Promotor: dr.ir. R.A.A. Oldeman Hoogleraar in de bosteelt en bosoecologie

Co-promotor: dr. C. Padoch Scientist Institute of Economie Botany The NewYor k Botanical Garden piôO&dpi t2000

WIL DE JONG

DIVERSITY, VARIATION, AND CHANGE IN RIBERENO AGRICULTURE AND AGROFORESTRY

Proefschrift ter verkrijging van de graad van doctor in de landbouw- en milieuwetenschappen op gezag van de rector magnificus, dr. C. M. Karssen, in het openbaar te verdedigen op woensdag 25 oktober 1995 des namiddags te vier uur in de Aula van de Landbouwuniversiteit te Wageningen.

ff 933 CIP-DATAKONINKLIJK E BIBLIOTHEEK, DENHAA G

1995

Diversity, varation, and change in Ribereno agriculture and agroforestry / Wil de Jong. [S.I.;s.n.]. Fig.,Tab .

Thesis Landbouwuniversiteit Wageningen.Wit hréf .Wit h summary in Dutch.

ISBN 90-5485-469-3

Subject headings:agriculture ;Amazonia-Peru/varze aresourc eutilization/ribereno s

Riberenos, the native farmers ofth e lowland Peruvian Amazon region, subsist in an ecologically complex Amazonian varzea environment by practicing a highly diverseagriculture , and following individualistic agricultural strategies.A tota lo f 14 different agricultural methods, identified as agricultural types, and the varia­ tion inagricultura l strategies are described for two villages located atth e Ucayali river.Diversit y ofswidden-fallo w agroforestry onterr a firme lands,an d ofvarze a agroforestry is investigated. Riberefio agricultural diversity and variation inagri ­ cultural strategies can be explained as adaptations to the complex and dynamic conditions of the varzea. The case of ribereno resource use gives reason to ques­ tion several theories that have been formulated about varzea resource utilization.

i \NVBO!""''IJ:'^ •"•],\-2iT'y " ''•'O-.Cé.N A/^JÛÏ'iOt , 2.00Q

STELLINGEN î De diversiteit van ribereno landbouw in de varzea weerlegt Ross' stelling (The evolution of the Amazon peasantry. 1978) dat tegenwoordig boeren van het Amazone-gebied deze zone voornamelijk gebruikenvoo r extractieve doelen.

Riberenos voedingspatroon op basis van cassave en vis weerlegt Roosevelts theorie (Parmana: Prehistorie maize and manioc subsistence along the Amazon and Orinoco, 1980) dat slechts als gevolg van de introductie van maïs in het Amazone-gebied groepen met meer complexe sociale stratificaties zich konden ontwikkelen.

Riberenos onafhankelijke landbouw demonstreert dat er geen complexe sociale organisatie nodig is om landbouw te kunnen bedrijven in de varzea, en tevens dat onafhankelijke boeren efficiënter met de ecologische diversiteit en de onvoorspelbaarheid van dit gebied overweg kunnen dan georganiseerd opererendegroepen .

De bereidheid van riberenos om, zoals ze gewoon zijn, riskante landbouw-produktiemethodento e te passen, zal verminderen als de nu aanwezige mogelijkheden zullen verdwijnen om binnen een korte tijd de consequentiesva n mislukte oogsten op te vangen.

Agroforestry-systemen komen minder voor in varzea met haar vruchtbare bodems, maar worden daar intensiever beheerd dan die op de armere terra firme.

De ecologische complexiteit van de varzea en de landbouwdiversiteit van riberenos vereisen een landbouwplanning en -voorlichting die in eerste instantie gericht zijn op het verbeteren van de economische situatie van de landgebruikers, en pas in de tweede plaats op het optimaliseren van de produktie van het gebied.

De duurzaamheid van de landbouw is geen objectief meetbare eigenschap, maar een standaard die nog te weinig mede gedefinieerd wordt door diegenen van wie de huidige landbouw- produktiemethodenvaa k als niet duurzaambeoordeel d worden. 8 Het gebruik van niet-hout bosprodukten kan een belangrijke bijdrage leveren aan de economische ontwikkeling van kleinschalige boeren, maar het is te verwachten dat dit slechts van gering belang zal zijn voor het behoudva n niet of weinig verstoord tropisch regenwoud.

Het gebruik van systeemtheorie in de analyse van landbouw-produktiemethodendi e direct en feitelijk beschreven kunnen worden, introduceert een extra abstractieniveau, wat in strijd is met de regel dat theorieën zo eenvoudig mogelijk dienent e zijn. 10 Vele proefschriften maken een fase door waarin ze voor een hele lange tijd bijna klaarzijn .

Wil de Jong. Diversity, variation, and change inribereno agricultur e and agroforestry. 25 oktober 1995. A Carmen;

Esta obra tela dedico ati, miesposa, micompanera, miamiga. Alfinal de cuenta, esfélo quenos hace alcanzar a losdestinos quenos ponemos. La tuya ha sido mas grande delo que podia esperar. Sinella esta obra no hubiera sidoposible. CONTENTS

GLOSSARY x

PREFACE xvii

1 VARZEA RESOURCE USE: ECOLOGY AND HISTORY 1 Introduction 1 Varzea ecology and land use 3 Pre-contact varzea resource use 4 Contemporary varzea resource use 6 Ribereno history 8 Riberefio resource use 9 Ribereno agroforestry... 10 Research questions and rational 11 Theories on varzea resource management questioned 13

2 METHODOLOLOGIES AND METHODS 15 Introduction 15 Definition of concepts 16 Scheduling of the research 17 Selection of research units 18 Data gathered and research methods 19

3 THE PEOPLE AND THE PLACES 24 Introduction 24 A brief history of riberefios 25 Ribereno political organization andsettlemen t patterns 27 The Ucayali environment 30 Climate soils and vegetation 33 Santa Rosa: people and place 35 A history and geography ofYanallp a 37

4 AGRICULTURAL DIVERSITY IN FLOODPLAIN FARMING 40 Introduction 40 Agricultural types, a concept for land use classification 41 Rice agricultural types 44 Barreal rice fields 44 Restinga rice fields 48 Aguajal rice fields 50 Terra firme rice fields 51 Chacra agricultural types 52 Restinga chacras 52 Terra firme chacras 54 Plantain agricultural types 56 Restinga plantain fields 56 Terra firme plantain fields 57 Forest garden agricultural types 59 Restinga forest gardens 59 Terra firme forest gardens 59 Other foodplain agricultural types 60 Restinga corn fields 60 Playa cowpea fields 62 Restingajut e fields 63 Restinga vegetable fields 64

5 VARIATION IN AGRICULTURAL STRATEGIES 65 Introduction 65 Factors influencing ribereno agriculture 66 Data sets used and methods of data-analysis 69

vi Results 71 Frequency of agricultural types in 1985/1986 71 Changes in utilization of agricultural types 75 Magnitudes of farming enterprises 77 Changes in agricultural strategies 80 Discussion 82 Preferred farming strategies 82 Household production 84 Rice versus corn cash cropping 84 Strategic farming in the floodplain 87

6 FOREST GARDENING IN SANTA ROSA 90 Introduction 90 Swidden-fallow agroforestry 91 Santa Rosa forest gardens 93 Results 95 Diversity of managed species 95 Weeding patterns 100 Yield levels 102 Management ofriberefio fores t gardens 104 Pathways for intensification of riberefio agroforestry 106

7 FLOODPLAIN AGROFORESTRY INYANALLP A 109 Introduction 109 Yanallpafloodplain agroforestr y 111 Yanallpa agroforestry fields 112 Results 115 Diversity of managed species 115 Weeding patterns and yield levels 120 Varzea agroforestry strategies 122 Options for varzeaagroforestr y development 124

vu 8 CONCLUSIONS AND PROSPECTIVES 126 Introduction 126 Summary of findings 128 Agricultural diversity 128 Variation in agricultural strategies 129 Diversity of ribereno agroforestry 130 Implications for theories on varzea resource management 131 Varzea agriculture in Peru versus Brazil 131 Requirement of complex social organization for varzea exploitation 133 Dependence of pre-contact groups on corn protein 134 Evaluation of the research 135 The future of varzea agriculture in Amazonian Peru 138

LITERATURE CITED 141

NEDERLANDSE SAMENVATTING 165

LEVENSOVERZICHT WIL DE JONG 168

vin FIGURES AND TABLES

Figure 3.1 : General overview of the research area and ofth e villages 29 researched Figure 3.2 : Map of the surroundings of the villages researched 31 Figure 3.3 : Monthly water level in the Ucayali river, Jenaro Herrera, 1985 33 Figure4. 1 : General calendar agricultural types with marked seasonality ... 44 Figure 5.1 : Frequencies ofagricultura l types among Santa Rosa farmers,. 72 1985/1986 and 1989 Figure 5.2 : Frequencies of agricultural types amongYanallp a farmers, 74 1985/1986 and 1989 Figure 5.3 : Number of agricultural types per farmer in Santa Rosa and.... 76 Yanallpa, 1985/1986 Figure 5.4 : Number of agricultural types per farmer in Santa Rosaand.. . 78 Yanallpa, 1989 Figure 5.5 : Changes in number ofagricultura l types per farmer between .. 81 1985/1986 and 1989

Figure 6.1 : Non managed vegetation index (IDmv) in four forest gardens, 101 Santa Rosa

Table 2.1 : Inmv (non managed vegetation) index values to estimate 22 weeding patterns in forest gardens Table4. 1 : General features of theagricultura l types described in 45 chapter four Table4. 2 : Comparison of fourrice agricultura l types 47 Table 6.1 : Densities of managed plant species in seven forest gardens,.... 96 Santa Rosa Table 6.2 : Production per hectare of four forest gardens, Santa Rosa 103 Table 7.1 : Densities of managed plant species in nine varzea 116 agroforestry fields, Yanallpa

IX GLOSSARY

Acrisols Strongly weathered soils with low CEC and base saturation similart oUltisols ;FAO/UNESC O clas­ sification.

Agente municipal Village authority who represent the municipality to which the village belongs, and who is elected by the villagers.

Agricultural type Unique site-cropcombinatio n which hasa typica l set of agricultural techniques, scheduling of ac­ tivities,productio n levels,risk ,an dprincipa l des­ tination of output.

Aguajal Forest which is dominated by the palm Mauritia flexuosa, probably an early succesional phase of varzea forest.

Amerindians People belonging to nativeAmerica n groups.

Aquept Soil which issaturate d with watera tsom eperio d in the year; Inceptisol suborder USDA classifica­ tion.

Aquoll Soils with the characteristic of wetness; Mollisol sub order USDA soil classification.

Ashaninka Peruvian indigenousgrou prelate dt oCampas ,liv ­ ingi nth euppe rreache so fth eUcayl i riveran dit s tributaries.

Bank caving Cavingo fth erive rban ko rlevee s asa consequenc e of water erosion. Bank-side shoaling Formationo fshoals ,shallo wplaces ,alon gth e river levees duringhig hwate rlevel ,whic hbecom e mud flatso r sand bars after the receding of the water during low water level season.

Barbasco Lonchocarpus nicou,LEGUMINOSAE , a shrub, whichcontain s rotenone,a natura l insecticide, for which itwa sgrow ni nth e 1940s, butlate rreplace d by factory produced substances.

Barreal Mudflats, whichar eth edeposit s ofsilt ysediment s of the Ucayali and Amazon river which appear during low water level season.

Biotope A micro environment with a relatively uniform land form, climate, soil, and biota.

Bora Amerindian group, living in the border area be­ tween Peru and Colombia.

Caboclo Rural people the Brazilian Amazon region, countrepart of Peruvian riberefios (see also ribereno).

Chacra Swidden made on terrafirme, an d planted with a mixture of crops, but mainly with manioc.

Chiefdoms Sociopolitica l organization ofa triba lsociet yrule d by chiefs with limited political power.

Cocama Amerindian group livingalon gth euppe rAmazo n and Ucayali river, now largely integrated in the ribereno society.

Cocamilla Amerindian group livingalon gth eMarano nriver ,

XI now largely integrated in ribereno society.

Conibo Amerindian group living along the middle reaches of the Ucayali river.

Cowpea Vigna unguiculaia, LEGUMINOSAE, a pulse commonly planted on sand beaches (playas) in the varzea.

Debt-peonage Work obligation which is emposed on laborers by debts they have incurred through the purchase of goods from the employer.

Farina Roasted manioc which was first gridded, washed, and dried.

Fluvent Alluvial soils with very simple profiles; Entisol sub order USDA classification.

Fluvisols Soils deposited by waterwit h very little subsequent alteration; FAO/UNESCO classification.

Fundo Large agricultural estates inth e Peruvian Amazon, which appeared in the second half of the last cen­ tury, but most of which had been abandoned by 1960

Gleysols Poorly drained soils with mottled or reduced ho­ rizons due to wetness; FAO/UNESCO classifica­ tion.

Illite Clay sized soil mineral derived from mica, occur­ ring ina n environment of relatively high silica and alumina with partial stripping of potassium of interlayer.

Xll Non managed vegetation index, used to measure the state ofth e weed vegetation ina fores t garden field in order to assess the weedingpattern .

Lancha Passenger and cargo boat of thetyp e whichcom ­ monly travels the largerAmazonia n rivers.

Levee Naturalembankment ,th eresul to fdeposit so fsedi ­ ments which are carried by the river.

Loreto North eastern department, the largest administra­ tive subdivision, of Peru.

Lovilla Nameo fth esmal riverl whic h flows East ofSant a Rosa, and which joins the Ucayali down stream from of the village.

Manioc Manihot esculenta, EUPHORBIACEAE, a tuber crop widely cultivated in SouthAmerica , butals o elsewehre inth e world.

Mayoruna Amerindiangrou pwhic hinhabit sth eare abetwee n the Ucayali and theYavar i rivers.

Montmorillonite Clay mineral with expansible layers, occurringi n environmentswit hhig hsilic aan dalumin aconcen ­ trations with slow moving or stagnant water.

Non managed vegetation index See Inmv

Over-bank depositions Deposits ofsedimen t which occursbeyon d levees adjacent to rivers.

Oxisols Strongly weathered soils consisting of kaolinite, quartz and hydrated oxydes; order USDAclassi ­ fication.

Xlll Paleudult Ultisol withol ddevelopmen t ina humi dmoistur e regime; great group USDA classification.

Playa Sand beach, the result of deposition of the heavy sandy sediments carried byth e river.

Point bar Pointed sedimentation areas, caused by uneven deposition in a meandering river.

Pre-contact From before the arrival ofth e Spanish conquista­ dores.

Puesto Smaller estates found alongside the Ucayali or Amazon river which usually engaged in agricul­ ture,bu tals o extraction, trade,o rfirewoo d supply for steam ships.

Quichua Amerindiangrou p inhabitingth euppe rreache so f theNap o river

Restinga Peruvian word for land on a levee, subdivided in high and low restinga.

Ribereno Inhabitantso fth ePeruvia nAmazon ,mos to fwhic h live along the larger rivers, and who are mostly small farmers (seeals o chapterthree) .

Ridge Seelevee .

Runa Amerindian group living in EcuadorianAmazon .

Sand bar Sandy deposits in the Ucayali or Amazon river, equivalent to playa.

Swale Depressionalongsid ea ridg eo rleve ei nth evarze a landscape.

xiv Swidden Agriculturalfield whic h isth e result ofslashin ga forest, orshru bvegetation ,an dwhic hi sintensivel y cultivated only for one or a fewyears .

Swidden-fallow Fallow forest garden, the result of vegetation de­ veloped on land which was previously used as swidden.

Swidden-fallow agroforestry Agroforestry ofswidden-fallow , whichmos t often areactivel ymanage dfo rfruits , poolso rothe rprod ­ ucts.

Tapiche River whichjoin s the Ucayali near Requena, and which was an important destination for produce from Yanallpa when rubber was extracted there.

Taungya Land use system in which farmers are allowd to growannua lo rsemi-perrenia lcrop so nlan dwhic h hasbee nplante d totree s bya governmen tagency , until trees are to tall to allow inter-cropping.

Teniente gobernador Village authority, representing the government on thevillag elevel ,electe db yth evillager sfo ra lim ­ ited period.

Terra firme Interfluvial uplands of the Amazon lanscape, mostly of Tertiary or Pleistocene origin.

Tropodult Ultisol ina continuall y warm humidenvironment ; great group USDA classification.

Ucayali Together with the Maranon river one of the two main tributaries of theAmazo n river.

Ultisol Stronglyweathere d lowbas estatus ,lo wCE Csoil s order USDA classification.

xv Varzea Floodplain of the major rivers of the Amazon ba­ sin.

Yagua Amerindian group inhabiting the border area be­ tween Peru, Colombia and Brazil.

XVI PREFACE

This book is the product of an Odyssey which began in 1982 when, as a doctoral student at the department of forestry, I left for Peru to do six monthso f field work, then a curricular requirement at the Agricultural University at Wageningen.Th epla n wast o work ina projec t atth e UniversidadNacional de la Amazonia Peruana (UNAP) in Iquitos. It wasn't clear what I was going to do there,bu t I figured that since Iwoul d at least learnth e Spanish language,th etri p would be worth while. On my return my father counted the timetha t Iha dbee n away from home: five years, eight months and eleven days. I hadfinished m y studies at Wageningen by correspondence, and had worked at three local institu­ tions in Iquitos, doing research infive differen t locations in remote parts of the Peruvian Amazon. Mydiplom a had been waiting for mea t theAgricultura l Uni­ versity since 1984. In my thirdjo b I was hired by Christine Padoch, from the NewYor kBo ­ tanical Garden, who had started a research project onnativ e fruits inth eAmazo n incollaboratio n withth eInstituto deInvestigaciones dela Amazonia Peruana (IIAP) in Iquitos. IIAP has a research station at the Ucayali river, the Centrode InvestigacionesJenaro Herrera (CIJH),an dw ewer e invitedt ojoi n inth eresearc h on ribereho agroforestry in nearby villages. Our research results would provide the information based onwhic h agroforestry trials weret ob ese t upa t CIJH. We started the research in 1985,an d soon decided to expand its scoop to more gen­ eral ribereho resource utilization. The research on ribererno resource management contributes to two topics whichar eo fscientifi c interest: resourcemanagemen t inth evarzea, the floodplains ofth elarge rAmazonia n rivers,an d resourcemanagemen t ofnon-triba l indigenous Amazonian people, a group which until recently had received little attention by anthropologists. Thestud yappear s inth e midst ofa pos t UNCED environmental euphoria, atim e in which many ambitious plans for the sustainable development

xvii ofAmazoni a are being proposed. It is hoped that this work will provide useful baseline data for any of such endeavors. Astud y likethi s only canb e written with the help of manypeopl e and in­ stitutions. I wish to thank José Lopez Parodi, the director of the Centro de InvestigacionesJenaro Herrera, whenI di dm yfield work ,an d whoallowe d ust o do research in his program. I also thank Ruperto del Aguila and MarcioTorre s from the same center who assisted inth e initialfield surveys , as well as mysev ­ eral assistants in Santa Rosa andYanallpa , in particular Pedro Revilla. Funding for the field work came from Swiss Aid to IIAP, and a grant from the Exxon Corporation to the New York Botanical Garden. I also wish to thank Michael Balick, directoro fth eNe wYor kBotanica l Garden's Institute ofEconomi cBotany , who somehow magically found resources to allow me to come to NewYor kan d writethi swork . Anadditiona lwritin ggran t wasprovide db yth eHomelan dFoun ­ dation. Printing ofthi s book was madepossibl e with a grant from thePr oNatur a Foundation. More than to anybody else I owe thanks to Christine Padoch. She helped to think out the research, conceptualize the monograph, and took upon her the tedious task of readingth efirst draft s ofth e various chapters. Sheals o tookcar e of mewhen , sometimes, Ifel t lost inth e vastjungl e ofNe wYork . The monitor­ ing of the writing was continued by Dr. Oldeman, who kept encouraging me to discipline myself scientifically during my time in Peru and while in NewYork , and who accepted mea s a doctoral candidate. There are many people whohav e read, corrected, and edited thevariou s chapters:Will a Capraro, Michael Chibnik, RobertEwing ,Elys aHammond ,Andre wHenderson ,Pau l Matthews,Judit hMayer , Maria Potess, FreerkWiersu m and Gail Wadsworth who gaveth e descriptions for the soils terms in the glossary. I thank you all very much for your patience and help. Lastbu t not least, Iwan tt othan kth epeopl efro m SantaRosa ,an dYanallpa , my friends, who shared with methei r knowledge, their wisdom, their happiness, and sorrows. Theyofte n wonderedabou tth epurpos e ofm ywork ,bu tneve rques ­ tioned it,an d wereneve r impatient with mebotherin g them. Imis s theirjoy ,an d friendliness.

XVlll CHAPTER ONE

VARZEA RESOURCE USE: ECOLOGY AND HISTORY

INTRODUCTION

Inmos t SouthAmerica ncountrie swhic h havea par t ofthei rterritor y inth e Amazon region, theaspiratio n persists of developing agriculture inth e largearea s of this un-exploited domain. An increasing number of scientists and developers consider the varzea, or floodplains of the larger rivers of the Amazon basin, as havinggrea t potential forsuc hpurpose s(Parke r 1981;Sanche z 1988;Nascimient o &Homm a 1984;Deneva n 1984;Bunke r 1984;Smit h 1982). The predominantly fertile soils ofth evarze aar ethough t tob epromisin g foragricultur eusin gmoder n methods, applying modern technology and fertilizers and pesticides. Thisexpec ­ tation, however, is based on the belief that very few people actually live in the floodplain, that mucho fit slan d isunder-utilized , andtha t indigenous agricultural techniquesar eincipien t (e.g.Wagle y 1953; Norgaard 1981;Petric k 1978; Denevan 1984). Considering the myth of the potential for intensive agriculture which this ecological zone is supposed to possess, it is quite surprising how little isknow n about theAmazonia n floodplain areas. Varzea ecology, for instance, is still very poorly understood. Major ecological studies ofth e larger SouthAmerica n flood- plainsdi dno tappea runti lth elas tdecad e(Siol i 1984a;Withmor ean dPranc e1987) , and basic ecological information on vegetation succession and rivergeomorphol - ogy is only now being produced (Lamotte 1990; Salo et al. 1986; Kalliola etal . 1987, 1988). The lack of information on indigenous land use practices in the various parts ofth eAmazo n floodplains iseve n moresurprising . Onlya fewdis ­ persed groups of tribal Amerindians still livethere . The larger part of the indig­ enouspopulation s living ino rclos et oth evarzea , however,ar ecaboclos inBrazi l and riberenos in Peru. Chibnik (1991) qualifies these two groups as non-Indian,

1 non-settlers. Caboclos as well as riberenos have mixed cultural origins of Amerindians,Europeans ,an dimmigrant sfrom othe rpart so fSout hAmeric a(Parke r 1985; Padoch 1986). Ethnographers working in theAmazo n have been more in­ terested in tribal Indian groups located in remote terra firme areas, or the vast interfluvial upland areas of the basin, and they have ignored the large non-tribal river populations. Consequently moststudie so nindigenou sresourc emanagemen t practices inAmazoni a havebee n pursued inth e same habitat. In fact, sincesub ­ stantial archeologica l research has been done inth e varzea, moreha sbee n written onpre-contac tfloodplai n resourcemanagemen tpractice stha nabou t contemporary use. This book presents a case study on resource management of riberenos in thePeruvia nAmazon ,a larg epar t ofwhic htake splac ewithi nth e floodplain. The complex agricultural and agroforestry practices ofth e inhabitants oftw ovillages , namely Santa Rosa and Yanallpa, located onth e lower Ucayali river will bedis ­ cussed. This study will provide new information on varzea resource utilization, but its findings also have implications for several of thetheorie s proposed about varzea resourcemanagemen t ingeneral . For instance, the fact that most ribereno agriculture islocate di nth efloodplai n contradictsothe rresearcher s(e.g .Ros s 1978) who think that Amazonian peasants only use the varzea for extractive purposes. The fact thatriberenos operat e as independent farmers challenges theories which saytha t acomple x social organization is required to makeagricultur e possible in the Amazonian floodplains (Meggers 1971). The agricultural diversity among ribereno farmers contradicts Roosevelt (1980) who states that pre-contact varzea groups could not produce sufficient manioc inth e floodplain, or obtain sufficient protein fromfishing t o adequately feed themselves, and therefore must have re­ lied largely on seed crop production. Inthi sintroductor ychapte rth emos timportan t facts andidea stha t havebee n written on varzea resource management will be summarized. It will briefly an­ ticipate the wayi nwhic h this study willcontribut e toth eunderstandin g ofvarze a resource management, and howsom e of itsfindings questio n several ofth e ideas which have been formulated regarding past and present varzea resource use. VARZEA ECOLOGY AND LAND USE

The division ofth eAmazo n lowland region, the area enclosed by the Guiana Shield, Central-Brazilian Shield, and the Andes (Putzer 1984), into two ecologi­ cal zones, terra firme and varzea, is largely based on their different geomorpho- logical history. Terra firme is the land of the Amazon basin above flood level (Klammer 1984), and this includes most of the basin which is of Tertiary and Pleistocene origin (Irion 1984; Putzer 1984). The term varzea is used only for the seasonally inundated floodplain of the Amazonian white water rivers. Varzeas include the entire area between the rimming valley walls, or interfluvial lands, in which the river flows across recent alluvium, and in which it changes its channels (Prance 1979;Deneva n 1984). Varzea soils are alluvial deposits of much later date than soils of terra firme (Roosevelt 1980). The Amazon river is divided into lower, middle, and upper Amazon (e.g. Parker 1981; Sioli 1984b). The varzea of the lower Amazon, or the varzea de mar as Parker (1981) calls it, is subject to periodic flooding caused by tidal influ­ ences of the Atlantic ocean. Subsequently, ecological conditions are different between the floodplains of the lower Amazon, and the varzeas of its middle and upper reaches which are subject to periodic flooding caused by the yearly oscilla­ tion of the river level (Parker 1981; Sternberg 1975; Petrick 1978; Sioli 1984b). The following discussion deals only with the latter varzea areas. Sternberg (1975:10) calls terra firme and varzea "two basically different groups of landforms", indicating not only the differences between the two zones, but also the ecological pluriformity of each. The two principal processes which account for the ecological complexity in the varzea are erosion and sediment depo­ sition. Erosion of the river banks, which mostly consist of previously deposited sediment (Sioli 1984 b), is mostly in the form of bank caving. Deposition may occur within the channel by bank-side shoaling resulting in muddy or sandy beaches,o r asover-ban k deposition which results inth e formation of natural levees. Sometimes building up of levees and eroding of the river bank may occur on the same sites,bu t during different periods ofth e year (Sternberg 1975). In the Ucayali concave banks arebein g eroded (Lamotte 1990). The complex geomorphological dynamics of the Amazonian white water rivers create a complex landscape in which Denevan (1984: 314) distinguished a total of 9 different landforms. The most important landforms, also found in the Ucayalienvironment ,ar enatura l levees,poin tbars ,an dsan do rmu dbeaches . Since theuppe rAmazon , Ucayali,an d Maranon rivers aremeandering , and not braided like the middle and lowerAmazo n (Goulding 1980; Sioli 1984b ) the shifting of therive rchanne l here isfaste r than inth e lowerparts . Subsequently, varzealand s inthes e regionsar e subject to muchfaste r changes (Parker 1981; Goulding 1980; Denevan 1984). In the Ucayali deposits of more than one meter resulting from one flooding season could be observed, while in some places the river channel shifted laterally up to 50m ina singleyear . Thefloodplai n geomorphology results ina varie dpatter n ofsoi lqualit yan d fertility, and in a varied periodicity and length of inundation. Consequently it contains various landforms and therefore many diverse options for its land use (Hiraoka 1985a,b; Bergman 1980; Denevan 1984). Denevan (1984) proposes a typical sequence of floodplain crop production which follows the topographical varzea gradient from the river side to natural levee to permanently flooded backswamps. Onth e mud or sand barsrice an d beans are grown. On the levee foreslopes maize, sugarcane, and jute are the most appropriate crops. Bananas, manioc, and orchard gardens are found on the tops of the levees. On the back slopesjut ean dsuga rcan eca nb egrown ,whil ebeans ,alon gwit hpasture ,ar efoun d inth e backswamps.

PRE-CONTACT VARZEA RESOURCE USE

Anevaluatio no fcontemporar y indigenous varzearesourc eus ei softe nbase d on its comparison with pre-contact (i.e. before the arrival of Europeans in the Amazon) resource use. The indigenous population which lived near the varzea when the Europeans first entered the Amazon region experienced more dramatic culturalchange stha nmos tterr afirme group s(Deneva n 1976;Megger s 1971). The introduction ofalie n diseases devastated most varzeasocietie s and disrupted their way oflif e (Denevan 1976). Slaveraidin g byth ePortugues e sinceth eearl ysev ­ enteenth century and the establishment ofcolonie s bymissionarie s also hadgrea t influence on theriverine people . After the expulsion of the Jesuits from South America in 1767, ane w mercantile interest inAmazoni a emerged and this inten­ sified theexploitatio n ofth e indigenouspopulatio n (Ross 1978;Sa nRoma n 1975; Parker 1981;d'An s 1982). Finallyth erubbe rboom ,whic hlaste dfro m about 1870 until 1920,sa wlarg egroup s oftriba lan dnon-triba l localpeopl ebein gmercilessl y exploited. Usingreport sfrom earl ytraveler s (e.g.Carvajal ,Acufia , andFritz) ,Megger s (1971)provide sa brie f description ofth eOmagua ,a grou plivin gi nlarg ean ddens e populations in theAmazo n floodplain between the mouth of the Japuri riveran d midwaybetwee nth eCoar ian dPuru s rivers. From thereport s itappear s thatthe y were organized into chiefdoms with well developed hierarchical authority struc­ tures. Thefloodplai n habitatprovide dthes epeopl ewit h largequantitie s ofgame , fish, and agricultural products, which were obtained with very little labor expen­ diture. Bitter maniocan d corn were growna sth eprincipa l staples,togethe rwit h numerous othercrops . Roosevelt (1989) further investigated pre-contact varzea resource use pat­ terns and distinguished four different evolutionary stages in prehistoric resource management in Amazonia. The first people who entered the Amazon lived as hunter-gatherers (1989: 40),followe d bysemi-sedentar yoccupation so fpeopl ewh o mayhav epractice dprimitiv ehorticultur e(1989 :43) . Thethir d evolutionarystag e in varzea resource use patterns istha t of early horticultural villagers. Thisstage , accordingt oRoosevelt , resemblesth epresent-da y indigenous varzeapatterns ,an d iti scharacterize d byrelianc eo nfish , game,an dstarch yroo tcrop s(1989:45) . The increase inpopulatio n which began several centuries before Christ, ledt o thelas t phase, that ofth e agricultural chiefdoms. Roosevelt bases herdistinctio n ofth efou r stageso narcheologica l andeco ­ logical evidence. Ina n earlierpublicatio n (1980) sheargue d that the largepopu ­ lations who inhabited the floodplains during contact subsisted on seed croppro - auction of corn and pulses, and not on a diet offish wit h manioc. This subsis­ tencepatter n lasted untilth enativ esocietie s weredestroye d asa resul t ofcontac t with Europeans. During high water periods, according to Roosevelt, fish catch levels are too low to supply sufficient protein for varzea populations living in densities as high as the pre-contact groups. Roosevelt (1980: 125) furthermore arguedtha t manioc isa ninadequat e staplefo rth efloodplain . Maniocvarietie sd o not yield significant production in less than six months. Since, according to Roosevelt, most varzea soils are flooded for half of the year and remain water­ logged for sometim e longer, manioc can only beproduce d during a few months in the floodplain. On the contrary, corn can easily begrow n inth e varzea, as its growth cycle iscomplete d within four months,an dth eproduc e canb estore d for the rest ofth eyear . Corn contains proteins which, when supplemented withpro ­ teins from pulses, provide all basic amino acids. Thus, according to Roosevelt, the chiefdoms could only develop as a result of shifting to subsistence on seed cropproduction . Thechiefdom s didno trel yo nfish an d manioca sthei rprincipa l calorie and protein resources, as was thought previously by Meggers (1971) or Denevan(1976) . Meggers, as well as Roosevelt, argue that the pre-contact groups subsisted in the varzea onlybecaus e they were able to organize their resource management calendaradequately . The high flood season required that sufficient food washar ­ vested and preserved during low watert o bridgeth e flood-period scarcity of sup­ plies, especially in protein. The unpredictability ofth eflooding regim e required atigh t schedulingo factivities . Boththes e factors demanded that laborwa sorga ­ nized strictly sotha t laborgroup s could beallocate d to different types ofwor ka t themos tappropriat emoment . Thehierarchica l socialorganizatio no fth epre-con ­ tact chiefdoms made this possible (Meggers 1971).

CONTEMPORARY VARZEA RESOURCE USE

Research on contemporary varzea resource use started with a few notable studies. Oneo fth e few Indian groups that haspersiste d inth efloodplai n andha s maintained its ethnic identity are the Shipibos in Peru. A detailed description of their resource management practices has been given by Bergman (1974, 1980), whoprovide sman ydat ao nth eus eo fecologicall y different sites,agricultura ltech ­ niques,labo rinput ,an deconomi c returns. Thefertil e varzeasoil san dric haquati c resources allow Shipibo farmers to "provide an excellent diet" while only work­ ing between one and three hours per day (Bergman 1980: 288). The principal staple,however , isplantain , which producesa highe rcalori cretur npe r laborinpu t than manioc. Wagley (1953)briefl y describes caboclo subsistence in a study ofa Brazil ­ ian town on the Amazon river. The caboclos, according to Wagley, grow some corn, ricean d beans inth e floodplain, butthei r main agricultural activity isman ­ iocproductio n interr afirme fields. However ,agricultur e doesno tyiel d sufficient cash income, and most caboclos are therefore involved in extractive activities as well. The most important of several collected forest products is rubber (Hevea brasiliensis). Otherauthor s givesimila r descriptions ofth ecaboclo' s main focus onterr afirme manio c production together with extractive activities in the varzea (Parker et al. 1983;Frechion e et al. 1989;Norgaar d 1981;Petric k 1978;Barro w 1985). The floodplain, according to the same authors, is not significantly used for agriculture. Ross(1978 )trie st oexplai nth edifferenc e betweenth ewel ldevelope dvarze a resourceexploitatio n ofth epre-contac tgroups ,an dth evirtua l absence ofagricul ­ tureamon gcontemporar y floodplain inhabitants. Heargue stha t inth eproces so f acculturationo fth eindigenou s varzeapopulation , mucho fth eknowledg ean dskil l requiredt obenefi t from thevarze aenvironmen t hasbee nlost . Tribalorganizatio n could not be maintained, and varzea dwellers could not practice any significant agriculturesinc ethe yha d littletim elef t from theirextractiv eactivities . Rossals o stipulates that thecaboclos ' solitary existence is the most important factor which preventsthe m from practicing varzeaagriculture , and their lack ofa socialstruc ­ ture excludes the organized labor, required to cope with the constraints involved withvarze aagriculture ,especiall y itshig h risks. Fluctuation andunpredictabilit y of flooding add arisk facto r to varzea agriculture, which is absent onterr a firme. ThusRos sreasons ,althoug hsoi lfertilit y ishighe ri nth efloodplain , caboclosprefe r tofar mterr afirme t oavoi dthi srisk . Althoughi nsom elocalitie s inBrazi lcaboclo s havereturne d toth e floodplain to start growingjute , orrice (Wesch e 1985;Bahr i 1988; Guillaumet et al. 1990),man y still hold that floodplain exploitation is vir­ tually absent.

RIBERENO HISTORY

Itha sbee nemphasize d repeatedlytha tth enon-triba l indigenous population of the Amazonian floodplains received insufficient attention from the scientific community (Parker 1985;Padoc h 1987). This explains partly whyman yaspect s of varzea resource use are still poorly understood. Only recently, major publica­ tions oncaboclo s haveappeare d (Parker 1981, 1985). Extended research projects on caboclo resource management practices have been carried out in theAmazo n estuary region (Anderson et al. 1985, Parker 1981), but only minor studies have been conducted in the mid Amazon region (Frechione et al. 1989; Parker et al. 1983). Only a few recent studies focus on caboclo varzea agriculture and agroforestry close to Manaus(Bahr i etal . 1991;Guillaume t et al. 1990). Studies on the Peruvian Amazon have focussed on specific aspects of its history (d'Ans 1982; Golob 1982), although only a few include a discussion of the twentieth century (San Roman 1975; Meyers 1988). Ribereno resource use practices are briefly discussed by Hiraoka (1985a,b), Padoch et al. (1985), and Padoch & de Jong (1987, 1989).Chibni kan d deJon g (1989)discus sth eribereno labo rorgani ­ zation. Chibnik(1990 )evaluate d riskan duncertaint yi nribereno rice production , and heals o madea nattemp t togiv ea mor eprecis echaracterizatio n ofth esevera l "quasi-ethnicgroups" ,o rlocall ybor nnon-Indian so fAmazonia ,includin gcaboclo s andriberefios (Chibni k 1991). Padoch (1986: 4) defines riberefios as "a rural mixed population of detribalized Amazonian natives and their descendants, former immigrantsfrom neighboring Peruvian Departments ofSa nMarti nan dAmazona san dthei rdescen ­ dants, the children and grandchildren of immigrants from other South American

8 countries as well as from overseas, and the descendants of any unions between members ofth eabov egroups . Veryfe war erecen t immigrants,an dvirtuall ynon e are from non-tropical environments". Riberenos are the largest group of rural inhabitants of the lowland Peruvian Amazon region, who live mostly along the majorrivers (Arambur u 1984). Thesmal lvillage scalle dcaserios ar eisolate dan d usually canonl y be reached byboat . Most riberenos makea livinga s agricultur­ ists, but many also extract forest products or work as wage laborers ortrader s to obtain cash income (Padoch 1987). Ribereno agriculture is, for a large part, lo­ cated inth e varzeaan d combinesproductio n for household consumption with in­ tensive cash cropping. Thehistor y ofth eribereno peopl ean dcultur ei ssimila rt oth eon edescribe d for the Brazilian caboclos. Both histories are largely determined by economic events in the region (San Roman 1975;Padoc h 1987;Padoc h & de Jong 1989). During the period before the rubber boom, until about 1870,th e impact of slave raiding, forced migrations, and debt peonage on the indigenousriver populatio n in Peru was less than in Brazilian Amazonia. However, during the rubberboo m the Peruvian natives also experienced influences which, in many cases, totally disrupted their social organization, economic activities,an d evencultura l identity. Accordingt o Chibnik (1991),th ecrucia lperio d ofth eformatio n ofribereno iden ­ tity was between 1910 and 1950, about 100 years after the emergence of the caboclos as a cultural group. In general, more traces ofAmerindia n culture still persist among these Peruvian riverine people than among thecaboclos .

RIBERENO RESOURCE USE

During the rubber boom, much agricultural production in the Peruvian Amazon was abandoned, since large contingents of farmers and farm laborers became involved inextractiv eactivities . After therubbe rboom , however,agricul ­ ture was reestablished but in the form of larger estates orfundos to which most nativeIndia nan driberen opopulation sbecam eattache da slaborer s (Higbee1945 ; San Roman 1975;Padoc h &d eJon g 1990). Besides agricultural production, the fiindo owners engaged with their labor forces in extractive activities every time that certain forest products were in high demand in the international markets (Padoch 1987;Padoc h & deJon g 1990). Beginning inth e early 1960s morean d more fundos collapsed, and their owners left. Previous laborers on these estates stayed behind and became independent farmers. Hiraoka(1985a,b , 1986)show sho wriberefi o farmers usesite s inth e flood- plainwhic har eecologicall y verydifferen t from oneanother . Thesesite sca nonl y beplante d withcertai n crops and requirespecifi c farming techniques. Hereport s ona total of 14biotopes which the inhabitants of a village close to Iquitos each use for specific purposes. Abiotop e is defined by Denevan (1984: 311)a s ami ­ cro-environment with relatively uniform landform, climate, soil, and biota. The biotopes which are mostly used for agriculture are located on natural levees, on river islands, and on sand and mud bars. Atota l of six biotopes areuse d forag ­ ricultural production. These include the top, foreslopes and backslopes of the natural levees,mu dan d sand bars,an d riverislands . According to Hiraoka(198 5 b: 252), inth e yearly flooded biotopes seasonal agriculture ispracticed , whileo n the levee tops a short fallow agriculture with semi-perennials and perennials is carried out. Different biotopesca n beplante d to different crops, require different agricultural techniques, and produce different yields.

RIBERENO AGROFORESTRY

It is now generally acknowledged that practices which can be qualified as agroforestry are common among many Amazonian tribal and non-tribal farmers (Posey 1983; 1984, 1985;Deneva n et al. 1984;Deneva n &Padoc h 1987; Ander­ son 1990; Anderson & Jardim 1989; Anderson et al. 1985; Padoch et al. 1985; Padoch& deJon g 1987; 1989;Hiraok a 1986;Irvin e 1987; 1989;Bahr ie tal .1991 ; Guillaumet et al. 1990). However, agroforestry practices among indigenous Amazonians havedissimila r characteristics andprovid e a range ofdifferen t prod­ ucts and services. The variation ofagroforestr y practices in, for instance,a nare a liketh ePeruvia nAmazo nregio naroun dIquito si sconsiderabl e(Deneva n& Padoch

10 1987; Padoch etal . 1985;Hiraok a 1986;Padoc h & deJon g 1987). Both house­ hold and market oriented agroforestry practices have been reported for ribereno communities (Padoch etal . 1985;Hiraok a 1986;Padoc h &d eJon g 1987; 1989). Not only do these systems differ in destination of the output, but also in species composition,complexity ,an dmanagemen tpatterns . Agroforestry practices ofloca l farmers whoar emor e involved with regional economiesar e suggested topresen t moreusefu l modelsfo r resourceus eplanner stha nthos efoun d amongremot etriba l communities (Padoch & de Jong 1987: 193).

RESEARCH QUESTIONS AND RATIONAL

This book will explore answers to the following questions : (1) How diverse are theagricultura l practices among riberenos? (2) What areth e specific characteristics of different agricultural practices? (3) Howvariabl ear eagricultura l strategiesamon gfarmer s within singlevillages , and between villages? (4) How canth e variation in agricultural strategies be explained? (5) How common and diverse are agroforestry practices among riberenos? (6) How do agroforestry practices differ between villages? Although studies by Hiraoka (1985a,b; 1986; 1989a) have shown someo f thecomplexit y ofribereno agriculture , much quantitative datao nspecifi c agricul­ tural methods found amongriberenos i s lacking. As aconsequence , conclusions like interpretation of the relationship between biotope diversity and land use, or the distinction between agriculture in seasonally flooded and non-flooded varzea landsstil l needst ob e investigated further. After adiscussio n ofth emethodolog y in chapter two, and an introduction to the research area, villages, and people in chapterthree ,i nchapte rfou r adetaile d discussion ofth edivers eagricultura lprac ­ tices ofriberenos wil l bepresented . This chapter will demonstrate that, although short fallow swiddening does occuro nhighe r levees (Hiraoka 1985b) ,thi s isno t the only type of land use found there. Furthermore, on sites ofth e lowervarze a the soil quality, elevation, and subsequent yearpl y flooding varies. As a result, ribereno farmers ceaselessly modify the kind of agriculture, adapting it to differ-

11 ent sites. Finally it will alsob e showntha t landus e onth esam e site varies from yeart o year as a response to changes in ecological and economic conditions. In chapter five variation and changes in agricultural strategies of ribereno farmers will be investigated. Agricultural strategies can be defined as choices concerning agricultural activities, which farmers have to make in regards to the allocation of limited resources (land, labor, capital) in ordert o meetthei r subsis­ tence needs. In general, studies onagricultura l strategies explain agricultural be­ havioro fcertai ngroup s offarmer s (Barlett 1980a). Iti sals otru etha tagricultura l practices of many traditional societies change in response to internal or external influences (Padoch 1982; Roseberry 1989;Cancia n 1989). Both variation inag ­ ricultural strategies and changes in these strategies demonstrate adaptations to certain ecological or socioeconomic environments. As will be shown in chapterfive, th e complexity of ribereno agriculture is partly aresul t of thevariatio n inagricultura l strategies. In addition to longter m changes of ribereno life and livelihood due to increased contact with the outside world (Hiraoka 1985b :265) ,riberen o agriculture ischaracterize d bya flexibility which is reflected by changes over short periods. Both variation in agricultural strategies andflexibility in resource management arerelate d toth e dynamicenvi ­ ronment inwhic hriberenos live . Not onlyar eth e land forms ecologicallydiverse , but they may also appear and disappear overnight. Market opportunities may changecompletel y withina ver yshor tperiod . Toadap tt othos echanges , ribereno farmers modify their agricultural strategies profoundly The data presented in chapterfive wil l demonstrate that the variation inribereno agricultura l strategies, as well as the profound short term changes inthes e strategies, are adaptations to the highly diverse and highly dynamic ecological and economic environment of the floodplain. Inchapte rsi x siwdden-fallow agroforestry inth evillag eo f SantaRosa ,an d inchapte rseve nvarze aagroforestr y inYanallp awil lb ediscussed . Althoughstudie s onAmazonia n ribereno agroforestry indicatetha ta variability exists within single systems usually reported for separated villages,thi s aspect has not yet beenthor ­ oughly investigated. Padoch & de Jong (1987) previously indicated that among

12 farmers within single villages agroforestry practices may be different. Since agroforestry practices arecommo n inth e PeruvianAmazon , agroforestry options areapparentl y available toriberefio farmers , but not all will opt for thesam esys ­ tems. Understanding thisphenomeno n will increase the knowledge ofth epoten ­ tial for developing agroforestry systems for this and possibly otherregions .

THEORIES ONVARZE A RESOURCE MANAGEMENT QUESTIONED

Theresult so fthi sstud yquestio nsevera lo fth eabov eoutline didea so n varzea resourceuse . Riberefio agriculture isi nremarkabl econtradictio n toth esuppose dabsenc e of cabocloagricultur e in the varzea, as reported by Wagley (1953), Ross(1978) , andFrechion ee tal . (1989). Thiscontradictio n requiresa nexplanation . Themai n constraints to varzea agriculture are risk of field flooding, and the limited time available for cropping varzea lands. The several agricultural methods which riberenos apply differ in labor requirements and scheduling, destination of pro­ duce,yiel d level,an d risk. Bydevelopin g numerousagricultura l methods, andb y combiningthos ei nsingl efar m enterprises,ribereno ssuccee di nlowerin gth eoveral l risk and makebette r use of their available labor. This indicates that agricultural diversity, variation in agricultural strategies, and agricultural flexibility areadap ­ tationst oth econstraint s ofth evarze aenvironment . Althoughman yriberefio farm ­ ers combine varzea and terra firme agricultural production, it is possible to rely only on exploitation of the varzea environment. The exampleo fth eribereno s invalidates theargumen t ofRos s(1978) ,tha t caboclos do not farm the varzea because they areunabl e to coordinate theirsub ­ sistenceactivities . Riberefio varzea utilization furthermore contradicts the need of persons of higherstatu st o coordinate laboramon gpre-contac t groups inorde rt o providea n adequateagricultura l calendar (Meggers 1971; Roosevelt 1980). Even inarea so f highpopulatio n densities,ribereno soperat eindependently . Organizedlabo rgroup s andfestiv e laborpartie sar ecommo namon griberefios (Chibni k& d eJon g 1989),

13 but many fanners do not participate in them, and they are not indispensable in varzea fanning. Hiring laborers is generally avoided byriberefio farmers , and it onlyoccur s forcertai n cashcro pactivitie s (Chibnikpers .com . 1991). Operating asa nindependen t farmer appearst ob ea mor eappropriat e wayo fcopin gwit hth e tremendous diversity and dynamics ofth evarze aenvironment , rathertha noperat ­ ing in larger hierarchical groups. The agricultural diversity of riberenos also contradicts Roosevelt's (1980) hypothesis concerning the seed crop based subsistence strategies of pre-contact groups. In chapter four it will be shown that the production of corn and manioc areno ta tal l incompatible inth e floodplain. However,th erat e ofretur n forpro ­ tein per unit of labor for a corn crop is not nearly as high as protein yields ob­ tainedb yfishing , evendurin ghig h flood (Bergman 1974). Therefore, theexampl e ofsustaine d floodplain useb yriberenos invalidate sth etheor ytha t largerpre-con ­ tact populations could develop only because they changed to a seed based diet.

14 CHAPTERTW O

METHODOLOGY AND METHODS

INTRODUCTION

Indigenousagricultur eha sbee nstudie db yscientist sfro m manybackgrounds . Researchers have emphasized different aspects according to their particular disci­ plinary orientations (Spencer 1966). While more orthodox agricultural scientists havefocusse d onmechanica lan dbiologica l aspects ofagriculture , otherscientist s like agricultural economists, economic anthropologists, geographers, and rural sociologists have been more interested in socioeconomic aspects of small scale farming. Howdivers eth especifi c aspectso fagricultura lpractice sca nb ei sshow n by Conklin (1963) whoprovide s a long list of different research themes that can be selected when studying shifting cultivation. As aresult , little methodological uniformity exists in research ontraditiona l agricultural systems. A scientific approach to indigenous agriculture which has received more interest recentlyi sth esystem stheor ymodel . Fresco(1986 )use dthi smode lwhe n studying shifting cultivation in Zaire. Fresco distinguished different system lev­ els, defined as suprasystem and subsystems, the most important of which arere ­ gional systems, farming systems,croppin gsystems ,an d crop systems (1986: 47). To conduct her analysis, Fresco defined a number of observation units for each system levelan dspecifie d ase to fvariable sfo r eachunit . Thisresulte d ina sepa ­ ratese t ofdat afo r eachsyste m level. Datawa sderive d from variousprimar yan d secondary sources. Adifferen t approach hasbee n followed in studies whichar ewithi n theru ­ ral sociology or economic anthropology disciplines. In earlier anthropological studies,attempt shav ebee nmad et o demonstrateth euniquenes s ofcertai npeopl e asa socia lgrou p(Cancia n 1988),o rt o investigate whethero rno t economic mod­ els applied in western societies could be used in non-western societies (Plattner,

15 1988; Roseberry 1988; Johnson 1980; Chibnik 1990; Barlett 1980b). However, sinceth e 1970sstudent s ofpeasan t agriculture havebecom econcerne d withagri ­ cultural development issues (Barlett 1980b: 1)an d have searched for answers to research questions similar to those formulated in the present study. Many eco­ nomic anthropological studies investigated how variable natural, social, political, and economic conditions influence the agricultural strategies of farmers (Barlett 1980a). Which conditions are important and, therefore, need to be investigated, depends not only on the agricultural systems under study, but also on the ques­ tions which are asked. In Barlett's words (1980a: 549-550) "The wide range of variables that affect production strategies derives not only from the complexityo f choices but also from the diverse research problems that havebee n addressed." The consequence of such an axiom is that even if one accepts the systems theory approach, it is still the specific research questions that determine what in­ formation is needed and how that information is generated. This assumes that certain practices orprocesse s have their own rationale or causality, which can be understood by relating different phenomena interm s ofcaus e and effect. This is thesam eapproac htha t system theoryuses . Asyste mtheor yapproac h dividesth e realm ofa certain peoplea s far as it relates tothei ragricultur e into different sys­ tem levels. However,th ecausalit y ofth ephenomen adescribe d remainsth esame . This, in fact, means that using systems theory is not a conditio sinequa non t o find answers to the questions formulated in chapter one. In retrospect, the following research strategy was followed for this study. First it was necessary to decide which data needed to be collected. Second, re­ search units were selected, and the variables that had to be quantified were de­ fined. Finally, methods werechose n forth egatherin g ofdata . Thepresen tchap ­ terdescribe sthi sprocess ,an dth emethod stha t wereuse dt oconduc t theresearch .

DEFINITION OF CONCEPTS

Anumbe r of concepts are important inthi s book. In the previous chapter agricultural strategies was defined as choices whichfarmers haveto make asto

16 theallocation of limited resources (land,-labor, and capital) tomeet their subsis­ tenceneeds, and as far as these concern agricultural activities. Asecon d con­ cept,discusse di nchapte rfour , isagricultura ldiversity . Agriculturaldiversity ca n bedefine da sthe combination of different agricultural methods used by a specific, butdelimited group offarmers. Athir d concept, called the agricultural type, is also proposed in chapter four. An agricultural typei s defined as a unique site- crop (orcrop mixture) combination, which has aspecific set of agricultural tech­ niques, schedulingof activities, production levels,principal destination of output, andrisk. Afourt h importantconcep t used inchapter s four through seveni sman ­ agement. Management as an agricultural activity is a more general term than cultivation. Managementca nb edefine d asactively and purposely influencing the developmentof vegetations, bethey planted or spontaneously occurringindividu­ als of aspecies, inorder toderive a certain benefitfrom thisspecific vegetation. Management therefore coversth e intellectual agricultural decisiontaking ,a swel l asth ecarryin g out ofa number ofcultivatio n activities such asharvesting , clear­ ing, burning, planting, and weeding or clearing the direct surroundings of indi­ vidualplants .

SCHEDULING OFTH ERESEARC H

Asth efirst methodologica l decision, it waschose nt o conduct a long-term study amongfarmer s intw oribereno villages . Sinceth e study wast o bepar t of theresearc hprogra m ofth e Centro deInvestigación Jenaro Herrera, theresearc h sites hadt ob echose n from thosevillage s located inth eare ai nwhic hthi s center conducts its research (Lopez 1982). This area extends along the Ucayali river approximately betweenth etow n ofRequen aan d itsconfluenc e withth eAmazo n (Figure 3.1). InApri l 1985, the several collaborators of theprojec t conducted a general survey ofth evillage s inth eare a inorde rt o identify study sites. Ineac h villagea few farmers wereinterviewe d aboutth emos t commonagricultura lprac ­ ticesi nthei rcommunity . SantaRos aan dYanallpa , whichwer echose na sresearc h sites,represen t two different types ofriberefio villages . The most important dif-

17 ference is their location. Santa Rosa is located on terrafirme, whil eYanallp a is completely located within the varzea. Each of the two village types will bedis ­ cussed inchapte rthree . Since both Santa Rosaan dYanallp a are located atabou t the same distance from Requena, the regional marketan d administrative center, it was expectedtha t market opportunities would beroughl yth esam e inth etw ovil ­ lages. Field workwa sstarte d in SantaRos ai nJun e 1985,an d inYanallp a inMa y 1986. After September 1986 most field work was carried out in Yanallpa, and only short visits were made to Santa Rosa about once a month. Thefield wor k was concluded inAugus t 1987. In both villages an initial exploratory period provided an opportunity to become familiar with the people, thearea , andth e manydifferen t subsistenceac ­ tivities. Duringtha t period, farmers werejoine d inthei r dailyactivitie s andman y wereinterviewe d whenther ewa s opportunity to do so. This allowed familiariza­ tion with the villagers and their agricultural fields, and resulted in muchunstruc ­ turedinformatio n aboutresourc emanagement . Thespecifi c researchactivitie san d methods were developed largely on the basis ofthi s understanding of thepeopl e and villages acquired during the initial period.

SELECTION OFRESEARC H UNITS

Although in research on indigenous agriculture the family or farm house­ hold is often chosen as the basic unit of study (Barlett 1980 a; Wiersum 1988; Marten & Saltman 1986), single households in Santa Rosa and Yanallpa some­ times consisted ofmor etha n one economic unit. For instance, different families could live together temporarily, or family members might be onthei r wayt obe ­ coming independent farmers. If more than one member claimed to have his or herow nagricultura l fields within asingl ehousehold ,the ywer eeac h distinguished asa separat efarmin g unit. Suchseparat eownershi pbecam eapparen t whenmem ­ bers of a household were questioned about thefields the y possessed. Thus,th e farming unit wasselecte d asth e first researchunit . Inth e rest ofthi s bookwhen ­ everth e term farmer is used, it refers to such an independent farming unit.

18 Theagricultura lfield wa schose n asth e second research unit. Thedistinc ­ tion and classification of agricultural types, as discussed inchapte r four, isbase d on a survey of many single agricultural fields. The assessment of diversity in agroforestry practices in chapters six and seven is also based on the comparison of manysingl e forest gardens or varzeaagroforestr y fields.

DATAGATHERE DAN DRESEARC HMETHOD S

Members of farming units were interviewed inorde r to gather information about agricultural strategiesan d othereconomi c data onthei r farm enterprises. A listo fquestion sha dbee nmad ebefor e startingth einterviews ,an dthes ewer easke d duringeac hinterview . Oneo rmor emember so fth efarmin g unitwer equestioned ; inmos tcase sthi sinclude dth ehea do fth efamily . Ifth efamil y headcoul dno tb e located than another member ofth e farming unit whoappeare d to beknowledge ­ able about agricultural matters was interviewed. Inquiries wereals o madeabou t economic activities other than agriculture (e.g. fishing, trade, transportation of goods) since this was expected to explain at least partly, why farming units had certain agricultural strategies. Whenevera farme r mentioned specialcircumstance s orpractice s duringth e interview, questioning wasdirecte dtowar dtha tparticula rsubject . Theinterviews , therefore, did not merely follow the questionnaire, but rather were standardized in-depth interviews. Thus,althoug h aspecifi c list ofquestion swa saske d ofeac h farmer, often additional information wasals o obtained in informal conversations. For example, when a farmer hadjus t made a new agricultural field, an inquiry was made into how much labor hadbee n used for preparation orplanting . Some interviews lasted up to two hours. Most of the interviews in Santa Rosa were donedurin g Octoberan dNovembe r 1985. Some farmers weredifficul t tolocate , and had to be interviewed later. InYanallp a the interviews were conducted be­ tween September and October 1986. The information needed for the characterization of agricultural types was also gathered through participant observation and interviews,a s wella s note-tak-

19 ing by a few farmers. Data concerning agricultural types were gathered during the entire research period. Many farmers were interviewed several times, often through informal conversations. Farmerswer eaccompanie d tothei rfields an dth e researcherofte n participated inlabo rpartie s orcollaborativ e workdays . Onceth e distinction ofagricultura l types (see chapter four), wasmade , the interviews with farmers became more specific. Before starting the interviews inYanallpa , many agriculturalfields wer e surveyed again to adjust the classification of varzeaagri ­ cultural types. In both villages a number of forest gardens were selected to study the di­ versityi nagroforestr y practices. Fieldsselectio n wasbase do ndifference s inspe ­ cies composition, age, and management pattern. A detailed description of each field wasmade ,includin g itsexac tlocation ,topography ,an dvegetation . Thelan d use history, as told by the current owner or other informants, was recorded. In mostcases ,thes ewen t backa sfa ra sth efirst us eo fth efields. A transec tmetho d waschose nt osampl efields (e.g .Richard s 1952; Whitmore 1984;Hall ee tal . 1978; Michon 1985;Guillaume t et al 1990). The transects used in forest gardens were 30 m wide. This is wider than used in most other similar studies (e.g. Irvine 1987; Denevan & Padoch 1987; Michon 1985). Inpreviou sresearc h only 10 mwid etransect s wereuse d (deJon g 1984), butthi sca nb econsidere da ninsufficien t samplesiz efo rfields wit ha smuc h spatial variation as displayed by the forest gardens in Santa Rosa andYanallpa . Some of the varzeafields wer eto o small to make 30 mwid e transects. Inthes e cases only 20 m wide transects were used. It was attempted to include major ecologicalgradient si nth etransects ,i nmos tcase sth emaximu mslop eo fth e field. Transects were subdivided into 10x10 m2 subplots. In forest gardens a distinction was made between managed and non-man­ aged vegetation. Each adult individual ofa domesticated plant species wascon ­ sidered as managed (e.g. Vasquez 1990; Calzada Benza 1981, Cavalcante. 1976, 1978, 1980). Ifnatura l regeneration ofa domesticated tree species occurred,an d therewa sn oclea r indication whichplant s wereactuall y managed,the nonl yindi ­ viduals highertha n 1 mwer econsidere d to bemanaged . Inadditio n to domesti-

20 cated plant species, ribereno farmers protect many native forest species, most of­ ten in forest gardens (Padoch & de Jong 1987). Individuals of a non-domesti­ cated species were considered to be protected if the vegetation surrounding such plants appeared to havebee n recently cleared, or ifthe y grew ina nunusua l habi­ tat. Anexampl eo fth elatter ,fo rexample ,wa ssee nwhe na specie sfrom th evarze a wasgrow n interr afirme. Thi sindicate d thatth eowne rha d deliberately savedth e plant from weeding, or cleared it of surrounding vegetation. In several cases the owner of the field was asked if specific plants were protected or not. All the managed plant individuals inth etransect swer e mapped, providing data ondensi ­ ties of managed species inth e forest gardens. Spatial patterns ofth e studied for­ est gardens will not be discussed in this book. Animportan t variableuse d for thecompariso n ofth emanagemen t of forest gardens in Santa Rosa isth e weedingpatter n of fields. Toestimat e weeding pat­ terns of terrafirme fields a n Index of Non-managed Vegetation (I,jmv) was used. Every three months the non-managed vegetation in each 10x10m 2 subplot ofth e transects wasestimate db yattributin gt o ita nI nmv valuerangin gfrom 1 to 5. The weeding pattern ofan yterr afirme agricultura lfield i sreflecte d byth e state ofth e non-managed or weed vegetation. The weeding patterns of four of the seven for­ est gardens was compared by estimating the state ofth e non-managed vegetation everythre e months for atota l period of 18months . Inthre e otherfields thi sob ­ servation was notpossible . Twofields wer e slashed soon after the inventory was carried out, and onefield wa s locatedto o closet oth e villaget oallo wth emakin g of a permanent plot.

Inmv values from 1 to 5hav eth e following meanings: Inmv value 1indicate s that the non-managed vegetation is virtually absent, or lesstha n 20 cm high inth e whole 10x10 m2 subplot. The weed vegetation is heel high, thefield ha s been newly made or recently cleared.

Inmv value 2 indicates that weed vegetation ofa height between 20 cman d 50 cm covers more than 50%o f the subplot. The weed vegetation is not more than knee high. Although such weeds are competitive to crops (Tripathi 1977; Crossley et al. 1984), Santa Rosa farmers do not perceive this state of non-man-

21 Table 2.1: lnmv (non managed vegetation index) values used to estimate weeding patterns in Santa Rosa forestgardens .

Inmv Descriptive State of weed vegetation value in1 0x 10m 2 subplot 1 heel high less than 20 cm high 2 knee high weed vegetation covering more than 50% between 20 cm and 50cm . 3 waist high between 0.5 and 1.5 m for more than 50%o f the subplot 4 person high between 1.5 to 2.5 m for more than 50% of the subplot 5 secondary exceeds 2.5 m for more than50 % forest ofth e subplot agedvegetatio n asa significant threat to aterr afirme agricultura lfield wit hman ­ ioc, plantains, ortre e vegetation.

lnmvvalu e3 indicates anon-manage d vegetation between 0.5 and 1.5 mi n heightcoverin gmor etha n 50%o fth esubplot . Thewee dvegetatio n iswais thigh . Thisvegetatio n isperceive db yfarmer s ascompetin gwit hth ecro pvegetatio nan d it impedes access to the field for harvesting purposes. In most cases swiddens which have such weed vegetation are considered to require aweeding .

lamvvalu eo f4 wasgive nt o asubplo t when moretha n 50%o f it wascov ­ eredb y non-managed vegetation between 1.5 to 2.5 m in height. The weedveg ­ etation is person high. It means that the field had not been slash-weeded for a longtime .

Non-managed vegetation which exceeded 2.5 mi nheigh t wasgive n a Inmv value of 5, indicating afull y developed secondary forest. The Inmv values and their descriptions are summarized intabl e 2.1. In the forest gardens inYanallp a the management patterns were evaluated by revisiting the fields severaltime s during the stay in the village and observing

22 thegenera l state of the non-managed vegetation. Yield estimates could only be obtained in the terrafirme fields an d not in varzea fields. In addition to observing the state of the non-managed vegetation every three months, the harvestable yield for each 10x10m 2 subplot for thenex t three-month period was estimated. The estimates were specific for different spe­ cies. For instance, inth e case of fruit trees,th e numbero f fruits which couldb e expected toripen th e next three months were counted, or estimated if their num­ bers weret o large. Forthes e estimates the helpo fth e localfield assistant , Pedro Revillawa scrucial . Theyiel destimate swer erepeate dfive time sove ra 15 months production period.

23 CHAPTERTHRE E THEPEOPL EAN DTH EPLACE S

INTRODUCTION

The wordribereno originate s from the word ribera, river bank, and indi­ cates the rural populations close to the larger rivers in the Peruvian Amazon. Historically theriver bank s of the largeAmazonia n river systems have been the most important areas of population settlement, sincerivers provide d transporta­ tion facilities and access torich fishing ground s (Lathrap 1970; Denevan 1976; Roosevelt 1980). In contemporary Amazonia most of thepopulation' s economic activitiesan dsettlement s arelocate d inth efloodplai n ofth emajo r rivers(Deneva n 1984). Inth e PeruvianAmazo n 87% ofth e rural population lives onth emargin s ofth egrea trivers (Arambur u 1984). Theemergenc eo fth eriberenos a sa cultura lgrou pbega nwit hth einvasion s ofth eAmazo n region by Europeans. Researchers divide the history of thePeru ­ vian Amazon into several periods. (1) Acculturation began with the Jesuits' at­ temptst oChristianiz eth enativ epopulatio n(Sa nRoma n 1975;D'An s 1982; Meyers 1988; Golob 1982). (2)Afte r the Jesuits were expelledfrom al l SouthAmerica n territories in 1767,th eloca lpeopl ebecam eth eprincipa lwor kforc e inth eexploi ­ tationo fdifferen t forest products from thearea . (3)Th eperio d ofth erubbe rboo m had a substantial impact on the indigenous people of the Peruvian Amazon (San Roman 1975;Padoc h 1986), and was of great influence on the formation of the riberenos as they aretoda y (Chibnik 1991). Theinhabitant so fbot ho fth etw oresearc hvillages ,Sant aRos aan dYanallpa , fit verywel l intoPadoch' s generaldefinitio n ofriberenos a s mentioned inchapte r one. However,the yals o havethei r ownuniqu ebackground s and histories. Santa Rosa andYanallp a represent two types ofribereno village s because ofthei r loca­ tion intw odifferen t environments,an dbecaus eo fthei rdifference s insocia lstruc -

24 turean d history. Most ofth ehouse s ofSant aRos aar elocate d closet o eachothe r ina smal lvillag ecente ro na terr afirm ehillock ,bu twit heas yacces st oth evarzea . Its history is very much related to the history of afundo, or largeagricultura l es­ tate, located within a short distance of the village (Padoch & de Jong 1989). Yanallpa is located in the varzea and its houses are stretched out in a row along theriver . Thevillag eha sa muc h longerhistory . Yanallpadevelope d whenpeopl e started to exploit, rather independently, this floodplain site. Important aspects of resource management, including access to land, are characteristically different between the two villages because of their different locations. Althoughriberenos fro m thePeruvia nAmazo n haveexperience d muchcul ­ tural change throughout their history, they have continued to live in the varzea, and exploit similar habitats. Their knowledge and understanding ofthi s environ­ ment is profound. It is this ecological knowledge that allowed riberenos to de­ velopth e manyagricultura l methods that willb edescribe d inchapte r four. Inth e present chapter, the people whose resource management is the subject ofdiscus ­ sion will be introduced, and their social organization and physical environment described.

A BRIEF HISTORY OFRIBERENO S

Only ashor t time after the Europeans entered the SouthAmerica n tropical lowlands, the Jesuits gained control of much of the upper Amazon region and becameth emajo ragent so fearl yEuropea ninfluenc e onth eloca lpopulation . Their control ofth euppe rAmazo n protected the Indians from theunscrupulou s exploi­ tationwhic hbefel l theBrazilia nIndian s(Parke r 1981;Ros s 1978). However,activ e cultural transference by the Jesuits and population decimation as a result of dis­ eases disrupted much of the local social structures, especially of riverinegroups . During this period of the occupation of the region, productive activities included forest extraction and agriculture for local consumption, and for export to Quito (San Roman 1975;Golo b 1982;D'An s 1982). The local population becameac ­ customed to new commodities which could be obtained by participating in are ­ gional market economy only.

25 After the Jesuit expulsion in 1767, many former inhabitants of the settle­ ments, which had been created by the missionaries, became small farmers living in isolated homesteads or in small groups (San Roman 1975). Outsiders gained easier access to the many forest products found in the Amazon. The indigenous populationbecam eth eprimar yextracto ro fthes eproducts . Toa lesse rexten tthe y became agricultural laborers on fundos, the large agricultural estates which were established in the region in the mid 19th Century (Stocks 1981). They livedan d worked under feudal conditions, dedicating part ofthei r time to subsistence agri­ culture and to the extraction of forest products. During therubbe r boom,aroun d thetur n ofth e century, more remoteIndi ­ ansbecam eth e victims ofslave-raider s and were forced to worka s slaves inrub ­ ber extraction. Manyo fth e enslaved natives dieda s aresul t ofth e extremehard ­ ship ofthei rworkin gcondition s (Hardenburg 1912;Weinstei n 1983). Theevent s ofthi s period hada pronounced effect on demography and ethnicity in the Peru­ vianAmazo n(Chibni k 1991). Manyne wimmigrant stravele dt oth eAmazon ,an d remained in the area after rubber extraction stopped (Padoch 1986; San Roman 1975). Theco-existenc e oftriba lan d detribalized Indiansan do f immigrants from other SouthAmerica n countries, Europe and North America resulted in thecom ­ plexbackgroun dtypica l forth eribereno populatio n (Padoch 1986). Accordingt o Chibnik (1991), the process of detribalization and formation of a self-identified non-Indian, non-settlerribereno grou p occurred mostly during and after therub ­ berboom ,betwee n about 1910 and 1950. Mostagricultura l activities inth ePeru ­ vianAmazo n stopped when rubber extraction became more lucrative. Inth eperio dfollowin gth ecollaps eo fth erubbe rmarket ,man yo fth e former participants inth e exploitation ofthi sproduc t became farmers, living onpuestos, or smalleragricultura l estates, mostly onth eriver banks . Manypeopl eturne dt o subsistence agriculture, while commercial farming increased moderately (San Roman 1975). Large fundos changed to a mixture of activities including cash cropping, cattle ranching and extraction. The labor force which had tapped rub­ bero nthes eestate s stayedan dparticipate d inth ene wactivitie s(Sa nRoma n 1975; Higbee 1945). The owners of large fundos had usufruct over extended areas of

26 land (Higbee 1945)an d settlers who lived onthes earea s hadt opa ytribut e toth e owneri ngood s ori n labor(Sa n Roman 1975). From 1930t oabou t 1960severa l other forest products experienced minor booms for short periods (Padoch 1987). Fromabou t 1934unti l 1945 barbasco (Lonchocarpus nicou) ashru b whichyield s natural biocides, wasgrow n in large quantities inth eregion . Thefal l of itspric e caused one more economic crisis in the region and led to the collapse and aban­ donment of most of the larger fundos (Stocks 1981). The last event which had a major impact on theribereno populatio n was the discovery of oil inth e PeruvianAmazo n inth e early 1970s. This resulted in the migration from the rural areas ofthousand s of men who went to work for oil companiesi nremot earea so fth eregion . After severalyears ,th enumbe ro fpeopl e needed byth e oil companies dropped, and manybecam ejobless . However,mos t of those who had come from rural areas had become accustomed to a new life style inth e oil campsan d stayed in Iquitos. Thepopulatio n ofthi scit y increased dramatically during this period (San Roman 1975). The urbanward migration of young peoplestil l continues, resulting ina stable or decreasing number ofinhab ­ itants of manyribereno village s (Hiraoka 1985b).

RIBERENO POLITICAL ORGANIZATION AND SETTLEMENT PAT­ TERNS

In contemporary Peruvian Amazonia, the largest part of the rural popula­ tion areriberenos livin g in isolated villages along the rivers (Aramburu 1984). These villages may have one or two neighboring communities which can only sometimesb ereache do nfoot . Largemarket susuall yca nonl yb ereache db yboa t andar eofte n faraway . Onlyribereno village s locatedclos et olarg etown so rcitie s are connected to them by road. In the Departamento of Loreto the population density was only 1.4 per km2 in 1981. The population growth in the previous decade had been 2.6% per year. The province of Requena, in which Santa Rosa andYanallp aar elocated ,ha da populatio n densityo f 1.0person spe rkm 2 (Ferrando 1985). Although such low densities suggest that sufficient land is available for

27 agriculture in the PeruvianAmazon , this is only partly true. Varzea land appropri­ ate for agriculture covers a limited area, and fanners living in a floodplain village have to compete for agricultural sites. Farmers living in many terra firme villages need to go far to find unclaimed land and, each time, fields have to be made at increasing distances from their houses. Since harvested products have to be car­ ried to the village, larger distances to fields are less desirable. Ribereno communities have acertai n formal autonomy overth evillag e area. The boundaries of this area are defined by the regional representative of the Min­ istry ofAgriculture , in coordination with the village authorities. Ribereno villages have two principal political authorities: the teniente gobernador, who represents the local government, and the agente municipal who represents the mayor of the municipality to which the villages belong. Their responsibilities are administra­ tive and they are empowered to keep local order. Both authorities are elected by the villagers. At infrequent meetings, village matters are discussed. Newly arriv­ ing inhabitants have to request membership in the community and, after this is granted, an area of land is designated for use by the newcomers. Although only the regional representatives of the ministry of agriculture can formally grant land userights, eve nwithi nth evillag e boundaries, thisagenc yusuall y follows the advice of the village authorities. However, favoritism is not unusual and leads to con­ flicts. Santa Rosa andYanallp a each represent a different type of ribereno village, both common in the Peruvian Amazon. Many caserios like Santa Rosa were founded by a labor force which stayed behind when owners abandoned a fundo (Hiraoka 1985a; Chibnik 1990; San Roman 1975). Since many fundos collapsed in only a short period of time, many of these villages have only a short history. Furthermore, most fundos were located on terra firme and the villages which re­ mained after the estates collapsed often continued to be located on the same site. Since the fields of farmers in this type of village may be located at considerable distances, they often have second houses in their fields and may spend longer periods there when much work needs to be done. Yanallpa represents the secondribereno villag e type. After the rubber boom,

28 l J/M \_ kas 0 50 \ Iquitos f^Ç) i

>v\^^<-^ f jWj Tamshi yacu 4° v^ /t-^-^mazwVriver

_ Maranon river J>\

^—' V^ucayali river (S "> pd^Jenaro Herrera ( \ /(\ ^) Yanallpa \ H 5" KM Santa Rosa Y Peru )

I Requena A. i i 74« 73°

Figure 3.1 General overview ofth e research area and locationth evillage s researched many riberenos engaged in commercial and subsistence agriculture as well as ex­ tractive activities in isolated puestos, or smaller agricultural estates, located on levees. Often several of these puestos were located on the same higher levee, and

29 in many places theyfinally merge d into a single village. However, even when this happened, these villages maintained astructur e ofa long chain of evenlydis ­ tributed farm houses alongth eriver channel , each surrounded by fields. Farmers of varzea villages are less likely to own second houses than are farmers of terra firme villages, since agriculturalfields ar e usually located nearth ehouse . Theonl ypubli ctransportatio nbetwee nth elowe rUcayal iregion ,an dIquito s isb y lanchas, i.e. largepassenge ran dcarg o ships. Most farmers ofthi sare adea l with government agencies and banks, and sell and buy most of their products in Requena (Figure 3.1). Requena, a town of 10,000 inhabitants, contains govern­ ment agencies and an agricultural bank. It is also an important regional market wherefarmer s sellproducts . Itha ssecondar y schoolsan da hospita l offering lim­ ited medical care. Several inhabitants of Santa Rosa andYanallp a own a second house in this town and spend considerable time there, especially when theirchil ­ dren attend secondary school. Farmers from the region have less frequent busi­ ness interactions with Iquitos. Travel to Requena is mostly in small outboard equipped boats owned by farmers.

THE UCAYALI ENVIRONMENT

Thetw oresearc h villages, SantaRos aan dYanallp aar e located at 73°45'W - 4°55'S, on the trueright sid eo f theUcayal i river, about 150k m southwest and upriver from Iquitos and about 70 km above the confluence of the Ucayali and Maranon rivers (Figure 3.1). The landscape of the lower Ucayali consists of a large floodplain infront o f the two villages, and a large terrafirme are a east of Santa Rosa (Figure 3.2 D). Thefloodplain are a is shared by the Ucayali and Marafionrivers an d extendst oabou t20 0k mupstrea mfrom th erivers' confluenc e (Dumont et al. 1990). It consists of levees, swamps, and many small and large lakes(Lope z &Freita s 1990). Iti sa par t ofth eNationa l Reserve Pacaya-Samiria and restrictions onth eus e of local resources exist. Hqwever,agricultur e closet o theriver, harvestin g of non-timber forest products, and non-commercialfishing areallowe d(COREPAS A 1986). Theterr afirme land st owhic hSant aRos afarmer s

30 haveacces s areTertiar y deposits, dissected bynumerou s smalrivers,l resultin gi n a hillylandscap etypica l oflarg earea so fAmazoni a(e.g .Deneva n& Padoch1987 ; Hiraoka 1986). As explained in chapter one, the natural levees, or restingas, vary in soil characteristics and elevation and, therefore, have different agricultural potentials (Hiraoka 1985a,b). Mudbar s(barreales) ,an dbeache s (playas)ar esoil s whichar e yearlyrenewed . Onbarreale sriberenos obtai nth ehighes tyield so frice ,thei rmos t

Ucayali river

5 km _J

Lovilla river

Figure 3.2 Map of the surroundings of the villages researched. Santa Rosa: A= location villagean d terrafirm e hillock, B= low leveeacros s the river, C= lowleve e downstream from village, Dterr a firme across Lovilla river, E = Mauritia flexuosa swamp infron t of village hillock, J and Klocation s of mudflats . Yanallpa: F =lo ­ cation village and principal high levee, G location of sand beach, H= location of lower levee across river.

31 important cash crop during the 1980s. Barreales are usually located adjacent to levees(Deneva n 1984),bu tals o liei ndepression sbetwee nridges (Lamott e 1990). Mauritiaflexuosa swamps ,o raguajales , area ver ycommo n feature inlarg e areas ofAmazoni a (ONERN 1977). Although their ecology is still being studied and conclusions remain provisional (e.g. Kahn & Mejia 1990), aguajales appear to be an early succession stage of varzeas which have been cut off from the river channelb ynatura l leveesan dwhic hwer einvade db yMauritia flexuosa (Kah npers . com.). Aguajales have standing water for most of the year and these conditions lead to organic matter accumulation (Gonzales Boscan 1987). As long as these palm forests contain standing water, they are not used for agriculture. The flooding regime is one of the principal determinants of agriculture in the Ucayali floodplain. Thefluctuation o fth e water level at the Jenaro Herrera research station is shown inFigur e 3.3. The lowestriver leve l isreache d around September, after whichth e waterstart s rising. Itsleve lrises ver y slowly,t oreac h itshighes tpoin taroun dApri lo rMay . Onceth ewate rleve lstart st olower ,i tdrop s in lesstha n two months to only a little above its lowest point. Inabou t 25t o3 0 days,however ,mos tagricultura l landsemerg e(Hiraok a 1985b). Thispatter no fa gradual rise ofth e water level followed by a fast drop is an important feature of the flooding regime, and as such is well understood by farmers. Althoughth echangin go fth ewate rleve lo fth eUcayal riveri show sa fairly regular pattern, minor differences between years are of extreme importance for varzea fanning. Very small differences in the highest level can have profound consequences for how much land is flooded. In turn, this influences where and how much soil is deposited, and therefore what land use options exist after the flooding season. Theflood o f 1985, for instance, was very low. It was followed by a much higher flood in 1986. However, the difference between the highest leveli n 1985 andth ehighes t leveli n 1986wa sonl y 1.5 m. The 1985floo d barely coveredth e lower levees inth e Ucayali floodplain and left manyfield s unsuitable for cropping. In 1986,eve n the highest parts of theriver levee s were inundated, leaving fields onbot h lowan d high leveesclea n and replenished with silt forth e following season. The twofloods differe d also in periodicity. The 1986 flood

32 O) > re >» re o

"55 > O)

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Months

Figure 3.3 Monthly water level in the Ucayali river at Jenaro Herrera, 1985 reached its highest level much later than the one in 1985, leaving moretha n suf­ ficient timefo rfarmer st oharves tal lthei rlowe rvarze afield sbefor eth e 1986flood . Onth e otherhan dth e late 1986floo d left insufficient timet oharves tfields befor e the 1987inundatio nan d manyo fthos egrowin g riceo nbarreale s lost part ofthei r crops.

CLIMATE, SOILS,AN DVEGETATIO N

The climate of the research area is typical of equatorial tropical rainforest regions. Datagathere d from 1974throug h 1983a t Jenaro Herrera show anaver ­ ageannua ltemperatur eo f26. 5degree s Celsius. Monthlyaverage s rangebetwee n 25.4 degrees inJul y and27. 1 degrees in December. The lowest monthlyprecipi ­ tation measured was 53.7 mm, and the highest 537.7 mm. The average annual

33 rainfall is252 1mm , while yearly averages variedbetwee n 1836an d 2800 mmi n the 10yea rperiod . Althoughther e isn opronounce d dryseaso n inthi spar t ofth e Amazonregion ,averag erainfal l islowe rbetwee nJun ean dSeptembe ran dbetwee n December and March than during the rest of theyear . Terra firme soils in the research area areclassifie d by Sombroek (1984)a s ferricAcrisol s (Paleudults) ororthi cAcrisol s (Tropodults). Thesear e well devel­ oped Ultisols with goodt o moderate drainage and medium tofine texture . They are acidic (PH 4.0 to 5.0), with a low to medium organic matter content. The base saturation varies from 35%t o 40%, while aluminum saturation is 30%t o 70%. In general, these soils are of low quality for agriculture, but are slightly bettertha n many Oxisols found in BrazilianAmazonia . Inth efloodplain, soil sdiffe r greatlyi ntexture . Sandiersoil soccu ro nlevees , while backswamps and mud bars display finer soils. Considerable micro-varia­ tion insoi ltextur ema yoccu racros ssingl elandform s (Denevan 1984). Thehighe r levees are composed of Fluvisols (Fluvents) with sedimentary stratification and free internaldrainage . Thelowe rstretches ,suc ha sth ebackswamps ,ar epredomi ­ nantly Gleysols (Aquepts, Aquolls), with restricted internal drainage. The latter arenon-acidi cwit h high-activitycla yminera lassemblage s (illite/montmorillonite) (Sombroek 1984). Natural terrafirme an d floodplain vegetation inth eresearc hare avaries . In SantaRosa ,terr afirme is largel ycovere db yprimar ytropica lrainforest . Research­ ershav eoffere d classifications ofthes eforest s (Encarnacion 1985;Lope z& Freita s 1990; ONERN 1976), but cannot agree on forest types (Lopez & Freitas 1990). The vegetation ofth e varzeaundergoe s more impact byabioti c factors than does terrafirme fores t (Salo et al. 1986; Dumont et al. 1990). Anewl y initiated suc­ cession on recently deposited lands may be interrupted before a fully developed forest is reached. The vegetation on a particular floodplain site depends on its age, the floodwaters that inundate it, the texture of sediments, the sedimentation rate, andth eperiodicit yo fflooding (Jun k 1984). Within shortdistance sa mosai c of forest patches in different successional stages may be found (Lamotte 1990; Salo et al. 1986; Dumont et aj. 1990; Encarnacion 1985). Periodic abiotic im-

34 pacts result in forests whoseflora is , on the average, less high, less diverse, and younger than those onterr a firme (Denevan 1984).

SANTAROSA :PEOPL EAN DPLAC E

Severalo fth epeopl ewh ofounde d SantaRos ai n 1960 hadlive dan dworke d ona fund o called MonteCarmelo . This fundo was founded around 1930an dwa s located on terrafirme onl y a few miles upstreamfrom today' s location of Santa Rosa (Figure 3.1). Some ofth e detribalized Ashaninka Indians, who now livei n thevillage ,wer ebrough tt oth eregio nfrom th euppe rUcayal it oprotec tth e fundo from raiding Mayoruna tribesmen. After the ownerclose d his estate, a group of laborers went with him to the Putumayo region, currently the border region be­ tweenPer uan dColombia ,t o extractfores t products. Theytravele d inth eemplo y of the owner for probably about a decade, during which time they also were en­ gaged inagricultura l activities onth eAmazo n andNap oRivers . Eachplac ethe y settled,th egrou p wasjoine db yne wpeopl eo fdifferen t ethnicorigin . Inth eearl y 1960sthe yfinall y returnedt oth eUcayal ian dsettle da tSant aRosa ,a smal lpuest o on ahig h leveeo nth etru eright rive rban kslightl y upstream from Yanallpa. The group includedforme r memberso fsevera l lowlandsocieties ,includin gAshaninkas , Yaguas from the lowerAmazon , and Quichuas from the upper Napo. All these peopleha dabandone dtriba l lifean d spokeSpanis hwit heac hother . Otherpeopl e of Santa Rosa have Cocama, Cocamilla, and Conibo tribal backgrounds. The adventures ofth epeopl ewh ofounde d SantaRos ahav ebee nextensivel y discussed elsewhere (Padoch & de Jong 1990). The natural levee where Santa Rosa was first located eroded away by the end of the 1960s (Figure 3.2, C). At that time, farmers started makingfields o n the terrafirme peninsul a on which the village is located today. However, most Santa Rosinos built their houses on the levee across the river (Figure 3.2, B)be ­ cause they feared the Mayoruna Indians, who used the terrafirme area s as their huntinggrounds . Theleve eo nth elef t riverban kwa sslightl y lowertha nth eleve e ofth e previous location ofth evillage , and several"hig hflood s destroyed mucho f

35 thefarmers 'crops . Theselosse s finally encouragedth eSant aRos apeopl et omov e to terrafirme land s in the early 1980s. After the village school was relocated to thepeninsul a hillock in 1981,mos t SantaRosino sbuil t housesthere , establishing the village on its present site (Figure 3.2A) . Santa Rosa is now located onth e right bank ofth e Ucayali river. Thehill ­ ock onwhic h Santa Rosa is located, ispar t of aterr afirme peninsul a surrounded by periodically inundated lands. To the West, the peninsula is bordered by a Mauritiaflexuosa swamp , or aguajal, which separates Santa Rosafrom th e river (Figure 3.2 E). Toth e Souththi s swamp stretches inland and forms the southern limit ofth epeninsula , whichals o forms theborde r ofth evillag eare a(3. 2 K). To the North the aguajal changes into a lower levee area (Figure 3.2 C). A small watercourse called Lovilla and its floodplain separateth e SantaRos apeninsul aa t the northeast from the inland terrafirme (3. 2 D). TheLovill ajoin s theseUcayal i rivert oth e north ofth evillage . TheMauritia flexuosa swamp , levee,an dLovill a floodplain all flood during the high waterseason . Onth esmal lterr afirme hillock ,o nwhic hSant aRos ai s located,th ehouse s arebuil t intw o rows on either side ofa wideun-levele d lane. Toreac h the river theMauritia flexuosa swam pha st ob ecrosse db ya nunstabl ebridge . Thevillag e lane leads to the school with a small soccerfield i n front. Although most Santa Rosinos prefer to live in the village center, several houses are located alongside theboundar y betweenterr afirme an dth eMauritia flexuosa swamp . Afe w Santa Rosafamilie s still livecompletel y withinth eperiodicall y flooded varzea. In 1986 Santa Rosa had 326 inhabitants, belonging to4 6 households. SantaRos aformers us ebot hvarze aan dterr afirme site sfo ragriculture . The entireterr afirme peninsul a hasbee n farmed. Todaypar t of it isunde rswiddens , and part under swidden-fallow agroforestry fields. Newfields ar e being opened upi nth eterr afirme are aacros sth eLovill afloodplai n (Figure3. 2D) . TheLovill a floodplain itselfcanno tb euse d foragricultur e sincei ttend st o flood severaltime s duringth eyea rafte r heavyrains . Inth evarze ath e lowlevee s north ofth evillag e (Figure 3.2 C),a s well as those acrossth eriver (Figur e 3.2 B),ar euse d foragri ­ culture. Barreales appear infront o f leveesacros sth eriver (Figur e 3.2 J)an do n

36 theright ban k ofth eriver i n front of the aguajal (Figure 3.2 E, K). Only aver y small area of sandy playa appears in Santa Rosa each year, next to the principal barreal area (Figure 3.2 J). Aerial photographs from 1957 and 1983 show how the Ucayali river has movedmuc hclose rt oth e SantaRos aterr afirme, erodin g largepart s ofth eorigi ­ nal aguajal. The part that remains isbein g built up byriver deposit s whenever it floods, and can be used for agriculture during low water. TheMauritia flexuosa population, if not slashed, is dying, perhaps because ofthes e deposits. Sinceth e Santa Rosa farmers moved to terrafirme, th e aguajal area has been used princi­ pally forrice growing .

A HISTORYAN D GEOGRAPHY OF YANALLPA

Yanallpa is much older than Santa Rosa. According to local informants, thefirst ribereno inhabitant s ofth e levee on which the village is located (Figure 3.2 F)settle d therearoun dth etur n ofth ecentury . Although somerubbe rtappin g occurred close to the village, the few puestos, or smaller agricultural estates en­ gaged merely in providing agricultural produce andfirewood t o boats and to the nearbyriver Tapich eare awer erubbe rwa sheavil yexploited . Thevillag eo fYanallp a was founded on the 12th of October 1911a s SantaCruz de Yanallpa by a man named Francisco Lovera. Much of the early history of the village is irretrievably lost. The people who lived longest inYanallp a came there in the early 1940's. At that time the villageconsiste d of3 0 housesan d about 220 inhabitants. Italread yha da centra l roadalongsid e theriver, a centra lplaza ,a school ,an d villageauthorities . During thisperio dfarmer sgre wplantains ,rice ,an dmanio ci nresting afields. Som efarmer s had a few cattle. According to several informants, many farmers grew fruit trees including oranges, Quararibea cordata, and Pouteria caimito, species which are no longer found in the village as they do not tolerate flooding. ManyYanallp a inhabitants apparently left the village around 1948 to live elsewhere onth eterr afirme becaus e ofrepeated , extremely high floods. Farmers

37 had sold their cattle. Fruit trees which could nottolerat e waterlogging had died, and the village was almost abandoned. There were neither school nor village authoritiesanymore . Aftera fe wyear so frelativel ylo wfloods , however,ne wpeopl e again settled onth eYanallp a high levee. Theincrease ddeman dfo rnatura lrubbe rfrom Amazoni ai nth eSecon dWorl d War(Padoc h 1987)resulte d in somecollectio n ofrubbe r inth eYanallp aarea . In the 1940's and 50's Yanallpa inhabitants went on long extraction expeditions to other regions, but later returned to the village. Sinceth e early 1960'sjut e (prob­ ably Corchorus capsularis) becameth e principal cash crop, but after a few years it was replaced by Urena lobata. In the early 1970s the Peruvian agricultural bank started providing bank loans for corn and rice production and established controlled prices for these products. Since then many farmers inYanallp abega n to grown corn andrice a s their principal cashcrops . Asketc h map ofYanallp a drawnb ya topographe r inth e Ministry ofAgri ­ culture in 1970 shows 44 different farms in the village. In 1986 only 14 of the ownerso fthes efarm s orthei rspouse s stilllive di nth evillage ,bu tsevera l ofthe m had changed the location of their farm. In the last two decades the village has moved slightly upstream since the downstream part ofth eprincipa l levee isno w separated from theriver b y a newly built lower levee. The village still has its main road along theriver an d a village center with school and soccer field. The houses arestretche d outalon gth e mainroad , with onlya fe w houses surrounding thecentra l plaza. Somepeopl ewh ofar m withinYanallp aboundarie s livemos to f theyea ri nth enearb ytow no fJenar oHerrera . In 1986a tota l of 322peopl elive d permanently inYanallp a in 50households . Yanallpa is located ona convex higher levee, which is part of a large river meander(Figur e3. 2 F). Bothhouse san dagricultura fieldsl ar efoun d onthi sleve e upt o itsupstrea m border,sinc eth egroun d levelremain s fairly even. Towardsth e downstream borderth e levelfalls . Thereth e leveei sstil luse d foragriculture ,bu t notfo rpermanen t housing. Theprincipa l villageleve ei sslowl yerodin gupstream , while downstream a newridge i s being deposited infront o f an older one. The newly built levee becomes part of a large beach during low water, reaching up-

38 stream asfa r asth e villagecente r (Figure3. 2 G). Thewides tpar t ofth eprincipa l levee (measured during lowflood) i s 360 m. The principal levee drops abruptly nearth emai nriver channel . Onth ebac kslope ,however ,th elan ddescend sslowl y and ends in a forested swamp area. Across theriver fro m Yanallpa, a low levee stretches along the main river channel (Figure 3.2 H). The width of this levee varies; it is about 200 m at its widest point. Behind the levee,th e mixed landscape of swamps, lakes, and low levees continues (Lopez &Freita s 1990). Except for the higher principal village levee, the entire area floods yearly. The higher levee is inundated only during extremely high floods, like the one in 1986. Agriculture is pursued on the main levee(Figur e 3.2 F)an d itsbackslopes ,o nth elowe r leveeacros sth eriver (Figur e 3.2 H), on the sandy beach (Figure 3.2 G), and on the mud bars that appear in front of the levees and in depressions onbot h sides ofth e river. Duringth e 1986 flood thewate rstoo d between 1.5 and2. 5 mabov eth efields o nYanallp aleve eo n thetru e leftriver sid e (Figure 3.2 H). Soils inthes e lowerrestinga s areo f loamy texture,bu t lessfine tha nthos eo nth e mudbars . Theprincipa lYanallp a leveeha s soils ofcoarse rtextur e onth etop ,bu tloamie r soils onth ebackslopes . Thehigh ­ est parts ofth eprincipa lYanallp a leveewer ecovere d withabou t 0.5m ofwate ri n 1986.

39 CHAPTER FOUR

AGRICULTURAL DIVERSITY IN FLOODPLAIN FARMING

INTRODUCTION

Agricultural diversity(se epag e2 8fo ra definition ) amonga specifie d group offarmers , isth e result ofthei rus e ofdivers eagricultura l practices. Variationsi n agricultural strategies arepossibl e in a single group of farmers because they use moretha n oneagricultura l practice. Hence, adiscussio n of agricultural diversity is necessary before the variation in agricultural strategies canb e assessed, asth e next chapter attempts. Whichar eth emos t importantcharacteristic s ofcertai nagricultura lpractice s dependso nth escal eo fcomparison . Forexample ,Duckha man dMasefiel d (1969) usedfou r degreeso f landus eintensit y and landus etype sa sth etw o maincharac ­ teristics for aclassificatio n ofagricultur eo na world-wid escale . Turneran dBrus h (1987: 7)us eth etyp eo ftechnolog y applied, inadditio n to landus e intensityan d production type, as athir d distinctive characteristic to classify farming practices. Themor especifi c aclassificatio n ofagricultura lpractice sbecomes ,th emor echar ­ acteristics can be included in distinguishing categories. Inth ecas eo fribereno agriculture ,th enatura lcondition s ofagricultura lsite s constitute oneo fth efactor s which determineagricultura l practices. Asexplaine d inth epreviou schapter ,an d following theclassificatio n usedb yHiraok a(1985a,b) , ribereno farmers in the Peruvian Amazon use six different biotopes for agricul­ ture. Since in each one of those site conditions aie different, the latter is to be one of the characteristics which have to be used for a typology of ribereno agri­ cultural diversity. Different sites areuse d to grow different crops, and eachspe ­ cific site-and-crop combination requires its ownagricultura l techniques. Further-

40 more, as a result of flood and climate seasonality, land use activities take place duringdifferen t parts ofth eyea ri ndifferen t sites. Thesecharacteristic s co-define ribereno agricultural practices. Anorganizationa l concept, calledagricultural type, isuse d heret odescrib e the differentribereno agricultura l practices. An agricultural type is defined as:a unique site-crop (or crop mixture) combination, which requires a specific set of agricultural techniques,an da specifi c calendaro factivities . Eachagricultura ltype , furthermore, hasa certai naverag eproductio n level, aprincipa l destination ofout ­ put,an dth etyp ean dleve lo frisk . Inthi schapte rth eagricultura l diversityamon g ribereno farmers of Santa Rosa andYanallp a is discussed by describing fourteen agricultural types which were observed inthos e two villages. Previously it will be explained why the concept of agricultural type is an appropriate tool to illus­ trate diversity ofribereno agriculture .

AGRICULTURAL TYPES,A CONCEPT FOR LAND USE CLASSIFICA­ TION

The agricultural type concept is most similar to the cropping system con­ ceptuse db ysevera lauthor s(e.g .Fresc o 1986;Conwa y 1985; Shanere tal . 1982). Accordingt o Fresco (1986: 110),a croppin g system is" a landus euni ttha ttrans ­ formsplan tmateria lan dsoi lnutrient sint ousefu l biomass". Besidessevera lphysi ­ cal components, a cropping system has inputs of management and labor. It is interestingt onot etha t Frescodistinguishe sfield types toaccoun tfo rth evariatio n between agriculturalfields belongin g to the same cropping system. Field types are defined as "the location of specific combination ofcroppin g systemscompo ­ nents in a givenagro-ecologica l and socioeconomic environment"(Ibidem) . Although anagricultura l type, as defined above, includesa specifi c combi­ nation ofth ecomponent s which Fresco usest o distinguish cropping systems,th e twoconcept s areno t identical. The wordsite, i nth e definition ofa nagricultura l type, includesa specifi c agro-ecological environment,correspondin gt obiotope a s usedb yDeneva n (1984)an dHiraok a (1985a,b)i nthei rcharacterizatio n ofvarze a

41 resourceuse . However, specific laborrequirement san dscheduling ,typ ean dleve l of risk and uncertainty, specific agricultural techniques used, and output levels, are factors which are also characteristic for a specific agricultural type. These characteristics may vary infields withi n one single biotope, and they, in particu­ lar, differentiate the agricultural type concept from the cropping system concept. Thedefinitio n ofa nagricultura ltyp e isclose rt oFresco' s definition ofa field typ e than to the cropping system. An importantjustificatio n for aclassificatio n interm s ofagricultura l types is its intrinsic combination ofecological ,technical ,an d economic features. Since eachagricultura l type described belowha s its ownrequirement s ast oamoun tan d calendar of labor input, risk, and yield levels,th e choice of a certain agricultural typereflect s anoptio nfo ra specifi c allocationo fthos elimite dproductiv eresources . Thismean stha ta choic efo ron eagricultura l typereflect s partlya farmer s agricul­ tural strategy. Henceth e agricultural type concept is also appropriate to evaluate agricultural strategies. The distinction ofparticula r agricultural types inth epresen t study isbase d ona classification of agriculturalfields i n Santa Rosa andYanallpa . Thecatego ­ rieswer e derived from surveys of numerous fields. They are consistent with a classification whichfarmer sthemselve sappl y inth etw ovillages . Theassessmen t ofth e economiccharacteristic s ofagricultura l types was obtainedthroug hpartici ­ pant observation, interviews with many farmers, and notes from daily work dia­ ries kept by a few farmers. The names of most agricultural types combine the name of the land form on which it is found, and the principal crop or cropcom ­ bination that isgrown . It was attempted to specify asmuc h aspossibl e a single set ofcharacteris ­ tics to describe each of the fourteen agricultural types discussed below. These characteristics are: (1) site conditions, (2) crops grown, (3) techniques used, (4) calendaro fcultivatio n activities,(5 )estimate sfo r laborinputs ,(6 )productio nlev ­ els, (7) destination of the output, and (8) type and level of risk involved. How­ ever, the precision of the presented information can not be the same for every agricultural type. Someagricultura l typeswer e found amongonl ya fe w farmers. As a consequence less information could be obtained about these types. In the following discussion, the estimates for labor input and production levels all refer

42 to afield equivalen t of onehectare . Of the fourteen agricultural types discussed below, ten are located in the floodplain and four onterr afir e lands. Thefloodplai n agricultural types included six located on several parts of the restingas: restingarice fields, resting a chacra, restinga plantainfields, resting a forest gardens, restinga corn fields, restingaveg ­ etablefields, an d restingajut e fields. The barrealrice fields wer e located onth e barreales, playa cowpea fields on the playa, and aguajal rice fields in the Santa Rosa aguajal area. Four terrafirme agricultura l types were distinguished among Santa Rosa farmers: terrafirme rice fields, terr afirme chacra , terrafirme plantai nfields, an d theterr afirme fores t gardens. Someremark sconcernin gth erelatio nbetwee nterr a firme agricultural types isnecessary . Thevariatio n insoi lcondition s ofth e Santa Rosa terrafirme landscap e is mainly a result of its dissected topography. Inth e typicalhill ylandscape ,soi lfertilit yan dmoistur eincreas efrom th eto pt oth ebotto m of slopes, while soil drainage is better on the higher parts. Amazonian farmers often adjust theirplantin go fcrop st osuc hgradient si nthei rfields (Deneva n1984) . In Santa Rosa these gradients occur mostly over a distance of less than 100m. However, the variation in agricultural potential of different terrafirme fields i s principally influenced by the land use history of the site and not by variation in soil quality. Aterr afirme rice field, terr afirme chacra , and terrafirme plantai nfield al l can be started as a field newly slashed in primary forest vegetation. However, both the chacra and the plantain fields may also follow as a second agricultural type after thefield ha s been usedfirst a s aterr afirme rice field. I t is most com­ mon, however, for achacr at ob e started as a new swidden, whilea plantai n field normally succeeds toa terr afirme rice field. Onl yth e forest garden isa nagricul ­ tural type that always follows a chacra or aplantai n field.

43 RICE AGRICULTURAL TYPES

Barreal rice fields Ribereno farmers only use dry rice cultivation. Irrigated rice farming in the floodplain would be severely limited byriver floods , even with a frequency of once every several years. The rice cultivars which riberenos plant are short cyclic vari­ eties. They are short grain rice varieties oîOriza sativa sspjaponica (Purseglove 1975;William s 1975;Gupt a & O'Toole 1986). In Santa Rosa andYanallp a farm­ ers grow rice in four ecologically different zones.

Figure 4.1: General calendar of some agricultural types with marked seasonality: C=preparation; P=planting; W=weeding; H=harvesting

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec barreal rice PP W HHHHH CC restinga rice WW HH CC CCCC PP aguajal rice CC PP WWWWWWWW HHHHH terra firme rice HHHHH CCCC PPP restinga chacra H H H H HHH CCPP WW WW restinga corn CHCHCH CPCP WW CHCHCPP WW playa cowpea PP HH

Theprincipa l limitation for agricultural production on barreales isthei r short flood-free period. In Santa Rosa andYanallpa , barreales start to appear as flood water recedes, somewhere in June, and become flooded again towards the end of October or early November (Figure 4.1). In most years this period is just long enough to grow one crop of the rice varieties planted by ribereno farmers. How­ ever, only a slight difference in the flooding regime may shorten the flood-free period and hence may lead to losses of a considerable part of the rice crop. Rice production onbarreale s in the region wasreporte d as early as 1943(Higbe e 1945), but Santa Rosa andYanallp a farmers have used barreales for rice production only

44 oo u Js «cH Om f Ui Ui O u, 3 00 00 00 G X O O (O g g u,2<£ '•5 O I o c 00 o o u '-3 .ßX 3 c •£ 00 K X O e 2-S «8 —" « ö «s O "~1C •c l-i u o O O 2 — 00 •-" O oog E £ 2 O X X ca D 'C T3 *i .S3 '>» co X •*-• a. ? 00 3 •o § M ° s^ O ° s Cd _« o CV X tó 2 X I as 2 ^ x O c<2 £ x 53 CV, x cv J=. I T3 •o •o g o U 3 13 3 "O co S o 2 O O o oo in I.Î 1.Î o«o ag o jj q .g — e u c 00 m c s C/l C3 .S S _C«3 00 co ca « G T3 cd p co C co X x X al ë 3 00 ca 00 9> o oo 00 es 4—* »4-* 4—» X Ä »-) X) ce Si .a o. ^ x

co S o g ca c o 'ca 1 .,-1 x ca <ü o f) o e E u 0 1) ca 4-» 00 2 O o X 0) 4-* o CU &> to CO CJ 3 l-l o S f > 5 l-RH

co gr o £

ro est i ca 00 O CU ca 00

45 sinceth eearl y 1970s. Withoutban k loansan dcontrolle d prices, ricewa sneve ra favored cash crop in barreales. However, during the present study barreal rice production was one of the most important cash crop activities among riberefios (Chibnik 1990;Pined o 1986). Barrealesusuall y emergea sclea ndeposit s ofheav ysilt ymateria l onwhic h theric eca ndirectl yb ebroadcas t(Tabl e4.2) . Soils,however ,mus tb ewe tt omak e thisplantin g method successful. Oncebarreale s start emerging, farmers plantth e newly appearing parts at least every other day. As soon as the soil dries,broad ­ castingwil lno tb esuccessfu l anymore. Broadcasting iscontinue dunti lth enewl y exposed soil reaches a level which is estimated to stay flood free long enought o allow harvesting of the rice. The land lying below this level is not expected to remain out ofth e water long enough for the rice cropt o mature, andtherefor e is not planted. Abarrea lrice cro p which is flooded, even if it does not totallysub ­ merge, does not produce a good yield. Local farmers hold that the sediment of therive rwate rtha t stickst oth eric eplant s damagesth ecrop . Theheav y silt load of the Ucayaliriver probabl y inhibits the production of flooded rice as it is re­ portedi nfloodplains in othe rpart so fth eworl d(Hank s 1972; DeDatt a 1981;Te n Have 1982). Barreales are excellent soils for rice production because of their high soil fertility, andgoo dwate rretentio ncapacit y(Roosevel t 1980; Purseglove 1975). The lattercharacteristi c is especially important sincewate rsuppl y isth echie f limiting factor inuplan dric ecultivatio n(Purseglov e 1975;Gupt a& O'Toole 1986; DeDatt a 1981). Under ideal situations, barreales yield upt o 4000 kg/hao f unhusked rice (Laur undated; Chibnik 1990). However, an average was recorded of onlysom e 1,500 kg/haamon gal lth e farmers in Santa Rosa(Tabl e4.1) . This lowaverag ei s a consequence of farmers growingrice i nsite s wereth eriver deposite d sediment which is to sandy, or the mud layer was too shallow. Although the average of barreal rice production is much lower than the potential maximum yield, it still compares favorably withproductio n levelso fsmal l farmers inman yothe rpart so f the world (e.g. Williams 1975). Abarrea l usually is only slowly invadedb y weeds, since the newlydepos -

46 ited mud contains few or no seeds. After the rice harvest, just before the flood, the area expected toreceiv e mud deposits during thenex t flood iscompletel y slash- weeded. The labor invested in field preparation and weeding in a barreal is much lower than in any of the other sites used for rice production. Broadcasting re­ quires only about five workdays/ha (Table 4.2). As little as five workdays may also be spent on weeding, while the amount of labor used for harvesting depends on the yield level. Since a single worker harvests an average of 80 to 100kg/day , a yield of 1.5 metric tons would require no more than about 20 days of work. A considerable amount of labor is spent on post harvest treatment (Laur undated). The labor spent on slash-weeding abarrea l before the flood varies between 20 and 30 workdays/ha. Broadcasting, weeding and harvesting add up to 60 workdays/ ha. Hanks (1972)report s an average of4 2 workdays/ha for broadcasting systems. In the systems he describes, however, land preparation is done by plowing, while average yields are 1.3 tons/ha. Although barreales areth e most favored sites for cash cropping, farming these sites involves several constraints and risks (Table 4.1). When ripe, the rice has to be harvested within a period of only a few weeks to avoid losses due to flooding. Moreover, the crop may become over-ripe and in that case much rice will be lost during harvesting. In addition to hiring labor, farmers must invest money, for

Table 4.2: Comparison of four rice agricultural types

Agricultural Use Planting Labor expenditure (day/ha) Type Frequency Technique Préparât. Planting Weeding barreal rice yearly broadcasting 25 5 5 restinga rice first dibbling 40 25 10 year aguajal rice yearly broadcasting 50 5 25 terra firme first dibbling 50 25 5 nee year

47 instance, for transportation ofth e product (Chibnik 1990; Laur undated). Sizes of barreal rice fields farmed by single farming units range between a quarter of a hectare and ten hectares. During the field work farmers relied on bank loans to hire non-family labor to harvest their barreal rice. The difficulties and risks of using bank loans have been discussed in more detail by Chibnik (1990). Several other risks threaten barreal rice growing. The greatest is crop loss to an early flood. Such losses occurred, for instance, in 1986whe n some farmers lost as much as half of their crop. Sometimes large amounts of seeds of obnox­ iousweed s maywas h intoth e field, increasing weeding laborrequirements . Float­ ing meadows, mainly consisting of Paspalumrepens and Echinochloa polystacha (Junk 1970) may be stranded on fields still flooded. If this occurs, these floats aredragge d intoth erive r again. Certain partso fa nare a expected toyiel d barreales may have been cleared before the flood, but then left without sufficient mud de­ posits once the water drops. This means that the labor invested is wasted. The soil deposited by the river may turn out to be to sandy, or barreales may disappear altogether as a result of river erosion.

Restinga rice fields On restinga sites, rice is mostly grown as a first crop in a field cleared for the first time in the forest (Table 4.2). In older restinga fields the risk of heavy weed invasion is high. Growing rice in an older restinga field is not very com­ mon. In newly slashed fields either fewer seeds of unwanted weeds are present, or most of them are killed by burning the debris. Sometimes farmers abandon a restinga rice field because of excessive weed competition. Within close range of the two villages studied, only restinga areas at relatively low elevations still have high forest, whereth eunderlyin g soil isappropriat e forrice growing . After aflood, the forest soil is left to dry before a new field can be slashed. New restinga fields, therefore, are not made earlier than August (Figure 4.1, see also Hiraoka 1985b). Once the forest is slashed, the debris must dry for about one month before it can be burned. Before planting, most of the unburned smaller stems and branches are gathered into piles where they are burned again.

48 Rice is planted in a restinga field using the dibbling technique (Table 4.2). About two months after planting the seeds, a minor weeding is done. Children often stay in the fields to scare away birds which come to feed on the rice when the crop is ripening. Harvesting usually starts four months after the rice has been planted. Farmers reported an average yield of 1,100 kg/ha, which is much lower than the 2.0 to 2.5 metric tons Hiraoka (1985b) reported. The highest yield re­ ported by a farmer in Santa Rosa was 3 metric tons/ha. Only shortly before the field is flooded, a complete slash-weeding is carried out to prepare the land for the next year's use. Farmers are much less willing to hire laborers to work in restinga than in barreal rice fields because of the lower economic return per unit of labor in a restinga rice field. The labor required to produce a rice crop in a restinga field is higher than in abarrea l field (Table 4.2). An investment of4 0 workdays to slash and burn the field was recorded, which is lower, for instance, than the 53 to 56 workdays re­ ported by Gupta & O'Toole (1986) from Sierra Leone. A total of 25 workdays are required for planting, while close to 10day s are spent on weeding. The labor spent on harvesting depends again on the amount of riceharvested , but because of more difficult conditions in a restinga field a worker can harvest less in a single day than in a barreal rice field. An average production of 1,100 kg/ha requires 15 to 20 days of harvesting. Thus a restinga rice field needs a total of 90-95 work­ days for producing one crop only. The main risk in growing rice in a restinga field is that production will be low and with itth e economic return of the labor invested (Table4.1) . A low yield may be due to low soil fertility, but more likely to insufficient moisture during the growing cycle. Restinga soils are sandier than soils in barreales (Hiraoka 1985a), and have a much lower water retention capacity. A rice crop requires a minimum of 750t o80 0m mo frainfal l duringth egrowin g season (TenHav e 1982;Purseglov e 1975). Since a rice crop is especially sensitive to environmental stress during the reproductive phase (Williams 1975), short periods of dry spell, which are com­ mon in the Amazon region, also may cause significant yield losses (Gupta and O'Toole 1986).

49 Aguajalrice fields Growingrice i nth eaguaja l area, immediately West infront o f SantaRosa , differs insevera laspect s from thetw o other floodplain riceagricultura l typesde ­ scribed so far. Ane waguaja l field isprepare d andplante d in Septembero rOcto ­ ber, following a schedule similar to a restinga rice field. An aguajal field, how­ ever, canb euse d to grow rice for several consecutive years,probabl y becauseo f the high organic content and moisture of the subsoil (Gonzales Boscan 1987). Weedsar eth eprincipa lthrea tt oth erice crop . Anaguaja lfield therefor e canonl y be used in a given season if during the previous flood sufficient sediment was deposited to inhibit excessive weed growth. Shortly before an aguajal field is flooded, it is completely slash-weeded. SantaRos a farmers haveuse d the aguajal area in front ofthei r village only since 1984. Since then, the main river channel has moved closer to the aguajal, and each year a larger part of it isflooded wit h sediment-rich water. The increased use since 1986i sa consequenc e ofdepositio n ofsediment si na large rare ao fthi s landform. Accordingt ofarmers ,bot hth efloo d andsedimen t depositsreduc ewee d invasion. Therear en oalternativ ecrop s suitablefo r theaguajal . Ifrice canno tb e planted, the labor spent on pre-flood slash-weeding is wasted. In 1985, for in­ stance, only seven SantaRos a farmers had planted their aguajalfield becaus eth e previousflood ha dbee n too low. Inexistin gaguaja l fields,rice i sbroadcas t immediatelyafte r thesubmerge d land reappears out ofth e water. Existingaguaja lfields emerg e shortlybefor e the barreales,an dthu s broadcasting and harvesting are donea little earlier. Weeding starts about onemont hafte r thefield i splanted , and iscontinue dunti l therice i s ripe. The harvestingperiod s ofbarrea l and aguajalrice fields ma yoverlap . This increases pressure on the labor supply if a farming unit manages a barreal and aguajalrice field a t the sametim e (Figure4.1) . Theproductio n ofa naguaja l ricefield average d 2,500 kg/ha, which isver y high compared to many other upland small agriculturistsrice cultivatio n systems (Williams 1975; Hanks 1972;Gupt a& O'Tool e 1986). Thisestimate ,however ,i s based on data of only five fields. Labor requirements for planting and pre-flood

50 slash-weeding areabou t thesam ea s for abarrea l rice field. Theamoun t of labor spent onweedin g is2 5t o 30workdays ,bu tma y increaseunde runfavorabl e con­ ditions (Table4.2) . The principal risk in farming aguajal ricefields i s an excessive growth of weeds,whic hma yresul teithe ri nlarg elabo rrequirement s forweeding ,o ri ncro p losses whenpart s offields ar eabandone d (Table4.1) . Laborspen t ona pre-floo d slash-weedingha sa highe rris ko fbein gwaste dtha n ina barrea l orresting a field. However, there is very little risk of losing a crop to an early flood.

Terra firme rice fields Terrafirme soil s are much less suited for growing rice than arebarreales , restinga,o raguaja l soils. Althoughsoi lfertilit y isadequat eafte r burningo fslashe d vegetation (Sanchez 1973), the lack of sufficient rainfall during the four month cropping cycleaccount s forth elo wrice production . Farmers inothe rpart so fth e PeruvianAmazon ,wh ohav en oacces st o varzeagrounds ,commonl ygro wrice i n thisenvironmen t (Hiraoka 1989b). Duringth eperio do fresearc honl ya fe wSant a Rosa farmers producedrice i nterr a firme. Oncea ne wterr afirme field is slashed ,felled , andburned ,th erice is plante d usingth edibblin gtechniqu e (Table4.2) . Ifth efield is mad ei nprimar y forest, no weeding isrequire dbefor e theharvest . Aterr afirme rice field mad ei nsecondar y forest requires some weeding. Soon after therice plantin g isfinished plantain s may be planted. In such a case the field is continued as a terra firme plantain field onceth erice i sharvested . Terrafirme rice fields als oma yb ereplante d with manioc, or plantains after harvesting. Terra firme rice is usually grown from December to April (Figure 4.1). During this period of the year rainfall is highest. Not only is this period most appropriate because of the higher rainfall in February and March, it also allows farmers tofinish harvestin gth e uplandrice befor e the varzea lands emerge. The newterr afirme field ha s to be slashed around September or October, andburne d just after the short dry period in December or January. The yield of eight terra firmerice fields average d 763kg/ha , which is much lesstha nan y otherrice agri -

51 cultural type. Labor investments for weeding are low. Harvesting requires be­ tween 15an d 20 workdays. Theprincipa l risks involved ingrowin g aterr a firme ricefield ar ea poo rbur n ofth e forest debris, but more important, lowproductio n as a result ofunfavorabl e growing conditions (Table4.1) .

CHACRAAGRICULTURA L TYPES

Restinga chacras Thewor dchacra isofte n usedamon g local fanners to indicatean yagricul ­ turalfield i ngeneral ,bu t refers mostlyt oa multi-cro pfield i nwhic h maniocpre ­ dominates. Thewor d isuse d hereexclusivel y to indicate fields ofth e lattertype . Restingachacra san dresting aplantai n fields arebot hlocate do nresting asite swit h only higher elevation. Since plantains are extremely susceptible to waterlogging and do not produce in less than one year (Williams 1975;Purseglov e 1975),th e veryhighes tpart so frestinga sar eusuall yuse dfo rplantai nfields. Resting achacra s aremor e often located onslightl y lowersites . Restinga sites used for chacrasar e composed of the same sandy material asplantai n sites, but in contrast to the lat­ ter,the yar e flooded during mostyears . This meanstha t only cropswhic h canb e plantedan dharveste dwithi na tim espa no fabou teigh tt ote nmonth sca nb egrow n ina resting a chacra.Ther e arefewe r cultivated species inresting a chacras than in chacras onterr afirme (se e Hiraoka 1985b). The restinga chacra provides manioc, which like plantains constitute the principal staple crop amongriberenos (Tabl e 4.1). In Santa Rosa and Yanallpa only varieties of sweet manioc are grown, which have no toxic levels of prussic acid inth e ediblepar t ofth etuber s (Purseglove 1974;Co ble y 1976). Most ofth e yield ofa resting a chacrai sdestine d for household consumption. Whenth eman ­ ioc of a restinga chacra is harvested there is an excessive supply in local markets and prices are low, so little of it is sold. A significant part of the manioc pro­ duced is used to make manioc beer, which is an important food in the labor ex­ change customs among riberenos (Chibnik & de Jong 1989). The annualriver cycle forces farmers to harvest all the restinga manioc by the end of the season,

52 sinceth ecro p doesno ttolerat ewaterloggin g (Purseglove 1974),an dstarch yroot s rot within a few days after flooding. The manioc, which is harvested in large quantities before the flood, is most commonly processed intofarina or baked manioc flour, since the tubers cannot be preserved for a long time (Purseglove 1974). Farina is an important daily staple during the high water season in vil­ lages located entirely inth e floodplain. It can beconserve d and sold more easily than manioc, and most farmers sell some of it in Requena or inth e village itself. Restinga chacras are usually made in fields that have already been used previously. The sandy and well drained soil, which appears after the flood and whichwil lb euse da sa chacra , needst o dryou tfo ra shorte rperio dtha nth elowe r restingafields destine d togro wcom . Ifth e sitet ob euse d asa chacr awa sslash - weeded before theflood, the n very little weeding hast ob e donebefor e planting. Manioci sthe nplante dusin gste mcuttings ,an don emont h laterth esecon dweed ­ ing is carried out. Thefield i s weeded athir d timejus t before ordurin g thehar ­ vest. Althoughmanio cca nb egrow nfo ra smuc ha s 10 monthsi na resting achacra , certainvarietie salread yma yb eharveste dafte r onlyfou rmonths . Onlyshor tcycli c cultivars of manioc areplante d in restinga fields. According to Hiraoka (1985b) production of manioc infloodplain fields range s from six to 13metri c tons per ha. Althoughthi s is lowertha nth eaverage s givenb yWilliam s (1975) for Brazil, it is not much lower than production levels ofterr afirme fields i n Santa Rosa. Tocultivat e a restinga chacra only family labori sused . Planting a manioc field requires about 10workdays , while 25 workdays are needed for harvesting. Thethre eweeding srequir ea nestimate dtota l of5 0t o7 0labo rdays . Theupcom ­ ing flood may force farmers to harvest all the manioc early and convert it into farina, or to store it in underground pits wrapped in Calathea sp. leaves. In the latter case, the manioc must be madeint o farina after the flood. Aresting a man­ ioccro p has fairly constant yield, if it isplante d earlyan dca nb ecompletel yhar ­ vested. However, suddenflooding o fth efield ma yresul t in losses if farmers are notprepare d for it, orar eabsen tfrom th evillag ea tth etime . Planting stockma y belos tdurin gth eflooding period ,an dthe nmus tb ereplace dfrom elsewhere . Weed invasion may result inmor e labortha n expected, anda sudde n needt o harvestal l

53 the manioc may interfere with other urgent activities, such as harvesting rice or corn in restinga fields.

Terrafirme chacra s The Santa Rosa terra firme chacra is similar to the typical upland agricul­ tural fields described for many Amazonian farmers (e.g. Denevan 1971; Gasché 1980;Guyo t 1975;Eden &Andrad e 1987;Ede n 1980; Harris 1980; Ruddle 1974). Although the species composition mayb e different fromfield t ofield, a chacr ai s always planted with manioca sth e most important crop,an d often but notalway s with plantains,a sa n important second crop. If Santa Rosa farmers growmanio c and plantain on afield, th e plantain usually is plantedfirst, an d the manioc sec­ ond. This sequence is different, for instance, among Tamshiyacu farmers near Iquitos,o rBor afarmer s whofirst plan t maniocan dthe nplantain s (deJon g 1985; Padoch et al. 1985;Deneva n & Padoch 1987). If thetw o crops are mixed, man­ ioc isplante d inth e wholefield, whil e plantains areplante d only incertai nparts . Other very common chacra crops are pineapple and sugarcane, which are also planted in certain areas of the field. Several studies onAmazonia n swiddeners reportth e largenumbe ro fcrop swhic hca nb efoun d insimila ruplan d agricultural fields (Denevan &Padoc h 1987;Hiraok a 1986). Thebes t period to slash and fell the forest vegetation for a newterr a firme field isbetwee nJun ean dDecember ,since ,i ntha tperiod ,rainfal l isles stha ndurin g theres t ofth eyear . However,th e demand ofagricultura l activities onth e farming unit labor pool is lowest during the high flood season, so sometimes Santa Rosa farmers makethei rfields in th efirst hal f ofth eyear . Inth e lattercas ethe yaccep t the increasedrisk tha tth e debris willdr y lessthoroughly , resulting ina poo rbur n andlowe rfield quality . Afarme rusuall ywil lplan tmanio can dplantain sa squickl y as possible,bu t may continue to plant othercrop s over moretha n one year. The weedingpatter n interr afirme fields whic h weremad ei nprimar y forest isslightl y different fromfields whic h were made in secondary forest. Primary forest fields areweede dfo rth efirs t timefou rt osi xmonth safte r theburnin go fth efores t debris. The second weeding is combined with harvesting of the manioc. In secondary

54 forestfields th efirst weedin g begins one month after burning. The slashing and felling of the forest vegetation and planting need to befinished a s soon as pos­ sibleonc ethe yar estarted . Mostothe rmanagemen tactivitie so fterr afirme chacras , however, are scheduled in-between more urgent activities. The terrafirme chacr a is a variable agricultural type. Crop combinations vary among fields, and change during afields cycle . Furthermore, soil fertility changes during the life time of thefield (Ny e & Greenland 1960; 1964; Jordan 1981; 1989;Sanche z 1973), andweedin gpattern svar yaccordin gt oth efields lan d use history. A terra firme field is managed as a chacra for one to four years. Sometimes only one manioc crop mayb e harvested, before thefield i sconverte d to a forest garden. Complete fallowing ofa terr afirme chacr a isunusual . Farm­ ers more commonly plant a second manioc crop after the first one, or plantains may be replanted and harvested over a longer period. The weeding pattern ofa terrafirme chacr a changes as the production level of manioc and plantain drops, or when weed competition becomes toosevere . Between 26an d 59workdays/h a wererecorde d for the slashing ofprimar y forest, and2 5fo rsecondar yforest . Plantingo fa singl emanio ccro prequire sabou t 10workdays , and weeding and harvesting some4 0t o 50 workdays. Thesecon d manioc croprequire sa t least 80workday s intotal ,sinc emuc hmor etim ei sspen t onweeding . Ina terr afirme chacr aadditiona l labori sspen t onplantin g andhar ­ vesting of other crops. Production levels of terra firme chacras are difficult to measure since they are so variable, andfields ar e gradually harvested. Experi­ ments in the nearby Jenaro Herrera research station resulted in production levels ofmanio cbetwee nnin ean d 20tons/ha . Variation inproductio n levels,mor etha n anything else, isth e result of differences inth e careth efield receives . Thebul k ofth eyiel d ofa terr afirme chacr a is destined for household consumption (Table 4.1). Sometimes, however, farmers may sell considerable amounts of manioco r plantains when prices are high, for instance a few months after a high flood pe­ riod. Chibnik (1990) haspointe d outtha triberefios attemp t toassur ethemselve s of a steady food supply from theirfields, eve n if they take higher risks in their

55 cash croppingactivities . Therisk involve d ingrowin g aterr afirme chacr ai sver y low. Pests and diseases areo f minor importance in Santa Rosa chacras (Lourde pers.com. 1986). The most important risk isa perio d ofunfavorabl e weather fol­ lowing the clearing of afield, resultin g in a poor burn, and therefore lower soil fertility.

PLANTAIN AGRICULTURAL TYPES

Restinga plantain fields Plantains and bananas (both cultivars of Musa paradisiaca) are important cashcrop s aswel l as food cropsamon griberenos . Plantains require welldraine d soils, but at the same time demand a sufficient water supply (Purseglove 1975). In the Peruvian Amazon this crop is grown both in the floodplain and on terra firme. Thebette r soils for plantains are located inth e floodplain. Inth e village survey, plantains were the third most important cash crop inYanallpa . Although all Musa paradisiaca cultivars are in fact annual crops, perennation occurs as a result ofvegetativ ereproductio n throughshoo t forming (Williams 1975). Planta­ tions ofthi s crop therefore mayreac h advanced ages (Purseglove 1975). Aplan ­ tain field on a high restinga can continue to produce for a period of at least 15 years without any need of fallowing. After this period the production decreases, probably as a result of nutrient deficiency. The fallow period of apreviou s plan­ tainfield nee dno tb e longertha ntw oo rthre eyears . Ifa plantai nfield i s flooded, the crop is almost completely destroyed. In such a situation, the field is often first usedfo ra differen t agriculturaltype ,befor ebein greplante dt oplantain . Even ifplantai nfields ar eno t flooded, a high subsoil water level doesaffec t aplantai n crop during highfloods, ofte n resulting in toppling ofth eplants . Plantainfields onl yhav et o bereplante d whena cro pi sdestroye d by flood­ ing, or after more thanfifteen year s of continuous plantain production. Shoots areplante d ona gri dpatter n ofabou t4 m by4m . This isa lowerdensit ytha no n terrafirme, wher e plantains are planted on a grid of about 3m by 3m. Although plantains can be planted at higher densities in the more fertile floodplain soils

56 (Purseglove 1975;William s 1975)riberenos, recogniz e that in later stages ofth e fields, through the crop's vegetative reproduction, these higher densities will be reached anyway. A mature plantainfield need s only one or two slash-weedings peryear . Theshad eo fth edens eplantai ncro pconsiderabl y reducesth egrowt ho f weeds (Williams 1975; Purseglove 1975). Most farmers schedule these slash- weedings duringth e endo fth ehig hwate rseaso nwhe nther ear efe wothe rurgen t tasks. Plantains are harvested whenever the racemes aremature . Onlylittl edat awa scollecte d onyield s ofplantai nfields. Hiraoka' s(1985b ) estimate of 200 racemes per year, however, appears lower than estimates from Yanallpa. Oneplantai nagroforestr yfield tha t wassurveye dha d53 3 plantainclus ­ ters per ha, many of which had full-grown new shoots. A mature restinga plan­ tainfield ma y have well over 1000adul t plants, and, intha t case, a yield of 500 racemes per year is a conservative estimate (Table4.1) . The laborrequirement s ofa resting aplantai nfield begi n with 30workday s forplantin ga field. Duringth efirst yea ro fa field's existenc eabou t 50workday s have to be spent on weeding. A mature plantain field, however, only requires between 20 and 30 workdays for one ortw o slash-weedings.A tota l of 19 work­ days/ha were recorded for a slash-weeding of atw o year old terrafirme plantai n field. The first harvest of a new plantainfield, abou t one year after planting, is concentrated withina few months. Insubsequen t yearsth e spacing ofharvestin g becomes irregular(Purseglov e 1975;William s 1975),an dafte r afe wyear sa field produces during the whole year. Risks of plantain growing are low cash returns becauseo flo wmarke tprice s forplantain ,an dtota l crop losswhe na field floods . Atth etim eo ffield research , infection ofFusarium oxisporum, which wasalread y common inth e vicinity ofManau san d Iquitos attha ttime , did notoccu r inplan ­ tain inYanallp a and Santa Rosa.

Terrafirme plantai n fields Ina terr afirme plantai nfield, beside sth epreparatio n andplanting ,th eonl y activities required is oneslash-weedin g per year and harvest ofth e produce. As

57 in restingafields, terr afirme plantai nfields star t yielding racemes about oneyea r after planting, but, because of the composition of plantain and banana cultivars which have different maturing periods in olderfields, raceme s are harvested the year round (Purseglove 1975;Hiraok a 1989b). A recently slashed terra firme forest area is sufficiently fertile for a good first yearplantai n crop. Therainfal l regime ofth eregio nassure s aminimu mpre ­ cipitation of5 0m mpe rmont hrequire d forplantai n onth ewel ldraine dterr a firme soils (Williams 1975). However, plantainfields i n terrafirme ar eonl y harvested for about four years, after which the production declines. It is not clear whether the decline inyiel d isa resul t ofdecrease d soil fertility orexcessiv e weedgrowt h (Purseglove 1975). Thetota l laborinpu t required for aterr afirme plantai nfield i s much lower than for achacra . The slash-weeding ofa two year old terrafirme plantai n field took 19workdays/ha . Since aplantai nfield i sharveste d piecemeal allyea rlong , it is difficult to accurately estimate howmuc htim e is spent onharvesting . Farm­ ers estimate that a two year old field yields about 400 racemes. In subsequent yearsproductio n is lower. Hiraoka (1989b) reports total yieldsbetwee n 1300an d 1800 racemes in similar fields in a village close to Iquitos. Harvesting of 400 racemes foron eyea rwoul drequir eapproximatel y2 0workdays . Althougha singl e person could harvest moretha n 20raceme s perday ,farmer s mostly collect onlya few each time they harvest a plantainfield, an d thus spend relatively more time looking for the produce of aterr afirme plantai n field. The produce of a terrafirme plantai n field is largely sold in the market in Requena, and a small part of it is consumed by the household members (Table 4.1). A terrafirme plantai n field is therefore principally a cash crop agricultural type, unlike aterr afirme chacra . The mainrisks involve d in terrafirme plantai n fields are lowproductio n because ofpoo rsoi l quality,o r lowprice s inth emarket .

58 FORESTGARDE NAGRICULTURA LTYPE S

Restinga forest gardens Theresting afores t gardens whichwer esurveye d inYanallp aar emor evari ­ able than any otheragricultura l type described so far. Agroforestryfields ar elo ­ catedo nbot h lowan d highrestingas ,an dthei rcompositio nan d weedingpattern s differ considerably. Somefores t gardensconsis t oftree sonl ywhil ei nother stree s are grown in combination with corn, manioc orplantains . Padoch& d eJon g(1989 )poin t outtha t worki nfores t gardens isdon ever y infrequently and only for very short periods. This makes labor investment and production in these fields very difficult to estimate. Moreover, since weeding patternsan d species composition offields ar ehighl y variable,th e labor inputan d production levelsals ovar ywidely . Inchapte rseve nth evariatio n inspecie scom ­ position and weeding patterns of different forest gardens inYanallp a will bedis ­ cussed. A species found in manyYanallp a agroforestry fields is Cedrela odorata. This tree species is native to the floodplain environment and has been incorpo­ rated by local farmers inthei rfields fo r its marketable timber. Some forest gar­ densproduc etre efruit s whichar esol d inth emarket ,bu tmos tproduct sharveste d from forest gardensar efo rhousehol d consumption. Asreporte d inchapte rthree , farmers grewwater-sensitiv ecrop s inagroforestr yfields i nth epast ,bu tthes eha d all disappeared after the 1986 flood. It can be expected that farmers will plant specieslik eorange ,o rQuararibea cordata againi fth ehig hresting adoe sno tfloo d for severalyears .

Terrafirme fores t gardens Terrafirme fores t gardens arecharacterize d bya dominant treevegetation . Most ofth eSant a Rosafores t gardens were oncechacr a orplantai n fields. How­ ever,th eter m forest garden isused , and not fallow, since thesefields continu et o be of significant economic importance. Evenfields whic h receive very little at­ tention still contain a largenumbe r of individuals of managed species, and yield

59 harvestable products. Managing afield a s a forest garden is economically more attractive than complete fallowing, from which it should be distinguished. Forest gardens onterr afirme ar ea sdivers ea sthe yar ei nth evarzea . Fields differ incompositio n and density ofmanage d species, as well as in management patterns. The principal management activities in forest gardens are weeding and harvesting, whilei nsom efields ne wspecie scontinu et ob eplanted . Comparedt o theothe ragricultura ltypes ,productio n levelsan dlabo rinvestmen tar elo w(Padoc h & deJon g 1989). Althoughsom ehighl ymanage dfores t gardensma yb emaintaine d forman y years, usually they have a limited life span. At one point, suchfields ar e used again for different agricultural types. The large number of products yielded by forest gardensar eprimaril y destined forhousehol dconsumption . Afe wproduct s mayb e sold inth e market (Padoch &d e Jong 1989). Very littlerisk i s involved in managing a forest garden since little is invested and not much canb e lost. In chaptersi xth evariatio n inspecie scomposition , weedingpatterns ,an dproductio n levels ofterr afirme fores t gardens will be discussed indetail .

OTHER FLOODPLAIN AGRICULTURAL TYPES

Restinga corn fields Although corn is known as a crop which needs good soil fertility, it hasa lowermoistur erequiremen t than rice. Foron egrowt hcycle ,cor nrequire sa mini ­ mum rainfall of only 200 mm, which makes it an appropriate crop in regions to dryfo rrice (Purseglov e 1975). Thisma yexplai nwh yproductio n levelso fresting a cornfields ar eles svariabl etha no fresting aric efields. O nth eothe rhand , riberefio farmers dogro w rice interr afirme, wher e moisture isth e limiting factor for this crop. Ribereftos dono tattemp t togro wcor n interr afirme fields. Th ehig hnitro ­ gen demand of this crop, especially beginning about two months after planting (Purseglove 1975;Cunar d 1967a)make s itscultivatio n unsuccessful unless fertil­ izersar eapplied . Although nitrogen is sufficiently availableafte r burning aterr a firmefield, th eavailabl e amount ofthi snutrien t becomes limited after aonl y few

60 weeks (Nye & Greenland 1960, 1964; Sanchez 1973;Jorda n 1989;Ewe l 1986). Corn isver y sensitivet ofree aluminu m caused bya lo wsoi l PH(Cunar d 1967b), and this makes terrafirme field s less suitable for this crop. Farmers in the Peru­ vianAmazo nwh od ogro wcor n interr afirme d othi s inslightl y morefertil e soils, orus esom esor to flocall yproduce dgree nmanur e(Vicker s 1976;1978 ; Casanova 1975; Ruddle 1974;Deneva n 1971; Johnson 1983;Bale e 1984; Brown 1984). Arip ecor ncro pdoe s nothav et ob eharveste d alla tonce ,bu tca nb e left in the field for a considerable period. During this additional period the moisture content ofth egrai n isreduce d (Purseglove 1975). Thus harvesting ofa cor ncro p can be scheduled overa longerperio d than rice, and therefore canb e carried out with family labor only. There is less pressure on labor resources ofth e farming unit whena corn crop isripe tha n inth e case of aripe rice crop . Even if a ripe, oralmos tripe, cor nfield i sflooded, th eyiel d is not seriously threatened. There­ fore, corn canb e grown infields whic hflood quit e early, ortw o successivecrop s ofcor nca nb e grown inon eyea r ina single field. It is, for instance,commo n in Yanallpaan dSant aRos at o see farmers standing waist deep inth ewater , harvest­ ing a corn field. Planting corn in old restingafields result s in excessive weed growth. Al­ though this may reduce the production (Cunard 1967b), corn appears to tolerate weed competition much better than rice. Aresting afield ha st o dryfrom on et o threemonth safte r it emergesfrom th ewater ,befor e itca nb eplante dt ocorn . By the time it is dry, a thorough slash-weeding has to be done to eliminate newly emerged weeds. Corn isthe n planted using the dibbling technique. After about one month a second weeding is done, but after this second weeding thefield i s notcleare d again untilth ecor n isripe (Figur e4.1) . Bytha ttim eth efield i scov ­ ered with dense weeds. Ribereöo farmers may weed thefield before , during, or after harvesting. Bergman (1974) reports that Shipibo farmers usually maintain varzea cornfields free o f weeds throughout the growingcycle . In most restinga corn fields, a second crop isplante d immediatelyafte r the first has been harvested. The schedule ofth efirst cro p is followed again. When ripe,thi s second crop is weeded to facilitate harvesting as well as to make sure

61 that the field is slash-weeded before it is flooded. Lower restingafields ca n be used continuously, but sometimes are fallowed for one or more years if they are not needed (Table 4.1, see also Bergman 1974). A corn crop may yield as much as 3000 kg/ha of seeds according to local informants. A detailed record ofth efield o f one farmer was kept, who only har­ vested 1300kg . Theplantin g ofa cor nfield require s some 10workdays , andth e harvesting 20 to 25 workdays. The first two weedings require some 10 to 15 workdayseach ,whil efo rth efinal weedin gabou t2 5workday smus tb espent . Post- harvesting corn treatment requires more labor than a rice crop does. The labor input for the production of one single corn crop is about 85 workdays. This is lower than for a restinga rice field. Because of the relatively stable market de­ mand for corn and lower risk as compared to a restingarice field, cor n was the second most important cash crop among Yanallpa and Santa Rosa farmers. The lowerrisk involve d ingrowin g corn ina resting afield correspond s to a lowerre ­ turn for the labor invested, according to the price levels of 1986an d 1987.

Playa cowpeafields Playas, or the very lowly elevated sand deposits in front of restingas, are outo fth e waterdurin g even shorterperiod s thanbarreale s are. Peanutsan dmel ­ ons have a growth cycle which is short enough to yield a crop that can be har­ vested, and they can produce well on the sandy soils (Hiraoka 1985a; Bergman 1974). These two crops are commonly grown by riberenos in other villages,bu t Yanallpaan d SantaRos a farmers only grew cowpeas onplaya s during theperio d offield research . Cowpeasgrow n forthei rgrain sproduc ea cro p inabou t9 0day s (Figure4.1 , Purseglove 1974;Muleb a& Ezuma h 1985). Furthermore,the ytoler ­ ate drought and other soil stresses well (Rachi 1985) and are moretoleran t than, for instance, Phaseolus spp. (Purseglove 1974). Although the latter crop gets a better market price than cowpeas, they are not grown in playafields becaus e of their difficult production. Taking care of a cowpea field is mainly a woman's responsibility. Playa fields areplante d immediatelyafte r theyemerge . Cowpeasar eplante db ymakin g a small hole with a machete and dropping several seeds in it. The broadcasting

62 technique,commo n inAfric a (Purseglove 1974),i sno t used. Playacowpe a fields are not weeded, sincefields appea r after theflooding seaso n free of weeds and without a soil seed bank. Weed competition, or pests, which are the most com­ moncause so flo wcowpe ayield si nAfric a (Rachie 1985;Muleb a& Ezumah 1985) are absent. Theentir epod s ofth ecowpe acro par eharveste d inth efield, an dar eshelle d at home. Thetota lyiel d of shelled cowpea mayreac h 700kg/h afo r asingl ecro p (Table 4.1, Hiraoka reports 0.6 to 0.9 metric tons), which compares favorably to levels of 200 to 300 kg/ha reported for cowpea yields of small farmers in Africa (Purseglove 1974; Rachie 1985;Muleb a & Ezumah 1985). Cowpeas are grown for bothmarke t salean dhousehol d consumption. Sincethi s crop does nothav ea guaranteed marketprice , it isgrow nmuc hles s frequently thanrice o rcorn . How­ ever,cowpea sca nb estore d for alon gtime ,an dfarmer s usuallyconserv esom eo f apreviou s harvesta t homea sa protei n sourcewhe nfish ar escarce . Farmerssel ­ domris kgrowin gcowpea so nsite s liablet o flood before thecro pca nb eharveste d because of its limited economic importance.

Restinga jute fields Restinga jute (Urena lobata) was only produced during the second inter­ viewperio d in Santa Rosaan dYanallpa . This agricultural type is found inman y otherplace san dfarmer s ofth etw ovillage sstudie dreporte dt ohav egrow n itbefor e (see chapter three). Jute was replaced by rice and corn asth e principal cashcro p around 1970. Growingjut e is disliked by farmers since the crop has to bepro ­ cessed while standing in the water, which is considered unhealthy. During the timeo fresearch , however,a controlle d pricean dth eavailabilit y ofban k loans for jute production persuaded many farmers to return tojut e growing. Jute is planted by broadcasting seeds on restinga about one month after a field dries. Onlyon eslash-weedin g isrequire dt oge trid o fth ecompetin gweeds . After sevent onin emonths ,ofte n whenth efield i salread yfloode d again,th ecro p is harvested. The harvestedjut e is left in the water for 18t o 20 days so that the plants rot. Then theplant s havet ob e washedan dbeate n thoroughly sotha tonl y thefibers, o rth esclerifie d coveringo fth evascula rbundles ,remain . Subsequently

63 thefibers ar e dried on poles, and rolled up in bundles. These bundles are very heavy, and difficult to transport. Laborrequirement sar eabou tthre eworkday s forbroadcastin gth ejut eseeds . The weeding takes about 10workdays . Harvesting, washing, drying and rolling thejut erequir eabou t4 5 to5 0workday s(Bergman , 1974; Pinedopers.com.) . Often farmers employ hired labor forjut e production.

Restinga vegetable fields Only very little information could be obtained about growing vegetables in a restinga field. During the first interview period only one farmer in Yanallpa reporteda resting avegetabl e field. The vegetablesgrow ni nthi sagricultura ltyp e includedtomatoes , severalpeppers ,cucumber s and Cyclantherapedata L. These arespecie sgrow nonl y inmediu mt o lowelevate dresting afields whic hhav emor e loamiersoil s thanfields o nth ehighe r restingado . Oncethes efields fal l drythe y are totally cleared and planted. Vegetable production continues until thefield i s flooded again. Since quite a few vegetables planted have cycles of only two to three months, several crops mayb e grown during oneseason . Weedingactivitie s ina vegetabl efield ar ecarrie d outa t least twicea week , if not more often. Different products are harvested at different times. Restinga vegetable gardens require very intensive monitoring and are mainly a cash crop venture (Table 4.1). Like the playa cowpea agricultural type, restinga vegetable fields aremainl ymanage db ywomen . Themai nrisks ar elos so fcrop st opest so r diseases, and low marketprices .

64 CHAPTER FIVE

VARIATION IN AGRICULTURAL STRATEGIES

INTRODUCTION

In chapter four it was discussed how ribereno farmers respond to the eco­ logically diverse landscape and economical diversemarke t opportunities by farm­ inga largenumbe r ofagricultura l types. Ribereno agricultural complexity is fur­ ther increased by the individual responses of farmers to such large numbers of available options. FewSant aRos aan dYanallp a farmers follow the sameagricul ­ tural strategy inmakin ga living . Furthermoreriberenos, displa ytremendou s flex­ ibility inadaptin gt one wcircumstances ,an dthe yconstantl ymodif y theirresourc e management strategies inrespons et oth echangin gcondition s ofth ephysica lan d economic environment. Ribereno agricultural strategies indicate howthes e farm­ ers adapt to the complexity of the environment in which they live. In chapter one agricultural strategies were defined as: agricultural choices in the allocation of limited resources (land, labor, capital) on which farmers rely to meet their subsistence needs, including needs for cash. Studies onagricultura l strategies essentially try to explain certain styles of agricultural behavior of ade ­ limited group of farmers. To be able to do so, it is assumed that farmers make individual agricultural choices, but also that there are certain explicit reasons for making such choices. It is assumed that specific external factors exercise selec­ tivepressur eupo nth edecision swhic hfarmer s take. Suchfactor s are,fo rinstance , access to land, labor and capital, access to markets,pric e levels for products,an d levels of risk and uncertainty carried by certain choices (Barlett 1980a,b, 1982; Cashdan 1990; Cancian 1989;Roseberr y 1989). Understanding the mechanisms of the influence of such factors on agricultural behavior of farmers, allows for explaining and understanding agricultural strategies. Riberenos makeagricultura l choices,fo r instance, whenthe ywan tt omak e a new field. In such situations a decision has to be made as to the agricultural

65 typeo fthi s new field. Thefactor s insuc ha situatio n which influence thischoic e areacces s to land, availability of labor resources, and availability ofcapital . The purpose ofth epresen t chapter ist o demonstrate the variation inagricultura l strat­ egies among farmers in Santa Rosa andYanallpa , and to attempt to explain this variation. Theagricultura l strategieso fth e inhabitants ofSant aRos aan dYanallp aar e evaluatedb yassessin g thecombinatio n ofagricultura l types which farmers hada t a specific moment. Achoic e for acertai n agricultural type impliesth e allocation of limited resources. A certain combination of agricultural types is the result of severalpreviou sdecision s onresourc eallocation . Henceth ecombinatio n ofagri ­ cultural types isa nappropriat ecriterio n forth e evaluation offarmers ' agricultural strategies. As in agricultural diversity, the variation in agricultural strategies among riberenos is remarkable. General agricultural strategies were quite different be­ tween Santa Rosaan dYanallpa ,principall y a result ofth e different farming envi­ ronments in each village. However, the variation in agricultural strategies within eachvillag e wasequall y large. Diversity, rathertha n similarity,appear s tob eth e rule. Furthermore, very few farmers maintain the sameagricultura l strategy over a longperiod . Someriberenos chang ethei r wayo f farming fundamentally within a time span of only asingl eyear . Suchchange s are responses to newopportuni ­ ties, or new conditions inth e ecological or economic environment.

FACTORS INFLUENCING RIBERENO AGRICULTURE

In order to conduct an analysis of agricultural strategies, a few principles shouldb eaccepted , whichar etru efo r small farmers ingeneral . First,w eassum e thatriberenos ar e rational economic actors who operate in fundamentally similar ways, but who experience a specific economic environment that limits their op­ tions (Barlett 1980b). Second, allriberenos tr y to match households needs with the resources available to them. Third, choices are madeb y single farming units withinth ecultura l and institutional environment inwhic hthe yar elocated . These

66 principlesca nb esummarize das :sinc eeac hfarmin g unit hasit sow nspecifi c needs and resources,an dsinc enatura lan dsocioeconomi c environmentsar evariable ,ag ­ ricultural strategies are individualistic and may vary over periods oftime . Forth ecas eo fribereno agricultura lbehavior ,si xfactor s canb edistinguishe d which directly influence agricultural choices, andtherefor e agricultural strategies. (1) Oneo fth ebasi c factors isth enee dt o find anadequat e balance betweenpro ­ ductionfo rhousehol dconsumptio n andcas hcropping . Riberenos,lik es o many othersmal l farmers, seekt o fulfil twogoal s inthei ragricultura l activities. Onth e one hand they want to produce their own food, while at the same time they de­ pend largely on agricultural production to provide cash needed to buygood san d services they cannot produce themselves. Althoughriberenos d o indeed wish to ensure a steady food supply (Chibnik 1990: 299) this does not diminish the im­ portance of the second goal. Chapter four shows that household production and cashcroppin geac hpartl ydepen do ndifferen t agriculturalactivities ,an dmos t farm­ ers havebot h kinds ofagricultura l types. It is,however , not unusual that farmers employ only agricultural types that produce either mainly for sale or mainly for household consumption. (2) The principal factor which influences the kind of commercial agricul­ tureriberenos wil l undertake, provided a sufficient ecological basis exists, is the opportunity tosel l products ata goo d profit. SantaRos aan dYanallp a farmers marketmos t ofthei ragricultura lproduct si nRequena . Atth etim eo fthi sresearc h rice could be sold at government buying centers. Although a government corn buyingcente rdi dexis ti nRequena ,i tseldo m operated,an dmos tfarmer s soldcor n to private companies or intermediaries. Jute could also be sold at government buying centers. Prices forrice, cor n andjut e approximately followed increasing inflation ratesbetwee n 1985an d 1987(Chibni k 1990). All otherproduct s soldb y farmers hadprice s established bymarke t forces. Prices for products which dono t havea controlle d price mayvar y tremendously within veryshor t periods (Padoch 1988, 1992; IPA 1977). Usually farmers sell such products to intermediaries in the market in Requena, oreve nt obuyer s inth e village. Theyals osometime sas k boat owners inth e village who aregoin g to Requena, to sell small quantities of

67 products for them. (3)Th e third factor which influences economic behavior ofribereiio farm ­ ers isth erisk involve d with agricultural activities. Risk is defined here asth e probability ofa los so rhazar d(Cashda n 1990). Inchapte rfou r therisk o f eacho f the fourteen agricultural typesha sbee n indicated. Althoughriberenos reduc e risk as mucha spossibl e theyd ono t avoidrisky choice s ifthi s mayb e rewarded with thepossibilit y ofa relativel yhig hpayoff . Infact , asdemonstrate dbelow, ,th emos t popular cash cropping activity in SantaRos a wasgrowin grice i na barreal , quite arisky optio n as explained in chapter four. Rather than avoiding or minimizing risk,riberenos see k"satisfactor y amounts of bothprofi t andrisk" (Chibni k 1990: 281). (4)A facto rwhic hinfluence sribereno agricultura lstrategie si sacces st oland . Land is often considered to be unlimited in Amazonia with its low population density. In chapter three it has been explained why this is not generally true. Certain varzea landforms are more appropriate for certain crops than others,an d access to these lands determines the agricultural type a farmer can opt for when deciding to make a new agricultural field. Furthermore, the specific location of land isals o important. Farmers maymaintai nrice fields i nbarreale s even onlo ­ cations outside ofth e villagearea . However, if new,stil luncultivate d terra firme land is located at a long distance from the village, than farmers may decide to slasha terr afirme fores t gardenwhil ei t isstil lyoun gan dconver t it intoa chacra . Alternatively they may makea chacra in the floodplain. (5)Th enex t important factor infarmer s agricultural decision-making isth e availabilityo flabor . Amongriberenos, bot h menan dwome nengag e inagricul ­ ture. Children already participate at an early age in household and farm work, but, in most analyses offamil y labor, only membersfifteen year s old orolde rar e considered full participants inth eagricultura l activities. Ribereno farmers gener­ allyus ethei r laborpoo lt o expand agricultural activities,bu t excessive laborma y also be allocated to other, non-agricultural production. Moreover, farming units maydiffe r significantly inth eamoun to fwor kthe yaccomplis hwit ha simila rlabo r pool (Barlett 1982).

68 (6) Directly related to theavailabilit y of labor isth e influence of accesst o capital onfarmin g strategies. Riberefio farmers obtain extra-family laborb yhold ­ ingfestiv e laborparties ,o rparticipatin g inlabo rgroup s (Chibnik& d eJon g 1989; deJon g 1987). Forsom especifi c labor-demanding occasionsfarmer s havet ohir e extra-family labor,and ,durin gth etim eo ffield work ,the yofte n relyo nban kloan s to obtain the cash needed to pay workers (Chibnik 1990). Criteria for eligibility for loansfro m theAgricultura l Bankchange d very muchbetwee n 1985an d 1989.

DATASET S USEDAN DMETHO D OF DATA-ANALYSIS

Both quantitative and qualitative methods can be used to analyze how the above factors influence agricultural strategies. If decision-making processes are the focus, participant observation and ethno-scientific interviews aremor e impor­ tant (Chibnik 1980). Such methods allow for an evaluation ofspecifi c alternative choices bybreakin g those down into a hierarchy ofsub-choice s which canb ere ­ latedt ocertai n key-factors. Thedecision-makin g processca nthe nb erepresente d withth eus eo fa decisio ntre eo rflowchart (e.g .Gladwi n& Murtaugh 1980; Barlett 1977; Chibnik & de Jong 1989). If the principal scientific interest ofth e research is to explain the variation inagricultura l strategies ofa certai n group of farmers, then analternativ e method ist o relate certain characteristics ofsubgroup s of farmers to certain kinds ofagri ­ cultural strategies. Thecharacteristic s which canb euse d tosubdivide ,o r stratify, the group of fanners are, for instance, the family labor pool, and the access to lando rcapital . Ifth e latterapproac h isfollowed , standardized questionnaires are often usedt o obtain quantitative data,an d theresult sar eanalyze d usingstatistica l procedures. However, a profound familiarity with the daily life of the farmers remains necessary to allow adequate explanations of the results. Although etho- scientific interviews and participant observation were methods also used in the research, the following analysis mainly follows the second approach. The variation in agricultural strategies among farmers in Santa Rosa and Yanallpawa sanalyze db ycomparin gth especifi c combination ofagricultura ltype s

69 within each farming unit. First, the frequency of each of the agricultural types reported by farmers is analyzed. Thisprovide d a general indication ofth e differ­ ences in agricultural strategies between thetw o villages. Second, the numbero f agricultural types reported by individual farming units were compared. Accord­ ingt o need, potential, and preference, eachfarmin g unit hasa specific numbero f agricultural types. This ishighl yindicativ e ofagricultura l strategies ofindividua l farmers. Third,th e discussion ofchange s inth enumbe r ofagricultura l typesan d combination ofagricultura l typesbetwee n 1985/1986an d 1989,reporte db yindi ­ vidual farmers, demonstrates changes in agricultural strategies during this three yearperiod . Theanswer sfrom th e general surveysconducte d in 1985 in Santa Rosa,i n 1986 inYanallpa , and in 1989 again in both villages were used as the principal data set for the discussion inthi s chapter. The farmers within each village were subdivided (statistically stratified) into groups according to certain differences in agricultural strategies, or combination of agricultural types. Next it was calcu­ lated whethero rno tth eaverage s ofa numbe ro fsocioeconomi c characteristicso f thewhol egrou po ffarmer s differed significantly betweenthes esubgroups . Ifthi s wasth ecase , it implied that the specific variable probably wasa factor of impor­ tance in differentiating the agricultural strategy in question. At-tes t wasuse d for this procedure. Differences inagricultura lstrategie swhic hwer eteste dagains tsocioeconomi c characteristics were, for instance, whether or not farmers reported practicing one of the principal agricultural types. A second stratification criterion was made betweenfarmer s whoha dthre eo rles sagricultura ltypes ,an dthos ewh oha dmor e than three. Means were calcultated for variables like: the number of household members,o rmember s whower e 15year sol d orolder ,th enumbe r ofyear s farm­ ersha dlive di nth evillage ,o rho wlon gfarmer s hadbee nusin gban kloans . Means are reported when they were significantly different at a one tailed t-test signifi­ cance level of 0.05 or less. Together with the means the one-tailed t-testprob­ ability isreporte d as@ value i.e.th echanc etha tth enotice d difference inmean s wouldoccu runde rth esituatio n whenthe yare ,i nfact , notdifferent . Inth eanaly -

70 sis ofdata ,an d its subsequent explanation, the results ofth e extensive interviews, and findings from participant observations, are also used. In Santa Rosa members of4 6 farming units were interviewed in 1985, and members of4 7 farming units in 1989. Out of the 47 farmers surveyed in 1989, ten had not been interviewed in 1985;som e were new immigrants, others newly established farmers. Out ofth e4 6 farmers who were interviewed in 1985,seve n hadlef t thevillage ,an dtw ofarmer s couldno tb elocate d in 1989. Thereremaine d only 37farmer s forwho mth eagricultura l strategies forbot h 1985an d 1989 could becompared . InYanallp a 50 farmers were interviewed in 1986,an d 68 in 1989. Of the 68 farmers surveyed in 1989, 19 had not been interviewed before. Nine farmers who lived inYanallp a in 1986 had left the village by 1989. For only4 1 farmers could the agricultural strategies becompare d both in 1986 and in 1989.

RESULTS

Frequencies of agricultural types in 1985/1986 Asshow n inFigur e 5.1,Sant a Rosa farmers managed 12agricultura l types in 1985. The terra firme agroforestry field and the terra firme chacra were the most common ones, found among 32 (70%) and 29 (63%) farmers respectively. The high frequency of these two low-risk, household-oriented agricultural types sustainsChibnik' s (1990)observatio ntha triberenos alway str yt oassur ethei r food supplyfirst, regardles s ofwhethe rthe ytak ehighe rrisks i ncas hcropping . Atota l of 10 fanners (22%) had a terra firme chacra together with a restinga chacra. However, a surprising 11Sant a Rosa farmers (24%) did not manage aterr a firme chacra, buta resting a chacrainstead . These farmers either lived inth e floodplain, ortemporaril y locatedalmos tal lo fthei ragricultura l activitiesthere . Farmerswh o lived on the terra firmehillock , and who had no terrafirme chacr a in 1985,al l made such anagricultura lfield betwee n 1985an d 1989. Only six farmers (13%) inSant aRos aha d neithera terr afirme chacra , nora resting achacra . Theirhouse ­ holds had fewer members (3.9 versus 6.9; @= 0.018) as well as fewer members

71 barreal ree

aguajal rice

rest ree

terfir rice

terfi chacra

rest chacra Q. *-< rest plantai

ra terfir plan

rest agrofor O 'i— O) terfir agrof < rest corn

playa co.vpea

rest jute

rest /egetab

O 20 40 60 80 100

Percentage of farmers Figure 5.1: Percentage of farmers in Santa Rosa who reported to farm the listed agricultural types in 1985 and 1989. over 15 years old (2.1versu s 3.3;@ = 0.009). Several ofthe m also lived outside the village most of the time, had only recently left their family to become inde­ pendent farmers, or were engaged in other productive activities outside agricul­ ture. The high proportion of the 39 farmers (85%) in Santa Rosa managing at least onerice agricultura l types reflects the importance ofric ea sa cas h crop,an d

72 probably also asa source of cash credit, during the time ofresearc h (Chibnik 1990). However, only 27 farmers (59%) had a barreal rice field. Both in 1985 and 1989 farmers who possessed a usufruct certificate for a barreal had been living in the village longer than those who did not have such a certificate (1 8year s versus 11.4 years in 1985; @ = 0.122, and 18.8 years versus 8.6 years in 1989 @ = 0.001). Barreales are the most favored cash cropping sites, and access to this landform is only available to a limited number of farmers because it is limited in area. In order to gain access to barreales, farmers have to buy the right of usufruct from another farmer, or to occupy a newly appearing barreal site, and it can take years before an opportunity to do so occurs. Not all farmers had access to a barreal and therefore some grew rice in a restinga or aguajal field. However, some farmers who had barreal rice fields, also had one or more other types of rice agricultural types. Of the 20 farmers (44%) who had a restinga rice field, nine also had a barreal, and of the seven farmers (15%) who had an aguajal rice field two also had a barreal rice field. Generally, only young farmers managed an aguajal ricefield bu t had nobarrea l ricefield bot h in 1985 (age 34.2 years versus age 41.0; @= 0.123) and in 1989 (age 26.8 versus age 49.2; @ = 0.000). Since for producing rice an aguajal is the second best al­ ternative to growing it in a barreal, younger farmers who did not yet have access to a barreal would prefer this alternative. The fact that only seven farmers man­ aged an aguajal rice field in 1985,whil e in 1989 a total of 13di d so was a result of the extremely low flood in 1985 which prevented five farmers from using their aguajal fields that year. Only four farmers (9%) produced rice in a terra firme field, and two of those had a barreal rice field at the same time. The mostnotabl e difference ingenera l farming strategies between Santa Rosa and Yanallpa was the absence of any terra firme agricultural types in the latter village, a result of the difficult access to terra firme. Some Yanallpa farmers re­ ported that they had owned fields on the Santa Rosa terra firme hillock around 1970, but had abandoned them because of the long traveling time needed to reach those fields. Yanallpa farmers managed only eight agricultural types in 1986. The most common agricultural type was the restinga chacra, which was in use by 44

73 of the farmers (88%, Figure 5.2). Although this number is much higher than the 29 Santa Rosa farmers (63%) who managed a terra firme chacra, it is not very different from the 40 farmers (87%) in Santa Rosa who had either a terra firme, or a restinga chacra, or both. Yanallpa farmers, like those in Santa Rosa, subsist onthei r own food production. Those who had neither aresting a chacra nor a plan­ tain field in 1989 lived, most of the time,outsid e the village, or were single-mem­ ber farming units. Others had recently become independent farmers, were new residents, or were farmers who obtained sufficient income from cash cropping or other work to buy the staples they needed. Agroforestry fields, on the other hand, were much less common inYanallp a (34%) than in Santa Rosa (70%). The 1989 data showed that heads of farming units in Yanallpa who had a restinga agroforestry field, tended to be older (50.1 versus 37.7; @ = 0.013) and had lived longer in the village (28.3 versus 11.1, @ = 0.003). Since there is less of an ecological need to establish agroforestry fields

barreal rice rest rice

O) rest chacra Q. rest plantai TO rest agrofor

rest corn ^^^^^^^^^^^^^^^^^^ ~3 playa cowp&a 1986 < rest jute rest vegetab ^1989 J L 0 20 40 60 80 100

Percentage offamer s Figure 5.2: Percentage of farmers in Yanallpa who reported to farm the listed agri­ cultural types in 1985 and 1989.

74 inth e floodplain than onterr afirme, fewe r farmers arewillin gt o dedicate landt o treeproduction . Perhaps this isbecaus e landpressur e has increased overth e last decades and only older fanners, who had had thesefields longe r still maintained agroforestryfields. Thever ylo wnumbe ro ffive Yanallp afarmer s (10%)wh oha dresting aplan ­ tainfields in 1986doe sno treflec t therelativ eimportanc eo fthi sagricultura ltype . Almost all theplantai nfields wer e destroyed that yearbecaus e ofth e high flood, and farmers were replanting their high levee fields. Many farmers, at that time, used the sites,usuall y planted to plantainsfirst, fo r otheragricultura l types,whil e they were restoring the plantain crop. This explains why there were 18 farmers (36%) who had a restingarice field i n 1986, while in 1989 only one did so. In 1989, the restinga plantainfield ha d become the most common agricultural type inYanallpa , where it was found in 63 ofth e farming units(93%) . Another noticeable difference, between the two villages studied, wasp th e emphasis ondifferen t cashcrops . In 1985/1986,Yanallpa , 39farmer s (78%)gre w corn inrestinga ,whil eonl y3 4farmer s (64%)ha don eo rmor etype s ofric e fields. Anotherconspicuou sdifferenc e betweenth etw ovillage swa sth enumbe ro ffarmer s whogre w cowpea onplay a land. Atota l of 29Yanallp a farmers (58%) had such afield, a s contrasted with only seven Santa Rosa farmers (15%). The access to land, i.e. the large playa in front ofYanallpa , explains this difference.

Changes in utilization of agricultural types In 1989Sant aRos afarmer s manageda tota lo f 12 different agriculturaltypes , the samenumbe ra s in 1985. Nonetheless important changes had occurred inth e kind of agricultural types which farmers preferred. The most striking change in agricultural strategies in Santa Rosa wasth e decline in rice production. Only2 8 ofth e Santa Rosa farmers (60%) plantedrice i n 1989, a decrease of 25.2%sinc e 1985 (Figure 5.2). None of the farmers grew rice in terra firme in 1989 while only 6gre wresting a rice. The decrease inth e number of farmers growingrice i n barreales from 27t o 21 wascause d byth ereductio n ofth etota l areao fbarrea li n the village.

75 Other cash crop agricultural types increased in importance in Santa Rosa between 1985 and 1989. More farmers (11, 23%) grew corn in a restinga field, and Santa Rosa farmers practiced two new agricultural types: restingajut e fields, and restinga vegetable fields. In the latter, mainly beans were grown. Both agri­ cultural types were found in 6 farming units (13%). Since none of the farmers

0. 1. 2. 3. 4. 5. Numbero fagricultura l types

Figure 5.3 Distribution of farming units according to farm size, expressed in num­ ber of agricultural types per unit, in Santa Rosa and Yanallpa, 1985/1986.

76 had both of the new agricultural types, 12 farmers (26%) chose a new cash crop activity in 1989. The number of farmers who grew plantain in restinga fields in­ creased from 6 to 12. InYanallp a all farmers managed nine agricultural types in 1989, compared to only eight in 1986. A decline in rice production, similar to that observed in Santa Rosa, occurred in this village. As mentioned above, the number of farmers who managed a restinga rice field especially decreased. The number of farmers growing rice in barreales increased from 20 to 25. The absolute number of farm­ ers who managed restinga corn fields and restinga chacra fields increased between 1986 and 1989, but in both years they represented around 80% of farmers in the village. The most dramatic change in the agricultural behavior ofYanallp a farm­ ers, however, was the increase in farmers growing plantains, as explained above.

Magnitudes of farming enterprises In Santa Rosa, in 1985,th e average number of agricultural types per farm­ ing units was 3.8. A total of 32 (70%) farmers managed between three and five agricultural types. The highest number of agricultural types managed by a single farming units was seven, reported by two farmers. The lowest number reported, also by two farmers, was one (Figure 5.3). In 1989 the average number of agri­ cultural types per farming unit had decreased from 3.8 to 3.6. That year, more farmers managed only three, instead of four agricultural types, and fewer farmers managed six or seven agricultural types than four years earlier (Figure 5.4). Although a comparable number of Santa Rosa farmers managed one or two agricultural types in 1985 and in 1989, they were mostly different farming units. In 1985, farmers who had two or less agricultural types had taken bank loans for an average of 2.4 years, while farmers managing three or more agricultural types had an average of 7.4 years of bank loans (@ = 0.017). Bank loans are mainly used to increase the labor capacity by paying hired workers, and the number of agricultural types practiced by one farmer therefore appears to be directly related to the amount of labor available. All farmers who managed fewer than three ag­ ricultural types in 1985,i n 1989 either had left the village, or had increased their

77 (A E ra

o ra *-» c a> u a.

Number of agricultural types

Figure 5.4 Distribution of farming units according to farm size, expressed in num­ ber of agricultural types per unit, in Santa Rosa and Yanallpa, 1989. number of agricultural types. One farmer whodecrease d the number of agricul­ tural types from three to two, was reported to have had a serious illness in his family. In 1985som e stratification couldb enotice d among fanners who managed five or more agricultural types. They had an average of 9.5year s of using bank loans, whereas farmers who managed four or less agricultural types only had an

78 average of 5.2 years of bank loans (@ = 0.026). This again can be explained in terms of farmers who have bank loans being capable of hiring extra-family labor, and therefore able to increase the number of agricultural types in their enterprise. In most cases the farmers with five or more agricultural types had a lower number in 1989. In some of these farming units household members had left, or they now focused on other non-agricultural work. In Yanallpa, in 1986 farming units managed an average of 3.4 agricultural types, a number slightly lower than that in Santa Rosa. Only two farmers had six agricultural types, and two farmers had one agricultural type (Figure 5.3). In 1989, however, a shift towards more complex combinations could benotice d (Figure 5.4). The average of agricultural types per farming unit increased from 3.4 to 3.9, a number even higher than in Santa Rosa in 1985. The number of farmers manag­ ing six agricultural types increased from only two (4%) in 1986 to seven in 1989 (10%). The number of farmers managing seven agricultural types increased from zero to four (6%) in the same period. The lower average number of agricultural types per farming unit in 1986 is a result of the flooding of the whole Yanallpa area whichha d destroyed many agricultural fields that year. Forinstance , few farm­ ers reported restinga plantain fields in 1986. One newYanallp a inhabitant in 1989 reported to have no agricultural fields at all, since he had only very recently ar­ rived in the village. In Yanallpa in 1986 farmers with two or less agricultural types had fewer fanning unit members (5.3 against 7.2; @= 0.028), and fewer members who were fifteen years or older (2.1 versus 3.3; @ = 0.004). In 1989 farmers with two or less agricultural types still had both fewer farming unit members (2.8 versus 5.8; @= 0.000), and fewer membersfifteen o rolde r (1.6 versus 2.9; @= 0.001) . Again the complexity of farms appeared tob edirectl y related to the availability ofhouse ­ hold labor. On the other hand, all the farmers in Yanallpa who managed one or two agricultural types in 1986 increased the diversity of their enterprise in the fol­ lowing years. Farming units with fewer household members may have less than the average number of agricultural types, but only temporarily. Farmers in Yanallpa who had five or more agricultural types in 1986 had

79 been working longer using bank loans (5.8 years versus 2.2 years; @ = 0.025). It is remarkable that several of the farmers in Yanallpa who managed six or seven agricultural types in 1989 had managed only one or two in 1986. Except for one case, they all had inherited fields from aparen t who had left the village for Iquitos. The most exemplary indication of the complexity of agricultural strategies among Santa Rosa andYanallp a farmers is the variation observed in combinations of agricultural types. In Santa Rosa in 1985 a total of 39 different combinations of agricultural types occurred. In 1989 among the4 7 farmers 43 different agricul­ tural type combinations were found. InYanallpa , in 1986,amon g the 50Yanallp a farmers 29differen t combinations of agricultural typeswer e observed, and in 1989 this number had increased to 44 combinations among 68 farmers.

Changes in agricultural strategies Between 1985 and 1989 Santa Rosa farmers increased or decreased the number of agricultural types they managed by between zero and three types. The most common change in number of agricultural types was three. Farmers who had changed most were those who had either very few, or numerous agricultural types. Again this indicates that farmers may have more or less than the average number, but after some time will return to those average numbers. In Yanallpa, farmers increased or decreased the number of agricultural types they managed by between zero and four types. The largest groups of farmers, however, had the same number of agricultural types both in 1986 and 1989 (Figure 5.5). The changes in the specific combination of agricultural types of individual farmers between 1986 and 1989 were much larger in the two villages. Farmers, both in Santa Rosa andYanallpa , sometimes replaced upt othre e agricultural types withthre e othersbetwee n 1985/1986 and 1989. No stratification could be detected among farmers according to the changes in combinations of agricultural types. However, the following two examples illustrate what could motivate farmers to change their agricultural strategies. One farmer in Santa Rosa had replaced four out ofth efive agricultura l types he managed in 1985 with three others by 1989. He had abandoned his barreal

80 rice field since this was not suitable for rice growing anymore. In 1989 he had a cowpea field on the same location. The corn he grew in a restinga field had given him unsatisfactory returns, and in 1989 he made a new restinga jute field. Fur­ thermore this farmer had been sick for a long period in 1988 and had not been able to maintain his terra firme chacra. This field was overgrown with secondary

Change in total number of agricultural types Figure 5.5 Distribution offarmin g units accordingt ochange s infar msiz ebetwee n 1985/1986 and 1989, expressed as increase or decrease in number of agricultural types per unit.

81 vegetation. The farmers had also lost a small restinga plantain field to the flood in 1986, which had not been replanted. OneYanallp a farmer managedfive agricultura l types during both interviews, but had replaced three that she managed in 1986 with three others by 1989. In 1986 she lived and worked together with her husband, who left the village in 1987. The woman continued to farm on her own with only the help of her children and using extra-family labor. To finance hiring laborers, or organizing labor parties, she started selling liquor and groceries at home. She abandoned the barreal field which had become too sandy, and her cowpea field because it was located to far from her house. She had started growing vegetables on the restinga field, which in 1986 had been used for rice growing, and began to grow corn and plantain in restinga fields. She maintained a restinga chacra, and an agroforestry field.

DISCUSSION

Preferred farming strategies The distribution of numbers of agricultural types per farming unit in Santa Rosa suggests that an inhabitant of this village preferably farms simultaneously three to five agricultural types. The frequencies of agricultural types indicate that a farmer is most likely to manage a terra firme chacra, a terra firme agroforestry field, and a barreal rice field. In Yanallpa a preferred agricultural strategy con­ sisted ofthre et ofive agricultura l types,bu twoul d include aresting a chacra, restinga plantain field, and restinga corn field. In Santa Rosa production for household consumption stems mainly from the terra firme chacra. InYanallpa , both restinga chacra and plantain fields provide food. The latter combination increases the se­ curity of the staple supply. The differences in barreal rice versus restinga corn production, as the main cash crop activity, is a result of ecological differences in the two villages. Under certain circumstances riberenos may have either fewer, or more agricultural types, than the preferred three to five agricultural types. The number of agricultural types managed by farmers may be influenced

82 by temporarily ecological factors, as was the case inYanallp a 1986 when an ex­ tremely high flood had destroyed all plantain fields. The principal factor which determines thenumbe r ofagricultura l types of a single farming unit isth e amount of labor available. Especially inYanallpa , a direct relation appeared between the labor pool of a farming unit and the number of agricultural types that were man­ aged. Also, those farmers who had made longer use of bank loans appeared to manage more agricultural types. This effect can also be explained as the result of labor availability, since such farmers are more accustomed to hiring workers. In most cases farmers who are permanent residents in the two villages only maintain unusually high or low numbers of agricultural types for a short period of time, and soon change back to more usual numbers. Several other factors explain why farmers managed fewer than three agri­ cultural types. Riberenos are very mobile (Padoch 1986)an d in any community a number ofrecen t immigrants are always found, who are still establishing their farm enterprise. In addition, a number of farmers are often about to leave the village, and therefore already have abandoned certain agricultural types they managed before. Young villagers may have recently become independent farmers, or they may be only single member farming units,an d thus able to manage fewer agricul­ tural types. Other farmers may be engaged in other productive activities, or live for part of the year outside the village. Sometimes farmers specialize in only one or two agricultural types, for instance the farmers who managed large areas of barreal rice fields. The main reasons why farmers want to change the number of agricultural types they manage at a certain time is because they have either too few or too many. Farmers may be overly ambitious at certain moments and try to increase their production beyond what the farming unit can handle. Sometimes a high number of agricultural types may reflect a transition period in which a farmer is trying out new agricultural types, whereas his old ones have not yet been com­ pletely abandoned.

83 Household production Whether or not riberenos will follow the strategy with the lowest risk to produce their own food, depends, more than anything on a personal willingness to take risk. Most Santa Rosa farmers relied on a terra firme chacra for their basic food supply during the two survey periods. A number of farmers, however, ob­ tained their staple food from aresting a chacra. Thismean stha tthe yaccepte d higher risk in return for easier access to other restinga fields and to the river. The loca­ tion of terra firme lands and arable varzea land within such a close distance is rather unusual in the Peruvian Amazon. In many other villages riberenos have no other choice than to locate all their agricultural activities in the floodplain, includ­ ing their food production. ManyYanallp a farmers obtain their principal staple food from both a restinga chacra and a restinga plantain field. The high number of farmers who had both of these agricultural types in 1989 indicates that they do not substitute, but rather complement each other. Combining a restinga chacra and plantain field increases food security, but also allowsYanallp a farmers to increase cash returns. A restinga chacra usually yields manioc for about four to six months per year. A plantain field yields plantains during most of the year, and surplus production can be sold. Plantain fields, however, will certainly be destroyed ifthe y are flooded, and farm­ ers consider it risky to only rely on this agricultural type for their food supply.

Rice versus corn cash cropping The preference of rice as a cash crop among Santa Rosa farmers versus a much stronger emphasis on corn cash cropping in Yanallpa is a result of the dif­ ferent ecological conditions in the two villages. Santa Rosa farmers have access to a large area of barreal, and between 1985 and 1989 concentrated most of their cash cropping activities on this landform. Santa Rosa farmers also have access to the aguajal, a biotope which is suited to rice production. Furthermore, much of the lower restinga in this village was, until 1986, still covered with high forest, and hence new fields could still be opened there for rice production. Barreales are,withou t doubt, the most preferred rice production agricultural

84 sites. Since these lands are scarce, new farmers have to begin by growing rice on different land forms, or by producing other cash crops until they have the oppor­ tunity to buy an existing barreal site or to occupy a new one. If a farmer has sufficient labor available, this labor may be invested in a second or a third cash crop agricultural type. Such a choice is also taken to reduce the risk of crop fail­ ure from one cash crop agricultural type. In the close vicinity ofYanallpa , a much smaller area of barreal terrain than in Santa Rosa emerges every year. While barreal rice production is important in Yanallpa, it is an additional cash crop activity, rather than the main source of monetary income. Much of the lower restinga terrains inYanallp a are located on thebac k slope ofth eprincipa l levee (see chapter three) and already had been used for many years. Weed competition makes these sites inappropriate for rice grow­ ing. A few areas of restinga on the left river bank still were covered by high for­ ests, which were being converted by some farmers to rice production. The larger part of theYanallp a lower restingas, however, is better suited to growing corn. Thedeclin e inric eproductio n inSant a Rosa andYanallp a wasmostl y aresul t of a fewer farmers growing rice in restinga fields. The conditions for growing rice in restinga fields were good in 1985/1986 in the two villages. In Santa Rosa the shift of the Ucayali river had resulted in increased sedimentation of the lower restinga area on the right bank, downstream from the village center. Santa Rosa farmers had increasingly been using this lower restinga land for rice production. By 1989 most new restinga sites in Santa Rosa had been deforested. In Yanallpa the high flood of 1986 caused favorable rice growing conditions on many higher restinga sites. Because no recent flooding had occurred which had rejuvenated the medium and higher elevated restinga sites in Yanallpa, rice was less planted on these lands. A second explanation for the decline in restinga rice growing wasth e chang­ ing services of the government agencies in charge of regulating rice production. Obtaining bank loans from the Peruvian Agrarian Bank for restinga rice produc­ tion used to be difficult, especially for new applicants, and the administrative pro­ cesses were tedious and time consuming (Chibnik 1990). After the APRA gov-

85 emment came into power in 1985, they improved both credit policies, and ser­ vices of the agencies in charge of buying rice from farmers, which increased rice production in restinga fields. However, three years later, when the economic poli­ cies of the new government had largely failed, services of the government agen­ cies again worsened. Farmers sometimes had to wait for several months to get paid for products they had delivered. These declining services for rice producers depressed rice production inresting a fields, since profits from this agricultural type are smaller than from barreal or aguajal rice fields. Many of the Santa Rosa farmers who had grown abarrea l in 1985,bu t were not doing so in 1989, grew a cowpea field in that year, mainly on the sites where thebarrea l hadbee n located before. This not only provided an alternative income, but also allowed the farmer to maintain the user's right over the site, in the hope that thequalit y will improve again in following years. Since inYanallp a barreales appeared in different locations in 1989 than those of 1986, about the same num­ ber of farmers managed this agricultural type in both years, but the group con­ sisted of different farmers in 1989. InYanallpa , in 1986,th e playa cowpea fields was themos t popular agricul­ tural type managed as a second cash crops. The drop in number of farmers, from 58% to 35% ,who had a playa cowpea field inYanallp a between 1986 and 1989 can largely be explained byth e increase in farmers having plantain fields. In 1986, farmers faced reduced incomes since they could not sell any plantains that year. Although the net returns for cowpea are less than for other cash crops, there was no additional land available to grow rice or corn. Hence farmers chose to grow cowpea to compensate for the losses in income from plantain production as a re­ sult of the flood in 1986. In 1989 the plantain fields were restored, and many farmers gaveu pthei r playa cowpea fields. Furthermore vegetableproductio n also increased, probably in response to a higher demand in Requena.

STRATEGIC FARMING IN THE FLOODPLAIN

According to Gladwin & Murtaugh (1980) studies of agricultural decision-

86 makingo fsmal lfarmer s should finally lead tooth epredictio n ofthei r future eco­ nomic behavior. The previous discussion did not reach this goal. Although this may be unfortunate, because such an understanding would be useful, it teaches something very important about ribereno agriculture. In the previous analysis it waspossibl e todemonstrat eth ecomplexit y ofriberen o farming, andth ecomple x wayi nwhic h ecological,economic ,an d social factors interact inth emechanism s leadingt oagricultura l choiceso friberen o farmers. Muchwor kstil lremain st ob e done before a sufficient understanding of ribereno resource management can be achieved. This understanding is necessary if any well-founded extension or de­ velopment action ist o be carried out in the floodplain. Thepreviou s discussion probably, moretha n anything else,ha sshow ntha t riberenoagricultur e isver y individualistic. Thediversit yan dvariatio n whichha s beendiscusse dabov eindicat etha tther ear eman yway st omak ea livin gfo rriberen o farmers. They also indicate that riberenos have many possibilities for resource management which probably have not yet even been explored. Complexity of riberenoresourc e management doesno t completely defy systematization. Strati­ fication of groupso f farmers asfoun d in SantaRos a andYanallp a isno t selfevi ­ dent, and it is difficult to sort out how single factors influence ribereno agricul­ tural decision making. Characteristics used for stratification inothe r studies like size of holdings (Barlett 1977), or number of farming unit members (Chayanov 1986;Chibni k 1980),onl y partly explain ribereno agricultural strategies. Rather than focussing onsingl efactors , interaction amongsevera l factors shouldb euse d toexplai n specific strategies. The individualistic approach of ribereno farmers canb e explained as are ­ sponset oth e ecological complexity of agricultural sites andth edynamic s ofth e floodplain. Thevariou s agricultural options inthi s environment, which are are ­ sult ofth evariatio n inappropriatenes s of land for agriculture purposes, involvea total different amount of labor, scheduling of activities,risk an d returns. There­ fore, farmers very carefully choose the strategies which best meet their specific possibilities and needs. As argued before, access to certain types of land is an important factor which influences the direction of agricultural strategies adopted

87 by individual farmers. The amount of labor farmers have available also signifi­ cantly influences how agricultural enterprises are expanded. Decisions on new agricultural activities are also influenced by which agricultural activities are al­ ready taking place. Given the complexity of the environment, agricultural decisions are very site specific. This requires detailed monitoring of agricultural fields. Single agricul­ tural fields mostly cover a limited area, sotha t their ecological conditions are rather uniform, and they can be planted to a single crop or crop combination. Monitor­ ing of such agricultural fields and making agricultural decisions indeed is best car­ ried out by single farmers. Independent farmers are more capable of doing both than any other organized group. Both individual farming and variable agricultural strategies are therefore adaptations to the dynamic floodplain environment. The agricultural diversity and variation inagricultura l strategies found among riberenos has consequences for research methodologies. Individual cases never give an adequate picture of the whole group (Padoch & de Jong 1987). Further­ more, a study of a single community, common in most ethnographic or peasant studies, is not a sufficiently large enough sample for larger groups like the whole ribereno population of Peru. General agricultural strategies in Santa Rosa and Yanallpa are sufficiently different from each other to justify, and even demand, research in at least two villages. Although, as was argued in chapter two, the two communities represent two village types from the PeruvianAmazon , it isals oclea r that numerous variables vary considerably in other places, e.g. market opportuni­ ties, local history, or specific ecological conditions. The two cases discussed above therefore cannot be considered representative for the whole PeruvianAmazon , and elsewhere ribereno resource management is most likely different. According to Hiraoka (1985b) increased involvement with regional and na­ tional economies, especially through greaterpossibilitie s for marketing agricultural and forest products, has significantly changed ribereno livelihood since the 1950s. Indeed, in most villages few people still accept the isolated lifestyle of several decades ago. Improved schooling, increased material goods, and leisure trips to Iquitos have become normal for many riberenos. On the other hand, riberenos have no choice but to share in the economic malaise which Peru has endured for several years. Although the long term changes which Hiraoka (1985b) reported may not have become evident from the data presented above, it ismos t likely that some tendencies observed are linked to such larger trends and will continue for longer periods. The economic policy towards a free-market economy, which the Peruvian government now advocates, has resulted in the abolishment of a guaranteed price for rice. Most likely this will decrease the importance of rice as a cash crop, a trend which was already evident in the two villages in 1991. However, the pre­ ceding discussion has sufficiently demonstrated that ribereno farmers are accus­ tomed to changes. This means that they will continue to adapt to new circum­ stancesb ymodifyin g their agricultural practices. Thesemodification s will be long- term changes, hidden below the short-term and visible variation in agricultural practices as described in the present chapter.

89 CHAPTER SIX

FOREST GARDENING IN SANTA ROSA

(A slightly revised version of this chapter has been accepted for publication in Agroforestry Systems)

INTRODUCTION

Nearly two decades have passed since agroforestry became an established scientific sub-discipline among the agricultural sciences (Nair 1990;Kin g 1989). Although agroforestry was initially hailed asa landus e option which could bring sustainable development to regions where other agricultural methods had failed (Nair 1990),i tha sno tye tfulfille d thisexpectation . Exampleso fsuccessfu l imple­ mentation ofne wagroforestr y practicesremai n few(e.g .Buc k 1990;Scher r 1990; Bishop 1983; Boonkirdetal . 1984; Bourke 1985;MacDicke n 1990; Getahun1990) . Much hopetoda y is founded on developing indigenous agroforestry, and callsar e made for continued and profound research on these practices (Budd et al. 1990; MacDicken 1990;Rochelea u 1987; Nair 1990). Manyscientist san d development workers advocate that designing improved systems should be based on existing agroforestry practices and make use ofpeople' s knowledge and experience (Beer etal . 1990;Lagema n& Heuveldorp 1983;Duchhar te tal . 1990;Buc k 1990; Scherr 1990). Somescientist s expect that important advances inagroforestr y inth enex t 10 to2 0year sma ycom efrom developmen t experiences,rathe rtha nfrom researc h (Beer et al. 1990). However, westil l understand too little about how indigenous agroforestry systems function (Nair 1990). According to MacDicken &Vergar a (1990: 1),"th escienc eo fagroforestr y lagsfa rbehin dth ear to fexistin g agroforestry practices". Shifting cultivation, or swidden agriculture, in which crops and trees are deliberately combined, are among the oldest and most widespread forms of agroforestry (Raintree & Warner 1986; MacDicken 1990; Gholz 1987;Vergar a

90 1987). Many scientists suggest that shifting cultivation practices should be im­ proved applying agroforestry technologies (Raintree and Warner 1986;Ra o etal . 1991;Alle n 1985;Wiersu m 1983; 1985;Andriess e 1978). In most cases, how­ ever, littleattentio n hasbee npai dt oth eexistin gagroforestr y component oftradi ­ tional swidden agriculture (e.g. Raintree &Warne r 1986;Duboi s 1990). Padoch en deJon g (1987) showed that swidden-fallow agroforestry within a single region may vary as the result of factors like market access and otherag ­ ricultural activities of fanners. In this chapter the diversity of swidden-fallow agroforestry in a single village, Santa Rosa is discussed. In chapter four it was explained why forest gardens in Santa Rosa are to be considered as a separate agricultural type. Cultivation measures include planting, weeding, and harvest­ ing. The amount and frequencies of these activities often vary betweenfields, resulting in a variation of forest gardens differing in species composition, struc­ turean d amount ofproduc ewhic hca nb eharvested . Thisdiscussio n also demon­ stratestha t asingl eagricultura l typema ysho wconsiderabl e variation,an d further emphasizes the complexity ofribereno agriculture .

SWIDDEN-FALLOW AGROFORESTRY

Manipulationo fwood ysecondar yvegetatio no folde rswidden swa sreporte d inearl ywork so nnatura lresourc emanagemen t oftraditiona lswidde nfarmer s(e.g . Conklin 1957; Spencer 1966; Clarke 1971;Deneva n 1971;Harri s 1971; Gordon 1969, 1982). Research carried out largely in the last decade has shown that in­ deed many forms of shifting cultivation include active management of economi­ cally important tree crops in the fallow vegetation. Denevan and Padoch (1987) suggested that the term swidden-fallow agroforestry should be used to describe thesepractices . Ribereno swidden-fallow agroforestry differs in the resulting vegetation of the forest gardens, the amount of labor they receive, and the destination of the produce. While some ofth eagroforestr y systems developed byriberefio farmer s areprincipall ymarket-oriente d (Padoch etal . 1985;Hiraok a 1986), theyals opro -

91 videa mi xo fproduct suse di ndail ysubsistenc e (Denevan& Padoch 1987; Padoch & de Jong 1987, 1989). Market opportunities, individual needs, and other pro­ ductive activities of farmers are the most important socioeconomic factors which influence the variation inagroforestr y practices (Padoch &d eJon g 1987). Since thesefactor sar edifferen t inalmos tever ylocalit yi nth ePeruvia nAmazon ,n osingl e and unique ribereno agroforestry practice exists. Rather, there are a large varia­ tion of practices, which differ in degrees from each other. Swidden-fallow agroforestry, as practiced by most Amazonian farmers, is characterized by a change in the management pattern of a swidden after several years of farming. Fanners do not continue the intensive use of afield fo r more than two to four years because decreasing returns and increasing weed invasion make it less profitable than changing the field to a forest garden and making a newswidde n elsewhere(Deneva n 1971;Ewe l 1986;Hoekstr a 1987). Older fields continue to be economically important, and are not simply abandoned. Products may be harvested from planted or naturally occurring species. Besides harvest­ ing, weeding and planting continue to be carried out in olderfields (Deneva n& Padoch 1987;Deneva n et al. 1984; Irvine 1987). The change in the management pattern from aterr afirme swidden , be ita plantainfield, o r chacra, to a forest garden is anticipated by most riberenos long before it actually happens. Whileth efield i sstil l achacr a orplantai n field, farm­ ersplan t several tree species which willyiel d products tob e harvested after con­ version into a forest garden. Nonetheless, how the forest gardenfinally wil l be managed mayb e influenced bysevera l other factors, someo fwhic har e not atal l related to what trees were planted in the field. The structure of a forest garden varies not only fromfield t ofield, bu t also from yeart oyea r(Alcor n 1990: 143). Such structure changes wereshow n inchapte rfive t ob erelate dt o changes inth e agricultural strategies of individual farmers. Onceth eprincipa l crops ofa swidde nceas et ob eharvested ,th esubsequen t frequencyan dthoroughnes s ofweedin gals odiffer s betweenfields. Th eare aaroun d planted ortende d trees is cleared infrequently. Sometimes tree crops receive no attention atal l for severalyear swhe nthe y startyieldin g fruits. Oncesevera l fruit

92 trees start to produce,fields ofte n are slash-weeded again eitherpartiall y orcom ­ pletely. It is also not uncommon for farmers continue to slash-weed terra firme fields once ortwic e a year during the transition period to a forest garden. In order to assess diversity in swidden-fallow agroforestry in Santa Rosa, sevenfields wer esurveyed . Thecompositio n ofmanage dspecies ,productio nlev ­ els, and weedingpattern s ofeac ho fthes efields ar ecompare d below. First agen ­ eral description ofth e sevenfores t gardens willb eprovided , includingfield loca ­ tion, vegetation, and the land-use history as recorded from the current owners. Subsequentlyth ecompositio n ofmanage dplan t specieswil lb ecompare dbetwee n fields. Finally the intensity and frequency of weeding, and the yield levels are discussed.

SANTAROS AFORES TGARDEN S

Boerboom (1974)ha spointe d outth eproblem stha t occurwit hth estud yo f secondary succession oftropica l forests. Iti salmos t impossiblet ofind a numbe r ofplot s of secondary forests for which only aver y specific set of factors vary,s o that only the influence of those factors on the forest development can be evalu­ ated. More recently Van Rompay (1993) stated that homogeneous plots do not exist intropica l forests andtherefor e neitherrepetitio n ofexperiments , norrepeti ­ tion of observations ofth e same forest arepossible . Similarproblem soccurre ddurin gth estud yo fth emanagemen t ofSant aRos a forest gardens. Instead of attempting an homogenous selection,fields wer e sur­ veyedwhic h represented asbroa d avariatio n offores t gardens ascoul db e found. Fields therefore differ not only in the way they were managed, but also in their age,an d land-use history. This makes it moredifficul t to drawconclusion s from thefindings, sinc ether e areman y factors which canexplai n the variation. How­ ever,bein ga first assessmen t ofdiversit y inribereno agroforestr y ina singl ecom ­ munity it wasnecessar yt o describeth e largevariatio nfirst rathe rtha n attempting to standardize the sampling as much as possible. The following provides a de­ scription of each field.

93 Forestgarde nA wastw oyear s and 10month s old when it was surveyed. At the time of the surveys, fruit species and tended forest species were evenly distributed throughout the field. Originally the field wasplante d asa chacra ,an d the crops grown included manioc, plantains, sugarcane, peppers, and pineapple. Many fruit species hadbee nplante d at an early stage,bu t other species likepal m heart (Euterpeprecatorid), andtropica lceda r (Cedrela odorata), hadbee nplante d onlywhe nth efield wa stw oyear sold . Whenth efield wa ssurveyed , itwa sbein g weeded for the third time, and for thefirst tim e since the last manioc had been harvested. Forestgarde n Bwa sbetwee n 20an d 25year s old,accordin g toth e sono f the original farmer. It had a closed high tree vegetation. The originalfield ha d been slashed in primary forest, planted as a chacra, and managed as an orchard ever since. Most of the area of this forest garden was slash-weeded about twice per year, but different locations in different periods. Forest garden C was between 30 and 35 years old. The field is uniquely positioned at the foot ofth e hillock onwhic h SantaRos a is located, forming part of the village center itself. The main village street leads through this mixed or­ chard,an d thehous e ofth e owner is located insideth e field. Thefield wa smad e asa swidde ni nth eearl ysixtie swhe nth eorigina l founders ofSant aRos areturne d from theirtravel sfrom th eNap orive r(Padoc h& d eJon g 1990 andchapte rthree) . Just like forest garden B, most ofthi sfield i s kept clean ofunwante d vegetation by irregular, but almost continuous weeding. Forestgarde n Dwa sseve nyear s old. Thefield ha da welldevelope d tree vegetation but it wasn't as high norwer eth e diameters as large asthos e in forest gardens Ban d C. It was originally cut from aprimar y forest area,an dplante da s achacr awit hmanioc ,plantain san dfrui t trees. Thecurren tweedin g intensitywa s much lowertha n that of forest gardens A, B, and C. Patches ofth efield ha dre ­ cently been slash-weeded, but others were covered with high secondary forest. Forest garden E was estimated to be about ten years old. The vegetation wassimila rt o the one infores t garden C. The manwh omad eth efield ha ddied , andonl ya smal lpar t ofi t wasuse db ya differen t farmer. Shortlyafte r surveying

94 this field it was largely cut and a house was built on the site. This field had a similar spatial variation in weeded and non-weeded areas as forest gardenD . Forest garden F was sevenyear s old. It had aclose d secondary treeveg ­ etation with only a few species ofplante d trees. Thefield ha d been made intw o previousfores t gardens,whil eth eorigina lnatura lfores t vegetationha sbee nslashe d 13an d 14 years ago. This field, therefore, had been farmed as a chacra twice. Very little slash-weeding appeared to have taken place inthi s field. Forestgarde n Gwa s six years old. It was also covered with a densesec ­ ondarytre e vegetation, with very few planted tree crops. Thefield ha d originally' been made in aprimar y forest and had been managed as a chacra for two years. Forestgarde n G wasslashe dagai nan dconverte d intoa chacr ashortl yafte r itwa s surveyed. It appeared not to have received any significant recent slash-weeding.

RESULTS

Diversity of managed species Atota l of9 2manage d species (seechapte rtw o for definition) werepresen t inth e seventerr afirme fores t gardens (Table6.1) . Thenumbe ro f managedplan t species per single forest garden ranged from 14t o 48. Of the 92 managed spe­ cies,onl y4 7 weredomesticate d plant species,commo n inman yAmazonia nagri ­ cultural systems (Hiraoka 1986; Denevan & PadochT987; Calzada Benza 1980; Cavalcante 1976, 1978, 1980). Theothe r4 5specie s werenativ especie sfrom th e forest which were planted in these fields, or occurred spontaneously and were tended. In some casestree s ofth e original forest cover were spared while slash­ ing the field. Such managing of native forest species is common in many agroforestry systems(e.g .Okafo r & Fernandes 1987;Leuschne r& Khaleque1987 ; Fernandes et al. 1984; Michon et al. 1986;Anderso n et al. 1985; Posey 1985; Rod-eleau 1987). Onlyseve n species werepresen t inal lth e seven forest gardens (Table6.1) , all cultivated fruit trees. They were Rollinia sp., Pouteria caimito, Inga edulis,

95 Table 6.1: Densities of managed plant species in seven forest gardens, Santa Rosa (Tot= Total, Rk = Rank)

Binomial, collection reference, Forest garden and vernicular name A B C D F G H Tot Rk

Alchornea triplinervia (de Jong 125, 42 0 0 0 0 0 0 42 25 zancudo caspi) Anacardium occidentale L. (casho) 0 4 8 0 10 0 0 22 37 Ananas comosus L. (pina) 683 7 0 33 1248 6 0 1977 1 Annona muricata L. (guanabana) 0 0 3 0 0 0 0 3 78 Annona sp. (de Jong 164, anonilla) 0 0 13 0 0 0 0 13 51 ANNONACEAE (de Jong 138, sacha anona) 4 0 0 3 0 0 0 7 63 ARECACEAE (inayuga) 8 0 0 0 5 0 0 13 49 Artocarpus incisa L. (de Jong 57, pandisho) 0 15 0 0 14 3 0 32 28 Astrocaryum chambira Burret. (chambira) 0 0 0 0 5 0 0 5 68 Astrocaryum huicungo Damm, (huicungo) 0 0 3 13 5 0 0 21 39 Bactris gasipaes (pijuayo) 92 122 27 50 157 3 95 546 4 Bixa sp. (de Jong 144, sacha achiote) 4 0 3 0 0 0 0 7 63 Calathea allouia (Aubl.) Lindl. (daledale) 0 48 0 0 0 0 10 58 20 C. sp. (de Jong 142, uchpa vijau) 8 0 0 127 0 3 0 138 11 C. sp. (de Jong 145, huira vijau) 0 0 0 0 14 0 0 14 47 Calycophyllum spruceanum (Benth.) Hook. 8 0 0 3 0 0 0 11 54 (de Jong 35, capirona) Capsicum spp. (de Jong, 53,55,56,62,71, aji) 4 0 0 0 0 0 0 4 73 Caricapapaya L. (papaya) 17 4 0 7 0 0 5 33 27 Cecropia membranacea Tree. 4 0 0 0 0 0 0 4 73 (de Jong 163, cetico) Cedrela odorata (de Jong 26, cedro) 0 70 11 0 43 0 5 129 12 Citrus aurantifolia (Chisten) Swingle 0 26 45 0 0 0 0 71 17 (de Jong 72, limon dulce) C. limon (L.) Burm (de Jong 154, limon) 0 15 11 0 24 0 0 30 21 C. nobilis var. deliciosa Swingle 0 4 21 0 0 0 0 25 36 (de Jong 153, mandarina)

96 Table 6.1 continued Species name, collection reference, Forest garden and vernicular name A B CD F G H Tot Rk C. paradisi Macfaden (de Jong 152, toronja) 0 11 24 0 0 3 0 382 6 C. sinensis (L.) Osla (de Jong 76, naranja) 83 44 96 7 0 6 10 246 8 Citrus sp. (de Jong 75,pomelo ) 0 0 13 0 0 0 0 134 9 Coffea arabica L. (cafe) 0 33 13 0 0 24 0 701 8 Colocasia esculenta (L.)Schott (de Jong68 , 8 0 3 0 0 0 0 115 4 huitina) Couma macrocarpa Barb.Rodr. (Peters 111, 0 7 5 10 0 0 5 27 34 lèche caspi) Crescentia cujete L. (de Jong 52,huingo ) 0 4 8 0 0 0 0 125 3 Curcuma longa L. (de Jong 11, guisador) 12 4 0 0 0 0 0 164 4 Dioscorea trifida L. (sachapapa ) 25 0 0 7 24 3 0 591 9 Eugenia stipitata McVaugh (de Jong 151, 0 0 0 7 0 0 0 76 3 araza) Euterpeprecatoria (husai) 0 7 5 0 5 0 0 174 2 Ficus sp. (de Jong 141, oje) 12 0 .3 0 0 0 0 154 6 Ficus sp. (deJon g 89,renaco ) 0 0 0 0 10 0 0 105 7 Genipa americana L. (de Jong 134,huito ) 0 0 8 0 0 0 0 85 9 Griasperuviana Miers (de Jong 50, 4 26 8 0 5 0 0 432 4 sachamango ) Himatanthus cf. sucuuba (Spruce)Woods . 0 4 0 3 10 0 0 17 42 (Peters 166,bellac o caspi) Hura crepitans L. (de Jong 149,katawa ) 0 0 3 0 0 0 0 37 8 Ingadensiflora (Benth., de Jong 128a, vacapaca)0 0 0 5 0 0 0 0 56 8 I. edulis Mart, (guava) 6 7 67 93 71130 38 452 5 I. macrophylla H.& B.e xWilld . (de Jong 133, 01 1 0 0 0 0 10 213 9 guava peluda) I. pilosula (Rich.) Macbr. (de Jong81 , 12 0 11 3 5 0 0 31 30 shimbillo) Jacaranda copaia Aubl. (huamansamana) 4 0 0 0 0 0 0 47 3 Jessenia bataua (Mart.) Burret (ungurahui) 0 7 0 0 19 0 0 263 5 Lonchocarpus nicou (Aubl.) DC. (barbasco) 33 7 0 0 0 3 33 761 6

97 Table 6.1 continued Species name, collection reference, Forest garden and vernicular name ABCD FGHTotRk Mangifera indica L. (mango) 81830 030 32 28 Mauritiaflexuosa L . (aguaje) 0 0 29 0 0 0 0 29 32 Miconiapilgeriana Ule (de Jong 129, rifari) 29 0 0 0 0 0 0 29 32 Miconia sp. (de Jong 165,rifar i bianco) 192 0 0 0 0 0 0 192 9 Musaparadisiaca L. (platano) 471 15 317 7 3310 332 4 1126 2 Ochroma lagopus Swartz (topa) 8000 000 8 59 Oenocarpus mapora Karst, (de Jong 7, 0 016 0 14 0 0 30 31 sinamillo) Palicourea duroia (de Jong 162,palt a moena) 0000 500 5 68 Paspalum rvpens Berg. (de Jong 130,cariso ) 0 0 13 0 0 0 0 13 51 Persea americana Mill, (palta) 4033 000 10 57 Petiveria alliacea L. (de Jong 16,mucura ) 0030 000 3 78 Phytolephas macrvcarpa R.& P (yarina) 015 3 0 000 1841 Poraqueiba sericea Tul. (de Jong 41, umari) 810 0 32 23 5 15 86 269 7 Pourouma cecropiifolia Mar. ex Miq. (uvilla) 50 26 27 70 81 61 19 334 6 Póuteria caimito (R&P) Radlk. (caimito) 2520 4 61 97 167 82 5 641 3 Psidium guayava L. (guayaba) 21 22 24 20 29 12 0 128 13 Quararibea cordata (H.& B.) Vischer (zapote) 0 22 0 0 0 0 0 22 37 Rheedia sp. (de Jong 78,charichuelo ) 0450 500 14 47 Rollinia mucosa (Jacq.) Baill. (de Jong 48, 25 26 13 10 24 48 33 179 10 anona) Saccharum officinarum L. (cana) 83 0 3 7 0 6 0 99 14 Scheelea brachyclada Burret (Ruiz 1424, 0053 000 8 59 shapaja) Senna multijuga (Rich.) I.& B.(Rui z 1464, 4000 000 4 73 pashaco) Socratea exorrhiza (Mart.) Wendl. (de Jong 49, 011 50 000 1644 cashapona ) Solanum stramonifolium Jacq. (de Jong 64, 4030 000 7 63 coconilla) Spondias dulcis L. (taperiba) 0003 000 3 78

98 Table 6.1 continued Species name, collection reference, Forest garden and vernicular name A B CD F G H Tot Rk Spondias mombin L. (uvos) 0 0 5 0 0 0 0 56 8 Syzygium malaccensis (L.) Merr.& Perry 0 26 19 0 0 3 0 482 3 (mamey) Tabebuia crysantha (Jacq.) Nicholson 8 0 0 0 0 0 0 85 9 (de Jong 166,tahuari ) Theobrvma bicolor Humb.& Bonpl. (macambo) 4 22 21 3 0 0 0 502 1 Theobroma obovatum Klotzsch ex Bernoulli 0 4 3 0 0 0 0 76 3 (de Jong 156,cacahuillo ) Vernoniapatens HBK (de Jong 131,ocuer a 4 0 0 0 0 0 0 47 3 negra) Vismia angusta Miq. (de Jong 132,pichirina ) 92 0 0 0 0 0 0 921 5 Vitex sp. (deJon g 137,almendra ) 4 7 0 0 0 0 0 115 4 ? (de Jong 42, sushana caspi) 0 0 0 3 0 0 0 07 8 ? (zapohuasca ) 0 0 0 0 5 0 0 56 8 ? (sachaajos) 0 0 3 0 0 0 0 3 78

Total individuals 2157 1009 719 785 2042 51767 8 Total species 40 38 48 26 28 19 14

Bactris gasipaes, Musa paradisiaca, Poraqueiba sericea, andPourouma cecropifolia. Atota l of 63 species occurred in only three or less forest gardens, with 28 of those species occurring in only one of the fields sampled. Miconia pilgeriana the species which ranked number nine in total number of plant indi­ viduals, was tended in only one forest garden (gardenA) . This species is asec ­ ondary forest tree which is used for construction purposes. Other forest species rankinghig h intota l numbero fplan t individuals wereCedrela odorata grown for its timber, Vismia angusta grown to be used as round wood in construction, and Calathea sp.grow nfo r its leavesuse d forwrapping . Thecas eo fth eCalathea sp. is especially interesting since its natural habitat is restricted to the floodplain. It had been introduced onterr a firme byth e owner of forest garden D, where itre -

99 produced easily and became very abundant. Forest garden D is now a source of this Calathea sp. leaves for many people in Santa Rosa.

Weeding patterns The weeding patterns of four forest gardens (A, B, D, and F) were com­ pared by estimating the state of the non managed vegetation (weeds) every three months for a total of 18 months, in the same transects in which densities of man­ aged species were measured. No permanent plots were established in forest gar­ den C because of its location close to the village, nor in forest garden E and G because they were slashed soon after the first survey. The use of the Index of Non-Managed Vegetation (Ij,^) is explained in chapter two. Forest gardenA showe d a regular weeding pattern relatively uniform through­ out the whole field (Figure 6.1). During the first observation, the field was in the process ofbein g slash-weeded, and the largerpar t (83%) had only heel-high weeds

(Inmv 1). A smaller part (17%) was covered with taller knee-high weeds (Inmv 2). During the second observation this slash-weeding wasfinished an d the wholefield had a cover of only heel high weed vegetation (Figure 6.1). During observations three, four, and five the field showed a gradual increase of areas with waist-high weed vegetation (Inmv 3). But between the fifth and sixth observation the field was slash-weeded completely again, resulting in a total cover of knee high weed vegetation (Inmv 2) during observation six. The weeding pattern in forest garden B was much more irregular than in the previous field. During the first observation, two thirds of the field (67%) had a only a heel-high weed cover (Inmv 1). In the other third of the field (33%) the weed vegetation was knee- and waist-high (Inmv 2 and 3). This latter area of for­ est garden B apparently received much less attention from the owner (Figure 6.1). Weed vegetation increased in forest garden B through the second and third obser­ vations resulting in waist to person-high weeds (Inmv 3 and 4). However, by the fourth observation this tendency was partly reversed, and the area with knee-high weeds (Inmv 2) increased again. During the third and fourth observation 22% of the field was covered by a person-high vegetation (Inmv 4).

100 Observation: 2 3 4 6 forest garden A

ro JLLJI_J_L 1 ' « ' •!!.. 1 I 1 . I .11.1 1 7 7 4 I 1 1 ! 4 4 forest garden B

J_ JL_ _Jl_ JlL JULJli _ 114 4 1114 4 .114 4 i 7 7 7 1 7 1 1 T 4 , , 7 7 1 forest garden D O

JlLJ k JIL _JiL_ul L jJ forest garden F

114 4 1114 1 1114 1 Inmvvalu e

Figure 6.1 Non-managed vegetation Index (lnmv) in four forest gardens. Each single graph indicates the percentage of the total area of a single field covered with weed vegetation represented by a certain lnmv value at a single observation. Observa­ tions were repeated six times at quarterly intervals. Inmv values: 1 = heel high; 2 = knee high; 3 = waist high; 4 = person high; 5 = area covered with secondary forest

101 Betweenobservatio n four andfive, part so ffores t gardenB whic hha dkne e andperso n high weeds (Inmv 2an d 3)wer e slash-weeded again, while otherarea s with person high vegetation were left untouched. At the last observation, how­ ever, the area with person high weed vegetation (Inmv 4) decreased, and a larger area showed only knee high weeds (Inmv 2). In the period between the last two observations theare awhic h hadgon ewithou t weedingfo rth e longest periodwa s cleared, while those areas cleared inth e previous period were left undisturbed. Atth efirst observation , forest gardenD showe d alarg eare a (60%)covere d withperso nan d forest highwee dvegetatio n (Inmv 4 or 5,Figur e 6.1). Duringal l six observations the area covered by such a high weed vegetation was neverles s than 47%o fth e field. However, the other half ofth efield wa sslash-weede d fol­ lowinga regula rschedule , similart otha t offield A . Betweenth efirs t andsecon d observation, asmalle r area was slash-weeded, resulting ina slight increase inth e part with knee high weeds (I^v 2). The next two observations showed an in­ crease of areas with a taller weeds,bu t during thefifth an d especially duringth e sixth observation theare a with lowerweed s increased againa sa resul t ofa slash - weeding ofthes eareas . In forest garden F much less weeding was carried out than in any of the previousfields durin gth e 18month s ofobservation . Duringth e secondan dthir d observations asmalle r areadi dhav ea wais t high weedvegetatio n (1^^ 3,Figur e 6.1). The rest of the field had waist and person high weeds throughout the 18 months. At observationsfive an d six, 100%o f thefield ha d a forest high weed vegetation (Inmv 5).

Yield levels Yield levels of economically important products in the same four forest gardens wereestimate dusin g methods described inchapte r two. Atota l number of4 2 different plant species, or 67%o f allth e managed species found inth e four gardens yielded harvestable products. In forest garden Ath e number of species yielding harvestable products was 24, or 60%o f all the managed species inth e field. In forest garden Bthi s was 23,o r 61% ofth e managed species. In forest

102 Table 6.2: Production perhectar e offou r forest gardens inSant a Rosa duringa fifteen months period. Yield levels arei nnumbe r offruit s unless otherwise indi­ cated.

Binomial and Spanish name Forest garden A B D Ananas comosus, pina 1333 4 7 0 Artocaprus incisa, pandisho 0 85 0 0 Astrocaryum huicungo, huicungo (trees) 0 0 3 0 Bactris gasipaes, pijuayo (racemes) 100 400 120 0 Capsicum spp.,aj i 208 0 0 0 Carica papaya, papaya 242 0 27 0 Calathea alloiua, daledale 0 26 0 0 Calathea sp., huira vijau (plants) 4 0 0 0 Calathea sp., uchpa vijau (plants) 8 0 127 3 Cedrela odorata, cedro (stems) 0 7 0 0 Citrus auranthifolia, limon dulce 0 504 0 0 Citrus limon, limon 0 411 0 0 Citrus sinensis, naranja 0 2482 0 0 Coffea arabica, cafe 0 11 0 6 Colocasia esculenta, huitina 4 0 0 0 Couma macrocapra, leche caspi (trees) 0 7 0 10 Dioscorea trifida, sacha papa 8 0 27 21 Euterpe precatoria, huasai (racemes) 0 15 0 0 Grias peruviana, sacha mango 0 15 0 0 Inga edulis, guava 4162 0 6523 2394 /. macrophylla, guava peluda 0 44 0 0 Jacaranda copaia, huamansamana (stems) 4 • 0 0 0 Lonchocarpus nicou, barbasco (stems) 38 0 0 0 Miconia pilgeriana, rifari (stems) 29 0 0 0 Miconia sp., rifari bianco (stems) 192 0 0 0 Musa paradisiaca, platano (racemes) 329 26 160 91 Phytolephas macrocarpa, yarina (trees) 0 15 0 0 Poraqueiba sericeae, umari 0 5630 0 0 Pourouma cecropiifolia, uvilla 3558 482 757 0 Pouteria caimito, caimito 0 1518 337 333 Psidium guajava, guayaba 621 0 133 0 Quararibea cordata, zapote 0 2111 0 0 Rollinia mucosa, anona 288 37 0 0 Saccharum officinarum, cafia (stems) 188 0 3 0 Senna multijuga, pashaco (stems) 4 0 0 0 Socratea exorrhiza, casha pona (stems) 0 11 0 0 Solanum sessiliflorum, cocona 25 0 0 0 Solanum stramonifolium, coconilla 171 0 0 0 Syzigium malaccensis, mamey 0 3926 0 0 Theobroma bicolor, macambo 0 70 20 0 Vernonia patens, ocuera negra (stems) 4 0 0 0 Vismiaangusta, pichirina (stems) 92 0 0 0

103 garden Donl y 13,o r 50% ofth emanage d speciesproduce d harvestableproducts , and infores t gardenF onl y seven,o r 37%o fth e speciesyielde d harvestableprod ­ ucts. Other amounts of harvestable products per species are presented in Table 6.2. It appearstha t notal lth especie s which wereplante d actuallyyielde dprod ­ ucts for which they wereplanted . This is a direct result of the opportunisticap ­ proach which ribereno farmers follow in their resource management activities. Riberefios havet o anticipate many different possible situations when makingag ­ ricultural decisions. Farmers may intend to manage a forest garden in a certain way,but , in responset o other factors, inth e end decide nott o follow the original plan. Hence,specie s mayb eplanted , that inth e endar eno ttende d properly,an d hence can not be harvested.

MANAGEMENT OFRIBEREN O FORESTGARDENS .

According to Fresco (1986: 110) management and labor are the most im­ portant inputs ofcroppin g systems amongshiftin g cultivators. Inman y studieso f agroforestry systems such management is reported to have different levels of in­ tensity (e.g. Nair 1989), although it is seldom specified how such variable man­ agement intensity can be measured. In chapter two, management of an agricul­ turafieldl wa sdefine d aspurposel y influencing thedevelopmen t ofit svegetation . Cultivationpractice stha tmos t influence thevegetatio n ofa fores t gardenar eplant ­ ing and weeding. The densities and diversity of managed species, and weeding patterns,therefore , areth emai nindicator so fdifferen t management patterns inth e of Santa Rosa forest gardens. Theresult spresente dabov esho wfou rdifferen t weedingpattern so fth efield s studied. Forest gardenA showe da regula rpatter n ofon ecomplet e slash-weeding peryear . However, asth efield becam e olderthi s slash-weeding became lessthor ­ ough from yeart o year. The largerpar t of forest garden B was kept fairly clean, but other parts were only cleared when the weed vegetation began turning into dense secondary forest. The slash-weeding activities here were spread out over

104 the whole year, but covered different areas at different times. Forestgarde nA an d B,an d forest gardenC whic hha dweedin gpatter nsimi ­ lar to forest garden B, had similar numbers of managed species (respectively 40, 38, and 48, Table 6.1). The difference in weeding patterns is attributed to the difference inag eo fth etw o fields. Forest gardenA ,whic hwa sno tye tthre eyear s old,whe nth esurve ybegan ,require da mor ethoroug h weedingi norde rt oachiev e thedesire d composition and structureo fth e vegetation. Inth e olderfores t garden B, which had a well established forest vegetation, less effort was necessary to maintain the existing composition and structure. This is also reflected by the amount of labor input needed for weeding each field. The amount of labor re­ corded byth e ownero f forest gardenA i n a diary during oneyea rwa s 13.5wor k days. The owner of forest garden B recorded only 0.75 work days spent in the same activity (Padoch & deJon g 1989). Forest garden D had a one year slash-weeding schedule, but only half of the field was cleared. This isconsisten t with the lowernumbe r ofmanage d spe­ cies in this field. Forest garden D, and forest garden E, in which the weeding pattern is similar, had 26 and 28 managed species. Only the areas where those managed species are present are kept under control, while in the other parts the vegetation hadbee n allowed todevelo p freely. It isno t impossibletha tth eowne r of forest garden D spent the same amount of labor on weeding as the owner of forest garden B. The slash-weeded area in forest garden D increased section by section duringth eobservatio nperiod . This isa consequenc e ofmor etree sreach ­ ing sizes at which they start yielding harvestable products, at which point their direct surroundings are than slash-weeded. Once the managed vegetation of for­ est garden D becomes fully grown, this field may well be managed in the same way as forest garden B. Such a strategy makes sense, since a more intensive management ofth efield woul d requirehighe r laborinput s atyounge r stagestha n when thefield i solder . The last forest garden F had very few managed species, and subsequently was virtually not cleared at all. It is unlikely that this will change sinceth ecur ­ rentstatu s ofth e non-managed vegetation andth elo wdensit yo fmanage dspecie s

105 dono tjustif y such investment. Theowne ro fthi s field stated that hewil l makea newfield o n the same site onceth e secondary forest is dense enough to suppress thegrasse s stillpresent . Thisfield an dfores t garden G aremos tclosel y relatedt o the abandoned swidden, which is still to often considered typical of shifting cul­ tivation practices. Forest garden F,however , retained atota l of 19manage dspe ­ cies, and forest garden Gha d 14manage d species.

PATHWAYSFO RINTENSIFICATIO N OFRIBEREN OAGROFORESTR Y

According to Nair (1990) most of the existent agroforestry practices inth e tropics need to be developed into modern agroforestry technologies through sci­ entific improvement. Some suggestions for such improvements have beengive n by Raintree &Warne r (1986) who propose integral taungya, enriched fallows, or tree crop alternatives to intensify shifting cultivation practices. Dubois (1990) suggests a sequence of swiddening, enriched forest fallows, and perennial crop plantations forAmazonia n Brazil, whileNai r (1990) presents cacao grownunde r widelyspace d rubbertrees , coconuts,an dpeac hpalm s asa noptio nt o farmers for additional cash income and economic stability. It is useful to discuss briefly (1) whethero rno tsuc ha nimprovemen t isnecessar y forterr afirme agroforestr y prac­ tices in Santa Rosa, (2) how such an improvement could be achieved, and (3)t o what extent the diversity of existing agroforestry practices hast ob etake n incon ­ sideration for such an improvement. Increased population pressure among upland farmers often has resulted in shortening of the fallow period of swidden-fallow systems (Boserup 1965; Ruthenberg 1980; Clarke 1966). However,othe rexample sdemonstrat etha thighe r population densities can also lead to intensified management of homegarden s or forest gardens (e.g. Fernandes et al. 1984; Leuschner & Khaleque 1987; Soemarwoto et al. 1985; Michon 1985; Michon et al. 1986; Wiersum 1983; Christanty et al. 1986). Such intensification are often responses to a shortage of forest resources, rather than land shortage. Agroforestry intensification can also result from new market opportunities, as demonstrated by examples from the

106 Amazon (Padoch et al. 1985;Hiraok a 1986) or elsewhere (e.g. Eder 1981). In general, Santa Rosa farmers practice more intensive agroforestry than farmers liketh eBor aIndian swh oliv ei na remote rare ao fPer u(Deneva n& Padoch 1987), or the Runa Indians of Ecuador (Irvine 1987; 1989). The inhabitants of SantaRos ad o notsel lan ysignifican t amounts ofproduct s from agroforestry fields inth emarket ,bu tus emos t ofi t forhom econsumptio n(Padoc h &d eJon g 1989). As such it can be concluded that agroforestry intensification in Santa Rosa is a result of increased pressure on forest resources, and also a result ofa greaterde ­ mand for family labor. The increased pressure on forest resources is a result of the sedentary life style of the Santa Rosa inhabitants. The distance to high forest becomes greater everyyear ,an dth efores t resourcewithi nclos evicinit yt oth evillag ear eharveste d faster than they regenerate. Since Santa Rosa farmers engage in commercial ag­ riculture, they have to minimize their labor expended on subsistence work. Pro­ ducing forest resources in agroforestry fields therefore, becomes an economically attractive option. Although riberenos are definitely in need of additional cash income and would eagerly welcome economic stability, it will not be easy to propose new agroforestry technologies which can provide such benefits as suggested by Nair (1990). Cacao,rubber , and coconut aredifficul t toproduc e onterr afirme terrain , while the market for peach palm is extremely limited. Timber species are slow- growing, although some farmers expressed interest in planting Cedrela odorata and Cedrelinga cataneiformis, to produce marketable timber. Thefocu so nsubsistenc eproductio n inSant aRos aagroforestr y alsoexplain s part ofth e diversity inagroforestr y practices. Sincen o obvious increase ofmon ­ etary incomei st ob egaine d from singlecro p intensification inagroforestr y fields, farmers valueth eeconomi cbenefit s from agroforestry fields indifferen t ways. It becomesa questio no fpersona lpreferenc e whichcrop sar et ob eproduce d insuc h fields and how much effort they are worth. Farmers who live in the village of Tamshiyacu focus mainly on Poraqueiba sericeaproduction , because they can market this product in Iquitos. Because earning cash income isth e main goalo f

107 agroforestryfields in Tamshiyac u(Padoc he tal . 1985;Hiraok a 1986,d eJon g1984) , agroforestry intensification ismuc hmor eunifor m inthi svillage . Sincesuc hmarke t stimuliar eminima l inSant aRosa ,farmer s makefores t gardens inaccordanc ewit h their personal circumstances and desires. Onlyagroforestr y practicestha twoul dgenerat ecas hincome ,ca nb eexpecte d tob eadopte d widely in SantaRosa . Giventh egrea t agricultural diversityamon g these farmers, it isno t at all evident that agroforestry isth e most appropriatewa y toreac hincreas emonetar yreturns . A different strategyt o increasehousehol dwell - beingcoul db et o increaseproductio n offores t resources for local use. There isa tendency among many farmers to replace forest products with products that can be bought in the market. Since cash incomes among riberenos fluctuate greatly from year to year, producing stable forest resources in agroforestry fields may become more attractive to farmers. Since household economic conditions vary within single villages, interest in such subsistence production would also vary. Therefore, diversity in individual agroforestry practices will persist even if agroforestry intensification assuggeste db yNai r(1990 )Raintre ean dWarne r(1986 ) and Dubois (1990) could be achieved.

108 CHAPTER SEVEN

FLOODPLAIN AGROFORESTRY INYANALLP A

INTRODUCTION

Indigenous agriculture of the Amazon floodplain is still a new field of re­ search. It is,therefore , nosurpris e that varzea agroforestry hasbee n less described than other agricultural systems of this ecological zone. Agroforestry practices are less common in varzea-like environments than in uplands, because the former are much rarer than dry-land,an d because periodically flooded land ismor e often used for intensive cropping of annuals or non-wood perennials (e.g. Hopkins 1987; Ruthenberg 1980). Likewise, farmers who live in flooded areas in countries like Bangladesh and India, locate forest gardens and homes on dry land rather than in lands that are periodically inundated (Leuschner & Khaleque 1987;Al i 1987;Nai r and Sreedharan 1986). Varzea agroforestry and forest management have been studied in the Ama­ zon estuary near Belem (Anderson 1990;Anderso n et al. 1985;Gel y 1989). The lands on which this kind of agroforestry occurs, however, flood daily and cannot be used for any other type of cropping. Frechione et al. (1989) and Parker et al. (1983) mentioned the occurrence of floodplain agroforestry among inhabitants of the middle reaches of theAmazon . Hiraoka (1985b) described briefly how flood- plain farmers in Peru transform restinga fields to forest gardens after several years of swiddening. In these forest gardens, planted fruit trees and spontaneously oc­ curringusefu l speciesar eharvested ,bu tthe yreceiv e littletending . Bergman (1974) reports on treegrowin g inplantai n fields aswel l as in orchards in floodplains along the Ucayali river. In the floodplains of theAtrat o river in the Colombian Chocó region,agro-silvicultura lpractice s including thegrowin g oftropica l cedar,Cedrela odorata, are reported (Leguizamo 1983,cite d in Dubois 1990). A more thorough study on varzea agroforestry has been conducted in the

109 close vicinity ofManau s (Bahn et al. 1991; Guillaumet et al. 1990;Bah n 1988). Bahn etal .(1991 )distinguis h four different types offores t gardens located onth e higherlevee so fCareir o island. These leveesar eno t inundatedb ya naverag ehig h flood. Thetype s of forest gardens are:(1 ) mixed rubber and cacao stands which wereplante d aroundth etur n ofth ecentur y and are still exploited, (2) similarol d plantations but enriched with fruit trees, (3) mixed fruit orchards with no rubber trees, and (4) pure rubber stands with pasture underneath. In these gardens 35 tendedspecie swer efound , includingfrui t trees,medicina lplants ,an dspecie swhic h yield a mixture of otherusefu l materials (Guillaumet et al. 1990). Theclos edis ­ tancebetwee n Careiro islandan dth e Manausmarke t allows somesal eo fth epro ­ duce from those fields. According to the researchers ofthes e varzea agroforestry systems, only economic and historical factors influence the variation in species composition and management of the forest gardens, since ecological site condi­ tions are uniform. In the present chapter diversity in varzea agroforestry practices in the vil­ lage of Yanallpa will be discussed. It will be argued that varzea diversity in agroforestry is partly a result of ecological conditions as well as economic cir­ cumstances of farmers. In this aspect,Yanallp a varzea agroforestry differs from the systems described for Careiro island in Brazil. In chapter four it wasargue d that, inth e varzea variation insit econdition s ison eo fth eprincipa l factors which determines its agricultural use. In the previous chapter, however, it was demon­ strated that differences in ecological site conditions dono t play aprimar y rolei n the variation in species composition and management of terrafirme forest s gar­ dens. Forest gardens inYanallpa ,however , are located onsite s which vary insoi l structure and fertility, and which experience different flooding regimes. Soiltex ­ ture, fertility and the flooding regime co-determine which type of agroforestry system is possible on a particular site. To begin the discussion of varzea agroforestry, the relation between agroforestry practices and other varzea agricultural types will be discussedfirst. Thiswil l explainwh yagroforestr y isles scommo n inYanallp atha n inSant aRosa , as noted inchapte rfive, an dconve y someawarenes s ofth e natureo fth e diversity

110 among agroforestry fields. Subsequently, the agroforestry fields studied will be described, and a brief land-use history included. The latter was recordedfrom interviews of current field owners. The composition of the managed vegetation and thegenera l weedingpatter n ofth e samefields wil lb ecompare d before more general conclusions are drawn from these findings.

YANALLPA FLOODPLAIN AGROFORESTRY

Agroforestry fields inYanallp aar elocate d onsevera l parts ofth eprincipa l village levee,qualifie d ashig hrestingas ,an do nlowe rrestinga s onth ebackslope s ofth esam e levee. Someagroforestr yfields ar elocate d onsmalle rlevee sa tlarge r distances from the river. Fertility and water retention capacity increase because the soil texture becomesfiner a s weapproac h the lower restingas. On the lower restingas, below a certain elevation,fields flood ever y year. Higheru po nth elevees ,besid etree swhic hresis t waterlogging for longerperiods , shortcycli ccrop ssuc ha smanio cca nb egrown . Onth ehighe rpart so fth elevees , plantains can be grown continuously for 15year s or longer without any need of fallowing, ifthe yd o not flood inearlie ryears . This means that inYanallp aman y different agricultural landus eoption sexis t for landwhic h isuse dfo r agroforestry. Onterr afirme, th e change from managinga field a s achacr a ora plantain field toa fores t garden isth eresul t ofth edecreas eo fsoi l fertility andth eincreas e ofwee dinvasio nwhic hmak ei teconomicall yunattractiv et ocontinu eannua lcrop ­ ping. Continuing afield a s a forest garden is moreprofitabl e than total abandon of a field. InYanallpa , however, farmers have to choose between using afield either foragroforestr y orfo r something else. Sincearabl e land ismor e limited in Yanallpa than in Santa Rosa, farmers less often decide to make an agroforestry field. This explains why fewer farmers in Yanallpa than in Santa Rosa have agroforestryfields. The nine agroforestryfields studie d inYanallp a differed in siteconditions , land-use history, dominant vegetation, and weeding patterns. They were located on both higher and lowerpart s of theprincipa l village levee, and onanothe r dif-

111 ferent levee away from theriver (se e Figure 3.2 chapter three). Two mixed or­ chards which were not very different in appearance from those studied in Santa Rosa were surveyed. Also one field with predominantly mixed tree vegetation, andon efield dominate db ybot hCedrela odorata, andCalycophyllum spruceanum trees wereselected . Oneagroforestr yfield marke dth eboundar ybetwee nterrain s belonging totw o different farmers. In addition to these agroforestry fields four fields in which farmers com­ bined annual crops withtre e production were inventoried. In two fields, farmers grew plantains and several tree species together. In two otherfields th e owner grew corn with Cedrela odorata trees. It should be emphasized here that these fields wereclassifie d asplantai nfields, an d restinga cornfields whe n conducting the agricultural type inventory amongYanallp a households. Inthi s chapter their tree component is compared with the other agroforestryfields.

YANALLPAAGROFORESTR Y FIELDS

One ofth e two mixed orchards which were studied inYanallp a (field A) began asa manio can dplantai nfield i nwhic hth e ownerha dplante d afe w Citrus paradisi andBactris gasipaes trees. Untilth e early 1970sth e owner ofthi s field lived ina hous elocate d ona par t ofth e leveei nfron t ofth eactua l mixedorchard . This part of the levee was eroded by theriver an d the house was relocated to its present siteinsid eth eorchard . Sometree swhic hwer elocate dclos et oth ewater' s edge already were lost as a result of erosion. This phenomenon is an important factor in the dynamics of varzea agroforestry fields in general (e.g. Bahri et al. 1991), but especially on the Ucayali. The Ucayali, liketh e Maranon, isa mean­ deringriver, a soppose dt oth eAmazo nwhic hfo r itslarges tpar ti sa braidin grive r (Goulding 1980;Siol i 1984,se echapte r 3). Theshiftin griver channe l mayresul t inlos s ofagroforestr y land,an d subsequently inlos so fplante dcrops . Asa resul t ofsedimentation , landma yals obecom e locateda tgreate rdistance s from the river channel, and as a consequence fields may lose their importance as intensively managed agroforestry systems.

112 In the orchard several of the Syzigium malaccensis, and Cedrela odorata treeswer eplante dbetwee n 1965 and 1970. Othertree s likeScheelea brachyclada orCouepia sp.wer eplante d morerecently . TheMauritia flexuosa tree s alsodat e from before 1970, but they appeared spontaneously from seeds disposed by the owners. According to the owner, several tree species which he grew inthi s field aren olonge rpresen tanymore . Specieslik eorange ,Pouteria caimito, Quararibea cordata, Rollinia mucosa andPersea americana (avocado),wer eal l destroyedb y thehig hfloo d in 1970. This loss discouragedth e ownerfro m replantingan yspe ­ cies vulnerable to flooding. Thesecon d mixedorchar d (field B)whic h wasals o located onth eprinci ­ pal levee,ha d aspatia l pattern which was different from field number one. Trees were clustered in separated locations inside the field. In one area Citrus species were grown nearly exclusively. Five smaller areas, at the boundary of the field, were usedt o grow manioc andplantains . In two areas a combination of spices, medicinals,ornamentals ,an dsugarcan ewer egrown . Thefield belonge dt oa nolde r woman who came toYanallp a in 1937. Since then she started planting this area withth e species still present today. Onlyth eagroforestr y areas ofthi sfield wer e inventoried. The mixed-tree agroforestryfield (fiel d C), also located onth e principal higherlevee , wasslashe d insecondar y forest in 1970. Thefiel d was first planted to corn, then to manioc, and later to plantains. The Cedrela odorata and Calycophyllum spruceanum trees, which were most abundant in this field, all stemmed from natural generation. The ownerals o planted a number ofScheelea brachyclada trees, several of which had died, probably as a result of the sand deposited bysevera l floods. Until threeyear s before the inventory, thefield wa s managed as a mixed chacra-agroforestry field. When manioc and plantain pro­ duction decreased, the owner decided to stop growing these crops and continued treeproductio n only. Inth e second half of 1987, however, shortlyafte r finishing theinventory ,muc ho fth etre evegetatio nwa sslash-weeded . Thefield wa splante d againwit hplantain san dmanio cunderneat ha nope ntre ecove ro fCedrela odorata, Calycophyllum spruceanum and some fruit trees. Before the slash-weeding, the

113 owneralread y hadharveste d about 75 Calycophyllum spruceanum treesfrom thi s field, which he haduse d to construct three different houses ofhi s own. Someo f thetree s heha d sold to the village to construct the school building. None ofth e Cedrela odorata trees wereye t of a harvestable size. Themixe dCedrela odoratalCalycophyllum spruceanum agroforestry field (field D)containe d only tree and shrub vegetation, and very few plantains. This field waslocate d ona lowresting abehin d theprincipa l village levee(Figur e 3.2). The soils ofthi s site were offiner texture ,an dflooded ever yyear . Ifno t useda s anagroforestr yfield, th esit ecoul db euse dt ogro w rice,corn ,o rvegetables . The sitewa sfirst slashe d in 1943,an dcorn ,beans ,an dabou t6 0Cedrela odorata trees were planted. The field was managed for a few years as a corn field, but then continueda sa nagroforestr y field. Allwood y vegetation exceptth e6 0ceda rtree s appeared spontaneously. A common agroforestry practice found amongYanallp a farmers is the use of a small strip of trees to separate lands belonging to different farmers. The boundary-stripagroforestr yfield (fiel d E)wa sno tplante db yth efarme r in whose terrain it was located, and he did not know who the original planter was. The vegetationo fthi sagroforestr yfield consiste d ofsevera lfrui t treeso fdifferen t sizes, with a dense ground cover of Calathea sp.,a perennial herb. One of the mixed plantain/agroforestry fields (field F), located on the principal high levee,wa sslashe d forth efirst tim e in 1966. It wasfirst plante dt o manioc, and laterturne d into aplantai n field. While slashing thefield, a numbe r ofCedrela odorata treeswer espared ,fou r ofwhic hwer eharveste dtw oyear slater . Newtree s appeared spontaneously. Atth e time ofth efield inventory , thebroke n crown cover of the tree vegetation had regrown so dense that its shade affected plantain production. Thefield ha d been aplantai n agroforestryfield eve rsinc ei t wasmade . Thesecon d plantain/agroforestry field (field G),als o located onth ehig h restinga, wasslashe d in 1932. That sameyea rth e ownerplante d several Cedrela odorata trees, of which healread y had harvested ten atth etim e ofth e inventory. Manyo fth eothe rtree so fth esam especie spresen t inth efield ar eo fa harvestabl e

114 size, but the owner preferred not to sell them yet, since he had no urgent need for the money. Timber trees in particular are often kept as a savings in case of emergency by farmers inYanallp a (Padoch & de Jong 1987). The two corn/Cedrela odorata/agroforestry fields (fields H and I) be­ longed toth e same owner. Bothfields wer e located onth eprincipa lYanallp a levee, but in a much lower area than the village center. In 1985 the farmer grew a crop of rice followed by a crop of corn in both fields. The rice yield was very unsat­ isfactory, and in 1986 and in 1987th e owner grew two crops of corn. In 1989 the two fields were planted with plantains, a rather risky endeavor because of the rela­ tively low elevation of these fields and hence the high risk of flooding. Both fields have been used for annual cropping for several decades. They were made by the father of the current owner before she wasborn , and theCedrela odorata trees have been present in these fields as long as she can remember. In 1989th e owner harvested 35tree s from one of the two fields, and used the money obtained to finance the schooling of two sons in Lima.

RESULTS

Diversity of managed species Of the 78 species which were managed in the 9 varzea agroforestry fields only 35ar e domesticated plant species, while the other4 3 species are non-domes­ ticated species which occur naturally in native forests. Although plantain ranked as the species with the highest number ofplan t individuals in the nine agroforestry fields, Calycophyllum spruceanum and Cedrela odorata, both native forest spe­ cies, ranked number 2 and 3 (Table 7.1). For instance in the Cedrela odorata/ Calycophyllumspruceanum agroforestry field (field D) these species accounted for 61% of all the managed plant individuals. Some species from one of the mixed orchards (field A) are particularly interesting. Mauritia flexuosa, Inga densiflora, and Oenocarpusmapora ar ethre e forest species grown for their fruits. Inth e same field Scheelea brachyclada is grown, which provides leaves which are used for

115 Table 7.1 Densities of managed species in nine varzea agroforestry fields in Yanallpa (Tot = Total, Rk = Rank)

Binomial, collection number and Agroforestryfield vernicular name ABCDEFGHITotRk Allamanda cathartica L. (de Jong 169, 03 00 0 0000 3 67 Alocasia macrorrhiza (L.) Schott 03 00011000 14 48 (de Jong 67,huitin a blanca) Anacardium occidentale L. (casho) 017 0000000 17 47 Artocarpus incisa L. (de Jong 57, 12 0 33 0 67 0 0 0 0 112 15 pandisho) Bactris gasipaes HBK. (pijuayo) 00 0000500 561 Bixaorellana L. (de Jong 66,achiote ) 30 0000000 371 Calathea sp. (de Jong 142, uchpa vijau) 19 7 48 0 0 50 86 0 0 210 7 Calycophyllum spruceanum (Benth.) 19 2 210930 0 6 0 0 0 1167 2 Hook, (de Jong 35, capirona) Canavalia ensiformis DC. (de Jong 183, 0 0 0 0 0 22 0 0 0 22 41 nescafé) Capsicum annuum L. (de Jong 55, 012 0000000 12 50 aji dulce) C.spp . (de Jong 53,71,56,62, aji) 470 0 0 0 0 0 0 0 472 8 Caricapapaya L. (papaya) 41 8 0 0 03 3 191 4 2 1171 4 Casearia sp. (de Jong 27,limo n casho) 0 0 06 0 0 0 0 0 0 602 3 Cecropia membranacea Tree. 0 0 276110 13 0 0 0 0 399 5 (de Jong 163, cetico) Cedrela odorata (cedro) 110 0 124640 0 89 52 42 12 1069 3 Cedrelinga cataneiformis (tornillo) 010000000 172 limon dulce) Citrus aurantifolia (Chisten) Swingle 0 3 5 0 13 11 0 0 0 32 35 (de Jong72 , C.limo n (L.) Burm (de Jong 154,limon ) 92 0000000 11 53 C.nobili s var. deliciosa Swingle 0 0 0 20 27 0 0 0 0 47 28 (tangarina) C.paradis i Macfaden (de Jong 152, 91 60 100 0 0 0 5 0 0 256 6 toronja)

116 Table 7.1, continued

Binomial, collection number and Agroforestry field vernicular name A B C D E F G H I Tot Rk Coccoloba densifrons Mart. 0 0 0 0 13 0 0 0 0 13 49 (de Jong 94, vino huayo) Colocasia esculenta (L.) Schott 12 0 0 0 0 0 0 0 0 12 50 (de Jong 68, huitina) Couepia sp. (parinari) 3 2 0 0 13 0 0 0 0 18 46 Crescentia cujete L. (de Jong 52, 12 3 5 0 0 6 0 0 0 26 39 huingo) Cymbopogon citratus (DC.) Stapf. 0 1 0 0 0 0 0 0 0 1 72 (de Jong 105,hierb a luisa) Dahlia sp. (de Jong 168, dalia) 0 1 0 0 0 0 0 0 0 1 72 Ficus sp. (de Jong 141,oje ) 0 0 10 40 0 6 0 0 0 56 24 FLACOURTIACEAE (timareo) 6 0 0 70 0 0 0 0 0 76 17 Genipa americana L. (de Jong 134, huito)12 9 5 0 27 11 10 1 0 75 18 Gossypium barbadense L. (de Jong 65, 3 2 0 0 0 0 0 0 0 5 61 algodon) Griasperuviana Miers (de Jong 50, 0 0 10 0 40 0 0 0 0 50 25 sacha mango) Guarea maerophylla Vahl (de Jong 177, 0 0 0140 0 6 0 0 0 146 11 requia) Hevea sp. (de Jong 69, shirinaga) 6 1 5 50 67 0 0 0 0 129 12 Inga densiflora (Benth., de Jong 128a, 12 1 10 0 13 6 0 0 0 42 32 vacapaca) I. edulis Mart, (guava) 38 33 43 30 0 0 24 0 0 168 10 I. perizifera Benth. (de Jong 25, 0 6 14 0 13 0 0 0 0 33 34 rujino shimbillo) I.pilosula (Rich.) Macbr. (de Jong 81, 0 2 0 20 13 0 0 0 0 35 33 shimbillo) Inga. sp. (de Jong 173, guavilla) 3 0 0 0 67 0 0 0 0 70 20 Ipomoea batatas (L.) Lam. (de Jong 54, 3 0 0 0 0 28 0 0 0 31 36 camote)

117 Table 7.1, continued

Binomial, collection number and Agroforestryfield vemicular name ABCDEFGHITotRk I. quamoclit L. (de Jong 4, enredadera) 0 1 0 0 0 0 0 0 0 17 2 Jatrvpha curcas L. (de Jong86 , 6 0 0 0 0 0 0 0 0 6 60 pinonbianco ) LAURACEAE (moena) 0 0 0 10 0 0 0 0 0 105 5 Mangifera indica L. (mango) 6 5 24 0 0 0 10 0 0 453 1 Manihot esculenta Crantz (yuca) 82 0 0 0 033 3 0 0 0 415 4 Mansoa alliacea (Lam.) A.Gentr y 3 0 0 0 0 0 5 0 0 85 9 (de Jong 146,aj o sacha) Margaritaria nobilis L.F.(d e Jong 24, 0 1 0100 0 6 0 0 0 107 16 loromicuno ) Mauritiaflexuosa L . (aguaje) 22 4 5 10 13 01 4 0 0 682 2 Muntingia calabura L. (de Jong 176, 0 0 52 0 0128 5 0 0 185 9 yumanasa) Musaparadisiaca L. (platano) 69 0 110 10 053 327 6 71 1 1270 1 Myrciaria dubia McVaugh (camucamu) 0 1 0 0 0 0 0 0 0 17 2 Myrciaria. sp. (de Jong 170, 0 1 0 30 0 0 0 0 0 313 6 sacha camucamu) MYRTACEAE (de Jong 179,lanz a caspi) 0 0 0 0 10 0 0 0 0 0 105 5 Oenocarpus mapora Karst, (de Jong 17, 9 1 0 0 0 01 0 0 0 204 3 vacavilla) Ormosia amazonica Ducke, (de Jong 172, 30 0000000 3 67 huayruro) Passiflora quadrangularis L. (tumbo) 0 1 0 0 0 0 0 0 0 17 2 Phytolephas macrocarpa R.& P (yarina) 0 1 0 0 0 0 0 0 0 17 2 Pouteria caimito (R&P) Radlk. (caimito) 3 1 0 0 0 6 0 0 0 105 5 Psidium guayava L. (guayaba) 121 2 71 10 0 11 0 3 0 1191 3 Pterocarpus amazonum (Benth.) Amshoff 00 0 0 20 0 0 0 0 0 204 3 (de Jong 181, parinari) Pterodium densiflorum (de Jong 182, 0 0 14 0 13 0 0 1 0 28 38 casha moena)

118 Table 7.1, continued

Binomial, collection number and Agroforestry field vernacular name A B C D E F G H I Tot Rk

Rheedia sp. (de Jong 78, charichuelo) 16 1 5 0 0 0 0 0 0 22 41 Saccharum officinarum L. (caria) 3 0 0 0 0 0 0 0 0 3 67 Scheelea brachyclada Biuret 28 5 14 0 0 0 0 0 0 47 28 (Ruiz 1424, shapaja) Senna multijuga (Rich.) I.& B. 0 3 119 40 13 22 5 0 0 202 8 (Ruiz 1464, pashaco) Sloanea laxiflora Spruce ex Benth. 0 0 5 0 0 0 0 0 0 5 61 (de Jong 181, sepanchina) Solanum sessiliflorum Dunal 47 1 0 0 0 0 0 0 0 48 26 (de Jong 60, cocona) Solanum stramonifolium Jacq. 19 0 0 0 0 0 0 0 0 19 45 (de Jong 64 coconilla) Spondias mombin L. (uvos) 0 5 0 0 0 0 0 0 0 5 61 S. dulcis L. (taperiba) 0 0 5 0 0 0 0 0 0 5 61 Syzygium malaccensis (L.) Merr.& Perry 38 12 10 0 0 0 10 0 0 70 20 (mamey) Tessaria integrifolia R.& R 0 0 0 0 0 11 0 0 0 11 53 (de Jong 102, pajaro bobo) Theobwma bicolor Humb.& Bonpl. 3 5 5 10 0 0 0 0 0 23 40 (macambo T. cacao L. (de Jong 70, cacao) 3 12 14 0 13 6 0 0 0 48 26 T. obovatum Klotzsch ex Bernoulli 6 0 0 0 0 6 0 0 0 12 50 (de Jong 156, cacahuillo) VERBENACEAE (de Jong 167, mullo 0 1 14 20 40 0 0 0 0 75 18 huayo) Xylopia frutescens Aubl. (de Jong 171, 3 0 0 0 0 0 0 0 0 3 67 espintana) ? (asna panga) 0 0 5 0 0 0 0 0 0 5 61 ? (mamey limena) 0 1 0 0 0 0 0 0 0 1 72 ? (zorro caspi) 0 0 0 10 0 0 0 0 0 10 55

Total individuals 1042 254 1370 2390 478 1347 536 132 15 Total species 40 43 31 23 18 23 15 6 3

119 thatching,togethe rwit htimare o (FLACOURTIACEAE, sp.indet.) ,a constructio n roundwoo dspecies ,Ormosia amazonica, yielding fruits used for handicrafts, and Mansoa alliacea, amedicina l herb. Inth eothe rmixe d orchard(fiel d B) Myrciaria dubia andSpondias mombin weregrown . Thesear ebot hedibl efruit sfrom neigh ­ boring forests whichar eharveste d forconsumptio nan d salei nth emarke t(Peter s 1990; Peters &Vasque z 1989). Other managed forest species from this field are Senna multijuga, mullo huayo (VERBENACEAE, sp. indet.), and Margaritaria nobilis, which areuse d for construction. The number of species in each of the ninefloodplain agroforestr y fields rangedfrom 6 8i nth emixe dorchar dfield (fiel d B)t ofive in th ecor n agroforestry field (field I;Tabl e 7.1). The large number of managed species inth e mixed or­ chard is a result using thisfield a s both a mixed orchard and as a house garden. However, when analyzed separately, the agroforestry section of field B had 43 species. Thisnumbe r doesno t differ greatlyfrom th e number ofspecie s found in other mixed orchards in Santa Rosa (see chapter6) . The important role of native forest species inYanallp a agroforestry is also reflected inth edistributio no fal lusefu l plantsi nal fieldsl surveyed . Noon especie s was found in all nine fields. However, three species were found in seven fields: plantains, Cedrela odorata, and Genipa americana.Th e latter isa nativespecie s found in a flooded forest habitat but which is commonly planted. Twotre e spe­ cies, Mauritiaflexuosa an d Senna multijuga, both also native to the floodplain, werepresen t insi x ofth e nine agroforestry fields. Atota l of 32o fth e7 8specie s were present each in only one of the nine agroforestry fields surveyed.

Weeding patterns and yield levels The intensive use of thetw o mixed orchards (agroforestryfields A and B) allowed very little weed invasion. Infrequent clearings kept most of the area of thetw ofields totall y weed free. Products wereharveste d wheneveravailabl ean d needed. Fruit production of some species like the Citrus paradisii n the mixed orchard offield B exceeded household demand and some of the surplus produc­ tion was sold. Some of the fruits, however, were not harvested at all. Cedrela

120 odorata trees are harvested only occasionally. The owner of mixed tree agroforestry field (field C) spent very little time weeding. Selected tree species, especially Calycophyllum spruceanum, werelib ­ eratedfrom surroundin g vegetation and occasionally cut. Fruit species werehar ­ vested whenever they yielded fruits. After the slash-weeding in 1987, however, the field was planted with manioc and plantain and was managed as a restinga chacra. Some of the smaller Calycophyllum spruceanum trees were tended just like they had been before the slash-weeding. The Cedrela odorata/Calycophyllumspruceanum agroforestryfield (fiel dD) , andth eboundary-stri pagroforestr yfield (fiel d E)receive dlittl eattentio nfrom thei r owners. TheCedrela odoratafield ha dno tbee nweede dfo ra lon gtime ,althoug h theowne rreporte dtha th ewoul dsoo nstar tclearin garea saroun dsom eo fth eceda r trees. Someincidenta lharvestin g oftree s for variousdomesti cuse soccurre d from this field. The Cedrela odorata trees were not yet large enough to be harvested for timber. Inth e boundary-strip agroforestry field (field E)onl yfruits wer ehar ­ vested. Weeding was onlycarrie d out atth e edges of thefield t opreven t theun - derstory vegetationfrom enterin g neighboring fields. The largerpar t ofth eplantai n agroforestryfield (fiel d F)wa sslash-weede d about oncepe ryear ,jus t likean yregula rresting aplantai nfield (se echapte r four). Asmal lpar t ofthi sfield receive d lesscare ,resultin g ina higher density of spon­ taneously occurring tree species. Plantains were harvested and replanted. Some ofth e smaller trees wereread yt o be used asconstructio n material, but the larger Cedrela odorata trees were not harvested yet. During the time of the inventory a dense weedy vegetation of grasses and kudzu {Calopogonium caeruleum) covered the second plantain agroforestryfield (field G). Shortlyafte r thefield wa ssurveye d the owner started a slash-weeding, andgraduall yth ewhol efield wa scleared . In 1989th e wholefield wa sstil lwith ­ outan ysignifican t weedvegetation . Theowne rreporte dtha th econducte da simila r clearing inthi sfield abou t everyfou r years, at which time heals o replanted most of the plantains and some Cedrela odorata trees. The plantains were harvested whenever they yielded products.

121 Themanagemen t activities ofth emixe dcor nagroforestr yfields focuse d on the corn crop. Weeding, planting and harvesting ofcor n followed a well-defined schedule (seechapte r four). No management of Cedrela odorata trees was done othertha n the harvesting of the 35tree s in 1989. No replanting oftree s in either of thefields wa s observed.

VARZEAAGROFORESTR Y STRATEGIES

Theagroforestr yfields discusse dabove ,ar eessentiall yo ftw odifferen t types. FieldsA throug h Eca nb e described as forest gardens since the larger part ofth e vegetation consists oftrees . Inthes efields th efocu s isprincipall y ontre eproduc ­ tion, albeit incombinatio n with a significant number ofnon-tre e species in fields Aan d B. Onth e other hand infields F throug h Ith e main emphasis lies onpro ­ ducingplantain , maniocan dcorn . Treesar egrow n inadditio n tothes eannua l or short-lived crops. Thecombination s ofcommo nagricultura lcrop swit htree si nfields F throug h Iar eonl ypossibl ebecaus eth esoil s ofthes efields allo wsuc ha mixture . Farmers choose to include trees inthes efields becaus e they seetha t the increase invalu e represented byth etree s outweighs the inherent loss incurred to theannua lcrops . Trees in these fields are maintained at lower densities than in the plantain agroforestryfields becaus e corn is less shade tolerant thanplantain . Most ofth e trees in thesefields ar e Cedrela odorata trees. This timber species has an open crown and sheds its leaves shortly after the high waterperiod ,jus t when the first corn crop is planted. As a result Cedrela odorata trees produce less shade than many other treespecies . Infields wit h a highchanc e of flooding crop combinations othertha ncor n and Cedrela odorata, such as plantain with trees (likefields F and G) or mixed orchards(lik efields A andB )ar et orisky . Anotherreaso n Cedrela odorata greatly dominatesa sa tre ecomponen t inresting acor nfields is becaus ethi s speciesha sa highcommercia l value. This species isgrow n fortimber , and soldt oth esawmil l in Requena. Sincecor n isa principa l cash crop, farmers are less willing to inter-

122 plant trees that will not increase economic returnsfrom th e field. Combinations ofothe rope ncrowne dtre e species,an d corn ormanio c isals o reportedfrom Jav a where soil fertility is higher than in most locations where swidden fallow agroforestry is common (van der Poel &va n Dijk 1987). Inth eplantai n agroforestry fields Cedrela odorata trees aregrow ntogethe r with several other tree species many of which are destined for local use. Plan­ tains grown onth e fertile varzea soils are less affected by shade than a corncro p and can grow under a more dense canopy cover. Plantains are a less important cash crop than corn. The additional returns of varioustre e products destined for household consumptionjustifie s a decrease in the production ofth e plantain. Ofth efores t gardens described above,th e mixed orchards (fields Aan dB ) occur onth e higherpart s ofth e levee,an d could not occur onsite s whichexperi ­ encemor efrequent flooding, sinc ethi s would too often destroy many ofth enon - tree species present. Tree species which are common in upland forest gardens, butwhic hd ono ttolerat eth e infrequentflooding, ar eabsen fromt thes etw o fields. Althoughspecie scomposition sar edifferen t ineach ,th eYanallp aagroforestr y fields A and B provide the owners with a similar range of products as the orchards in SantaRosa , orothe rhome-garden s reportedfrom elsewher e(e.g . Soemarwotoan d Soemarwoto 1984;Christant y etal . 1986;Wiersu m 1983;Fernande s etal . 1987). During most of the life cycle of the mixed-tree agroforestry field (field C) trees weregrow n incombinatio n with annual cropso rmanioc . Duringth eperio d without manioc, the emphasis was shifted totre eproduction . Whenth efield ha d been cleared again the larger Cedrela odorata trees weremaintaine d and sower e some ofth e smaller Calycophyllum spruceanum trees. Hence inthi sfield perma ­ nent tree crops are combined with a cyclic planting of manioc and plantain. In fact, the field is partly fallowed for manioc and plantain. During such a quasi fallow period attention to aspects oftre e production, especially ofCalycophyllum spruceanum, is intensified. The field dominated by the Cedrela odorata /Calycophyllum spruceanum vegetation (field D) is managed solely for the production of these tree species, although someadditiona ltre especie sar eallowe dt ogrow . The sitecoul db euse d

123 for corn production, which would provide much faster monetary returns, but the ownerpreferre d not to because he had enough other fields in which hecoul dpro ­ duce rice and corn. If the owner would not have enough land, this field would probably be converted into a mixed com/Cedrela odorata agroforestry field. A mixed orchard, orplantain/Cedrela odorata agroforestry isno t viableo nthi ssite , because it is subject to frequent flooding. Sinceth emai n function ofth eboundary-stri p agroforestryfield (fiel d E)i s to separate areas of land belonging to different owners, economic returns are less important than inth e other fields. As aconsequenc e thesesite s generally receive little attention and weeding from their owners. In many suchfields Artocarpus incisa hasa prominen tpresence . This species, originally from Asia, isa n impor­ tant food source when no other staples are available. A shortage of staples may arise whena very high flood destroys all plantainfields, a s occurred in 1986. In time of abundance, fruits ofthi s tree species are fed to domestic animals.

OPTIONS FORVARZE AAGROFORESTR Y DEVELOPMENT

Thespecia lcircumstance s inth efloodplain allo wsevera l other agroforestry options which are not feasible in terra firme. Since there is no permanent crop­ ping ofannual s on the latter, combinations of short cycle crops and trees areun ­ usual. Occasionally, terrafirme farmer s leave trees standing when slashing new swiddens (Denevan et al. 1984), and such practices are also found among Santa Rosafarmers . However,eve nsuc h initially openagroforestr y swiddens willeven ­ tually develop intoclose d forest gardens. Henceth e most obvious difference be­ tween Santa Rosa and Yanallpa agroforestry is the permanent mixture of annual crops and tree crops in the latter. Thesecon d notabledifferenc e betweenYanallp aan d SantaRos aagroforestr y is the larger role of native forest species in Yanallpa. Cedrelaodorata and Calycophyllum spruceanum, the native forest species most common inYanallp a agroforestry appear spontaneously infields i n the varzea and are easily incorpo­ rated inth emanagement . Yanallpafarmers ' strongerfocu s onthes especie si sals o

124 explainedb ythei rdecrease dacces st oforest s asa sourc eo fproduct sfo rthei rdail y household use. It therefore makes economic sense to grow such species, even if limitedfield spac emus tb euse dfo rsuc ha purpose . Furthermore, Cedrela odorata hasa marke t value,grow sfas t inth e varzea lands, and can begrow n incombina ­ tions with many other crops, including light demanding ones like corn or even rice. Agroforestry species in the two villages have a similar range of uses, but market-oriented timber production is more important inYanallpa . Thevariatio n inagroforestr y practices inYanallp aallow s for moredevelop ­ mentpossibilitie stha ni nSant aRosa . Anincreas eo fcas hincom eca nb eachieve d through improvedproductio n ofCedrela odorata. Somesimpl esilvicultura lmea ­ sures like pruning, or selection of improved planting material achieve this. Be­ sides improving existingpractices ,combinin g Cedrela odorata withdifferen t cash crops should be tried. Moreover, production for household consumption, or for salei nYanallp a itselfo ri nneighborin gvillages ,ma y improveth e incomeo fsom e farmers. The lack of forest resources within close vicinity ofYanallpa , and the interest offarmer s toproduc ethos e forest products themselves indicatestha t such agroforestry practices oriented towards household consumption are possible in Yanallpa. Production of Calycophyllum spruceanum, for instance could satisfy both a local and a regional need. Just like in Santa Rosa, agroforestry inYanallp a will continue to bea vari ­ able business. No single improved system can possibly be earmarked asth ebes t and only alternative. Rather, several of the practices discussed above should be useda sa bas efo rth edevelopmen t ofimprove dsystem swhic hca n increaseeithe r cash income orth esuppl yo f locallyneede d resources. Itshoul db ekep t inmind , however,tha t inYanallpa , anyne wagroforestr y strategy will always havet ocom ­ petewit hothe rland-us eoptions . Newagroforestr y practices will onlyb eadopte d if they provide returns which are valued higher than returns from other agricul­ tural types.

125 CHAPTER EIGHT

CONCLUSIONS AND PROSPECTIVE

INTRODUCTION

The fate ofth e world's natural resources is currently a great concern. The potential consequences ofth e unsustainable use ofthes e resources for largepart s ofth epopulations ,especiall yo flesse rdevelope d countriesurgentl y demandmodi ­ fication ofcurren t ways and levels of exploitation. Todaysproblem s of resource depletion, degrading environments,an drelate d decreases ofth e living standard of manyrura lpopulation sar ealread ysuc htha tdirec tinterventio n isneede d(Nationa l ResearchCounci l 1992). However,a sa resul to fth eincrease dcomplexit y ofprob ­ lems concerning the sustainable use of natural resources, solutions must become moresophisticated . Inman ycases , suchsolution s havet ob esite-specific , requir­ inga profoun d understanding ofloca lcondition s andpossibilities . Consequently, research which provides the required information for such understanding remains paramount. In this book a study is reported of the diversified use of natural resources by riberenos ofth e PeruvianAmazon . Thevarzea ,o rperiodicall y flooded plains alongth eAmazon ,ar eno wbecomin g an important area for agricultural develop­ ment. Unlessmethod s aredevelope dt ous ethi sare awisely ,th e increasedexploi ­ tationo fth efloodplain wil llea dt obot henvironmenta lan dsocia ldestruction . This present studyattempt st ocontribut et obasi c information onth ecomple x resource management practices ofloca lfarmer s whoinhabi t and exploit varzealands . Iti s hoped that, by learning their methods, wewil l be able to benefit from the riches of theAmazo n landscape without ending up overtaxing them. Most research inth e varzeaha sbee nconducte d bytw o different groups of scientists. Onth eon ehand ,agronomist s anddeveloper sbecam eintereste d inthi s ecological zonea sa possibl e location for thedevelopmen t ofmoder nagriculture .

126 Their interest wasmainl y triggered byth e fertile soils ofth evarzea , comparedt o the lowfertilit y ofth eterr afirme soils . Onth eothe r hand,archaeologist s studied the varzea as the main route of human occupation of lowlandAmazonia . Scien­ tists of both groups formulated theories concerning resource utilization of the Amazon floodplains, some of which have been summarized in chapter one. It is assumed inthi s booktha t examples ofcontemporar y varzearesourc e exploitation cancontribut e toth etheorie s about future orpas t varzea exploitation. Suchtheo ­ ries have insufficiently been tested against contemporary use. This is, to a large extent, a result of the general belief that present varzea resource use is virtually absent (e.g. Ross 1978). Other studies (e.g. Hiraoka 1985a,b, Bergman 1974),a s well as the present study proves such belief to beuntrue . In this book, resource management among riberenos was investigated. Riberenosar erural ,non-tribal , indigenouspeopl ewh oliv ei nisolate dvillage smos t ofwhic har elocate dalon gth elarge rrive rsystem so fth ePeruvia nAmazon . Among these people variability can be observed in, for instance, degree of acculturation, local history, and involvement with regional economies. However, a number of general common features justify their distinction asa separate economic andcul ­ tural group (Chibnik 1991). Riberenos practice agriculture, fishing, hunting and collecting for both household consumption and markets. Theirresourc emanage ­ ment practices, however, havebee n ignored by the scientific community untilre ­ cently. Sinceriberenos ar e the largest group of rural inhabitants in the Peruvian Amazon, they play an important role inth e regional economy. This role can be expected to increase inth e future. Riberenos are also the group that will mostly be affected by any expansions of government sponsored industrial agriculture. Development of the varzea in the Peruvian Amazon should start in collaboration withribereno farmers . This will not only lead to a more equally distributed de­ velopment,bu ti twil lals obenefi t from anaccumulate d experiencean dunderstand ­ ing of the environment and its potentials. Inthi slas tchapte rth emos t importantfindings an dthei rimplicatio n forsom e of the theories that have been formulated earlier concerning varzea resource ex­ ploitation willb esummarized . Theresearc hwhic h isa tth ebas eo fthi sboo kwil l

127 briefly beassessed . Finally some thoughts will begive n to possible scenarios of the future ofriberefio agriculture , and the possible implications of this study for development or extension efforts among theriverine populatio n of the Peruvian Amazon.

SUMMARY OFFINDING S

Agricultural diversity This study investigated the adaptation ofriberenos t o a complex and dy­ namic varzea and adjacent terra firme. Specifically, it documented and analyzed diversity inagricultura l practices, variation inagricultura l strategies,an d diversity in agroforestry practices in tworibereno communitie s in the lower Ucayali river ofPeru . Agricultural diversity, variation inagricultura l strategies,an ddiversit yi n agroforestry practices couldadequatel yb edemonstrated , butcoul d only partlyb e explained. This is not a consequence of insufficient, or inappropriate research, but rather ofth e complexity ofribereno agricultura l practices. The principal factor determining the complexity ofriberefio agricultur e is theecologica ldivers eenvironmen ti nwhic hriberefios live . Especially inth e varzea, the land forms which are appropriate for agriculture can be planted to different crops,an drequir etotall ydifferen t farming techniques. Asa consequenc eagricul ­ tural practices vary according to land form. Among the economic factors which influence agricultural complexity, market opportunities are essential. Riberenos areth e suppliers of manyagricultura l and forest products for local markets. De­ mands for products and prices vary andriberefios ar e sensitive to such fluctua­ tions. One of the social factors which increase agricultural complexity in the PeruvianAmazo ni sth ehig hmobilit y ofriberenos. Man yo fthes eriverine peopl e have lived inmor etha n one location during their life time. Migrating to thecit y and back toth e countryside, but also relocation to a different village iscommon . This mobility also leads to an increased complexity ofriberefio agriculture . Agricultural diversityamon griberefios coul db eadequatel ydemonstrate db y describing fourteen different agricultural types. Anagricultura l type is defined as

128 a unique site-crop combination which has its own specific set of agro-economic characteristics. Fouro fthes eagricultura l types were located onterr afirme land s and ten in the fioodplain. Ecological conditions of sites, as well as economic opportunities determine the crops whichriberenos wil l grow, and where. Aspe ­ cific site and crop combination implies its own specific allocation of labor, eco­ nomic returns, andrisk o f failure. Alarg epar t ofth eprimar y research described inth epresen tboo kfocuse d onquantifyin g theagro-ecologica l andagro-economi c characteristics of the fourteen agricultural types. Agricultural diversity isa commo ncharacteristi co fribereno agriculture . By definition, this characteristic is neither uniform nor static. Since ecological con­ ditions as well as economic opportunities vary in almost any other place in the Peruvian Amazon, it isunlikel y that in otherribereno village s any identical form of diversity will be found. This appeared, for instance, from the comparison of agricultural diversity in Santa Rosa andYanallpa . It is also unlikely that in the future the same agricultural types which are described here will continue to have the sameimportanc e in SantaRos aan dYanallpa :whil eth eecologica l complexity of the environment will probably remain the same, economic opportunities will change. Consequently agricultural practices will and doalter . Thiscoul dalread y be observed between 1985an d 1989.

Variation in agricultural strategies Asecon d levelo fagricultura lcomplexit y isth evariatio ni nagricultura lstrat ­ egies among individualribereno farmers . Variation of agricultural strategies is a result ofindividua l responseso ffarmer s toth eman yoption s available,an dthere ­ fore a direct result of the agricultural diversity. Ribereno farmers are unique in that the variation in agricultural strategies that can be found within a single vil­ lage exceeds case studies reported from elsewhere (e.g. Barlett 1982). The large variation inagricultura l strategies amongriberenos mad e it diffi­ cult to understand the rules which govern this phenomenon. No one variableex ­ plains all the observed variation. Instead, various interacting factors influenced the wayriberenos farme d in Santa Rosaan dYanallpa . Riberenos generally grow

129 their ownstapl e food, mainly maniocan dplantains ,an dthe y producecrop s tob e sold in the market. Production of food for household consumption and growing crops for monetary income are mostly carried out in different agricultural types. Agriculturalstrategie sar efurthe r influenced byavailabilit y oflabor ,acces st oland , availability ofban k loans, and the willingnes to take risk. Not all farmers have their principal food-producing fields on terrafirme, although these sites haveth e lowest risk for growing manioc. Togai n easierac ­ cess to lands, many farmers relocate both food and cash crop production to the floodplain.I fthi shappens ,tha nfoo d production isofte n carriedou t inmor etha n oneagricultura l type. This lowers therisk tha t all thefields wit h food cropswil l be innundated, which would leaveth e household without a daily staple. InSant aRos aan dYanallpa ,mos t ofth ecas hcroppin goccurre d inth e flood- plain. Principal cashcrops , during thetim e ofresearch , wererice, corn ,an dt o a lesser extent plantains,jute , and vegetables. Ecological factors largely accounted for the difference of rice being the principal cash crop in Santa Rosa, while in Yanallpacorn ,ric ean dplantai n wereth eprincipa lcas hcrops . InSant aRos amor e land was appropriate for rice production than inYanallpa . Riberefio farmers ac­ cepted the higherrisk o f growing rice in barreales, the annually appearing mud­ flats, as atrade-of f for higher returns for invested labor. However, only a limited numbero ffarmer s had accesst othi spreferre d landtype . Thosewh odi dno thav e accesst obarreale s hadt orel yo nothe ragricultura l typest o obtaincash . Farmers havet owai t sometimes for manyyear sbefor e theyca nbu ya barrea l from another farmer, orunti l newbarreale s are formed byriver dynamics .

Diversity inriberefio agroforestr y Agroforestry practicesdiffere d betweenSant aRos aan dYanallpa , andi neac h villagethe y showed important variation. In SantaRosa ,variatio n in swidden-fal- low agroforestry could be demonstrated by comparing the densities of managed plant species, weeding patterns, and production levels of a number of terra firme agroforestry fields. Although some of the forest gardens in Santa Rosa were in­ tensivelytended ,fores t gardens with verylo wmanagemen t intensity stillprovide d important economic returns from species planted in the previous swiddens. The

130 more important role of swidden-fallow agroforestry in Santa Rosa, compared to other locations from Peru and Ecuador (e.g.Deneva n & Padoch 1987;Irvin e 1987) is a consequence of the higher pressure on forest resources in Santa Rosa, and a higherinvolvemen t ofriberen o farmers inth eregiona l market economy. This results in an economic incentive to produce a number of fruit species and forest products in forest gardens, which therefore are given more attention. The variation in agroforestry practices within the village is a result from these practices being ori­ ented toward production for household consumption. As a consequence, farmers have different preferences in how much labor they want to invest in forest gar­ dens. Santa Rosa inhabitants will not easily find possibilities to become market- oriented agroforesters, as riberenos who live close to Iquitos have (Padoch et al. 1985; Hiraoka 1986). In Yanallpa, fewer farmers had agroforestry fields than in Santa Rosa. In Yanallpa arable land is scarce, while alternative land-use options always exist for fields which are destined for tree growing. Agroforestry fields inYanallp a had a more permanent character than in Santa Rosa, but were also more integrated with other agricultural activities. Agroforestry practices inYanallp a included the grow­ ing of corn, manioc, and plantains together with trees in single fields. Growing of Cedrela odorata trees for commercial purposes was more important in Yanallpa than in Santa Rosa. Among agronomists, agroforestry in the floodplain is still quite novel and unexplored, but it appears to have great potential for further de­ velopment. Especially the combination of trees with other agricultural crops of­ fers an interesting prospective.

IMPLICATIONS FOR THEORIES ON VARZEA RESOURCE USE

Varzea agriculture in Peru versus Brazil The contradiction between the complex and diversified use of the varzea by riberenos in Peru on the one hand, and the limited uses of this ecological zone which is found among caboclos in Brazil, (Wagley 1953;Ros s 1987;Frechion e et

131 al. 1989;Bunke r 1981; Wesche 1985) is remarkable and requires an explanation. More recent studies conducted around Manaus (Bahri et al. 1991)indicat etha t at least incertai n locations the Brazilianriverine population s have developed diver­ sified agricultural activities comparable to those found in the Peruvian Amazon. Itshoul d notb eexclude dtha t somereport s ofth eabsenc eo fvarze aagricultur e in Brazil werebase d on insufficient field work. But even if Brazilian varzeaactivi ­ ties are still insufficiently documented, then it still appears that utilization ofth e Amazonian floodplains by non-tribal indigenous populations is more developed in Peru than in many parts in Brazil. The observed dissimilarity in varzea occupation and use can be explained byth e differences inhistorica l and economicprocesse s inth ePeruvia n versusth e Brazilian Amazon. Although the native populations in both regions suffered se­ vereimpact sresultin g from Europeans exploiting forest productsan dslaves ,thes e impactswer emor esever ea ta nearlie rtim ei nth eBrazilia nAmazo n(Parke r 1981; Meyer 1988). During thenineteent h centuryth e floodplain ofth ewhol eAmazo n region had low population densities. The populations of the Brazilian region, however, hadalread yexperience d moreprofoun d culturalchange stha nth epeopl e from the upperAmazon . The impact of Europeans onth e lives of upperAmazonia n groups changed dramatically with thecomin g of the rubber boom around the turn ofth ecentury . Duringthi sperio dth eloca lpopulation s inth euppe rAmazo nexperience da nimpac t which was at least as severe as groups inth e Brazilian Amazon had undergone a century ortw o earlier. Therubbe rboo m also brought a largenumbe ro fmigrant s from elsewheret oth eregion . Onceexploitatio no frubbe rha dceased ,a larg egrou p ofpeopl ewh oha dcom et oth efloodplai n toparticipat e inrubbe rexploitatio nstaye d inth earea . Duringth e following decades, many local dwellers lived andworke d on largeragricultura l estates which emerged following the eloxpitation ofrubber . When theseriver dweller s became independent farmers they were acclimated to varzea agriculture, and to active participation in regional economies. Since the nineteen-sixties, market opportunities for agricultural products have improved as a result of increased urban populations in small and large cities, as well as im-

132 provedrive rtransportation . Duringth elat esixtie san dearl yseventie s thegovern ­ ment promoted commercial agriculture in the region. As a response, riberefios experimented with new agricultural methods, and with the cultivation ofconven ­ tionalan dnove lcrops ,ofte n onsite swhic hha dno tbee nuse dfo ragricultur ebefore . This lead to the present diversification of agricultural activities, as described in this study.

Requirement of complex social organization for varzea exploitation Meggers (1974),an d with her Ross (1978),an d Roosevelt (1980),hav ear ­ gued that pre-contact groups could subsist in the floodplain because they had chiefdom like social organizations in which persons of higher status coordinated subsistence activities. Such coordination was considered necessary to be able to copewit hth econstraint so fth evarzea ,especiall ywit hth eunpredictabilit yo ffloods . This lack of social organization was one ofth ereason s which, according toRos s (1987), accounted for the lack of any significant agriculture inth e floodplain. Thisstud ypresent sa nexampl eo fcomplexit y ofriberefio agricultur ewhic h contradicts that acomple x social organization isa requiremen t for varzeaagricul ­ ture. Not onlyd oriberefios operat ea s independent farmers,thei ragricultura l strat­ egiesar eals oindividualistic . Technologicaldifference s betweenpre-contac tgroups , andcontemporar yriberefios ar estil lminimal . Althoughriberefios d odiversif y their agriculture partlybecaus ethe yproduc e for an external market,thi s only increases thecomplexit y ofthei ragricultur e further, as has been arguedabove . Hencepre - contact agriculture wasprobabl y lesscomplex . Themor ecomple x varzeaexploi ­ tation does not hinderriberefios in still continuing their independent farming. Unfortunately therear en oexac tan dcomparabl efigures o npopulatio nden ­ sities, but it is questionable whether, during pre-contact periods in the upper Amazon, population densities of floodplain groups were higher than they areto ­ day. Atpresen t the PeruvianAmazo n sites whichar eappropriat e for livingar eal l occupied. Ifdurin gpre-contac tperiod ssuc hhighe rdensitie swoul dhav eoccurred , the distribution of limited resources, likelan d orharvests ,ma yhav erequire dper ­ sons with higher social status. It would not have made it impossible for farmers

133 to operate independently from the largergroup . Nocomple x societies withsocia l stratification are necessary to practice complex agriculture in the floodplain. It could evenb e argued that individualistic farming is more efficient inth e floodplain. Theus e ofsevera l land forms not only allows farmers to differentiate production, butals ot o spread laboradequatel y sotha t itbes tfits th efarmer s own laborpoo lan dt ominimiz eth eris ko fcro pfailure . Someo fth eagricultura ltype s require precise scheduling, since only a very limited time isavailabl e to produce a certaincro p ina certai n site. Giventh ecomplexit y ofth e environment, agricul­ tural decisionsar esite-specific . This requiresa precis e monitoring ofagricultura l fields. Ifdecision sconcernin gagricultura l activitieswoul db emad eb ya nauthor ­ ityo rth e whole group,the n the process of information gathering, decisions mak­ ing,an dimplementatio nwoul dnecessaril yhav etake nlonger . Sucha proces s would therefore likely make agricultural activities in the floodplain less successful than single-household farming.

Dependence of pre-contact groups on corn protein Data on the diversity of agricultural methods challenge the ecological evi­ dencewhic hRoosevel t (1980)use dt osuppor the rhypothesi sconcernin gth eseed - crop based subsistence strategies of pre-contact groups. As shown by Bergman (1974; 1980), Hiraoka (1985a,b, 1986), and in the present book's chapter four, production ofcor nan dmanio car eno ta tal lincompatibl e inth efloodplain . Many higherresting asite sar esand yan dar einundate donl ydurin gextremel yhig h floods. They allow manioc production for as long as 10months , while production levels of single crops areth e same as interr afirme fields. Althoug h manyriberenos in Santa Rosa and Yanallpa grew corn and rice in 1985/1986 and 1989, they sub­ sisted largely on manioc or plantain together with fish, even during high water periods. A one hectarefield plante d with regional corn varieties yields no more than 1800 kg, orabou t 180 kg of low quality protein. This production however requires no less than 85 labor days with thetechnolog y used byriberenos today . This isth eequivalen t ofa rati o of2. 1k go fprotei npe rlabo r day. Reportedcatc h levelso fShipib ovarze ainhabitant sexcee d0. 5k goffis h perhou r(Bergma n 1974),

134 an efficiency rate much higher than in corn protein production. It is unlikely that on a regional level total fishing catch levels during pre- contact periods were higher than they are today. Today commercial fishermen, using modern equipment, providefish fo r largepopulation s in cities likeIquitos , Manaus and Pucallpa during the whole year. Even if pre-contact groups lived in higher densities inth e varzea, withth e fishing techniques they had availablethe y probably did not overfish their fishing grounds to sucha n extent that catchlevels , as reported among Shipibo fishermen, were severely reduced. Adense rpre-con ­ tact population most likely could subsist well on a fish and manioc diet, as do mostriverine peopl e inPer u today. Corn obviously wasa n important part ofpre - contact diets (Roosevelt 1980; 1989),bu t the ecological evidence contradicts the hypothesis that larger pre-contact populations could only develop because they changed to a seed-based diet.

EVALUATION OF THE RESEARCH

In this study agricultural diversity and variation in agricultural strategies amongribereno farmer s in Peru has been demonstrated. This means that itsob ­ jectives havebee n fulfilled. In ordert o grasp the complexity ofribereno agricul ­ ture it was necessary tofind a n organizing concept. The concept of "agricultural type"ha sprove nt ob ea nadequat etoo l for bothth estud yo fagricultura l diversity as well as variation in agricultural strategies. Usingth e agricultural typeconcep t the many farming methods observed amongriberenos in Santa Rosaan dYanallp a can adequately be described, and their agro-ecological as well as agro-economic features can be characterized. Each ofth eagricultura l types described hasa specifi c set ofagro-economi c characteristics. Afarme r whocombine ssevera l specific agriculturaltype sno tonl y chooses certain production types, but also decides for a specific allocation ofre ­ sources. Agricultural strategies were defined as the individual allocation ofsuc h resources,an dtherefor e theagricultura l typeconcep t also wasprove nt ob e useful in the analysis of variation in agricultural strategies. The agricultural type con-

135 cept could be used for both the analysis of agricultural diversity and variation in agricultural strategies,becaus e itsdefinitio n assumesa functiona l relationbetwee n the technological features of agricultural production types and the related agro- economic characteristics. Participant observation and interviewing of farmers wereth emethod suse d for the quantification ofth eagro-ecologica l and agro-economic characteristics of agriculturaltypes . Thesequantification s couldb e improvedusin g directmeasure ­ ments of a sufficient large selection offields belongin g to each agricultural type. This would require much more time and more researchers. Because of the large number of agricultural types described, it was not possible to do such measure­ mentsi nthi s study. Sincesom eo fth eagricultura ltype swer emor ecommo ntha n others,mor edat acoul db egathere d onth ecommo n ones. Hence,th eaccurac yo f information presented here differs for each agricultural type. A more thorough quantification ofribereno agricultura l types will also re­ sult in a better understanding of the variation of agricultural strategies among ribereno farmers, since the economic implications of certain choices will be un­ derstood better. However, since agricultural strategies are very individualistic, as has been demonstrated in chapter five, a much closer monitoring of individual farmers will havet ob econducte d tob eabl et o fully understand agricultural strat­ egies. It is questionable whetherribereno agricultur e can ever be understood so welltha t farmer's behaviorca nb epredicted . Riberenosar ever y dynamic inthei r agricultural strategiesbecaus ethe yhav et ob ever yopportunisti ca sa consequenc e of the highly variable ecological and economic environment. In the discussion of diversity in agroforestry practices, the composition of agroforestry fields, their yield levels, and the weeding schedules followed were compared. Forth ecompariso n ofcompositio n ofagroforestr yfields a distinction was made between managed and non-managed vegetation. Swidden-fallow agroforestryfields ar ecovere d witha mi x oftrees , shrubs,an d weeds. Some,bu t inmos tcase s notal l ofthi s vegetation ispurposel y andactivel ytended . Thepro ­ cedurefo rth edistinctio n ofth emanage d versus non-managedvegetatio nha sbee n described inchapte rtwo .

136 The comparison of the composition and yield levels of only the managed vegetationexclude sa non-manage d vegetationwhic hma yb econsiderabl e insom e forest gardens. Such vegetation does have an economic importance. However, comparing only the managed vegetation is consistent with the practice of forest garden exploitation used by the Santa Rosa and Yanallpa inhabitants. For this exploitation the wordpurmear isused . Purmear means going to a forest garden toharves tproduct swhic hon eknow sar ether ebecaus ethe ywer eplante do rtended . Collecting other forests products from the spontaneous non-managed vegetation occurs but, at least in Santa Rosa and Yanallpa, it does not have the same eco­ nomic importance as harvesting managed species. Acompariso n ofth e managed vegetation of Santa Rosa andYanallp a forests gardens is therefore in fact acom ­ parison of the relative economic importance of these forest gardens. Weeding patterns of the Santa Rosa terra firme forest gardens were mea­ sured by looking at changes in the status of the non-managed weed vegetation. The relative cover and height of the non-managed vegetation were estimated ev­ ery three months for single fields. From the changes observed a management patternwa sderived . Theprocedur euse dfo rthi sestimatio nha sals obee nexplaine d inchapte rtwo . Anove lmetho dt oestimat eyiel dlevel so fagroforestr yfields wa sals oimple ­ mented. Everythre emonth si ntransect s inagroforestr yfields, th eproduc ewhic h could be harvested from the field in the next three months was estimated. This methodallowe d to obtain yield estimates without havingt o rely onmor e difficult methods likelabelin go ffruits, usin gfruit traps ,o rrecordin g ofproduct sharveste d by users of forest gardens. Marking and collecting fruits has to be done with painstaking precision, and therefore is more time-consuming. Even the latter method stillca n leadt o errorswhen , for instance, fruits areharveste db ychildren , or other people whopas s through the forest. Recording of produce harvested by owners is much less accurate since usually many people, other than the owner, harvesfromt fores t gardens. Estimatingfutur e production forth enex tthre emonth s is lessaccurat e than counting, but errors are within acceptable limits for the pur­ poses required.

137 THE FUTURE OFVARZE AAGRICULTUR E INAMAZONIA N PERU

Acceleratedchange so fth eriberefi o societyan deconom ybega ni nth e1950' s when improved transportation and communication, increasing demands for the areas'resources ,an dgovernmenta l regional integrationan ddevelopmen t attempts began to affect many facets ofribereno lif e and environment (Hiraoka 1985b). Riberenos livelihood activities became adapted for a market-oriented economy. Hiraoka (1985b) expected that inth e Peruvian Amazon the streamlining of agri­ culturalland-us ewoul dincreas edisparitie si nacces st olan dan dwealth ,an dstimu ­ late rural emigration to cities and changing flora and fauna. Among caboclos in Brazil such changes have been reported (Wesche 1985;Bunke r 1981;Schmin k 1985). In the vicinity of Manaus opening of roads to caboclo communities has leadt o" aprogressiv edestabilizatio no fth elivelihoo do fcabocl osociety "(Wesch e 1985: 136). Urbanmigratio n increased,an dcaboclo shav ebecom eparticipant si n an incipient labor class (Wesche 1985). The market oriented caboclo communi­ ties in centralAmazoni a are not able to resist pressure from apowerfu l local en­ trepreneurialclas swhic hdominate sloca leconomie sespeciall yaffectin g traditional caboclo cash activities and taking over most of the land (Bunker 1981). It re­ mainst ob esee nwhethe rsuc hdevelopment swil loccu ri nSant aRos aan dYanallpa . An increased economic stratification of rural communities is taking effect in the area around Iquitos. Increasingly, local entrepreneurs are dominating the agricultural economy. As a consequence land conflicts and even expulsions of farmers from the terrain they farm havebecom e more common (Chibnik 1990). Unclear land tenure, lack of access to bank loans, or unfamiliarity with govern­ mentagencie smak eman yribereno sles scompetitiv etha nlarge rfarmer s from the villageso rth ecity . Thesear e signs ofsimila rproces stha tma y leadt odestabili ­ zationo fribereno livelihoo d asdescribe db yWesch e(1985 )fo r caboclos inBra ­ zil. Changes, as noticed by Hiraoka, inribereno communitie s close to Iquitos began in Santa Rosa and Yanallpa during the mid- and late sixties. However, becauseo fthei rgeographi c isolations,an dbecaus eo frecen tdevelopment s inth e Peruvianeconomy ,changes ,suc ha soccurre damon gcaboclo si nBrazi lan dwhic h

138 are taking place inth e vicinity of Iquitos, are not likely to occur soon in villages on the lower Ucayali river. Although upstream from Requena larger farmers or patrones still appear to dominate local economic life in some villages, the owner of the Monte Carmelo agricultural estate,jus t North of Santa Rosa, had stopped mosto fhi sactivitie sb y 1990. Nolarge rentrepreneur s willfind incentive st oinves t inagricultura l production inthi s region, sincether ear eto o few market opportuni­ ties in acit y like Requena. Transportation costs areto o high to ship agricultural produce to largermarkets . It is also unlikely that increased economic stratification of farmers within any of the villages will occur soon. Net monetary returns for cash crops have decreased thelas tsevera lyears ,especiall y sinceth egovernmen tagencie s incharg e ofbuyin g rice was dissolved. Although an official price still exists for rice, local intermediaries pay as little as half or athir d of this price. The agricultural bank doesno tprovid ean y loanst o smallagriculturis t anymore. Rathertha n increasing income inequality, current tendencies appear to lead to a homogenization of the economic levels of individual riberenos. Becauseo fdeclinin gprice sfo ragricultura lproducts ,income samon griberen o farmers havedroppe d the last several years, and will probably continue to doso . Because of the decrease in economic activities in this part of Peru there will be less economic interchange between the lower Ucayali region and Iquitos. The accelerated urban migration of the early seventies has stopped. Although many young people still go to the city to complete secondary education, their number will probably decrease because of declining inçqjne levels. Changes in Santa Rosa and Yanallpa will mainly be the result of internal regional processes. Of the two villages studied, the largest changes can be ex­ pected to takeplac ei n Santa Rosa. Plans for a roadconnectio n ofthi s villaget o Requena have existed for a long time. Ifthi s road isfinally built , then more in­ habitants ofSant a Rosawil l livei nth ecity ,an d become migrant farmers. Subse­ quently,the ywoul d spend lesstim edoin gagricultura lactivities ,whic hwoul dlea d toa genera ldecreas eo fth ecomplexit y ofthei ragricultura l enterprises. Suchfarm ­ erswoul dprobabl ymaintai non ecas hcro pan da foo dproducin gagricultura ltype ,

139 oreve nonl y oneo fth etwo . Ifn oroa d connection willb eestablishe d soon,the n SantaRos a farmers willcontinu et o lookfo r opportunities to obtaincas hincome , bymean s ofagricultura l or otheractivities . Their agriculture will continue tob e opportunistic, andtherefor e diversified and dynamic,a s it was between 1985an d 1989. The same is also true forYanallp a farmers. Riberefio agriculture can most certainly be improved. Unfortunately, agri­ culturalresearc han dextensio n inth ePeruvia nAmazo n lacksprecis efocu s onth e real problems andpossibilitie s of local farmers and conditions. One ofth emai n researchproject s ofth eMinistr yo fAgricultur econtinuou st ob eproductio no fwate r buffaloes inuplan d and varzeaterrain . Production of buffaloes is not necessarily a bad idea, as long as it is investigated and promoted on a small scale, as was attempted previously at the Jenaro Herrera Research Station. Otherwise it will benefit onlya fe wan dbrin gn odevelopmen t for largergroup s ofriverine farmers . The local station of the Institute Nacional deInvestigation Agricola has no re­ searchproject s inth evarze aye t (Mendoza andPined ounpublished) . Atth eUni ­ versity of Iquitos and at the Institute Veterinario de InvestigationTropical Amazonica afe wpromisin g experimentationan dextensio nproject s arebein gcar ­ riedout . One of these includes, for instance, production of Myrciaria dubia,a riparian shrub ofwhic hth e fruits havea hig hconten t ofvitami n C,bu ti nth epas t has mainly been harvested inth e wild (Peters 1990). It is these kinds of smaller projects that can improveriberefio agriculture . Local research and extension agencies should recognize thatriberefio agricultur e isdiversified , andthe yshoul d offer apackag eo fresearc han dassistanc e forman y ofth e activities that riberenos undertake. Riberenos have no need for largescal e or capital intensive agricultural methods. Caboclos andriberenos ar e the best adapted inhabitants ofth efloodplai n andthu sth emos tvaluabl eintermediarie s for technological change andtransfe r inthi s region (Hiraoka 1992). Butthi stechno ­ logical change should start from existing resource management practices already present in the varzea. Whatriberenos nee d is a development which starts from their problems, knowledge and skills.

140 LITERATURE CITED

Alcorn,Jani sB . 1990. Indigenousagroforestr y strategiesmeetin gfarmers 'needs . In:A.B.Anderso n (ed.)Alternative s to deforestation: Stepstowar d sustainableus e of theAmazo n rainforest, pp 141-151. Columbia University Press,Ne wYork .

Ali,Ab u M.S. 1987. Intensive paddy agriculture in Shyampur, Bangladesh. In: B.L. Turner II and S.B. Brush (ed.) Comparative farming systems, pp 276-306. The Guilford Press,Ne wYork .

Allen, Bryant J. 1985. Dynamics of fallow succession and introduction of ro- busta coffee in shifting cultivation areas in the lowlands of Papua New Guinea. Agroforestry Systems 3:227-238.

Anderson,Anthon yB . 1990. Extraction and forest management byrura l inhab­ itants inth eAmazo nestuary . In:A.B.Anderso n (ed.)Alternative st o deforestation: Steps toward sustainable use ofth eAmazo n rainforest, pp 65-85.Columbi aUni ­ versity Press,Ne wYork .

Anderson, Anthony and MarioAugust o G. Jardim. 1989. Costs and benefits of floodplain forest management by rural inhabitabts in the Amazon Estuary:A case study of açai palm production. In: John O. Browder (ed.) Fragile lands of LatinAmerica , pp 114-129. Westview Press, Boulder.

Anderson,A.B. ,A .Gely ,J . Strudwick,G.L . Sobelan dM.G . Pinto 1985. Um sistemaagrofloresta l navarze a doestuari o amazonico (Unada sOnças , Municipio de Barcarena, Estado do Para).Act aAmazonica , Supplement, 15(1-2):195-224 .

Andriesse, J.P. 1979. From shifting cultivation to agroforestry or permanent agriculture. Royal Tropical Institute Bulletin,303:35-43 .

141 Aramburu, CarlosE . 1984.Expansio n ofth e agrarianan d demographic frontier inth ePeruvia nselva .In :M.Schmin kan dC.H.Wood , (eds.)Frontie rexpansio ni n Amazonia,p p 153-179. University of Florida Press, Gainesville.

Bahri, S. 1988. Les systèmes agroforestiers. In: Conditions écologiques et économiques de la production d'une ile de varzea: File de Careiro, pp 336-347. ORSTOM, INPA,CEEE .

Bahri, S,F .Grenand ,P .Grenand ,J . Guillaumet, M.Lourd . 1991. Lavarze a est-elle un don de l'Amazonie? In: J.Gallais et al. Sahel Nordeste Amazonie: Politiques d'aménagement en miliux fragiles, pp 104-144. UNESCO,Paris .

Balée, W.L. 1984. The persistence of Ka'apor culture. Dissertation Columbia University NewYork . University Microfilms International, AnnArbor .

Barlett,Pegg yF .1977 . Thestructur eo fdecisio nmakin g inPaso .America nEth ­ nologist 4:285-307.

Barlett, Peggy F. 1980a. Adaptive strategies in peasant agricultural production. Ann. Rev.Anthropol . 9:545-573.

Barlett,Pegg yF .1980b . Introduction:Developmen t issuesan deconomi canthro ­ pology. In: Peggy Barlett, (ed.) Agricultural decision making: Anthropological contributions to rural development, pp 1-18. Academic Press,Ne wYork .

Barlett,Pegg yF .1982 .Agricultura lchoic ean dchange :Decisio nmakin gi na Cost a Rican community. Rutgers University Press,Ne wBrunswick .

Barrow, C.J. 1985. The development of the varzeas (floodlands) of Brazilian Amazonia. In: J. Hemming (ed.) Change in theAmazo n basin vol 1: Man' s im­ pact on forests andrivers, p p 108-128.Mancheste r University Press, Manchester.

142 Beer,J. ,R .Borel ,an dA .Bonneman . 1990.On-far m agroforestry researchplan ­ ning in Costa Rica. In:W.W .Bud d et al. (eds.) Planning forAgroforestry , pp58 - 78. Elsevier,Amsterdam .

Bergman,Ronal dW .1974 . Shipibosubsistenc e inth euppe rAmazo n rainforest. Dissertation dep. Geography Unversity of Wisconsin. University Microfilms In­ ternational, AnnArbor .

Bergman, Ronald W. 1980. Amazon economics:Th e simplicity of Shipibo In­ dian wealth. Dellplain Latin American Studies 6. University Microfilms, Ann Arbor.

Bishop, J.P. 1983. Tropical forest sheep on legume forage/fuelwood fallows. Agroforestry Systems 1:79-84.

Boerboom,Ja nH.A . 1974. Succession studies inth ehumi dtropica l lowlandso f Surinam. In:Proceeding s ofth efirst internationa l congress ofecology ,De nHaag . Wageningen.

Boonkird, S.A., E.C.M. Fernandes, and P.K.R.Nair . 1984. Forest villages:a n agroforestry approacht o rehabilitating forest landdegrade d byshiftin g cultivation in Thailand. Agroforestry Systems 2:87-102 .

Bourke, R. M. 1985. Food, coffee and casuarine: an agroforestry systemfrom the PapuaNe w Guinea highlands.Agroforestr y Systems 2:273-279 .

Boserup, Ester. 1965. The conditions of agricultural growth.Aldin e Publishing Company, Chicago.

Brown, M.F. 1984. Una paz incierta: Historia y cultura de las comunidades Aguarunasfrent e alimpact od el acarreter amarginal . CAAAP,Seri eAntropologic a No 5, Lima.

143 Budd W.W. et al. (eds.) 1990. Planning for agroforestry. Elsevier,Amsterdam .

Buck, L. 1990. Planning agroforestry extension projects, the Care International approach inKenya . In:W.W .Bud d etal . (eds.) Planning for agroforestry, pp 101- 131. Elsevier, Amsterdam.

Bunker, Stephen G. 1981. The impact of deforestation onpeasan t communities inth e MedioAmazona s ofBrazil . In:Wher ehav e allth eflowers gone : Defores­ tation inth eThir d World. Studies inThir d World Societies 13: 45-60.

Bunker,Stephe n G. 1984. Underdeveloping theAmazon . University of Illinois Press, Chicago.

Burley,CEI . andP .Vo n Carlowitz.1983 .Multipurpos etre egermplasm . ICRAF, Nairobi.

Calzada Benza, Jose. 1981. 143frutale s nativos. Lima

Cancian Frank. 1988. Economic behavior in peasant communities. In: Steward Plattner (ed.) Economic anthropology, pp 125-170. Stanford University Press.

Casanova, J. 1975. El sistema de cultivo Secoya. In: P. Centlivres, J. Gasché and A. Lourteig (eds.) Culture sur brûlis et évolution du milieu forestier en Amazonied uNord-Ouest ,p p 129-143.Bulleti n del aSociét éSuiss ed'Ethnologie .

Cashdan, Elizabeth. 1990. Introduction. In: E. Cashdan (éd.) Risk and uncer­ tainty in tribal and peasant economies, pp 1-16. Westview Press, Boulder.

Cavalcante. 1976, 1978, 1980. Frutas comestiveis daAmazonia . Vol I, II, III. Museo Paraensis Emilio Goeldi,Belem .

Chayanov, A.V. 1986. The theory of peasant economy. The University ofWis ­ consin Press, Madison. 144 Chibnik,Michael . 1990.Double-edge drisks an duncertainties :Choice sabou tric e loans in the PeruvianAmazon . In: E.Cashdan (ed.) Risk and uncertainty intriba l and peasant economies, pp 279-302. Westview Press, Boulder.

Chibnik, Michael. 1991. Quasi-ethnic groups inAmazonia . Ethnology.

Chibnik, M. and W.d eJong . 1989. Agricultural labor organization in riberefio communities of the Peruvian Amazon. Ethnology28:75-95 .

Christanty, Linda. 1986. Shifting cultivation and tropical soils: Patterns,prob ­ lems, and possible improvements. In: G. Marten (ed.) Traditional agriculture in SoutheastAsia : Ahuma n ecologyperspective ,p p226-240 .Westvie wPress ,Boul ­ der.

Christanty L., O.S. Abdoellah, G.G. Marten, and J. Iskander. 1986. Tradi­ tional agroforestry in West Java: The pekarangan (homegarden) and kebun-talun (annualperennia l rotation)croppin g systems.In :G.G .Marte n(ed. )Traditiona lag ­ riculture in Southeast Asia:A hum a ecology perspective, pp 132-158. Westview Press, Boulder.

Clarke, William C. 1966. From extensive to intensive shifting cultivation:A successionfrom Ne w Guinea. Ethnology 5:347-359 .

Clarke,Willia mC . 1971. Placean dpeople : Anecolog y ofa Ne wGuinea ncom ­ munity. University of California Press, Berkeley.

Cobley, Leslie S. 1976. An Introduction to the botany of tropical crops. Second edition. Longman, London.

Conklin,Harol dC .1957 . Hanunooagriculture .FA O ForestryDevelopmen tPaper , No 12. FAO,Rome .

145 Conklin,Harol d C. 1963. Thestud yo fshiftin g cultivation. PanAmerica nUnio n Studies and Monographs VI. Washington, D.C.

Conway, Gordon R. (1985).Agroecosyste m analysis. Agricultural Administra­ tion 20:31-55 .

COREPASA. 1986. Planmaestr od el aRéserv aNaciona lPacaya-Samiria .Iquito s Peru.

Crossley D.A.Jr., G.J. House, R.M. Snider, R.J. Snider, B.R. Stinner. 1984. The positive interactions in agroecosystems. In: RXowrance, B.R.Stinner and G.J.House(eds. )Agricultura lecosystems :Unifyin g concepts,p p73-82 .Joh nWile y and Sons,Ne wYork .

Cunard, A.S. 1967 a. Maize agronomy: Part II. World Crops 19(3):24-28.

Cunard, A.S. 1967 b. Maize agronomy: Part IV.Worl d Crops 19(6):45-58. d'Ans, A.M. 1982. L'Amazonie péruvienne indigene. Payot, Paris.

DeDatta , Surijat K. 1981. Principlesan dpractice so fric eproduction .Joh nWile y & Sons,Ne wYork . deJong ,Wi lA .1984 . Elsistem ad eproduccio n deumari .In :Memori ade ltalle r de trabajo de un plan de investigaciones sobre frutales nativos en la Amazonia Peruana September 1984,Iquitos . de Jong, Wil A. 1985. Floristiese samenstelling en abundantie van soorten en botaniese families in primair bos en antropogeen beinvloed sekundair bos in de PeruviaanseAmazonia . Thesis (MS)Agricultura l University, Wageningen. deJong , WilA . 1987. Organizacion detrabaj o en laAmazoni a Peruana: Elcas o

146 de las sociedades agricolas deTamshiyacu . Amazonia Indigena 7(13)11-17.

Denevan,Willia mM . 1971. Campasubsistenc e inth eGra nPajonal , easternPeru . Geogr.Rev. 61:496-518.

Denevan, William M. 1976. The aboriginal population of Amazonia. In: W.M.Denevan (ed.)Th e native population ofth eAmerica s in 1492,p p205-234 . University ofWisconsi n Press, Madison.

Denevan,Willia mM . 1984. Ecological heterogeneityan dhorizonta lzonatio no f agriculture inth eAmazo n floodplain. In:M.Schmin k andC.H.Woo d(eds. )Fron ­ tier expansion inAmazonia ,p p311-336 .Universit y ofFlorid aPress ,Gainesville .

Denevan,W.M. ,J.M .Treacy ,J.B .Alcorn ,C .Padoch ,J .Denslo wan dS .Flores - Paitan. 1984.Indigenou sagroforestr y inth e PeruvianAmazon :Bor aIndia nman ­ agement of swiddenfields. Intercienci a 9(6):346-357.

Denevan, W.M and C. Padoch 1987. Swidden-fallow agroforestry inth e Peru­ vianAmazon . Adv.in Econ.Bot.5.

Dubois, Jean. C.L. 1990. Secondary forests as a land-use resource in frontière zones ofAmazonia . In:A.Anderso n (ed.)Alternative s to deforestation: Stepsto ­ ward sustainable use ofth eAmazo n rainforest, pp 181-229.Columbi a University Press,Ne wYork .

Ducchart, I., R. Van Haeringen, and F.Steiner . 1990. Introduction: Integrated planning for agroforestry. In: W.W. Budd et al. (eds.) Planning for agroforestry, pp 1-17. Elsevier, Amsterdam.

Duckham, A.N. and G.B. Masefield. 1969. Farming systems of the world. Praeger Publisher, NewYork .

147 Dumont,J.F. , S. Lamotte and F.Kahn . 1990. Wetland and upland forest eco­ systems in Peruvian Amazonia: Plant species diversity in the light of somegeo ­ logical and botanical evidence. For.Ecol.and Manag. 33/34:125-139.

Eden,M.J . 1980. Atraditiona l agro-system in theAmazo n region of Colombia. In:J.I . Furtado (ed.)Tropica l ecologyan d development. Proceedings ofth e Fifth International Symposium ofTropica l Ecology, 16-21Apri l 1979,Kual aLumpur , Malaysia,Par t 1, pp 509-515. TheInternationa l Societyo fTropica lEcology ,Kual a Lumpur.

Eden, M.J. and A. Andrade. 1987. Ecological aspects of swidden cultivation among theAndok e and Witoto Indians of the Colombian Amazon. Human Ecol­ ogy 15(3):339-359.

Eder, James F. 1981. From grain crops to tree crops in the Cuyunon swidden system. In:H .Olofso n (ed.)Adaptiv estrategie san dchang ei nPhilippin eswidden - based societies,p p 91-104.Fores t Research Institute College, LagunaPhillipines .

Encarnacion,Filomeno . 1985. Introducciona l aflor ay vegetacio nd el aAmazoni a Peruana: Estado actual de los estudios, medio natural y ensayo de una clave de determinacion de las formaciones végétales en la llanura amazonica. Candollea 40:237-252.

Ewel, John J. 1986. Designing agricultural ecosystems for the humid tropics. Annual Review of Ecology and Systematics, 17:245-271 .

Fernandes E.C.M., A. O'King, and J. Maghembe. 1984. The Chagga homegardens: Amultistoreye dagroforestr ycroppin gsyste m onMoun t Kilimanjaro (Northern Tanzania).Agroforestr y Systems 2:73-86.

Ferrando,Delicia . 1985. Situacion demograflca. In:Pefi a Herrerae tal . Laselv a Peruana: Realidad Poblacional, pp 35-57. EdicionesAnidep , Lima.

148 Frechione,J. ,D.A . Poseyan d L.Francelin o daSilv a 1989. Theperceptio n of ecological zonesan d natural resources inth e BrazilianAmazon : An ethnoecology of lake Coari. In: D.A.Posey and W.Balée (eds.) Resource management in Amazonia: Indigenous and folk strategies.Adv.i n Econ.Bot. 7:260-259.

Fresco,Louis e O.1988. Cassave inshiftin g cultivation. RoyalTropica l Institute, Amsterdam.

Gallais,J . etal . 1991. SahelNordest eAmazonie : Politiques d'aménagement en milieux fragiles. UNESCO,Paris .

Gasche,Jurge . 1980. El estudio comparativo de los sistemas decultiv o nativos ys uimpact osobr ee lbosqu eamazonico .In :Consult ascientific a subregionalsobr e lasactividade s decort ey quem ae ne lecosistem ad ebosqu etropical .Ma nan dth e Biosphere Program, Iquitos.

Geertz,Clifford . 1963. Agricultural involution:Th eproces s ofecologica lchang e in Indonesia.

Gely, Anne. 1989. Une réponse stratégique face au risque en agriculture: les systèmesagroforestier s del'estuair eamazonien .In :M.Eldi nan dP.Millevill e(eds. ) Le risque en agriculture. ORSTOM,Paris .

Getahun,A . 1990.Agroforestr y fordevelopmen t inKenya : Anoverview .In : W.W. Budd et al. (eds.) Planning for agroforestry, pp 183-202. Elsevier, Amsterdam.

Gholz, H.L. 1987. Introduction. In: H.L. Gholz (éd.) Agroforestry: Realities, possibilities, and potentials, pp 1-6. Martinus Nijhof f Publishers, Dordrecht.

Gladwin, H. and M. Murtaugh. 1980. The attentive-preattentive distinction in agricultural decision making. In: Peggy Barlett, (ed.)Agricultura l decision mak-

149 ing:Anthropologica l contributions to rural development. Academic Press,Ne w York.

Golob, Ann. 1982. The upper Amazon in historical perspective. Dissertation. University Microfilms International, AnnArbor .

Gonzales Boscan, Valois C. 1987. Los morichales de los llanos orientales. Un enfoque ecologico. Ediciones Corpoven.

Gordon, Burton L. 1969. Anthropogeography and rainforest ecology in Bocas del Toro province, Panama. University of California, Department of Geography, Berkeley.

Gordon, Burton L. 1982. A Panama forest and shore: Natural history and Amerindian culture in Bocas delToro . The Boxwood Press, Pacific Grove.

Goulding, M. 1980. The fishes and the forest. University of California Press, Berkeley.

GuillaumetJ.L .e tal . 1990.Le sjardins-verger s familiaux d'Amazonies centrale: un example d'utilisation de l'espace. Turrialba 40(1) pp 63-81.

Gupta,P.C .an dJ.C.O'TooIe .1986 .Uplan drice: A globa lperspective . IRRI, Los Banos.

Guyot,M . 1975. Lesystèm ecultura l Bora-Mirana. In:P .Centrelivres ,J . Gasché and A. Lourteig (eds.) Culture sur brûlis et évolution du milieu forestier en Amazonie du nord-ouest. Musée d'Ethnographie, Switzerland.

Halle,F. ,R.A.A . Oldeman, P.B.Tomlinso n 1978.Tropica l treesan d forests:A n architectural analysis. SpringerVerlag ,Ne wYork .

150 Hanks,Lude n M. 1972. Ricean d man:Agricultura l ecology in SoutheastAsia . AHM Publishing Corporation, Arlington Heights, Illinois.

Hardenburg, W.E. 1912. The Putomayo, the devil's paradise. T.Fisher unwin, London.

Harris, D.R. 1971.Th e ecology of swidden cultivation in in the upper Orinoco rainforest, Venezuela. Geogr. Rev. 61:475-495.

Higbee, Edward C. 1945. The river is the plow. Science Monthly 60:405-416.

Hiraoka,Mario . 1985a . Floodplain farming inth ePeruvia nAmazon .Geographi ­ cal Review ofJapa n 58(1):1-23 .

Hiraoka, Mario. 1985 b. Changing floodplain livelihood patterns in the Peru­ vianAmazon .Tsukub aStudie s in Human Geography IX. Institute ofGeoscience . University ofTsukuba .

Hiraoka, Mario. 1986. Zonation of mestizoriverine farmin g systems inNorth ­ east Peru. National Geographic Research2(3):354-371 .

Hiraoka,Mario . 1989a. Agricultural systems onth e floodplains ofth ePeruvia n Amazon. In: J.O.Browder(ed. )Fragi lland so fLati nAmerica ,p p75-101 .Westvie w Press, Boulder.

Hiraoka, Mario. 1989b. Riberenos changing economicpattern s in the Peruvian Amazon. Journal of Cultural Geography 9(2):103-119 .

Hiraoka,Mario . 1992. Cabocloan driberen oresourc emanagemen t inAmazonia : A review. In K. Redford and C. Padoch (eds.) Conservation ofNeotropica l for­ ests, pp 134-157. Columbia University Press,Ne wYork .

151 Hoekstra,D.A . 1987. Economiesi nagroforestry . In:J.W .Beer ,H.W .Fassbende r and J. Heuveldop (eds.) Advances in agroforestry research, pp 36-49. CATIE, Turrialba.

Hopkins, Nicholas S. 1987. Mechanized irrigation in upper Egypt:Th e role of technology and the state in agriculture. In: B.L. Turner II and S.B. Brush (ed.) Comparative farming systems, pp 223-247.Th e Guilford Press,Ne w York.

Irion, G. 1984. Sedimentation and sediments ofAmazonia n rivers and evolution ofth eAmazonia n landscapesinc ePliocen etimes . In:Haral d Sioli(ed. )Th eAma ­ zon, pp 201-214. Junk Publishers, Dordrecht.

Irvine, Dominique. 1987. Resource management by the Runa Indians of the Ecuadorian Amazon. Dissertation Dep. of Anthropology, Standford University. University Microfilm International, AnnArbor .

Irvine, Dominique. 1989. Succession management and resource distribution in an Amazonian rainforest. In: D.A.Posey and W Balée (eds.) Resource manage­ ment inAmazonia : Indigenous and folk strategies.Adv.i n Econ.Bot. 7:223-238.

Johnson,Allen . 1983. Machiguengagardens .In : R.B. Vickers andW.T . Vickers (eds.)Adaptiv e responses ofnativ eAmazonians , pp 29-64.Academi c Press,Ne w York.

Jordan, CF. 1981.Nutrien t leaching from agro-ecosystems in theAmazo n Ba­ sin,an d implications for recovery ofth eforest . In:J.I .Furtad o(ed. )Tropica lecol ­ ogyan d development. Proceedings ofth eFift h Internatinal Symposium ofTropi ­ calEcology , 16-21Apri l 1979,Kual a Lumpur, Malaysia, Part 1, pp 553-561.Th e International Society ofTropica l Ecology, Kuala Lumpur.

Jordan, Carl F. 1989. Shifting cultivation. In: Carl F. Jordan (ed.)Amazonia n rainforest: Ecosystem disturbancean drecovery ,p p9-23 .Springe rVerlag , NewYork .

152 Junk, W.J. 1970. Investigations on the ecology and production-biology of the "floating meadows"(Paspalo-Echinocloetum ) onth e middleAmazon . Part I.Th e floating vegetation and its ecology.Amazonia n 2(4):449-495.

Junk,W.J . 1980.Aquati c macrophytes: Ecology and use inAmazonia n agricul­ ture. In:J.I . Furtado (ed.)Tropica l ecology and development. Proceedings ofth e Fifth Internatinal Symposium of Tropical Ecology, 16-21 April 1979, Kuala Lumpur, Malaysia. Part2 ,p p763-770 . TheInternationa l Societyo fTropica lEcol ­ ogy, Kuala Lumpur.

Junk,W.J . 1984.Ecolog yo fth evarzea floodplain, o fAmazonia n Whitewater riv­ ers. In: Harold Sioli, (ed.) TheAmazon : Limnology and landscape ecology of a mighty tropical river and its basin, pp 215-244. Junk Publishers, Dordrecht, The Netherlands.

Kahn, F.an d K. Mejia. 1990. Palm communities in wetland forest ecosystems of PeruvianAmazonia . For.Ecol.and Manag. 33/34:169-179.

Kalliola, R., J. Salo andY .Makinen . 1987. Regeneracion natural de selvas en laAmazoni a Peruana 1: Dinamica fluvial y succesion riberena. Memorias del Museo de Historia Natural Javier Prado,No . 18. Universidad Nacional Mayord e SanMarcos ,Lima .

Kalliola, R.,Y .Makine n and S. Salo. 1988. Regeneracion natural deselva s en laAmazoni a Peruana 2:Autecologi a dealguna s especies succesionales. Memoria del Museo deHistori aNatura l JavierPrad oNo . 18. Universidad Nacional Mayor de San Marcos, Lima.

Klammer,Gerhard .1984 .Th erelie fo fth eextra-Andea nAmazo nbasin .In : Harald Sioli (ed.) TheAmazon , pp 47-83. Junk Publishers, Dordrecht.

153 Lageman, J. and J. Heuveldop. 1983. Characterization and evaluation of agroforestry systems:Th ecas e ofAcosta-Puriscal , Costa Rica. Agroforestry Sys­ tems 1:101-115 .

Lamotte,Sandrina . 1990. Fluvialdynamic san dsuccessio n inth elowe rUcayal i river basin, Peruvian Amazonia. For.Ecol.and Manag. 33/34:141-156.

Lathrap, Donald. 1970. The upperAmazon . Praeger, NewYork .

Laur, Pedro nd. Como calcular los costos de produccion. Mimeograph, COPAPMA, Iquitos, Peru.

Leuschner,W.A .an dK .Khaleque .1987 . Homesteadagroforestr y inBangladesh . Agroforestry systems 5: 139-152.

Lopez Parodi, José. 1982. A peruvian experience for the development of Amazonia:Th eregiona l development organization ofLoret o (Ordeloreto)an dth e JenaroHerrer asetllemen t project. In:S.B .Hech t (ed.)Amazonia :Agricultur ean d land-use research. CIAT,Cali .

Lopez,J . andD .Freitas . 1990. Geographical aspectso f forested wetlands inth e lower Ucayali, Peruvian Amazonia. For.Ecol.and Manag. 33/34:157-168.

MacDicken, Kenneth G. 1990. Agroforestry management inth e humidtropics . In: K.G. MacDicken and N.T. Vergara (eds.) Agroforestry: Classification and management, pp 98-149. John Wiley & sons,Ne wYork .

MacDicken,K.G. ,an dN.T .Vergara .1990 .Introductio nt oagroforestry . In: K.G. MacDickenan dN.T .Vergar a(eds. )Agroforestry : Classification andmanagement , pp 1-30. John Wiley & sons,Ne wYork .

154 Marten, G. and D.M. Saltman. 1986. The human ecology perspective. In: G. Marten (ed.)Traditiona l agriculture in SoutheastAsia : Ahuma n ecologyperspec ­ tive, pp 20-53. Westview Press,Boulder .

Meggers, Betty J. 1971. Amazonia: Man and culture in a counterfeit paradise. Aldine-Atherton, Chicago.

Meyers,Thomas . 1988. Theorige no fth eRiberenos .Pape rpresente da tth e46t h International Congress ofAmericanists ,Amsterdam .

Michon, Genevieve. 1985. De l'homme de la foret au paysan de l'arbre: Agroforesteries indonésiennes. These Académie de Montpellier, Université des Sciences etTechnique s duLanguedoc .

Michon, G., J. Bompard, C. Ducatillon and P.Hecketsweller . 1983. Tropical forest architectural analysis as applied to agroforests in the humid tropics: The example oftraditiona l village forests inWes tJava . Agroforestry Systems 1:118- 129.

Michon, G., F.Mar yan dJ . Bompard. 1986. Multistoreyed agroforestry garden system in West Sumatra Indonesia.Agroforestr y Systems,4:315-338 .

Muleba, N. and H.C. Ezumah. 1985. Optimizing cultural practices for cowpea inAfrica . In: S.R. Kingh and K.O Rachie (eds.) Cowpea: Research, production, and utilization, pp 289-295. John Wiley & Sons,Ne wYork .

Nair, M.A. and C. Sreedharan (1986). Agroforestry farming systems in the homesteads of Kerala, southern India. Agroforestry Systems4:339-363 .

Nair, P.K.R. 1985. Classification of agroforestry systems. Agroforestry Systems 3:97-128.

155 Nair, P.K.R. 1989. Ecological aspects of agroforestry. In: P.K.R. Nair (ed.) Agroforestry systems in the tropics, pp 63-86. Kluwer Academic Publishers, Dordrecht.

Nair, P.K.R. 1990. Classification of agroforestry systems. In: K.G. MacDicken andN.T .Vergar a(ed. )Agroforestry : Classification andmanagement ,p p31-57 .Joh n Wiley & Son, NewYork .

Nascimiento, C. andA .Homma . 1984. Amazonia: Meioambient ee tecnologi a agricola. EMBRAPA-CPATU, Belem.

National Research Council. 1992. Conserving Biodiversity:A research agenda for development agencies.Nationa lAcadem y Press, Washington, D.C.

Nye,P.H . and D.J. Greenland. 1961. Thesoi lunde rshiftin g cultivation. Com­ monwealth Agricultural Bureau, Reading.

Nye,P.H . and D.J. Greenland. 1964.Change s inth esoi l after clearing tropical forest. Plant & Soil 21:101-112.

Norgaard, Richard B. 1981. Sociosystems and ecosystems coevolution in the Amazon. Journal of Environmental Economics and Management 8:238-254.

Okafor, J.C and E.C.M. Fernandes. 1987. Compounds of southeastern Nige­ ria: Apredominan tagroforestr y homegardensyste mwit hcrop san dsmal llivestock . Agroforestry Systems 5:153-168.

ONERN. 1976. Inventario, evaluacion eintegracio n de los recursos naturales de la zona Iquitos, Nauta, Requena y Colonia Angamos. Oficina Nacional de Evaluacion deRecurso s Naturales, Lima.

ONERN. 1977. Use of remote sensing systems evaluating the potential of the

156 aguaje palm tree in the Peruvian jungle. Oficina National de Evaluation de Recursos Naturales, Lima,Peru .

Padoch, Christine. 1982. Migration and its alternatives among the Iban of Sarawak. Verhandeling van het Koninklijk Instituut voor Taal-,Land- en Volkenkunde 98.Martinu sNijhoff , TheHague .

Padoch, Christine. 1986. Thecampesino s of Santa Rosa: History and ethnicity ina nAmazonia ncommunity .Pape rpresente da tth eAnnua lMeetin go fth eAmeri ­ canAnthropologica l Association, December 1986, Philadelphia.

Padoch,Christine . 1987. Theeconomi c importance and marketing of forest and fallow products in the Iquitos region. In: W.M.Denevan and C.Padoch (eds.) Swidden-fallow agroforestry inth e PeruvianAmazon .Adv.i n Econ.Bot. 5:74-89.

Padoch, Christine. 1988. Aguaje (Mauritia flexuosa L.f.) in the economy of Iquitos, Peru. In: M.Balick (ed.) The palm-tree of life: Biology utilization and conservation. Adv. in Econ. Bot. 6:214-244.

Padoch, Christine. 1992. Marketing of non-timber forest products in Western Amazonia: General observations and research priorities. In: D. Nepstad and S. Schwartzman (eds.)Non-timbe rproduct sfrom tropica l forests.Adv .i nEcon .Bot . 9:43-51.

Padoch, C. and W.de Jong. 1987. Traditional agroforestry practices of native and ribereno farmers in the lowland Peruvian Amazon. In: H.Gholz (ed.) Agroforestry: Realities,possibilities ,an dpotentials ,p p 179-194. MartinusNijhofF , Dordrecht, TheNetherlands .

Padoch, C. and W. de Jong. 1989. Production and profit in agroforestry: an example from the Peruvian Amazon. In: J.O.Browder (ed.) Fragil lands of Latin America, pp 102-113. Westview Press, Boulder.

157 Padoch, C. and W.d e Jong. 1990. Santa Rosa: The impact of forest products trade ona nAmazonia n place and population. In: G.T.Prance and M.Balick(eds. ) New directions in the study ofplant s and people.Adv.i n Econ.Bot. 8:151-159.

Padoch, C, J. Chota Inuma, W. de Jong and J. Unruh. 1985. Amazonian agroforestry: Amarket-oriente d system in Peru. Agroforestry Systems 3:47-58.

Parker, Eugene P. 1981. Cultural ecology and change:A caboclo varzea com­ munity in the Brazilian Amazon. Dissertation University of Coloradoa t Boulder. University Microfilms International. AnnArbor , Michigan.

Parker, Eugene P. 1985. The Amazon caboclo: Historical and contemporary perspectives. Studies inThir d World Societies,Vol . 32.Willia m and MaryPress , Williamsburg.

Parker, E.P.,D .Posey ,J . Frechione and F.d aSilv a L. 1983. Resourceexploi ­ tation inAmazonia : Ethno-ecological examples from four populations. Annals of Carnegie Museum 52:163-203.

Peters, Charles. 1990. Population ecology and management of forest fruit trees inPeruvia n Amazonia. In:Anthon yAnderso n (ed.)Alternative s to deforestation. Columbia Unviersity Press,Ne wYork , pp86-98 .

Peters,C .an dA .Vasquez . 1989. Estudiosecologico s decamu-cam u [Myrciaria dubia): Produccion de frutos enpoblacione s naturales. FoliaAmazonic a 1(1):83- 98.

Petrick, Carsten. 1978. The complementary function of floodlands for agricul­ turalutilization :Th evarze ao fth eBrazilia nAmazo nregion .Applie d Sciencesan d Development 12:24-46.

Pinedo Vasquez, Miguel. 1986. Conservation and local control of resources in

158 the Peruvian Amazon. Paper presented at the Annual Meeting of the American Anthropological Association, December 1986, Philadelphia.

Pinedo Vasquez, M.; D. Zarin and P. Jipp. 1990. Land-use in the Amazon. Nature348:397 .

Plattner, Steward. 1988. Introduction. In: Steward Plattner (ed.) Economic an­ thropology, pp 1-20. Stanford University Press.

Posey, DarrelA . 1983. The keepers of the forest. Garden 6(1) pp 18-24.

Posey,Darre lA . 1984. Apreliminar y report ondiversifie d management oftropi ­ cal forest byth e Kayapo Indians ofth e Brazilian Amazon. In: G.T. Prance andJ . Kallunki (eds.)Adv.i n Econ.Bot. 1:112-126.

Posey, Darrel A. 1985. Indigenous management of tropical forest ecosystems: The case of the Kayapo Indians of the Brazilian Amazon. Agroforestry Systems 3:139-158.

Prance, Ghillean T. 1979. Notes onth e vegetation ofAmazoni a III.Th etermi ­ nology ofAmazonia n forest types subject to innundation. Brittonia 31(l):26-38.

Purseglove,J.W .1974 . Tropicalcrops :Dicotyledons .Thir dimpression .Longma n Group Limited, London.

Purseglove, J.W. 1975. Tropical crops: Monocotyledons. Second impression. Longman Group Limited, London.

Putzer, Hannfrit. 1984. The geological evolution of the Amazon basin and its mineral resources. In: Harald Sioli (ed.)Th eAmazon , pp15-46 .Jun kPublishers , Dordrecht.

159 Rachie, K.O. 1985. Introduction. In: S.R. Kingh and K.0 Rachie (eds.) Cow- pea: Research, production, and utilization. John Wiley & Sons,Ne wYork .

Raintree,J.B . and K.Warner . 1986. Agroforestry pathways forth e intensifica­ tion of shifting cultivation. Agroforestry Systems 4:39-54.

Rao,M.R. , CS. Kamara,F .Kwesig aan dB .Duguma . 1991. Agroforestry field experiments:methodologica l issuesfo r research onimprove d fallows. Agroforestry Today, 2(4):8-12.

Richards, P.W. 1952. The tropical Rainforest. University Press, Cambridge.

Rocheleau M., Dianne A. 1987. The user perspective and the agroforestry re­ search and action agenda. In: H.L. Gholz (ed.)Agroforestry : Realities, possibili­ ties and potentials, pp 59-88. Martinus Nijhof Publishers, Dordrecht.

Roosevelt,Ann e C. 1980. Parmana: Prehistoric maize and manioc subsistence along theAmazo n and Orinoco. Academic Press,Ne wYork .

Roosevelt,Ann e C. 1989. Resource management inAmazoni a before the Con­ quest:Beyon dethnographi cprojection . In:D.A.Pose yan dW.Balé e(eds. )Resourc e management inAmazonia :Indigenou san d folk strategies.Adv.i nEcon.Bot . 7:30- 62.

Roseberry, William. 1988. Peasants and the world. In: Steward Plattner (ed.) Economic anthropology, pp 108-126. Stanford University Press.

Ross, Eric B. 1978. The evolution of the Amazon peasantry. Journal of Latin American Studies 10(2):193-218 .

Ruddle, Kenneth. 1974. TheYukp acultivatio n system:A stud y of shifting cul­ tivation in Colombia and Venezuela. University of California Press, Berkeley.

160 Ruthenberg, Hans. 1980. Farming systems in the tropics. Second Edition. Clarendon Press, Oxford.

Salo,J. , R. Kalliola, I. Häkkinen,Y .Makinen ,P .Niemela ,M . Puhakka,P.D . Coley. 1986. Riverdynamic s andth e diversity ofAmazo n lowland forest. Nature 322:254-258.

SanRoman ,Jesus . 1975. Perfiles historicas del aAmazoni aPeruana . Ediciones Paulinas, Lima.

Sanchez, PedroA . 1973. Soil management under shifting cultivation. In: P.A. Sanchez(ed. )A revie wo fsoil sresearc h intropica l LatinAmerica ,p p46-67 . Tech. Bui. No. 219. Soil Science Department N.C. State University at Realeigh.

Sanchez, Pedro A. 1988. Deforestation reduction initiative: An imperative for world sustainability in the twenty-first century. Paper presented at the Bureau of Science and Technology, US-AID. Washington D.C.

Scherr, S.J. 1990. The diagnosis and design approach to agroforestry project planning and implementation: Examples from Western Kenya. In: W.W.Bud d et al. Planning for agroforestry, pp 132-160. Elsevier, Amsterdam.

Shaner,W.W. ,P.F .Philip pan dW.R .Schmehl . 1982. Farmingsystem sresearc h and development. Westview Press BoulderColorado .

Sioli, Harold (ed.) 1984a. TheAmazon :Limnolog y and landscape ecology ofa mighty tropical river and itsbasin . Junk Publishers, Dordrecht, TheNetherlands .

Sioli, Harold. 1984b. The Amazon and its main affluents: Hydrography, mor­ phology ofth e rivercourses , and rivertypes . In: Harold Sioli (ed.)Th eAmazon , pp 127-165.Jun k Publishers, Dordrecht, TheNetherlands .

161 Schmink,M . 1985. SaoFeli x do Xingu:A cabocl o community intransition . In: E.P.Parke r (ed.)Th eAmazo n caboclo: Historical andcontemporar y perspectives, pp 143-166. Study in Third World Societies, Vol 32. William and Mary Press, Williamsburg.

Smith, Nigel J.H. 1982. Rainforest corridors. University of California Press, Berkeley.

Smole,Willia mJ . 1976. TheYanoam a Indians: Acultura lgeography . University ofTexa s Press,Austin .

Soemarwoto, Oan d I. Soemarwoto. 1984. The Javanese rural ecosystems. In: A.T. Rambo & P.E. Sajise (eds.) An introcution to human ecology research on agricultural systems insoutheas t Asia,p p 254-287. University ofth e Philippines, Los Banos.

Soemarwoto, O, I. Soemarwoto, Karyono, E.M. Soekartadireja and A. Ramlan. 1985. TheJavanes ehom egarde na sa nintegrate dagro-ecosystem . Food andNutritio n Bulletin 7:44-47.

Sombroek, W.G. 1984. Soils of theAmazo n region. In: H.Sioli (ed.)Th eAma ­ zon:Limnolog yan d landscapeecolog y ofa might ytropica riverl an d itsbasin ,p p 521-536.Jun k Publishers, Dordrecht, TheNetherlands .

Spencer,J.E . 1966. Shifting cultivation in SoutheastAsia .Publication s inGeog ­ raphyNo .19 , University of California Press, Berkeley.

Sternberg,Hildegar d O'Reilly.1975 . TheAmazo nrive ro fBrasil .Geographisch e Zeitschrift, Beihefte 40.

Stocks,Anthon yW .1981 . Losnativo sinvisibles .Nota ssobr el ahistori ay realida d actual de los Cocamilla del Rio Huallaga, Peru. CAAAP, Lima.

162 ten Have, H. 1982. Production andconsumptio n ofrice an dth e socio-economic importance ofrice i nth e tropics. In: International Crongress ofth e 12thagricul ­ tural machinery exhibition. January 19& 20, 1982,Amsterdam .

Tripathi, R.S. 1977. Weed promlem: An ecological perspective. Tropical Ecol­ ogy 18:138-148.

Turner II, B.L. and S.B. Brush (1987). Purpose, classification and organiza­ tion. In:B.L .Turne rI Ian d S.B. Brush(eds. )Comparativ e farming systems,p p 3- 10. University of California, Davis. vande rPoel ,P , andH .va nDijk . Householdeconom yan dtre egrowin gi nuplan d Central Java.Agroforestr y Systems 5:169-184.

Vasquez, Rodolfo. 1990. Plantas utiles de laAmazoni a Peruana. Iquitos, Peru.

Vergara,Napoleo nT .1987 . Agroforestry: Asustainabl e landus efo r fragile eco­ systems in the humid tropics. In: H.L. Gholz (ed.)Agroforestry : Realities,possi ­ bilities and potentials, pp 7-19. Martinus Nijhoff Publishers, Dordrecht.

Vickers,Willia mT .1976 . Cultural adaptation toAmazonia n habitats:Th eSiona - Secoyao fEaster nEcuador . PhDThesis .Universit y Microfilms International, Ann Arbor.

Vickers,Willia mT .1978 . NativeAmazonia nsubsistenc e indivers ehabitats : The Siona-Secoya of Ecuador. In: Emilio F. Moran (ed.) Changing agricultural sys­ tems in Latin America. Studies inThir d World Societies, No 7,Williamsburg .

Wagley,Charles . 1953. Amazontown : Astud yo fma ni nth etropics .MacMillan , NewYork .

Weinstein, Barbara. 1983. The Amazon rubberboom: 1850-1920. Standford University Press, Standford. 163 Wesche, R. 1985. The transformation of rural caboclo society upon integration into Brazil's Amazon frontier: A study of Itacoatiara. In: E.P. Parker (ed.) The Amazoncaboclo :Historica l andcontemporar yperspectives ,p p 115-142.Stud yi n Third World Societies, Vol. 32.Willia m and Mary Press, Williamsburg.

Wiersum, K. Freerk. 1983. Tree gardening and taungya on Java: Examples of agroforestry techniques in the humid tropics.Agroforestr y Systems 1:53-70.

Wiersum,K . Freerk. 1985. Forestryaspect s ofstabilizin g shifting cultivation in Africa. Agricultural University, Wageningen.

Wiersum,K . Freerk.(ed.) . 1988. Viewpoints onAgroforestry : Second,renewe d edition. Hinkeloord, Agricultural University Wageningen, TheNetherlands .

Williams,C.N . 1975. Theagronom y ofth emajo r tropical crops.Oxfor d Univer­ sity Press, Kuala Lumpur.

Whitmore, T.C. 1984. Tropical rainforests of the far east. Second edition. Clarendon Press, Oxford.

Withmore, T.C. and G.T. Prance (eds.). 1987. Biogeography and Quaternary history in tropical America. Clarendon Press, Oxford.

Yen,D.E .1984 . Arboriculture inth esubsistenc eo fSant aCruz , SolomonIslands . Economic Botany 28:247-284.

164 NEDERLANDSE SAMENVATTING

DIVERSITEIT, VARIATIE, EN VERANDERING IN RIBERENO LANDBOUW EN AGROFORESTRY

Destroombeddinge nva nd egrot erivieren va nhe tAmazon ebekke nworde n in het algemeen aangeduid als varzea. Deze varzea wordt gekenmerkt door een gevarieerd landschap van natuurlijke dijken, uiterwaarden, en modder- en zandplaten. Daar de hoogste van de natuurlijke dijken slechts een keer in vele jaren overstromen, is het mogelijk om in deze rivierbeddingen landbouw te bedrijven, en te wonen. De varzea was een belangrijk leefgebied voor de pre- columbiaanse Indiaanse groepen inhe tAmazon e gebied. Dezebevolkin g maakte gebruikva nd erijkdo m aanvi se nvruchtbar elandbouwgron d aanwezigi nd evarze a De tegenwoordige inheemse bevolking van de varzea worden aangeduid als riberenos in Peru, en als caboclos in Brazilië. Wetenschappers hebben verschillendetheorieë n geformuleerd betreffende hetvoorhistorisc h entegenwoordi g landgebruik ind evarzea . VolgensRos s (1978), islandbou w doord einheems ebevolkin g ind evarze anagenoe gverdwene nomda t zij, alsgevol gva nsocial e desintegratie,nie t devereist ecoördinati eka nopbrenge n diehe tgebrui kva ndez egecompliceerd eomgevin gvereist . Meggers(1971 )betoog t dathe tgebrui kva nd evarze aee nsocial eorginasati eme tpersone nva nhoger estatu s nodig heeft. Roosevelt (1980) stelt dat de voorhistorische, hoger ontwikkelde samenlevingen, diei nd evarze awoonde n voordatd eEuropeane n erkwamen ,zic h slechtskonde nontwikkele nn ad eintrodukti eva nmaiz ,daa rdi tgewa she t mogelijk maakte een voldoende hoeveelheid eiwitten te produceren zodat een talrijkere bevolking gevoedko nworde n danme the tallee n etenva ncassave envi s mogelijk was. In dit boek wordt de diversiteit, variatie, en verandering in landbouw en agroforestry inSant aRos ae nYanallpa ,twe eribereno dorpe naa nd eUcayal i rivier in Peru, geanalyseerd. Yanallpa ligt op een natuurlijke dijk, totaal ingesloten in

165 derivier bedding ,terwij l SantaRos ao pterra firme, he tvast elan dwaari nd e rivier zichee nwe gheef t gebaand,gelege n is. Riberenos beoefenen eengecompliceerd e envernuftig e landbouw, endoe ndi t als individueleboere n zonder leiders diehu n daarbij organiseren. Riberenos verbouwen weliswaar maiz en ook rijst, echter voornamelijk voord everkoop . Zij zijn echtervolledi g instaa t zichzelft evoede n metvis ,cassave , enbananen . Hetvoorbeel d vanribereno landgebrui k ind evarze a weerlegt dedri ebovengenoemd e theorieën, diederhalv e herformulering vereisen. Riberenos zijn de afstammelingen van de oorspronkelijk Indiaanse bevolking van het Peruviaanse Amazone gebied, die als gevolg van de, meestal onvrijwillige, deelname aan verschillende ekonomische aktiviteiten in dit gebied, envermenginge n metpersone nva nander e inheemsegroepe ne nimmigranten ,ee n nieuwe sociale en culturele identiteit verworven hebben. Zij leven vooral van landbouw, visvangst, handel, en het verzamelen van bosprodukten. Het varzea landschap waar zij voor het grootste gedeelte hun landbouw bedrijven kent een vescheidenheidaa nlandvormen . Dezeverschille ni nhoogt eniveau ,i nd e frequentie enmat ewaari nz eoverstromen ,e nal sgevol gdaarvan ,i nbode mtyp ee nkwaliteit . Eennieu wconcept ,genoem dlandbouwtyp e{agricultural type), isgebruik t omd elandbou wdiversitei t vanriberenos te kunne nbeschrijven . Eenlandbouwtyp e isgedefinieer d als eenuniek ecombinati e vangroeiplaat s engewas ,da t specifieke landbouwtechnieke nvereist ,ee nbepaald eaktiviteite nkalende rheeft , enbovendie n een begrensd gemiddeld produktie niveau, een eigen doel van de opbrengst, en eenbepaal d type engroott e vanrisiko heeft . In detwe ebestudeerd e dorpen werd een totaal van veertien verschillende landbouwtypes aangetroffen. Tien daarvan kwamen in de varzea voor, terwijl vier opterr afirme gevonde n werden. Variatiei n landbouwstrategieën tussen deboere n vand etwe ebestudeerd e dorpen werdganalyseerd . Desamenstellin g vand elandbouwtype s vanel kbedrijf , en de verandering daarin tussen 1985/1986 en 1989 werd opgenomen. Deze samenstellingen werden gekorreleerd aan verscheidene sociaal-economische variabele kenmerken van de boer en zijn familie. Het bleek dat boeren in Santa Rosa bij voorkeur drie tot vijf landbouwtypes hebben, bestaande uit rijst op de vruchtbare modderplaten, een gemengde cassave akker, en een agroforestry veld

166 opterr a firme. Deboere n inYanallp a hebben eveneens bij voorkeur drie tot vijf landbouwtypes. Echter indi tgeva lbestaa ndi eui t maisvelden,bananenvelden ,e n een gemengde cassave akker. In de twee dorpen zijn veel boeren die om verschillende redenene vandi t patroonafwijken , enhe t isderhalv e niettypisc ht e noemenvoo rriberen o landbouw. Eenverschi l innadru ko prijs tprodukti ei nSant a Rosa, en maiz produktie in Yanallpa, beiden verbouwd voor de verkoop, is het gevolg van verschillende mogelijkheden diehe t landschap biedt inbeid e dorpen. Alsgevol gva nveranderd eecologisch eomstandigheden ,veranderin gva nhe tsociaa l economische klimaat, of omstandigheden ind ehuishoudelijk e sfeer, veranderden demeest eboere ni nmeerder eo fminde rmat ehu nlandbouwstrategieë ntusse n1985 / 1986 en 1989. Diversiteit interr afirme agroforestr y inSant aRosa ,e nvarze aagroforestr y inYanallp awerde n geanalyseerd. InSant a Rosa heeft dit landbouwtype devor m vanhe tbehere n vanbospercele n oplan d dat eerdervoo r eenande r landbouwtype gebruikt werd. Agroforestry velden verschillen in desamenstellin g vanbeheerd e soorten, in de mate waarin de velden onderhouden worden, en in de opbrengst. Varzea agroforestry inYanallp a heeft een meer permanent karakter, en wordt in hogeremat egekombineer d met deprodukti e vanjaarlijk s gewassen, dan inSant a Rosa. InYanallp aword t bovendien meer denadru k gelegd opprodukti e vannie t gedomesticeerde soorten, alsgevol gva nhu ngroter e schaarstedaarva n indi tdor p dat verderaf gelegen is vanprimai rbos .

167 Levensoverzicht Wil de Jong

Wild eJon gwer dgebore ni nVelden ,Limburg ,o p7 februar i 1956.I n 1975 begon hij zijn studies aan de toen nog genoemde Landbouwhogeschool Wageningen. Na het afronden van het kandidaats programma, aan de vakgroep bosbouw, volgdehi j het inhaalprogrammafilosofie aa nd eKatholiek e Universiteit Nijmegen. Vervolgens deed hij zijn eerste doctoraal onderzoek naar de mogelijkheden omgrovedennenbosse nt everandere ni nmee rgemengd ebosse no p het landgoed Rozendaal bijArnhem . In 1982vertro khi j naarPer u ome r praktijk te doen aan de Universidad Nationalde laAmazonia Peruana, en deed hij een tweededoctoraa londerzoe knaa rd efloristische samenstellin gva nsecundair ebosse n op voormalige akkers van Bora Indianen in het Amazone gebied nabij de Colombiaanse grens. In 1984 werkte hij eenjaa r aan het InstituteNational de InvestigationAgricola in Iquitos, in een onderzoek naar Poraqueiba sericea agroforestry produktie inTamshiyacu , een klein dorp aan deAmazon e rivier. In 1985bego n hij het onderzoek naar ribereno landbouw en agroforestry aan de Ucayali rivier, het thema van dit boek, in een projekt dat het Institute de Investigaciones de la Amazonia Peruana uitvoerde in samenwerking met hetIn­ stitute ofEconomie Botany van deNew YorkBotanical Garden. In 1988vertro k hij naarNe wYor k om daart e werken aan de voltooiing vanzij n proefschrift. In 1992 begon hij als onderzoeks medewerker aan hetzelfde instituut aan een onderzoeknaa rDaya kbosbehee ri nWes tKalimantan . Op 1 November 1995 begint hij een nieuweaangstellin g aan het Centrefor InternationalForestry Research in Bogor, Indonesië.

168