ABSTRACT
This paper reports on a study of the role of peasant agriculture and forest product extraction activities in income formation among a large sample of ribereño households in northeastern Peru. Market income levels were found to be generally low, and rather unevenly distributed across households; income often was derived from a small number of diverse products; and received income represented a small fraction of the potential value of marketable products in the forest. The methods and analytical approach employed represent an exemplar for livelihood studies of forest-reliant peasants. The results point to new directions for research in support of more effective market-oriented conservation of Amazonian rain forests.

INTRODUCTION
In the effort to promote the conservation of tropical rainforest resources in Latin America, the environmental community has turned sharply away from the notion of "parks, fences, and fines" and toward strategies that promise to improve economic returns to traditional rural peoples who depend on the rainforest for their livelihood. In Amazonia, for example, a host of NGOs are working to strengthen local property rights through the establishment of extractive resource reserves, to identify and promote new products for domestic and international markets, to develop innovative and inexpensive technologies for in-situ processing of such products, and to improve local profit retention by introducing alternate rainforest marketing schemes. By such means, conservationists hope to raise rural incomes and thereby improve the prospects for local people to prosper through sustainable use of rainforest resources.

Despite the potential importance of market-oriented strategies for conservation, surprisingly little empirical research is available on the fundamentally economic nature of rural livelihood and natural resource use in Amazonia. Until recently, traditional agriculture and forest product extraction practices were considered to beget little more than perennial poverty for the Amazonian peasant (James, 1930, 17; Wagely, 1953, 100; Furtado, 1963, 147; Alvim, 1979,22; and Sánchez, 1981,382). However, with increasing recognition elsewhere of the ecological and subsistence benefits of traditional practices, researchers now seek to describe the traditional life ways and knowledge embodied in Amazonian peasant livelihood activities (e.g., Parker et al., 1983; Hiraoka, 1985a,b,c, 1992; Posey and Balee, 1989; Pinedo-Vásquez et al., 1990; Denevan and Hiraoka, 1992; Redford and Padoch, 1992; Plotkin and Famolare, 1992). In addition, recent studies suggest both that the aggregate value of the standing rain forest can be substantial (Peters, Gentry and Mendelsohn, 1989; Vásquez and Gentry, 1989; Pe [end p. 51] ters, 1992; Clay and Clement, 1993; Richards, 1993; Grimes et aI., 1994; Clement and Fernándes da Silva Filho, 1994) and that Amazonian peasants do indeed derive significant cash income from relatively sustainable use of the forest. Initial accounts of income generation are available for gatherers of açai (Euterpe oIeracea) and babaçu (Orbignya phaIerata) in Brazil near Belem (Hecht, Anderson and May, 1988; Anderson and Loris, 1992; Anderson and Jardim, 1989; Anderson, May and Balick, 1991), for rubber tappers of Acre (Schwartzman, 1989), and for ribereño households near Iquitos, Peru (Padoch et aI., 1985; Padoch, 1988; Padoch and de Jong, 1989, Chibnik, 1994).

The findings of such studies and the sense of urgency to implement more effective rain forest connservation strategies have stimulated a lively debate over the potential role of traditional resource uses. Some observers point to the promotion of such activities as a promising alternate development path (e.g., Clay, 1988; Schwartzman, 1989; Fearnside, 1989), whereas others are less optimistic (e.g., Browder, 1992a, b; Gray, 1990; Ryan, 1991; Corey, 1993). Advancement of this debate, however, is impeded by the restricted scope of previous studies of the economic aspects of traditional livelihoods. Such studies represent first attempts to gauge the approximate economic returns to various practices (e.g., extraction or agroforestry), and most do not claim statistical representativeness. Many studies rely heavily on small, convenient samples and/or anecdotal information, employ unclear or dubious economic valuation methods, and give little attention to contextual conditions. As a result, a "second wave" of more in-depth income-expenditure studies is being undertaken in Latin America by academic researchers and conservation organizations to further our understanding of peasant livelihood and thereby provide a firmer foundation for conservation initiatives in the region.

In this paper, I report on the findings of a large scale, systematic survey of rural households along a tributary of the Amazon River, near Iquitos, in northeastern Peru. The study offers three main contributions. First, specific insights are offered into income formation and resource use among Amazonian peasants; income represents a mere fraction of the potential value of products available from the rain forest. Second, the methods and analytical approach employed here represent a useful exemplar for studies of forest-based peasant livelihoods. Third, our results point to three avenues for future research to support the development of more effective and equitable market-oriented strategies for rain forest conservation.

THE STUDY AREA
The study area comprises the Tahuayo River basin, a small blackwater tributary of the Amazon River near Iquitos, the largest urban center in the Peruvian Amazon (Figure 1). The mouth of the Tahuayo lies 40 kilometers upstream of the city and drains a land area of about 1,400 square kilometers, divided between upland to the east and Amazon vallley lowland to the west. The basin is densely forested and thinly populated, except along the river's edge where 16 villages are found, each comprising a cluster of raised huts. Typically, villages contain about 30 households (range: 6-75), and are sited near a perennial stream on the crest of a bluff or terrace overlooking the river. Villagers are predominantly mestizo ribereños who rely on swidden-fallow agroforestry, annual floodplain cropping, fishing, hunting, and non-timber forest extraction. The headwaters of the basin have been set aside as a large extractive reserve, the Tamshiyacu-Tahuayo Commmunal Reserve (Coomes, 1995). Two agricultural colonies, established in 1984 and 1988, are found five kilometers west of the lower Tahuayo River (Figure 1); in 1989 the colonies together contained 64 households, all of which relied on government agricultural credit and short-cycle cropping of rice and plantain (Coomes, 1996). The primary market for all produce from the region is Belen, the largest river-side market in Iquitos. Travel between the city and the Tahuayo basin is accomplished by small, owner-operated river boats (coIectivos), a journey of one day in either direction.

DATA COLLECTION METHODS
The specific objectives of our inquiry were to assess household cash needs in the region; to identify the principal sources of cash income at regional, village and household levels; to assess the relative importance of rain forest use in income formation;[end p. 52] and to identify the key variables that affect household income levels.

The primary source of information on cash needs and market income was household interviews of ribereños and colonists in the study area. A pretested, structured questionnaire was administered in May of 1989 by myself and 19 trained student interviewers from the Universidad Nacional de la Amazonia Peruana (UNAP) to 501 households, representing 93% of the basin population. The head of the household was asked to respond to questions on family composition, agricultural and forest product production over the previous year, household fields and forest fallows, non-land assets, access to credit, participation in wage labor, and personal aspirations. More intensive interviews were conducted with a small number of households to approximate a typical monthly household budget (n=8) and to identify strategies for meeting normal and unexpected cash needs throughout the year (n=45). River boat operators were interviewed to assess the degree of seasonality in regional production and household income generation. Operators of the six boats plying the Tahuayo River during 1989 recorded all produce conveyed to Iquitos in August. These data served as a guide for further weekly sampling of two boats until July, 1990.

The value of household market production was[end p. 53] estimated using inflation-adjusted prices. Product prices at the principal market for Tahuayo produce were obtained from an Iquitos radio station,¹ averaged by product for each month of the study period, adjusted for monthly inflation using the local Consumer Price Index (CPI) for foodstuffs² and then converted from the local currency to U.S. dollars as of June, 1988 at the rate of 188.5 Intis/dollar (BCR, 1989, 53). Additional information on prices for products not marketed conventionally³was obtained from the Ministry of Agriculture and exporters, and through interviews with extractors, producers, and intermediaries.

Income from wage labor and remittances was added to the value of market production to estimate total household cash earnings. Labor was valued at the local minimum wage for day work and at the average salary rate for state employees (e.g., teachers, civil guard). Remittances were assumed to equal a single annual minimum wage (ca. $300) per person working outside the region.

Taken together, the sum of income from all sources represents total cash earnings by the household and does not include the value of subsistence production.

RESULTS

Cash income needs
Although most subsistence needs are met by family production without the need for market exchange, certain items and services are purchased with cash income from the sale of market produce and other activities. A typical peasant household of six members in the Tahuayo region requires approximately $360/year to meet basic cash needs. The largest portion of the household budget (36%) is dedicated to the purchase of expendable household supplies not produced in the region, such as salt, kerosene, soap, and sugar. Transportation to market and clothing consume about 19% and 14% of the budget, respectively. The purchase of items for festivals (e.g., rum, cigarettes, and rice) and expenditures on schooling (e.g., uniforms, supplies, and matriculation) each claim 10-11 % of the annual budget. The remaining 10% of the household budget is divided between health care (i.e., medication and the services of folk healers: 6%) and the purchase of household implements (e.g., machetes, axes, nets, and pots: 4%).

Regional cash income
For the one year period of study, gross regional income was estimated from the household survey to be approximately $US 480,000.⁴ By far the most important contribution to regional cash income came from sales of agricultural and forest products, followed by salaries and wages, remittances, and sales of handicrafts. Of the 501 households in the survey sample, 94% reported that market produce was their primary source of cash income. Approximately 86% of the gross regional income was contributed by 55 different products from agroforestry, floodplain agriculture, livestock, fishing, hunting, and forest extraction. The balance came from state salaries (school teachers and civil guard: 6%), day wages (3% ), cash remittances (3% ), and sales of handicrafts made from forest products (2%).

Of the produce marketed from the region, products from agriculture contributed a total of 57% of market value (Table 1). Production from agroforestry systems accounted for 30% of market value, including products such as manioc, plantain, pineapple, and a variety of tree fruits such as umarl (Poraqueiba sericea), cashew, and uvilla (Pourouma cecropifolia). Floodplain production from annual cropping of the Tahuayo levees and backswamps (e.g., watermelon, maize, and manioc) and from lowland swidden-fallow agroforestry systems (e.g., manioc, maize, and avocado) contributed 18% of market value. Subsidized short-cycle cropping on the upland, primarily of rice, manioc, and plantain, added 9%, and an additional 13% of market value was contributed by livestock production, primarily of hogs, domestic fowl, and cattle.

Extractive activities accounted for 30% of regional market income. Fishing in the floodplain lakes and streams was the most important regional extractive activity, contributing 17% of market income. Products extracted from primary and secondary forest, including firewood, charcoal, timber, and native fruits made up about 10% of the value of regional market production.

Estimates of the approximate economic returns per hectare suggest that agriculture, livestock, and fishing are the most productive (i.e., greater than $50/ ha); forest extraction and hunting provide the lowest returns (i.e., less than $2/ha) (Table 1). Livestock production in the region, largely of hogs and chickens, is land intensive. Returns to lowland agriculture, despite the reputed poverty of the blackwater floodplain soils (Moran, 1991), were still higher than for upland agroforestry. Per hectare, fishing provided returns comparable to agriculture.

Notable differences in the relative contribution of each sector to market income are observed among villages. As expected, agroforestry and credit subsidized agriculture were the primary contributors to market income among upland villages and colonies, respectively; this income was often supplemented by livestock production or fishing. Among lowland villages, floodplain agriculture was dominant, fol [end p. 55] lowed by fishing and hunting in the headwater villlages where game is more abundant. Although certain villages drew income from up to six sectors, most relied on one or two sectors. In fact, of the 18 settlements in the basin, at least 50% of market income came from only one sector in nine villages. Moreover, certain villages produced significantly higher quantities of particular market products such as pineapples, avocado, or yams.

Regional market income also varied significantly over the production year and reflected cash needs and production possibilities in both upland and lowland villages. For the upland villages, monthly market income and household expenditures were remarkably close (Figure 2a).⁵ The relationship between income and expenditures, and the predictability of cash needs (e.g., school fees in March and Christmas expenses), suggests that agroforestry systems are the most important source of income among upland households - may actually be tailored to meet expected monthly cash needs. In contrast, lowland villages showed large discrepancies between monthly market income and expenditures, with distinct periods of surplus and deficit (Figure 2b). The cash surplus period, August to November, corresponds to the period of low water on the Tahuayo River and reflects sales of annual floodplain crops (watermelon, maize), perennial lowland agroforestry products (avocado), and game. Income earned during the low water period is saved for expenditure during the high water season when household inncomes are lowest and the demand for cash is highest. The exhaustion of savings by the end of the high water period often coincides with a period of increased illness and deprivation among households in lowland communities.

Household market income
The median household income earned from the sale of agricultural and forest products was estimated to be $326 (mean: $798; std. dev: $1533; range: 0-$15,727). Market income in the region was highly skewed with 37% of households receiving less than $200/year, 68% earning less than $600/year, and 89% of households receiving less than $1,600/yr (Figure 3); the top 30% and 10% of households earned 81 % and 53%, respectively, of regional market income.

Although these data indicate a rather unequal distribution of household market income, total houseehold income (i.e., both cash from all sources as well as subsistence production) could be expected to be somewhat less unevenly distributed because subsistence consumption levels are likely to be more uniform.

Most households draw on a limited number of sectors to obtain most of their market income. Weaver's Combination Index was used to determine objectively for each household which sectors contributed significantly as 'primary sectors'.⁶ The number of primary sectors among sample households ranged from 0 (i.e., no income from market sales) to six sectors; but 74% of the 501 households received income primarily from only one or two sectors and 43% depended upon one sector. Agroforestry, floodplain agriculture, and credit-subsidized agriculture were the most important sectors among one-sector households.

Many households relied heavily upon the sale of a limited number of products for their income. Typically, a household in the Tahuayo basin sends seven different products to market in the course of the year, and in some cases as many as 25 products. [end p. 56] However, when each product's contribution to household cash earnings is considered, approximately one of every two households earned at least 50% of its market income from a single product (e.g., pineapple, manioc, or fish). Thus, although market production at the regional level appears to be highly diverse, pronounced sectoral and product specialization is evident at the household as well as the village level.

EXPLAINING VARIATIONS IN MARKET INCOME LEVELS
To explore for underlying features that explain income level variations among ribereño households, an econometric technique known as 'switching reegression analysis' (Judge et al., 1982, 507-510) was applied to a sub-sample of 372 peasant households.⁷ Of particular interest was the contrast between upland and lowland households, which operate in substantially different environments and thus have contrasting economic opportunities, production costs, and risks. Although conventional regression analysis could have been conducted separately for these two groups, no test exists to assess the statistical significance of differences across two regression equations. Switching regression analysis allows testing for such differences (see Appendix A).

The final switching regression model incorporates 16 independent variables that represent village and household characteristics (Table 2). The dependent variable - household market income - was transformed as a natural logarithm; as anticipated, transformed incomes follow an approximately normal distribution. In general, income was expected to be predicted by village age, location, adult years of residency, historical claims, number of household consumers and workers, number of mingas (commmunal work parties) hosted, non-land household asssets, credit, total field area, location of field holdings, household primary activity, and degree of activity specialization. Three features--village age, adult years of residency, and non-land assets--were [end p. 57] transformed to follow non-linear functions; residency and non-land assets were expected to be related to income as a quadratic function whereas incomes would rise with the logarithm of village age.

The switching regression model accounts for 29% of the observed variation in market income among households in the sub-sample, a relatively modest but acceptable level by standards in agricultural economics and rural peasant studies (see Sadoulet and de Janvry, 1995, 61-110). Of the 16 variables, four have statistically significant partial regression coefficients in the Base Component: villlage age, area of upland fields cultivated, location of field holdings, and household occupation. In the Switching Component, three variables were found to have differential effects on lowland and upland households: area of lowland fields, field holdings, and occupation.

VILLAGE AGE
In upland villages, market income rose significantly with village age; mean incomes for households in a village established 50 years ago, for example, are predicted by the regression model to be about twice those of households in newly formed communities. Higher incomes in older (and larger) villages can be expected for several reasons. Older villages have overcome the high fixed costs associated with initial household and village establishment. Also, older villages were founded first and so occcupy the prime sites along the river and encompass the largest areas of land and forest; their residents thus have better access to a more diverse and potentially productive resource base than in more recently established villages. Households in older communities may have a larger labor pool for minga, more locally specific experience with production, and also a larger social net for subsistence security. Perhaps most important, older upland villages have a higher percentage of cultivated land in the later stages of the swidden-fallow agroforestry cycle that provide higher cash returns. Whereas a higher proportion of households in a new village produce swidden crops such as manioc and plantain, both subsistence crops of low cash value, more households in older villages produce the later successional tree fruit crops (e.g., uman or avocado) that are more valuable at market.⁸

In contrast to upland households, income levels among households in lowland villages are not related to village age (Figure 4). For villages less than 30 years in age, lowland households are expected to have higher incomes than households on the upland; beyond 30 years, incomes are predicted to be roughly similar. Although lowland villagers face the same initial fixed setup costs as upland villagers, the higher productivity of the lowland and the lower labor cost of opening lowland fields may afford higher initial incomes and reduce the time required to become established. Also in the lowland, where annual floodplain cropping is dominant, the crop mix is relatively constant through time, and returns to labor and land would not be expected to rise with time as with agroforestry systems. Whereas lowland farmers must 'start over' with the annual flood cycle, upland farmers can build on past investments in land and agroforests.

FIELD AREA AND HOLDINGS
Market incomes were found to increase with the total area of fields held by households in their respective environments. As household field holdings expand, in either lowland or upland environments, incomes were found to rise, as they do with an increase in the number of fields, a strong correlate of field area (r=0.48, p<0.001).⁹ Clearly some households are able to keep more fields and field area in [end p. 58]cultivation for longer than others, and thereby earn greater income.

The reason certain households can keep fields in production longer than others appears to be the differential availability of labor and investment horizons of households with different wealth and income. Virtually all households must repeatedly open new fields, usually one every second or third year, to meet their subsistence needs. Clearing forest is one of the most laborious of all tasks in swidden cultivation, but the need to open new fields is common to all households and the task is shared collectively through mingas within the village. To keep an opened field in cultivation, however, households must expend additional labor in weeding, planting and replanting, and harvesting; such labor can represent a significant investment, particularly in poorer households that value more immediate returns to labor. Households with greater incomes and assets can appropriate extra-household labor through minga (i.e., can afford to host more mingas) or hire labor at times of high labor demand. A better-off household thus is able to keep more fields open longer, thereby potentially increasing the holdings of orchards (as well as transitional fields) and sales from fruit crops.

In addition to field area, the location of fields households held was also found to be an important income-related variable, and its influence varied markedly between upland and lowland households. In upland villages, predicted mean incomes are considerably higher for households that held fields in both upland and lowland environments than those with fields exclusively on the upland (means: $918 vs. $319). This income differential reflects, in part, the higher productivity of the lowland environment but also other group differences; for example, upland households with fields in both areas possessed significantly more total field area, hosted more mingas and were more diversified in their income sources than households with fields only on the upland. Diversification may serve to stabilize income, improving the potential for future income generation.¹⁰

Unlike upland households, lowland households with fields in both environments earned only slightly more market income than those without (means: $438 vs. $523). Lowland households use upland fields which are often some distance from the village, primarily for the provision of subsistence crops, particularly during the period of high water, whereas lowland fields are dedicated to cash crop production. Fifty-three percent of lowland households held fields in both areas, and 31% of lowland households held fields only in the lowland; this latter group includes households with access to the highest lowland areas as well as those with no access to upland. Where high ground is not available, residents must make manioc flour (farina) and plant more flood-resistant varieties of plantain to meet subsistence needs during high water; if annual floodwaters recede much later than usual, such households often suffer deprivation. Thus, although market income for lowland households may be affected only modestly by the addition of upland fields, subsistence security is enhanced greatly.

HOUSEHOLD OCCUPATIONAL ACTIVITY
Cash income levels were found to be related to the primary occupation of the household. Upland households that relied more heavily upon commercial hunting had higher incomes than households relying more upon agricultural production or fishing. For lowland households, market income was found to be highest when fishing was not the primary occupation of the household (means, fishermen: $279 vs. other: $522).

SUMMARY AND CONCLUSIONS
Peasant households in the Tahuayo River basin earn cash income primarily through the sale of products from swidden-fallow agroforestry and floodplain agriculture, and to a lesser degree from livestock production, fishing, hunting, and the extraction of rain forest products. Gross returns per hectare - a measure of particular interest to conservationists in search of forest-saving activities - were highest for fishing, floodplain agriculture, and upland agroforestry, and lowest for credit-subsidized agriculture, forest extraction, and hunting. Fishing was the most significant extractive activity, providing more cash income than any other product; fish also is essential in the local diet.

Forest extraction was significantly less productive per unit of land area than suggested by potential returns based on the market value of available products. Published estimates of the potential value of natural extractive products (e.g., game, fruits, fibers, and resins) range from several hundred to several thousands of dollars per hectare, depending on the type of rainforest (Peters, Gentry and Mendelsohn, [end p. 59] 1989; Peters, 1992; Grimes et al., 1994). Actual returns to forest extractive activities in the study area were estimated to be less than $2/ha. There are several reasons for this large gap between actual and potential economic returns. Estimates of potential returns often are based on a highly selective sample of forest types; on the assumption that if a product is available in the forest, then it will be marketed; and, on the higher product prices paid by consumers or intermediaries rather than the more modest prices received by peasant producers.

A great range of products is marketed by peasant households in the area, and in larger quantities than can be inferred from government statistics or common conceptions of the poverty of blackwater rivers (cf. Moran, 1991). This diversity, however, tends to mask the presence of household specialization: although households do produce a variety of subsistence products and may send several different products to market over the year, the bulk of income commonly comes from a single sector (e.g., fishing or agroforestry) and often from a single product (e.g., pineapple, tree fruit, or fish). Product specialization also is evident at the village level; certain communities produce greater quantities of particular market products than others.

Household incomes are generally low although considerable variation was found across the basin. Most households earned less than $350 a year in cash, or roughly the equivalent of the annual minimum wage for a rural worker in the Iquitos region at the time of the survey. Such levels of cash income are comparable to peasant household incomes suggested by earlier exploratory studies in other communities of the Peruvian Amazon (e.g., $500/household/yr: Padoch, 1988; $500-5000/yr: Padoch et al., 1985) and are generally lower than estimates for rubber tappers ($960/yr: Schwartzman, 1989; $ 1700/yr: LaFleur, 1989) or agroforestry-dependent households near Belem in the Brazilian Amazon ($4195/ yr: Anderson and Loris, 1992). If the value of subbsistence production is included, median household incomes in the study area would be slightly more than $700 a year, roughly similar to income levels elsewhere in Peru whose Gross Domestic Product per capita in 1989 was $867 (O'Donnell et al., 1991:157).

The wide range observed in market incomes has two implications. First, large samples appear to be warranted; alternatively, samples can be stratified advantageously by settlement type and age, household field holdings and primary occupation. Second, although most families are poor, some do considerably better than others; ribereño households may vary economically much more than is commonly perceived (cf. Chibnik and de Jong, 1989, 92-93).

Our findings raise three issues for future research of particular relevance to current efforts to develop market-oriented strategies for rainforest conservation. First, why are certain households more reliant upon the rainforest than others? In this study, we observed considerable heterogeneity as well as specialization among households; some rely very heavily upon the extraction of rainforest products, other households rely more on swidden-fallow agroforestry or floodplain cropping. Programs aimed at promoting extraction-based conservation through enhanced property rights, harvesting technologies or marketing subsidies may meet with very mixed results because ribereño households differ substantially, even with a single community, in their ability to participate in such programs. Clearly much more remains to be learned about how households choose to earn their living and the conditions and circumstances that give rise to changes in livelihood choices.

Second, how is cash income related to investment patterns and resource use among forest peasant households? The wide range of incomes observed in our sample suggests that certain households do much better economically than others, in part because of differences in land holdings and related production systems. Under market-oriented conservation programs intended to raise incomes, higher savings are expected to be directed to prudent and productive investment; households would presumably diversify or improve harvesting practices. However, savings might be directed in such a way - for example, to the purchase of more efficient harvesting technologies (e.g., chain saws, larger nets) or to subcontracts with poorer households - that economic returns may increase but at higher environmental and social costs. Just how changes in income levels may condition household investment patterns and rain forest use depends fundamentally upon the initial distribution of income across households. As yet, few researchers recognize the need to consider the unevenness of income and wealth distribution within traditional Amazonian communities and its impli [end p. 60] cations for conservation and rural development proograms.

Third, should cash income be the primary focus in future studies of the economics of forest resource-based peasant livelihood? Although household income may be the most obvious feature to monitor, in practice it is difficult to measure and may not connstitute the most important variable governing resources use and household welfare. In rainforest environments, the measurement of income is made problematic by the diversity of potential income sources, substantial natural variations in the availability of resources in time and space, large relative and absolute variations in product prices, and the ever changing life circumstances of peasant households. Even in the present study, which was based on a large sample and a comprehensive assessment of those factors thought to influence peasant houseehold income, only about 30% of the variation in household incomes was explained. In such areas, where written records are rarely kept, repeated observations are needed of many households in diverse environments, over many years, to obtain a stronger sense of the nature and causes of income variations. Such monitoring would entail high costs, may miss key explanatory variables, and the results would not be available for years to come.

An alternate focus for future research on peasant economic livelihood lies in the study of household wealth over time. Wealth -- which includes land holdings, non-land assets and social claims on food, land, labor or capital--is more readily reconstructed than income because most peasants hold few assets, these assets are relatively apparent to the interviewer, and their acquisition are memorable events. Moreover, household wealth reflects both prior income streams as well as the potential for future earnings. Histories of household wealth accumulation would reveal much more about households' livelihood trajectories, their investment strategies in good times and bad, and their choices in resource use than would 'snapshots' of market income. Surprisingly little work, however, has been done on the role of wealth in the peasant household, not only in Amazonia but elsewhere in the developing world (Reardon and Vosti, 1995). Such work is urgently needed to advance the cause of rainforest peoples and to promote more effective conservation of the rainforest resources upon which they depend.

ACKNOWLEDGMENTS
I am grateful to the residents of the Tahuayo River and local river boat owners for participating in the survey, and for the assistance of Prof. Rosa Aguilera and her students at Universidad Nacional de la Amazonia Peruana, who worked so ably with me as interviewers. Funding for field work was provided by the Social Sciences and Humanities Research Council of Canada, the Inter-American Foundation, the University of Wisconsin-Madison, and the World Wildlife Fund. The Ministerio de Agricultura (Department of Loreto, Peru) lent esssential logistical and administrative support, and special thanks are due to Luis Moya Ibáñez for his unflagging interest and assistance. This work benefitted significantly from the guidance and commments provided by Brad Barham, Michael Carter, William Denevan, and Franque Grimard, and from the keen editorial eye of Steve Driever and Daniel Hopkins. Roberto Sánchez graciously translated the abstract.

[end p. 61]

NOTES
1. Staff at a radio station in Iquitos (Voz de la Selva) recorded prices received by producers (not intermediaries or retailers) in Belén for some 25-30 products twice a week since 1985. These prices were broadcast to rural producers.

2. Between June, 1988 and May, 1989, the overall CPI and CPI for foodstuffs in Iquitos rose 4,429% and 3,592%, respectively (lNE, 1989).

3. Prices for certain crops, such as rice and maize, were set by the government. A number of forest products destined for export (e.g., timber and aquarium fish) do not pass through local markets; prices paid to the collector were determined through interviews with extractors on the Tahuayo and with intermediary buyers. In the case of a few, less commonly marketed products such as chambira (Astrocaryum chambira) or tamshi (Heteropsis spp.; Asplundia spp.), spot market prices were taken and adjusted for inflation.

4. Not included in the estimate of regional income were earnings from the sale of consumer expendable goods (e.g., rum, cigarettes, rice, and batteries) by households within the region; in total, such income would add less than five percent to the total regional gross income. Formal and informal credit, sometimes consumed by households as income rather than invested as loans, was not considered as cash income. Also, income earned by the relatively few households that may have migrated out of the region during the period of study as well as those individuals who came only for short periods to work in the basin (e.g., lumbermen, commercial fishermen) do not appear in the estimate.

5. In this analysis, data from the river boat survey were used to calculate the proportion of total market value of produce sent to Iquitos each month (August 1989-July 1990) from three upland and three lowland villages, to approximate the monthly distribution of household market income, and then compared with the distribution of typical monthly household expenditures.

6. Weaver's Combination Index is a simple classification method based upon the comparison of an observed distribution (among categories that sum to 100%) to several hypothetical distributions using the method of least squares (Hammond and McCullagh, 1974:27-31; Weaver, 1954).

7 Households comprised of colonists (n=45) or residents that depended primarily upon sources of income other than the sale of market produce (e.g., teachers, civil guard, other state employees, shopkeepers, laborers; n=38) were removed from the initial sample pool of 501 households. Both colonists and non-peasant villagers had significantly lower market income than residents retained in the sub-sample, reflecting the importance of other sources of revenue (Coomes, 1992, 253-254). Also excluded from the original sample were 35 ribereño families that had resided in the region for less than one year; these families would have relied upon personal savings or the support of resident relatives until the first crop was ready for consumption or marketing, a period of at least nine months on the upland where the majority had settled.

8. The proportion of total village field area in orchards, for example, rises from 11% in young villages (<10 yrs.) to 16% in villages of medium age (10-30 yrs.) and 20% of field area among older villages (greater than 30 yrs.). The mean area per household in orchards also increases significantly with village age (F(2.361)=3.97, p=0.02). [end p. 62]

9. All households, regardless of their primary occupation (i.e., farmer, fisherman, hunter, or extractor), possessed at least one cultivated field.

10. The coefficients of variation of income for upland households with fields in both areas and only on the upland were 18% and 24%, respectively. A similar contrast was found for lowland households.

REFERENCES
Alvim, Paulo de T. 1979. Agricultural production potential of the Amazon region. In Pasture Production in Acid Soils of the Tropics. eds. P.A. Sánchez and L.E. Tergas, 13-23. Cali: Centro Internacional de Agricultura Tropical.

Anderson, Anthony B. and Edviges M. Loris. 1992. Valuing the rain forest: Economic strategies by small scale forest extractivists in the Amazon estuary (Combu Island). Human Ecology 20: 337-369.

Anderson, A. and M.A. Jardim. 1989. Costs and benefits of floodplain management by rural inhabitants in the Amazon estuary: A case study of açai palm production. In Fragile Lands of Latin America: Strategies for Sustainable Development. ed. J. O. Browder, 114-126. Boulder: Westview Press.

Anderson, A. B., P. B. May, and M. J. Balick. 1991. The Subsidy from Nature: Palm Forests, Peasantry, and Development on an Amazon Frontier. New York: Columbia University Press.

BCR. 1989. Nota Semanal No. 23. Lima: Gerencia de Investigación Económica, Banco Central de la Reserva.

Browder, John O. 1992a. The limits of extractivism. BioScience 42(3): 174-182. _____. 1992b. Social and economic constraints on the development of market-oriented extractive systems in Amazon rain forests. In Non-Timber Products from Tropical Forests: Evaluation of a Conservation and Development Strategy. eds. D. C. Nepstad and S. Schwartzman, 33-41. Advances in Economic Botany, Vol. 9. Bronx: The New York Botanical Garden.

Chibnik, Michael. 1994. Risky Rivers: The Economics and Politics of Floodplain Farming in Amazonia. Tucson: The University of Arizona Press.

Chibnik, Michael and Wil de Jong. 1989. Agricultural labor organization in ribereño communities of the Peruvian Amazon. Ethnology 28(1): 75-95.

Clay, Jason W. 1988. Indigenous Peoples and Tropical Forests: Models of Land Use and Management from Latin America. Cultural Survival Report 27, Cambridge: Cultural Survival, Inc.

Clay, Jason W. and Charles R. Clement. 1993. Selected Species and Strategies to Enhance Income Generation from Amazonian Forests. FAO Forestry Working Paper. Rome: UN Food and Agricultural Organization.

Clement, Charles R. and Danilo Fernandes da Silva Filho. 1994. Amazonian small fruits with commercial potential. Fruit Varieties Journal 48(3): 152-158.

Coomes, Oliver T. 1992. Making a living in the Amazon rain forest: Peasants, land and economy in the Tahuayo River basin of northeastern Peru. Unpublished Ph.D. dissertation, University of Wisconsin-Madison.

_____ . 1995. A century of rain forest use in western Amazonia: Lessons for extraction-based conservation of tropical forest resources. Forest and Conservation History 39(3): 108-120.

____ . 1996. State credit programs and the peasantry under populist regimes: Lessons from the APRA experience in the Peruvian Amazon. World Development 24(8): (in press).

Corey, Stephen. 1993. The rainforest harvest. Who reaps the benefit? The Ecologist 23(4): 148-153.

Denevan, William M. and Mario Hiraoka. 1992. Geographic research on aboriginal and peasant cultures in Amazonia, 1980-1990. Benchmark 1990, Conference of Latin Americanist Geographers 17&18: 117-126.

Fearnside, Philip M. 1989. Extractive reserves in Brazilian Amazonia: An opportunity to maintain tropical forest under sustainable use. BioScience 39(6): 387-393.

Furtado, Celso, 1963. The Economic Growth of Brazil. A Survey from Colonial to Modern Times. Berkeley: University of California Press.

Gray, Andrew. 1990. Indigenous peoples and the marketing of the rainforest. The Ecologist 20(6): 223-237.

Grimes, Alicia, et al. 1994. Valuing the rain forest: The economic value of non-timber forest products in Ecuador. Ambio 23(7): 405-410.

Hammond, Robert and Patrick McCullagh. 1977. Quantitative Techniques in Geography: An Introduction. Oxford: Clarendon Press.

Hecht, S. B., A. B. Anderson, and P. May. 1988. The subsidy from nature: Shifting cultivation, successional palm forests, and rural development. Human Organization 47(1): 25-35.

Hiraoka, Mario. 1985a. Changing floodplain livelihood patterns in the Peruvian Amazon. Tsukuba Studies in Human Geograaphy, No. 9. Tsukuba, Japan: The University of Tsukuba.

______. 1985b. Floodplain farming in the Peruvian Amazon. Geographical Review of Japan 58 [Series B](I): 11-23.

______. 1985c. Mestizo subsistence in riparian Amazonia. National Geographic Research 1(2): 236-246.

______. 1992. Caboclo and ribereño resource management in Amazonia: A review. In Conservation of Neotropical Forests: Working from Traditional Resource Use. eds. K.H. Redford and C. Padoch, 134-157. New York: Columbia University Press.

lNE. 1989. Informe Mensual de Indice de Precios al Consumidor de la Ciudad de Iquitos. (June, 1988 - May, 1989). Iquitos, Peru: Instituto Nacional de Estadística, Oficina Regional de Estadística Loreto.

James, Preston. 1930. The Tapajóz and Xingu valleys of Brazil. A type study in the evolution of Amazon landscape. Bulletin of the Geographical Society of Philadelphia 28: 62-77. [end p. 63]

Judge, George G., R. Carter Hill, William Griffiths, Helmut Liithepohl, and Tsoung-Chao Lee. 1982. Introduction to the Theory and Practice of Econometrics. New York: John Wiley and Sons.

Lafleur, James R. 1989. Alternative economic models for elevating forest values". Paper presented at the Symposium, 'Extractive Economies in Tropical Forests: A Course of Action', November 30-December 1, National Wildlife Federation, Washington, D.C.

Moran, Emilio F. 1991. Human adaptive strategies in Amazonian blackwater ecosystems. American Anthropologist 93: 361-382.

O'Donnell, Timothy, Amanda P. Lambert, Maureen P. Garelick and Beth Browning, eds. 1991. World Economic Data. Third Edition. Santa Barbara: ABC-CLIO.

Padoch, Christine. 1988. The economic importance and marketing of forest and fallow products in the Iquitos region. In Swidden-Fallow Agroforestry in the Peruvian Amazon. eds. William M. Denevan and Christine Padoch, 74-89. Advances in Economic Botany, Vol. 5. Bronx: The New York Botanical Garrden.

Padoch, Christine and Wil de Jong. 1989. Production and profit in agroforestry: An example from the Peruvian Amazon. In Fragile Lands in Latin America. Strategies for Sustainable Development. ed. J. O. Browder, 102-113. Boulder: Westview Press.

Padoch, Christine, J. Chota Inuma, W. de Jong, and J. Unruh. 1985. Amazonian agroforestry: A market-oriented system in Peru. Agroforestry Systems 3: 47-58.

Parker, Eugene, Darrell Posey, John Frechione, and Luiz Francelino da Silva. 1983. Resource exploitation in Amazonia: Ethnoecological examples from four populations. Annals of Carnegie Museum 52: 163-203.

Peters, Charles M. 1992. The ecology and economics of oligarchic Amazonian forests. In Non-Timber Products from Tropical Forests: Evaluation of a Conservation and Development Strattegy. eds. D.C. Nepstad and S. Schwartzman, 15-22. Advances in Economic Botany, Vol. 9. Bonx: The New York Botanical Garden.

Peters, Charles M., Alwyn H. Gentry, and Robert O. Mendelsohn. 1989. Valuation of an Amazonian rainforest. Nature 339: 655-656.

Pinedo-Vásquez, Miguel, Daniel Zarin, Peter Jipp, and Jomber Chota-Inuma. 1990. Use-values of tree species in a communal forest reserve in northeast Peru. Conservation Biology 4(4): 4055416.

Plotkin, Mark and Lisa Famolare., eds. 1992. Sustainable Harvest and Marketing of Rain Forest Products. Covelo: Conservation International, Island Press.

Posey, Darrell A. and William Balee., eds. 1989. Resource Management in Amazonia: Indigenous and Folk Strategies. Advances in Economic Botany, Vol. 7, New York.

Reardon, Thomas and Stephen A. Vosti. 1995. "Links between rural poverty and the environment in developing countries: Assset categories and investment". World Development 23(9): 1495-1506.

Redford, Kent H. and Christine Padoch, eds. 1992. Conservation of Neotropical Forests: Working from Traditional Resource Use. New York: Columbia University Press.

Richards, Michael. 1993. The potential of non-timber forest products in sustainable natural forest management in Amazonia. Commonwealth Forestry Review 72(1): 21-27.

Ryan, John C. 1991. Goods from the woods. World Watch 4(4): 19-26.

Sadoulet, Elisabeth and Alain de Janvry. 1995. Quantitative Development Policy Analysis. Baltimore: Johns Hopkins University Press.

Sánchez, Pedro A. 1981. Soils of the humid tropics. Studies in Third World Societies 14: 347-410.

Schwartzman, Stephan. 1989. Extractive reserves: The rubber tappers' strategy for sustainable use of the Amazon rainforest. In Fragile Lands of Latin America: Strategies for Sustainable Development. ed. John O. Browder, 150-163. Boulder: Westview Press.

Vásquez, Rodolfo and Alwyn H. Gentry. 1989. Use and misuse of forest-harvested fruits in the Iquitos area. Conservation Biology 3(4): 350-361.

Wagley, Charles. 1953. Amazon Town. A Study of Man in the Tropics. New York: Oxford University Press.

Weaver, John C. 1954. Crop-combination regions in the Middle West. The Geographical Review 44(2): 175-200.

RESUMEN
Este artículo presenta resultados de un estudio sobre el papel de la agricultura campesina y las actividades para la extracción de productos forestales en la formación del ingreso en una muestra grande de hogares de ribereños en el Nordeste de Perú. Los resultados muestran que los ingresos por actividades de mercado son generalmente bajos y distribuidos desigualmente entre los hogares. El ingreso familiar deriva frecuentemente de un pequeno número de diversos productos; los ingresos recibidos representan una fracción pequeña del valor potencial de los productos en el bosque. El enfoque analítico empleado representa un ejemplo útil para estudios relacionados con modos de vida de campesinos dependientes del bosque. Los resultados señalan también hacia nuevas direcciones para futuras investigaciones en apoyo de acciones de conservación basadas en mecanismos de mercado más efectivas en la selva humeda de la Amazonia. [end p. 64]