any decisions based on military (i.e., reduction) of indigenous areas in the region. and geopolitical arguments, and to concentrate at­ In 1995, most of the functions of the Calha tention only on technical advances against soil re­ Norte Project were incorporated into plans for SI­ straints. I would suggest that soil and related VAM - a major military project to maintain radar physical conditions are much more "given" than surveillance of Amazonian air space. Highways are restraints that result from the country's social

Rev. Brasil. Biol., 57 (4): 531-549 AGRICULTURE AND RANCHING IN BRAZILIAN AMAZONIA 545

and decision-making hierarchy, and that the latter current population of rural Amazonia and their de­ category is where attention should be focused. scendants is to tap the potential monetary value of What are the ingredients of a rational deci­ the environmental services provided to the rest of sion on the question of attempting or not attempt­ the world by the natural forests in Amazonia ing to overcome a limitation on development? The (Feamside, 1997). starting point must be a clear definition of the ob­ jectives of development. For example, if the ob­ CONCLUSIONS jective of development is to provide a sustainable 1. Limiting factors restrict both the intensifi­ livelihood for the populations of the region, then cation of agriculture and ranching uses and the little benefit will be achieved by augmenting the scale to which these land uses can be expanded. productivity or the life expectancy of cattle pas­ 2. The expression of limiting factors on de­ tures on large ranches by supplying fertilizers and velopment is mediated through human planning. improving management. Many efforts to push Perception of limits to agricultural yields and of back limits to agricultural crop propuction have as the severity and probability of environmental im­ their rationale supporting an ever-larger popula­ pacts can lead to decisions to limit agricultural ex­ tion of farmers, for example, of immigrants who pansion. come to Amazonia from other parts of Brazil. This 3. Limits to intensification of agriculture in­ is not necessarily in the best interests of Ama­ clude agronomic limits on per-hectare yields, zonia 's current population and their descendants. technological limits and research, and cultural It would be better to recognize that the ability of limits. Amazonia to support population is limited and to 4. Limits to expansion of agricultural areas guide development in such ways that the popula­ include physical resource limits such as phosphate tion size and environmental impacts are kept deposits, limits of social values, institutional lim­ within those limits (Feamside, 1996b). its (including the credibility of institutions), limits There is no such thing as sustainable devel­ on human habitation (such as health), and limits to opment for an infinite number of people, nor for a environmental risks. fixed population that is infinitely rapacious. There 5. Limits from considerations in political is also no way that development aimed at increas­ and military spheres often override "rational" de­ ing the size of the pie can address problems that cisions based on land capability and environ­ are rooted in highly unequal distribution of the mental consequences. This kind of "interference," pie. Many physical limits represent restrictions however, can cause a variety of impacts that, if that need to be respected and lived with rather properly evaluated, would likely make the net re­ than as an agenda of items to be attacked. Recog­ sult of such development projects a negative one nition of this fact forces one to face fundamental for Brazil's national interests. problems of development that many people would 6. While some limiting factors can be prefer not to think about - resulting in a tendency pushed back through technological advances and to deny the existence of limits. Admitting to the fi­ institutional changes, many of the restraints on ag­ nite potential for growth of the pie does not con­ riculture and ranching development must simply demn the poor to poverty, but rather condemns the be accepted and lived within. Rather than expan­ rich to dividing the pie (Feamside, 1993a). sion of agriculture and ranching, other strategies A formidable array of limiting factors stands should be pursued to support Amazonia's popula­ in the way of sustaining production in large areas tion, especially by tapping the value of the envi­ of Amazonia if forests are converted to agriculture ronmental services of intact forest. and ranching. This does not mean that the outlook need be gloomy for sustaining the region's current Acknowledgments - This paper was presented at the "Work­ population, provided the means of support is de­ shop: Sustainable Biosphere Project: Amazon Basin Case rived from the forest itself rather than through re­ Study," 18-20 June 1996, Manaus, Amazonas, Brazil. I thank the conference organizers for permission to publish this ver­ placing it with nonforest land uses. This author sion. Portions of the text have been adapted from Fearnside, believes that the best long-term strategy for pro­ 1987, 1990a and 1990b. E. Salati and S. V. Wilson made valu­ viding a sustainable basis of development for the able comments on the manuscript.

Rev. Brasil. Biol., 57 (4): 531-549 546 PHILIP M. FEARNSIDE

REFERENCES EDEN, M. 1., 1978, Ecology and land development: The case of Amazonian rainforest. Transactions of the Institute ALLAN, W., 1965, The African Husbandman. Barnes and ofBritish Geographers, New Series, 3(4): 444-463. Noble, New York, U.S.A. 265 pp. FALES!, I. C., 1974, 0 solo na Amazonia e sua rela<;ii.o corn a ALLEGRETTI, M. H., 1990, Extractive reserves: An alterna­ defini<;lio de sisternas de produ<;lio agricola. In: Em­ tive for reconciling development and environmental presa Brasileira de Pesquisa Agropecuaria (EMBRAPA) conservation in Amazonia. In: A. B. Anderson (ed.) Al­ Reuniao do Grupo Interdisciplinar de Trabalho so­ ternatives to Deforestation: Steps toward Sustain­ bre Diretrizes de Pesquisa Agricola para a able Use of Amazonian Rain Forest. Columbia University Press, New York, U.S.A. pp. 252-264. Amazonia (Tropico Umido), Brasilia, May 6-10, 1974. EMBRAPA, Brasilia, Brazil. Vol. l, pp. 2.1-2.11. ALVIM, P. DE T., 1981, A perspective appraisal of perennial crops in the Amazon Basin. Interciencia, 6(3): 139- FALESI, I. C., 1976, Ecossistema de Pastagem Cultivada na 145. Amazonia Brasileira. Boletirn Tecnico No. 1. Centro BATES, H. W., 1863 [1989], The Naturalist on the River de Pesquisa Agropecuaria do Tr6pico Urnido (CPATU), Amazons. Penguin Books, New York, U.S.A. 383 pp. Belem, Para, Brazil. BEISIEGEL, W. DE R. & DE SOUZA, W. 0., 1986, Reservas FAO, 1980, Report of the Second FAO/UNFPA Expert Con­ de fosfatos - Panorama nacional e rnundial. In: Instituto sultation on Land Resources for Populations of the Brasileiro de Fosfato (IBRAFOS) III Encontro Na­ Future. Food and Agriculture Organization of the cional de Rocha Fosfdtica, Brasilia, 16-18/06/86. United Nations (FAO), Rome, Italy. 369 pp. IBRAFOS, Brasilia, Brazil. pp. 55-67. FAO, 1981, Report on the Agro-Ecological Zones Project, BLACKMAN, F. F., 1905, Optima and limiting factors. An­ Vol. 3. World Soils Resources Report 48/3. Food and nals of Botany, 14(74): 281-295. Reprinted, in: E. J. Agricultural Organization of the United Nations (FAO), Korrnondy (ed.) Readings in Ecology. Prentice-Hall, Englewood Cliffs, New Jersey, U.S.A. (1965). pp. 14- Rome, Italy. 251 pp. 17. FAO, 1984, Land, Food and People. FAO Economic and So­ BOSERUP, E., 1965, The Conditions of Agricultural cial Development Series No. 30. Food and Agriculture Growth. Airline, Chicago, Illinois, U.S.A. 124 pp. Organization of the United Nations (FAO), Rome, Italy. 96 pp. + 4 maps. BRAZIL, EMBRAPA, 1988, Delineamento Macro-Agroe­ cologico do Brasil. Ernpresa Brasileira de Pesquisa FEARNSIDE, P. M., 1978, Estimation of Carrying Capacity Agropecuaria (EMBRAPA), Servi<;o Nacional de Le­ for Human Populations in a part of the Transama­ vantarnento e Conserva<;lio de Solos (SNLCS), Rio de zon Highway Colonization Area of Brazil. Ph.D. dis­ Janeiro, Brazil. 114 pp. sertation in biological sciences, University of Michigan, BRAZIL, RADAMBRASIL, Ministerio de Minas e Energia, Ann Arbor. University Microfilms International, Ann Projeto RADAMBRASIL, 1973-1982, Levantamento Arbor, Michigan, U.S.A. 624 pp. de Recursos Naturais. Departarnento de Produ<;lio FEARNSIDE, P. M., 1979a, Cattle yield prediction for the Mineral (DNPM), Rio de Janeiro, Brazil. 27 vols. Transarnazon Highway of Brazil. lnterciencia, 4(4): BROOKFIELD, H. C., 1972, Intensification and disintensifiea­ 220-225. tion in Pacific agriculture: A theoretical approach. Pa­ cific Viewpoint, 13: 30-48. FEARNSIDE, P. M., 1979b, The development of the Amazon rain forest: Priority problems for the formulation of DAVIDSON, J., 1987, Bioenergy tree plantations in the trop­ ics: Ecological implications and impacts. Commission guidelines. Interciencia, 4(6): 338-343. on Ecology Papers No. 12. lnternational Union for Con­ FEARNSIDE, P. M., 1980a, Black pepper yield prediction for servation of Nature and Natural Resources (IUCN), the Transarnazon Highway of Brazil. Turrialba, 30(1): Gland, Switzerland. 47 pp. 35-42. DE LIMA, J.M. G., 1976, Perfil Analitico dos Fertilizantes FEARNSIDE, P. M., 1980b, The effects of cattle pastures on Fosfatados. Ministerio das Minas e Energia, Depar­ soil fertility in the Brazilian Amazon: Consequences for tarnento Nacional de Produ<;ao Mineral (DNPM) beef production sustainability. Tropical Ecology, Boletirn No. 39. DNPM, Brasilia, Brazil. 21(1): 125-137. DE OLIVEIRA FILHO, J. P., 1990, Frontier security and the new indigenisrn: Nature and origins of the Calha Norte FEARNSIDE, P. M., 1980c, Land use allocation of the Tran­ Project. In: D. Goodman & A. Hall (eds.) The Future sarnazon Highway colonists of Brazil and its relation to of Amazonia: Destruction or Sustainable Develop­ human carrying capacity. In: F. Barbira-Scazzocchio ment? Macmillan, London, U.K. pp. 155-176. (ed.) Land, People and Planning in Contemporary DOS SANTOS, B. A., 1981, Amazonia: Potencial Mineral e Amazonia. University of Cambridge Centre of Latin Perspectivas de Desenvolvimento. Editora da Univer­ American Studies Occasional Paper No. 3, Cambridge, sidade de Sao Paulo (EDUSP), Sao Paulo, Brazil. 256 pp. U.K. pp. 114-138.

Rev. Brasil. Biol., 57 (4): 531-549 AGRICULTURE AND RANCHING JN BRAZILIAN AMAZONIA 547

FEARNSJDE, P. M., 1983a, Development alternatives in the FEARNSIDE, P. M., 1996b, Human carrying capacity estima­ Brazilian Amazon: An ecological evaluation. Intercien­ tion in Brazilian Amazonia: Research requirements to cia, 8(2): 65-78. provide a basis for sustainable development. In: R. Lie­ berei, C. Reisdorff & A. D. Machado (eds.) Interdisci­ FEARNSJDE, P. M., 1985, Environmental change and defores­ plinary Research on the Conservation and tation in the Brazilian Amazon. In: J. Hemming (ed.) Sustainable Use of the Amazonian Rain Forest and Change in the Amazon Basin: Mans Impact on For­ its Information Requirements. Forschungszentrum ests and Rivers. Manchester University Press, Man­ Geesthacht GmbH (GKSS), Bremen, Germany. pp. chester, U.K. pp. 70-89. 274-291. FEARNSJDE, P. M., 1986a, Human Carrying Capacity of FEARNSIDE, P. M., 1997, Environmental services as a strat­ the Brazilian Rainforest. Columbia University Press, egy for sustainable development in rural Amazonia. New York, U.S.A. 293 pp. Ecological Economics, 20(1): 53-70. FEARNSIDE, P. M., 1986b, Settlement in Rondonia and the FEARNSIDE, P. M. & FERREIRA, G. DEL., 1984, Roads in token role of science and technology in Brazil's Ama­ Rondonia: Highway construction plans and the farce of zonian development planning. Interciencia, 11(5): 229- unprotected forest reserves in Brazil's Amazonian for­ 236. est. Environmental Conservation, 11(4): 358-360. FOLHADE SAO PAULO [Sao Paulo] 11August1993, "Con­ FEARNSJDE, P. M., 1987, Rethinking continuous cultivation selho aprova piano para a Amazonia." Section 1, p. 8. in Amazonia. BioScience, 37(3): 209-214. FORESTA, R. A., 1991, Amazon Conservation in the Age of FEARNSJDE, P. M., 1988, Yurimaguas reply. BioScience, Development: The Limits of Providence. University 38(8): 525-527. of Florida Press, Gainesville, Florida, U.S.A. 366 pp. FEARNSIDE, P. M., 1989, Extractive reserves in Brazilian GALLOPJN, G. C. & WINOGRAD, M., 1995, Ecological pro­ Amazonia: Opportunity to maintain tropical rain forest spective for tropical Latin America. In: T. Nishizawa & under sustainable use. BioScience, 39(6): 387-393. J.I. Uitto (eds.) The Fragile Tropics ofLatin America: FEARNSIDE. P M., 1990a, Human carrying capacity in rain­ Sustainable Management of Changing Environ­ forest areas. Trends in Ecology and Evolution, 5(6): ments. United Nations University Press, Tokyo, Japan. 192-196. pp. 13-44. GERAIS, B. B. & RIBEIRO, T. C., 1986, Relato de casos. On­ FEARNSIDE, P. M., 1990b, Predominant land uses in the Bra­ cocercose: primeiro caso aut6ctone da regiiio Centro­ zilian Amazon. In: A. B. Anderson (ed.)Alternatives to oeste do Brasil. Revista da Sociedade Brasileira de Deforestation: Towards Sustainable Use of the Ama­ Medicina Tropical, 19(2): 105-107. zon Rain Forest. Columbia University Press, New HALL, A. E., 1990, Physiological ecology of crops in relation York, U.S.A. pp. 235-251. to light, water and temperature. In: C. R. Carroll, J. H. FEARNSIDE, P. M., 1993a, Forests or fields: A response to the Vandermeer & P. M. Rossel (eds.) Agroecology. theory that tropical forest conservation poses a threat to McGraw-Hill, New York, U.S.A. pp. 191-234. the poor. Land Use Policy, 10(2): 108-121. HECHT, S. B., 1981, Deforestation in the Amazon basin: prac­ FEARNSIDE, P. M., 1993b, Migra«iio, coloniza«iio e meio­ tice, theory and soil resource effects. Studies in Third ambiente: 0 potencial dos ecossistemas amazonicos. World Societies, 13: 61-108. Cadernos da Saude, 9(4): 448-457. HECHT, S. B., 1983, Cattle ranching in the eastern Amazon: FEARNSIDE, P. M., 1995a, Agroforestry in Brazil's Ama­ environmental and social implications. In: E. F. Moran zonian development policy: The role and limits of a po­ (ed.) The Dilemma of Amazonian Development. tential use for degraded lands. In: M. Cliisener-Godt & Westview Press, Boulder, Colorado, U.S.A. pp. 155- 188. I. Sachs (eds.) Brazilian Perspectives on Sustainable Development of the Amazon Region. UNESCO, Paris, HIGGINS, G. M.; KASSAM, A. H.; NAIKEN, L.; FISCHER, France & Parthenon Publishing Group, Carnforth, U.K. G. & SHAH, M. M., 1982, Potential Population Sup­ pp. 125-148. porting Capacities of Lands in the Developing World. Technical Report of Project INT/75/P13 "Land FEARNSIDE, P. M., 1995b, Potential impacts of climatic Resources for Populations of the Future," Food and Ag­ change on natural forests and forestry in Brazilian Ama­ riculture Organization of the United Nations (PAO), zonia. Forest Ecology and Management, 78: 51-70. Rome, Italy. 139 pp. FEARNSJDE, P. M., 1996a, Review of: The Fragile Tropics HOLDEN, C., 1980, The Reagan years: Environmentalists of Latin America: Sustainable Management of tremble. Science, 210: 988-991. Changing Environments. Edited by Toshie Nishizawa HUBBELL, S. P., 1973, Populations and simple food webs as & Juba I. Uitto (United Nations University Press, To­ energy filters. I. Onespecies systems. American N atu­ kyo, 1995), Environmental Conservation, 23(4): 382. ralist, 107: 94-121.

Rev. Brasil. Biol., 57 (4): 531-549 548 PHILIP M. FEARNSIDE

INTERNATIONAL BANK FOR RECONSTRUCTION AND REGIS, M., 1989, "IBGE e Embrapa divergem sobre melhor DEVELOPMENT (IBRD), 1981, Brazil: Integrated ocupac;cao para Amaz6nia." Jornal do Brasil [Rio de Development of the Northwest Frontier. The World Janeiro] 9 July 1989, Section 1, p. 17. Bank (IBRD), Washington, DC, U.S.A. 101 pp. ROSENN, K. S., 1971, The jeito: Brazil's institutional bypass JORNAL DO BRASIL [Rio de Janeiro] 12 August 1991, "Uma of the formal legal system and its development implica­ rodovia muito po!emica." Cademo Ecologia, p. 3. tions. American Journal of Comparative Law, 19: KOSTER, H. W.; KHAN, E. J. A & BOSSHART, R. P., 1977, 514-549. Programa e Resultados Preliminares dos Estudos de SALATI, E.; DOS SANTOS, A; LOVEJOY, T. E. & Pastagens na Regiiio de Paragominas, Para, e nor­ KLABIN, I., nd. Porque Sa/var a Floresta deste de Mato Grosso, junho 1975-dezembro 1976. Amazonica. Instituto Nacional de Pesquisas da Superintendencia do Desenvoivimento da Amaz6nia Amaz6nia (INPA), Manaus, Amazonas, Brazil. (in (SUDAM), Convenio SUDAM/Instituto de Pesquisas press). IRI, Belem, Para, Brazil. SALATI, E. & VOSE, P. B., 1984, Amazon Basin: A system in LIEBIG, J., 1840, Organic chemistry in its application to vege­ equilibrium. Science, 225: 129-138. table physiology and agriculture. Jn: Professor Liebig's SANCHEZ, P. A; BANDY, D. E.; VILLACHICA, J. H. & Complete Works on Chemistry. T. B. Peterson, Phila­ NICHOLAIDES, III, J. J., 1982, Amazon basin soils: delphia, PA (1841). Reprinted, in: E. J. Kormondy (ed.) Management for continuous crop production. Science, Readings in Ecology, Prentice-Hall, Englewood Cliffs, 216: 821-827. New Jersey, U.S.A. (1965) pp. 1-14. SAYER, J., 1991, Rainforest Buffer Zones: Guidelines for MALINGREAU, J. P.; STEPHENS, G. & FELLOWS, L., Protected Area Managers. The World Conservation 1985, Remote sensing of forest fires: Kalirnantan and Union (IUCN), Gland, Switzerland. 94 pp. North Borneo in 1982-83.Ambio, 14(6): 314-321. SERRAO, E. A S. & FALESI, 1. C., 1977, Pastagens do MEGGERS, B. J., 1994, Archeological evidence for the impact Tropico Umido Brasileiro. Empresa Brasileira de Pes­ of mega-Nifto events on Amazonia during the past two quisa Agropecuaria-Centro de Pesquisa Agropecuana millenia. Climatic Change, 28(1-2): 321-338. do Tr6pico Umido (EMBRAPA--CPATU), Belem, Para, Brazil. MORAN, E. F., 1981, Developing the Amazon. Indiana Uni­ versity Press, Bloomington, Indiana, U.S.A. 292 pp. SERRAO, E. A S.; FALESI, I. C.; da VIEGA, J. B. & TEIXEIRA NETO, J. F., 1979, Productivity of culti­ MUELLER, C., 1980, Frontier based agricultural expansion: vated pastures on low fertility soils in the Amazon of the case of Rond6nia. Jn: F. Barbira-Scazzocchio (ed.) Brazil. Jn: P. A Sanchez & L. E. Tergas (eds.) Pasture Land, People and Planning in Contemporary Ama­ Production in Acid Soils of the Tropics: Proceedings zonia. Cambridge University Centre of Latin American of a Seminar held at CIAT, Cal~ Colombia, April 17- Studies Occasional Paper No. 3, Cambridge University, 21, 1978, Series 03 EG-05. Centro Internacional de Cambridge, U.K. pp. 141-153. Agricultura Tropical (CIAT), Cali, Colombia. pp. 195- NICHOLAIDES, III, J. J.; BANDY, D. E.; SANCHEZ, P. A; 225. BENITES, J. R.; VILLACHICA, J. H.; COUTU, A J. SETUBAL, 0. E.; COUTO, R. C.; SAYAD, J. & DENYS, R. & VALVERDE S., C., 1985, Agricultural alternatives B., 1986, Desenvolvimento e seguranc;ca na regiao norte for the Amazon Basin. BioScience, 35: 279-285. das calhas dos rios Solirn6es e Amazonas, Projeto Calha ODUM, E. P., 1969, The strategy of ecosystem development. Norte, Exposic;cao de Motivos (No. 770) para o Presi­ Science, 164: 262-270. dente Jose Sarney. Brasilia, Brazil. PACHAUSKI, F., 1994, "Trabalha, escravo." Jsto E [Brasilia] SHELDON, R. P., 1982, Phosphate rock. Scientific American, 4 May 1994, pp. 32-35. 246(6): 31-37. PAEZ, G. & DUTRA, S., 1974, Algumas considerac;coes sobre SHELFORD, V. E., 1911, Physiological animal geography. o delineamento de sistemas de produc

Rev. Brasil. Biol., 57 (4): 531-549 AGRICULTURE AND RANCHING IN BRAZILIAN AMAZONIA 549

STERNBERG, H. O'R., 1973, Development and conservation. VALVERDE, 0.; JAPIASSU, A. M. S.; LOPES, A. M. T.; Erdkunde, Archiv fur vissenschaftliche Geographie, NEVES, A. M.; EGLER, E. G.; MESQUITA, H. M.; Band 27, 1 fg. 4 Bonn, Germany. pp. 263-265. (repub­ DA COSTA, I. B.; GARRIDO FILHA, I.; DE BU­ lished as: Reprint No. 447, Center for Latin American LHAES, M. G.; MESQUITA, M. G. G. C. & FER­ Studies, University of California, Berkeley, California, REIRA, N. A., 1979, A Organiza~iio do Espa~o na U.S.A.). Faixa da Transamazonica. Volume 1: Introdm;ao SUBLER, S. & UHL, C., 1990, Japanese agroforestry in Ama­ Sudoeste Arnaz6nico, Rond6nia e Regiiies Vizinhas. zonia: A case study in Tome-A<;u, Brazil. In: A. B. An­ Instituto Brasileiro de Geografia e Estatistica (IBGE), derson (ed.) Alternatives to Deforestation: Towards Rio de Janeiro, Brazil. 258 pp. Sustainable Use of the Amazon Rain Forest. Colum­ bia University Press, New York, U.S.A. pp. 152-166. WALKER, B. H.; LAVELLE, P. & WEISCHET, W., 1987, SUTTON, A., 1994, Slavery in Brazil-A Link in the Chain Yurimaguas technology. BioScience, 37(9): 638-640. of Modernization. Anti-Slavery International, London, WAUGH, D. L.; CATE, JR., R. B.; NELSON, L. A. & MAN­ U.K. ZANO, A., 1975, New concepts in biological and eco­ UHL, C. & BUSCHBACHER, R., 1985, A disturbing syner­ nomical interpretation of fertilizer response. In: E. gism between cattle-ranch burning practices and selec­ Bornemisza & A. Alvarado (eds.) Soil Management in tive tree harvesting in the eastern Amazon. Biotropica, Tropical America: Proceedings of a Seminar held at 17(4): 265-268. CIAT, Cali, Colombia, February 10-14, 1974, North UHL, C. & KAUFFMAN, J.B., 1990, Deforestation, fire sus­ Carolina State University, Soil Science Department, ceptibility, and potential tree responses to fire in the Raleigh, North Carolina, U.S.A. pp. 484-563. Eastern Amazon. Ecology, 71(2): 437-449. WELLS, F. J., 1976, The Long-Run Availability of Phospho­ UNITED STATES, COUNCIL ON ENVIRONMENTAL QUALITY (CEQ) & DEPARTMENT OF STATE, rus: A Case Study in Mineral Resource Analysis. 1980, The Global 2000 Report to the President. Per­ Johns Hopkins University Press, Baltimore, Maryland, gamon Press, New York, U.S.A. 3 Vols. U.S.A.

Rev. Brasil. Biol., 57 (4): 531-549