TropSoils technical report

Material Information

TropSoils technical report
Soil Management Collaborative Research Support Program
Place of Publication:
Raleigh N.C
TropSoils Management Entity, North Carolina State University
Creation Date:
Publication Date:
Physical Description:
v. : ill. ; 28 cm.


Subjects / Keywords:
Soils -- Periodicals -- Tropics ( lcsh )
Soils -- Periodicals -- Latin America ( lcsh )
Soil management -- Periodicals -- Tropics ( lcsh )
Soil science -- Periodicals -- Tropics ( lcsh )
serial ( sobekcm )


Dates or Sequential Designation:
General Note:
"TropSoils is one of the Collaborative Research Support Programs."
General Note:
Latest issue consulted: 1988-1989.
Electronic resources created as part of a prototype UF Institutional Repository and Faculty Papers project by the University of Florida.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact Digital Services ( with any additional information they can provide.
Resource Identifier:
16150153 ( OCLC )
sn 90040074 ( LCCN )

Related Items

Preceded by:
TropSoils triennial technical report

Full Text
/ 7/^








Edited by
Neil Caudle, TropSoils Editor, Department of Agricultural
Communications, North Carolina State University
Charles B. McCants, Director, TropSoils Management Entity,
North Carolina State University

Published June, 1987

TropSoils is one of the Collaborative Research Support Programs (CRSPs) created to implement Title XII,
"Famine Prevention and Freedom From Hunger" of the Foreign Assistance Act. Primary funding is
provided by the U.S. Agency for International Development under grant no. DAN-1311-G-SS-108 3.

Copies of this report may be obtained by writing TropSoils Management Entity,
Box 7113, North Carolina State University, Raleigh, N.C. 27695-7113, U.S.A.


Preface. vii
Collaborators. viii


Introduction... 1
Low-Input Cropping Systems...5
Central Low-Input Experiment... 6
Tillage-Phosphorus Interactions Under Low-Input Cropping Systems...7
Calcium and Magnesium Movement in Low-Input Cropping Systems... 9

Legume-Based Pastures...11
Legume-Based Pastures: Central Experiment...12
Potassium Dynamics in Legume-Based Pastures... 15
Pasture Germplasm Evaluation and Agronomy... 17
Pasture Reclamation in Steeplands...19

Agroforestry Systems...21
Trees as Soil Improvers in the Tropics?...22
Alley-Cropping on Ultisols...23
Peach Palm as a Soil Management Option on Ultisols...26
Gmelina arborea: Intercropping, Coppicing and Nutritional Requirements...28
Contrasting Effect of Pinus caribaea and Gmelina arborea on Soil Properties...29
Improved Fallows... 31
Forest and Soil Regeneration... 35
Comparative Soil Dynamics...43
Collection and Propagation of Agroforestry Species...43
Living Fences...44

Continuous Cropping Systems...45
Land-Clearing and Post-Clearing Soil Management Practices...46
Tillage With Tractors in Continuously Cropped Ultisols...50
Continuous Cropping: Central Experiment...52
Production Potential of Corn-Peanut Intercrops in the Humid Tropics... 54
Phosphorus, Zinc and Copper Fertilization... 56
Potassium, Lime and Magnesium Interactions and Corn Yields... 59
Weed Population Shifts Under Continuous Cropping Systems...65
Chemical Weed Control in Corn...66

Paddy Rice in Alluvial Soils...69
Intensive Management of Alluvial Soils for Irrigated Rice Production...70

Soil Characterization and Interpretation... 73
FCC Adaptation to Wetland Soils... 74
Volcanic Ash Influence on Transmigration Areas of Sumatra...77
FCC and Site Characterization in Relation to Caribbean Pine... 77
Alluvial Soils of the Amazon Basin...78
Ultisol Dominated Landscapes in Southeastern Peru... 80
Soils of Pichis Valley Extrapolation Sites...82
Soil Survey of the Puerto Maldonado Experiment Station...84

Soil Management Research Network...85

Brazil: Extrapolation to Clayey Oxisols...87
Soil Nutrient Dynamics and Fertility Management for Sustained Crop Production on Ox-
isols in the Brazilian Amazon...88
Phosphorus Management in Humid Tropical Oxisols...94
Potassium Management in Humid Tropical Oxisols...97
Guarana Fertilization...100
Lime Requirements and Downward Movement of Ca and Mg...102
Planting Dates in Relation to Weather Pattern at Manaus...107
Micronutrient Fertilization on a Typic Acrorthox at Manaus...109
Management of Green-Manure Nitrogen on Oxisols at Manaus...lll
Conditions Other Than Extractable Nutrient Concentrations in the Soil Test Interpretations
for P and Zn...113

Extrapolation of Soil Management Technologies to the Pichis Valley...115
Low-Input Cropping Systems for the Pichis Valley... 116
Legume-Based Pastures in the Pichis Valley... 118
Agroforestry in Steeplands of the Pichis Valley...120
Soil Erosion and Reclamation...120

Sitiung: Extrapolation to Transmigration Areas of Indonesia...121
Reclamation of Bulldozed Lands...122
Liming in Transmigration Areas... 125
Phosphorus Management in Transmigration Areas... 131
Effect of Green Manure Applications on Soil Fertility and Crops... 134
Potassium Management of Upland Crops...137
Sulfur Fertilization... 139

Contributions to North Carolina and U.S. Agriculture... 139




Site Characterization: Soil Variability... 145
Soil Variability in Forest Land Mechanically Cleared...146

Soil Management and People... 155
Indigenous Knowledge Systems Related to Soil Management... 156
Intrahousehold Decision-Making... 157
Collaborative Research With Farmers on Upland Fields... 158
Time-Allocation Study... 159
Nutrition/Income Survey...162
Collaborative Research With Farmers on Home Gardens...163
Farmers' Perceptions of Constraints to Agriculture...164
Minang Tree-Farming Study... 165

Productivity in Farmers' Fields...167
Matching Crop Requirements of Rice, Maize, Soybean and Peanut to Soil Characteristics
with Crop Simulation Models...168
Effects of VA Mycorrhizae on Cowpea Response to P Fertilization and Lime in High
Manganese Soil...171
Modeling Phosphorus-Lime Interactions... 175
Application of Expert Systems to the Transfer of Soil Management Technology...177
Pasture Grasses and Legumes for the Humid Tropics... 181

Land Reclamation: Soil Physics and Soil Conservation...18 3
Management of Organic Material in Indonesian Farming Systems...184

Contributions to Hawaii Agriculture...187



Introduction... 189

Nitrogen Management... 191
Nitrogen Availability From Legume Crop Residues and Green Manures to Succeeding
Nonlegume Crops...192

Evaluation of the Mineralization Potential of Legume Residues Through Laboratory Incuba-
tion Studies...194
Fertilizer Nitrogen Movement in Cerrado Soils...195

Water and Chemical Budgets...197
The Effects of Gypsum Amendments on Charge Properties in Cerrado Soils...198
Ion Movement in Cerrado Soils: The Effects of Amendments on Sulfur Availability...200

Soil Constraints to Management...203
Soil Morphology and Water Table Relations in Some Oxisols of the Cerrado Region...204
Characterization of Root Restricting Zones in Cerrado Soils...206

Contributions to New York Agriculture...208




Soil Data Base...213
Soil Genesis, Phosphorus and Micronutrients of Selected Vertisols and Associated Alfisols of
Northern Cameroon...214
Clay Dispersibility in Sandy Soils of the Sahel, West Africa...218
Iron Oxide Properties Versus Strength of Ferruginous Crust and Iron-Glaebules in Soils...221
Soil Crusting: Compaction of Soil Particles Due to Impact of Raindrops and Drying...222
Soil Properties Versus Crust Strength of Some Texas and West African Soils...223
Soil Geomorphology-Hydrology Relationships of Semi-Arid Tropical Landscapes...225
Calibration of Two-Probe Gamma-Gauge Densitometers...228
Field Calibration of Neutron Meters Using a Two-Probe Gamma Density Gauge...229
A Simple Method to Calculate Distribution of a Scaling Factor From Soil-Water Retention
Simulation and Measurement of Evaporation From a Bare Soil...2 32
Causes and Control of Pronounced Plant-Growth Variability... 233
Water and Energy Balance of Sahelian Soils...236

Technology for Rainfed Agriculture...239
Influence of Tiller Removal on Growth and Production of Millet...240
Pearl Millet (Pennisetum typhoides) Response to Soil Variability in Sandy Ustaffs...242
Phosphorus Fertilization and Relationships of Root Distribution and Soil Water
Extraction... 244
Sorghum Water-Use Efficiency and Fertilizer Relationships...247

Evaluation of the Sandfighter Under Sahelian Conditions... 249
Potential of Contour-Strip Water Harvesting for Cereal Production...250
Soil Moisture Relations of Sandy Soils of Niger...251

Technology for Forest Lands...255
Soil and Water Management in Degraded Sahelian Soils...256

Agroclimatic Data Base...259
Quantification of Rainfall Characteristics, Patterns and Hydrology of Representative Cropped
Soils of Niger...260
Influence of Neem-Tree Windbreaks on Microclimate and the Growth and Yield of Cereals
Between the Rows...263

Contributions to Texas Agriculture...266



TropSoils' goal is to develop and adopt improved soil-management technology that will reduce con-
straints to plant growth, and to ensure that this technology is agronomically, economically and ecological-
ly sound for developing countries in the tropics. Because it is impractical to do this in every tropical
nation or region at once, the program has situated and developed its research projects in a way that
makes their results applicable over broad areas having similar soils and environments. These areas,
or "agro-ecological zones," are the basic units of TropSoils' organization.

This document reports the progress of TropSoils research in three agro-ecological zones: the humid
tropics, the acid savannas and the semiarid tropics. Each participating university has taken a lead role
in one of these zones. For more information about any of the projects covered in this report, contact
the Management Entity or one of the program coordinators listed below.

Program Coordinators

Pedro A. Sanchez
Soil Science Department
Box 7619, N.C. State University
Raleigh, NC 27650-7619

Goro Uehara
Dept. of Agronomy &Soil Science
University of Hawaii
Honolulu, HI 96822

Lloyd Hossner
Dept. of Soil and Crop Science
Texas A&M University
College Station, TX 77843

Douglas Lathwell
Department of Agronomy
Cornell University
Ithaca, NY 14853

Collaborators, N.C. State University

INIPA (Instituto Nacional de Investigacion y Promocion Agraria-Peru)

EMBRAPA (Empresa Brasileira de Pesquisa Agropecuaria-Brazil)

IVITA (Instituto Veterinario de Investigacion del Tropico y Altura)

UEPAE-Manaus (Unidade de Execucao de Pesquisa de Ambito Estadual de Manaus)

IICA (Instituto Interamericano de Cooperacion para la Agricultura)

PEPP (Proyecto Especial Pichis-Palcazu)

CPAC/EMBRAPA (Centro de Pesquisa Agropecuaria dos Cerrados)

Universidad Nacional Agraria, La Molina, Peru

Center for Soils Research, Indonesia

CIAT (Centro Internacional de Agricultura Tropical)

University of Hawaii

University of Georgia

Wageningen University, The Netherlands

Institute of Development Studies, Finland

Reading University, United Kingdom

Cornell University

U.S. Agency for International Development

Collaborators, University of Hawaii

Center for Soil Research, Indonesia

Institute Pertanian Bogor, Nutrition Department

Andalas University, Padang

U.S. Agency for International Development

North Carolina State University

Collaborators, Cornell University

EMBRAPA (Empresa Brasileira de Pesquisa Agropecuaria-Brazil)

CPAC/EMBRAPA (Centro de Pesquisa Agropecuaria dos Cerrados)

U.S. Agency for International Development

North Carolina State University

Collaborators, Texas A&M University

ICRISAT (International Crops Research Institute for the Semi-Arid Tropics)

ILCA (International Livestock Center for Africa, Ethiopia)

CARE (Cooperatives for American Relief Everywhere)

INTSORMIL (USAID Collaborative Research Support Program on Sorghum and Millet)

IRAT (Institut de Recherches Agronomiques Tropicals, France)

INRAN (Institut National de Recherche Agronomique du Niger, Niger)

University of Niamey, Niger

IFDC (International Fertilizer Development Center, U.S.A.)

University of Wageningen, The Netherlands

FLUP (Forest and Land Use Planning Project, Niger)

USAID/NIGER (U.S. Agency for International Development, Niger Mission)

USAID/Mali (U.S. Agency for International Development, Mali Mission)

OICD (Office of International Cooperation and Development, USDA)

SAFGRAD (Semi-Arid Food Grain Research and Development, Burkina Faso)

ICARDA (International Center for Agricultural Research in the Dry Areas, Syria)

IFRSC (Institute Francais de Recherche Scientifique et Cooperation former ORSTOM-France)

SMSS (Soil Management Support Services, USDA/AID)

I.R.E. (Institut d'Economie Rural, Mali)

AGRHYMET (Centre Regional de Formation et d'Application en Agrometeorologie et Hydrologie
Operationnelle, Naimey, Niger)



Traditional shifting cultivation with slash-and-burn clearing has been singled out as a major con-
tributor to deforestation. This deforestation has raised concern world-wide over the prospect of economic
and ecological disaster in the tropics. The pressure for new clearing arrises in part from the nature
of tropical-rainforest soils, which are typically acid and infertile. Lacking the capital and technology
to overcome these soil constraints, shifting cultivators slash and burn the forest, raise one or two marginal
crops in its fertile ash, then move on to clear a new field. Many hectares of cleared land are abandon-
ed after only one or two years of use.
The objective of North Carolina State University's TropSoils program in the humid tropics is to
develop and transfer, together with national institutions and other universities, improved soil-management
technologies that are agronomically, economically and ecologically sound for productive and sustain-
ed farming systems in the humid tropics and acid savannas. The program has been rooted in the premise
that a stable, productive agriculture is the best means for conserving tropical forests and improving
the living standard of farm families in developing countries. The emphasis in on relieving the pressure
for new clearing by increasing and stabilizing the production of lands already cleared.
Rather than advocate a single solution for the humid tropics, the program seeks to provide a series
of management options to farmers in the process of transition from shifting cultivation to settled
agriculture. These management options cover the principal soils, landscape positions and levels of in-
frastructure development in the humid tropics, and include low-input cropping, continuous cultiva-
tion, agroforestry, legume-based pastures, paddy-rice production, and reclamation of humid tropical
The program's primary research site is the Yurimaguas Experiment Station in Yurimaguas, Peru.
Research is also conducted in Indonesia (supporting TropSoils work in the Sitiung transmigration
settlements); at Manaus, Brazil; in the Cerrado of Brazil; and at Pucallpa and Pichis-Palcazu, in the
Selva of Peru. These sites represent a range of environments. Soils and climate at Manaus, for exam-
ple, are intermediate to Yurimaguas (loamy Ultisols and a weak dry season) and the Cerrado clayeyy
Oxisols with the strong dry season). Pichis-Palcazu has a perudic soil-moisture regime (3400 mm rain-
fall, Ultisols and Dystropepts), and Pucallpa a near-ustic regime (1500 mm). Technologies showing
promise at one site are tested and adapted at others, and are introduced to other countries through
a research network, so as to improve the transfer and application of results.
Some of the work reported here is complete; some is only beginning to produce results. While
the research is presented as individual projects, most projects were conceived as parts of broader in-
vestigations supported by many collaborating scientists and institutions sharing a common goal: to
increase food production while conserving natural resources in developing countries in the tropics.

Low-Input Cropping Systems... 5
Central Low-Input Experiment...6
Tillage-Phosphorus Interactions Under Low-Input Cropping Systems... 7
Calcium and Magnesium Movement in Low-Input Cropping Systems... 9

Legume-Based Pastures... 11
Legume-Based Pastures: Central Experiment... 12
Potassium Dynamics in Legume-Based Pastures...15
Pasture Germplasm Evaluation and Agronomy... 17
Pasture Reclamation in Steeplands...19

Agroforestry Systems...21
Trees as Soil Improvers in the Tropics?...22
Alley-Cropping on Ultisols...23
Peach Palm as a Soil Management Option on Ultisols...26
Gmelina arborea: Intercropping, Coppicing and Nutritional Requirements...28
Contrasting Effect of Pinus caribaea and Gmelina arborea on Soil Properties...29
Improved Fallows... 31
Forest and Soil Regeneration... 3 5
Comparative Soil Dynamics...43
Collection and Propagation of Agroforestry Species...43
Living Fences...44

Continuous Cropping Systems...45
endearing and Post-Clearing Soil Management Practices...46
Tillage With Tractors in Continuously Cropped Ultisols...50
Continuous Cropping: Central Experiment... 52
Production Potential of Corn-Peanut Intercrops in the Humid Tropics... 54
Phosphorus, Zinc and Copper Fertilization... 56
Potassium, Lime and Magnesium Interactions and Corn Yields... 59
Weed Population Shifts Under Continuous Cropping Systems...65
Chemical Weed Control in Corn... 66

Paddy Rice in Alluvial Soils...69
Intensive Management of Alluvial Soils for Irrigated Rice Production... 70

Soil Characterization and Interpretation... 73
FCC Adaptation to Wetland Soils... 74
Volcanic Ash Influence on Transmigration Areas of Sumatra...77
FCC and Site Characterization in Relation to Caribbean Pine... 77
Alluvial Soils of the Amazon Basin...78
Ultisol Dominated Landscapes in Southeastern Peru...80
Soils of Pichis Valley Extrapolation Sites...82
Soil Survey of the Puerto Maldonado Experiment Station...84

Soil Management Research Network...85

Brazil: Extrapolation to Clayey Oxisols...87
0 Soil Nutrient Dynamics and Fertility Management for Sustained Crop Production on Ox-
isols in the Brazilian Amazon...88
Phosphorus Management in Humid Tropical Oxisols...94
Potassium Management in Humid Tropical Oxisols...97
Guarana Fertilization...100
Lime Requirements and Downward Movement of Ca and Mg...102
Planting Dates in Relation to Weather Pattern at Manaus...107
Micronutrient Fertilization on a Typic Acrorthox at Manaus...109
Management of Green-Manure Nitrogen on Oxisols at Manaus...lll
Conditions Other Than Extractable Nutrient Concentrations in the Soil Test Interpretations
for P and Zn...113

Extrapolation of Soil Management Technologies to the Pichis Valley... 115
Low-Input Cropping Systems for the Pichis Valley...116
Legume-Based Pastures in the Pichis Valley... 118
Agroforestry in Steeplands of the Pichis Valley... 120
Soil Erosion and Reclamation...120

Sitiung: Extrapolation to Transmigration Areas of Indonesia...121
Reclamation of Bulldozed Lands...122
Liming in Transmigration Areas...125
Phosphorus Management in Transmigration Areas...131
Effect of Green Manure Applications on Soil Fertility and Crops...134
Potassium Management of Upland Crops... 137
Sulfur Fertilization... 139

Contributions to North Carolina and U.S. Agriculture... 139

Publications... 141


Because it is unlikely that many farmers in the humid tropics can abruptly switch from traditional
shifting cultivation to a fertilizer-based agriculture with continuous cropping, an intermediate system
of low-cost, "low-input" technologies might be useful as a transitional agriculture that could produce
ample food while reducing the need to clear more forest land. The emphasis of this low-input ap-
proach is on adapting plants to the soil, rather than correcting soil constraints to meet the plants'
Previous research and interviews with farmers have led to an experimental low-input system largely
based on traditional farming practices, with innovations introduced in stages. The strategy developed
at Yurimaguas includes slash-and-burn clearing, aluminum-tolerant cultivars, a rotation of upland rice
and cowpeas, zero or minimum tillage, chemical weed control, modest rates of fertilizers, and crop-
residue management. As soil fertility and crop yields decline with time, the system would shift to
such options as continuous cropping, pastures, tree crops or managed fallows.
The low-input system is considered transitory, but has remained productive considerably longer
than expected. In the central experiment, seven continuous crops in three years have yielded a total
of 13.8 t/ha of rice and cowpea grain, without application of lime or fertilizers in a soil with pH
4.4 and 68% Al saturation. Studies presented here indicate that soil fertility is not the major con-
straint in this system, as crop yields remain high with only small amounts of phosphorus and potassium
after the second year.
As predicted by farmers during the interviews, weeds are a primary constraint to sustained produc-
tion. Results are not yet available from a study of weed control in low-input systems, but preliminary
observations suggest that, without tillage, weeds will limit the system's useful life to about six crop
cycles. Research is planned to develop practical weed-control strategies that will prolong the produc-
tivity of the low-input system and increase its appeal as a soil-management option in the humid tropics.


Central Low-Input Experiment
Jose R. Benites, N.C. State University
Marco A. Nurefia, INIPA
Pedro A. Sanchez, N.C. State University

A central experiment was established at Yurimaguas,
Peru, to determine the potential of a low-input, crop-
production system based on a rotation of upland rice
and cowpeas, and to determine how long the system
might remain productive. A one-hectare plot of a ten-
year-old secondary forest fallow was cleared by slash
and burn in July, 1982. In August, a study was
established consisting of upland rice and cowpea with
two treatments: one-half hectare fertilized at the rate
of 30 kg N, 22 kg P and 48 kg K/ha per rice crop,
beginning with the second rice crop, and the other
half-hectare not fertilized. The traditional upland rice

Table 1. Productivity of a low input system during the first 34
Planting Grain Yields
Crop and Cultivar Date Not Fertilized Fertilized*
Month t/ha _
Rice, Carolino Sept. 82 2.4 2.4
Rice, Africano Feb. 83 3.0 3.1
Cowpea, Vita 7 Sept. 83 1.1 1.2
Rice, Africano Dec. 83 2.8 3.2
Cowpea, Vita 7 May 84 1.2 0.9
Rice, Africano Sept. 84 1.8 2.0
Rice, Africano Feb. 85 1.5 2.5
Total 34 Months 13.8 15.3
* 30 kg N/ha, 22 kg P/ha, 48 kg K/ha to Aficano rice crops.

variety was sown with a planting stick (tacarpo) at the
wide spacing common to the region; a post-emergence
herbicide was used to control broad-leaf weeds. After
the first rice harvest, at the time farmers typically aban-
don the field, several practices were introduced:
1. All the rice straw was cut and spread evenly.
2. "Africano Desconocido," an acid-tolerant, im-
proved rice cultivar, was planted with tacarpo at 30
x 50 cm spacing.
3. Rice was followed by an acid-tolerant cowpea
(cultivar Vita 6 or Vita 7), also planted with tacarpo.
4. After threshing, all the rice straw or cowpea stover
was spread evenly on the field.
5. The rotation continued for 34 months, fertiliz-
ing only the rice crops in the fertilization treatment.
6. Pre-plant application of 2-4 D (1.5 L/ha) and
Paraquat (2.5 L/ha) were used for weed control.

Crop Yields
Yields of both rice and cowpea were high. Table
1 shows the yields of seven continuous crops harvested
within three years after the experiment began. A total
of 13.8 t/ha of rice and cowpea grain was produced
during this period without any addition of fertilizer
or lime. These results contrast sharply with those from
the continuous-cropping system, in which yields ap-
proached zero without fertilizers within a year. The
use of Al-tolerant cultivars, maximum residue return
and zero tillage are believed to be responsible for this
difference. The first six rice crops showed no response
to the fertilizers applied. A sharp yield response to fer-
tilizer was observed in the seventh crop, indicating a
fertility decline in the check plots, and modest NPK
applications became important at the end of the third

Table 2. Topsoil (0-15cm) fertility dynamics within the first 34 months of the low input cropping system at
Months after Exchangeable Al Avail.
Clearing Fertilized' pH Al Ca Mg K ECEC Sat. P OM
c mol/L % mg/kg %
3 No 4.4 1.10 0.30 0.09 0.13 1.62 68 20 2.12
14 No 4.6 1.46 0.92 0.28 0.19 2.85 51 13 2.06
Yes 4.7 1.14 0.97 0.27 0.19 2.58 45 18 2.07
34 No 4.6 1.65 1.00 0.23 0.10 2.99 53 5 1.92
Yes 4.6 1.23 1.16 0.20 0.16 2.76 44 16 1.77
CV%2 6 46 46 41 43 9 37 39 20
LSD.052 0.1 0.25 0.17 0.04 0.03 0.20 7 2 0.15
1 Cumulative amount over 34 months: 120 kg N/ha, 88 kg P/ha as OSP, 192 kg/ha as KCI.
2 Comparisons do not include sampling at three months after clearing.


Soil Properties
Topsoil chemical properties (Table 2) improved dur-
ing the period three to 14 months after clearing, in
response to the fertilizer value of the ash, which in-
creased the base status. From 14 to 34 months, there
was little change in pH, organic matter and ex-
changeable bases, and a more favorable Al saturation
level was maintained. It is noteworthy that soil organic
matter decreased only slightly, a sharp contrast to the
25% decrease observed in similar soils under a
continuous-cropping system. Apparently this was due
to the residue return and absence of tillage.
The check plots showed a pattern of declining soil
fertility less drastic than results from continuous-
cultivation experiments. Available P and exchangeable
K decreased below the critical levels (12 ppm for P
and 0.15 cmol/L for K). The small P and K additions
in the fertilized treatment were apparently sufficient
to offset this decrease.

It seems reasonable to assume that this low-input
system could be sustained by modest fertilizer applica-
tions. The crucial limiting factor is a gradual buildup
of grassy weeds, particularly during the rice crops. The
effect of fertilizer application on weed growth does
not appear to be important. The studies on weed con-
trol for low-input systems show that much needs to
be learned about how to control these weeds
economically by herbicides, and, as a consequence,
avoiding tillage does not help, either. It is possible to
control the weeds with hand labor economically, or
with herbicides at a prohibitively high cost. Conse-
quently, we have reached a crossroads in this transi-
tion technology. For the low-input system to succeed,
effective and affordable weed control measures are
needed to bridge the gap between year two and year
Results are promising for the low-input strategy as
a transition from shifting agriculture to a more per-
manent system of management. With relatively sim-
ple practices farmers can grow seven crops where they
were able to grow only one. This system cannot be
considered stable at this time, and is viewed as a tran-
sition technology.

Tillage-Phosphorus Interactions
Under Low-Input Cropping Systems
Mwenja P. Gichuru, N.C. State University
Pedro A. Sanchez, N.C. State University

This experiment was begun in May 1982 to study
the management of phosphorus in the low-input crop-
ping system being developed on an Ultisol at
Yurimaguas, Peru. Its objectives were to study the ef-
fect of no-till versus rotovation on continuous cropp-
ing without liming, using acid-tolerant crops, and to
determine efficient rates and sources of phosphorus
for a rotation of acid-tolerant upland rice (Oryza sativa
L.) and cowpea (Vigna unguiculata).
Relevant soil chemical properties at the initiation
of the experiment are shown in Table 1. The main
plot treatments were tillage methods: 1) no-till, with
broadcast fertilizers, and 2) rotovation before each crop,
with fertilizers broadcast and incorporated to a depth
of about 8-10 cm. Subplot treatments were phosphorus
sources, ordinary superphosphate and Sechura
phosphate rock. The sub-subplot treatments were 0,
25, 50, 100 and 200 kg P205/ha. Crop residues were
left on the surface.

Tillage Effects
Figure 1 shows the influence of tillage on relative
yields of five consecutive crops (relative yield is a
percentage of maximum absolute yield). The first crop
produced significantly more grain in rotovated
treatments compared with no-till plots. The better
growth in rotovated plots is probably due to improv-
ed soil physical properties and a better distribution of
nutrients from the ash left by slash-and-burn.
The second and third crops' grain yields showed no
differences due to tillage treatments. The advantage
of rotovation in terms of improved physical proper-
ties may have disappeared, probably due to constant
traffic during weeding and harvest. Rotovation follow-
ed by human traffic is likely to result in greater soil
compaction and poorer crop performance, compared
with no-till.
In the fourth and fifth crop dramatic yield reduc-
tions occurred in rotovated treatments compared with
no-till plots. Rotovated treatments produced 80% and
76% relative yields, respectively, in the fourth and fifth
harvests (Figure 1). The reason for this sudden negative
response to tillage is not clear. Bulk density
measurements in the 0 to 7.5 cm depth, taken im-
mediately after harvesting the fifth crop, showed high


Table 1. Relevant soil chemical properties at the initiation of the tillage-phosphorus experiment.
Soil Exchangeable Effective Al Avail P
Depth pH Acid Ca Mg K CEC Sat. (Mod. Olsen)

cm c cmol/L % mg/kg
0-15 4.5 1.9 1.3 0.4 0.15 3.7 53 14
15-30 4.3 3.9 0.5 0.1 0.11 4.6 84 4
30-45 4.3 4.1 0.4 0.1 0.09 4.7 88 3

density values both in rotovated plots (1.46 g/cc) Phosphorus Effects
in no-till treatments (1.42 g/cc). These high values The grain yields ranged from a low of 1.41 t/ha
believed to be mainly due to human traffic, to a high of 3.26 t/ha among the five crops. However,
no significant differences were found between the
(3.6) (2.4) (1.4) (2.5) (2.7) phosphorus sources in any of the crops. Thus,
0 phosphate rock was comparable to ordinary super-
No-tl (95%) phosphate in supplying P.
The data in Figure 2 show that, in general, only
the control produced less than 80% relative yield in
the first three consecutive harvests. The first P incre-
ment produced slightly over 80% while higher rates
generally produced over 90% relative yields during
Rotovated (91%) the same period. The absence of a strong P response
Sis believed to be due to the initially favorable P status
1 2 3 4 5 in the soil, as shown in Table 1.
P R R C R R The fourth crop showed vigorous growth, probably
ure 1. Relative yields of five consecutive crops as due to N fixed by the cowpea crop, but severe lodg-
uenced by tillage. Numbers in parenthesis at the ing occurred during the grain-filling stage. All treatments
are maximum yields in tons/ha, and numbers to produced yields over 90% of the maximum yield, but
right are average yields of five crops. R = rice; the grain quality was very poor; it had a high percen-
cowpea. tage of half-filled grain because of the lodging. The
fifth harvest produced yields at least 80% of maximum,
(3.1) (2.7) (1.6) (3.1) (3.1) 1
with the 100 kg P20,/ha rate producing the max-
1'OO 96 imum yield.

1 2 3 4 5
Crop R R C R R
Figure 2. Relative yields of five consecutive crops as
influenced by phosphorus application. Numbers in
parenthesis are maximum yields in tons/ha, and
numbers to the right are P rates (kg P = 20 = 5/ha) and
average relative yields, respectively.

1. The effect of tillage appears to follow a trend in
which rotovation is superior to no tillage during the
first crop, about equal for the second and third crops,
and inferior during the fourth and fifth rice crops. A
combination of initial tillage followed by no-till ap-
pears advantageous, but firmer conclusions require ad-
ditional data.
2. Rock phosphate at the rate of 50 kg P20,/ha,
applied to the surface, was sufficient to produce 95%
of the maximum yields in the low-input system, bas-
ed on crop varieties highly tolerant to aluminum. A
total of 12.9 t/ha of rice and cowpea grain was pro-
duced by five crops on an Ultisol with pH 4.5 and
no lime application.
3. The data show no significant interaction between
tillage and phosphorus.

are 1



C =



Calcium and Magnesium Movement
In Low-Input Cropping Systems
Mwenja P. Gichuru, N.C. State University
Pedro A. Sanchez, N.C. State University
Jos6 R. Benites, N. C. State University

This experiment was initiated in May 1982 to study
the effect of small additions of dolomitic limestone or
gypsum on the downward movement of Ca and Mg
as part of the low-input, crop-production strategy under
development at Yurimaguas. A second objective was
to determine the effect of tillage methods on the rates
at which these cations moved into the subsoil.
Main plot treatments were a combination of tillage
methods and nutrient incorporation: 1) no-till and
broadcast fertilizers with no incorporation; 2) strip-
tillage, with fertilizers applied in strips 15 cm wide
(about one-third the total area) and incorporated with
a hoe to a depth of approximately 8 to 10 cm, and
3) rotovator tillage with fertilizers broadcast and in-
corporated to rotovator depth (about 8 to 10 cm).
Subplot treatments were calcium sources dolomiticc
limestone and gypsum). Sub-subplot treatments were
0, 33, 100, 300 and 600 kg Ca/ha. Treatments with
gypsum as the Ca source were supplemented with Mg
from MgSO4. 7H20 in amounts equivalent to that
supplied by dolomitic limestone. The crop rotation
was rice (Oryza sativa L.), rice, cowpea (Vigna
unguiculata), rice, rice, cowpea.

Exchangeable Ca
Strip-tillage resulted in higher initial levels of ex-
changeable Ca in the topsoil because the calcium was
applied to about a third of the soil surface area, but
there were no significant tillage effects at other depths,
and the effect on the topsoil had disappeared at twen-
ty months.
The pattern of exchangeable Ca distribution appears
nevertheless to be influenced by tillage treatments.
Generally, seven months after application, dolomitic
limestone resulted in significantly higher exchangeable
Ca in the topsoil compared with gypsum treatments.
The lower exchangeable Ca in the topsoil of gypsum-
treated plots was due to downward movement, as in-
dicated by higher exchangeable Ca at lower depths
compared with the check and dolomitic-limestone
treatments. Downward movement of Ca was more
pronounced when gypsum was incorporated than
when it was applied to the surface, as indicated by a
larger bulge at the 15 to 16 cm depth in both strip
tillage and rotovated treatments compared with no-
till treatments (Figure 1).
The effect of gypsum on downward movement of
Ca was still measurable 20 months after application.
However, the exchangeable Ca bulge at the 15 to 45
cm layer had slightly shrunk despite a substantial
decrease in exchangeable Ca in the above layers.
Slight increases in exchangeable Ca were found at
100 cm, suggesting that some Ca from gypsum had
moved beyond the sampled depth. The application of

1 2

- i No Till
S7 mo.


1 2
I '--| --

-r om
*- I No till
20 mo. -

- b

Exchangeable Ca (cmol (p+) L)

1 2 3 0

-r --

Strip till

7 mo.

Strip till
-! 20 mo.

1 2

1 2
'I I

- Fp Rotovated -
-L f 20 mo.
LJ kg/ha Source

i 600 Gypsum
,I --- 600 Dolomite

Figure 1. Exchangeable Ca as influenced by Ca source and rate under three tillage systems.




0 0





Table 1. Exchangeable Mg as influenced by the application of dolomitic limestone and gypsum.
Calcium No-Till Strip-Till Rotovated
Applied Dol' Gyp2 Dol Gyp Dol Gyp

kgCa/ha c mol/L

7 Months After Application
0 .21 .28 .27 .26 .22 .20
33 .27 .25 .35 .29 .30 .20
100 .31 .17 .75 .23 .37 .39
300 .67 .23 1.03 .24 .55 .23
600 .56 .12 .98 .16 .90 .18
Rate LSD (.05) .25 NS .45 NS .19 NS
Source LSD (.05) 0.09 0.14 0.08

20 Months After Application
0 .15 .17 .18 .20 .11 .13
33 .17 .15 .19 .21 .14 .10
100 .17 .17 .47 .20 .20 .24
600 .58 .26 .97 .41 .53 .29
Rate LSD (.05) .14 NS .25 .16 .10 .11
Source LSD (0.05) 0.06 0.08 NS

1 Dol = dolomitic limestone
2 Gyp = gypsum

300 kg Ca/ha showed a similar trend (data not
presented here), but the magnitudes were smaller com-
pared with the application of 600 kg Ca/ha. An ex-
ception was observed in no-till treatments, where the
calcium bulge was comparable to that of the 600 kg
Ca/ha rate in the rotovated treatment.

Exchangeable Mg
Statistical analysis of exchangeable Mg data reveal-
ed that treatments had little or no significant effect
below the 0 to 15 cm layer. This was probably due
to the initial variability of exchangeable Mg in the soil
(coefficients of variation of 42, 87, 108, 55, 67 and
59% at increasing depth intervals, respectively). Ex-
changeable Mg status of the topsoil, however, was im-
proved by the application of dolomitic limestone (Table
1). But little or no change in exchangeable Mg occurred
in gypsum treatments after seven months, although
they had received supplemental Mg in equivalent
amounts supplied by dolomitic limestone. The high
solubility of the Mg source may have resulted in rapid
leaching beyond the sampled depth. A similar trend
was observed 13 months later, except that the highest
rates of gypsum produced some increase in ex-
changeable Mg. In these high rates of gypsum the sup-

plemental magnesium was split over the cropping cycles
and applied at planting.

1. The applications of 300 or 600 kg Ca/ha as gyp-
sum resulted in substantial downward movement of
Ca in less than two years, whereas dolomitic limestone
application resulted in little or no change in ex-
changeable Ca below the 0 to 15 cm depth. However,
it was difficult to detect the effect of rates lower than
300 kg Ca/ha, probably because of variability in the
field, which had recently been cleared by slash and
2. Rotovation resulted in greater and more uniform
downward movement of Ca, compared with surface
3. Gypsum application will result in subsoil enrich-
ment with Ca in a short time.
4. The data from this experiment suggest that, 20
months after application, some Ca had moved
downward beyond the sampling depth.
Results of this study indicate that relatively low rates
of gypsum can promote a significant movement of
calcium into the subsoil.


Pastures are good news and bad news for soil management in the tropics. Well-managed, they pro-
tect the soil, require relatively few cash inputs, make good use of soils unsuitable for food crops, and
produce milk and meat with grazing animals, which recycle most of the nutrients they consume. But
poorly managed pastures are an economic and ecological liability. The use of pasture species badly
adapted to tropical soils and environments leads to poor animal nutrition and therefore low produc-
tivity. Many thousands of hectares of rainforest have been cleared for pastures, only to be abandoned
as the pastures became degraded by overgrazing, soil compaction and erosion.
Pastures research at Yurimaguas, Peru has been closely integrated with the Tropical Pastures Pro-
gram of Centro Internacional de Agricultura Tropical (CIAT), and with INIPA's National Selva Pro-
gram, which is now conducting most of the agronomic studies. Research reported here has for the
most part concentrated on the relationships among soil fertility, pasture quality, pasture persistence,
grazing, the transfer of N from legumes to grasses, and the recycling of nutrients. Several promising
grass-legume pastures have produced stable pastures and animal weight gains many times greater than
those on the typical humid-tropical farm. After years of sustained production at Yurimaguas, the best
of these associations and management techniques are being tested at extrapolation sites.
Some of these extrapolation studies have been initiated near Pucallpa, Peru, but are not reported
because they are still in the establishment phase. Two projects are examining the N contribution of
legumes to mixed pastures, and determining the optimum schedule for herbicide application during
pasture establishment.
Central in the work reported here is the role of legumes in pasture associations. Legume-based pastures
have not been studied extensively in humid tropical environments, and the information about their
response to such management variables as fertility, grazing pressures and establishment methods are
expected to be of use throughout the humid tropics.


Legume-Based Pastures:
Central Experiment
Rolando Dextre, INIPA
Miguel A. Ayarza, N. C. State University
Pedro A. Sanchez, N. C. State University

The central experiment with pastures, begun in
1980, has sought to develop a practical management
system for improving tropical pastures with stable mix-
tures of acid-tolerant legumes and grasses. Since its in-
itiation, the experiment has evolved in response to new
information gained from companion experiments and
from collaboration with the International Center for
Tropical Agriculture (CIAT) and its Tropical Pastures
Network. Work so far has shown that proper manage-
ment of some grass-legume associations can greatly im-
prove the stability and productivity of previously
degraded tropical pastures, while conserving the soil-
resource base.
The objectives of this experiment are 1) to measure
pasture and animal productivity on different associa-
tions, in terms of daily weight gain and annual

liveweight production; 2) to evaluate the compatibili-
ty and the persistence of the different grass-legume mix-
tures under grazing, and 3) to evaluate changes in soil
properties as a consequence of long-term pasture
Four associations remain unchanged, but during the
four years the project has been in progress, Panicum
maximum + Pueruaria pbaseoloides was replaced by An-
dropogon gayanus + Centrosema macrocarpum 5056 in
October 1984. Table 1 shows the species, animal
management and years of evaluation for each associa-

Animal Production and Botanical Composition
Changes in animal production are shown in Figure
1. During the first year, most associations yielded above
600 kg/ha/yr of liveweight gains. However, only C.
pubescens and the Brachiaria-based pastures were able
to maintain that level of productivity beyond the first
A remarkable performance by Centrosema pubescens
438 has been observed. After the first year, A. gayanus
disappeared from this mixture because of establishment

Table 1. Grazing trial mixtures with starting dates of continuous and alternate grazing. (Planting dates: March
to July, 1980 for first four, February, 1982 for fifth and October, 1984 for sixth mixture.)
Treatments Initiation of Grazing Initiation of Grazing Days
Grass Legume Continuous Grazing Days Alternate Grazing Through Sept. 1985
B. decumbens/D. ovalifolium 350 Nov. 15, 1980 238 Oct. 7, 1981 370
P. maximum/P. phaseoloides 9900 Nov. 20, 1980 238 Oct. 6, 1981 1005
A. gayanus/S. guianensis 134-186 May 15, 1981 57 Oct. 6, 1981 1370
C. pubescens 438 0 Oct. 7, 1981 1370
B. humidicola/D. ovalifolium 350 -0 Oct. 10, 1982 1056
A. gayanus/C. macrocarpum 50561 0 May 1985 120
1 Replaced Pm/Pp.

Table 2. Nutrient levels of grass and legume mixture components and tannin content of legumes May, 1983.
Species N P K Ca Mg S Zn Tannin
% ppm %
Centrosema pubescens 438 4.46 0.26 1.30 0.93 0.26 0.18 31 2.5
Desmodium ovalifolium 350 2.69 0.16 0.72 0.83 0.21 0.12 10 21.0
Stylosanthes guianensis 136 3.89 0.22 0.32 1.13 0.32 0.16 37 4.0
Pueraria phaseoloides 9900 3.90 0.22 1.30 0.45 0.32 0.12 25 4.0

Brachiaria decumbens 606 2.38 0.24 1.57 0.42 0.42 0.11 14 -
Brachiaria humidicola 1.70 0.21 1.66 0.28 0.27 0.10 16 -
Androgogon gayanus 2.11 0.17 1.05 0.36 0.14 0.11 10 -


Years: 1 2 3 4 5 1 2 3 1 2 3 4 1 2 3 4 5 1 2 3
Mixture: B. decumbens B. humidicola Centrosema A. gayanus P. maximum
D. ovalifolium D. ovalifolium pubescens 438 S. guianensis P. phaseoloides

Figure 1. Changes in animal production from grass-legume pastures with time in Yurimaguas. A number at
the top of a bar is an average annual percentage of legume in the mixture.

problems. In spite of this, the legume has persisted and
maintained high levels of production. This must be
related to the high quality of the species (excellent
palatability and high nutritional content) in comparison
to the other legumes in the experiment (Table 2).
The change from continuous to alternate grazing has
favored the presence of grass in the mixtures of B.
decumbens + D. ovalifolium and B. bumidicola + D.
ovalifolium (Table 2). This has resulted in sustained
animal gains based mainly on consumption of the grass,
since the legume is of low palatability (Figure 1). Levels
of D. ovalifolium in the mixture with B. decumbens,
however, decreased sharply in May 1984. This was
related to an unusual intake of legume and a rejec-
tion of the grass during a short drought at that time.
In the mixture of A. gayanus + S. guianensis the an-
nual animal performance was closely related to the con-
tent of legume in the association (r2 = 0.75). The
legume almost disappeared in the same period as in-
dicated for D. ovalifolium with b. decumbens.
Poor animal gains were observed in P. maximum

800 -



0 25 50 75 100
Legume in Forage on Offer (%)

Figure 2. Effect of the content of legume in the forage
on offer on animal gains in three grass-legume
pastures growing in an Ultisol of Yurimaguas. (%
legume expressed as annual mean value.)



E 3

5 2


1 2 3 4 5

1 2 3 4


o 4



4. Brachiaria decumbens +
Desmodium ovalifolium
5. Brachiaria humidicola +
Desmodium ovalifolium

D 1980


1 2 3 4 5

Figure 3. Changes in the topsoil chemical properties
in five grass-legume associations after four years of

+ P. pbaseoloides after three years of evaluation. This
appears to be related to the increasing content of
Pueraria. The negative effect of this legume on animal
gains is reflected in a negative correlation coefficient
of -0.77 (Figure 2).
Centrosema pubescens 438 as a pure legume and the
B. bumidicola + D. ovalifolium mixture were the best
pastures in terms of individual animal gains and kg/ha

Table 3. Average annual productivity of five associations under
grazing. Stocking rate 4.4 animals/ha.
Years of
Association Grazing Liveweight Gains
kg/ha/yr g/an/day
B. humidicola/D. ovalifolium 2 691 429
B. decumbens D. ovalifolium 4 626 379
Centrosema pubescens 438 3 619 459
A. gauanus S. guanensis 4 467 357
P. maxim/P. phaseoloides 3 455 296

weight/ha (Table 3). The association of B. bumidicola
+ D. ovalifolium performed well but had only two
years of evaluation.

Changes in Soil Chemical Properties
In order to follow the changes in the soil properties
as a function of time in these pastures, the treatments
were sampled in 1980 (before grazing, and six months
after fertilization for establishment) and in 1983 and
1984. Figure 3 shows that some of the properties have
changed in the 0-20 cm depth in several pastures. It
is interesting to observe that exchangeable acidity has
decreased in all associations except P. maximum +
P. pbaseoloides and that pH has increased to 5.0 in B.
decumbens + D. ovalifolium. Phosphorus levels have
increased in all associations, probably due to
maintenance applications of phosphorus (25 kg P
ha/yr). Most changes have occurred in the 0-20 cm

Progress in 1985
Several changes have been introduced in the manage-
ment of the associations. D. ovalifolium and S. guianen-
sis were replanted in their respective associations in
order to investigate the effects of different grazing-
management procedures.
Grazing periods were reduced from 42 to 28 days,
and stocking rates were increased from 4.4 to 5.5
animals per ha during the rainy season, and maintain-
ed at 4.4 during the dry season, in pastures of B.
decumbens + D. ovalifolium and B. bumidicola + D.
ovalifolium. These changes more efficiently use the high
levels of available forage and prevent losses in the quali-
ty of the grass when the pasture remains ungrazed for
longer periods.
The C. pubescens pasture is being maintained with
the same 4.4 animals/ha and 28-day grazing and resting
periods. Stocking rates for the associations A. gayanus
and S. guianensis and A. gayanus and C. macrocarpum
have been adjusted to 3.3 animals/ha and 20 to 28
days of grazing.
The new management has produced positive results
in B. bumidicola and D. ovalifolium. In seven months
of grazing, animal gains passed the annual gains for
the previous two years. Furthermore, individual gains
are excellent at this time, and if this trend continues
for the remaining four months of the year, it will be
possible to reach levels up to 900 kg liveweight per
year in this pasture. The sward is in excellent shape
with a 40% legume intimately mixed with the grass.
The Brachiaria decumbens and D. ovalifolium pasture


Table 4. Preliminary results of animal production levels in five associations under grazing until Sept., 1985,
Yurimaguas, Peru.
Stocking Rates Animal Gains Legume
Associations Rainy season Dry Season Kg/live weight/ha gm-head/d %
Brachiaria humidicola + 5.5 4.4 697.6 597.6 40
D. ovalifolium1
B. decumbens + 5.5 4.4 530.2 470.3 40
D. ovalifolium1
Centrosema pubescens 4381 4.4 4.4 399.4 419.8 100
A. gayanus + 3.3 3.3 387.2 751.4 20
C. macrocarpum 50662
A. gayanus + Stylosanthes 3.3 3.3 282 641.8 60
1 Grazing from January-September
2 Grazing from May-September

is once again showing the problems observed in 1984.
Although the grass appears to be doing well, the
animals are grazing only the legume and making little
use of the grass. This seems to indicate that this grass
is very sensitive to drought stress normally occurring
during the dry season. As a result, digestibility pro-
bably falls to levels preventing its consumption.
Measurements will be conducted to test this statement.
Centrosema appears likely to maintain the animal pro-
ductivity observed in 1984. The other two pastures
have only six months of evaluation and it is too early
to make any conclusions. A. gayanus + C. macrocar-
pum 5065 is well established with a 15% legume base.

1) To date, Centrosema pubescens 438 as a pure legume
and the B. bumidicola + D. ovalifolium mixture have
provided the best pastures in terms of individual animal
gains and kg/ha weight/ha.
2) The association of B. humidicola + D. ovalifolium
performed well but had only two years of evaluation.
3) Exchangeable acidity has decreased and soil P
levels have increased in most of the pastures during
the course of this study.
4) Adjustments in stocking rates and grazing periods
have improved the quality of grasses and made more
efficient use of available forage in several of the grass-
legume associations.

Potassium Dynamics
In Legume-Based Pastures
Miguel Ayarza, N. C. State University
Pedro A. Sanchez, N. C. State University
Rolando Dextre, INIPA

Tropical pastures on acid soils are stable and pro-
ductive only when nutrients are sufficient to sustain
a vigorous forage crop. Maintaining this fertility re-
quires a management method that takes into account
the nutrient leaching common in areas of high rain-
fall, as well as the cycling of nutrients among soil,
forage and animals. This study, which was conducted
at the Yurimaguas Experiment Station, concentrated
on one nutrient, potassium. Its objectives were 1) to
quantify leaching losses of K in pastures under clipp-
ing and grazing; 2) to monitor the effect of K levels
on the productivity of the pasture and on the dynamics
of K in the soil; 3) to estimate the effect of K return
by animal excretions, and 4) to compare estimated K
losses from pastures with losses from crops grown in
the same area.
The grazing experiment was a factorial of three an-
nual rates of K fertilization (0, 50 and 100 kg K/ha)
by two grazing pressures (11 and 7.8 kg green forage
dry matter/100 kg liveweight), with three repetitions.
Two additional experiments were established on 3 x
4 m plots with K rates of 0, 25, 50, 75, 100, 150,
and 300 kg K/ha/yr. The first, in which some plots
had clippings removed while others had clippings
returned, provides a comparison of the effect of graz-
ing on K dynamics. The second was a bare-plot ex-
periment designed to account for soil chemical and


Table 1. Effect of cumulative rainfall on exchangeable K status
of 0-5 cm layer in bare and clipped plots.
Bare Plots Clipped Plots
Cumulative Rainfall, mm Cumulative Rainfall, mm
K Rate 25 150 25 150
kg/ha Exchangeable K, cmol/L
0 0.07 0.06 0.07 0.07
25 0.16 0.09 0.10 0.07
50 0.26 0.12 0.15 0.16
75 0.28 0.14 0.17 0.18
100 0.32 0.18 0.18 0.19
100 0.42 0.29 0.25 0.24
300 0.78 0.47 0.61 0.48
LSD .05 0.09 0.28 0.09 0.28

Exchangeable K (cmollL)

Figure 1. Effect of application of potassium on the
distribution of exchangeable K in the profile of a san-
dy loam Ultisol, following 25 mm of rain.

physical properties related to K leaching, and to
estimate the effect of plant growth on K dynamics.
Four hectares were planted with a mixture of
Bracbiaria bumidicola and Desmodium ovalifolium in
December, 1984. Potassium treatments were applied
on May 13, 1985, and grazing began on July 4.
Potassium distribution in the soil with depth was
monitored as a function of precipitation. Changes in
soil and plant K were determined in the small plots
and grazing experiments. Amounts and composition
of plant residues were evaluated under grazing. The
effect of urine on the return of K to the soil is being
studied, comparing plant growth and changes in soil
K in affected vs. unaffected areas under grazing.

Exchangeable K Dynamics
The Ultisol in the experimental area was characteriz-
ed as having a sandy loam topsoil texture, Al satura-
tion of 74% and K contents of 0.06 cmol/L, far below
the critical range of 0.15 to 0.20. Most of the K ap-
plied to the bare soil, however, remained in the top
5 cm, even though 25 mm of rain fell during applica-
tion (Figure 1). After an additional 159 mm of rain,
topsoil exchangeable K levels remained constant, ex-
cept in plots receiving 300 kg K/ha, which lost 0.13
cmol K/L (Table 1). In the clipping plots, where the
grass-legume mixture was growing rapidly, this decrease
at the 300 kg K/ha rate was more pronounced (Table

Pasture Response
Results of the first cut of the mixture growing in
the small plots are presented in Table 2. Total dry
matter was not affected by the potassium rates,
although the content of the grass in the mixture tend-
ed to increase with K rate. Foliar K in the grass in-

Table 2. Effect of seven K rates on dry matter yields and uptake of Brachiaria humidicola and D. ovalifoliumb
growing in association in an Ultisol of Yurimaguas, Peru (first cut).

Grass Plant K Content Total Uptake
K Applied Dry Matter Component Grass Legume B.humidicola D.ovalifolium
kg/ha g/m2 % % K kg/ha
0 196.6 a 54 0.96 c 0.74 d 1.32 0.54
25 217.8 a 61 2.21 b 0.91 d 1.83 0.74
50 219.5 a 62 1.51 b 0.90 d 2.29 0.66
75 183.6 a 63 1.62 b 1.31 ab 1.80 0.93
100 201.5 a 51 1.65 b 1.06 bcd 1.72 1.04
150 229.3 a 59 2.19 a 1.24 abc 3.00 1.12
300 268.0 a 71 2.56 a 1.58 a 4.95 1.18
LSD 67.5 14


Table 3. Effect of K applications on dry yields and botanical composition in a pasture of B. humidicola and
D. ovalifolium before grazing.
Dry matter K in K in
K Applied Yield Grass Legume Grass Legume
kg/ha t/ha %
0 2.51 53 a 47 a 0.98 1.02
50 3.03 65 b 35 b 1.51 1.29
100 2.95 68 b 32 b 1.72 1.38

creased with potassium applications, and the highest
application rate multiplied foliar K 2.6 times the level
found in grass with no K applied. Compared to the
grass, the legume accumulated less foliar K in response
to K fertilization.
The relation between the potassium content in leaves
and dry matter yield was not consistent in either
species, suggesting a high luxury consumption by the
grass. Only at 300 kg K/ha was there an increase in
dry matter of B. humidicola.
The effect of K fertilization on the growth and K
content of leaves before grazing is shown in Table 3.
As observed in the clipping plots, grass composed a
larger share of the forage mixture when the pasture
was fertilized with K. This was true under grazing,
as well.

Animal Behavior
Grazing has shown an expected animal preference
for the grass, regardless of the potassium level. Little
consumption of the legume was observed, even at the
higher grazing pressure. However, grass recovery after
grazing was excellent, especially when potassium was
present. After 86 days of grazing there was an overall
increase of 36 kg per animal. Individual gains appear
to be slightly better in the 50 kg K/ha treatment.

The first stages of this continuing study yield the
following initial conclusions:
1) Rainfall did not significantly reduce exchangeable
K in the topsoil except on plots with the highest rate
of K fertilization (300 kg K/ha).
2) Plots receiving K showed an increased percen-
tage of grass in the forage mixture.
3) Foliar K levels increased sharply with K fertiliza-
tion in grass, but only slightly in the legume.

Pasture Germplasm
Evaluation and Agronomy
Rolando Dextre, INIPA
Miguel Ayarza, N.C. State University
Jose M. Toledo, CIAT
Mario Calder6n, CIAT
Jill Lenn6, CIAT
Esteban Pizarro, CIAT

Twenty-three species of grasses and legumes have
been tested for their adaptation to Yurimaguas condi-
tions according to the methodology suggested by CIAT
for regional trials type B. The objectives of these studies
are 1) to introduce new acid-tolerant grass and legume
accessions through regional trials and seed production,
and 2) to evaluate tolerance to spittlebug in grasses
and to anthracnose in legumes.
Species are harvested at four cutting intervals (three,
six, nine and 12 weeks). Cover and resistance to pests
and diseases are also recorded.

Yields at Cutting Interval
Figure 1 presents the effect of four intervals of cut-
ting on four legumes and three grasses growing on an
Ultisol at Yurimaguas. At the 12-week cutting inter-
val, Bracbiaria dictyoneura produced yields similar to
those of B. decumbens and Andropogon species. Ground-
cover ability and vigorous growth of this species make
it a promising grass for further evaluation under graz-
ing. Among legumes, the Centrosema macrocarpum
5065, 5062, and 5452 appear to have potential for
the Yurimaguas areas. Stable yields during dry and wet
periods is an important attribute for their selection.
Centrosema pubescens 5189 produces yields similar to
those from the 438 ecotype, which is the most suc-
cessful legume at present under Yurimaguas conditions.
New Desmodium accessions have not shown superior
yields to the 350 ecotype.
These species seem to require a minimum of six
weeks to recover from cutting.


Minimum Precipitation
A C. macrocarpum 5065
* C. pubescens 438
O D. ovalifolium 350
O Z. latifolia 728 A /

i I I 12
3 6 9 12

A A. gayanus 621
* B. decumbens
D B. dictyoneura

I I I I II 10
3 6 9 12 0 3 6 9 12 a E
E 8n 8
Figure 1. Effect of four intervals of cutting on four > 6
legumes (a) and three grasses (b) growing in an 4
Ultisol of Yurimaguas. (Mean of two years.) E N 2
z 2

Seed Production of Promising Forage Species
Four grasses and five legumes, drawn from the most
promising accessions described above, are under evalua-
tion for their potential to produce seed, an important
attribute of a good forage species. All grasses except
A. gayanus have to be propagated vegetatively, as they
do not produce viable seed under Yurimaguas condi-
tions. Between four and five hectares could be planted
with the available material of the Brachiaria accessions.
Relatively good yield potential was observed for Cen-
trosema accessions 5713 and 5452.

Tolerance to Pests and Diseases
Twenty-six accessions of Stylosanthes guianensis have
been studied for tolerance to anthracnose and 26
ecotypes of Brachiaria have been studied for tolerance
to spittlebug. Results after two years of observation
indicate a wide range of tolerance to anthracnose in
most S. guianensis species. Although the disease is pre-
sent, it seriously affects only five accessions: 97, 1091,
1017, 1951 and 1893. S. guianensis 136 and 184 are

Maximum Precipitation

0 U

m B. decumbens C3 B. emini
= S.P. hibrido E3 B. ruziziensis
ME" B. brizantha Q B. radicums
MI B. humidicola
40" M S B. dictyoneura


a 10

Figure 2. Animal preference for 26 brachiaria acces-
sions ranked by number of times grazed every 15
minutes in 18-hour period, and by time animal spent
grazing in 18 hours.

the most tolerant accessions to this disease. Spittlebug
does not appear to be a major problem in any of the
species of Brachiaria under testing. High levels of in-
festation can be observed in B. bumidicola pasture under
grazing; however, there is no apparent damage.

Animal Preference
A new criterion was added to the Brachiaria acces-
sions. Animal preference was tested in March, 1984
(rainy season) for a period of 18 hours using four
animals. The entire area was fenced and the animals
were left to graze after spending a night without food.
The number of times the animal grazed each acces-
sion was recorded every 15 minutes and the total
period of time the animals grazed each species are il-
lustrated in Figure 2. Results indicate a difference in
preference not only among species but also within
species. Brachiaria decumbens 6009, B. hybrid 6298 and
B. bumidicola 629 were among the most preferred
species. It is interesting to observe the low preference
for B. dictyoneura accessions. A new evaluation is pro-
posed for the dry season next year.

2400 -

1800 -

1200 -




1800 -

1200 -


3 6 9 12


2400 -

1600 -

800 -


Pasture Reclamation in Steeplands
Rolando Dextre, INIPA
Miguel A. Ayarza, N.C. State University
Pedro A. Sanchez, N.C. State University

There is a substantial area of pasture land in the
humid tropics that has very low productivity because
of poor soil management, overgrazing or badly adapted
forage species. The purpose of this project is to develop
a simple technique for reclaiming degraded pastures
in Ultisol steeplands.
A two-factor experiment was installed in a degrad-
ed pasture occupying a 5.18 ha watershed with
sideslopes of 20 to 50%. Treatments were establish-
ed in an amphitheater fashion, following slope con-

New-Project Update
This project has not been under way long enough
to yield substantive reports, but should be mentioned
because of its importance to the program as a whole.

tours, with tillage methods as main plots and improv-
ed species as subplots. The tillage treatments are 1)
zero tillage (pastures planted in an array of holes 20

cm in diameter); 2) minimum tillage (pastures planted
in 50 cm wide, rototilled furrows 2 m apart); and 3)
total tillage (50 cm wide furrows, with spaces between
furrows gradually cultivated as forage crops grow). The
only fertilizer was Bayovar rock phosphate, applied
at the rate of 50 kg P/ha in the hole or furrow.
Although data are not yet available, visual evalua-
tion of the effect of treatments on the replacement
of the native species (torourco) can be summarized as
follows: The grasses, Brachiaria bumidicola and B.
decumbens, which are planted by vegetative propaga-
tion, are both well established, but B. bumidicola ap-
pears to be better in the zero-tillage treatment. Both
grasses have almost replaced the torourco between fur-
rows in the minimum-tillage treatment.
The legumes tested, which were planted by seed,
initially did not compete as well with the torourco. Cen-
trosema pubescens 438 is doing better than D. ovalifolium,
but both require some tillage before planting, in order
to diminish competition for light and nutrients in the
early stages of growth.
Evaluation of biomass production and percent cover
of native grass is planned at four and eight months
after planting. The information gathered so far sug-
gests that some tillage is required to establish grass and
legumes successfully.


Anyone regarding the luxuriant growth of a rainforest might gather that trees are the natural voca-
tion of the humid tropics. Because a variety of commercially valuable species can succeed in this en-
vironment, agroforestry, the production of trees alongside crops or pastures, is an important soil-
management option. In many cases, annual crops may be the wise choice for alluvial sites and soils
with a high base status, while trees might be better suited to rolling, upland sites, where they require
relatively little maintenance, tolerate acid soils, and afford long-term protection against erosion. Even
so, most of the data available on tropical tree production are from areas with fertile soils. Collection
and screening of the wide variety of germplasm available for use on acid soils have only begun.
Agroforestry projects at Yurimaguas are focused on tree-soil relationships on well-drained, acid soils.
This work has been linked to agroforestry programs supported by INIPA, the International Council
for Research in Forestry (ICRAF) and the International Development and Research Center of Canada
(IDRC), and the principles being tested are expected to be applicable in the humid tropics world-wide.
The studies presented here have explored three general areas: 1) the selection and production of
trees and shrubs valued for fruit and timber production or soil improvement; 2) the integration of
trees and annual crops in systems such as alley-cropping; and 3) the role of secondary forest fallows
in the improvement of soil properties. Because trees require months or years to become established
and productive, these experiments are long-term, and some are in the earliest phases.


Trees as Soil Improvers
In the Humid Tropics?
Pedro A. Sanchez, N. C. State University

It is commonly believed that trees are the best op-
tion for producing food and fiber on a sustained basis
in the humid tropics. Because tree plantations resem-
ble the natural ecosystem more closely than do an-
nual crops, tree management might be expected to re-
quire fewer inputs. The accumulation of large amounts
of biomass on acid, infertile soils, seemingly due to
rapid and efficient nutrient cycling, suggests that
tropical forests function in a fundamentally different
way than do annual crops and pastures. It is not known
if the same conditions exist in production-oriented tree
crops. The objective of the activity reported here was
1) to bring together reliable information on the effects
of deliberately planted tree crops on soil properties in
the humid tropics, and 2) to develop working
hypotheses for agroforestry research.
Available information on the effect of deliberately
planted tree crops on soil properties in the humid
tropics was compiled, examined, and, whenever possi-
ble, compared to alternative systems such as native
forests, annual crops, pastures or fallows. Only data
sets meeting a set of soil uniformity criteria were us-
ed in the analysis. A complete report has been publish-
ed. (See publications list.)

Working Hypotheses
The main conclusions or working hypotheses are:
1. Tree crops make the soils initially more vulnerable
to runoff and erosion than annual crops or pastures
because of their lower rate of canopy development dur-
ing the establishment phase.
2. The degree of changes in soil properties during
the tree establishment phase depends largely on land-
clearing methods.
3. Protection of the soil surface with well-managed
leguminous covers during the tree-establishment phase
can largely prevent deterioration of physical and
chemical properties.
4. When trees close their canopy, they begin to ex-
ert four major positive effects on soil properties: (a)
soil-surface protection with a double layer (canopy and
litter); (b) opening soil pores via root expansion and
decomposition; (c) capture of nutrients and storage in
biomass, and (d) recycling nutrients back to the soil.
5. Closed tree canopies provide virtually complete
protection against erosion unless the trees are

deciduous, and provided the litter layer is not remov-
ed or burned.
6. Closed tree canopies tend to improve soil struc-
ture and decrease topsoil bulk density, but this effect
varies substantially with tree species.
7. Closed tree canopies do not increase topsoil
organic matter contents. In most cases, soil organic
matter is maintained relative to pre-clearing levels.
When products such as rubber or oil palm are
harvested, soil organic matter decreases and then
reaches a new equilibrium level.
8. Some tree species tend to increase topsoil Ca and
Mg by mechanisms not clearly understood. The ef-
fect is marked with Gmelina arborea, which appears
to be a calcium accumulator in Nigeria and Brazil. Ex-
changeable K often decreases to very low levels and
may trigger deficiencies. Tree species differ in their abili-
ty to alter soil acidity.
9. Leaching losses occur mainly during the tree-
establishment phase. When well managed tree plan-
tatins develop a full canopy, leaching losses are as low
as in undisturbed forests. The nutrient-cycling
mechanisms of many perennial tree crops appear to
be very efficient. Sometimes their efficiency is enhanced
by fertilization.
10. Fertilization and other management practices are
likely to be needed for second rotations of timber crops,
as has been clearly demonstrated with perennial crops.
Expectations are likely to be erroneous that sustained
tropical forestry is possible in acid soils of the humid
tropics without using fertilizers.
11. Trees, therefore, generally maintain or improve
soil properties in the humid tropics after they have
established a closed canopy. Maintenance can be in-
itiated early with a leguminous cover. The main ad-
vantages of trees over annual crops or pastures seem
to be related to the longer period of time during which
they exert their influence on soil properties.


Alley-Cropping on Ultisols
Lawrence T. Szott, N. C. State University
Charles B. Davey, N. C. State Universtiy
Cheryl A. Palm, N. C. State University

In areas with increasing demographic pressure, tradi-
tional forms of shifting cultivation must be supplanted
by production systems that yield more food on the
available land. One technique, shown to be promis-
ing in Alfisols of West Africa, is the combination of
rows of leguminous trees with annual crops grown bet-
ween them. Prunings from the trees form a mulch that
may aid in weed control and provide nitrogen and
other nutrients, cycled from deep in the soil, to the
crops. The use of such organic additions may prolong
the productivity of the acid, infertile soils found in
much of the humid tropics.
This study was conducted at the Yurimaguas Ex-
periment Station. Its objectives were: 1) to assess the
suitability of various leguminous trees or shrubs in an
alley-cropping system, an assessment based on survival,
biomass production, ability to withstand repeated prun-
ings, and litter-decomposition characteristics; 2) to
determine the appropriate spacing between tree rows,
as it affects crop yield; 3) to study changes in soil
chemical properties and how they are affected by the
amount of prunings added, and 4) to measure the ef-
fects of pruning additions on crop yields and yield
Six leguminous species were chosen: Inga edulis,
Erytbrina sp., Cajanus cajan, and Cedrelinga catenaefor-
mis were obtained locally while Leucaena leucocephala
and L. diversifolia were obtained from the Nitrogen
Fixing Tree Association (NFTA) in Hawaii. Cedrelinga
was replaced in late January, 1985 by Desmodium
gyroides, direct-seeded.
Most species were raised from seed in the nursery
and after four to six months were transplanted (Oc-
tober 1984) to a field that had been slashed, burned
and planted with rice.
An experiment was established with variable alley
spacings, using a randomized, complete-block design
and four replications. Three rows of trees were planted
in 9.5 x 2 3 m plots; the middle tree row was periodical-
ly staggered to provide six different intervals between
adjacent tree rows. These intervals accommodate two
to seven rows of annual crops (Figure 1). All the trees
in each plot received one of three fertility treatments:
1) none, 2) two tons lime/ha or 3) two tons lime +
100 kg P/ha, which were applied once, on an

equivalent-area basis, only to the 1 m wide tree rows.
In March, 1985, some of the trees showed signs of
K deficiency. Therefore, 100 kg K/ha was applied on
an equivalent-area basis to the 1 m wide tree rows in
the lime + P treatment in both experiments. The alley
crops receiving K were Cajanus, Inga, and Erytbrina.
Food crops were grown between the tree rows,
without direct fertilization and at a constant spacing
within and between rows. The checks were two ad-
ditional treatments of crops grown without trees -
"sole" crops, with and without fertilization. Soil
chemical properties were monitored after each crop
Trees were pruned before each planting or soon
thereafter, and during crop growth as needed. Prun-
ing biomass was measured and subsamples taken to
measure dry matter and nutrients. Prunings from the
plot were divided into six equal parts; each part was
spread over one spacing interval. Therefore, the
amount of prunings per area varied with each spac-
Weed biomass was measured approximately two
weeks before every crop harvest. Crop yield was in-
itially recorded by row position and fertility level.
Subsequently, yield data were obtained by spacing in-
terval, row position and fertility treatment.
Crop Rows Trees

[l I I IF I I
2. m I4.OrI

B I I I I I E lI I I I
I 12.e m i lm 4.6 m' I I I

13.m3.5 I

75 cm 50cm
Figure 1. Plot design, variable spacing plots of alley-
cropping experiment. Capital letters identify spacing


The first rice crop was harvested in February 1984.
Most of it was severely infected with rice blast caused
by Pyricularia oryzae and was left in the field.
Pruning of Cajanus began in March 1984; of Inga
in August 1984; and of Erytbrina in April 1985. Cedrel-
inga was eliminated in January 1985 and replaced with
Desmodium gyroides. By April 1985, biomass produc-
tion and growth of the two Leucaena species were poor.
At this time, the average tree height per plot was
estimated, all plants were pruned 1 m above the soil,
pruning yields were measured, and the prunings were
placed around the trees.
L. leucocephala yielded 576, 667, and 859 kg of dry
prunings per hectare for the no-input, lime, and lime
+ P treatments, respectively; L. diversifolia prunings
averaged 481, 384, and 438 kg/ha for the same
treatments. While in some cases vertical growth was
good, reaching over three meters, the majority of the
plants appeared spindly, without much foliage. The
Leucaenas have subsequently been eliminated from the
By mid-August 1985, Cajanus had been pruned four
times; four crops (corn, cowpea, rice, and rice) were
harvested from these plots. Inga had been pruned four
times also; three crops were harvested cowpeaa, rice,

0 4 8 12 16 20
Months After Transplanting

Figure 2. Cumulative pruning dry matter yields of
alley crops during the first 18 months after transplan-
ting. Alley length: 3160 m/ha.

and rice). Erytbrina had been pruned once and one
rice crop was harvested.

Pruning Yields
Cumulative pruning yields of the alley-crop species,
averaged over fertility treatment and replications, are
shown in Figure 2. Yields, based on 3160 m of tree-
row length per hectare, are 8.3 tons of dry matter per
hectare for Inga and 3.1 t/ha for Cajanus. Erytbrina
has been pruned only once, producing slightly higher
biomass than Inga and Cajanus at six months of age.
There was no response to lime additions, but the
lime + P treatment, in comparison with the no-input
treatment, yielded approximately 8% more prunings
in Inga and 6% for Cajanus. Erytbrina appears to res-
pond to P. The lime Ca + P treatment yielded 625
kg or 38% more prunings than either the no-input
or the lime treatment. In addition, there appears to
be a positive response to increases in the clay content
of the topsoil in all alley-crop species.
The rate of biomass production by Inga appears fairly
constant over time and averages 8.7 t/ha/yr after the
first pruning. In contrast, Cajanus production averages
1.8 t/ha/yr and appears to decline with time. Much
of this decrease is due to plant senescence. By January
1985, only about 65% of the original Cajanus plants
were still alive, indicating the need for continual
replacement. Because of different establishment prac-
tices, some species were ready for a first pruning before
The decomposition characteristics of the prunings
differ by species. Erytbrina prunings decompose rapidly;
few last longer than one month. A significant propor-
tion of the Inga prunings, on the other hand, can still
be observed after three months. The decomposition
rate of Cajanus prunings is intermediate to those of
Erythrina and Inga.

Weed Control and Biomass
Weeds were controlled by herbicide use prior to
planting and hand weeding during the crop, as need-
ed. Herbicide application before planting was the on-
ly form of weed control used in the corn and cowpea
crops. Weeds in the first rice crop following cowpea
were controlled with preplant herbicides and two hand-
weedings; in the subsequent rice crop, all treatments
received preplant herbicides and one hand-weeding.
In addition, the Cajanus and control plots required a
second hand-weeding.
Due to the weed-control regimens used, and the dif-
ferent lengths of time plots have been in cultivation,


the most appropriate comparisons of weed-biomass
data are within alley-crop species.
A comparison of weed biomass by spacing subplot
for the most recent sampling (7/85) shows surprisingly
little relation between subplot and weed biomass (Table
1). It might be expected that weed levels would be
lowest at the closest spacing due to shading and a
greater mulch concentration, especially in the case of
Inga, which has a long-lasting mulch. The data,
however, do not appear to support this hypothesis.
There is a suggestion that weed biomass decreases
with closeness to the tree rows and that the reduc-
tion is likely due to shading. The effect appears to be
less pronounced in the Cajanus than in the Inga
treatments, presumably due to less biomass and less
shade production in the former.
For each alley-cropping species, there appear to be
significant differences in weed biomass among replica-
tions, but the pattern over time is variable and dif-
ficult to explain.
In general, weed biomass is usually higher in the
untilled, unfertilized control plots than in the alley-
cropping treatments. It is also interesting to note that
fertilization was related to decreased weed biomass in
all sole-crop checks.

As might be expected, the effectiveness of weed con-
trol by the prunings is less pronounced in Cajanus than
Inga due to the reduced quantity of Cajanus mulch
produced. One should note that similar levels of weed
biomass were measured for the Inga and Cajanus
treatments (7/1985), despite an additional hand
weeding in the Cajanus plots requiring approximately
60 man-hours of labor.
Soil Chemical Properties
In general, topsoil chemical properties degrade with
time after burning. Exchangeable cations and available
P decrease while acidity and exchangeable Al increase,

for all treatments with the exception of the fertilized
check. There are only minor differences between the
unfertilized check, Cajanus, and Inga treatments. The
Inga plots have slightly higher levels of Ca + Mg and
K, but they have also been used for one less crop than
Cajanus. Both Inga and Cajanus have slightly higher
levels of soil organic matter than the unfertilized check.
In an attempt to further define whether mulch ad-
ditions affect soil chemical properties, the soil was
sampled in March 1985 by spacing sub-plot (six com-
posited samples per plot) for the Cajanus and Inga
treatments. Data for the closest and farthest spacings
for each species are shown in Table 2. For Inga, the
closest spacings, which have received 1.75 times more
pruning dry matter than the widest spacings, have
slightly higher levels of exchangeable Ca, Mg, and K.
The closest spacing for Cajanus, on the other hand,
has virtually the same soil nutrient levels as the widest
spacing. The topsoil base status under Inga is higher
than under Cajanus, and it appears that, at least in the
case of Inga, there is some slight improvement in soil

Crop Yields
Crop yields have been adversely affected by insects,
disease and weather. Consequently, yields are low, bear
Table 1. Weed biomass as affected by alley crop species and
by spacing sub-plot, July 1985.
Tree Aliey crop species
spacing Cajanus Inga Erythrina

m g/m2
2.0 40+34 35+39 87+43
2.5 20 + 26 42 + 32 65+55
3.0 3724 47+30 48+21
3.5 43+42 40 26 58+23
4.0 39 + 27 51 + 33 57+34
4.5 32+15 36+27 89+61

Table 2. Effect of alley crop species and mulch application rate on topsoil chemical properties 18 months after planting
tree crops (3/85).
Alley crop Alley Mulch Exchangeable Al Avail.
species spacing added pH Al Ca Mg K ECEC Sat. P
m t/ha/18 mos. cmol/L % ppm
Cajanus 4.5 2.6 4.5 1.6 0.64 0.16 0.07 2.47 65 8
2.0 4.5 4.5 1.4 0.65 0.16 0.07 2.28 61 9

Inga 4.5 6.9 4.6 1.6 0.85 0.18 0.10 2.73 59 7
2.0 11.8 4.6 1.5 1.03 0.23 0.13 2.89 52 8


little relation to treatments, and are probably not in-
terpretable. In general, row position appears to have
an effect on yield, but it is uncertain whether these
differences are significant. The data suggest that, with
rice and cowpea, yields increase with distance from
the tree rows. It may be too early in the cropping se-
quence to identify meaningful patterns. Furthermore,
since soil texture seems to affect yields, covariance
techniques may be necessary to pick out trends in yield.

1. Of the six original leguminous trees or shrubs
assayed in an alley-cropping system, Leucaena
leucocepbala, L. diversifalia, and Cedrelinga catenaeformis
have been eliminated due to poor survival and low
biomass production. Cajanus cajan is also unsuitable
due to increased plant senescence and decreased pro-
ductivity at about one year of age.
2. Both Inga edulis and Erytbrina appear to have good
survival and coppicing ability. Inga biomass produc-
tion is high (8.3 t dry matter/ha/yr, based on 3160
m of tree-row length per hectare); Erytbrina produc-
tion must be assessed for a longer period. Cajanus, In-
ga and Erytbrina production appears to respond to P
3. Inga prunings resist decomposition, Erytbrina
prunings are readily decomposed, and Cajanus prun-
ings are intermediate in their rate of decomposition.
4. Weather, insects and disease severely reduced crop
yields and made them for the most part uninter-
5. A correct comparison of weed levels can only be
made among spacings within a given alley-crop species
in this study. There appears to be little relation bet-
ween weed biomass and spacing/mulching.
6. Soil chemical properties decline with time and
are similar in all alley-crop treatments and in the unfer-
tilized check. Soil chemical properties have improved
in the fertilized check. Comparisons of the highest and
lowest pruning-addition levels show increases in ex-
changeable cations at the highest mulching rates for
Inga and no difference between rates for Cajanus.
New work in this project will investigate the effect
of mulching, nutrient transfers between prunings and
crops, the effects of different types of prunings, and
competition between the trees and crops.

Peach Palm as a Soil Management
Option on Ultisols
Jorge Perez, INIPA
Charles B. Davey, N. C. State University
Robert E. McCollum, N. C. State University
Beto Pashanasi, INIPA
Jos6 R. Benites, N. C. State University

The production of peach palm, Guilielma gasipaes,
has several advantages as a management option for the
Amazon Basin. It is indigenous and adapted to acid
soils, does not require yearly tillage, has been known
to remain productive for 20 years, and has potential
as a source of fruit, heart of palm and lumber for par-
quet. Results from peach palm experiments establish-
ed at Yurimagaus in 1980 indicate that the crop can
produce fruit at the rate of 15 ton/ha/yr, beginning
with the fifth year. At present prices (U. S. $0.17/kg
fruit), gross proceeds from a peach palm plantation pro-
ducing 15 ton/ha/yr would be about S2600/ha/yr.
Although prices would vary with production and
market conditions, the economic potential of peach
palm appears to compare favorably with paddy rice
production on alluvial soils, the most economically at-
tractive option in Yurimaguas so far. Because of this
potential, several projects related to peach palm pro-
duction have been conducted at the Yurimaguas Ex-
periment Station.

Collection and Propagation
The objectives of this work were 1) to collect and
maintain a permanent collection of spineless-trunk and
spiny-trunk peach palm germplasm from Amazonia
nations in order to improve certain characteristics for
higher agronomic value, and 2) to establish a
phenological calendar for each accession.
Yurimaguas is a center of origin for spineless-trunk
peach palm. Over 120 spineless accessions have been
collected in the area, varying considerably in fruit
characteristics. In addition, 80 spiney-trunk accessions
in an international collection from Brazil, Colombia
and Ecuador have been planted in an area sufficiently
distant from the spineless collection to prevent
Peach palm accessions have been planted in two
agroforestry sequences. The spineless collection had
an upland rice-cowpea rotation, while the spiny one
was interplanted with cassava, which yielded 10-30
ton/ha. After the cassava harvest, a ground cover of
Desmodium ovalifolium, was established.


The range in nutritional composition of fruits col-
lected around Yurimaguas is shown in Table 1. It has
higher mean protein content than sweet potatoes with
comparable values. The range indicates the possibili-
ty of selecting fruits with very high protein or fat

Nutritional Requirements of Peach Palm
A fertilization experiment was established in peach
palms transplanted in August, 1982 in order to deter-
mine optimum levels of N, P, K, and Mg and the
response to lime and Zn in peach palm production.
Trees were set at a 3 x 3 m spacing on an Ultisol with
topsoil of pH 4.4, 0.1 cmol/L of Ca + Mg, 90% Al
saturation and 3.5 ppm available P. The main response
has been to N. This response was linear during the
second year but developed a clear optimum rate of
100 kg N/ha by the third year (Table 2). Trees without
N showed strong chlorosis. No response to P, Mg or
lime has been detected. Potassium responses became
evident the second year with a clear peak at 50 kg
K/ha/yr (Table 3). There was also a clear response
to 2 kg Zn/ha the third year.

Cover Crops in Plantations
This project's objective was to observe the effect of
different leguminous ground covers on peach palm
development and production. Peach palms were
planted in October, 1980 at 5 x 5 m spacing in a 0.7
ha plot. Four legume covers were planted in February,
1982, in strips without replications. The legumes were:
Pueraria phaseoloides (kudzu), Desmodium beterophyllum,
D. ovalidolium and Centrosema hybrid 438. Two of the
legumes D. beterophyllum and Centrosema had disap-
peared, apparently because of drought and shade,
despite the fact that sunlight passed through the peach
palm canopy. The other two legumes were well
adapted, and began invading adjacent areas. Each pro-
duced dry matter of about 800 kg/ha/yr.
Legume ground covers showed no significant dif-
ferences in their effect on tree growth. After 28
months, peach palm trunk diameter breast height
(DBH) ranged from 17.4 to 18.5 cm, and height ranged
from 5.0 to 6.1 m.
Three-year topsoil samples show sharp increases in
acidity and base depletion. No fertilizers or lime have
been added to this field to date.

Table 1. Nutritional composition of peach palm fruits collected
around Yurimaguas. Fresh-weight basis.










Table 2. Effect of applied N on the growth in height
of peach palm.
Applied N (kg/ha)
0 50 100 200
Tree age (years) Tree height (m)
One 0.6 1.0 1.2 1.2
Two 1.8 2.3 3.1 4.1
Three 2.6 5.5 7.1 7.4

Table 3. Effect of applied K on the growth in height
of peach palm.
K Applied kg/ha
0 50 100 200
Tree age (years) Tree height (m)
One 0.8 1.0 1.0 0.8
Two 2.2 3.4 3.4 3.2
Three 5.8 7.0 6.8 4.6


Gmelina arborea: Intercropping, Coppic-
ing and Nutritional Requirements
Jorge Perez, INIPA
Charles B. Davey, N. C. State University
Robert E. McCollum, N. C. State University

Gmelina arborea is a promising, fast-growing
timber species for the humid tropics. The first stand
in Yurimaguas was planted in April 1981, and several
experiments were conducted in order to evaluate
Gmelina's potential in agroforestry, including systems
of intercropping. The objectives of the experiments
reported here were 1) to observe the effects of ground
covers on growth and coppicing behavior of Gmelina
arborea, and 2) to determine this tree's response to ap-
plications of N, P, K, Mg, lime and Zn.

Effects of Understories
Gmelina planted at a 3 x 3 m spacing grew quickly
and reached a height of over 7 m with a 10 cm
diameter at breast height (DBH) in three and a half
years (Table 1). Gmelina did not grow significantly less
when pineapple, plantain, or a rotation of three an-

Table 1. Effects of understories on the growth of
Gmelina arborea 3.5 years after planting.
at Breast Tree
Understory Height Height
cm m
None 10.0 7.4
Pineapple 11.8 7.3
Corn-rice-soybeans 11.3 6.8
Plantain 9.1 7.0
Cassava 8.5 5.3
Pueraria phaseoloides 9.7 5.9
Desmodium ovalifolium 8.9 5.7
Desmodium heterophyl//um 7.8 5.6
Brachiaria decumbens 7.9 5.8
Brachiaria humidicola 6.7 5.3

nual crops were raised in its understory. Tree growth
was retarded by cassava and the legume species, and
stunted by the grasses. The effect of the Bracbiaria was
so severe that it may be allelopathic.

Soil Properties
This experiment was installed in an area which was
limed to pH 5.5. With time, soil acidity increased and
available P declined. Reports from other regions sug-
gest that Gmelina is a topsoil-calcium accumulator.
Although soil tests do not provide evidence of this hap-
pening up to this time, analysis of leaves reveals a very
high Ca content, about triple that of pasture grasses
and leguminous plants grown at Yurimaguas.

Coppicing Behavior
When the trees were five years old, they were cut
and allowed to coppice. A cutting height variable was
introduced but did not show any influence on the
growth of sprouts. The remarkable aspect is the rate
of regrowth-3 m in 60 days. An upland rice crop
planted in the area grew so poorly that no yield was
produced. Apparently the stump regrowth was too
competitive for water and perhaps nutrients.

Nutritional Requirements
The Gmelina fertilization experiment was planted
in November, 1982 on an Ultisol at pH 4.3 and 75%
Al saturation in the top 15 cm. Trees were spaced at
3 x 3 m. During the first year, trees attained an average
height of 4 m and a DBH of 3.6 cm. Eighteen months
later, average height reached 9.7 m (a growth rate of
32 cm per month) and DBH reached 11.2 cm. By the
third year, the canopy had closed, impeding the growth
of weeds. Damage by leaf-cutting ants continues, but
is not critical. Gmelina arborea is susceptible to dry
periods greater than 60 days. Such droughts cause a
general chlorosis and strong defoliation.
No significant responses to N, P, K, Mg, Zn and
lime have been observed. Variability in growth is
primarily related to areas that are poorly drained.


Contrasting Effect of Pinus caribaea and
Gmelina arborea on Soil Properties
Pedro A. Sanchez, N. C. State University
Charles E. Russell, Institute of Ecology,
University of Georgia

Reliable data are scarce on soil-fertility dynamics
under fast-growing timber species in the humid tropics.
This project was conducted to complement thesis
research by Charles E. Russell of the University of
Georgia, who sought additional soils data to quantify
the influence of Gmelina arborea and Pinus caribaea
plantations on soil properties of Typic Paleudults at
Jari Florestal Agropecuiria in the state of Para, Brazil.
The discussion that follows draws on his thesis and
laboratory analysis conducted at N.C. State University.
Gmelina arborea significantly increased soil pH and

pH (soil surface)

exchangeable Ca in the top 1 m of this soil, while Pinus
caribaea decreased these parameters as well as available
P, exchangeable K, Mg and total N (Figure 1). A syn-
thesis of the changes in total nutrient stocks during
the course of plantation establishment and growth is
presented in Figure 2. Total nutrient stock is defined
as the sum of all the nutrients in the plant biomass
(aboveground, litter, detritus, roots) plus total N,
available P (by the Mehlich I method), and ex-
changeable K, Ca, and Mg in the top meter of the
soil. This estimate, therefore, ignores the total P, Ca,
Mg, and K contents of the soil.
Total plant biomass decreased to about 40 to 60%
that of the virgin rainforest at the end of the first rota-
tion of Gmelina or Pinus. Most of the losses are quan-
titatively accounted for by the newly planted trees and
the dry matter extracted by harvest.

Exch. Ca (tonslha/100 cm)

I Ine

Exch. K (tonslhal100 cm)



8.5 9.5

SExch. Mg (tons/ha/100 cm)

" Avail. P (kg/ha/100 cm)


0.5 8.5 9.5

Years after Clearing Virgin Rainforest
Figure 1. Effects of Pinus caribaea and Gmelina arborea plantings on surface soil pH and nutrient content of
the top 1 m of a sandy Ultisol in Jari, Brazil.




| I


120- P3

0-1 -

I `L I -

.d .o 0 d


a: 0 .

Figure 2. Changes in nutrient stock in fast-growing
tree plantations on a sandy Ultisol in Jari, Brazil.
Amount equivalent to 100 indicated in t/ha. H =
harvest loss; L = leaching loss; RF = rainforest;
P-0.5 and P-9.5 = Pinus caribaea plantations of 0.5
and 9.5 years old, respectively; G-8.5 = Gmelina ar-
borea plantation of 8.5 years old; PII = 1.5 year Pinus
caribaea second rotation after harvesting eight-year

Nutrient Stocks
The plantations, of all ages, contained approximately
60% of the total N stock of the rainforest. Most of
the decrease in N occurred at clearing. Because none
of the plantation trees were legumes and no legume
cover crops were used, no N buildup occurred. The

ecosystem, therefore, lost 40% of its total N, and then
reached a new equilibrium.
A remarkable conservation of P is shown in Figure
2. Nutrient stocks ranged from 76 to 116% of the
rainforest values. The decrease in P at the second rota-
tion is largely accounted for by the P removed in the
first rotation harvest. These values reflect only the frac-
tion of total P extracted by the Mehlich procedure.
Calculations from total elemental analysis of soils of
the Amazon by Marbut and Manifold in 1926 indicate
an average total P content of the top meter on the
order of 60 times the total P stock indicated in Figure
Significant losses of K occur when rainforests are
converted into tree plantations. Potassium stocks after
clearing decreased to about 32 % those of the native
forest. Most of the losses are accounted for by the
removal of harvested trees and the rapid leaching losses
recorded during this period. After clearing, there were
slight increases to about 40% of the rainforest value.
The overall stocks and losses of exchangeable K
presented in Figure 2 (1.06 t/ha), however, are small
considering the total K content of these soils, estimated
at 73 t/ha of K (Marbut and Manifold, 1926). It is
not surprising that research on perennial crops such
as rubber and oil palm on Ultisols shows rapid deple-
tion of K and the need to fertilize the trees with this
The calcium nutrient stock decreased to about 56%
of the rainforest value upon planting the first Pinus
rotation. The losses were again accounted for by
harvest and slight leaching. This level remained relative-
ly stable with Pinus but increased to above pre-clearing
levels with Gmelina (Figure 2). The second rotation
started at a lower level, but much of the loss was related
to the amount removed by Gmelina harvest. Losses
were slight compared to the total Ca in the top meter,
about 13.6 t/ha.
The magnesium nutrient stocks decrease with age
of Pinus plantations, but mature Gmelina plantations
maintained a steady level of about 75% of the level
in the rainforest. The 25% loss appears to be related
to the harvest of the rainforest. The overall losses are
small relative to the total Mg content of these soils,
about 14.4 t/ha of Mg at the 100 cm depth.


Improved Fallows
Lawrence T. Szott, N. C. State University
Charles B. Davey, N. C. State University
Cheryl A. Palm, N. C. State University
Pedro A. Sanchez, N. C. State University
Much of the land available to shifting cultivators
remains idle each year, due to the long fallow periods
required for secondary forests to restore the produc-
tivity of abandoned agricultural fields. The purpose
of this project, which was conducted at the Yurimaguas
Experiment Station, was to determine whether pro-
ductivity in such fields might be regenerated more
rapidly with the use of selected, high-biomass, nitrogen-
fixing fallow species, and to measure the effects of these
species on soil physical and chemical properties, weed
suppression, and, subsequently, crop yield.
A one-hectare, 15-to-20-year-old secondary forest
on a loamy topsoil and clay loam subsoil was cut, burn-
ed, and planted with upland rice using traditional
methods in August 1983. The rice was harvested in
January 1984, and the following treatments were in-
stalled in 100 m2 plots, with four replications, in a
randomized complete block design: Natural purma
(secondary vegetation); Cajanus cajan; Inga edulis;
Stylosanthes guianensis 136; Centrosema macrocarpum;
Desmodium ovalifolium 350; Pueraria pbaseoloides (kud-
zu); high-input cropping check (fertilization and
mechanization); low-input cropping check (without fer-
tilization or mechanization).
Changes in soil and vegetation properties during the
first 16 months after fallow establishment are reported.
The experiment will be continued for at least 32 ad-
ditional months, after which the plots will be prepared
for low-input crop rotations. Crop harvest data will
be used to assess fallow performance.

Fallows may restore the productivity of abandoned
agricultural land by one or more of the following: 1)
enrichment of the topsoil-vegetation system by retain-
ing nutrients added in rainfall, dust, or mineral weather-
ing, or by direct contributions from N2 fixation or the
recycling of nutrients from the subsoil; 2) improve-
ment in soil physical properties; 3) control of weeds.

Aboveground Living Biomass
The importance of possibility (1) can be ascertain-
ed through the construction of a nutrient budget for
the topsoil-vegetation system. While complete soil and
plant-tissue nutrient analyses are presently lacking, it

is worthwhile to consider biomass accumulation, since
nutrient immobilization is a function of both the quan-
tity of biomass and its nutrient concentration. Con-
sideration of the planted-fallow biomass only shows
that there is little difference among treatments after
16 months. On average, living aboveground biomass
approaches 7.5 t/ha; kudzu accumulation is about 2.5
t/ha lower, and that of Desmodium 2 t/ha greater
(Figure 1). There are, however, differences in the rate
of biomass accumulation. Stylosanthes and Cajanus pro-
duction, for example, is concentrated in the first eight
months; that of Centrosema and Desmodium appear
linear with time; and a large part of the Inga and kudzu
accumulation occurs between eight and 16 months.
Measurements of total living aboveground biomass
(planted fallow plus other vegetation), on the other
hand, show treatment-related differences (Figure 2).
At 16 months, Desmodium has outperformed the
natural purma, wile the Inga and Cajanus treatments
are about equal to it. The biomass of the "spreading"
type fallows (kudzu, Centrosema, and Stylosanthes) is
much lower. The difference between the two groups
is due to the presence of non-planted fallow vegeta-
tion, primarily trees, bushes, and lianas. This vegeta-
tion is naturally excluded from the "spreading" fallows,
but has readily invaded the tree or bush fallows. It
is noteworthy that at 16 months after establishment,

10 A Desmodium
0 Inga
A Centrosema
8 Kudzu
0 Pigeonpea

0 4 8 12 16
Months After Establishment
Figure 1. Living aboveground biomass of planted


weed (grass and broad-leaf herbaceous plants) popula-
tions are low in all treatments and are lower in the
planted fallows than in the natural fallows.

Root Biomass
The pattern of root-biomass accumulation at 16
months parallels that of aboveground biomass (Table
1). No planted fallow has a greater root biomass than
the natural purma, although the Inga and Cajanus

20.0 Desmodium
0 Calanus
0 Inga
A Centrosema
1] Check

10.0 --

s.0 --

*significantly different from the check
o 'I 1 '
0 4 8 12 1
Fallow Age (months)

jre 2. Total living aboveground biomass ac-
Amulation in fallow treatments.

fallows are similar to it. In general, the "spreading"
type fallows have much less root biomass. Kudzu root
biomass, for example, is only about 40% that of the
natural purma. The large root biomass observed in the
Stylosantbes treatment is due to the presence of a large
root ( >10 mm in diameter) remaining from the
previous forest. If this root is discounted, root biomass
for Stylosantbes is similar to that of the other
"spreading" fallows.
The quantity and distribution of fine roots, most ac-
tive in water and nutrient uptake, should also taken
into account. Moreover, their presence in significant
quantity in the subsoil may indicate nutrient "pump-
ing," one means by which the topsoil-vegetation system
may be enriched.
In general, fine root ( < 3 mm in diameter) biomass
comprises around 85% of the total root biomass,
although this proportion is only 65% in the natural
purma. Again, fine root biomass patterns parallel
aboveground biomass--treatments with great
aboveground biomass also have greater fine-root
biomass. The majority of fine roots are found in the
upper 15 cm of the soil for all species (Figure 3). Signifi-
cant quantities (- 700 kg/ha) of fine roots, however,
are also found in the subsoil in the Inga, Cajanus, and
natural purma treatments. The Centrosema and
Desmodium treatments contain about 500 kg of fine
roots per hectare, and the Stylosantbes and kudzu
fallows approximately half that much. Hence, the
possibility for recycling nutrients from great soil depths
appears limited for the Stylosantbes and kudzu fallows.

Vegetation Structure
The speed of leaf-canopy development has a number
of important consequences. The development of the
photosynthetic machinery affects vegetative growth
rates and concomitantly water and nutrient uptake,
while canopy formation reduces rainfall impact and

Table 1.Total Root Biomass, by Treatment and Depth.
Treatments kg/na
Depth Centro Stylo Inga Cajanus Kudzu Desmo Check
0-5 1074ab 585 b 748 b 1268ab 949ab 998ab 2014a
5-15 576 b 1776a 1019ab 478 b 262 b 621 b 891ab
15-30 256 768 499 457 134 399 297 ns
30-50 234 383 543 515 110 170 563 ns
Treatments significantly different (0.05) from the check at given soil depths are shown by different letters.




5 0


Fine Root Biomass (kg/ha) 24 hours after fairly heavy rainfalls. Depending as it
does on the amount and intensity of rainfall and the
100 200 400 600 800 1000 initial moisture status of the soil, a comparison of field
I 0 capacity over time may not have the validity of com-
parisons among treatments at a given time. For all
fallows, field capacity at 16 months is greater than or
equal to that measured at field abandonment.
However, there are few treatment-related differences
in the values at a given time. At 16 months, field
Capacity for Desmodium and Inga is lower, and higher
,2 for Centrosema, than that of the purma. Few differences
among treatments are observed in infiltration rates
measured at 16 months. Differences, if present, may
be obscured by variability in the data. All planted
fallows, with the exception of Stylosanthes, have in-
Q filtration rates that equal or exceed that of the natural

Figure 3. Fine root (less than or equal to 3 mm diam.)
with soil depth. Treatments differing significantly
(.05) in fine root biomass, at given soil depths, are
shown. The average biomass is shown when there
are no significant differences when comparisons are
made across treatments.

the potential for erosion, as well as moderating the
soil microclimate. Indicators of foliage development
are the leaf-area index (LAI) and the percentage of
ground cover.
In general, LAI correlates well with increasing
biomass, although Centrosema and the natural purma
have lower LAIs at 16 than at eight months, and ranges
between four and eight at 16 months. It is also notable
that all fallows develop a fairly high LAI (LAI = 5-6)
and an almost complete ground cover within four to
eight months after establishment for all fallows.

Soil Physical Properties
Soil physical properties are affected by vegetation
growth and development and associated changes in the
physical environment. Changes in soil physical pro-
perties are shown in Table 2. Bulk density, while in-
creasing in the first eight months after field abandon-
ment, appears to decrease to pre-cultivation levels by
16 months. Generally, the treatments with a high fine-
root biomass (purma, Cajanus, Inga) have lower bulk
densities than those with little fine-root production
(kudzu). These differences are not statistically signifi-
cant, however.
Field capacity (see Table 2) was measured in the field

Soil Chemical Properties
Soil chemical data for the first eight months after
establishment indicate a reduction with time in ex-
changeable Ca + Mg and available P levels in the top-
soil. Topsoil K, on the other hand, increases during
the first four months, probably due to K release from
the decomposition of 2 t/ha of rice straw. It soon
declines to levels similar to those measured at fallow
establishment. The reduction in topsoil nutrients can
probably be attributed to plant uptake and immobiliza-
tion in the biomass. The lack of significant differences
among treatments in soil chemical properties can pro-
bably be attributed to all treatments having similar
biomass at eight months, and the relatively short time
span during which measurements were recorded.
There seems to be a direct relationship between top-
soil organic matter and available P contents. Both
parameters increase after burning, drop precipitously
during the first eight months of fallow, and increase

Table 2. Representative values for topsoil bulk density, field
capacity (0-15 cm) and infiltration rate at various times follow-
ing fallow establishment.
Months After Establishment
0 8 18
Bulk Density (g/cm3)
0 7.5 cm 1.16 1.19 1.11
7.5-15 cm 1.32 1.33 1.24
Field Capacity (% H20) 26 28 29
Infiltration Rate
at 3 hrs (cm/hr) 19


Weed Control
As noted previously, all planted fallows are successful
at controlling grasses and broad-leaf weeds (Figure 4).
The best and quickest weed control is afforded by the
"spreading" type of fallow. In effect, these fallows
screen out almost all other types of vegetation. The
bush- and tree-type fallows, while allowing a somewhat
greater invasion of weeds, also permit the establish-
ment of other bushes and trees, resulting in greater
total biomass in those treatments. This may be advan-
tageous in that more biomass may result in greater
nutrient immobilization. Moreover, a mixture of
vegetation types may exploit the soil more complete-
ly. Furthermore, as observed in the fine-root biomass
data, treatments with significant quantities of tree, bush
and liana biomass not only tend to have more fine
roots, but also the roots are encountered at greater
soil depths, thus raising the possibility of the recycl-
ing of nutrients from the subsoil.

Comparisons With Continuous Cultivation
Grain yields for the mechanically incorporated fer-
tilizer treatment are about double those from the unfer-
tilized plots 10 t/ha vs. 6.3 t/ha (Table 3).
However, it is significant that the latter treatment con-
tinues to yield reasonable quantities of grain two years
after clearing. Similar phenomena have been observ-
ed elsewhere at the Yurimaguas Experiment Station
and suggest that factors other than the decline in soil
fertility weed control, for example are critical
to a farmer's decision to abandon his land.
In comparison with the fallows, soil chemical pro-
perties are more favorable to crop production in the

Table 3. Grain yield from continuously cultivated plots included
in the managed fallow experiment.
Crop Harvested Treatment Grain Yield
Rice 01-14-84 None 2.82 + .38
Corn 06-04-84 Fertilized 0.93 + .32
Unfertilized 0.44 + .11
Cowpea 08-29-84 Fertilized 0.86 + 0.21
Unfertilized 0.48 + 0.14
Rice 12-30-84 Fertilized 2.3 + 1.0
Unfertilized 1.0 + 0.4
Rice 05-14-85 Fertilized 1.91 + 0.45
Unfertilized 0.82 + 0.19
Cowpea 07-30-85 Fertilized 1.26 + 0.58
Unfertilized 0.76 + 0.36
Total Grain Production Fertilized 10.08
Unfertilized 6.32

0 4 8 12 16
Fallow Age (months)
*significantly (.05) different from check treatment
Figure 4. Changes in weed biomass with time in
fallow treatments.

continuous-cultivation treatments. This is expected in
the fertilized treatment due to periodic nutrient addi-
tions. In the unfertilized, continuously cultivated treat-
ment, the level of nutrient removal via harvest, which
may be likened to long-term nutrient immobilization
in the fallows, is likely to be much less. For example,
only the nutrients contained in 6.3 t grain/ha have
been removed from the cultivated plots vs. the nutrients
contained in 8 to 17 t biomass/ha in the fallows.
Furthermore, decomposition of soil organic matter and
crop residues releases nutrients to the soil. With time,
of course, soil nutrient levels in the cultivated, unfer-
tilized treatment should approach or decrease below
that of the fallows due to continued nutrient removal
during harvests, a reduction in the quantity of nutrients
recycled as plant production declines, and leaching

Comparisons of the rates at which natural and im-
proved fallows restore productivity can be based on
soil physical properties, the quantity of nutrient stocks
in both soil and biomass, the weed population, and
crop yields after the fallow. Considering these factors,


the following conclusions can be drawn:
1. Physical properties improve with time under all
fallows, and there are few treatment-related differences
in infiltration rate, field capacity or topsoil bulk den-
sity. Field capacity and topsoil bulk density improve
with time, the bulk-density values at 16 months ap-
proaching those following clearing.
2. Available nutrient levels in the topsoil decrease
with time in all treatments, except continuous cultiva-
tion, probably due to immobilization in the biomass.
Biomass nutrient stocks are determined by the quan-
tity of biomass present and the concentrations of
nutrients in the tissues.
3. After 16 months of growth, total biomass ac-
cumulation is highest in the bush or tree fallows (ap-
proximately 14 t/ha) and lowest in the spreading types
(approximately 7-10 t/ha). The natural purma biomass
is 14.1 t/ha. High biomass in the bush or tree fallows
is due primarily to the invasion of trees, since the
planted-fallow biomass is similar in the majority of
treatments at 16 months, about 7 t/ha. These dif-
ferences in total biomass are accentuated with time.
4. Planted fallow biomass accumulation was greatest
in Desmodium (9.7 t/ha) and least in Cajanus (5.1 t/ha).
The difference between total and planted fallow
biomass is due to the presence of other vegetation,
primarily trees.
5. Foliage development and the establishment of an
almost complete ground cover occur within four to
eight months in all treatments.
6. Weed control is better than the natural purma
except for Cajanus cajan in all planted fallows. Con-
trol is quickest and most effective with the "spreading"
fallow types.
7. Based on biomass production, fine-root produc-
tion and distribution, weed control, and changes in
soil properties, Desmodium may serve as a good short-
term (16-month) fallow. Desmodium, Centrosema, or
Stylosantbes may be suitable for fallows of eight months.

Forest and Soil Regeneration
Lawrence T. Szott, N.C. State University
Charles B. Davey, N.C. State University
Jorge Perez, INIPA
Cheryl A. Palm, N.C. State University

The period of vegetative regrowth following the
abandonment of agricultural fields has often been
credited with the restoration of site productivity. Sur-
prisingly, there have been few studies of soil and vegeta-
tion dynamics in shifting cultivation fallows, even
though such studies are necessary for an understan-
ding of how the most common agricultural produc-
tion system in the humid tropics functions, and may
point the way to potential improvements in the system.
Therefore, two complementary projects, addressing dif-
ferent aspects of old field secondary succession, were
undertaken. One compared secondary successional sites
of different ages but similar soils (Udults), in order to
determine how soil properties and vegetation struc-
ture and composition change over fairly long periods
of time. The other project is a long-term study, at a
single site, of the effects of different levels of soil fer-
tility on secondary succession in an abandoned
agricultural field.
The two studies are complementary in the sense that
they address different aspects of old field secondary
succession: how soil properties and vegetation change
over fairly long periods of time, and how soil fertility
affects these processes. They physically complement
each other in that the non-fertilized control plots in
the fertility study have been used to provide data points
in the study of different-aged purmas.

Soil and Vegetation Dynamics
In Shifting-Cultivation Purmas
Three purmas within approximately 1 km of one
another were identified. According to their owners,
all were previously farmed in the traditional manner
and had suffered little disturbance since abandonment.
The purmas were approximately three, seven and 11
years old. A fourth purma, nearby, used for the un-
disturbed check plots of the fertility study, was included
to provide data for purmas of age zero (field abandon-
ment) to 17 months. Soil texture in all the purmas
appeared similar, and are classified as sandy loams. Fur-
ther analysis revealed that the zero-year purma had
a lower clay content, with a classification of loamy


0 1 I I I 1 1I I I I I I
0 2 4 6 8 10 12
Purma Age (years)
Figure 1. Changes in aboveground biomass with timi
by vegetation type.

Table 1. Changes in purma vegetation structure with time after
Average Leaf
Time After Dominant Area
Abandonment Height, m Index Cover, %
0 1.35 + .28 1.4 + .7 58 + 17
4 mo. 2.28 + .29 3.6 + 1.1 93 + 12
8 mo. 3.56 + .47 6.3 + 1.5 98 + 5
17 mo. 4.04 + .51 4.8 + 1.3 98 + 5
3 years 5.99 + 1.52 not measured
7 years 12.89 + 1.39 not measured
11 years 14.68 + 2.24 not measured

The older purmas were sampled during a two-month
period. Soil properties measured included infiltration
rate, bulk density, field capacity, pore-size distribution,
organic-matter percentage, pH, exchangeable Al, and
nutrient contents (exchangeable Ca + Mg, K, and
available P). Biomass estimates of above-ground vegeta-
tion, litter, and roots to a 50-cm depth were also ob-
tained. Subsamples of each vegetation component were
taken for nutrient analysis and included tissue samples
of all trees greater than 2.5 cm DBH (diameter 1.4
m above the ground), by species. Heights and diameters
of all trees greater than 2.5 cm DBH were also

Changes in Biomass
Changes in living above-ground biomass are shown
in Figure 1. Biomass accumulation shows a typical
growth curve with a decline after seven years. The
average rate of biomass accumulation, 6 t/ha/yr, is
a bit low for humid tropical forests, which average
around 10 t/ha/yr. The seven-year-old purma,
however, accumulated biomass at a rate of 8.3 t/ha/yr.
In general, grasses dominate during the first year of
succession and decrease thereafter, presumably due to
shading by trees and other vegetation. The biomass
of lianas appears to increase in the early years of suc-
cession, maintaining itself at around 2 t/ha thereafter.
Tree biomass, while low during the first year, increases
rapidly thereafter and composes the majority of the
total biomass after three years. Recruitment of trees
into the population, as represented by the understory
(trees < 2.5 cm DBH) biomass, remains fairly con-
stant with time (about 1.5 t/ha).
Root biomass, down to a 50-cm depth, is shown
in Figure 2. Discounting roots having diameters greater
than 10 mm, root biomass appears to increase with
time. The large quantity of roots and the number of
large roots found at abandonment likely represent rem-
nants of the previous forest. It is generally cited that
root biomass is usually about 20% that of the above-
ground living biomass. Clearly that is not the case here,
as root biomass declines from 39% at 17 months of
age, to 21%, 10% and 9% at three, seven and 11 years
of age, respectively. Root turnover, which may be rapid
and hence contribute greatly to production estimates,
was not measured.
The standing-crop biomass of fine roots-those most
important in the uptake of water and nutrients-is also
shown in Figure 2. In general, fine root biomass in-
creases with time. This pattern is maintained to a soil
depth of 30 cm; at greater depths, all purmas, regardless


of age, have similar fine root biomass. In general, fine
root biomass declines with depth. The majority of the
fine roots are in the upper 15 cm of the soil, while
very few are found at depths greater than 30 cm.

Changes in Vegetation Structure
Rapid changes in vegetation structure occur during
the first 17 months (Table 1). Foliage development,
as measured by the leaf-area index (LAI) and percen-
tage of ground cover, occurs quickly. Within eight
months there is almost complete ground cover and a
fairly high LAI. The decrease in LAI between eight
and 17 months is probably due to the senescence of
grasses, herbs, and many of the dominant, short-lived
trees. At the same time, early vertical growth of trees
is extremely rapid (5.33 im/yr at eight months), even-
tually decreasing to an average growth rate of 1.33
m/yr at 11 years.
These changes have a number of important conse-
quences. The development of a multi-layered canopy
will tend to moderate soil temperature and humidity
and buffer any changes in these environmental
variables. Microbial activities and the decomposition
of organic matter may also be affected. The develop-
ment of a high photosynthetic potential also allows
rapid growth to occur and, with it, a high demand
for water and nutrients. Nutrients arriving at the roots,
via water uptake and diffusion along concentration gra-
dients, and their uptake in the biomass, are a conser-
vation mechanism by which losses in leaching, runoff,
or erosion can be reduced.

The diversity of the purmas, i.e., the number of
genera of trees greater than 2.5 cm DBH present, ap-
pears to decrease with time. The three-, seven- and
11-year-old purmas contained 60, 56, and 51 genera
per sampling unit, respectively. Very few genera pre-



A 0
0 1.4 yr
0 3yr
* 7yr
A 11 yr

Purma Root Biomass
Age (yr) All Roots
0 5,070
1.4 3,350
3 3,400
7 10,310
11 5,790

(kglha) to 50 cm soil depth
Fine Roots (< 3mm diameter]

Figure 2. Distribution of fine-root biomass in purmas
of different age.

sent in the three-year-old purina, however, are also
found in the seven- or 11-year purmas, indicating a
very rapid species turnover or nonhomogenous plant
population distributions. The only genera in common
to the three-, seven-, and 11-year-old purmas were
Cecropia (cetico), Inga (shimbillo) and Pollalestra
(yanavara), and these accounted for 60%, 62.5% and
39.2% of the individuals in the three-, seven-, and
11-year-old purmas, respectively. Apparently, although

Table 2. Effects of purma age on topsoil field capacity, infiltration rates and bulk density (g/cm3 using Uhland
Field Capacity Infiltration Rate Soil Depth (cm)
Purma Age 0-15cm at 3 hrs. 0-7.5 7.5-15

% cm/hr g/cm3
0 14.0 + 2.2 -1.36 + .07 1.53 + .03
17 mo. 14.9 + 1.91 53.9 + 8.41 -
3 yrs 16.0 + 0.8 22.2 + 13.4 1.28 + /10 1.48 + .11
7 yrs 19.1 + 2.7 29.9 + 5.8 1.18 + .05 1.35 + .10
11 yrs 16.0 + 1.3 24.8 21.9 1.21 + .13 1.42 + .13
1 Measured at 15 months.


the number of genera present may decrease with time,
dominance also decreases.

Soil Physical Properties
Bulk density was measured using two different
methods, Uhland cores and the excavation of a known
volume of soil. Results obtained using Uhland cores
suggest that topsoil bulk density decreases with time,
the greatest change occurring in the first year or two.
The bulk density in the seven-year old purma is a bit
low and may be due to the lower sand content in this
soil. The excavation method shows similar results. Bulk
density at the 0-7.5 and 7.5-15 cm soil depths generally
decreases with time, presumably due to organic-matter
additions and root growth. At greater depths, however,
soil bulk density is higher and does not appear to
change significantly with time.
Topsoil field capacity, measured in the field approx-
imately 24 hours after a heavy rainfall, appears to in-
crease slightly with time (Table 2). Once again, the
higher value for the seven-year-old purma may be due
to the lower sand content and, hence, macropores in
this soil. Infiltration rate was measured in the field using
a double-ring infiltrometer. There are probably no
significant differences in infiltration rates for the three-,
seven- or 11-year-old purmas (see Table 2). The rate
measured at 17 months is significantly higher and is
probably due to the high sand and low clay content
of this soil.

Soil Chemical Properties
In general, topsoil acidity, as measured by pH value,
and exchangeable Al, both increase with time (Table
3). There is a suggestion that exchangeable nutrient
cation levels in the topsoil decrease. A fairly large
decrease in cation levels appears in the first year after
abandonment. This is followed by a period (six to seven
years) of little change in the exchangeable cation con-
tent of the topsoil before a subsequent decrease is
observed in the 11-year old purma. A similar pattern
is observed at the 15-30 cm soil depth, while little
change with time occurs at depths greater than 30 cm.
A number of factors make the interpretation of this
pattern rather difficult. It is unclear, for example,
whether the changes observed in the purmas of dif-
ferent ages are, in fact, due only to the growth of the
vegetation and passage of time, or whether there are
significant differences in such things as site history and
clay mineralogy. Additionally, there is a fair amount
of variability in the data, especially in the results for
the seven-year old purma, which obscures trends and
makes interpretation difficult.
The decline in topsoil nutrients during the first year
can probably be attributed to nutrient immobilization
in living biomass (6.3 t/ha above ground) and leaching
losses. After this decline, topsoil nutrients appear to
change little for six years, while living above-ground
biomass increases nine-fold during the same period.
It is likely that the nutrients sustaining this growth are

Table 3. Changes in soil chemical properties with time after abandonment.
Soil Months After Abandonment Years After Abandonment
Parameter Depth 0 4 10 3 7 11

pH 0-15 5.1 + .2 5.0 + .3 4.8 + .1 4.8 + .2 4.6 + .1 4.5 + .2
15-30 4.4 + .1 4.5 + .1 4.6 + .04 4.7 + .1 4.4 + .1 4.5 + .2
30-45 4.4 + .1 4.6 .04 4.8 + .1 4.5 + .1 4.8 + .1
Exch Al 0-15 0.2 + .3 0.9 + .3 0.8 + .4 0.6 + .1 1.8 .4 1.1 + .3
(cmol Li) 15-30 1.3 + .4 1.6 + .4 2.0 + .8 1.6 + .4 3.4 + .6 2.2 + .7
30-45 1.7 .3 2.5 .4 2.3 + .4 3.3 + .6 2.8 + .6
Exch Ca* 0-15 1.3 + .9 0.8 + .3 1.0 + .2 1.0 + .2 1.2 + 15 0.5 + .2
(cmol Li) 15-30 0.5 + .1 0.4 + .1 0.5 + .1 0.5 + .1 0.5 + .1 0.4 + .1
30-45 0.3 + .1 0.4 + .1 0.3 + .1 0.4 + .1 0.3 + .1
Exch K 0-15 .10 + .07 .07 + .03 .09 .02 0.9 .02 .11 01 .08 + .01
(cmol Li) 15-30 .08 + .05 .07 + .03 .08 + .03 .06 + .01 .07 + .02 .06 .02
30-45 .06 + .03 .075 .02 .06 + .01 .07 .02 .05 + .01
Avail P 0-15 14 + 4 8 + 2 7 + 1 5 + 1 5 + 1 4 + 1
(ppm) 15-30 7 2 4 + 2 4 + 1 3 + 0 2 1 2 + 1
30-45 4 + 1- 3 + 1 3 + 1 2 + 1 2 + 1
* Values for exch Ca for 0 and 4 months are exch. Ca + Mg.


Table 4. Changes in soil organic matter with time.
Soil Depth Months After Abandonment1
(cm) 0 2 9 12
S% organic matter
0-5 2.11 + 1.00 1.25 1.34 + .37 1.69 + .50
5-15 1.50 + .46 0.96 0.95 + .17 1.20 + .11
15-30 1.26 + .18 -
Soil Depth Months After Abandonment
(cm) 0 36 84 132
0-15 2.07 + .40 1.05 + .13 1.67 + .19 1.21 + .20
15-30 1.26 + .18 0.63 + .06 0.84 + .32 0.85 + .23
30-45 -0.44 + .24 0.83 + .17 0.50 + .17
45-100 -0.23 + .09 0.65 + .11 0.29 + .10
1 0-12 month data were measured at the same site; 3-, 7-, and 11year-old data are from different sites.

provided by atmospheric inputs and the decomposi-
tion of organic matter in and on the soil, and
retranslocation within the vegetation itself. A build-
up in soil organic matter, presumably due to litter fall,
between three and seven years is observed (Table 4).
Changes in the biomass of the forest floor have been
measured, but the data are not available at this time.
In any case, further analysis, including the contribu-
tion of atmospheric inputs and the construction of a
nutrient budget, is needed.

1. Secondary succession is in general very dynamic.
Pioneer vegetation quickly establishes its photosyn-
thetic material and a soil cover. Photosynthesis and
transpiration set up soil-water and nutrient gradients
that enable rapid biomass accumulation and nutrient
immobilization in the biomass.
2. Rapid vegetation growth affects soil properties
through root growth and turnover, organic matter ad-
ditions, vegetation-mediated changes in soil
microclimate, and the canopy's ability to reduce the
impact of rainfall on the topsoil, all probably con-
tributing to the decrease in bulk density and improv-
ing the retention of soil moisture. It is interesting to
note that where root biomass is similar, bulk density
is also similar. And, while differences in sand content
may influence the increase in field capacity over time,
this increase during the first fifteen months agrees well
with increases in soil organic matter over the same
3. Nutrients become impoverished in the topsoil over
time, largely as a consequence of nutrient uptake and
immobilization by vegetation.
4. It is unclear whether the higher biomass accumula-
tion and productivity of the seven-year-old purma

should be attributed to a growth phase common to
this stage of succession, or to differences in sites. Data
show that the site of the seven-year-old purma was
less sandy and somewhat more fertile than the others.
5. A number of questions remain. The effect and
importance of changes in soil physical properties on
vegetation growth remain largely unknown, as does
the effect of soil fertility on biomass production and
vegetation composition. The relative importance of
atmospheric inputs, N fixation, soil organic matter,
and litter as sources of nutrients for developing vegeta-
tion is likely to vary according to soil type, and these
factors also need further evaluation. It is important
to conduct long-term studies at permanent sites, in
order to avoid the problems created when site and time
are confounded.

Effect of Soil Fertility
On Shifting Cultivation Fallows
The purpose of this study was to answer some of
the questions regarding the effect of soil fertility on
secondary succession in abandoned agricultural fields.
Specifically, the study was designed to quantify the
size of potential or actual nutrient pools in an aban-
doned agricultural field, and to investigate the effects
of soil chemical properties on secondary succession by
manipulating soil fertility.
A 0.5 ha abandoned field that had been cropped
with a rice-corn mixture was sampled on 24 plots of
64 m < 2. Analyses included total soil nutrients to
a 1 m depth; exchangeable Al, Ca + Mg, K; available
P contents in the upper 100 cm; and soil organic
Biomass measurements and subsamples for nutrient
analysis were obtained for litter, crop residue, and living
vegetation. The leaf-area index (LAI), average heights


of the five dominant individuals per plot, and percen-
tage of cover were also measured.
Three rainfall collectors were installed, as well as
six zero-tension leachate collectors. The lysimeters were
placed in zones of textural change, usually between
15 and 20 cm depth.
Soil physical properties-texture, bulk density, field
capacity, and pore size distribution-were also sampled.
After the initial sampling, four fertility treatments
were established: 1) undisturbed controls, 2) litter
removal, 3) litter additions (the equivalent of 3.25 t
of rice straw and 17.4 t of wood per hectare) and 4)
fertilization with 100 kg P/ha, 100 kg N/ha, 100 kg
K/ha and and 50 kg Mg/ha.
Soil and vegetation dynamics were studied by means
of periodic measurements. These included: topsoil
organic matter approximately every two months; ex-
changeable Al, Ca + Mg, K; available P; pH at four,
ten and 17 months after abandonment; topsoil bulk
density at zero, ten and 17 months after abandonment;
field capacity at 0 and 15 months; infiltration rate at
17 months; and pore size distribution at zero, ten and
17 months. The heights of the five dominant in-
dividuals per plot, LAI, % cover, and biomass (weight
and nutrient concentration by vegetation type, i.e.,
grasses, lianas, herbs, and trees) were measured at zero,
four, ten and 17 months. Root biomass down to a
50 cm soil depth was measured at one and 17 months.
Rainfall and leachate samples were collected within
two days after a rain, the quantity collected was
measured, and a subsample was taken for chemical

Nutrient Pools
Soil total nutrient analyses are shown in Table 5.
The increase in nutrients observed in the 15-60 cm
horizon is due to an increased clay content. The
relatively large amounts of total K and Mg reported
may be due to the presence of vermiculite-like minerals
having imperfect substitution or Al oxide minerals hav-

Table 5. Total soil nutrient stocks present at field abandonment
(mean of three replications).
Depth (cm) Kjeldahl Ca Mg P K
0-15 857 292 361 173 669
15-60 2410 67 1264 613 3507
60-100 1431 25 1041 505 2393

ing K or Mg entrapped in the interlayers. These
nutrients are, for the most part, probably unavailable.
It has been reported that, after kaolinite, vermiculite
is the next most common soil mineral in the
Yurimaguas soils.
In general, the total quantity of nutrients in the soil
appears sufficient to support more than one crop
harvest. However, the slow release of minerals into
plant available forms and the possibility of Ca defi-
ciencies or Ca/Mg imbalances would probably
negatively affect crop productivity. Moreover, available
nutrient levels in the topsoil, while not high, general-
ly appear sufficient, although K availability may be
a limiting factor. The existence of large total nutrient
pools deep in the soil profile, on the other hand, raises
the possibility of nutrient "pumping" by deep-rooted
Preliminary analyses of atmospheric inputs suggest
that inputs may be sizeable for K, Na, and Ca, but
may be within the range reported in the literature. An
unexplained anomaly is the rainfall pH value of 6.9.
Rainfall should have a pH of around 5.5. A higher
pH value may indicate contamination. In any case,
atmospheric inputs fall far short of supplying the
nutrients needed for continual crop production. Ad-
ditions over a ten-year period, however, may be suf-
ficient to support one or two crop harvests, assuming
total nutrient conservation and similar atmospheric in-
put levels for all years.
The importance of large and small litter pools pre-
sent at abandonment cannot be assessed at this time
due to the absence of nutrient analyses of litter. Con-
sidering that litter biomass ranges between approx-
imately three and 11 t/ha, litter decomposition may
supply large quantities of nutrients to vegetation
In comparison to nutrient additions, leaching losses
appear relatively small. Losses were greatest for K, as
might be expected based on its chemical characteristics.
In general, leaching losses varied by treatment. Losses
were greater in those treatments with larger quantities
of available nutrients.

Vegetation Dynamics
In general, a great deal of within-treatment variability
is observed in the measurements of living above-ground
biomass (Table 6). In many cases, coefficients of varia-
tion are 100% or more. Such variation is due to the
clumpyy" nature of vegetation distribution and the
quadrat sampling techniques used. Such variability
obscures trends and makes interpretation difficult.


Table 6. Changes in living aboveground biomass following field adandonment (kg/ha).
Broad-leafed Trees, Trees, Crop
Treatment* Grass Weeds Lianas <2.5cm dbh >2.5cm dbh Remnants
1 Month After Abandonment
1. 28 43 214 108 0 2590
2. 28 35 447 821 0 2113
3. 204 34 249 234 0 2617
4. 210 203 99 272 0 2030
4 Months After Abandonment
1. 928 165 445 763 0 144
2. 520 361 421 531 0 511
3. 1029 477 894 825 0 147
4. 1006 449 1555 3243 0 72
10 Months After Abandonment
1. 3372 628 741 1539 0 0
2. 2085 89 382 720 625 0
3. 2190 54 241 1300 1475 0
4. 2394 549 315 1211 9150 0
17 Months After Abandonment
1. 2467 14 1481 1745 1841 0
2. 2775 177 287 2934 5562 0
3. 1072 72 616 872 5037 0
4. 456 346 638 2217 7038 0
* Key to Treatments: 1) Natural secondary succession control; 2) residue removed; 3) residue added; 4) fertilized.

A number of patterns in biomass accumulation are
apparent, regardless of fertility treatment (Table 6 and
Figure 3). Above-ground living biomass decreases dur-
ing the first four months after abandonment as crop
plants remaining after harvest senesce. The crop plants
are quickly replaced by grasses and trees which are
codominant for a while. The grasses eventually
decrease in biomass as tree biomass continues to in-
crease both absolutely and in relative importance. Her-
baceous weeds and lianas, although present, are of
much lesser importance. Their biomass generally peaks
between four and eight months and decline thereafter.
There appear to be very few treatment-related dif-
ferences in biomass accumulation. Up to ten months
after abandonment, total above-ground living biomass
is similar in the undisturbed check and the residue
treatments. Biomass levels, considered by vegetation
type, are also similar across these treatments. During
this period, however, greater biomass levels were
recorded in the chemically fertilized treatment, primari-
ly due to a greater tree biomass. These differences may
not be significant due to variability in the data.
Biomass measurements at 17 months show little dif-
ference between the check and residue-addition
treatments, while the residue-removal and chemical-


10 t-

0 1
0 2
A 3
0 4

r I I I I I
0 4 8 12 16
Months After Abandonment

Figure 3. Changes in living above-ground biomass
with time.


ly fertilized treatments have similar levels of biomass,
averaging 2-3 t/ha more than the check. The data sug-
2.2 gest that tree growth responds quickly to chemical fer-
tilizer additions. There is little difference among
treatments in topsoil or subsoil chemical properties at
ten months, with the exception of higher P levels in
2.0 Treatment .
20 1 the chemically fertilized treatment. This suggests that
S\ 2 the nutrients which were added were lost either in sur-
S0 3 face runoff or erosion, converted to an unavailable
-1.8 0 4 form, or immobilized in litter and biomass. The lat-
ter possibility, at least, can be investigated via nutrient-
budget calculations.

0 \Soil Organic Matter
S1.4 Topsoil (0-5 cm) organic matter declined in the first
cn six months after the field was abandoned, but increased
from six to 12 months (Figure 4). There was no ap-
1.2 parent effect of the different treatment. Similar but
less marked changes were also observed in the 5-15
cm layer.
1.0 -
1. The recovery of vegetation in an abandoned
0.8 agricultural field appears to respond to the application
0 2 4 6 8 10 12 of 100 kgN, 100 kgK, 100 kgP, and 50 kgMgper
hectare. This response is mainly apparent in increas-
ed tree growth. Despite differences in tree growth,
Figure 4. Changes in soil organic matter with time there was little effect of treatment on soil physical or
(0-15 cm). chemical properties. This suggests the amounts of
nutrients added or removed in vegetation residue is
small relative to soil pools and that nutrients added
via the application of fertilizers may only be available
to the vegetation for a short period of time before they
are taken up, leached, or converted to chemical forms
unavailable to plants.
2. Further work is needed on the quantity and rate
of release of nutrients from soil pools as well as in-
puts to and losses from these pools.


New-Project Update
This project has not been under way long enough to yield substantive reports, but should be
mentioned because of its importance to the program as a whole.

Comparative Soil Dynamics
Julio C. Alegre, N. C. State University
Pedro A. Sanchez, N. C. State University
Luis Arevalo, N. C. State University
Jorge Perez, INIPA
Manuel Villavicencio, INIPA

This project, established on a tract of ten-year-old
secondary forest at the Yurimaguas Experiment Sta-
tion, compares the effects of various crop-management
systems on changes in the physical, chemical and
biological properties of an Ultisol upon clearing. The
experiment consists of six management options as
treatments, which are: 1) shifting cultivation, with plots
contracted out to a farmer who planted upland rice
intercropped with cassava and plantain; 2) mechaniz-
ed continuous cropping, with corn and soybean, on
a plot cleared with a tractor and straight blade, then
logged, burned, disked and fertilized according to soil-
test recommendations; 3) low-input technology, in
which a rotation of two crops of rice followed by
cowpea will be carried out until productivity declines;
then it will be placed in managed fallow; 4) combin-
ed tree and crop production, with rice interplanted
with tornillo (Cedrelinga catenaeformis), and Inga edulis
planted between the Cedrelinga; 5) peach palm
(Gulielma gasipaes), interplanted with the first crop in
a sequence of rice, rice and cowpea; and 6) a forest
fallow check, in which the plots were not disturbed.
Plots were sampled before and after clearing, and
after burning. Soil chemical, physical and biological
properties are being monitored intensively.

Collection and Propagation
Of Agroforestry Species
Angel Salazar, INIPA
Jorge Perez, INIPA
Cheryl A. Palm, N.C. State University
Alwyn Gentry, Missouri Botanical Garden
Kenneth MacDicken, University of Hawaii

Most tree and shrub species used for agroforestry
in the humid tropics are successful in high base status
soils. This project seeks to identify species that can suc-
ceed on acid soils. There was progress during 1985
in several areas:
1) At Yurimaguas, seed of eleven legumes with
potential in agroforestry were collected, identified and
planted at the nursery.
2) Some 200 trees, shrubs and vines, collected
around Yurimaguas during a survey by the Flora of
Peru Project, are being identified.
3) A trial to evaluate trees for alley-cropping was
established, with eight of twenty tree species planted.
The trees will be studied for survival, growth rate,
regrowth after pruning, biomass production and
nutrient accumulation.
4) In collaboration with the Nitrogen Fixing Tree
Association (Hawaii), a site was prepared for a trial
designed to measure biomass production and N ac-
cumulation in species considered strong N-fixers on
acid soils.


New-Project Update

Living Fences
Jorge Perez, INIPA
Jose Benites, N.C. State University

The objectives of this project are 1) to determine
the most suitable tree species for use in living fences,
and 2) to test the idea that living posts, sprouted from
stakes or poles, could persist on acid soils, reducing
the high cost of fence maintenance on farms in the
humid tropics.
Research was begun at Yurimaguas with three pro-
mising species: mullohuayo (Seca floribunda), llambo
pashaco (Eslorabium sp.) and huina caspi (genus and
species unknown).

Sprouting percentage was determined with 0.5 or
1.0 m long mullohuayo stakes as a function of days
of storage after cutting.
Over 80% sprouting was achieved with seven-day-
old stakes of either size. Similar results were obtained
with 2.5 m long stakes.

Llambo pashaco
For this species, 2.5 m stakes were used. Seven days
after planting, 80% of the stakes had sprouts, but the
vigor disappeared quickly and all stakes died.

Huina caspi
This species grows on upland soils. It is traditional-
ly used as fencing material in the gardens of pueblos
jovenes of Yurimaguas. The same experiment was car-
ried out as with mullohuayo, using 0.5 m and 1.0 m
stakes. Results were negative because the macroporosi-
ty of this species allows a very quick loss of its water
content. Ninety days after planting the survival rate
was 82%. At 180 days after planting, only 26% of
the living plants were found. The causes of mortality
are not known but sprouts began drying and fell from
the trunks.
Further study will test ways to stimulate rooting
and rapid growth. Native legumes with a high poten-
tial for vegetative propagation will also be tested in
the living fences.


Continuous cropping is in many ways an attractive system to those in the humid tropics who seek
to produce more food on less land, conserving natural forests while developing a more stable agricultural
base. Research has shown that the effective use of modern fertilizers and lime can dramatically in-
crease crop yields in the acid, infertile soils of the humid tropics. Before this technology can be applied
successfully in developing nations, however, it must first be adapted to local economic, social and
environmental conditions.
A stable system of continuous cultivation must overcome several soil-related constraints. The com-
mon humid-tropical weather pattern of intense rainstorms followed by dry periods promotes soil ero-
sion and water stress in crops. Most of the available soils are acid and infertile, and are subject to
nutrient leaching during heavy rains. Tropical weeds, which have traditionally been controlled by
extended fallows and very brief periods of cropping, soon dominate food crops if they are allowed
to establish themselves in the fields.
The reports that follow describe research aimed at managing each of these primary constraints. Most
of the research has been conducted at the Yurimaguas Experiment Station in Peru. Several projects
examine various tillage methods and their effect on soil physical and chemical properties. A second
group reports experiments in fertilizer management, while a third deals with methods of weed con-
trol. The central continuous-cropping experiment incorporates information and methods generated
by each of these experiments, so as to develop and refine a comprehensive soil-management program
for the continuous cultivation of food crops in the humid tropics.


Land-Clearing and Post-Clearing
Soil Management Practices
Julio C. Alegre, N. C. State University
D. Keith Cassel, N. C. State University
Dale E. Bandy, N. C. State University

In the humid tropics, large areas of cleared land have
been abandoned because their soils were too com-
pacted, eroded or infertile to support crops. Much of
this damage has been blamed on the use of bulldozers
with straight blades in land-clearing. There has been
a need for information on alternatives to straight-blade
bulldozing, as well as for soil-management practices
that could improve the productivity of cleared land.
The purpose of this project, which was conducted at
the Yurimaguas Experiment Station, was 1) to deter-
mine the rate of change of selected soil physical pro-
perties resulting from alternative land-clearing and soil-
management practices, and 2) to evaluate crop per-

formance as affected by land-clearing and soil-
management practices of a humid tropical Ultisol.

Soil Physical Properties
Clearing methods and burning and tillage treatments
are shown in Table 1, along with bulk-density
measurements for two depth intervals, taken 15 days
before harvesting the first and last crops. Soils in the
treatments were 8 to 10% clay, and particle-size
distribution at the 0 to 15 cm depth was not altered
by land-clearing. Topsoil removal and mixing of top-
soil and subsoil are probably more a function of the
bulldozer operator than of the clearing method itself.
Clearing tended to increase the variation in Db
(Table 2), as indicated by the higher standard devia-
tions. Compaction occurred for slash-and-burn clear-
ing due to foot traffic during slashing and trunk
removal. The greatest numerical increase in Db oc-
curred for straight-blade clearing, but it was not
significantly different from the shear-blade and slash-

Table 1. Mean bulk density and standard deviation (in parentheses) of Yurimaguas soil at two depths for two
times as a function of land clearing methods, burning and soil tillage treatments.
8 Months After Clearning 23 Months After Clearing
Treatment 0 to 15 cm 15 to 25 cm 0 to 15 cm 15 to 25 cm
Slash/burn 1.17*(0.10)a** 1.37 (0.10) b 1.32(0.04) c 1.38 (0.04)b
Straight blade/rototill 1.26 (0.16)a 1.44 (0.19)ab 1.42 (0.06)ab 1.56 (0.15)a
Straight blade/chisel/rototill 1.25 (0.09)a 1.49 (0.09)ab 1.46 (0.11)a 1.57 (0.07)a
Shear bladelburn/disk/rototill 1.18 (0.11)a 1.45 (0.15)ab 1.34 (0.11) bc 1.52 (0.11)a
Shear blade/rototill 1.28 (0.24)a 1.52 (0.12)a 1.32 (0.12) c 1.58 (0.04)a
Shear blade/disk/rototill 1.31 (0.14)a 1.56 (0.16)a 1.33 (0.11) bc 1.57 (0.06)a
Each value is the mean of nine measurements in cores of 76 x 76 mm.
* Means in a given column with the same letter are not significantly different at the 5% level by the Waller-Duncan
multiple comparison test.

Table 2. Mean bulk density and standard deviaton (in parentheses) of Yurimaguas soil prior to and three months
after land clearing.
Depth (cm)
Time Clearing Method 0 to 15 15 to 25
Before clearing* 1.16 (0.09)b*** 1.39 (0.08)b
Three months
after clearing** slash/burn 1.27 (0.07)a 1.37 (0.10)b
straight blade 1.42 (0.12)a 1.49 (0.12)a
shear blade 1.28 (0.25)a 1.50 (0.15)a
Each value is the mean of 36 measurements.
** Each value is the mean of 18 measurements.
*** Means in a given column with the same letter are not significantly different at the 5% level with the Waller-
Duncan comparison test.


and-bum clearing methods. This lack of difference may
be attributed to the high variability produced by the
mechanical clearing methods.
Bulk density in the 15 to 25 cm depth increased
for both types of mechanical clearing, but not for slash
and burn. No differences were found at the upper
depth before harvesting the first crop, but at the lower
depth greater Db values were observed for some ran-
dom treatments compared to slash and burn. After 23
months, Db of the upper depth for the slash-and-burn
treatments was significantly lower than Db for two
other random treatments. At the lower depth, all
mechanized land-clearing treatments had greater Db's
compared to the slash-and-burn treatment.

Water-Holding and Infiltration
The soil water characteristics for the 0 to 15 cm
depth prior to clearing are shown in Figure lA. The
vertical bar through each point is two standard-
deviation units long. The amount of water held bet-

-1500 -40 -30 -20 -10 0
Pressure Head (k Pascal)
0.19 6.5 7.4 9.8 14.7 29.4
C. Pore Neck Diameter (pM)

Three Months After Clearing
15 25 cm depth
A Slash and Burn
0 Straight Blade
A Shear Blade T

I LSD .05

-40 -30 -20 -10 0
Pressure Head (k Pascal)

ween soil water pressures of -13 to -1,500 KPa was
assumed to approximate the soil's capacity for holding
plant-available water before clearing, and was equal
to 0.187 m3/m3. The diameter of the largest pore neck
that retained water at the applied soil water pressures
is also shown in Figure lA. Pores with neck diameters
> 23 /. m were drained at in situ field capacity.
The soil water characteristics at the 0 to 15 cm depth
three months after clearing are presented in Figure lB.
Straight-blade clearing increased soil water content for
soil water pressures < -2 KPa when compared to the
slash-and-burn and shear-blade treatments. These
higher water contents are attributed to destruction of
macropores by compaction, which, in turn, increased
the volume of micropores. Soil water characteristics
for the 15 to 25 cm depth showed the same trend three
months after clearing (Figure 1C). The soil water
characteristics for the 0 to 15 cm depth for the three
soil-management subtreatments eight months after
clearing are shown in Figure ID. The bedded treat-
ment had the greatest total porosity and greatest


0.4 r 0.4

0.2 0.2
0 0

0.4 E 0.4

0.2 0.2

Three Months After Clearing
0 15 cm depth
A Slash and Burn -
Straight Blade
A Shear Blade

I LSD .05

-40 -30 -20 -10 0
Pressure Head (k Pascal)


Eight Months After Clearing
0 15 cm depth
o Flat Planted
A Flat Planted/Fertilizer, Lime
O Bedding/Fertilizer, Lime

I LSD .05

-40 -30 -20 -10 0
Pressure Head (k Pascal)

Figure 1. Soil water characteristics of Yurimaguas soil before and several times after clearing.



volume of large pores because less soil disturbance oc-
curred due to land preparation by hand.
The mean infiltration rate before clearing was 420
mm/hr over a two-hour period (Figure 2A).
Mechanical clearing with both the straight blade and
the shear blade significantly reduced infiltration rate.
Mean infiltration rates during the first two hours of
infiltration were 304, 14, and 32 mm/hr three months
after clearing for the slash-and-burn, straight-blade and
shear-blade methods, respectively. Because infiltration
measurements were so time-consuming, it was not
possible to measure cumulative infiltration for all
treatment-subtreatment combinations. However,
measurements were replicated six times for those com-
binations measured. Cumulative infiltration during a

2B. Contin


Figure 2. i
before an
methods (

2 3 months after straight-blade clearing resulted in a
low infiltration rate when no other soil-management
treatment was used before planting. However, when
straight-blade clearing was coupled with chisel plow-
ing before planting the first crop, cumulative infiltra-
tion values were similar to those for the slash-and-burn
During the 23-month-long period of continuous
cropping, cumulative infiltration over a two-hour
period decreased from 800 to 200 mm for the slash-
and-burn treatment (compare Figures 2A and 2B). On
the other hand, cumulative infiltration increased for
the straight blade/chisel and the shear blade/disk

period for selected land-clearing sub- Crop Response
23 months after clearing is shown in Figure Grain yield of rice, the first crop seeded after the
uously cropping the Yurimaguas soil for treatments were imposed, was highest for the slash-
A. and-burn treatment (Table 3). This was expected
because slash and burn supplies nutrients in ash and
Slash and Burn (before clearing) leaves the topsoil in place. The shear blade/burn/disk
O Slash and Burn
Shear Blade 3 mos. after treatment also incorporated nutrients from ash into
Straight Blade clearing the soil and produced the second highest yield. Very
little removal of topsoil or mixing of subsoil with top-
soil occurred for the shear-blade treatment. The
ST bed/fertilization subtreatment had the highest grain
LSD .05 yields followed by the flat/fertilization. Soil in the beds
maintained good structure, partly because field laborers
never walked on the elevated beds. The highest grain
yield for the second rice crop (the fourth consecutive
I I I I I -- I crop in the rotation) was for the shear/burn/disk treat-
B. ment followed by the slash-and-burn and the straight-
blade/chisel treatments (Table 3).
0 Slash, Burn, Flat Planted
A Straight Blade, Chisel, Flat Planted In general, rice grain yield was 0.5 to 0.7 Mg/ha
t Straight Blade, Flat Planted less for the fourth crop compared to the first crop.
This was especially true for those plots where no fer-
tilizer was added because most of the nutrients had
been removed by the previous crops.
A very poor soybean crop resulted from poor ger-
LSD .05 mination due to dry conditions and low-quality seed.
Yields (Table 4) were low compared to the 2.5 Mg/ha
yield obtained from other studies nearby. The highest
grain yields occurred for the slash-and-burn, straight-
blade/chisel, and shear-blade/burn/disk treatments.
I I I |I I I I The response to chiseling land cleared by straight blade
0 20 40 60 80 100 120 was 0.38 Mg/ha. Based on the general response to
chiseling and disking, it appears that soil compaction
Time (min) constrained soybean growth and yield.
Cumulative infiltration of Yurimaguas soil Corn height for the third crop was significantly
d after clearing for three land-clearing greater for the slash-and-burn and shear-
A), and cumulative infiltration for selected blade/burn/disk treatments (Table 5). For both
ing methods 23 months after clearing (B).


treatments involving burning, some available nutrients
still remained and plant height was 0.85 m. Grain
yields for the fifth crop were greatest for the slash-
and-burn and shear-blade/bum/disk treatments (Table
5). There was good response to all treatments receiv-
ing disk or chisel tillage. Without fertilizer and lime
plants did not survive. Although application of fertilizer
and lime appeared to compensate for some of the ef-
fect of compaction, corn growth and yield were bet-
ter on chiseled land. Soil structure in the bedding
system was favorable to germination and root distribu-
tion, and bedding produced higher grain yields.
For the slash-and-burn treatment, the bed/fertiliza-
tion management resulted in the best soil physical con-
ditions, less lodging and higher yields. Physical pro-
perties of the subsoil for this treatment were never

altered by clearing or management from their initial
condition, and therefore provided a good environment
for roots.

1. Most of the changes in soil physical properties
occur during the land-clearing process rather than after
2. Degradation of physical properties due to
mechanical land-clearing is sufficient to decrease crop
yields if no effort is made to improve soil physical con-
3. The least deterioration in soil physical proper-
ties occurred in the slash-and-burn treatment.
4. Chiseling and disking after mechanical land-
clearing tended to offset some of the undesirable ef-

Table 3. Rice grain for the first and fourth consecutive crops as affected by land clearing, tillage and soil
Treatment Flat/No Fert. Flat/Fert. Bed/Fert. Mean
First crop
Slash/burn 3.11 3.56 3.98 3.55a **
Straight blade 0.91 2.75 3.38 2.35 cd
Straight blade/Chisel 1.14 2.84 2.85 2.28 cd
Shear blade/burn/disk 2.39 3.06 3.68 3.04 b
Shear blade 1.27 3.02 3.20 2.49 c
shear blade/disk 0.91 2.58 2.74 2.07 d

Fourth Crop
Slash/burn 0.75 3.48 2.82 2.35 b
Straight blade 0.26 1.56 2.00 1.27 d
Straight blade/chisel 0.70 3.04 2.91 2.22 b
Shear blade/burn/disk 1.58 3.44 2.96 2.66a
Shear blade 0.97 2.13 2.28 1.80 c
Shear blade/disk 0.63 1.83 1.96 1.47 d

Table 4. Soybean grain yields for the second crop as affected by land clearing-tillage and soil management.
Grain (Mg/ha)
Treatments Flat/No Fert. Flat/Fert. Bed/Fert. Mean
Slash/burn 0.42 2.32 1.86 1.53ab **
Straight blade 0.10 1.03 1.50 0.88 d
Straight blade/chisel 0.32 1.44 2.03 1.26abc
Straight blade/burn/disk 0.48 2.17 2.05 1.57a
Shear blade 0.18 1.37 1.87 1.14 cd
Shear blade/disk 0.12 2.01 1.56 1.23 bc
** Means in a given column followed by the same small letter and means in rows followed by the same capital
letter are not significantly different at the 5% level by the Waller-Duncan multiple comparison test.


Table 5. Corn height for the third crop and corn grain yield for the fifth crop harvest as affected by land clear-
ing tillage and soil management.
Treatment Flat/no Fert. Flat/Fert. Bed/Fert. Mean
Plant height,m

Third crop
Straight blade
Straight blade/chisel
Shear blade/burn/disk
Shear blade
Shear blade/disk

Fifth crop
Straight blade
Straight blade/chisel
Straight blade/burn/disk
Shear blade
Shear blade/disk



58 2.34
48 2.20
29 2.35
63 2.23
51 2.25
43 2.03
.Grain yield, Mg/ha

*** Means in columns for a given harvest followed by the same small letter and means in rows followed by the
same capital letter are not significantly different at the 5% level by the Waller-Duncan multiple comparison test.

fects of land-clearing on soil physical properties. For
example, the infiltration rates for soil cleared by the
mechanized methods were increased by disking and
chisel plowing prior to planting the first crop.
5. Declines in soil physical properties under con-
tinuous cropping were minimized when soil was bedd-
ed using a hand hoe.
6. Slash-and-burn clearing resulted in higher yields
for rice, soybean and corn compared to mechanized
clearing. Of the mechanical land-clearing and tillage
methods examined, shear blade/burn/disk produced
the highest rice, soybean and corn yields. All crops
showed a positive yield response to treatments in which
soils were chiseled or disked after clearing.

While all of the land-clearing methods examined
adversely affect these soils to some extent, traditional
slash and burn does the least amount of damage.
Machinery can apparently be used to clear secondary
forests for crop production if its use is coupled with
effective soil-management techniques after clearing. Of
the mechanical methods examined, the best alternative
to slash-and-burn clearing was a combination of fell-
ing with a shear blade, then burning the vegetation
on-site as traditionally practiced. However, unless
followed by disking or chisel-plowing, this method will
tend to produce lower crop yields and less favorable
soil properties than traditional slash-and-burn clearing.

Tillage With Tractors
In Continuously Cropped Ultisols
Robert E. McCollum, N. C. State University

Before January, 1984, the long-term continuous-
cropping experiment at Yurimaguas had been tilled
with a small hand rototiller to a depth of 7 to 10 cm
since it was first cleared in 1972. While rototilling had
been considered a possible intermediate step between
hand tillage and tractor-drawn implements, it has
several disadvantages: Tillage with rototillers has
resulted in a shallow root system limited to the zone
of fertilizer incorporation; small rototillers do not pro-
perly incorporate crop residues; and rototilling is slow.
The objectives of this project are 1) to determine
whether tillage with tractor-mounted farm equipment
is possible on humid tropical Ultisols; and 2) to deter-
mine whether tillage with tractors is agriculturally and
ecologically sound in humid tropical environments.

Tillage Practices
An abandoned portion of Chacra I, which is the site
of the continuous-cropping experiment at Yurimaguas,
was prepared duringJuly-August, 1983 via the follow-
ing steps: 1) Chop crop residues with rotary mower;
allow to dry; burn; repeat as necessary; 2) disk as
necessary to reduce remaining residues to a manageable

1.23 d
1.64 bc
1.74 b
1.59 c

0.88 c
1.04 c
0.96 c


Table 1. Crop yields on Chacra I after initiating tractor-mounted tillage. Months refer to harvest date.
Comparisons Rice Corn Corn Corn Corn1
Jan 84 Jan 84 July 84 Jan 85 July 85
Grain yields, th/a, and (number of observations)
General Mean 2.47 (101) 2.96 (42) 3.15 (216) 3.48 (125) 2.90 (36)
Clearing Method
Burned, 1972: 2.56 ( 41) 3.00 ( 8) 3.27 ( 99) 3.50 ( 64) 2.70 (15)
Bulldozed, 1972: 2.54( 43) 2.93 (33) 2.93( 90) 3.70( 32) 2.87 (12)
Peripheral Areas2 2.10 ( 17) 3.49 ( 1) 3.42 ( 27) 3.19 ( 29) 3.28 ( 9)
Lime + P
None 2.74( 16) 2.35 (17) 2.71 ( 56) -
Lime + P3 2.42 ( 85) 3.38 (25) 3.30 (160) -
Banded P
None 2.54(15) -
Banded P 3.19 (27) -
1 Low Corn population (29,000 plants/ha)
2 Clearing method unknown
3 2.5 t lime/ha + 100 kg/P/ha

bulk; 3) plow to 20-25 cm with a moldboard plow
to incorporate plant residues and soil amendments; 4)
disk again; 5) make tractor tracks as row markers; 6)
construct 75 cm beds; 7) prepare seedbed with
rototiller, and 8) plant crop, either by hand or with
tractor-mounted Cole planters. Lime and fertilizer ap-
plications were consistent with practices developed for
Chacra I. Weeds were controlled chemically with
Corn and upland rice were planted in August, 1983
and harvested in January, 1984. The same tillage opera-
tions were repeated prior to planting three consecutive
corn crops, which were harvested in July, 1984,
January, 1985 and July, 1985.

Product yields during four biological cycles follow-
ing initiation of tractor-mounted tillage are shown in
Table 1. While there were no treatments in an ex-
perimental sense, some comparisons based on previous
treatment and current management can be made regar-
ding clearing method, and the effects of lime and P.

Clearing Method
The site was cleared in 1972 by bulldozing or by
traditional cutting and burning. The data show that
any differences between initial clearing methods were
virtually eliminated by tilling to 20 cm.

Effects of Lime + Phosphorus
Lime and phosphorus, broadcast before the first crop,
increased corn yields but had not effect on rice. In-
row banded P resulted in a 26% increase in corn yields
during the first cycle.

Initial observations have been that mechanized tillage
is possible on Ultisols in the Yurimaguas environment.
Tillage can be accomplished once a year, during the
"dry" season, July-September. Mechanized tillage at
other times should only be undertaken with strict at-
tention to soil moisture. Continuing work in this area
will seek to determine if mechanized tillage is
agriculturally and ecologically sound in humid tropical


Continuous Cropping:
Central Experiment
Robert E. McCollum, N. C. State University
Pedro A. Sanchez, N. C. State University
Dale E. Bandy, N. C. State University

This experiment is a long-term demonstration of a
continuous-cropping system for acid soils of the humid
tropics. The system is based on the judicious use of
fertilizers and the best available soil-management prac-
tices. It occupies eight 10 m by 28 m main plots in
each of two fields known as Chacra I (site one) and
Chacra III. There were seven fertility-management
treatments under each of two rotations, rice-corn-
soybeans and rice-peanut-soybeans, and four replica-
tions each in 4 x 10 m plots. These plots have been
cropped continuously since September of 1972, pro-
ducing 31 crops to date in Chacra I.
From May, 1982 until January, 1984, the seven
fertility-management treatments were:
1. A check, with no tillage and no fertilizer or lime
2. Complete fertilization plus 1 kg Mn/ha, applied

one month after planting, to the foliage of either corn
or peanut.
3. Complete fertilization as follows (in kg/ha/crop):
rice, 100 N; corn, 80 N, 30 P, 100 K; and 30 Mg
to all crops.
4. Complete fertilization plus 2 kg Mn/ha, applied
as in treatment 2.
5. Previous crop residues left on soil and
6. Previous crop residues left on soil but not incor-
7. 1.5 times the complete fertilization.
This report describes results fromJuly, 1981 toJu-
ly, 1985, including the harvests of crops 25 through
31 (Table 1). Prior to January, 1984, all of the plots
except the check (treatment 1) were rototilled with
a hand tractor to a depth of about 7.5 cm. Tillage with
tractor-drawn implements to a depth of 20-25 cm was
introduced with crop 29, and was used thereafter in
the corn crops discussed in this report.
Continous cropping on Chacra III was discontinued
in September, 1983. No significant yield differences
were noted between Chacras I and III during the last
four harvests. Twenty-one crops were harvested from
Chacra III within a span of 112 months.

Table 1. Crop yields of continuously cultivated plots from January, 1982 to July, 1985.
Crop No., Species and Harvest Date1
25 25 26 27 27 28 28 29 30 31
Treatments2 Corn Peanut Rice Corn Peanut Corn Rice Corn Corn Corn
1/82 8/82 1/83 6/83 6/83 1/84 1/84 7/84 1/85 7/85
1. Check3 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
2. Complete + Mn 2.97a 4.63a 3.04 c 1.93b 2.71a 1.21c 1.51b 1.82ab 2.56c 1.85ab
3. Complete 3.28a 4.39a 3.45b 1.62b 2.74a 1.91bc 1.55b 2.76b 3.02bc 1.56b
4. Complete + 2 Mn 2.87a 4.75a 3.68b 1.88b 2-.76a 2.48ab 1.99a 3.05ab 2.94bc 2.06ab
5. Complete + residue 3.21a 4.54a 4.48a 1.68b 2.42a 2.76a 2.00a 3.58ab 3.73a 2.80a
6. Complete + residue 3.20a 3.28b 3.71b 2.04b 2.51a 2.61ab 1.44b 3.91a 3.21ab 2.48ab
7. 1.5 x complete 3.49a 4.54 3.54b 2.66a 2.50a 2.31ab 1.80ab 3.06ab 3.15 1.98ab
Mean 3.17 4.35 3.65 1.97 2.61 2.21 1.72 3.20 3.10 2.12
(No. of obs.) 8 8 16 8 8 4 4 8 8 8
C.V. (%) 19 14 12 28 15 23 14 27 17 39
1 Crop yields from 1/82 through 6/83 are pooled data from Chacras I and III; subsequent harvests are from Chacra
I only.
2 Treatments up to crop #28: uniform fertility and conventional tillage to 20-25 cm to all but treatment 1 for crops
29, 30, 31.
3 Treatment 1 dropped from data array for statistical analysis.


Table 2. Effects of selected treatments on relative yields January, 1982 January, 1984. Crops 25 28
in Chacra I.
Corn Rice Peanut
Treatment (3 crops) (2 crops) (2 crops)

Relative Yieldl
3. Complete + residue removed 91 b 94 b 103 a
5. Complete + residue incorporated 104 a 121 a 99 a
6. Compete + residue as mulch 107 a 98 b 86 b
7. 1.5 x complete + residue removed 116 c 99 b 100 a
* Mean yield over all treatments within a cycle = 100.

Crop Performance
Yields of crops 25 to 31 from Cbacra I: Check-plot
yields have been zero for the last ten years. The first
five harvests include pooled data from the two chacras
because there were no chacra effects. Crop 28 (rice)
was affected by pathogen attacks on the grain. Corn
crops 29 and 30 showed nitrogen-deficiency symptoms
from tasseling to maturity. Crop 3 l's stand was
decreased to 26,700 plants/ha by an intense rainfall
(225 mm in six hours) immediately after planting. The

ECEC + Tillage to 20 cm
4.5 Initiated


2. 0 Trt. 1: Check E A
O Trts. 2, 3, 4: Complete
0 A Trts. 5, 6: Complete + Residues
E 1.0 1 Trt. 7:1.5 x Complete c o h g i

4 ( --C) Exch. Ca + Mg

2 -

123 130 136 142 148 154 160
Months After Clearing
Figure 1. Trends in effective cation exchange capaci-
ty (ECEC), exchangeable acidity and extractable Ca-
plus-Mg of soil on continuously cultivated plots in
Chacra I.

following observations can be gleaned from crop per-
formance, expressed as relative yields in Table 2.
Incorporating residues produced significantly higher
yields than residue removal in corn and rice, but not
on peanut (Table 2). Leaving residues as mulch had
a positive effect on corn, neutral on rice, and negative
on peanut. Peanuts are more susceptible to attack from
soil disease organisms such as Sclerotium rolfsii, which
causes southern stem rot, when residues remain in the

123 130 136 142 148 154 160
Months After Clearing
Figure 2. Trends in soil pH, % Al saturation and ex-
tractable soil P in soil on continuously cultivated plots
in Chacra I. The shaded bar at month 148 represents
the first tillage with tractor-mounted tools.


The "1.5 x complete" fertilization treatment pro-
duced significantly higher corn yields than others, sug-
gesting the need for N rates greater than 80 kg/ha
for this crop. In view of the deficiency symptoms
observed, nitrogen may have been the limiting nutrient.
Rice and peanut did not display this effect (Table 2).

Topsoil Properties
Figures 1 and 2 show relevant topsoil fertility in-
dices from January, 1982 to January, 1985, during
the growth of crops 25 to 30. Treatments 2, 3 and
4 were pooled as the "complete" because they were
similar with respect to measured soil properties.
Treatments 5 and 7 were likewise pooled to summarize
the effect of residue cycling.
The check treatment had the least desirable soil pro-
perties in all fertility categories. Returning crop residue
showed consistent and statistically higher fertility in-
dices except for P. Obviously, treatments 5 and 6 have
received less fertilizer P than treatments 2, 3 and 4.
The positive effects of residue return on physical pro-
perties were apparent each time the land was tilled.
Lime was applied to these plots at the rate of 2 t/ha
four years before the results shown in Figures 1 and
2. Soil-acidity indices were stable with less than 20%
Al saturation in treatments other than the check plot.
This observation indicates a considerable residual ef-
fect of lime.

Effects of Tillage to 20-25 cm
There was an abrupt change in most fertility indices
when the soil was tilled to 20 cm. Soil pH and available
P dropped while exchangeable acidity, percent Al
saturation and effective cation exchange capacity in-
creased. Exchangeable Ca + Mg remained stable,
however. These changes are a consequence of mixing
a less fertile and heavier-textured, 7-20 cm soil layer
with the 0-7 cm topsoil. The situation was beginning
to stabilize by the third crop under the 20 cm tillage
regime, because the lime and incorporated fertilizers
were then well mixed with a 20 cm plowed layer.

1) Returning crop residues, either by incorporation
or as mulch, had a slight positive effect on corn and
rice, but not on peanut.
2) An apparent response of corn yields to fertilizer
in the "1.5 x complete" treatment, together with
observed symptoms of N deficiency, suggest that N
fertilization should be increased for this crop, though
not for rice.

3) Most indices showed abrupt declines in soil fer-
tility immediately after the introduction of tillage to
20 cm and the mixing of subsoil with topsoil. Testable
fertility improved by the third crop, however, as lime
and fertilizers were more thoroughly mixed in the soil.

After 31 harvests in 12 years, yields of corn, rice
and peanut remain high by local standards. With
judicious use of lime and fertilizers, it is apparent that
acid soils in the humid tropics will produce acceptable
yields of short-cycle food crops under continuous
cultivation. It is also apparent that soil chemical pro-
perties can be improved while producing high yields.

Production Potential of Corn-Peanut
Intercrops in the Humid Tropics
Jos6 R. Benites, N.C. State University
Robert E. McCollum, N.C. State University
Andres Aznaran, INIPA

This experiment was conducted to compare the pro-
ductive efficiency of a corn-peanut intercrop with
monocultures (sole crops) of the intercrop components
grown in rotation and with continuous corn. A se-
cond objective was to determine the effect of nitrogen
fertilization on cropping-system efficiency. Corn (Zea
mays L.) and peanut (Arachis hypogea L.) were grown
in a "strip-intercrop" arrangement for three biological
cycles (trimesters). The strip-intercrop consisted of two
75 cm rows of corn in an alternating pattern with three
38 cm rows of peanut. This row arrangement permit-
ted the intercrop to be grown as a corn-peanut rota-
tion. Monocultures of the interplanted species, with
corn in 75 cm rows and peanut in 38 cm rows, serv-
ed as the reference standard (monocultre check) dur-
ing each cycle. These monoculture checks were also
grown as a corn-peanut rotation. The third cropping
system was continuous corn.
Corn in both monoculture and intercrop received
three levels of N fertilization (0, 100, or 200 kg N/ha),
and the experiment was arranged in a split-plot design
with four replications. Main plots were cropping
system, and subplots were N fertilization. The experi-
ment site was an Ultisol that had been limed and
phosphated before initiating the experiment.
All three cropping systems were planted with tractor-
mounted planters on the same date. Within-row
seeding rates were the same for each species in each


Table 1. Yields of corn grain and peanuts as influenced by applied nitrogen and cropping systems.
Cropping Species & Crop N Rates (kg/ha)
Systems Trimester duration 0 100 200

days yield, kg/ha
Strip Intercrop:
Corn/Peanuts: Corn (1) 108 779 1751 1602
Peanuts (1) 110 1034 782 795
Corn (2) 109 1806 1980 2297
Peanuts (2) 111 243 175 210
Corn (3) 119 1652 2295 2365
Peanuts (3) 122 357 299 322
Suequential monocultures:
Corn Peanuts -* Corn:
Corn (1) 108 917 2625 3218
Peanuts (2) 111 1086 1086 1086
Corn (3) 119 2595 3450 3185
Peanuts Corn -- Peanuts:
Peanuts (1) 110 2275 2275 2275
Corn (2) 109 3191 3800 3758
Peanuts (3) 122 859 1092 855
Corn Corn Corn:
Corn (1) 108 861 2757 3227
Corn (2) 109 2441 3590 3561
Corn (3) 119 2334 3106 3692
Trimester Trimester Trimester
1 2 3
Corn LSD 0.05 Cropping system 452 303 330
LSD 0.05 Nitrogen level 232 257 389
LSD 0.05 Int. CS x N 557 472 640
Peanuts LSD 0.05 Cropping system 544 184 226
LSD 0.05 Nitrogen level 56 36 119
LSD 0.05 Int. CS x N 547 188 260

system. While this methodology provided near-
equivalent total plant densities for each cropping
system, it should be noted that the corn-peanut inter-
crop had only one-half as many corn plants and one-
half as many peanut plants as its companion

Product Yields
Corn was virtually unaffected by its association with
peanuts (Table 1); the intercrop produced more than
60% of its reference monoculture during each cycle
(average relative yield of intercropped corn during three
cycles = 0.64). The N response was positive for all
cropping systems, and yields were near-maximal at 100
kg N/ha. There is some evidence that corn following
peanuts was less responsive to N than corn following
corn (Figure 1).

In contrast to corn, the yield of interplanted peanuts
was severely reduced by overstory corn. When averag-
ed over three cycles, the intercrop produced 31 % of
its monoculture check. Peanut plants were severely
affected by Cercospora leaf spot during the second cy-
cle, and the detrimental effect of this pathogen seem-
ed more pronounced in the intercrop.

Intercrop Efficiency
Table 2 shows the effect of N fertilization on area-
time equivalency ratios (ATER) during each cycle
(ATER= LER because intercrop duration equals
production-cycle duration for each species). When corn
was grown without N fertilization, the corn-peanut
intercrop used area and time more efficiently than
monocultures (ATER > 1.0) during two of the three
cycles. When corn was fertilized with 100-plus kg


Table 2. Effects of N fertilization to corn in a corn-peanut inter-
crop on Area-Time Equivalency Ratio during three biological

kg N/ha Cycle N
to corn 1 2 3 Mean


0 1.34 0.78 1.05 1.06
100 0.97 0.67 0.94 0.86
200 0.84 0.81 1.00 0.88
Cycle Mean 1.05 0.75 1.00 0.93
lATER relative to sole-crop corn and sole-crop peanuts in the corn-
peanut rotation.

Table 3.Effects of N fertilization and cropping systems on rate
of caloric yield and relative croppping-system efficiency.

Cropping N Fertilization (kg N/ha) CS
System 0 100 200 Mean

Caloric Yield
__ M cal/ha/year
Cont. Corn 16.36 27.44 30.42 24.74
Corn-Peanut 20.06 25.29 25.12 23.49
Corn-Peanut 20.34 23.68 24.71 22.91
N Mean 18.92 25.47 26.75

Relative Efficiency (Continuous Corn = 100)
Corn-Peanut 123 92 83 99
Corn-Peanut 124 86 81 97
N Mean 124 89 82

Second Crop Cycle Prior
Peanut 4

LSD .05



Third Crop Cycle
h crop'

-K- Corn

/ r Peanut
ILSD .05

- I

-. I

0 54 100 200 0 33 100 200
N Applied (Kg N/ha)

N/ha, however, there was no evidence of any inter-
crop advantage (ATER= < 1.0). Low ATER's dur-
ing the second biological cycle were due exclusively
to extremely low relative yields of interplanted peanuts
(relative peanut yields = 0.19 for cycle two versus
0.40 for cycles one and three). This observation lends
support to the earlier supposition that Cercospora leaf
spot was more severe in the intercrop.

Cropping-System Efficiency
Continuous corn was compared with the two corn-
peanut rotations by converting absolute yields to caloric
equivalents and summing over three cycles. Effects of
cropping system and N fertilization on the rate of
caloric yield (Mcal/ha/yr) are shown in Table 3. (All
entries in Table 3 are comparable because each entry
sums over three harvests of the relevant species). Im-
plications from the Table 2 data are quite clear: 1)
Without fertilizer nitrogen, the two systems that in-
clude peanuts are significantly superior to continuous
corn in storing energy; but 2) continuous corn with
adequate N (100-plus kg/ha) gives a higher rate of
caloric yield than the corn-peanut intercrop or
monocultures of its components grown in rotation.
If, in fact, caloric yield were the principal basis for deci-
sion making, the only near-equivalent system to well-
fertilized continuous corn would be continuous peanuts
(approximately 25 Mcals/ha/year, not shown).

A two-crop intercrop of corn and peanuts may have
merit under low-N regimes in humid tropical en-
vironments. In most such environments, however, the
inherent soil condition is high acidity (high ex-
changeable aluminum) and low phosphorus. These in-
itial conditions must be corrected with massive doses
of lime and phosphorus before either species can be
expected to produce the yields reported here.
There are some other favorable aspects to the in-
tercrop system described here:
1. Three cycles per year are possible because each
species can be grown year-round and because all have
nearly identical production-cycle durations.
2. Most field operations can be done with machines
or by hand.
3. Since the strip-intercrop system can be managed
as a corn-peanut rotation, there may be some nitrogen
carryover from the peanuts to the following crop of

Figure 1. Apparent N carryover from proceeding
peanut crops to subsequent corn crop.


Phosphorus, Zinc and
Copper Fertilization
Robert E. McCollum, N. C. State University
Luis Arevalo, INIPA
Andres Aznaran, INIPA

This project evolved after observations on several
Yurimaguas sites suggested that band-applied
phosphorus induces a micronutrient deficiency in corn.
The phenomenon was first observed in an experimental
field that had been limed and fertilized with N, P, K,
Mg, Cu and Zn at recommended rates. The first corn
crop had been machine-planted in January, 1984 with
TSP in the fertilizer hoppers. Within ten days after
corn emergence, virtually the entire planting had
developed a chlorosis symptomatic of Zn deficiency.
The possibility of a micronutrient toxicity was ruled
out because the rate of Zn or Cu applied was only
1 kg/ha.
In response to these observations, a project was
designed to meet the following objectives: 1) to deter-
mine the effect of banded phosphorus fertilizer on
growth and yield of corn on continuously cropped
Ultisols; 2) to explore the hypothesis that band-applied
phosphorus exacerbates zinc deficiency on low-Zn soil,
and 3) to determine if the nutritional abnormalities
induced by band-applied phosphorus can be
ameliorated by applying zinc or copper to the soil.
In July-August of 1984, eight plots (two contiguous
sets of four plots each) that had not received lime or
phosphate recently were selected for a "banded-P (BP)
by soil zinc" experiment. All plots received applica-
tions of lime (2.5 t/ha), phosphate (100 kg P/ha) and

copper (4 kg Cu/ha). Zinc variables were to have been
0, 2, 4, and 8 kg Zn/ha in 7.5 m by 12 m plots with
eight replications, but an error resulted in a double
application of P, Cu and Zn on one four-plot group
and a double application of lime on two of them. In-
stead of one experiment with eight replications, this
dosage error made it necessary to consider the "zinc
by banded-P" endeavor as two separate experiments:
1) a three-factor experiment (2 lime x 4 zinc x 2 BP)
with two replications, and 2) a two-factor experiment
(4 zinc x 2 BP) with four replications.
At the same time (July-August, 1984), a second con-
tiguous, eight-plot area that had received lime and
phosphate a year earlier was selected for a "zinc by
copper by banded-P" experiment. Copper was applied
at 0 and 4 kg/ha in factorial combination with 0, 2,
4 and 8 kg Zn/ha. All plots were plowed, disked, bedd-
ed, and rototilled. For the first corn crop (September,
1984 to January, 1985), the banded-P treatment was
achieved by planting four rows without banded P and
six rows with banded P. This P-banding procedure was
reversed for the following crop.

Crop Yields
Corn showed statistically significant, positive
responses to applications of banded P, broadcast Zn
and broadcast Cu in terms of grain yields and plant
populations (Tables 1 and 2), but there was no "Zn
x banded P" or "Cu x banded P" interaction. Plant
population was much lower in the second crop because
of intense rainfall (226 mm in six hours) immediately
after planting (Tables 1 and 2), and poor product yields
for this cycle are primarily the result of low plant

Table 1. Corn grain yields and plant population as affected by boradcast ZN and banded P application in two
consecutive crops
Corn Crop Zn Banded P Applied Banded P Applied
(Harvest Date) Applied No Yes Mean (Zn) No Yes Mean (Zn)
Yield, t/ha 1000 Plants/ha
Jan 85 None 2.42 3.08 2.75 36.0 38.2 36.8
Yes* 3.32 3.43 3.38 36.3 37.3 37.1
Mean (BP)** 3.09 3.34 36.2 37.6
July 85 None 1.79 2.01 1.90 20.5 25.5 23.0
Yes 2.28 2.59 2.44 24.3 27.1 25.7
Mean (BP) 2.16 2.45 23.4 26.7
Means of 2, 4 and 8 kg Zn/ha (36 plots) vs. 12 plots for 0 Zn.
** Weighted means.


Table 2. Corn grain yields and plant population as affected by broadcast Cu and banded P application in two
consecutive crops.
Corn Crop Cu Banded P Applied Banded P Applied
(Harvest Date) Applied' No Yes Mean (Cu) No Yes Mean (Cu)

kg/ha Yield, t/ha 1000 Plants/ha
Jan 85 0 3.06 3.27 3.17 33.6 35.0 34.3
4 3.72 3.71 3.72 34.8 37.7 36.2
Mean (BP) 3.39 3.49 34.2 36.3
July 85 0 1.80 2.51 2.15 20.1 25.0 22.6
4 2.39 2.52 2.46 23.3 24.6 24.0
Mean (BP) 2.09 2.52 21.5 24.8
1 Extractable Cu levels in the soil were 0.94 and 2.05 ppm Cu for levels 0 and 4 kg Cu/ha.

Banded P
Banding P increased corn yields by 8 and 13% in
the P-Zn plots (Table 1), and by 3 and 21% in the
P-Cu plots (Table 2), with an average overall effect
of 11%. While some of these apparent treatment ef-
fects were related to increased plant population, there
was a marked growth response to banded P, which,
unlike similar experiences in North Carolina, was
reflected in grain yields.

Zinc Response
Corn responded significantly to the first increment
of Zn (2 kg/ha). Since there was no significant response
to the second or third increment in applied Zn, all plus-
zinc treatments (2, 4, 8, or 16 Zn/ha) were pooled,
and the data analyzed as a minus-Zn versus plus-Zn
experiment. Data from all relevant plots in the "zinc
by copper by banded-P" experiment were analyzed
with the Zn-banded P data, and are included in Table
Zinc plus banded P applications caused a 42% yield
increase in the first crop with no effect on plant popula-
tion (Table 1). Zinc alone caused a 23% yield increase.
Zinc plus banded P and Zn alone caused a similar yield
increase in the second crop, and a 12% increase in
plant population.

Copper Response
Response trends for the "copper by banded-P" data
are highly positive. Incorporating 4 kg of Cu/ha in
the top 20 cm of soil caused a 6% population increase
during each cropping cycle but a 14-17% increase in
yield (Table 2).

Plant Population
This study underscores the difficulty of assessing
treatment effects in crops whose populations are
variable or inadequate because of such factors as heavy
rainfall, poor seed quality or inconsistencies in sow-
ing and thinning. A clear-cut estimate of treatment ef-
fects is confounded by the fact that additions of banded
P and micronutrients affect not only the growth and
yield of individual plants but also the number of plants
that emerge.
The positive relationship between band-applied P
and plant population has been apparent in virtually
every banding vs. no-banding comparison to date.
Figure 1 shows the relationship between plant popula-
tion and yield when the comparison was first made
(January, 1983 corn harvested from Chacra I, one of
the original experimental fields at Yurimaguas). Banded

4.4 ABanded P
Yield = -524 + 0.0988 x POP;
4.0 R- 2 = 0.25
ONo Banded P
'S Yield = 525 + 0.0628 x POP;
3.6 R2 = 0.36

c 2.8
2.4 -


24 28 32 36 40 44 48
Corn Population (x10-3)/ha
Figure 1. Relationship between corn population and
grain yield with and without banded-in-row P.


P not only resulted in a higher plant population at
harvest, but the slope of the yield vs. population curve
was also steeper. Yield versus population curves have
also been generated for other examples, and they show
the same trends for more surviving plants with band-
applied P. The slopes of the curves show that each
additional 1000 plants up to 50,000 plants/ha pro-
duces an additional 60 to 120 kg of grain. These
calculations also show that the corn population is usual-
ly inadequate. In no instance has there been a signifi-
cant quadratic response to population, and only rare-
ly have there been 50,000 plants/ha.

While it is too early to draw firm conclusions from
this study, preliminary observations suggest the
1) Corn yield and plant populations significantly
responded to banded P and broadcast Zn or Cu.
2) Although there has been no apparent effect of
banding P on the response of corn to Zn or Cu in
the first two crops (no measureable banded P x
micronutrient interaction), yields were low because of
reduced population density. Since plant population was
also treatment-related, the effects of treatment on pro-
duct yield are confounded with population effects.
3) While the observed chlorosis was corrected by
micronutrients, the initial hypothesis is still unproven.

Potassium, Lime and Magnesium
Interactions and Corn Yields
Rob Schnaar, Wageningen University
Robert E. McCollum, N.C. State University

A lime-by-potassium study was initiated at
Yurimaguas during July and August, 1984. The ob-
jectives of this study were 1) to construct a potassium
response curve for continuously cropped Ultisols in
humid tropical environments; 2) to estimate a critical
soil K level for corn in the soils; 3) to quantify the
recycling of K via crop residues in the soils; and 4)
to determine the effect of lime on K responses by food
crops, K utilization by crops, and K retention in the
soils. A between-site magnesium variable was introduc-
ed because two liming materials were used. While this
was not part of the original plan, site-related yield dif-
ferences provided some useful insights about cation
balance and Mg nutrition.
Two experimental sites were selected because the
soil represents the textural extremes for upland posi-
tions in the Yurimaguas environment (Table 1). At
site #1, the top 45 cm of soil is a clay loam and its
texture grades to clay between 45 and 60 cm. Site #2
has a transitional sandy loam-sandy clay loam sur-
face (0-20 cm) with little or no textural change to 60

Each site was cleared of secondary forest with a
bulldozer in 1980 and left to regrow without chemical
amendments. The sandy site was recleared and made

Table 1. Textural characteristics and effective cation exchange
capacity (ECEC) of two Ultisols used for lime-by-potassium
studies at Yurimaguas.
Range Site Sand Clay Textural ECEC
cm No. Class
% % meq/100cm3
0-201 1 40 31 Clay Loam (C1) 4.66
2 62 21 Trans. S1-SC1 2.69
20-30 1 35 37 C1 5.20
2 58 25 SC1 3.28
30-45 1 38 36 C1 5.80
2 52 29 SCi 3.63
45-60 1 36 42 C 6.86
2 52 32 SC1 3.89
1 Texture of surface layer determined after mixing to 20cm


tillable in mid-1983. It was then used for about one
year to screen germplasm (rice, cowpeas, corn) for
aluminum tolerance with a lime differential of 0 and
2.0 t/ha [lime source = Ca(OH)21 as main plots. The
lime was incorporated to 20 cm by routine tillage when
the site was acquired.
The clayey site was renovated by mowing and plow-
ing to 20 cm. The soil was sampled in detail. Initial
soil properties were: pH = 4.6; Ac = 3.78; Ca+Mg
= 0.89; Al satn. = 81%; K = 0.07., P = 3ppm.
A factorial experiment consisting of three rates of lime
and five rates of K was established. Lime levels (whole
plots) were chosen to neutralize 0, 50% and 100%
of the exchangeable acidity in the top 20 cm (0, 2,
and 4 t/ha). One-half of the intended lime dose
dolomiticc limestone) and one-half of the intended
blanket doses of P, Zn, and Cu (100, 8, and 4 kg/ha,

80 I Al Sat'n.

20 -
0 I

20 I


2 -

Exch. Acidity

2 -

Exch. (Ca + Mg)
0 -
0 1 2 3 4
Tons CaCO, equiv./ha
A Clay loam, limed Aug., 1984
0 Sandy loam, limed Aug. 1983

Figure 1. Effect of liming on two Yurimaguas Ultisols
on pH-related properties. Both sites sampled in
January, 1985.

respectively) were applied to the once-plowed soil and
incorporated by plowing again with a two-bottom
moldboard plow. After the second plowing, the re-
maining half of the various soil amendments [lime
source = Ca(OH)2] was applied and incorporated by
routine disking, bedding and rototilling.
Potassium treatments (sub-plots) were selected to in-
crease the K level in the top 20 cm of soil by 0, 0.05,
0.10, 0.15, or 0.20 meq 100/cm3 (0, 39, 78, 117,
and 156 kg K/ha). The potassium (as KC1) was hand-
drilled on the bedded rows and incorporated by routine
pre-plant rototilling.
Corn was machine-planted on each site in late
September of 1984 (harvested January, 1985).
Nitrogen fertilization (as urea) was 150 kg N/ha with
one-third of the total applied pre-plant and the remain-
ing two-thirds at about 40 days after corn emergence.
Total dry matter accumulation at early ear formation
(maximum K accumulation) was estimated by
harvesting and processing for analysis six whole plants
per subplot. The soil was sampled to 60 cm (0-20,
20-30, 30-45, and 45-60) at corn maturity and ana-
lyzed for relevant properties.
A second corn crop was planted in late March of
1985 with the same K additions as indicated for cycle
one, but excessive rainfall of high intensity resulted
in a low plant population as well as poor weed con-
trol, and no meaningful treatment-related yield data
were obtained. The following summary of treatment
effects on soil properties as well as product yield is
for the first biological cycle only.

Soil Properties
Two tons of lime applied to the sandy loam soil in
1983 had reduced aluminum saturation from 62% (pH
= 4.4) to 38% (pH = 4.7) when the soil was sam-
pled about 18 months later (Figure 1). On the clay
loam soil, aluminum saturation at about six months
after applying 0, 2, or 4 tons of lime was 68% (pH
= 4.3), 52% (pH = 4.5) and 34% (pH = 4.8),
respectively. Other pH-related properties followed
predictable trends (Figure 1), but there was no
measurable increase in effective cation exchange capaci-
ty (ECEC) due to liming.
Extractable soil potassium in the top 20 cm of soil
was a linear function of K applied on each site (Figure
2), but the steeper response slope for the sandy soil
shows that a higher percentage of the K applied was
recoverable by the extractant used. These results are
typical for soils of differing texture.


Product Yields
Grain yields on the sandy soil were less than 50%
of those measured on the clay loam soil (Table 2).
While there was a positive response to lime on each
site, the only measurable response to K was on the
clay loam soil, and grain yields on this soil were near-
maximal with 78 kg K/ha. Figure 3 shows the rela-
tionship between aluminum saturation and corn yield
when data from the two sites were pooled. Since ab-
solute yields from the two soils differed drastically,
a "relative yield" was first calculated (Relative yield
= 100 absolute yield/mean maximum yield at each
site), and relative yields were regressed on the percen-
tage of aluminum saturation. The data show that corn
yields are maximal when aluminum saturation is
around 30% of the cation exchange capacity. They
are therefore in close agreement with results from
similar studies on Utisols in southeastern U.S.
Since the clay loam soil was the only site with a
measurable K response, yield data from this experi-
ment were used to estimate the "critical" soil K level
for corn in this environment. Figure 4 shows that yields
were near maximal when extractable soil K was 0.12
meq/100 cm3. Given the fact that these were among
the highest corn yields ever recorded for upland posi-
tions at Yurimaguas, the data should provide the most
reliable estimate to date of the critical K level; but
similar data from succeeding corn crops and other soils
are needed to confirm this estimate.

K Applied (kglha)

0 0.05 0.10 0.15
K Applied (Meq/100 cm3 in surface 20 cm)

A Slope (Meq. extra. soil K) = 0.37 (t 0.04), R2 = 0.62
SMeq. K applied /
0 Slope Meq.extr. soilK = 0.87 ( 0.12), RW = 0.60
p Meq. K applied /
Dashed line shows approximate critical level of soil K (see Fig. 4)

Figure 2. Effect of applied potassium on extractable
soil K in two Yurimaguas Ultisols after one biological
cycle (six months).

Table 2. Effects of lime and potassium applications to two Yurimaguas soils on corn yield. Jan. '85 harvest.
Tons of Lime/ha

Soil Kg K/ha 0 2 4 K Mean
kg grain/ha
Clay Loam 0 3724 4525 4968 4406
39 3823 5409 5535 4922
78 3750 5460 6074 5095
117 4197 5587 5619 5134
156 3653 5810 5598 5020
Lime Mean 3830 5358 5559 4916
LSD (0.05); K = 48; Lime = 503; Lime x K = NS.

Sandy Loam 0 968 2653 1810
39 1889 2517 2203
78 1522 2443 1982
117 1672 2761 2216
156 1592 2794 2193
Lime Mean 1529 2634 2081
LSD (0.05); K = NS; Lime = 819; Lime x K = NS.


----- Plateau -- --------------- Quadratic --------------

Z-A Aa s


A .A

0 A

A Clay loam; Y max. = 6.07 tlha
O Sandy loam; Y max. = 2.79 t/ha

I I I I i i i


10 20 30 40 50 60 70
Al. + + Saturation (%)
Figure 3. Effect of Al saturation in two Yurimaguas
Ultisols on corn yield. Quadratic part: Relative yield
= 82.64 = 1.045 (AI ++ sat'n)2, R2 = 0.48. Solid
symbols: K = 0.

4---------- Quadratic ---------- ----------- Plateau ----.-----


^ ^

0.05 0.10 0.15 0.20
Extractable Soil K (meql100 cmrn3)
Figure 4. Effect of soil potassium in a heavy-textured
(clay loam) Yurimaguas Ultisol on corn yield (each
symbol is the mean yield of one to eight observations
at indicated level of soil K). Quadratic part: Yield =
2.65 + 49.66K 205.5K2. "Critical" K level = 0.12


Potassium Recycling
Whole-plant samples taken at early ear formation
were used to estimate the amount of K returned to
the soil in corn stover.
Since K accumulation by corn is maximal at ear for-
mation, any K not removed in the grain is returned
to the soil. Harvested corn grain was not analyzed for
K, but its concentration in mature corn seed is vir-
tually constant at about 0.30%; and this value was
used to estimate K removal. "Recycled" K was then
estimated as total plant K at ear formation minus K
removed in the grain, and a relationship between pro-
duct yield and recycled K was shown by least-squares
multiple regression (Figure 5).
Several features of Figure 5 merit special comment:
1) Data from the two sites could not be pooled because
the sandy loam produced as much vegetative dry mat-
ter as the clay loam but less than one-half as much
grain; 2) recycled K was a linear function of product
yield at each site, but the rate of K recycling was greater
on the low-yielding sandy loam soil-(recycled K (san-
dy loam) = 0.022 kg K/kg grain versus 0.016 kg
K/kg grain (clay loam))-because a smaller percentage
of silking-stage potassium was stored in grain; 3) recycl-
ed K is highly correlated with silking-stage dry mat-
ter; and 4) all of the Figure 5 data serve to emphasize
the point that nutrient cycling via crop residues is a
critical component of fertility maintenance.

Site #1 versus Site #2
With one exception (the exception being soil tex-
ture and other properties associated with texture), these
two experiments were supposed to be conceptually
identical. Obviously, they were not identical in prac-
tice, and it seems worthwhile to speculate on a pro-
bable cause for the two-fold yield difference between
the two sites. The clues point to a problem in cation
balance (Table 3).
Limed plots on the clay loam soil received a sizeable
dose of magnesium (120 or 240 kg Mg/ha) because
dolomitic limestone (12% Mg) was used. The sandy
loam soil, by contrast, was limed with Ca(OH)2 and
no magnesium was applied. By virtue of using two
different liming materials, widely differing soil chemical
environments were created on the two experimental
sites (Table 3-A), and silking-stage cation concentra-
tions, as well as concentration ratios in corn plants
(Table 3-B), are a direct reflection of the suite of
nutrient cations in the soil that produced them.
While the Table 3 data do not "prove" that the
sandy loam soil was deficient in magnesium, they do


show that limed plots on the clay loam site were well
supplied with this element, and several pieces of
evidence support the view that magnesium nutrition
was at least a part of the problem on the sandy soil:
1) In the unaltered state, each soil was "nutritionally
low" in Mg (0.18 and 0.12 meq/100 cm3 soil) because
Mg saturation of the exchange complex was less than
5% (5% Mg saturation is considered a "limiting value"
for many crops).
2) After liming with dolomite, extractable Mg as
well as percent Mg saturation in the clay loam soil
increased in direct proportion to the amount of Mg
applied; two tons of slaked lime on the sandy soil had
no effect on extractable Mg nor Mg saturation.
3) Two tons of dolomite on the clay loam lowered
the Ca:Mg ratio in the soil by 22%; two tons of slak-
ed lime on the sandy loan raised this ratio by 48%.
4) On the clay loam soil, whole-plant Mg concen-
tration in silking-stage corn increased in direct propor-
tion to extractable soil Mg. Corn grown on the san-
dy loam had less tissue Mg than the unlimed check
of the clay loam, and it was virtually unaffected by
5) Without lime, the ratio of Ca to Mg in corn tissue
was the same on both sites. This ratio was decreased
by liming the clay loam soil with dolomite; it was in-
creased when soil acidity in the sandy soil was neutraliz-
ed with Ca (OH)2.
6) On the sandy soil, there was a significant positive
effect of K fertilization on the K:Mg ratio in silking-
stage corn plants (K/Mg = 2.15 when K = 0., K/Mg
= 4.19 when K = 156 kg/ha); the effect of K treat-
ment of the K:Mg ratio was not measurable on the
clay loam.
7) There was a highly significant positive response
to K on the clay loam soil (Table 2); there was no
measurable response to K on the sandy soil.
None of these observations constitutes direct cause-
and-effect support for the Mg-deficiency hypothesis.
They do show, however, that the suite of cations in
the two soils and in the plants differed appreciably
because the practices followed were not comparable.
It is also clear that a high-lime, high-Mg clay loam
soil produced 5.5 tons of corn per hectare during the
same period that a high-lime but low-Mg sandy soil
was producing less than half as much. Furthermore,
the yield-component data of Table 3 suggest that the
sandy-soil problem was associated with pollenation and
grain filling. Each site produced comparable amounts
of silking-stage vegetative dry matter and would ap-
pear to have similar yield potentials. Yet the clay loam

100 -

E -80

60 '0



1 2 3 4 5
Product Yield (103 kg grain/ha)

Figure 5. Relationship between 1) corn yield and
silking-stage dry matter (open symbols), and 2) corn
yield and potassium recycled in corn stover (closed
symbols). Corn grown at the same time on differing
soils in the Yurimaguas environment. Data plotted
over range in product yield for each site.

soil produced nearly 50% more ears per plant (0.94
vs 0.64), and the harvested ears were more than twice
as large (208 vs 98 g/ear). The relatively minor dif-
ference in soil texture is not considered to be the cause
for the large differences in reproductive behavior. The
Mg-deficiency hypothesis is therefore the more plausi-
ble; it is supported, though indirectly, by the data at

In summary, the explanation for the widely differ-
ing corn yields between the clayey site and the sandy
site follows: Initially, the soil on each site was too high
in exchangeable aluminum and too low in bases (K,
Ca, and Mg) to produce corn. When the clay loam
was limed with dolomite, the Mg problem was resolv-
ed simultaneously with the acidity and Ca problems,
and corn responded to K fertilization as hypothesiz-
ed. Liming the sandy soil with Ca(OH)2 resolved two
of the initial problems (acidity and Ca), but it inten-
sified the inherent Mg deficiency.
This "imbalance" in nutrient cations was further ex-
acerbated by each increment in K fertilization, and corn
could not respond to K because inadequate Mg had
become the principal growth-limiting factor.


Table 3. (A) Properties of the soil on two sites used for lime-by-K experiments at Yurimaguas; and (B) some characteristics
of corn plants grown on each site under near-equal aerial environments.
A. Soil Property Site No. B. Plant Characteristic Site No.

1 2 1 2
1. Textural class (0-20cm) Clay loam Sandy loam 1. Harvested population (plants/ha) 36350 39560
2. ECEC (meq/100cm3) 4.66 2.79 2. Total dry matter at silking (Kg/ha) 5884 5528
3. Mg applied (kg/ha)1 3. Product yield (Kg grain/ha) 4916 2081
Lime (t/ha) = 0 0 0 4. Ears per plant 0.94 0.64
Lime (t/ha) = 2 120 0 5. Weight per ear (gm) 208 98
Lime (t/ha) = 4 240 6. [Mg] at silking (%, whole-plant)
4. Extractable Mg (meq/100cm3)2 Lime (t/ha) = 0 0.18 0.14
Lime (t/ha) = 0 0.18 0.12 Lime (t/ha) = 2 0.26 0.17
Lime (t/ha) = 2 0.36 0.13 Lime (t/ha) = 4 0.34 -
Lime (t/ha) = 4 0.58 7. [cation] at silking (meq/0lgmO)3
5. Mg saturation (%) Lime (t/ha) = 0 67 62
Lime (t/ha) = 0 3.8 4.3 Lime (t/ha) = 2 86 73
Lime (t/ha) = 2 7.7 4.7 Lime (t/ha) = 4 89 -
Lime (t/ha) = 4 12.4 8. Ratio: total cations/Mg, at silking4
6. Cal mg Lime (t/ha) = 0 4.75 5.52
Lime (t/ha) = 0 6.82 9.25 Lime (t/ha) = 2 4.14 5.41
Lime (t/ha) = 2 5.31 13.73 Lime (t/ha) = 4 3.36 -
Lime (t/ha) = 4 4.54 9. Ratio: Ca/Mg, at silking4
Lime (t/ha) = 0 1.10 1.13
Lime (t/ha) = 2 0.97 1.45
Lime (t/ha) = 4 0.83 -
10. Ratio: K/Mg, at silking4
K (Kg/ha) = 0 2.06 2.15
K (Kg/ha) = 39 1.88 2.51
K (Kg/ha) = 78 2.06 3.47
K (Kg/ha) = 117 2.20 3.52
K (Kg/ha) = 156 2.40 4.19
1 One-half of the limestone applied to site -1 was dolomitic (12% Mg); site -2 was limed with Ca (OH)2.
2 All soil chemical data are based on samples taken at crop maturity.
3 [cation] = summation of Ca, Mg, and K (chem. equivalents).
4 Ratios are as chemical equivalents.


Weed Population Shifts Under
Continuous Cropping Systems
Jane Mt. Pleasant, N. C. State University
Robert E. McCollum, N. C. State University

In traditional slash-and-burn agriculture, fields are
abandoned as weeds begin to dominate food crops,
and a forest fallow is the primary agent in weed con-
trol. Stable continuous-cropping systems, however,
could be expected to require a comprehensive program
of weed management, probably including the use of
chemical herbicides. The objective of this project was
to test the following hypotheses:
1) A given set of weed-control measures, if practic-
ed over time, will cause a change in the spectrum of
weed species. With intensive chemical control, a few
species will become dominant, requiring new control
2) Effective weed-management programs can be
devised for high-input, continuous-cropping systems
in the Amazon Basin.
A split-plot experimental design was used in a rice-
corn-soybean-rice-corn rotation. Weed-control prac-
tices in rice were the main plot treatments; methods
of weed control in corn and soybeans represented split-
plot treatments. In rice, the herbicides used were pro-
panil and oxadiazon; in corn, metolachlor, and in soy-
beans, metolachlor, sethoxydim, and bentazon. In the
second year of the experiment, a no-till treatment was
introduced in which paraquat was used to kill existing
vegetation. In all crops, hand-weeded and check
treatments (no weed control) were also included.

Analysis of data from this experiment has not yet
been completed. Therefore, only preliminary conclu-
sions and observations will be given.
Grassy weeds are by far the most important weed
problem in continuously cultivated, short-cycle food
crops (Table 1). Rottboelia exaltata, an annual grass, is
potentially the most noxious weed (Figure 1). Rott-
boelia cannot be controlled in corn except by hand
weeding, and controlling it in rice and grain legumes
requires a large herbicide input. In corn and soybeans,
the continued use of metolachlor alone eliminates all
weeds except R. exaltata, which establishes pure stands
among the crops.

60 -

50 -

40 -

30 -

20 -

A---6--A dry wt. g/m2
- -0-- % plots infested

I II 1
Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5
Figure 1. Trends in the level of infestation by R. ex-
altata during five production cycles. Yurimaguas,



100 M
80 0

60 (


Table 1. Weight and composition of weed population during five production cycles.
Total dry wt* BLW** Grasses Other Monocots
Cycle/Crop g/m2 % of total
1/Rice (11-83)*** 188 20 52 15
2/Corn (3-84) 46 15 70 15
3/Soybean (9-84) 128 27 69 4
4/Rice (1-85) 266 14 63 23
5/Corn (6-85) 178 6 79 15
Data are an average of four or five weed control treatments per crop, which included hand weeded,
chemcial control and no weed control.

** BLW = Broad leaf weeds

*** Numbers in parentheses are date of weed sampling. In each case samples were taken near crop


Cyperaceae species are not competitive with either
crops or grassy species. In addition, they are readily
controlled with most of the herbicides in use at the
station. The majority of the cyperaceae found in
Yurimaguas are annuals, rather than nutlet-forming
perennials. This probably accounts for their ease of
control. With few exceptions, broad-leaf species are
not important weeds.
Crop species have been observed to differ greatly
in their ability to compete against weeds. Lack of weed
control in upland rice often means crop failure. Corn,
however, appears to be far more competitive. Yields
may be reduced without weed control, but they are
greater than zero.

Preliminary Conclusions, Implications
Even though data are still being analyzed, direct
observations over the course of this experiment strongly
suggest the following: Weeds can be controlled in in-
tensively managed, short-cycle food crops in this en-
vironment, but the cost is likely to be high. With the
products and rates used in this experiment, the average
price of chemical control is approximately $100/ha.
This is not economical control within the present
price/profit structure in Yurimaguas.
Observations during this study also suggest that
upland rice should be removed from the high-input
cropping system, unless yields can be significantly in-
creased to offset the high cost of weed control. Crops
more competitive against weeds, such as corn and grain
legumes, offer a broader range of weed-control options,
and may therefore be more practical than rice in the
high-input system.
This experiment will be continued for at least two
more cropping cycles, with corn replacing rice in the

Chemical Weed Control in Corn
Jonathan Lopez, INIPA
Jane Mt. Pleasant, N. C. State University
Robert E. McCollum, N. C. State University

In the selva, chemical weed control in corn may be
practical when hand labor is scarce or expensive. There
are several herbicides available in Peru for controlling
a broad spectrum of weeds in corn. In Yurimaguas,
where the primary weed problem is grasses,
metolachlor has generally given good control.
Metolachlor is a preemergence herbicide effective
against grasses. It also controls a large number of broad-
leaf weeds. The objective of this project was to deter-
mine whether other herbicides such as atrazine, used
alone or in combination with metolachlor, might im-
prove weed control, or expand the spectrum of species
controlled, and thereby improve corn yields.
Atrazine was selected for the experiment because
it has been used extensively in temperate regions. Ap-
plied together, atrazine and metolachlor control a much
broader spectrum of weeds than does either alone. In
some treatments, metolachlor was also combined with
other herbicides effective against broad-leaf weeds.
Corn was grown for two cycles using eight weed-
control treatments in a randomized complete block
design: Hand weeding; no control; atrazine pre (2.25
kg/ha); metolachlor pre (2.25 kg/ha); metolachlor pre
+ bifenox pre (2.25 + 1.25 kg/ha); metolachlor pre
+ atrazine pre (1.75 + 1.55 kg/ha); metolachlor pre
followed by 2, 4-D post (2.25 + .30 kg/ha), and
metolachlor pre followed by bentazon post (2.25 +
1.0 kg/ha).
Weeds were counted by species in each plot seven
weeks after planting. Prior to corn harvest,
aboveground dry-matter weights of weed species were
recorded. Corn yields on all plots were very low, rang-
ing from 1754 to 2349 kg/ha, due to poor germina-
tion and irregular stands. As a result, there were no
significant differences in yield among treatments.

Weed Counts
Weed counts taken early in the season indicated large
differences among treatments. Check plots (those with
no weed control) had 874 weeds/m2, while treatments
with chemical control or hand weeding ranged from
16 to 109 weeds/m2. Grass species were the domi-
nant weeds in all plots, accounting for 47-87% of the
total weed population (Table 1). Three grass species,
Axonopus compressus, Digitaria sanguinalis, and


Table 1. Effect of weed control treatment on number of weeds and composition of weed population in corn
seven weeks after planting.
Treatment Total Weeds All Grasses All BLW+ All Cyperac. Commelina
No. Plants/m2 % of total weeds
1. Hand weed 109 47 41 6 5
2. No control-check 874 53 34 10 2
3. Atrazine 88 84 1 11 3
4. Metolachlor 56 61 14 23 2
5. Metolachlor + bifenox 31 52 3 33 16
6. Metolachlor + bifenox 16 56 0 44 0
7. Metolachlor + 2, 4-D 42 67 0 31 2
8. Metolachlor + bentazon 46 87 7 4 2

+ BLW = broad leaf weeds
Level of significant main
effects ** ** ** ns ns
Species within grasses
A.compressus ns
D.sanguinalis **
P.paniculatum *

Paspalum paniculatum, represented 75-96% of the grass no weed control had significantly more weeds than
weeds in all treatments. other treatments. There was no difference between
Cyperaceae species and commelina species were not hand-weeded plots and those receiving chemical con-
present in significant numbers in any treatments, while trol. Metolachlor alone was equal to or better than
broad-leaf weeds were found in large numbers in on- all other treatments in controlling both the total
ly two treatments, hand-weeded and check. In both number of weeds and also the different components
cases Lindernia bumilis comprised more than 85% of of the weed population. No additional control of any
the broad-leaf population. This low-growing weed is weed group was obtained by the use of bifenox,
uncompetitive with cultivated plants and is of little atrazine, bentazon, or 2, 4-D with metolachlor. In ad-
importance. edition, the data showed that atrazine alone did not
Planned comparisons were used to determine which control D. sanguinalis. When atrazine alone was com-
of the treatments were responsible for the significant pared with metolachlor alone or metolachlor plus
differences identified by the F test. Single degree-of- atrazine, plots with atrazine alone had a significantly
freedom comparisons for relevant treatments are shown greater number of D. sanguinalis.
in Table 2. For all categories of weeds listed, plots with

Table 2. Tests of significance for differences in weed counts for planned treatment comparisons.
Level of Significance
All All All P. D.
Contrast Weeds Grasses BLW Paniculatum Sanguin.
1. Handweed vs. all chemical trts. ns ns ns ns ns
2. No control vs. all others ** **
3. Metolachlor vs. atrazine ns ns ns ns
4. Metolachlor vs. (metol. + atrazine) ns ns ns ns ns
5. Atrazine vs. (metol. + atrazine) ns ns ns ns
6. Metolachlor vs. (metol. + bifenox) ns ns ns ns ns
7. Metolachlor vs. (metol. + 2,4-D) ns ns ns ns ns
8. Metolachlor vs. (metol + bentazon) ns ns ns ns ns
ns = Not significant; ** = Highly significant


Weed Weights Before Harvest
In all treatments, the poor stand of corn provided
little or no competition to emerging weeds, which at-
tained heavy growth by late season. Total dry weights
ranged from 147 to 366 g/m2. Grasses were the domi-
nant species, comprising 83 to 98% of the total weed
weight. Two species, A. compressus and D. sanguinalis,
comprised 65-98% of the grass population. Broad-leaf
weeds, cyperaceae species and commelina species were
unimportant components of the weed population.
Since A. compressus was the only weed affected by
treatment, single degree-of-freedom contrasts were
made in order to determine which treatments were
relevant to its control. Plots with metolachlor alone
had much higher weights of this species than plots with
atrazine alone or atrazine plus metolachlor. All
chemical-control treatments also had a much higher
weight of A. compressus than did hand-weeded plots.
The data indicated that atrazine controls A. compressus
while metolachlor does not.

Despite the lack of corn-yield response to weed-
control treatments, two conclusions can be drawn from
the first cycle of this experiment:
1. Use of metolachlor plus atrazine increased the
spectrum of weeds controlled compared to either her-
bicide alone. Atrazine alone did not control D.
sanguinalis early in season, while metolachlor alone
failed to control A. compressus later in the season.
2. Broad-leaf weeds and non-grass monocots such
as cyperacea sp. and commelina sp. are not important

components of the weed population. It is unlikely that
there would be any benefit in using additional her-
bicides in combination with metolachlor to control

D. sanguinalis is an important species in most fields
at the station. Research in temperate regions has shown
that atrazine has little effect on this grass. Confirming
this information under Yurimaguas conditions enables
the development of more effective weed-control
measures for corn. A. compressus, a perennial grass, has
not been an important weed species in short-season
food crops in Yurimaguas. An additional cycle of this
experiment is required to determine A. compressus'
resistance to metolachlor and its importance as a weed
in corn.
Weeds are a critical factor in continuous cropping
systems. They are as important in limiting yields as
soil fertility while their management is considerably
more difficult. Two years of research has shown that
the weed populations will change in response to
chemical control practices. Species resistant to her-
bicides dominate with time, and their control becomes
increasingly difficult and expensive. At present,
chemical weed control represents an enormous
economic input; it averaged $100/ha for upland crops
in Yurimaguas. Hand weeding is often much cheaper,
but in many cases labor is simply not available. Con-
tinued research effort will be required to develop weed
management practices that are agronomically effective
as well as economically viable.


Technology tested at Yurimaguas for sustained irrigated rice production in fertile alluvial soils of
the Amazon has been validated and is now being transferred to producers through Peruvian extension
programs. Peru recently became self-sufficient in rice, due in part to the expansion of flooded-rice
agriculture into the Amazon. Several general principles of Amazon flooded-rice production have been
established, including the following:
1) Land can be cleared by slash and burn or by bulldozing, as the usually detrimental effects of
soil compaction by bulldozers do not seriously affect paddy rice. Care must be given not to displace
topsoil during the land-leveling operation.
2) Supplemental irrigation every two weeks increases yields by about 50% as compared to yields
from crops dependent entirely on rainfall. The source of water may be gravity canals or pumping
from rivers.
3) Transplanting provides higher yields than broadcasting seeds for the first two crops due to insuffi-
cient leveling. Broadcasting pregerminated seeds is highly advantageous after the paddies are adequately
4) Fertilization will be minimal. No significant responses to N or P fertilization have been observed
during eight consecutive crops grown in a four-year period. N deficiencies are expected to appear
with continuous use.
5) A combination of herbicides provides satisfactory weed control.
6) Two crops a year with recommended short-statured varieties can produce annual yields of 12
to 15 ton/ha or 5.2 to 6.6 t/ha/crop. Considering that one hectare of acid soils must be cleared
every year to produce one ton of upland rice, every hectare under irrigated rice production might
save from 12 to 15 hectares of tropical forests annually from deforestation.
Research and extension activities are now the responsibility of INIPA's National Rice Program.
The work reported here is conducted to further test and refine this technology for long-term flooded-
rice production.


Intensive Management of Alluvial Soils
For Irrigated Rice Production
Luis Arevalo, N. C. State University
Robert E. McCollum, N. C. State University
Jos6 R. Benites, N. C. State University
Alfredo Rachumi, INIPA
Cesar Tepe, INIPA
Kristinaa Hormia, Institute of Development
Studies, Finland

Research on this important management option has
progressed to the point of widespread technology
transfer, contributing to a 40% increase in rice pro-
duction on fertile, alluvial soils of the Amazon Basin
of Peru. The objectives of this project, which was con-
ducted at the Yurimaguas Experiment Station, were
1) to determine the best methods of planting to achieve
maximum yields in paddy rice; 2) to determine the
best fertilizer sources, schedules and rates; 3) to deter-
mine optimal irrigation frequency, and 4) to determine
the effect of water-level fluctuation on the survival of
paddy-rice seedlings.

Transplanting vs. Direct Seeding
A project was initiated in August, 1981 with the
objective of determining the best planting methods in
paddies newly developed on an Eutric Haplaquept
clayeyy, mixed, isohyperthermic), on a high terrace near
the Shanusi river at Yurimaguas. The results shown
in Table 1 indicate that annual mean production was
only slightly less with direct seeding than with
transplanting. Two new experiments were initiated in
1985 to develop seeding methods and weed-control
practices for direct-seeded rice. The first found that
there was no significant difference in yields between
crops broadcast-seeded by hand and those broadcast

with a Cyclone-type seeder. One man can seed 1.5
to 2.0 ha per day by hand, but the same person can
plant 5.0 ha in one day with the Cyclone seeder .
In the second experiment, three types of herbicides
were tested with direct-seeded rice. Ten days after
seeding, the thiobencarb and oxadiazon treatments pro-
duced harmful effects. The check plots gave 100% ger-
mination, but all the other herbicide treatments gave
only 30 to 50%. The effects on yields are shown in
Table 2. Using 2-4 D amine alone at a rate of 2.0 L/ha
resulted in a 20% higher yield than the check plot and
performed better than the other herbicides. The low
yields obtained in this experiment are probably due
to herbicide toxicity and an attact of molluscs Arion,
the two together affecting initial plant growth severely.

Nitrogen and Phosphorus Fertilization
After eight consecutive rice crops, there have been
no significant responses to either fertilizer N at rates
up to 200 kg/ha or to P at rates up to 100 kg
P205/ha. In the N experiment, mean grain yields
were in the range of 6 to 7 t/ha for all treatments.
In the P experiment, mean grain yields were also
around 6 t/ha regardless of P rate or P source.

Supplementary Irrigation
The paddy-rice production system developed by the
project includes supplemental irrigation. Table 3 shows
that the best yields were obtained with supplemental
irrigation once every two weeks, as compared with
rainfall dependency. Table 4 shows the effect of dif-
ferent water depths on crop yields through three
harvests. The data indicate that highest yields were
obtained when water depth was maintained between
10 and 20 cm. At higher and lower levels, yields tend
to decrease.

Table 1. Performance of flooded IR4-2 rice in different land preparation systems in an Eutric Haplaquept
at a restingg" in Yurimaguas, during the first 26 months after clearing.
Land Planting First Second Third Fourth Fifth Mean per annual
preparation system crop crop crop crop crop crop production1
Grain Yields, ton/ha
Puddled: Transplanted 7.9 5.2 7.1 6.0 6.8 6.6 15.2
Broadcast/direct-seeded2 3.2 4.9 6.4 4.8 6.7 5.2 12.0
Dry: Transplanted 8.3 6.7 6.2 5.6 6.3 6.6 15.3
Broadcast/direct-seeded2 6.3 5.6 4.9 4.6 6.0 5.5 12.6
1 Assuming 2.3 crops per year
2 Hand-weeded


1) Direct-seeded paddy rice produced mean annual
yields only slightly lower than those from transplanted
rice. Direct seeding eliminates pre-plant soil puddling
and thereby reduces labor costs.
2) Broadcast-seeding rice with a cyclone-type seeder
was as effective as broadcasting by hand, and required
less than half the manpower.
3) Of the three herbicides tested for use in broadcast-
seeded rice, 2-4 D amine used alone gave the best
4) Best rice yields were obtained with supplemen-
tal irrigation once every two weeks, which maintain-
ed the water level between 10 and 20 cm.

Table 2. Effects of week-control methods on direct-
seeded IR4-2 paddy rice.
Herbicide Rate, L/ha Rice Grain Yield, t/ha
Thiobencarb 7.0 + 3.0 4.64
2-4 D amine

Thiobencarb 8.0 + 2.0
2-4 4 D amine

2-4 D amine

2-4 D amine

2-4 D amine

2.0 + 3.0

4.0 + 2.0


Check --- 3.90

Recommended varieties of paddy rice can produce
yields in the range of 12 to 15 ton/ha/yr on alluvial
Amazon soils without fertilization for the first three
years. Direct seeding with a Cyclone-type broadcaster,
coupled with judicious herbicide use, can save substan-
tially on labor, an important factor because skilled
laborers are scarce in this area. Despite the heavy rain-
fall in the Amazon, supplemental irrigation from rivers
or ponds every two weeks improves rice production
50%. These results are being tested in farmers' fields
in the Tupac Amaru settlement near Yurimaguas.

Table 3. Effect of the irrigation frequency on the rice yield for
cultivar IR4-2.
Irrigation Number of Harvests
Frequency 1st 2nd 3rd Mean

Once/2weeks 5.78 6.74 6.00 6.17
Rainfall only 4.08 5/13 3.99 4.40

Table 4. Grain yield as affected by different water levels. Rice
variety IR4-2.
Number of Harvests
Water Levels 1st 2nd 3rd Average
cm L/ha
0 5.48 3.80 5.13 4.80
10 6.77 5.03 5.44 5.75
20 6.45 5.18 6.18 5.94
30 5.66 4.92 4.48 5.02


Knowledge of the properties and distribution of soils in the humid tropics serves as the basis for
soil management. Proper selection of sites for extrapolation work requires good soil characterization
and classification by Soil Taxonomy, as well as interpretations practical in agronomic terms. The Fer-
tility Capability Classification (FCC) system is being adopted in many areas of the world as a basis
for research planning and technology transfer because it helps identify the soil characteristics that af-
fect crop production. Coupled with Soil Taxonomy, the FCC system is an effective tool in the develop-
ment of soil-management technologies adapted to specific sites and conditions.
An example of the widespread applicability of this approach is an FCC map of Africa, being developed
by the Food and Agriculture Organization of the United Nations (FAO). By providing planners with
an inventory of Africa's soils characterized by their productive potential, the FAO's map will assist
in long-range agricultural research and development. N.C. State University collaborators on this pro-
ject are developing software for personal computers that will allow soil maps to be digitized, so that
users might have at their fingertips maps with any soil grouping desired. The work described here
concentrates on two general areas. The first is the adaptation and refinement of the FCC system for
various tropical ecosystems and crops. The second is the study of soils in several humid tropical regions
on the agricultural frontier, where soil characterization lays the groundwork for future research and


FCC Adaptation to Wetland Soils
Pedro A. Sanchez, N. C. State University
Stanley W. Buol, N. C. State University

The Fertility Capability Classification system (FCC)
has been tested on many sites around the world in
order to adapt it to various soils and conditions. In
each case, the primary aim has been to identify soil
constraints to crop production and to guide decisions
about how to relieve or offset these constraints. At
the request of the International Rice Research Institute
(IRRI) and Soil Management Support Services (SMSS),
the FCC system was applied to soils with aquic soil-
moisture regimes in order to relate soil classification
with soil-productivity parameters that are important
for flooded-rice production.
Interpretations for FCC soil types and substrata types
for wetland soils are shown in Table 1, and for con-
dition modifiers in Table 2. The FCC system iden-
tified specific soil characteristics directly related to most
of the physiological disorders of rice, except for iodine
and boron toxicity. Iron toxicity caused by Fe-rich in-
terflow from adjacent uplands requires an FCC
classification of such upland soils. One additional con-
dition modifier was necessary to include in the FCC:
a g' modifier for constantly flooded soils.

The FCC interpretations of aquic soils were tested
by workshop participants from the International Net-
work on Soil Fertility and Fertilizer Evaluation for Rice
(INSFFER) and Soil Management Support Service
(SMSS), during a five-day field trip in Central Luzon,
Phillipines, where 16 profiles were examined. Infor-
mation from these profiles was related to the condi-
tion of rice plants growing on adjacent plots established
for INSFFER fertilizer trials. The FCC system was
successful in predicting Zn deficiency by the presence
of either the b (calcareous) or the g' (prolonged
flooding) modifiers. Two other characteristics, ease of
puddling and difficulties in regenerating the puddled
structure for rotation with upland crops, were also
readily identified by FCC classes. The possibility of
low N fertilizer efficiency indicated by the v (vertic)
and b (calcareous) modifiers was confirmed by the
results of the INSFERR trails. Table 3 shows the Soil
Taxonomy and FCC designation of the pits studied
and the fertility problems encountered.

The workshop's soil-fertility group recommended
the following in relation to FCC:
1. The FCC system should be tested and applied
to wetland rice soils as a means for grouping together
soils with similar constraints.

Table 1. Interpretations of FCC type and substrata types for rice cultivation in aquic soil moisture regimes.
High in infiltration, low water-holding capacity, more difficult to do thorough puddling; traffic pans infrequent; relatively
easy to regenerate structure for rotation with other crops. Level of management (nutrients and water) required for
high rice yields is higher than in L or C soils.

Medium infiltration, medium water-holding capacity, usually easy to puddle (except Lx), and medium difficulty in
regenerating structure. Traffic pans important in these soils except for Lx. L soils are generally more productive
for rice than S soils and less than C soils, provided condition modifiers are similar.

Low infiltration rates, high water-holding capacity (except Ci), easy to puddle and difficult to regenerate previous
structure (except Ci); traffic pans not common. Generally higher productivity for rice than L or S soils provided con-
dition modifiers are similar.

Deep organic or peat soils, with little to no potential for rice production.

OC, OL or OS
Shallow organic soils with a mineral layer of less than 50 cm depth. Potential for rice production

Somewhat better water-holding capacity and thus better suitability for rice production than S soils.


Table 2. Interpretations of FCC condition modifiers for rice cultivation in aquic soil moisture regimes.
When only one modifier is included in the FCC unit, the following limitations or management requirements apply
to the soil. Interpretations may differ when two or more modifiers are present simultaneously or when textural types
are different.

Defines wetland soils. Preferred moisture regime for rice cultivation.

Prolonged submergence causes Zn and perhaps Cu deficiency.

Topsoil moisture limited during dry season unless irrigated. Generally only one rainfed rice crop can be grown a
year. Irrigated rice during the dry season has higher yield potential and responds to higher N rates.

Low inherent fertility because of low reserves of weatherable minerals. Management levels higher than in soils without
this modifier. Potential K deficiency depending on base contents of irrigation water.

Low ECEC reflects less gradual N release, more exacting N management. Identifies degraded paddy soils with SLa
or LCa and low organic matter contents. If so, potential H2S toxicity can occur if (NH4)2SO04 is used as N source.
Potential Fe toxicity if adjacent uplands have Fe-rich soils.

Aluminum toxicity will occur in aerobic layers. Soil test for identifying P deficiency recommended.

Potential P deficiency under continuous rice cropping. Otherwise optimum aerobic pH for flooded rice production.
If combined with SLe or LCe, potential Si deficiency.

High pH may induce Fe deficiency when aerobic, and Zn deficiency when water-logged. High N volatilization loss
potential from broadcast N applications. NH4+ fixation by 2:1 clays possible. Mollusk shells indicative of Zn

High P fixation by Fe; P deficiency likely; Fe toxicity potential; soils difficult to puddle and will regenerate original
structure rapidly. Inter-flow from Ci uplands may cause Fe toxicity to e soils with lower topographic position.

Volcanic materials indicate high inherent fertility with no potential Si deficiency; N and P deficiencies common and
soil may fix large quantities of P; soils difficult to puddle and will regenerate original structure rapidly.

Soils will shrink and crack when dry, causing excessive percolation losses afterwards. Easy to puddle but difficult
to regenerate structure. P deficiency suspect and should be determined by soil tests. Soil fixes applied NH4+ and
releases it later to the rice crop a positive attribute. Cracks may not close after reflooding due to ripening-water
percolation and additional N losses.

Defines saline soils. Drainage needed but must consider conductivity of irrigation water.

Defines alkali soils. Reclaiming with drainage and gypsum applications may be needed.

Acid sulfate soils causing Fe and S toxicity when anaerobic and Al toxicity when aerobic. Depth at which c modifier
occurs determines feasibility of rice production. Strong P deficiency likely and Al toxicity when aerobic.

Presence of gravel limits land preparation and water holding capacity.

Skeletal soils with limited potential for rice production.

The higher the % slope, the narrower the paddies will

2. The FCC modifier for acid-sulfate soils (c) needs
further refinement to establish a better limit. An ad-
ditional modifier for cation imbalance ratios (r) should
be developed; additional modifiers for high organic
nitrogen in the topsoil (q), and for high available native
topsoil phosphorus (p), should be investigated and in-
corporated into the system if reliable quantitative limits
can be identified.
3. Field trials in rice fertility and soil management
should have the soil classified according to Soil Tax-
onomy at the family level. Emphasis should be given
to mineralogy characterization, in relation to fixation
and release mechanisms. FCC should not be considered
an alternative to Soil Taxonomy, but as a technical
system that facilitates its interpretation for agronomic
4. A Hydrological Capability Classification (HCC)
system should be developed along similar lines as FCC

be and the higher the rise between terraces will be.

to characterize in a systematic and quantitative basis
extremely important factors such as: 1) water-table
depth during dry and wet seasons, 2) frequency, depth,
speed, and duration of natural flooding, 3) quality of
irrigation, flood or ground water and other relevant
hydrological parameters. A working group should be
established to develop the HCC. Hydrological con-
straints often override soil constraints in rice produc-
tion. The development of this technical system is,
therefore, considered an urgent matter.
In addition, the INSFFER network meeting con-
cluded that Soil Taxonomy is to be used to characterize
network sites and that the FCC system is to be tested
by all countries participating in the network, which
are: Burma, Bangladesh, China, India, Indonesia,
Malaysia, Nigeria, Nepal, Pakistan, Philippines, Sri
Lanka, Thailand, and Vietnam. Tests are under way
in many of these countries.

Table 3. Field testing of FCC system in 14 pedons of Luzon, Philippines, and nutritional deficiencies observ-
ed in adjacent INSFERR trials.
Zn P N Mg/K
Taxonomy FCC deff response inefficiency imbalance
Andaqueptic Fluvaquent CLg' X
Vertic Haplaquoll Cg'v X
Vertic Tropaqualf Cgh X
Udorthentic Pellustert Cgdv X X
Aeric Tropaquept Cg
Entic Pellustert Cgdhv X X
Aeric Tropaquert LCgd
Fluvaquentic Haplustoll CLdbg X
Entic Chromustert Cdvbg X
Andic Palehumult Lkax no trial
Typic Haplaquoll Lgdbk X X
Cummulic Haplaquoll LCgdbk X X
Typic Tropaquept Lgd
Aquic Ustifluvent LSdeh


New-Project Update
Several projects in this series have not been under
way long enough to yield substantive reports, but should
be mentioned because of their importance to the pro-
gram as a whole.

Volcanic Ash Influence on
Transmigration Areas of Sumatra
Hardjosubroto Subagjo, Center for Soil
Stanley W. Buol, N.C. State University
John R. Thompson, University of Hawaii
Michael K. Wade, N.C. State University
Mohammed Sudjadi, Center for Soil
I. Putu Gedjer, Center for Soil Research
Agus B. Siswanto, Center for Soil Research

The objective of this project is to determine the
amount, thickness and effect on phosphate chemistry
of amorphous material in the major soil and geographic
areas around the Sitiung-Bangko transmigration set-
tlements, on the island of Sumatra. The western part
of Sumatra belongs to the Bukit Barisan mountain
range, where faulting and folding have been accom-
panied by volcanism. Along the foot of these moun-
tains lies a vast, undulating and rolling plain.
Two transects extending from the lowland to
volcanoes were studied, along with sites intermediate
to them in the transmigration areas. Profiles were
described and soils were sampled to a depth of 2 m
at each site. The tentative classification of these soils
is generally Paleudults or Haplorthox at the lower
altitudes (40-160 m), Topudults at the middle altitudes
(300-350 in), and Dystrandepts or Hydrandepts on
the lower slopes of the volcanoes (1200-1350 m). As
this project continues, soils will be further characterized
and analyzed at the Center for Soil Research and at
N.C. State University.

FCC and Site Characterization
In Relation to Caribbean Pine
Leon H. Liegel, USDA Southern Forestry
Station, Puerto Rico
Stanley W. Buol, N.C. State University
Robert E. Hoag, N.C. State University
Pedro A. Sanchez, N.C. State University

The purpose of this project was to evaluate the Fer-
tility Capability Classification system (FCC) in rela-
tion to an important commercial tree crop, Caribbean
pine, and to determine if the system needs modifica-
tion for use with perennial tree crops. To date samples
have been taken at 46 sites in Venezuela, 44 in Jamaica,
and 29 under Caribbean pine (Pinus caribaea, var.
honurensis). These samples were analyzed for particle-
size distribution, pH value, extractable Al, Ca, Mg,
K, and P. Brief profile notes were made at each site.
Using these data, each site was classified by FCC
A preliminary summary of the data reveals that the
majority of the Venezuela sites were coarse to medium
in texture, types S, L, or SL, and had high Al concen-
trations (a), low CEC (e), an ustic soil moisture regime
(d), and low potential to supply potassium (k). The
Jamaican sites were medium to fine in texture, types
L or C, with many having no obvious chemical con-
straints. Some did have acidity constraints (h), and a
few had Al constraints (a). A low potential to supply
K was also present at several sites. Soils studied in Puer-
to Rico were also medium and fine in texture, types
L and C, with no subtype texture modifier. These
generally contained more Al, and were universally low
in their potential to supply K. Further work on this
project will compile soils data, comparing tree growth
and FCC grouping for each site.


Alluvial Soils of the Amazon Basin
Robert E. Hoag, N.C. State University
Stanley W. Buol, N.C. State University
Jorge Perez, INIPA

Alluvial soils are usually considered to be of high
native fertility. The purpose of this work was to test
such a hypothesis, which would be useful in ex-
trapolating soil-management options for alluvial soils
in the humid tropics. To do so, the investigators sampl-
ed soils from three different types of deposits in the
Amazon Basin of Peru, determined their physical,
chemical and mineralogical properties, and developed
a means of predicting the occurrence of the contrasting
soil properties on flood-plain landscapes in the region.
The 20 sampling sites, placed into three groups, were
selected on the basis of the geologic formation from
which the tributaries originate. Representative data are
given in Table 1. Sampling sites in Group One were
along rivers that originate within the Eastern Peruvian
Cordillera. As predicted, these soils have relatively high
pH values throughout their profiles, ranging from 6.5
to 8.5. (Complete data for each of the profiles sampl-
ed are in Mr. Hoag's thesis).
Sampling sites in Group Two were along rivers that
originate in the foothills of the Peruvian Andes, where
carbonaceous and non-carbonaceous sandstones
predominate in the headwaters. These soils have
chemical properties similar to those sampled along cer-
tain rivers that originate in the Ecuadorian Cordillera.
Although upper elevations of the Ecuadorian Andes
are composed predominantly of acid igneous and
Table 1. Summary of topsoil fertility and FCC classification
representative of sampling sites in three groups of soils in the
Amazon Basin of Peru. (Chemical values are for top 20 cm)
Group 1 Group 2 Group 3
Soil Property Rio Mayo Rio Cashiboya Rio Yavari
FCC Classification Cgvb Cg Cga
Ph 7.5 5.6 4.0
AI Saturation 0 0 78
Ca, meq/100 g 39.7 29.7 1.8
Mg, meq/100 g 9.7 7.0 0.5
K, meq/100 g 2.02 0.69 0.36
Mn, ppm 90 130 35
Cu, ppm 6.2 4.9 2.3
Zn, ppm 4.5 3.2 3.3
P, ppm 145 29 6

volcanic rock strata, the tributaries dissect limestone-
bearing marine deposits along the eastern flank of the
mountains. This group of nine sampling sites has pH
values throughout their profiles of 5.0 to 6.5, and tend
to be near a value of 6.0 in the upper horizons.
Mineralogy of the sand fraction is mixed, and mont-
morillonite dominates the clay fraction. Characteriza-
tion data for the sampling site along the Cashiboya
are representative of this group, although textures may
be loamy rather than clayey.
The third group of soils includes those sampled along
rivers that originate among pre-weathered, within-basin
sediments of Peru. Chemical properties of soils sampled
along two rivers that originate from within-basin
sediments and northern portions of the Andes in
Ecuador are also included with this group. These soils
are strongly acid, with pH values ranging from 4.0
to 5.0. The clay fractions of these soils are dominated
by either montmorillonite or kaolinite, with both
minerals being present in abundance. Aluminum
saturation is high and may exceed 85"% of the exchange
The soil profiles were classified according to soil tax-
onomy and results are presented in Table 2. Classifica-
tion according to the Fertility Capability Classifica-
tion system (FCC) was based upon data obtained from
samples submitted to the N.C. State University Soil
Testing Laboratory.

Physical, chemical and mineralogical data support
the premise that information describing the geologic
formations from which tributaries in the Amazon Basin
of Peru originate may be useful in predicting soil pro-
perties on floodplains. Soils along rivers with head-
waters in the Eastern Peruvian Cordillera are general-
ly of high base status and pH values. Montmorillonite
dominates the clay fraction of these soils. There may
be some question as to the availability of P and
micronutrients due to completing at the high pH
levels. Soils developing in sediments eroded from the
calcareous sedimentary deposits of the Andean foothills
in both Peru and Ecuador tend to be slightly acid with
no serious chemical or mineralogical problems. In the
Eastern portion of the Peruvian Basin, the floodplain
soils tend to be strongly acid with very high levels of
aluminum saturation. Repeated sequences of weather-
ing, erosion and deposition over a long period of time
have apparently contributed to the leaching of solu-
ble bases and dominance of Al on the exchange


Table 2. Taxonomic classification of representative profiles of alluvial soils of the Upper Amazon.

Rio Aguaytia
Rio Blanco
Rio Cashiboya
Rio Cumbaza
Rio Cushabatay
Rio Mayo
Rio Mazan
Rio Nanay
Rio Napo
Rio Nucuray
Rio Paranapura
Rio Pastaza
Rio Putumayo
Rio Samiria
Rio Tamshiyacu
Rio Tapiche (upper)
Rio Tapiche (lower)
Rio Tigre
Rio Utoquinea
Rio Yavari

-Typic Tropofluvent, clayey over loamy, mixed (nonacid) isohyperthermic.
-Aeric Tropaquept, fine, montomorillonitic (acid), isohyperthermic.
-Aeric Tropic Fluvaquent, very-fine, montmorillonitic (acid), isohyperthermic.
-Typic Tropofluvent, coarse-loamy, siliceous (nonacid), isohyperthermic.
-Aeric Tropic Fluvaquent, coarse-loamy, mixed (nonacid), isohyperthermic.
-Aquic Hapludoll, very-fine, montomorillonitic (calcareous), isohyperthermic.
-Aeric Tropic Fluvaquent, fine, kaolinitic (acid), isohyperthermic.
-Aeric Tropic Fluvaquent, fine-loamy, siliceous (acid), isohyperthermic.
-Typic Tropofluvent, coarse-silty over clayey, mixed (nonacid), isohyperthermic.
-Typic Eutropept, fine-silty, mixed, isohyperthermic.
-Typic Tropofluvent, coarse-loamy mixed (nonacid), isohyperthermic.
-Typic Fluvaquent, coarse-loamy, mixed (nonacid)
-Aeric Tropaquept, very-fine, kaolinitic (acid), isohyperthermic.
-Typic Tropofluvent, fine-silty, mixed (nonacid), isohyperthermic.
-Typic Fluvaquent, fine, montmorillonitic (acid), isohyperthermic.
-Aquic Fluvaquent, fine montmorillonitic (acid), isohyperthermic.
-Aquic Eutropept, fine-silty, mixed, isohyperthermic.
-Aeric Tropic Fluvaquent, fine-loamy, mixed (acid), isohyperthermic.
-Aeric Tropic Fluvaquent, fine-silty, mixed (acid), isohyperthermic.
-Typic Tropaquept, very-fine, montmorillonitic, (acid), isohyperthermic.

Soils of the humid tropics are diverse not only on
uplands and mountains, but also on floodplains. The
notion that all alluvial soils within the Amazon Basin
of Peru are homogeneous is clearly mistaken. As
chemical and physical properties diverge from one loca-
tion to another, so must management recommenda-
tions. Results of this study may help to identify the
geographical boundaries within which research data

may be extrapolated. The study may serve to iden-
tify regions that may require additional research if
agriculture is to expand in them. As an example, it
may be necessary to use lime or Al-tolerant genotypes
in the very acid alluvial soils in Eastern Peru. Data
from this study may also prove to be useful in selec-
ting drainage systems for more intensive soil-genesis
or fertility investigations.


Ultisol Dominated Landscapes
In Southeastern Peru
Laurie J. Newman, N.C. State University
Stanley W. Buol, N.C. State University
Rafael Chumbimune, INIPA-CIPA XVII,
Madre de Dios

An area of southeastern Peru was selected for a study
whose objectives where 1) to characterize the physical,
chemical and mineralogical properties of the soils of
the region, and 2) to determine the relationship of soil
properties to landscape position. The research site, 450
ha near Puerto Maldonado, Madre de Dios, is con-
sidered representative of the soils, climatic conditions,
landscapes and vegetation in the region. Results from
this project may assist in the extrapolation of research
to areas within the region where knowledge of the
soil resource is scarce.
The topography of the region is characterized by
level uplands, dissected side slopes and recent flood
plains. Lower base levels caused drainage entrenchment
and formation of the associated convex and planar side
slopes. The flood-plain soils are forming in Holocene,
local alluvium and organic parent materials.
Ultisols with ustic soil moisture regimes are the
predominant soils of the uplands. These Ultisols can
be characterized as having pH values ranging from 3.9
to 4.9, aluminum saturation values greater than 70%
of the effective CEC in the argillic horizons, and sur-
face horizon cation exchange capacities of 1 to 6 cmol
(+)/kg. Textures of the Ustults vary from clayey to
course-loamy, as a function of the texture of the in-
itial materials and position on the landscape. There
are few weatherable minerals in the sand size fractions
of these soils. The dominant clay mineral is kaolinite.
Some muscovite mica, vermiculite and hydroxy-Al-
interlayered vermiculite are present. Paleusults, located
in positions where water tables fluctuate within the
profile, have features associated with oxidizing and
reducing conditions such as mottling, plinthite and in-
durated iron.
The poorly drained soils formed in recent alluvium
are Placaquods and Troposaprists. The texture of these
soils is dependent on the depositional environment of
the alluvial materials. The sand fraction is primarily
quartz. Magnetite, pyrite and kyanite are present in
trace quantities. The soils are strongly acid to medium
acid and vary in base status.
Three geomorphic surfaces have been defined within
the area. Two of these are located on the upland, and

the other occupies the lower areas associated with
stream drains.
Surface 1 covers the level and nearly level uplands.
It is the oldest and most stable surface of the three.
Surface 2 consists of the side slopes and dissected por-
tions of the uplands. Within this group are landscape
positions with greater than 3% slope. Surface 3, on
the recent flood plains, occupies the smallest portion
of the study area. This surface is the youngest and may
be subject to rapid changes in morphology with the
movement of the nearby channel.
The soils have textures ranging from loamy fine sand
to clay. All of the upland soils in the sandy area have
an increase in clay content and a decrease in sand con-
tent with depth. In the flood plain of the third order
stream, organic soils are dominant. On the uplands
of the sandy area, clayey soil families are located beside
soils with coarse loamy control sections. These abrupt
changes in texture laterally across the landscape are
characteristic of areas where soils have developed in
alluvial parent materials.
Soil pH values for the upland soils are generally
higher at depth in the profile than at the soil surface.
Values determined in water range from 3.4 to 4.9
(Table 1). The soils in the recent flood plain of the
second-order streams have pH values ranging from 1.9
to 5.6. The pH values of the upland soils of the region
may be a result of soil formation in acid parent
materials, or of formation in higher pH materials where
bases have been leached out over time.
Soil reaction data from the flood plains of the Madre
de Dios and Tambopata Rivers indicate that recent
alluvium is neutral to slightly acid reaction, having pH
values ranging from 5.2 to 6.9. Because the present-
day rivers have a similar source as the sediments in
which the upland soils are formed, it is assumed that
the soils of the upland have developed in sediments
with neutral reaction and that post depositional
removal of bases is responsible for the increase in soil
Extremely low pH values of 1.9 to 2.1 have been
measured in the epipedon and subsurface horizon of
a buried soil in the second-order stream drain (Table
1). pH values of less than three are rare in saturated
soils, but have been recorded in sulphitic soils, a result
of oxidation after soil drainage. Pyrite (Fe2S) may be
present in these horizons and may be assumed to be
controlling the very acid soil reaction. The organic soils
associated with the Rio Chonta have pH values similar
in range to those on the upland.
The amount of exchangeable aluminum of the


Table 1. Physical and chemical properties of selected profiles in Puerto Maldonado, Peru.
Depth Clay Sand Org.C pH Al Ca Mg K ECEC ECEC Al sat.
(cm) --- % --- H20 --- --- c mol(+)kg --- --- cmol(+)kg clay %
Soils of the Level Upland (Surface 1):

Carretera; Typic Paleustult; clayey, kaolinitic, isohyperthermic

0-9 27 26 1.6 3.9 4.1 0.1 0.2 0.2 5 19 81
9-25 30 22 0.7 3.9 4.2 0.1 0.1 0.1 5 17 82
25-52 39 19 0.6 4.2 5.3 0.1 0.1 0.2 7 17 79
52-70 44 18 0.5 4.4 6.4 0.1 0.1 0.1 7 16 90
70-118 48 17 0.5 4.5 6.8 0.0 0.1 0.1 8 16 89
118-155 47 16 0.7 4.6 6.7 0.0 0.2 0.1 8 17 86
155-200 46 21 0.3 4.6 6.3 0.0 0.2 0.1 8 16 86

Estacion; Typic Paleustult; coarse-loamy, siliceous, isohyperthermic

0-9 3 81 0.4 4.3 0.3 0.4 0.4 0.1 2 53 18
9-24 2 70 3.9 1.0 0.1 0.2 0.1 2 73 60
24-53 5 69 3.9 1.7 0.1 0.1 0.1 2 43 73
53-82 15 62 4.1 1.4 0.2 0.2 0.1 2 14 68
82-110 16 68 0.3 4.1 1.4 0.0 0.0 0.1 2 12 74
110-153 18 66 4.1 1.4 0.0 0.0 0.1 2 10 79
153-200 19 64 4.1 1.7 0.0 0.0 0.1 2 10 84

upland soils in the study area range from 0.1 to 7.5
cmol(+)/kg. The maximum occurs in the lower B
horizon of the Palma Real profile, a somewhat poor-
ly drained soil with clayey textures throughout the
solum. The minimum occurs in the surface horizon
of the Amable profile, a sandy textured epipedon on
a 12% slope. In the upland soils, exchangeable
hydrogen is a significant portion of the acidity, com-
prising up to 25 percent in B horizons. In all well drain-
ed soils, the total exchangeable acidity is higher in B
horizons than in the surface horizons.
All upland soils in undisturbed forests have low con-
tents of exchangeable bases throughout. Soils that have
recently been cut and burned for agricultural use have
concentrations of calcium and magnesium in the sur-
face horizons of 0.30 to 0.41 cmol(+)/kg. Basic ca-
tion content ranges in most B horizons are 0.01 to
0.05 cmol(+)/kg for calcium, 0.0 to 0.22 for
magnesium, and 0.2 to 0.15 for potassium. The highest
values are found in soils with clayey family particle
size classes. The location of Estacion profile has been
used in the past for lime rate experiments. This may
account for the calcium concentrations of 0.10
cmol(+)/kg to 82 cm in the profile.

Soils of the recent flood plain have cation satura-
tion values for calcium, magnesium, and potassium in
mineral soils ranging from 0.06 to 0.89, 0.26 to 4.49,
and 0.02 to 0.06, respectively and values for calcium,
magnesium, sodium and potassium in organic horizons
ranging from 2.37 to 4.71, 8.95 to 18.10, 0.10 to
0.33 to 0.17 to 0.68 cmol(+)/kg, respectively. High
concentrations of calcium and magnesium could be
a result of the deposition of sediments from more basic
waters of the Tambopata that flood the Chonta River
during the wet season.
In the surface and some subsurface horizons of the
coarser-textured soils, effective CEC values range from
23 to 82 cmol(+)/kg. The presence of muscovite mica,
vermiculite and hydroxy-Al-interlayered vermiculite
with cation exchange capacities of 20-40, 100-150, and
10-40 cmol(+)/kg account for the greater CEC. Soil
horizons that have organic carbon contents greater than
2% have higher exchange capacities than clayey con-
tent CEC correlations could suggest as a result of the
100-300 cmol(+)/kg CEC associated with organic
matter. In these samples, effective CEC values range
from 44 to 275.
The pH-dependent charge ie., CEC, ECEC, increases



Coarse Loamy

Coarse Loamy


-- Mashco
Sandy Astillero Coarse Loamy
Placaquoda Fine Loamy Aquic
(0-2%/) Paleustults Palenstults
(2-36%) (5-12%)
Figure 1. Relationship of soil map units to landscape

with depth in all mineral soil profiles, reflecting an in-
crease in kaolinite in the clay fraction. This property
has been observed in Ultisols of the southeastern
United States.
X-ray diffraction was used to identify and quantify
the minerals present in the clay fraction of selected
horizons. Diffraction patterns show the predominance
of kaolinite in all of the horizons analyzed. Muscovite
mica, low-charge vermiculite and hydroxy-aluminum
interlayered vermiculite are present in varying quan-
tities, as are minor amounts of goethite, gibbsite, and
Soils with clayey family particle size classes have ver-
miculite as the second most abundant clay mineral.
All other soils in loamy, sandy and organic particle
size classes have muscovite mica as their second most
abundant clay mineral. The sandy, poorly drained soils
in the second-order stream drain have the lowest
amount of vermiculite of any of the soils. Gibbsite
is present in the B horizons of profiles that contain
plinthite. The trace contents of talc, which is a stable
product of metomorphism, have probably been
transported from the sedimentary parent material
source in the Andes.
In the clay fraction the sequence: mica- ver-
miculite (expanded hydrous mica)--hydroxy in-
terlayered vermiculite---kaolinite represents the suc-
cession of the stages of weathering.
All sand fractions examined by petrographic analysis
were greater than 95% quartz. Muscovite mica was
present in small amounts as were the heavy minerals,

Palma Real
(0-3%) /

anisotrapic kyanite and the isotropic magnetite.
All of the soils in the study area have developed from
unconsolidated sediments. Most have formed in an-
cient alluvium, and some are formed from more re-
cent alluvial deposits. Soil morphology and genesis dif-
fer in these soils as a result of differing textures of the
original parent materials, their geomorphic position
and their relationship with the present day water table.
The idealized block diagram in Figure 1 presents seven
major map units and their position on the landscape.

Soils of Pichis Valley Extrapolation Sites

Laurie J. Newman, N. C. State University
Dennis del Castillo, N. C. State University

The purpose of this project was to characterize the
soils where the Pichis Extrapolation Project is adap-
ting soil-management technologies developed at
Yurimaguas to a location in the High Selva of Peru.
The extrapolation sites are two neighboring experi-
ment stations situated on opposite sides of the Pichis
River. Compared to Yurimaguas, this area has lower
night temperatures, 50% more rainfall and steeper
slopes. The important landscapes for agriculture are
alluvial positions and gently rolling to steep uplands.
Four pedons were sampled in October 1984 and
samples were analyzed at the Yurimaguas laboratory.
Mineral family was inferred from the sum of cations
and clay content. The results shown in Table 1 in-
dicate that the alluvial terrace soils are Fluventic
Eutropepts, slightly acid, but with otherwise high
native fertility (Pedon A). Soils on the upland, rolling
sites at La Esperanza Station are clayey, kaolinitic Typic
Paleudults (Pedon B). Soils on the high, nearly level
terraces at both stations are Typic Palehumults and
Typic Tropohumults, with high topsoil organic mat-
ter contents (Pedons C and D).
On the basis of the four pedons described and analyz-
ed, soils at the extrapolation sites differ only slightly
from those at Yurimaguas. The Pichis soils contained
proportionately more clay and organic matter, and
there are probably more 2:1 clays. These
characteristics, taken together, indicate that these soils
will retain more exchangeable Al. Lime requirements
are expected to be higher, and percolation of bases less
than those on soils at Yurimaguas. The high base status
soils on the low terraces at the extrapolation site ap-
pear comparable to soils on the low terrace at


Table 1. Characteristics of four soil profiles on the Pichis Valley, Peru.
Exchangeable Al Organic
Horizon Texture pH Al Ca Mg K ECEC sat. matter
cm (H20) cmol(+)/L % %

El Vivero, low terrace, Fluvaquentic Eutropept, clayey, mixed, isohyperthermic. FCC:LCh.

Ap 0-9 silt loam 5.4 0.3 9.00 2.08 0.24 11.62 3 3.4
Bti 9-24 silty clay 5.5 0.5 9.45 1.83 0.14 11.92 4 1.3
Bw 254 silty clay 5.1 5.4 4.53 1.50 0.10 10.53 51 0.7
Ab 54-89 loam 5.3 3.7 3.77 1.24 0.09 8.80 42 0.3
Bwgl 89-130 5.4 1.8 6.75 2.48 0.15 11.18 16 0.3
Bwgl 130-151 silty clay 5.7 1.0 7.44 3,41 0.16 12.01 8 0.2
Cg 151-180 clay 6.0 0.4 8.59 3.79 0.16 12.94 3, 0.4

Esperanza Experiment Station, upland area secondary forest. 11% slopes. Typic Paleudult, clayey, mixed, isohyper-
thermic. FCC:LCak.
A 0-6 loam 3.8 5.3 0.20 0.20 0.16 5.86 90 5.2
Btl 6-33 clay loam 4.2 5.8 0.20 0.17 0.05 6.22 93 1.4
Bt2 33-65 clay loam 4.3 6.7 0.24 0.14 0.06 7.14 94 1.0
Bt3 65-114 clay 4.5 8.4 0.23 0,10 0.03 8.76 96 0.3
Bt4 114-150 fine sandy 4.7 9.3 0.20 0.09 0.05 9.44 99 0.3
BC 150-180 fine sandy 4.7 6.9 0.20 0.09 0.05 7.24 95 0.3

El Vivero, high terrace, 3% slope on gently undulating topograph.Typic Palehumult, claey, mixed, isohyperthermic.
Ap 0-2 silt loam 4.2 5.9 1.74 1.02 0.22 8.88 66 6.2
Apb 2-13 clay loam 4.3 6.9 0.40 0.36 0.16 7.82 88 1.7
A 13-36 clayloam 4.4 7.3 0.27 0.24 0.12 7.93 92 1.3
Bt 36-84 clay loam 4.4 8.8 0.27 0.23 0.08 9.38 94 1.2
Btg 84-123 clay 4.3 9.1 0.47 0.16 0.07 9.80 93 0.5
BC 123-165 clay 4.7 10.1 0.43 0.14 0.06 10.73 94 0.4
Cg 165-180 clay 4.8 15.3 0.20 0.14 0.06 15.70 97 0.2

Esperanza Experiment Station, virgin forest. 0-2% slopes. Typic tropo humult, mixed, isohyperthermic. FCC: LCak.
A 0-8 clay loam 3.7 7.9 0.23 0.18 0.12 8.4,3 94 4.2
AB 8-19 clay loam 4.1 5.6 0.27 0.18 0.10 6.15 91 2.3
Btl 19-46 clay loam 4.6 4.4 0.53 0.17 0.08 5.18 85 1.13
BT2 46-79 clay 4.5 7.0 0.23 0.10 0.03 7.36 95 0.8
Bt3 79-96 clay 4.5 6.1 0.20 0.10. 0.03 6.43 95 0.9
Bt3 96-120 clay 4.6 5.6 0.20 0.10 0.03 5.93 94 0.7
Btg 120-142 si. clay 4.7 5.2 0.20 0.10 0.03 5.53 94 0.2
Bx1 142-162 loam 4.7 5.1 0.23 0.09 0.03 5.45 94 0.2
Bx2 162-180 loam 4.8 4.8 0.23 0.09 0.03 5.15 93 0.4
Bx2 180-200 loam 4.8 4.8 0.13 0.09 0.04 5.06 95 -


Soil Survey of the Puerto Maldonado
Experiment Station
Laurie J. Newman, N. C. State University
Stanley W. Buol, N. C. State University
Rafael Chumbimune, INIPA, CIPA XVII,
Puerto Maldonado

The purpose of this study was to obtain an
understanding of soils, including their genesis and mor-
phology, at the Puerto Maldonado Experiment Sta-
tion, so that local agriculturalists will be able to recom-

mend management options for similar soils in the
region. The station is in the Acre alluvial basin of
southeastern Peru. The upland soils have ustic soil
moisture regimes, and the natural vegetation is seasonal
semi-evergreen forest.
Table 1 presents the soils classified by Soil Tax-
onomy and Fertility Capability Classification (FCC).
The complete Soil Survey of the Puerto Maldonado Ex-
periment Station (Newman, L.J. 1985) is available from
the Tropical Soils Research Program, Box 7619, N.
C. State University, Raleigh, NC 27695-7619, USA.

Table 1. Classification of the soils of the Puerto Maldonado Station, Peru
Soil Taxonomy
Name Sub Group Family FCC
Soils of the Level Upland
Carretera Typic1 Paleustult clayey, kaolinitic, isohyperthermic LCadk

Estacion Typici Paleustult coarse-loamy,siliceous, SLadek

Palma Real Plinthaquic1 Paleustult clayey, kaolinitic, isohyperthermic Cadgk

Soil of the Upland Sideslopes
Astillero TypicI Paleustult fine-loamy, siliceous, Ladek

Amable Typici Paleustult coarse-loamy, siliceous, Ladek

Mashco Aquic1 Palestult coarse-loamy, siliceous, SLadegk

Soils of the Recent Floodplain
Laberinto Aeric1 Placaquod sandy, siliceous, isohyperthermic Sdeghk

Chonta Typic Troposaprist dysic, isohyperthermic Ok
1 Sub group not presently defined in Soil Taxonomy (Soil Management Support Systems, 1985).


In the humid tropics, the limited resources and qualified professionals of national institutions often
prevent programs from pursuing improved soil-management practices beyond the primary research
sites. Linking programs and institutions into a network, however, can pool talent and resources, ac-
celerating the flow of information among scientists and specialists, and the transfer of soil-management
technology to producers.
Work led by TropSoils, REDINNA and IBSRAM scientists has launched the development of a
soil-management research network for the humid tropics. The network's objectives are 1) to develop
the capability of collaborating country personnel to conduct, interpret, and report user-oriented, soil-
management research, and 2) to validate and extrapolate available soil-management technologies to
other countries beyond the TropSoils primary research sites.

Work during 1985 was geared at promoting the establishment of such a network at three levels:
nationally in Peru in cooperation with INIPA; regionally in Tropical America in cooperation with
REDINNA and AID's Latin America and Caribbean Bureau; and tropics-wide in cooperation with
IBSRAM. An underlying assumption has been the need to link with other institutions, because of
the limited time and resources available for research validation in TropSoils and the fact that actual
technology transfer is the mandate of other projects or institutions. TropSoils' role in networking
is viewed as catalytic, educational and as a major source of new management technology for the humid
tropics. Agencies with a clear mandate for technology transfer such as national institutions like INIPA,
and network institutions like REDINNA and IBSRAM, are viewed as the means to conduct the net-
works. The U.S. Agency for International Development (USAID) participates in the network, and
the project has engaged the cooperation of several USAID bureaus, as well as USAID missions in
the humid tropics around the world.
Although countries with significant areas in the humid tropics differ widely in their stage of soil-
management technology development, they all share one common limitation: lack of up-to-date know-
how on management of humid tropical soils by scientists and extension agents stationed in the humid
tropics. They may have excellent backing from top scientists, most of whom are in the nation's capital.
Although they may exert considerable positive influence on program direction, it is the people in
the field who determine the success of technology transfer. The vast majority of them are the B.S.
level (ingeniero agr6nomo or similar title) who received university training based on now oboslete con-
cepts of tropical soil management (lime to pH 7, for example). This project intends to focus on such
individuals as the real validators and transformers of technology.

Catalyzing Events
In order to promote network development, TropSoils personnel in 1985 helped conduct a series
of workshops, courses and meetings, organized jointly with several agencies and institutions. Highlights
IBSRAM's Inaugural Workshop to Launch the Acid Tropical Soils Management Network,
in Yurimaguas, Manaus and Brasilia, April 24-May 3. There were 67 participants, representing 13
national institutions, seven donor agencies and a number of international centers and universities. The
sponsoring institutions were INIPA, EMBRAPA, ACIAR-Australia, ORSTRM-France, Soil Manage-
ment Support Services (SMSS), USAID and N.C. State University. Acting on expressed country in-
terests, participants determined that the Acid Tropical Soils Management Network will concentrate
its technology-validation activities on 1) pedology and fertility interactions, 2) soil acidity, 3) phosphorus
fertilization, 4) management of the soil surface, 5) rehabilitation of degraded, acid tropical lands, and
6) soil dynamics. A TropSoils scientist participates in the follow-up activities as a member of the
network's coordinating committee. Initial emphasis will be given to humid tropical Africa.
Also supported was IBSRAM's sister Inaugural Workshop on Land Clearing and Development,
held in Jakarta in August, 1985, which was attended by a scientist who summarized TropSoils research
experience on land clearing and reclamation in both Peru and Indonesia. The TropSoils scientist is
a member of that network's coordinating committee.
Agroforestry Research Course for the Humid Tropics, in Yurimaguas, June 3-22. This course
was co-sponsored by INIPA and ICRAF with IDRC-Canada. There were 27 participants from Bolivia,
Brazil, Colombia, Ecuador, Peru, and Venezuela. Follow-up activities include research validation in
soil management for agroforestry systems throughout the Amazon.
National Selva Program Planning Workshop, in Yurimaguas, October 24-31. Ninety Peruvian
research and extension workers of INIPA's National Selva Program met and determined priorities
for technology transfer through the Selva. Other participants came from Peruvian universities, from
IIAP and from the USAID-sponsored Selva Special Projects. A TropSoils senior scientist was assigned
the task of coordinating technical aspects of the Selva Program. The 22 soil scientists present developed
and recommended 11 specific activities for technology validation and transfer.

New Collaboration
Plans are to expand the network by capitalizing on collaborative research and training activities
already planned or conducted by NCSU and TropSoils in some 3 3 countries around the world, and
by maintaining contact with the several hundred students and professionals who have received train-
ing or degrees while participating in NCSU and TropSoils projects. Interest is especially strong in
Asia, where the INSFERR Network has begun widespread testing of the Fertility Classification System
(FCC). Countries with collaborative, network relationships with N.C. State University TropSoils are:

Latin America (13): Bolivia, Brazil, Colombia, Costa Rica, Dominican Republic, Ecuador,
Jamaica, Mexico, Panama, Peru, Puerto Rico, Trinidad, Venezuela.

Africa (8): Burundi, Cameroon, Congo, Ivory Coast, Madagascar, Nigeria, Zambia, Zimbabwe.

Asisa (11): Burma, China, India, Indonesia, Malaysia, Nepal, Pakistan, Phillipines, Sri Lanka,
Taiwan, Thailand.


Research conducted at Manaus, Brazil has offered the opportunity to select promising soil-management
technologies from the primary research site at Yurimaguas, Peru, and test them in an ecosystem with
different soils and climate. The ecosystem which includes Manaus occupies approximately 57 percent
of the Amazon Basin. It is near-ustic, and the predominant vegetation is semi-evergreen, seasonal,
primary forest, whereas vegetation at Yurimaguas is typically secondary forest. The soils at Manaus
are Oxisols, found in 45 percent of the Amazon Basin, mostly in near-ustic and ustic soil-moisture
regimes. They are clayey, very fine, kaolinitic Typic Acrorthox on flat plateaus, in contrast to the
fine loamy, siliceous Typic Paleudults of Yurimaguas. Phosphorus sorption capacity for the clayey
Oxisol at Manaus is intermediate between the high levels for the Cerrado Oxisols and the low levels
found in Yurimaguas Ultisols.
Research has been conducted to discover how these differences in soils and climate affect the adap-
tation of soil-management technologies to the ecosystem found at Manaus. The first experiment ex-
amined the response of crops under continuous cultivation to a range of nutrients. It was designed
to establish the times at which the application of each nutrient was required in order to sustain long-
term production. As this central experiment indicated the need for more detailed information related
to individual nutrients, satellite experiments were established. The work reported here is a cooperative
project conducted at the UEPAE/Manaus Station (Unidade de Execucao de Pesquisa de Ambito Estadual
de Manaus), conducted jointly with EMBRAPA (Empresa Brasileira de Pesquisa Agropecuaria).
This section also includes a report on work conducted in the Cerrado region, in collaboration with
the Centro de Pesquisa Agropecuaria dos Cerrados (CPAC/EMBRAPA).


Soil Nutrient Dynamics and Fertility
Management for Sustained
Crop Production on Oxisols
In the Brazilian Amazon
Thomas Jot Smyth, N. C. State University
Manoel Cravo, EMBRAPA
Joaquim B. Bastos, EMBRAPA

The objectives of this project were 1) to establish
the patterns of soil-nutrient depletion as a function of
time after clearing for a Central Brazilian Amazon Ox-
isol under continuous cultivation; 2) to determine the
fertilizer inputs required for sustaining continuous crop
production on these Oxisols; and 3) to determine how
a soil-fertility management system for the Manaus Ox-
isols would differ from one for the Yurimaguas Ultisols.
During the four years in which this study has been
conducted, eight crops have been harvested in the se-
quence described in Table 1. Table 2 shows selected
fertilizer treatments during continuous cultivation.
Other treatments included a check (no fertilization),
and a treatment with residue incorporation. A judicious
monitoring of soil and plant nutrient levels for every
crop determined when each treatment was establish-
ed. Treatments for copper and lime were initiated in
the second soybean crop, for sulfur in the second corn
crop, for boron and zinc in the third corn crop, and
for manganese in the third cowpea crop.
Topsoil nutrient-depletion patterns for this Oxisol
are shown in Figure 1 and Table 3 for the absolute
check treatment, which has not received any fertilizer
or lime. Soil nutrient levels initially increased as a result
of ash additions from burning the primary forest. The
liming effect of the ash reduced Al saturation values
from 73% to 18% during the first crop.

Table 1. Sequence of crops, varieties and time after
burning for the cultivation of each crop.
Crop Variety Planting to Harvest Time
Months After Burning
Rice lAC 47 3.0 7.4
Soybeans Tropical 8.9 12.6
Soybeans Tropical 18.5 22.3
Cowpeas Manaus 22.9 25.2
Corn BR 5102 27.6 31.3
Cowpeas VITA 3 32.7 34.9
Corn BR 5102 37.2 41.8
Soybeans Tropical 42.3 46.2

Ash analysis indicated that approximately 10 kg/ha
of P were added to the soil by the burn. Crop yields
for the absolute check treatment (Figure 2) suggested
that yields, in the absence of fertilizer inputs, would
be negligible after the first crop. Despite the initial in-




60 >

40 -




0 10 20 30 40 60
Months After Burning
Figure 1. Soil dynamics for carbon, total N, acidity,
exchangeable Al, Ca, Mg, K and Mehlich I P during
the initial 44 months after clearing the primary forest
on a Manaus Oxisol.


crease in Mehlich 1 extractable soil P (Figure 1), rice Potassium
yields were doubled by the broadcast application of Topsoil K levels had declined from 107 to 56 ppm
22 kg P/ha (Figure 2). Although all succeeding crops before planting the second crop (Figure 1). Treatments
showed a significant response to P fertilization, the evaluating three rates of K were, therefore, initiated
optimum P rate was related to the timing of fresh P with this crop. Yield responses to K and the incor-
applications and crop P requirements. Relationships portion of residues from the previous crop are/shown
between crop yield and soil-test P suggested optimal in Figure 3. The crop-residue treatment was included
Mehlich 1 P levels in the range of 5-10 ppm for corn, to evaluate the effects of returning residues when these
8-10 ppm for cowpeas and 10-15 ppm for soybeans. are harvested for threshing the grait. Yield for this
Cumulative yields for the total applications of 88, 176, treatment was low on the initial soy ean cropnsince
and 264 kg P/ha were 11.3, 13.5, and 16.7 t/ha, P was only applied prior to planting the thrrd crop
respectively. in the study (Table 2). Yields for the crop-residiie treat-

Table 2. Fertilization history for selected treatments in the nutrient-dynamics study.
P, P2 P3 N1 N2 N3 K, K2 K3
Crop Fert.
N 60 60 60 30 60 90
P 22 44 66 44 44 44
N 20 20 20 0 0 0 20 20 20
P 22 44 66 44 44 44 88 88 88
K 50 50 50 50 50 50 25 50 100
Mo .02 .02 .02 .02 .02 .02 .02 .02 .02
N 54 54 54 27 54 80 54 54 54
P 0 0 0 0 0 0 22 22 22
K 50 50 50 50 50 50 25 50 100
Cu 1 1 1 1 1 1 1 1 1
(no fertilizer applications in these treatments)
K 0 0 0 0 0 0 25 50 100
N 80 80 80 40 80 120 80 80 80
P 22 44 66 22 22 22 22 22 22
K 50 50 50 50 50 50 0 0 0
Lime (t/ha) 2 2 2 2 2 2 2 2 2
K 0 0 0 0 0 0 25 50 100
N 80 80 80 40 80 120 80 80 80
K 50 50 50 50 50 50 25 50 100
P 22 44 66 44 44 44 44 44 44
K 50 50 50 50 50 50 25 50 100
K 50 50 50 50 50 50 25 50 100
Crop failed in 43-day drought. Corn was cut and the stover incorporated.


Rice Soyb. Soyb. Cowp. Corn Cowp. Corn Soyb.

Figure 2. Grain yields for eight consecutive crops on
the absolute check and P treatments.

Applied K (kg/ha)
rI Residues
E22 25

S 100

Soyb. Soyb. Cowp. Corn Cowp. Corn Soyb.

Figure 3. Effects of K fertilization and crop residue
incorporation on yields of seven crops.

Table 4. Effects of K fertilization and crop-residue
stage during seven consecutive crops.

ment, relative to the K treatments, declined progressive-
ly following the first corn crop and yvere primarily
related to a continual decline in soil K (Table 4) and
increasing levels of soil acidity.
Yield response to fertilizer K has not exceeded 50
kg K/ha for any crop. Foliar K levels with this treat-
ment approached the recommended values in most
crops (Table 4). Topsoil K data suggested that residual
effects from fertilizer K were low (Table 4). This was

Table 3. Mehlich 1 extractable soil micronutrient
levels on the absolute check treatment as a function
of time after burning.
Time After Melich 1 Extractable
Burning Cu Mn Zn

months ppm


LSD .05



2 0.5

incorporation on soil K and foliar K levels at flowering

Applied Crop Sequence
K Soyb. Soyb. Cowp. Corn Cowp. Corn Soyb.

kg/ha soil K, ppm
25 61 20 29 16 30 18 23
50 91 26 48 19 50 25 48
100 78 30 68 19 97 64 110
Crop Residues 124 32 52 24 50 21 25
LSD .05 19 9 12 ns 15 11 15
leaf K, %
25 1.26 1.40 0.90 1.23 1.25 1.14 1.44
50 1.85 1.70 1.16 1.68 1.56 1.50 1.63
100 1.99 1.86 1.12 1.80 1.95 1.64 1.81
CropResidues 1.64 1.92 1.09 1.93 1.44 1.29 1.45
LSD.05 0.35 0.21 0.26 0.31 0.36 0.17 0.19