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
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 )
Crops ( jstor )
Pastures ( jstor )
Soil science ( jstor )
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


This item has the following downloads:















































































































































































































































































































































































Full Text
Technical Report for 1986-87
Draft, July 1987
Tropical Soils Research Program
Department of Soil Science
North Carolina State University

Draft, July 1987
Tropical Soils Research Program
Department of Soil Science
North Carolina State University Raleigh, NC 27695-7619
Supported by
The Soil Management Collaborative Research Support Program under a grant from the United States Agency for International Development
Edited by Pedro A. Sanchez and Cynthia L. Garver

- Highlights .................................................... 1
- Personnel ......................................................... 2
2. RESEARCH NETWORK .................................................... 7
- Yurimaguas Workshop .............................................. 8
- Network Development in Latin America: RISTROP ................. 16
- Network Development in Africa and Asia ......................... 19
3. LEGUME-BASED PASTURES .............................................. 24
- Persistence of Grass-Legume Mixtures under Grazing ............ 27
- Evaluation of Animal Preference in Small Plots ................ 42
- Nitrogen Contribution of Legumes in Mixed Pastures ............ 44
- Potassium Dynamics in Legume-Based Pastures ................... 50
- Sulfur Accumulation in Grazed Pastures ......................... 56
- Pasture Reclamation in Degraded Steeplands .................... 58
- Pasture Reclamation via Herbicides .............................. 63
- Legume Shade Tolerance .......................................... 67
- Extrapolation in Farmer Fields .................................. 69
4. LOW-INPUT SYSTEMS ................................................... 72
- Central Experiment: Transition to Other Technologies .......... 73
- Weed Control in Low-Input Cropping Systems ..................... 88
5. AGROFORESTRY SYSTEMS ............................................... 96
- Multipurpose Tree Selection for Alleycropping .................. 97
- Mulch Quality and Nitrogen Cycling ............................. 104
- Alleycropping in Ultisols ...................................... 112

- Alleycropping in Alluvial Soils ................................ 114
- Inga Rice Interface in Alluvial Soils ....................... 121
- Improved Fallows .............................................. 127
- Legume Cover Crops for Peach Palm ............................. 129
- Mycorrhizae Inoculation in Tropical Palm Nurseries ............ 132
- Nutritional Requirements of Peach Palm ........................ 134
- Nutritional Requirements of Other Amazonian Fruit Trees ....... 137 Nutritional Requirements of Gmelina arborea ................... 140
6. CONTINUOUS CULTIVATION ........................................... 143
- Conservation Tillage under Continuous Cropping ................ 144
- Conventional Tillage under Continuous Cropping ................. 156
- Central Continuous Cropping Experiment ........................ 160
- Nitrogen Carryover in Rotation and Strip Intercropping Systems. 164 Maintenance of Phosphorus Fertility under Continuous Cropping 173 Potassium Buffering in Yurimaguas Ultisols .................... 180
- Weed Population Shifts in Continuous Cropping ................. 186
7. COMPARATIVE SOIL DYNAMICS ......................................... 191
- Comparative Soil Dynamics under Different Management Options 192 Soil Organic Matter Pools as Affected by Management Options 206
- Root Production and Turnover in Different Management Options 210 Root Distribution in Pastures and Alleycropping Options ....... 216 Occurrence of Mycorrhizae in Crops, Pastures, and Tree Species. 219 Effect of Different Management Options in Mycorrhizal Infection 222 Rhizobium Nodulation in Grain Legumes ......................... 227
- Soil Macrofauna as Affected by Management Practices ........... 229

- NitrogenMineralizationandLeachingasAffectedby Management
Practices ..................................................... 233
- Nitrogen Mineralization and Soil Moisture Content ............ 241
- Utilization of the Fertility Capability ClassificationSystem 244 Alluvial Soils of the Amazon Basin ............................ 252
- Volcanic Ash Influence in Transmigration Areas of Sumatra ..... 271
- Nutrient Dynamics ............................................. 279
- Phosphorus Fertilizer Placement and Profitability ............. 281
- Lime and Gypsum Applications .................................. 284
- Nitrogen Management .......................................... 290
- Comparison of Mehlich-1, Mehlich-3, Bray-1, and Anion Exchange
Resin P Soil Test ............................................ 296
- Effects of Soil Texture,Zinc, and pH on Corn Yield and Plant
Zinc Concentration ........................................... 301
- A P-testInterpretation Model forKaoliniticSoilsusing
Mehlich-1 and Clay Content .................................... 307
- AP-testlnterpretation Model for OxisolsUsingMehlich-3,
Resin ......................................................... 310
- RunoffandErosion Process in a Primary Forest Catchmentof the
Humid Tropical Steeplands ..................................... 315
- Establishing a Plant Canopy in Eroded Ultisol Steeplands ...... 319
- RainfedLow-InputCrop Rotation Patterns inAlluvialSoils
Subject to Seasonal Flooding .................................. 323

- Evaluation of Pasture Germplasm under a Perudic Rainfall
Regime ........................... .... ................... 329
- Recuperation of Degraded Pastures Dominated by Homolepsis
aturensis ..................................................... 334
- Response of Upland Crops to Potassium and Lime Applications ... 337 Agroforestry Research Needs ................................... 349
13. PUBLICATIONS .................................................. 359

This is the fifteenth continuous year of operations of North Carolina State University's Tropical Soils Program and the fifth as part of TropSoils, the Soil Management Collaborative Research Support Program financed mainly by the U.S. Agency for International Development and collaborating host institutions: INIPA in Peru, EMBRAPA in Brazil, and AARD in Indonesia. The year 1986 was simultaneously very difficult but also very rewarding. We received a 25% budget cut from our donor agency in February 1986, which resulted in the discontinuation of our field research in Indonesia in August 1986 and in major adjustments in research activities in Peru, Brazil, and on campus. For the first time in the program's history, we were unable to offer new graduate assistantships.
But 1986 was also a very rewarding and productive year, partly due to the momentum of all our projects in full operation and partly by the implementation of two major initiatives: the soil management research network, and comparative soil dynamics with emphasis on tropical soil biology. INIPA's building program at Yurimaguas is almost complete and the station was officially inaugurated by the Minister of Agriculture. Although modest by world standards, the Yurimaguas Experiment Station now has sufficient office, laboratory, and computer facilities to support long-term experiments and a 30-person training and conference center for on-the-job training. Three major international workshops were held at Yurimaguas in 1986: a Latin American Agroforestry Workshop in cooperation with ICRAF; the Third Tropical Soil Biology and Fertility Workshop; and the Latin American Soil Management Workshop, which trained 31 professionals from the "planting stick to the computer" in 3 weeks and helped launch the network. Extrapolation activities were also facilitated in tropical Asia and Africa through IBSRAM's Acid Tropical Soils Network. These network activities have paved the way for meaningful technology validation and transfer for 37 countries throughout Latin America, Asia, and Africa.
Work reported herein is carried out in close collaboration with INIPA in Peru, EMBRAPA in Brazil, AARD in Indonesia, several international centers, USAID Missions, and our sister TropSoils universities.

Robert H. Miller Department Head
Pedro A. Sanchez Program Coordinator
Dale E. Bandy Chief, NCSU Mission to Peru
T. Jot Smyth Assistant Professor of Tropical Soils
Jose R. Benites Project Leader, Yurimaguas
Michael K. Wade* Project Leader, Indonesia
Julio C. Alegre Assistant Professor of Soil Physics,
Dennis del Castillo Project Leader, Pichis Palcazu
Stanley W. Buol Professor of Soil Genesis and
D. Keith Cassel Professor of Soil Physics
Fred R. Cox Professor of Soil Fertility
Charles B. Davey Professor of Forestry and Soil Science
Eugene J. Kamprath Professor of Soil Fertility
Robert E. McCollum Associate Professor of Soil Fertility
Kenneth Reategui Research Associate, Pichis Palcazu
Robert J. Scholes Post Doctoral Fellow in Plant Ecology
Mary C. Scholes Post Doctoral Fellow in Soil
Microbiology (Yurimaguas)
* Resigned during the year.

George C. Naderman, Jr. Extension Soil Management Specialist
Bertha I. Monar Program Administrator
Mariela Gonzalez Administrative Assistant, Lima Office
Sue Florindez* Administrative Assistant, Yurimaguas
Patricia Gowland Research Technician
Tonya K. Forbes Research Technician
01inda Ayre Soil Analysis Laboratory, Yurimaguas
Rafael Roman Plant Analysis Laboratory, Yurimaguas
Valeria Medeiros* Bilingual Secretary
Elizabeth Phillips Bilingual Secretary
Amparo Ayarza Translator, Yurimaguas
Graduate Students (with degree candidacy and nationality) Miguel A. Ara Soil-pastures, Pucallpa (PhD-Peru)
Miguel A. Ayarza Soil-pastures, Yurimaguas (PhDColombia)
Dan W. Gill Soil fertility, Indonesia (PhD-USA)
Ricardo J. Melgar Soil fertility, Manaus (PhD-Argentina)
Ibere D. G. Lins Soil fertility, Brasilia (PhD-Brazil)
Amilcar Ubiera Soil mineralogy, Raleigh (PhD-Dominican
Helmut Elsenbeer Soil physics, Pichis (PhD-Germany)
Jose R. Davelouis Soil fertility, Yurimaguas (PhD-Peru)
* Resigned during the year.

Cheryl A. Palm Soil agroforestry, Yurimaguas (PhD-USA)
Lawrence T. Szott Soil agroforestry, Yurimaguas (PhD-USA)
Erick C. M. Fernandes Soil agroforestry, Raleigh (PhD-Kenya) Carlos Castilla Soil pastures, Raleigh (PhD-Colombia)
Christopher W. Smith Soil classification, Raleigh (PhD-USA) Hadjrosuboto Subagjo Soil classification, Indonesia (PhDIndonesia)
Victor Ngachie Soil fertility, Raleigh (MS-Cameroon)
Eleazar Salazar Soil classification, Raleigh (MSVenezuela)
Eduardo Uribe Soil fertility, Raleigh (PhD-Colombia)
Jane Mt. Pleasant Weed control, Raleigh (PhD-USA)
Jonathan Hooper Soil classification Raleigh (MS-USA)
Marisa R. Fontes Soil chemistry, Raleigh (PhD-Brazil)
Abdul Karim Makarim** Soil fertility, Indonesia (PhDIndonesia)
Mwenja P. Gichuru** Soil fertility, Yurimaguas (PhD-Kenya) Laurie J. Newman** Soil classification, Raleigh (MS-USA) Robert H. Hoag** Soil classification, Raleigh (PhD-USA)
Victor Palma* Head of INIPA (and TropSoils Board
Lander Pacora Head of INIPA (and TropSoils Board
* Resigned during the year.
** Completed degree in 1986.

Manuel Villavicencio Director, Yurimaguas Station Angel Salazar Agroforestry, Yurimaguas
Jorge M. Perez Agrogorestry, Yurimaguas
Pedro 0. Ruiz Mycorrhiza specialist, Yurimaguas
Luis A. Arivalo* Soil fertility specialist, Yurimaguas
Beto Pashanasi Soil zoology, Yurimaguas
Miguel Bustamante Training Officer, Yurimaguas
Rolando Dextre* Pastures specialist, Yurimaguas
Daysi Lara Pastures specialist, Yurimaguas
Marco Galvez Corn-weed control specialist,
Andres Aznaran Tillage specialist, Yurimaguas
C~sar Tepe Paddy rice specialist, Yurimaguas
Alfredo Racchumi Uplandrice specialist, Yurimaguas
Jonathan L6pez Corn and sorghum specialist, Yurimaguas
Mercedes Escobar Economist, Yurimaguas
Wilfredo Guillen Grain legumes specialist, Yurimaguas
Jorge W. Vela Pastures specialist, Pucallpa
Luis Zui'ga Soil physicist, Pichis
Jos4 Merino Head, La Esperanza Station, Pichis
Hemilce Ivazeta Pastures specialist, Tulumayo, Tingo
Marta Gallo Soil specialist, Tulumayo, Tingo Maria
Rodolfo Schaus Pastures specialist, Pucallpa
Miguel Flores Farming systems specialist, Tulumayo,
Tingo Maria
* Resigned during the year.

Jorge Fuigueroa Farming systems specialist, Tulumayo,
Tingo Maria
Juan Lermo Agronomist, Proyecto Especial Pichis
Palcazu, Puerto Bermudez
R. Rufz Pastures specialist, Proyecto Especial
Pichis Palcazu, Puerto Bermudez G. Cantera Pastures specialist, Proyecto Especial
Pichis Palcazu, Puerto Bermudez E. Acuia Pastures specialist, Proyecto Especial
Pichis Palcazu, Puerto Bermudez
Erci de Moraes Chief UEPAE de Manaus Station
Manoel S. Cravo Soil fertility specialist, Manaus
Mohammed Sudjadi Director, CSR, Bogor
I. P. G. Widjaya Adhi Soil fertility, Country Coordinator, Bogor
Sri Adiningsih Head, Soil fertility division, Bogor
Fahmuddin Agus Soil physicist, Sitiung
M. Heryadi Soil fertility, Sitiung
Al-Jabri Soil fertility, Bogor
Putu Wegena Soil fertility, Jambi

Tropical soils research has progressed to the point that several management options for sustainable productivity in agronomic and ecological terms are ready to be widely tested by national research institutions. The different options for the humid tropics constitute the model for networking (see first report). A training workshop, held in Spanish at Yurimaguas during September 1986, provided on-the-job training for 31 front-line professionals from the planting stick to the computer. The workshop
participants created RISTROP (Red de Investigaci6n de Suelos Tropicales) with core experiments on low-input systems, agroforestry, continuous cropping, legume-based pastures, and paddy rice. This network is now operating in 11 Latin American countries. Continuing technical backstopping is being provided to IBSRAM (International Board for Soil Research and Management) in the establishment of two Acid Tropical Soils Networks in Asia and Africa. Given the striking similarities in soil constraints between tropical America and tropical Africa, a similar workshop is being planned in coordination with IBSRAM to train African soil specialists at Yurimaguas and establish a viable link between soil management technologies generated in Latin America and its potential users in Africa.

Yurimaguas Workshop
T. Jot Smyth, N. C. State University, Raleigh Josg R. Benites, N. C. State University, Yurimaguas, Peru Dale E. Bandy, N. C. State University, Yurimaguas, Peru Pedro A. Sanchez, N. C. State University, Raleigh
Tropical soils research has progressed to the point of grouping promising alternatives into soil management options that account for differences in physical and socioeconomic conditions within this ecosystem (Figure 1). Many of the key components for these technologies can now be transferred to national research institutions, allowing local investigators to adapt soil management options to their specific conditions. A concerted validation and extrapolation effort across tropical soil ecosystems in Latin America would not only encourage interaction among participating institutions but also identify refinements and modifications that should be pursued to improve existing management options. Such an effort requires the identification and training of capable on-site personnel at collaborating national institutions.
In September 1986, North Carolina State University's TropSoils conducted a 21-day workshop on tropical soils management at its primary research site in Yurimaguas, Peru, in cooperation with INIPA, USAID/Lima, and the Interamerican Institute for Cooperation in Agriculture (IICA). Purposes of the workshop were to acquaint key Latin American scientists with the most recent techniques in tropical soil characterization and management and to identify common interests and establish a soil research network. Criteria for selection of participants were based on the national research institutes': (a) interest and capabilities in pursuing soil management research in tropical ecosystems and (b) designation of workshop candidates, with at least a B.S. in agronomy or equivalent training, as the personnel responsible for conducting network investigations resulting from this workshop. The 31 participants who attended the workshop represented a total of 23 potential research sites distributed among 15 national institutions in 10 different countries (Table 1). A detailed description of the workshop and the experiments developed for the research network can be found in a

report to the U.S. Agency for International Development available in English and Spanish.
Activities were organized around the two workshop objectives and occurred simultaneously during the 3-week schedule. The sequence of topics
covered during the instructional component included: characterization of tropical ecosystems; diversity, classification, and. taxonomy of tropical soils; soil physics in relation to land clearing and tillage systems; and fundamental aspects of soil chemistry, soil fertility, soil testing, and plant analysis. Emphasis was placed on field activities that gave participants hands-on experience with the most recent field and laboratory techniques and computer software used in soil science research. Technology developed for humid tropical soil management was discussed as five distinct
packages: mechanized high-input continuous crop production, low-input crop production with acid-tolerant species, agroforestry systems, paddy rice production on alluvial soils, and legume-based pastures. Approximately one full day was devoted to field tours of experiments for each management option.
During their 3 weeks in Yurimaguas, participants installed a low-input experiment on a secondary forest site. This activity acquainted the group with the procedures, measurements, and decisions to be made during the
processes of site selection, forest clearing, soil and vegetation characterization, plot establishment, and planting the initial crop. The group performed nondestructive forest biomass measurements, burned the
slashed vegetation, collected ash and post-burn soil samples, and analyzed the ash and soil for nutrient content in the laboratories. Before
departing, participants were able to observe their experiment with an initial stand of upland rice.
Considerations for network development were initiated by participant presentations geared to provide opportunities for interaction and identification of common interests. Each member described the ecosystem, facilities, research program, and limitations of the experimental site where they performed studies for their national institutes. Comparative data among the participants' research sites (Table 2) indicate a broad spectrum of research thrusts and ecosystems.
Toward the end of the workshop, participants voluntarily chose to participate in individual working groups on each of the five soil management

options. Each working group was requested to (a) identify soil management factors that should be investigated in an extrapolation and validation network, (b) design experiments with clearly defined objectives, and (c) define experimental methodologies and basic requirements of equipment and facilities. A brief description of the network studies developed by each working group follows.
Low-input option
Obj ectives
1. Compare the sequence in which nutrient constraints appear in acid
soils, under different ecosystems, under upland rice-cowpea
2. Compare the effects of nutrient additions by ash from burning
different types of standing vegetation in different climatic
3. Establish soil nutrient levels to aid in formulating minimum
fertilizer recommendations for sustained upland rice-cowpea
production in acid soils.
Experimental approach
A total of 13 fertilization treatments, in a randomized complete block design with four replications, were developed as common to all network sites. Both udic and ustic moisture regimes exist among sites, and vegetation varies from primary rainforest to savanna. Post-clearing management will be constant among all
High-input option
Obj ectives
1. Evaluate, under continuous cultivation, crop responses to
increasing rates of K fertilization;
2. Evaluate the effects of crop residues on soil K dynamics;
3. Evaluate crop response to residual fertilizer K;
4. Determine the influence of K rates on K interactions with Ca and
Mg in soils and plants.
Experimental approach
Crop rotations will be corn-soybean or corn-cowpea. Yield

systems. Annual crops will be excluded from the experiment for
steep topography.
Paddy rice option
Obj ectives
1. Investigate alluvial soil management systems for paddy rice
2. Evaluate over time changes in soil nutrient availability in
alluvial soils under paddy rice production.
Experimental approach
Paddy rice production has not been practiced in large areas of the Amazon. Local expertise, therefore, is almost nonexistent.
Farmer acceptance of the system in the lower Amazon Basin is
believed to depend on the demonstration that (a) labor-intensive dike formation persists after seasonal river flooding, (b) constant flooding of rice paddies reduces weeding, and (c) broadcast seeding is a viable alternative to labor-intensive
After reviewing the proposals, participants were asked to identify, on a priority basis, the top three (if any) experiments they considered most applicable to their research station activities. Responses suggested primary interest in the low-input and agroforestry soil management options. This workshop activity provided valuable feedback information to the TropSoils program in Latin America, especially because these opinions were provided by professional field scientists after an intensive review of the primary research site program.
Participants agreed that a common goal of the network would be the transferral and validation of improved soil management technologies on acid humid tropical soils in Latin America. The group requested that North
Carolina State University coordinate information exchange and technical backstopping among network participants.
The quality of the projects developed by the group suggests that the workshop was successful in enhancing the soil management research capabilities of collaborating country personnel. Workshop program
evaluations by the participants highlighted the feasibility of using the

response to K will be evaluated over six K rates, ranging from 0 to 250 kg K/ha. Crop residue effects and interactions with Mg will be evaluated in four additional treatments. Potassium will
be monitored in both the soil profile and plant tissue.
Improved pastures option
Obj ectives
1. Determine the appropriate method for renewing pasture productivity
through legume incorporation;
2. Determine the effects of P fertilization on legume establishment
in pastures;
3. Evaluate the persistence of legume-grass associations as a
function of establishment treatments.
Experimental approach
The experiment will be conducted in degraded pastures and will be composed of two distinct phases: (a) legume establishment as a function of tillage and P fertilization and (b) P dynamics and legume persistence as a function of animal grazing. The initial phase will be conducted in a network experiment with three P rates, two tillage methods, and four legume species in a splitsplit plot design with three replications.
Agroforestry option
Obj ective
Improve soil fertility and control soil erosion and weed incidence
by improved fallows and tree crops in agroforestry systems.
Experimental approach
The group chose to develop one experiment for improved fallows and two experiments for tree crops, with a distinction in the latter f or steep and f lat topography. The improved f al low study will compare the effects of a selected tree + groundcover legume shortterm (3-5 years) fallow to a traditional secondary forest fallow (5-10 years) on the control of weeds, soil erosion, and soil fertility replenishment for subsequent crop production. The tree crop experiments will evaluate the productivity, nutrient distribution, and use of selected perennial crops in multistrata

Yurimaguas Experiment Station research program to provide scientists in the humid tropics with on-site exposure to knowledge of how to manage soils.
Table 1. Distribution of participants by country, national institutions,
and experiment stations at the Tropical Soil Management Workshop,
Yurimaguas, Aug. 31-Sept. 21, 1986.
National Research Number of
Country institution site participants
Bolivia IBTA Chapare 3
Brazil CEPLAC Itabuna 1
EMBRAPA: Bel6m 2
Manaus 1
Porto Velho 1
Rio Branco 1
Colombia UNIBAN Urabg 2
Costa Rica CATIE Turrialba 1
Univ. Costa Rica Rio Frfo 1
Dominican Republic ISA Santiago 1
Ecuador INIAP Pichilingue 1
Guatemala ICTA Izabal 2
Honduras Min. Rec. Naturales Danlf 1
Panama IDIAP Calabacito 2
Peru IIAP Iquitos 1
INIPA: Moyobamba 1
Iquitos 1
Pichis 1
Puerto Maldonado 1
Tarapoto 1
Tingo Maria 2
La Molina Satipo 1
Univ. Amaz. Peruana Iquitos 1

C14 cn Cl) C:> cv) C*4
C"s -"2 "IS -la
44 44
m m
ml Ell 0 0 EM Ej
1 44

Paddy Rice Continuous Cropping Low-Input Cropping Pastures
Agroforestry Forest/Farming Mosaic
Regenerating Slopes
Alluvial Soils Acid Soils Young Soils
Figure 1. Soil management options for sustainable production in the humid tropics used in the
research network.

Network Development in Latin America: RISTROP
T. Jot Smyth, N. C. State University, Raleigh
Participants requested that NCSU provide the central coordination for the network, which they chose to name RISTROP (Red de Investigaci6n de Suelos Tropicales). Assistance was requested for research site selection and characterization, support services for analyses and interpretation of resulting data, and information exchange between participating national institutions. Funding was not available to sponsor each participant's research activities in the network. Participants were therefore
individually responsible for obtaining approval and funding from their national institutes for their network activities. The limited budget available for network coordination also led to the stipulation that support services from NCSU's Tropical Soils Program would only be initiated for a network participant upon confirmation of the national institute's approval of network activities.
The network coordination has received positive responses from participants in 8 of 11 countries, for a total of 27 initiated or planned experiments, encompassing all management options presented during the Yurimaguas workshop (Table 1). With the exception of collaborators in Bolivia, participants have limited their commitments in 1987 to initiating network experiments, which they identified as first priority during the Yurimaguas workshop. Participants from IBTA/Bolivia intend to initiate experiments identified as both first and second priority for their institutional programs. INIPA stations in Puerto Maldonado, Moyobamba, and Iquitos plan to incorporate the low-input and agroforestry studies into experiments on native fruit tree production systems for peach palm, Brazil nut, and camu-camu. A similar approach is planned for the same network experiments at EMBRAPA/Manaus in guarana production systems. The Peruvian government recently established lime management as top research priority for the Selva region. High-input annual crop experiments at Tingo Maria and Moyobamba, therefore, will be directed toward comparisons of yield responses to locally available lime sources.
RISTROP collaborators in Ecuador recommended delaying network

initiatives in their country until funding is available through the recently established agricultural research foundation. Despite several inquiries, no response has been received on the network status in the Dominican Republic. A similar situation in Honduras was transformed into a positive commitment after discussions with the participant's superiors during recent travel to Central America. Additional correspondence with Colombian scientists, who were unable to attend the workshop, may result in implementation of two studies in that country. Venezuela also was not represented at the
Yurimaguas workshop; however, after reviewing the workshop report, collaborators from the Universidad Central de Venezuela notified RISTROP coordination of their intention to participate in three network experiments.
Since the completion of the Yurimaguas workshop, NCSU support
activities for the network have centered on technical visits to
participating institutions during implementation of field experiments. In addition to assisting participants in adjusting methodologies and field plans to local conditions, this action has also provided an opportunity to obtain on-site familiarity with the research programs of the national
institutions. Extensive discussions with participating network scientists and travel conducted to date have provided the following observations:
The specific needs for technical backstopping by national
institutions and the capacity for NCSU's Tropical Soils Program to provide this expertise extends beyond the existing scope of the budget
and/or conceptual development of the research network. The type of required technical support varies among institutions from assistance in
establishing functional soil testing laboratories to the identification of research priorities through interpretation of existing soils information. Such limitations often impede participants' abilities to implement knowledge gained during the Yurimaguas workshop on a broader
institutional scale.
Ongoing national institute research programs, in some of the
visited countries, are often unrelated and nonsupportive to the USAID Mission agricultural development programs. Quite often network
participants and their superiors have indicated unfamiliarity with ongoing USAID Mission programs. Synchronization of national institute and USAID Mission programs would capitalize on the investment made,

thus far, to transfer soil management technology to the network
Although it is not anticipated that soil science expertise in national
institutions will be fully implemented through participation in RISTROP, it is fitting to consider supportive measures which will ensure that experiences gained in the network will be maintained and used in future national research endeavors.
Table 1. Current stage of developments for RISTROP experiments in each
participating country.
Annual crops
Low High Agrofor- Improved Paddy
Country Institution input input estry pastures rice
Guatemala ICTA I
Honduras Min. Recursos Naturales I
Costa Rica Univ. C. Rica/CATIE I
Panama IDIAP A I
Dom. Repub. ISA P
Peru INIPA/Puerto Maldonado A A
INIPA/Tingo Maria A
INIPA/Iquitos A A
INIPA/Moyobamba A A A
INIPA/Yurimaguas I
Bolivia IBTA/Chapare I I I
EMBRAPA/Porto Velho I
Colombia ICA P
Com. Esp. Guaviare P
Venezuela Univ. Central Venezuela I I I
Initiated (I) 6 2 1 2 2
Approved Plan (A) 6 2 4
Potential (P) 3 2

Network Development in Africa and Asia
Pedro A. Sanchez, N. C. State University, Raleigh T. Jot Smyth, N. C. State University, Raleigh Stanley W. Buol, N. C. State University, Raleigh
TropSoils and IBSRAM signed a memorandum of understanding in which both institutions formally agreed to work together toward the development of an Acid Tropical Soils Network on a worldwide basis. TropSoils input has concentrated on providing technical leadership through the Network Coordinating Committees.
The Inaugural Workshop of the Acid Tropical Soils Network was held in Yurimaguas, Peru, and in Manaus and Brasilia, Brazil, from April 24 to May 3, 1985. It was organized by TropSoils/NCSU and co-sponsored by INIPA and EMBRAPA with support from various international organizations and donor agencies. After several days of observing ongoing long-term research in the humid tropics and acid savannas of South America, representatives from 13 developing countries (Brazil, Cameroon, China, Congo, Ivory Coast, Madagascar, Malaysia, Mexico, Panama, Peru, Thailand, Venezuela, and Zambia) decided to form the Acid Tropical Soils Network. The participants identified a defined target area, six principal research-validation topics, and several support services.
The Inaugural Workshop Proceedings provide a state-of-the-art review on management of acid tropical soils and its publication, edited by TropSoils/NCSU, is expected in early 1987. The proceedings may serve as the conceptual base of the network.
The first IBSRAM African regional workshop was held in Douala, Cameroon, on January 21-27, 1986. The five original African countries represented in the Inaugural Workshop were joined by Rwanda, Burundi, and Nigeria. A total of 55 individuals from 15 countries participated, under the sponsorship of the Cameroonian Ministry of Higher Education and Research, with several donor inputs. The five original countries have all initiated activities without waiting for additional funds. Sites have been selected for new experiment stations in Cameroon and Congo, where several of the humid tropical soil management options seen in Yurimaguas are planned to be implemented. Zambia

began implementing many of the ideas gathered after visiting Yurimaguas, Manaus, and Brasilia. Proposals for new sites were identified by representatives for Madagascar, Ivory Coast, Nigeria, Rwanda, and Burundi.
Common methodologies for evaluating edaphic parameters were agreed upon at the Cameroon Workshop and are shown in Table 1. To assist in project development, a list of equipment, supplies, and reagents for a fullyoperational laboratory to analyze the agreed-upon edaphic parameters was prepared by TropSoils/NCSU and submitted to IBSRAM headquarters. A second African meeting scheduled for April 1987 in Lusaka, Zambia, is expected to produce concrete experimental designs.
The first IBSRAM Regional Workshop on Soil Management under Humid Conditions in Asia was held at Khon Kaen and Phitsanulok, Thailand, from October 13 to 20, 1986, with 82 participants from 17 countries present. It was co-hosted by Thailand's Ministry of Agriculture and Cooperatives and IBSRAM, and was funded primarily by the Asian Development Bank (ADB) and the Australian Council for International Agricultural Research (ACIAR). A total of eight countries expressed interest in joining the Acid Soils Network. Included in the list are the three Asian participants at the Inaugural Workshop--China, Malaysia, and Thailand--and five additional countries-India, Indonesia, Philippines, Vietnam, and Western Samoa. The common theme was the limited knowledge base on how to produce food crops on acid upland soils of tropical Asia.
The participants agreed on two overall types of research activities, "core" experiment and component research, in addition to site characterization, a common activity of all networks. The core experiment is designed to compare current acid upland soil management practices with (1) low-input systems based on acid-tolerant cultivars, low levels of added P, and no change in the soil's acidity and (2) an intensive system with liming to neutralize exchangeable Al and appropriate fertilizer practices for cropping systems based on corn, soybean, or cotton. Although the specific cropping system will vary with site, the common thread will be the monitoring of soil dynamics as proposed at the Inaugural Workshop.
Component research projects focus on (1) liming, (2) screening of acidtolerant species and varieties, (3) residual effects of P fertilization, (4) organic inputs, initially focusing on determining the nutrient contents of composts, green manures, or animal manures, and (5) Fertility Capability

Participants in the African network have expressed to IBSRAM particular concern about the need to increase their expertise in acid tropical soil management techniques among their front-line scientists. Based on the recent success of a similar activity for RISTROP participants, and on the African network emphasis on technologies developed in the TropSoils program, IBSRAM has requested NCSU's assistance with on-the-job training of soil scientists who will be doing the work in the Acid Soil Management Network. The
experience accumulated in Latin America by the TropSoils program and the commonality of interests with IBSRAM in transferring such knowledge to African scientists makes it fitting that assistance be provided toward network implementation through a workshop/training course at the primary research site in Yurimaguas, Peru.
Acid tropical soils comprise approximately 1.7 billion ha of land area in 72 developing countries. Their geographical concentration is primarily in least-developed Third World regions, many of which are currently undergoing social unrest and face several food shortages by the next decade. Thirtyseven of these countries are now involved in RISTROP and IBSRAM and tropical soils networks (Figure 1). The network support activities offer the opportunity for concerted worldwide efforts in disseminating existing information and developing local expertise on management of this fragile ecosystem in needy countries. Feedback from validation and extrapolation of existing information would enhance TropSoils' role in the discrimination of agronomically and economically sound acid soil management technologies through the identification of refinements and modifications for ongoing research.

Table 1. Edaphic paramenters to be measured in IBSRAM Acid Tropical Soils
Network Experiments, as agreed in the Cameroon Seminar, January,
Parameter 0-10 10-20 20-50 Method
---------cm---------pH (H20) Yes Yes Yes 1:2.5 H20
pH (KC1) Yes Yes Yes
Exch. Al Yes Yes Yes
Exch. Ca Yes Yes Yes iN KCl
Exch. Mg Yes Yes Yes
Exch. K Yes Yes Yes Modified Olsen
Avail. P Yes Yes No "
Avail. Zn Opt.* Opt.* No "
Avail. Fe Opt.* Opt.* No "
Avail. Cu Opt.* Opt.* No "
Avail. Mn Opt.* Opt.* No "
ECEC Yes Yes Yes Exch. Exch. Exch. Exch.
Al + Ca + Mg + K
Al sat. Yes Yes Yes Exch. Al = ECEC x 100
P sorption Yes Yes No Fox and Kamprath, Juo and Fox
Org. C Yes Yes Yes Walkley Black
pH (NaF) Yes Yes Yes If allophane is suspect
ZPC (Zero Yes Yes Yes If subhorizon approaches
point of acric properties
Slope %
Bulk density Yes Yes Yes At planting time
1/3 bar H20 Yes Yes Yes
*Optional. Mn not optional where deficiencies or toxicities suspected.

---F - -----i--~c~r. --- --- -t-#munsnAT SCALE AONGmQT
Figure 1. Countries involved in acid tropical soils research networks supported by TBSRAN and TropSoils.

Cattle grazing for beef and milk production is one of the major land use activities of cleared rainforest areas in Latin America. We continue to find that when they are well managed, legume-based pastures protect 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 productivity. Several million 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 and Pucallpa, Peru, has been closely integrated with the Tropical Pastures Program of the Centro Internacional de Agricultura Tropical (CIAT) and with INIPA's National Selva Program, which is now conducting most of the agronomic studies. Research reported here is based on an overall long-term strategy shown in Figure 1. All but the last stage are fully implemented.
Long-term grazing studies show that high animal production levels can be sustained with three widely differing pastures in acid Ultisols with low inputs: a mixture of two stoloniferous grass and legume species (Brachiaria humidicola/Desmodium ovalifolium), a mixture of two erect grass and legume species (Andropogon gayanus/Stylosanthes guianensis), and one pure legume pasture of Centrosema pubescens. Other mixtures are failing because of low quality problems. After 4 to 6 years of continuous grazing, soil physical properties remain good and chemical properties have improved, because more than 80% of the P, K, and Ca applied as fertilizer is recycled to the soil. This year we obtained the first estimate of the N contribution of legumes to the associated grass under grazing in highly acid soil: the legume
contributed an equivalent of 150 kg N/ha of urea nitrogen. Potassium
dynamics, a Key issue for pasture persistence, continues to be quantified, and S accumulation in Ultisol subsoils was confirmed.
Perhaps the most exciting accomplishment of this year was the successful transformation of degraded steepland pastures into highly productive ones by

establishing grass and legume species with minimum tillage and phosphate rock in slopes ranging from 20 to 80%. In areas where herbicides are available, the proper combination of tillage and herbicide was determined to eliminate undesirable species and plant improved ones.

Figure 1. Strategy for developing the pasture soil management option for the humid tropics.

Persistence of Grass-Legume Mixtures under Grazing
Miguel A. Ayarza, N. C. State University, Yurimaguas, Peru Rolando Dextre, INIPA, Yurimaguas, Peru Pedro A. Sanchez, N. C. State University, Raleigh
The central experiment for the legume-based pasture soil management option is now in its sixth year of grazing. Its objectives are (1) to measure pasture and animal productivity in different associations, in terms of daily weight gain and annual liveweight production, (2) to evaluate the compatibility and the persistence of the different grass-legume mixtures under grazing, and (3) to evaluate changes in soil properties as a consequence of long-term pasture production.
Four associations remain unchanged, but during the 4 years the
project has been in progress, Panicum maximum + Pueraria phaseoloides was replaced by Andropogon gayanus + Centrosema macrocarpum 5056 in October 1984. A sixth association, Brachiaria dyctioneura + Desmodium ovalifolium, was established at a separate location with grazing initiated in March 1986. The species previously reported as Centrosema hybrid 438 was reclassified by plant taxonomists as Centrosema pubescens 438. The main features of the experiment are shown in Table 1.
Annual Production and Botanical Composition 1985-1986
Annual liveweight gains per hectare (the measure of overall pasture productivity), individual animal daily gains (an estimate of pasture quality), legume content, actual grazing periods and stocking rates used are shown in Table 2 for 1985 and Table 3 for 1986.
Annual liveweight gains were generally better in 1985 than in 1986 due to a more favorable rainfall pattern in 1985 and a 3-month delay in grazing initiation in 1986. Stocking rates were adjusted according to forage availability and divided into two semesters (January to June as the wetter period, and July to December as the less wet period). Brachiaria
humidicola/D. ovalifolium mixtures produced very high annual liveweight gains (843 kg/ha), way above the other mixtures. This mixture produced high available forage throughout the year (Figure 1), which permitted a higher

stocking rate. Liveweight gains decreased in 1986, in spite of legume content similar to that in 1985.
The B. decumbens/D. ovalifolium mixture performed quite inferiorly, producing about half the daily animal gains than the previous mixture in spite of using identical stocking rates (Tables 2 and 3). Although forage availability was higher with B. decumbens than with B. humidicola (Figure 1), sharp animal weight losses were observed with the B. decumbens/D. ovalifolium mixture during the second half of the year. Individual animal gains dropped drastically from June 1985 and continued dropping until December when the animals lost more than 300 g/day (Figure 2). The same situation occurred at the same time in 1986, which was reflected in low individual animal gains. This problem may be due to photosensitivity, which sometimes occurs with B. decumbens, or to an unknown nutritional problem during the drier part of the year. Its solution is likely to require the collaboration of animal nutrition specialists.
The pure legume pasture C. pubescens 438 maintained good levels of animal production in both years, in spite of the lower amount of forage on offer than the mixed pastures (Figure 2), which resulted in no weight losses during the drier period. Since forage quantity is lower than in the previous mixtures, the higher quality of C. pubescens 438 over D. ovalifolium (Table 4) is considered responsible for its good performance.
The mixture of erect species A. gayanus with Stylosanthes guianensis performed remarkably well in 1985 and 1986, particularly in terms of daily gains, suggesting a higher quality of the legume component (Table 4). Promising results were obtained in A. gayanus + C. macrocarpum 5065. Excellent individual animal gains were recorded during 1985 and 1986. Since this pasture has only been grazed since May 1985, it is not possible to compare it with the older mixtures, but it certainly shows good promise.
In sharp contrast, the new mixture established in 1985 which began grazing in March 1986 (B. dyctioneura/D. ovalifolium) performed very poorly even during the first year (Table 3). This mixture showed an excessively high content of legume in the forage on offer. This may significantly reduce animal performance due to the low quality of the legume, in comparison with the Centrosema species (Table 4). This mixture was eliminated from the trial in 1986. This is disappointing because this species is being considered an alternative to the other Brachiarias in the ustic tropics. Apparently this

is not the case in Yurimaguas with a udic soil moisture regime.
Long-term Trends in Persistence and Productivity
A summary of the 6 years of grazing was presented at the American Society of Agronomy meetings in November 1986. Its summary gives for the first time a vision of pasture persistence. Only the four most promising mixtures are included in this analysis. The overall summary (Table 5) indicates that there are several widely different options, all with very high animal productivity potential (annual liveweight gains on the order of 500 to 600 kg/ha/yr). Considering that cattle production from unimproved pastures is of the order of 50-100 kg/ha/yr, each mixture is very attractive. The quality problem of B. decumbens/D. ovalifolium definitely puts it at a disadvantage in relation to the other three.
Each of the remaining three represents a different approach: a mixture of creeper species (B. humidicola/D. ovalifolium), a mixture of erect species (A. gayanus/S. guianensis), and a pure, high-quality legume pasture (C. pubescens 438).
The yearly fluctuation in the above parameters gives a feel for pasture stability. Annual liveweight gains fluctuated considerably, but the greatest fluctuations were observed in the erect mixture (Figure 3). The legume content tended to fluctuate less than liveweight gains (Figure 4) and showed little relationship with annual liveweight gains. Legume content seldom dropped below 20% of the forage on offer, a level below which is considered undesirable. Daily liveweight gains (Figure 5) showed less yearly fluctuations than the previous two parameters. Nevertheless, these yearly averages include sharp weight losses in the B. decumbens/D. ovalifolium mixture during the last 2 years. Overall, Figures 3, 4, and 5 suggest a reasonable degree of stability in the other three mixtures.
Long-term Changes in Soil Properties
Measurements taken in December 1985 indicated a marked decrease in water infiltration rates in most of the pastures as a result of soil compaction produced by animals after 5 years of grazing (Table 6). Lowest values were found in A. gayanus/S. guianensis and A. gayanus/C. macrocarpum. This may be associated with the erect growth of these species so that they do not cover

the soil as do the other pastures. Statistical analysis, however, showed that infiltration differences among pastures were not significant, due to the high variability of the double ring infiltrometer measurement.
The infiltration rate prior to the start of grazing in November 1980 was 12.7 cm/hr with a range of 6.3 to 19.8. Five years later the average infiltration was 4.1 cm/hr with a range of 1.0 to 10.4 (Table 6). There is no question, therefore, that 5 years of trampling have decreased infiltration rates in this sandy loam Ultisol. The magnitude of the decrease has not produced visible runoff or erosion. This is because rainfall intensity values average much less than 41 mm/hr (see Continuous Cropping section) and also because these.pasture provide a year-round full plant canopy that protects the soil from raindrop impact.
Soil Organic Matter
Topsoil (0-20 cm) organic matter and total N for each pasture were determined in December 1985, and values were compared to those obtained 5 years before the initiation of grazing (Table 7). These data compare the overall levels of the experiment prior to grazing with the effects of individual pastures, and not on a plot-per-plot basis. In spite of this limitation, it appears that organic matter and total N contents either increased or remained the same during 5 years of grazing. The differences between pastures cannot be explained in terms of legume content, since both the highest and lowest levels were observed in mixtures with D. ovalifolium. The lowest levels were observed with B. humidicola/D. ovalifolium, the most productive pasture. Topsoil organic matter contents were maintained under this management option (Table 7).
Soil Fertility Paramenters
Status of soil chemical properties in the 0-20 cm depth is reported in Table 8. Topsoil pH and P values were higher in 1985 than in 1980 in most of the pastures. Acidity decreased greatly in topsoils under B. decumbens/D. ovalifolium. Additional soil characterization at the 0-5 and 5-20 cm depths on B. humidicola/D. ovalifolium showed that most changes observed in topsoils were restricted to the 0-5 cm soil surface layer (Figures 6, 7, 8, 9). This may indicate that nutrient accumulation and cycling occurs mainly at the soil 0-5 m layer.
Nutrient Cycling
The improvements in soil properties in the top 20 cm can probably be

attributed to the original and annual maintenance fertilization schedule shown in Table 1. It is noteworthy to observe that the annual maintenance fertilization was not applied in November 1983. The total amounts of P, K, and Ca added as fertilizers and lime during the 5-year period were compared by the amount calculated to be removed by animals in terms of annual liveweight gains. The results, shown in Table 9, indicate that 80% of the P, 98% of the K, and 92% of the Ca added as fertilizer was recycled to the soil in the B. decumbens/D. ovalifolium mixture. This high level of recycling is one of the real advantages of grazed grass-legume pastures, where the nutrients exported on the hoof are very low.
Table 1. Main features of the central, legume-based pastures experiment
Design: Randomized complete block, 2 replications Plot size: 0.45 ha
Pasture established: Sept. 1979 Stocking rate: 3.3-5.5 an/ha (150-kg Cebu steers) Grazing management: Continuous from Nov. 1980 to July 1981 Alternate 28-35 day cycles since July 1981 Fertilization:
22 kg P/ha as SSP (yearly) 42 kg K/ha as KC1 (yearly)
10 kg Mg/ha as MgS04 (yearly)
500 kg lime/ha (once)
Soil: Typic Paleudult, fine loamy, siliceous, isohyperthermic
Initial topsoil (0-20 cm) properties. Sept. 1979:
Clay: 13% Al sat.: 78%
0.M.: 1.85% ECEC : 3.87 cmol/L
pH : 4.3 Avail P : 2.2 pg/g

Table 2. Animal production, grazing periods, and percentage of legume in five
pastures under alternate grazing in 1985.
Liveweight gains
Stocking rates
Grazing Grazing Rainy Dry Total Indiv. Legume
Pasture years period period period prod. prod. content
----an/ha---- kg/ha/yr g/an/day %
C. pubescens 438 4 Jan. 14- 4.4 4.4 510 342 100
Dec. 19
B. humidicola + 3 Jan. 14D.ovalifolium 350 Dec. 19 5.5 4.4 843 482 30
B. decumbens + 5 Jan. 14D. ovalifolium 350 Dec. 19 5.5 4.4 469 259 26
A. gayanus + 5 April. 1S. guianensis 134-186 Dec. 3 3.3 3.3 363 594 49
A. gayanus + 2 April 23C. macrocarpum 5016 Nov. 20 3.3 3.3 502 775 13
Table 3. Animal production, grazing periods, and percentage of legume in five
pastures under alternate grazing in 1986.
Liveweight gains
Stocking rates
Grazing Grazing Rainy Dry Total Indiv. Legume
Pasture years period period period prod. prod. content
---an/ha---- kg/ha/yr g/an/day %
C. pubescens 438 5 Mar. 18- 3.3 3.3 498 636 100
Dec. 4
B. humidicola + 4 Mar. 18D. ovalifolium 350 Dec. 22 5.5 4.4 460 336 34
B. decumbens + 6 Mar. 18D. ovalifolium 350 Dec. 10 5.5 4.4 223 170 22
A. gayanus + 6 Mar. 18S. guianensis 136-184 Dec. 10 3.3 3.3 436 544 34
A. gayanus + 3 Mar. 18C. macrocarpum 5056 Dec. 22 3.3 3.3 632 755 33
B. dyctioneura + 1 Mar. 18D. ovalifolium 350 Oct. 10 4.4 4.4 131 149 52

Table 4. Nutrient quality of leaf blades (Dec. 1985).
Species protein P Tanninsa
--------------------------------------------------------------------------------------- % ---------------------Legumes
D. ovalifolium 12.4 0.22 21.0
C. pubescens 24.8 0.27 2.5
S. guianensis 21.4 0.35 4.0
B. decumbens 11.3 0.,22
B. humidicola 10.6 0.22
A. gayanus 10.6 0.27
-----------------------------------------------------------------a. Analyzed in 1983. Vanillin-HCL method.
-----------------------------------------------------------------Table 5. Average annual productivity of legume-based pastures in the central
grazing experiment at Yurimaguas (1980-86).
Pasture Years of Stocking Total Indiv. Legume
mixture grazing rate prod. prod. content
---------------------------------------------------------------------------------an/ha kg/ha/yr 9/an/day %
B. humidicola/D.ovalifolium 4 4.6 671 419 38
B. decumbens/D. ovalifolium 6 4.7 532 324 35
C. pubescens 438 5 3.8 573 471 96
A. gayanus/S. guianensis 6 3.2 477 427 31

Table 6. Infiltration values in five pastures under grazing in Yurimaguas for
several years (mean of two replications and four observations per
replication of each pasture).
Years of
Pasture grazing Infiltrationa S.E. x
B. decumbens +
D. ovalifolium 5 4.71 a 1.20
B. humidicola + D. ovalifolium 3 2.00 a 0.58
C. pubescens 4 10.44 a 2.60
A. gayanus +
C. macrocarpum 1 1.96 a 2.82
A. gayanus +
S. guianensis 5 1.00 a 0.19
LSDO.05 = 10.4
a. Numbers followed by the same letter are not statistically significant at
P = 0.05.

Table 7. Changes in organic matter total nitrogen in the topsoil (0-20 cm) of
five pastures under grazing after 5 years (1980-1985).
-------------------------------------------------------------------------Total N
Pasture Organic matter SDa Amountb
-----------------%-------------- kg/ha
Before grazing (Nov 1980) 1.93+0.4 0.0707+0.016 1837
B. decumbens +
D. ovalifolium 3.35+0.4 0.0895+0.015 2327
B. humidicola +
D. ovalifolium 1.35+0.6 0.0705+0.412 1085
C. pubescens 1.95 +0.2 0.0856+0.0025 2225
A. gayanus +
S. guianensis 2.02+0.2 0.0770+ 0.005 2002
A. gayanus +
C. macrocarpum 2.14+0.2 0.0757+0.009 1969
-------------------------------------------------------------------------a. Mean of two composite samples (10 subsamples each). Rest of pasture is
the mean of two replications (10 subsamples each).
b. Value calculated assuming 1.4 g/cc in the 0-20 cm depth. Mean of three
composite samples (10 subsamples each). Mean + standard deviation.

Table 8. Status of soil fertility of the topsoil (0-20 cm) of five pastures
after 5 years of grazing in Yurimaguas.
Pasture Year pH O.M. Olsen P Ca+Mg K Al sat.
% ppm ----cmol/L---- %
B. decumbens + 1980 4.2 1.9 3.9 1.7 2.7 55
D. ovalifolium 1985 4.6 3.3 8.3 1.0 0.10 0.9 53
B. humidicola + 1980 5.2 1.7 2.2 2.4 1.7 42
D. ovalifolium 1985 5.0 1.3 6.4 1.2 0.12 1.3 49
C. pubescens 1980 4.9 2.2 8.6 2.1 2.5 54
1985 5.0 1.9 6.5 1.8 0.18 2.4 54
A. gayanus + 1980 4.6 2.1 5.8 1.7 2.4 56
S. guianensis 1985 5.3 2.0 7.9 1.6 0.11 1.8 51
A. gayanus + 1980 4.3 1.8 4.3 1.5 2.4 60
C. macrocarpuma 1985 5.1 2.1 1.0 1.1 0.07 2.2 65
a. A. gayanus + Pueraria phaseoloides for the first 3 years.
Table 9. Nutrient balance in B. decumbens/D. ovalifolium during 5 years of
Balance (1980-1985) P K Ca
---------kg/ha--------Added as fertilizers and lime 112 160 505
Removed by animals 22 6 39
Balance 90 154 466
% "recycled" 80 98 92

2 a B. decumbens + 0. ovalifolium
S5 o B. humidicolo + 0. ovalifolium
SC. pubescens CD 4 a
Jon Mar May Jul Sep Nov Feb Apr Jun Aug Oct Dec
Grozing periods (1985)
Figure 1. Seasonal changes in available forage on offer of
three pastures in 1985.
-7 o B. decumbens + 0. oval i fol lum
a-o B. humidicolo + 0. ovolifolium
S oa C. pubescens
o 800
U) 400 C 0
- -400 0o
Jan Mar May Jul Sep Nov Feb Apr un Aug Oct Dec
Grazing periods (1985)
Figure 2. Seasonal changes in individual liveweight gains
in 1985.

'M 800

c: 800
0 12 345 8 1 2 3 4 1 2 3 4 5 12 3 4 5 8
Bd/Oo Bh/Oo Cp Rg/Sg
Posture and grazing year
Figure 3. Yearly fluctuations in liveweight gains in four
rotationally grazed pastures at Yurimaguas. Bd =
B. decumbens; Do = D. ovalifolium; Bh = B. humidicola; Cp = C. pubescens; Ag = A. gayanus; Sg =.S. guianensis.
, 100
S 80
a) C
o- 60
0D 40
( 20
123456 1234 12345 123456
Bd/Oo Bh/Oo Cp Ag/Sg
Posture and grazing year
Figure 4. Yearly fluctuation in average legume composition of
the mixture. Bd = B. decumbens; Do = D. ovalifolium;
Bh B. humidicola; Cp = C. pubescens; Ag = A. gayanus;
Sg = S. guianensis.

'N1 600
- 123458 1234 12345 123456
-j Bd/Oo Bh/Oo Cp Ag/Sg
Posture and grazing year
Figure 5. Yearly fluctuations in daily liveweight gains.
Bd = B. decumbens; Do = D. ovalifolium; Bh =
B. humidicola; Cp = C. pubescens; Ag = A. gayanus;
Sg = S. guianensis.
Avoiloble P (ppm)
2 4 6 8 10
C) 20-40- SD 3.9
CL 40-60CD
Figure 6. Available P profile 5 years after grazing a B.
humidicola/D. ovalifolium mixture.

Exchongeable RI (cmol/L)
2 4 6 8 10
0-5I 1
) 20-40 LSO 0.09
d 0.05

C 40-60
Figure 7. Exchangeable Al profile 5 years after grazing a B.
humidtcoldD. ovalifolium mixture.
Exchangeable Co + Mg (cmol/L)
0.2 0.4 0.6 0.8 1.0 1 2
5-20- Mg
0 20-40- Cc
CL 40-60
Figure 8. Exchangeable Ca and Mg profiles 5 years after grazing
a B.humidicola/D. ovalifolium pasture.

Exchangeob I e K ( cmo I/L ) 050 0.02 0.04 0.06 0.08 0.10 0.12
5-20 SO 0.50
U 20-40
0 40-60C
Figure 9. Exchangeable K profile 5 years after grazing a B. humidicola
D. ovalifolium pasture.

Evaluation of Animal Preference in Small Plots
Miguel A. Ayarza, N. C. State University, Yurimaguas, Peru Rolando Dextre, INIPA, Yurimaguas, Peru
Observation of animal preference at early stages of evaluating new forage accessions is considered an important tool in selecting desirable species from both the agronomic and the animal standpoints. Palatability is crucial in the acid-tolerant tropical legumes we are working with. Factors dealing with quality and preference can hardly be assessed without animals.
To test whether differences exist in animal preference for 13 legume
accessions at Yurimaguas.
An old regional Trial B, for which agronomic evaluation concluded in 1985, was used in this work. The experiment was fenced in March and animals were introduced in May 1986. Accessions were arranged in randomized complete blocks with 3 x 4 m plot size per species. Three animals (180 kg liveweight) were observed 6 hr/day for 3 days in each replication. Position of the animals and time spent grazing a particular accession were recorded every 15 minutes.
Before entering the next replication, animals were kept out of the experiment on a pure stand of B. humidicola for 3 days in an attempt to diminish preference effects between replications.
The first evaluation showed average grazing time per species differed even within a genus. Centrosema macrocarpum 5065 was preferred over the other Centrosema accessions (Figure 1), and Centrosema macrocarpum 5052 was hardly consumed at all. Desmodium ovalifolium 366 was preferred over D. ovalifolium 350.
These results indicate that differences in animal preference for legumes

exist, even within the same genus and species. Although Centrosema species are known to be palatable, there are important differences in palatability.
The preference for D. ovalifolium 366 is probably associated with its lower tannin content than D. ovalifolium 350, as has been reported by CIAT. A simple animal preference test can be included at the earliest stages of germplasm evaluation to obtain a comparative idea of animal preference with known species. This procedure could save significant time and effort by
eliminating undesirable accessions before starting a grazing trial.
N -) 1000 -E
Cen trosema Cen trosemo Desmod ium Zo rn ia
macrocarpum pubescens ovalI i folI i um sp.
Figure 1. Animal preference for 12 legumes at Yurimaguas.
ta) CC) 0
-Cefltf-osera Centrosemo Desmodium Zorn jo
macrocarpum pubeSCenS ovaIi foIlium sp.
Figure 1I Animal preference for 12 legumes at Yurimaguas.

Nitrogen Contribution of Legumes in Mixed Pastures
Miguel A. Ara, N. C. State University, Pucallpa, Peru Jorge W. Vela, INIPA, Pucallpa, Peru
Pedro A. Sanchez, N. C. State University, Raleigh
Mixed grass-legume pastures are rare in the humid tropics. Most of the successful mixtures occur in the ustic soil moisture regimes. The success of some mixed pastures in such areas is often related to the ability of the legume to provide protein during the strong dry season when grasses cannot. In the humid tropics, where the grasses remain green throughout the year, the role of the legume is different. The purpose of this work is to ascertain the role of legumes under humid tropical conditions. A long-term grazing experiment was established at the IVITA Principal Tropical Station west of Pucallpa, Peru, in collaboration with INIPA and IVITA.
1. To estimate the N contribution of two adapted pasture legumes
(Centrosema pubescens and Desmodium ovalifolium) to their respective mixtures with Brachiaria decumbens in terms of N yield, N availability
to the grazing animals, legume litter accumulation, and N release;
2. To evaluate the effect of different grazing pressures on the mixture B.
decumbens/D. ovalifolium in terms of N availability to grazing animals,
legume litter accumulation, and N release.
A 6-ha degraded pasture was planted to improved pastures to compare the N-supplying capability of two alternative N sources (legumes vs. different levels of N-fertilizer) on a grazed Brachiaria decumbens pasture, in terms of N yield of the standing biomass, N availability to the grazing animals, legume leaf litter accumulation, and N release. An additional variable, high grazing pressure, was included in the legume mixture (normal = 4.5 kg available forage dry matter per 100 kg animal liveweight; high = twice the pressure). The experimental design was randomized complete blocks with five treatments and three replications. Grazing was initiated in January 1986. Replications were utilized as paddocks of a 15-day grazing, 30-day rest

rotation. Brown Swiss and Holstein-Cebu halfbreeds were used. Two
esophagus-fistulated steers were always kept in the same treatment; intact animals were used as grazers. Animals were weighed before entering each replication, but only to adjust the grazing pressure.
Slope was used as a blocking criterion. Replication I was located on 05% slope, and replications II and III were located on 75% slopes. Soil was a Pucallpa series Paleudult. Selected soil properties at the initiation of the experiment are given in Table 1.
Serious problems in establishing the Centrosema pubescens/Brachiaria decumbens mixture forced us to discard this treatment.
Forage Availability
Dry matter availability and botanical composition were evaluated using a double sampling dry weight rank procedure with 100 samples per 0.33-ha paddock just before grazing began. Dry matter yield and botanical
composition are shown in Table 2 and the corresponding ANOVA for dry matter available to the treatments is shown in Table 3. No response was obtained for dry matter availability to the treatments, although a trend of N response is evident. Legume contents are adequate, but hover at high grazing pressure, as expected.
Nitrogen Content and Nitrogen Yield
This year we were able to install a micro-Kjeldahl unit on the IVITA soils laboratory. Results of N content and yield are the first from this laboratory. Unlike dry matter availability, treatment effects on N content and N yield of Brachiaria decumbens were well expressed (Table 2). The N fertilizer effect on N content and N yield was linear with no quadratic component (Figures 1 and 2). The presence of the legume significantly increased the N content of the grass by an average of 44% (1.06 vs. 1.53% N). The effect of grazing pressure of mixed pastures on grass N content was not significant.
On the average, mixtures gave a higher grass N yield than grass alone with no N fertilizer, equivalent to 151 kg of N (from the regression equation). This amount is contributed by 28% of the legume component in the

high grazing pressure treatment and 35% in the normal grazing pressure treatment. These numbers do not include the N content of the legume per se. Consequently, they provide an early indication that legumes contribute the equivalent of about 150 kg N/ha/yr to the associated grass during the first 10 months of grazing.
Leaf Litter N Accumulation
Four 0.5 x 0.5 m squares were selected in each legume paddock to cover the range of legume composition. The squares were fixed with iron stakes, and the Desmodium ovalifolium leaf litter was collected, dried, and weighed. Leaf litter dry matter accumulation averaged 14 g/0.5 m2/18 months for the high grazing pressure treatment and 29 g/0.5m2/18 months for the normal grazing pressure treatment.
Nitrogen content of the leaf litter averaged 1.60%. Nitrogen
accumulation in litter form during this 18-month period averaged 4.6 kg N/ha for the high grazing pressure treatment and 10.1 kg N/ha for the normal grazing pressure treatment. Consequently, the lower the grazing pressure, the more important will be the N transfer through the litter. This effect is being measured concurrently with N intake by animals and in excreta to fully calculate the N contribution of legumes to mixed pastures.
Conclusions to Date
1. There is a strong linear response of Brachiaria decumbens grazed
pastures to N fertilization as high as 300 kg N/ha/yr.
2. Including D. ovalifolium in the mixture contributes about 151 kg N/ha/yr
to the grass, in addition to the legume's N content.
3. Leaf litter N accumulation in mixed pastures decreases with increasing
grazing pasture.
These first year data show the obvious need for some sort of N input in Brachiaria decumbens pastures established in land that previously had degraded pastures. Addition of a legume in the mixture apparently makes a large contribution to meet this need. Continuing to gather data will demonstrate the effect on animal production and elucidate some of the N transfer processes involved.

Table 1. Selected soil properties at the initiation of the
experiment. 0-15 cm. December 1985.
Rep Sand Clay pH P Al Ca Mg K 0.M. N sat.
---- %----- pg/ml ------ cmol/L--------------------I 25.4 25.2 4.6 5.4 2.3 1.83 0.66 0.17 2.33 0.09 48 II 44.3 25.2 4.5 5.7 1.8 1.90 0.66 0.14 2.34 0.10 40
III 51.9 22.2 4.7 4.1 2.1 1.68 0.54 0.11 2.19 0.08 45
Table 2. Forage on offer and botanical composition of B. decumbens pastures
fertilized with N or mixed with D. ovalifolium at two grazing
pressures. Mean of three replications.
N Grazing Forage dry
source pressure mattera Legume Gross N content
kg/ha/yr tDM/ha/cycle % % N kg/ha/cycle
0 Normal 1.34 1.06 a 140
150 Normal 1.56 1.42 b 222
300 Normal 1.66 1.85 c 308
Legume Normal 1.42 35 1.38 b 218
Legume High 1.38 28 1.58 b 230
a. Mean of seven grazing cycles of 45 days duration each.

Table 3. ANOVA for forage dry matter available.
Degrees of Mean
freedom Squares
Replications 2 46.3047 ns
Treatments 4 140.8177 ns
N linear 1 398.53 ns
N quadratic 1 21.12 ns
Mixtures vs. NQ 1 18.40 ns
HGP vs. LGP 1 5.23 ns

Y = 1.0483 + 0.0026 x
0 I t N Fertilizer N
Z 0 Legume-normal grazing pressure
0 Legume-high grazing pressure I I
0 150 300
N (kg / ha/year)
Figure 1. Effect of fertilizer N and legume presence
on N content of Brachiaria decumbens pasture
during the first grazing year in Pucallpa,
0- Y =0.8967 + 0.0028x
1 I r=O.99**
"" Fertilizer N
So0 Legume-normal grazing pressure
- 0 Legume-high grazing pressure
0 150 300
N (kg/ha/year)
Figure 2. Apparent N contribution of legumes in
relation to N accumulation by fertilized
Brachiaria decumbens pastures in
Pucallpa, Peru.

Potassium Dynamics in Legume-Based Pastures
Miguel A. Ayarza, N. C. State University, Yurimaguas, Peru Pedro A. Sanchez, N. C. State University, Raleigh
Tropical pastures on acid soils are stable and productive only when nutrients are sufficient to sustain a vigorous forage crop. Maintaining this fertility requires a management method that takes into account the nutrient leaching common in areas of high rainfall, 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, and one of our most promising mixtures (Brachiaria humidicola/ Desmodium ovalifolium).
1. To quantify leaching losses of K in pastures under clipping 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;
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 annual rates of K fertilization (0, 50, and 100 kg K/ha) applied only once by two stocking rates (3.3 and 6.6 animals/ha), with three replications. Two additional experiments were established on 3 x 4 m plots with K rates of 0, 25, 50, 75, and 100 kg K/ha/yr. The first, a clipping experiment in which some plots had clippings removed while others had clippings returned, provides a comparison of the effect of grazing on K dynamics. The second was a bare-plot
experiment designed to account for soil chemical and 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 Brachiaria humidicola and Desmodium ovalifolium in December 1984. Potassium treatments were applied on May 13, 1985, and grazing began on July 4 of that year and terminated 2 years

later. 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.
Soil characterization of the area showed very low levels of exchangeable K in the soil profile (0.05-0.06 cmol/L), except for the 0-5 cm layer, which averaged 0.16 cmol/L. Soil texture was classified as sandy loam with topsoil clay contents of 17%.
Application of K treatments produced a significant increase in exchangeable K in the 0-5 cm of the small plot experiment. Applied K, however, moved down the profile as a function of precipitation and K rates. There were significant changes in the 0-5 and 5-20 cm depths after 970 and 1678 mm cumulative rainfall, especially in the 300 kg K rate (Figure 1). The presence of plants significantly reduced the levels of exchangeable K in the soil, and this effect increased with the level applied (Figure 2). Preliminary results about the effect of K rates and plant residues on the productivity of the association indicated a positive effect of addition of residues on yields (Table 1). There was also a response to K on the cumulative yields in both treatments. There were no significant differences of the effect. The return of plant residues on the soil properties after 1 year is shown in Table 2.
After 255 days of grazing, the effect of K and stocking rates on aboveground biomass available for grazing are summarized in Table 3. There was an increase of forage dry matter in response to the treatments; however,
it was not significant, probably due to differences among grazing cycles. On-site studies of the effect of plant cover on soil moisture were started in September 1986. In addition, K in the soil solution is being measured in the small plot experiment.
Conclusions to Date
1. There was a movement of applied K in the soil used in the experiment;
however, the degree of movement depended on the rates applied and on

2. Plants seemed to be able to significantly reduce the amounts of K
susceptible to leaching by absorbing most of the available K in the
3. Plant residues appear to be an important component in the stability of
Table 1. Effect of the return of plant residues on the cumulative yields of
a mixture of B. humidicola + D. ovalifolium (sum of six cuttings). Applied K No residue return Residue returned Increase
kg/ha -----------t dry matter/ha----------- %
0 9.75 14.87 34
50 11.74 15.34 23
100 13.30 17.93 26
300 15.96 18.05 11

Table 2. Effect of plants and residue return on topsoil (0-5 cm) properties
under a pasture of B. humidicola + D. ovalifolium after 1 year
(mean of three replications).a
--------------------------------------------------------------------Plots P Al Ca Mg K
ppm ------------------cmol/L----------------Bare 10.5 b 2.77 a 0.54 a 0.20 a 0.08 a
Planted, no
returned 8.9 a 2.53 a 0.49 a 0.09 b 0.06 a
returned 7.8 a 2.63 a 0.54 a 0.15 a 0.07 a
--------------------------------------------------------------------a. Columns and rows with the same letters are not statistically significant
at P = 0.05.
--------------------------------------------------------------------Table 3. Effect of two stocking rates and three potassium rates on levels
of available forage before grazing (mean of five grazing cycles).
--------------------------------------------------------------------Stocking rate K applied Green dry matter forage Grass
--------------------------------------------------------------------animals/ha kg/ha t/ha/cycle %
3.3 0 3.04 46
50 3.73 51
100 3.85 50
6.6 0 3.26 50
50 4.20 54
100 4.48 58

Exch. K (cmol/L)
0 0.04 0.08 0.12 0.16 0.20 0.25
40 50 K (kg/ha)
60 Cumulative Rainfall
* 25 mm 0 970 mm 100 A 1687 mm
. I I I I I
E 40 100 K (kg/ha)
C. 60100- 0.77
300 K (kg/ha) 100
Figure 1. Effect of potassium additions and cumulative
rainfall on distribution of exchangeable K in
bare plots.

Exch. K (cmol/L)
0 0.04 0.08 0.12 0.16 0.20 0.25
40- 50 K (kg/ha)
, I I II I I
E 40- 100 K (kg/ha)
- 6 0
100- 0.30 0.34 0.38 040
I I l l i ff l/ I I I
40- 300 K (kg/ha)
* with plants o0 no plants 100
Figure 2. Effect of plants on the distribution of exchangeable
K in the profile at three potassium rates (mean of
seven sampling dates).

Sulfur Accumulation in Grazed Pastures
Miguel A. Ayarza, N. C. State University, Yurimaguas, Peru Pedro A. Sanchez, N. C. State University, Raleigh
Ultisols have increasing clay contents with depth and often accumulate S in their subsoil because of their higher S sorption capacity. This situation is well documented in the Southeastern United States and is therefore expected to occur in Ultisols of Yurimaguas.
Soil samples were taken from a Brachiaria decumbens/ Desmodium
ovalifolium pasture under grazing for 3 years, which had received a total of 116 kg S/ha from yearly additions of ordinary superphosphate and magnesium sulfate. The topsoil of a degraded and not fertilized nearby pasture was also sampled to serve as a control. Twenty subsamples per site were
composited and extracted with 0.01M Ca(H2P04)2 for So determination.
Sulfur distribution in the profile of the fertilized pasture is shown in Table 1. Applied S apparently moved down and accumulated in the subsoil as expected. Nevertheless, S concentrations in the topsoil of the fertilized pasture were significantly higher than those in the degraded pasture.
Responses to S applications were observed in a pasture of D. ovalifolium grown in an Oxisol in Carimagua, Colombia, when extracted SO4 was 12 ppm. Thus if S assessment of the fertilized pasture were based solely on S status of the topsoil, a probable response would be predicted. This should not be the case when the potential contribution from sorbed S in the subsoil is taken into account. In addition, grass and legume components of this pasture are Al-tolerant species with extensive root systems in the B horizon.
Use of the weighed profile mean to make better predictions of S responses in soils has been suggested, and a critical level of 4 ppm of SO4 below which S responses must be expected was established by Australian researchers. Sulfur requirements in Ultisols of the humid tropics should be established using this approach.

Table 1. Extractable sulfur in an Ultisol under pasture in Yurimaguas.
---------------------------------------------------------------------Site Sampling depth S04 a
----------------------------------------------------------------------cm PPM
Not fertilized 0-20 10.30 d
Fertilized 0-20 13.20 c
20-40 26.58 a
40-60 21.42 b
60-100 14.06 c
----------------------------------------------------------------------a. Figures with the same letter are not statistically significant at
P = 0.05.

Pasture Reclamation in Degraded Steeplands
Miguel A. Ayarza, N. C. State University, Yurimaguas, Peru Rolando Dextre, INIPA, Yurimaguas, Peru
There are several million hectares of degraded, unproductive pastures in the Amazon, often on steep slopes. The purpose of this project is to develop a simple technique for reclaiming degraded pastures in Ultisol steeplands, using different establishment techniques.
A two-factor experiment was installed in a degraded pasture occupying a 5.18-ha watershed with sideslopes of 20 to 50%. Treatments were established in an amphitheater fashion, following slope contours, with tillage methods as main plots and improved 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 pastures grow). The only fertilizer was Bayovar rock phosphate, applied at the rate of 12 kg P/ha in the hole or furrow.
The species included Brachiaria decumbens, Brachiaria humidicola, Desmodium ovalifolium 350, and Centrosema pubescens 438. The species were planted in rows 2 m apart from each other. Initial soil chemical and
physical properties are supplied in Table 1. The main degraded pasture species were of torourco complex. The experiment was initiated in June 1986. Improved grasses were planted using vegetative propagules and legumes using several seeds. After 1 month, a standardization cut was done on grasses and a minimum handweeding on the zero tillage plots.
Performance of species was determined on the basis of percent cover 2 months after planting and the production of biomass 6 months after planting. Cover was measured using quadrants 4 x I m in size. The number of introduced species was counted and the percent cover estimated. Biomass production was determined in a 4 x 1 m area in the center of each plot. Two rows and two strips were included in the sampling area. Results were expressed as fresh weight of planted species and percentage of the total biomass.

Soil physical properties appeared more limiting than chemical properties. Penetrometer resistance values indicated that a compacted layer was present in the 5-10 cm depth, perhaps due to overgrazing (Figure 1). Relatively high pH and Ca values (Table 1) were associated with the age of the pasture. The area was burned and cleared 3 years ago to favor the regrowth of natural grasses.
Fresh biomass production of the four species 5 months after establishment is shown in Table 2. The two grasses were successfully established without tillage while the two legumes responded significantly to minimum or total tillage. The ability of the improved species to take over the degraded pasture is shown in Figure 2. More than half the area was covered by the improved species in all cases except where the legumes were planted without tillage. Total tillage, done progressively in order to avoid contiguous areas of exposed surfaces, resulted in almost complete cover within 5 months.
These results suggest that minimum soil disturbance is needed to establish grasses such as those used in the experiment. The stoloniferous species are able to rapidly cover new areas and compete strongly with species already present. On the other hand, legumes require at least minimum tillage. In general, Centrosema performed better than D. ovalifolium, due to a faster growth habit and a more aggressive tendency of Centrosema than D. ovalifolium to cover.
A second phase of the study was initiated in February 1986 to determine whether the persistence of a species depends upon grazing.
The experimental area was divided into three paddocks, each containing the four species and the three methods of establishment. Two 150-kg steers started grazing by replication. Animal management was adjusted to give the animals the chance to consume all available forage (18 days grazing and 36 days resting). Persistence was monitored by using transects across the plots for each species x treatment combination before each grazing period. Results were expressed as percentage of presence of shoots of the species over the total number of countings every 50 cm along the transect.

After 6 months of grazing, the percentage of grasses increased, whereas that of legumes decreased (Figure 3). This is probably the result of the capacity of these Brachiaria species to compete and displace existing vegetation. Although the legume population is decreasing, an excellent stand of Centrosema is present.
Promising methods exist to establish improved grasses and legumes in degraded pastures. The results of this experiment indicate that simple establishment methods can be successful in compacted Ultisol steeplands but that some minimum tillage is needed to establish the legumes. Although minimum tillage was performed by a hand tractor, it is possible to replace it by using animal power.
Table 1. Initial soil properties of the steepland Ultisol area used for
pasture reclamation in Yurimaguas (mean of seven samples).
Depth pH P Al Ca Mg K Bulk density
ppm ---------cmol/L----------- gm/cc3
0-20 4.7 8.1 3.8 2.88 0.73 0.12 1.37
20-40 4.8 4.4 5.9 2.42 0.64 0.10
Table 2. Effect of three tillage methods on the production of green
forage of two grasses and two legumes after 5 months of
Tillage treatment
Species Zero Minimum Total
---------t fresh wt/ha--------Brachiaria decumbens 7.70 a 10.6 a 10.3 a
Brachiaria humidicola 6.67 a 10.7 a 12.4 a
Desmodium ovalifolium 0.32 a 1.78 a 4.65 b
Centrosema 438 0.76 a 4.41 b 2.72 b
a. Figures followed by the same letter ard not statistically different
according to Duncan's multiple range test (P = 0.05).

Mechoan i cal resistance (kg/cm2)
0 1 2 3 4
Q. 9
a) O
Figure 1. Soil mechanical resistance measured by a
penetrometer in degraded pasture prior to
treatment initiation (average of 36
observations per depth).

M o 0 decubens
I B. hunidicolo O Centrosemo 438 a-__ 1110. ovalifoliun
No tilloge Min. tillaoge Total tilloge
Land preparat i on
Figure 2. Effect of tillage method on percentage of
biomass produced by four introduced species
invading a degraded pasture.

0 Prior to grazing I Six months afterwards
'oo (February 1986)
o ;2 3
0 40
Q- 0. ...
C. pubescens 0. ovolifolium
I. Zero tilloge
2. Minimum tilloge
3. Conventional tillage
0 Prior to grazing I Six months afterwards
too (Februar I986) 3 2
B.deumen.B hmiicl
E 50
0 40r l
L 0
B. decumbens B. humidicoIo
I. Zero tillage
2. Mininumn tilloge
3. Conventional t I looe
Figure 3. Effect of grazing on persistence of two
grasses and two legumes established on a
degraded steepland area in Yurimaguas.

Pasture Reclamation via Herbicides
Jorge W. Vela, INIPA, Pucallpa, Peru
Miguel A. Ara, N. C. State University, Pucallpa, Peru
Another approach to pasture reclamation is the eradication of the unsuitable species using herbicides. This approach is important in the Pucallpa region of the Peruvian Amazon where chemical inputs are more available than in Yurimaguas.
To obtain the optimum combination of herbicide rate and time of application after tilling for effective weed control during the
establishment phase of Brachiaria decumbens and Andropogon gayanus.
A trial was carried out from February to July 1986 at the IVITA station near Pucallpa. Factors under study were three times of herbicide application and planting after land preparation (after 30, after 45, and after 60 days) and three glyphosate rates (1, 2, and 4 L/ha) as Round-up. The treatment structure was a 3 x 3 factorial. The experimental layout consisted of 2 x 6 m plots in a randomized complete block design with six replications. Variables measured were dry matter yield of Andropogon and Brachiaria at establishment (120 days of growth); weed reinfestation at 30, 60, 90, and 120 days after tilling; and pasture cover at the same times.
The experiment was laid out on a degraded native pasture. Selected soil properties at initiation of the experiment are shown in Table 1.
Predominant weeds before land preparation by rototilling and 30 days after tilling are shown in Table 2. Predominance of weeds changed from before land preparation: grasses comprised 70% of the weeds before but only 10% 30 days after tilling. Ciperaceae comprised only 1.6% in the early stage but had the highest percentage (39%) after tilling.
The weed-control treatments had similar results in both Andropogon and Brachiaria so we will discuss only Brachiaria.

Treatment effects of glyphosate on weed cover 30 days after herbicide application and crop planting (reinfestation) were highly significant, both for herbicide rates and times (Table 3) and for their linear and quadratic interactions. The lowest rate (I L/ha) applied 30 days after tilling was able to reduce weed cover by 41%, but the highest reduction was for the highest rate (4 L/ha) applied 60 days after tilling, which reduced weed cover by 85%. Higher weed covers for the lowest rate applied at both 30 and 60 days after tilling were the product of a combination of effects-reinfestation at the "30 days after" treatment and insufficient herbicide at the "60 days after" treatment. This condition was more or less the same after 60 and 90 days, after herbicide application and rate. At 90 days, maximum reinfestation occurred for the "30 days after" treatment (60% weed cover). After that, the weed cover reflected more the competition with an already developed Brachiaria than the treatment effect (no significant effect of treatments).
Effect on the pasture itself is less clear. Thirty days after herbicide application and crop planting, the percentage of Brachiaria cover did not reflect any treatment effect but Andropogon showed a highly linear interaction between rates and times, which is believed to reflect seed quality and season more than the treatment itself (Table 4). At 60 and 90 days after herbicide application and planting, the Brachiaria cover slightly reflected weed control efforts, but at 120 days no treatment effect was observed and cover exceeded 50%. This condition was also observed in the pasture dry matter yield at establishment, which was not affected by any treatments.
No significant differences between the different treatments in dry matter yield at establishment were present, and it was concluded that the lowest rate (1 L glyphosate/ha) is as good as 4 L/ha and could be recommended. With regard to the "times after" treatments, convenience depends on particular cases, but when low grazing pressure is planned at the early establishment phase it could be advisable to wait 60 days after tilling before herbicide application and planting.

Table 1. Selected soil properties at initiation of the experiment, 0-20 cm
Sand Clay pH P Al Ca Mg K O.M. sat.
-------------------------------------------------------------------------------- % ------ Fg/ml -------- cmol/L ----------- ----- % ----43 20 4.7 3.8 2.2 .2.07 0.53 0.15 2.73 44
--------------------------------------------------------------------------Table 2. Predominant weeds before rototilling and 30 days later.
--------------------------------------------------------------------------Approximate coverage
Before 30 days after
Local name Scientific name rototilling rototilling
--------------------------------------------------------------------------------------- % --------------Torourco Homolepis, Axonopus, 70 10
Paspalum sp.
Pega-Pega Desmodium sp. 8 4
Mimosa Mimosa pudica 6 2
Matapasto Pseudoelephantopus 6 3
Sinchipichana unknown 4 19
Ciperaceae Ciperus sp. 2 39
Guayaba Psidium guayaba 3 0
Broadleaved weeds 0 24

Table 3. Weed cover at 30 days after glyphosate application and Brachiaria
planting. Treatment means of six replications. Highly
significant linear times x rates interaction.
Time of application
Rate of
application 30 days 45 days 60 days
L/ha ------------------%-----------------1 59 63 67
2 43 63 34
4 45 35 15
Table 4. Pasture cover for Brachiaria decumbens and Andropogon gayanus.
Means of six replications. Brachiaria: no significance.
Andropogon: highly significant linear times x rates interaction.
Time of application
Rate of
application 30 days 45 days 60 days
L/ha ------------------%-----------------Brachiaria
1 4 3 4
2 4 3 5
4 4 3 4
1 5 2 0.3
2 2 1 0.5
3 2 2 0.3

Legume Shade Tolerance
Jorge W. Vela, INIPA, Pucallpa, Peru
Miguel A. Ara, N. C. State University, Pucallpa, Peru
Acid-tolerant legumes used in the TropSoils pasture management options may play an important role in agroforestry systems where cattle grazing may take place under trees. Experience in Southeast Asia indicates a wide variability of legume response to shade.
To test our most promising legume germplasm under shade of a mature oil
palm plantation to determine their adaptability to this factor.
This experiment was established in November 1985 under an oil palm plantation managed by CIPA XXIII at km 44 of the Pucallpa-Lima highway. Three legumes (Desmodium ovalifolium, Pueraria phaseoloides, and Stylosanthes guianensis) are being evaluated for their dry matter productivity and feed quality in the presence and absence of oil palm shade. The experimental design is randomized complete blocks with five replications. Plots are 8 x 16 m, and each plot has two oil palm trees in the same position. A similar but smaller (3 x 5 plot) experiment was established outside the palm plantation and will be used as a reference; all the variables will be analyzed as percentages of full sunlight.
Dry matter production of all three legumes was severely affected by oil palm shade. Desmodium ovalifolium and P. phaseoloides performed similarly but yielded only about 22% of the full sunlight value even though they outperformed S. guianensis, which gave a poor 8% of the reference value. Crude protein content values for D. ovalifolium and S. guianensis were higher than full sunlight values, but D. ovalifolium values were highest (Table 1).
Preliminary data suggest that D. ovalifolium is the most shade-tolerant legume tested. Visual observations show S. guianensis performs quite poorly and tends to disappear. The stands of both D. ovalifolium and kudzu were

quite acceptable under shade in absolute terms. Since kudzu is a climber, periodical cutting around trees is required, whereas D. ovalifolium does not require such cutting.
Table 1. Percentage of dry matter yield and crude protein content
(CPC) of the legumes under shade relative to full
sunlight (means of five replications).a
S. guianensis D. ovalifolium P. phaseoloides
---------------% of full sunlight--------------Dry matter 7.7 b 23.3 a 21.8 a
Crude protein 113.2 b 128.4 a 84.2 b
a. Means with the same letter in the horizontal rows are not
significant at P = 0.01.

Extrapolation in Farmer Fields
Miguel A. Ayarza, N. C. State University, Yurimaguas, Peru Rolando Dextre, INIPA, Yurimaguas, Peru
After 6 years of work in the Yurimaguas Station, Brachiaria humidicola, Centrosema pubescens, Stylosanthes guianensis, and Desmodium ovalifolium have shown a high potential to increase animal production and good persistence under grazing. On the basis of the experience gained over the years, it was decided to test the potential of improved pastures to replace degraded native pastures, which are common in the area.
A validation trial was set up to demonstrate that degraded pastures can be put into production by substituting present vegetation or improvement of present pasture plus incorporation of new species. An agreement was signed with the Empresa Ganadera Amazonas to conduct a 20-ha validation trial in an area covered by degraded pastures.
Four pasture renovation options were designed:
1. Introduction of a legume on a degraded pasture of Brachiaria
2. Use of a crop as precursor to establishing improved grass-legume
3. Improvement of native pasture through introduction of a legume;
4. Reclamation of a steepland pasture using improved grasses, legumes,
and trees with minimum tillage.
Work started in December 1985 with soil characterization of the area. Extremely low native fertility and a predominantly sandy texture existed in this sandy Ultisol (Table 1).
For option 1, 4 ha of a 10-year-old B. ruzisensis field were disked slightly to break the soil surface and facilitate planting of C. pubescens 438 and to promote grass regrowth. The legume was broadcast over the area at a rate of 1 kg/ha. Rock phosphate was applied to supply 25 kg P205 and dolomitic limestone was applied to supply 50 kg Ca + 10 kg Mg.
Option 2 was installed on a flat area infested by weeds and unpalatable grass species. Two diskings and two rototiller passes were needed to

completely till 3 ha to eliminate existing vegetation. Cv. Africano desconocido, an upland rice variety known to be tolerant to Al toxicity, was planted in strips 12 m wide. The crop received 60 kg N, 40 kg K20, and 50 kg P205 per ha. Brachiaria humidicola was planted in alternate strips 2 months later.
After rice harvest, cowpea was planted in rows 50 cm apart and B. humidicola + D. ovalifolium was planted between rows.
Option 3 was installed on a 2-ha native pasture of torourco (Axonopus compresus). Strips 4 m wide were opened every 20 m and planted with Stylosanthes guianensis accessions 136 and 184. Phosphorus was applied at a rate of 12 kg P/ha using rock phosphate.
Option 4 was installed in a degraded pasture on a 20% slope. The area was tilled in strips 1 m wide and 4 m apart in contour. Land preparation was carried out with a 1-1/2 HP manual rototiller. Areas between rows remained untouched. Brachiaria humidicola and C. pubescens 438 were planted in alternate strips, and Erythrina popigiana was planted in rows every 12 m along the contour.
After 8 months, all pasture renovation options were fully established. In option 1, B. ruzisensis reacted positively to tillage and C. pubescens was spreading very well over the area. Measurements of botanical composition indicated 20% of Centrosema in the total biomass, a highly desirable legume content of a mixed pasture.
In option 2, rice yields were affected by short dry spells during January and February 1986. In spite of limitations, 1.5 t/ha of rice were harvested and cowpea yields reached about 800 kg/ha. Both are acceptable yields for low-input systems. After harvesting the cowpea, B. humidicola was almost fully established, although it was infested by annual weeds.
In option 3, Stylosanthes established rather quickly, although some handweeding was required.
Introduced pastures in the steepland covered almost the entire area in option 4. Both Centrosema and B. humidicola replaced most of the native species in the nontilled bands. On the other hand, Erythrina did not establish well. Growth was affected by soil conditions (sandy texture and K deficiency). Only two of six rows have established.

Costs of renovating degraded pastures are presented in Table 2. Total investment varied among systems. The most expensive system was the one with crops (option 2) because of labor and machine costs. Returns from crop yields were enough to pay for pasture renovation, however, and leave some extra profit before grazing.
Overall results showed that renovation of degraded pastures can be accomplished with low monetary inputs in some instances. In other cases where a complete renovation is required, crops as precursors pay for the entire cost of pasture re-establishment.
A second phase calls for milk production from cows grazed on the pastures in every system. This phase will start in 1987 as work supported by INIPA and the Ganadera Amazonas S.A. and will serve as a thesis for an undergraduate student from the Universidad Agraria at La Molina.
Table 1. Topsoil chemical property of the soil used for the
extrapolation work in K-17 (Ganadera Amazonas area) (mean
of five samples). December 1985.
pH P Al Ca Mg K ECEC Al sat. Sand Clay
ppm -----------cmol/L ------------ --------%------4.0 3.4 1.1 0.3 0.08 0.05 1.53 72 82 14
Table 2. Cost of renovating a degraded pasture by three methods
of introduction of improved species.
Option 1 Option 2 Option 3
B. ruzisensis + Rice-cowpea- Native pasture
C. pubescens pasture (Stylosanthes)
U.S.$/ha % U.S.$/ha % U.S.$/ha %
Labor 7 11 86 39 20 16
Machinery 9 13 65 29 43 34
Fertilizer 35 53 53 24 6 4
Seed 12 17 16 7 50 40
Other 3 5 11 5 6 5
Total 66 99 231 104 125 99

A low-input cropping system, reported last year as a transition technology between shifting cultivation and permanent agriculture, collapsed after seven crops in 3 years due to P and K deficiencies and increasing weed pressure. This year we report on the nutrient cycling and economic aspects of this first cropping period, the success of a 1-year kudzu fallow in overcoming weed constraints, the successful transition to other systems, and in-depth data on weed build-ups during the first cropping period.
Nutrient cycling from above and below plant residues returned to the soil more than 80% of the K and Ca accumulated by plants, about half the biomass produced and half the N and Mg accumulation, but only 37% of the P accumulated, largely because of grain removal. The system was highly profitable both with and without fertilizer applications: purchased chemical inputs accounted for only 8% of the total costs without fertilizer and 16% with fertilizer. After 1 year of kudzu fallow, the fields were largely devoid of weeds and showed a higher fertility status. A second low-input cropping period started with high yields, and the transition to fertilizerbased continuous cultivation was successful. Weed-control studies indicate the inability to grow more than five or six low-input crops continuously with the best combination of herbicides and manual weed control. The main problem
is controlling weeds in upland rice.

Central Experiment: Transition to Other Technologies
Jose R. Benites, N.C. State University, Yurimaguas, Peru Pedro A. Sanchez, N.C. State University, Raleigh
A low-input cropping system has been developed in Yurimaguas, Peru, to serve as a transition technology between shifting and continuous cultivation for acid soils of the humid tropics. Its principal components are (1) Traditional slash-and-burn clearing of forest fallow, (2) selection of acidtolerant cultivars capable of high yields without liming, (3) rotation of upland rice and cowpea cultivars (no tillage) and removing only the grain,
(4) no fertilizers, lime, or organic inputs are brought in and soil pH remains at about 4.5, (5) the rotation continues for 3 years, but increasing weed pressure and decreases in available P and K cause the system to collapse in agronomic and economic terms. A total of five upland rice and two cowpea crops were harvested during a 3-year period, as opposed to one rice crop under traditional cultivation. Crop yields and effects on soil properties during this first period were presented in the last TropSoils technical report. This report covers the nutrient cycling effects and economic analysis of the first cropping period and the transition phase initiated after the system collapsed at the seventh continuous harvest on June 30, 1985.
Nutrient Removal Cycling
Low-input systems should be efficient recycles of nutrients in order to minimize nutritional inputs that must replace nutrients extracted by crop harvests.
The nutrient composition of the acid-tolerant rice (cv. Africano) and cowpea (Vita 7) plant parts at harvest obtained in neighboring experiments are presented in Table 1. The calculated amount of nutrient accumulation by the seven crops is shown in Table 2. Even though only the rice grain and cowpea pods plus grain were exported from the field, the harvested products during the 3 year period represented considerable nutrient removal from the field (Table 2). The amounts of nutrients accumulated by the crops but returned to the soil as above- or belowground organic inputs was larger than the amount removed except for P. Crop residues plus root runover returned to

the soil 62% of the dry matter produced, 54% of the N, 59% of the Mg, 87% of the K, 94% of the Ca, but only 37% of the P accumulated by crops (Table 2). Root turnover, assuming 100% fine root decomposition, accounted for a relatively minor proportion of amounts recycled (14% of dry matter, 21% of N, 25% of P, 5% of K, 18% of Ca, and 12% of Mg). The actual amounts returned, therefore, are equivalent to an annual fertilization rate of 98-7-199-33-13 kg/ha of N-P-K-Ca-Mg. A proportion of the N returned as aboveground residue, however, may be lost before it enters the soil, via denitrification that may take place on the mulch-soil surface interface. Biological N fixation by cowpea, however, may counteract such losses, but neither process was measured. The P, K, Ca, and Mg inputs, however, are likely to be transferred entirely to the soil. Phosphorus, therefore, appears to be the critical nutrient, since about two-thirds of the crop uptake was removed by the harvested products giving this element the lowest percentage of recycling and the lowest absolute amounts returned to the soil among the five nutrients evaluated.
Weed Pressure
In the authors' view, increasing weed control difficulties was the single most important factor for the instability of this low-input system during its third year. The initial weed population was mainly broadleaved, which is typical of shifting cultivation fields in the area. With time, the weed population gradually shifted to grasses that are more aggressive and not subject to economically sound control by commercially available herbicides. Of particular importance was the spread of Rottboelia excelsa, a nonrhizomatous grass, particularly during rice growth. Cowpea was more
competitive with weeds than upland rice because cowpea covered the soil surface more thoroughly.
Studies on weed control in low-input systems at Yurimaguas indicate that the absence of tillage and burning promotes weed build-up (see next report). Rice straw mulch may decrease weed growth in cowpea, but cowpea residues do not have the same effect on rice, perhaps because of the fast decomposition rate visually observed with cowpea residues.
Cost records were kept in this 1-ha experiment. The summary for the

first seven crops in the plots without fertilization is shown in Table 3. Labor inputs for the first crop include land clearing; thus the subsequent crops averaged two-thirds of the first crop's labor. Returning and
redistributing crop residues averaged 10 man days/ha, or approximately U.S. $20/ha per crop. Another major labor input was bird watchers near harvest time. The next major cost items were interest on crop loans from the Banco Agrario and government fees for receiving and processing rice at the mills. Shifting cultivators routinely obtain bank loans, which are used primarily as an advance on their labor. Interest charges fluctuating from 40 to 101% on an annual basis in local currency reflect the high inflation rate in Peru, which averaged 125% annually during the study period. Even in U.S. dollar terms, the indirect costs averaged about 30% of the total production costs. In contrast, the cost of purchased chemical inputs (herbicides and insecticides) and others (seed, bags, thresher rent) comprised 8 and 19% of the total production costs, respectively.
The low-input system without fertilizer applications was highly profitable, averaging net returns of U.S. $1144/ha per year, or a 21% return over total costs (Table 4). The low-input system with fertilizers was also quite profitable, averaging an annual net return of US$ 1125/ha and a 100% return over total costs. Fertilizers accounted for 9% of the total cost in the system, but also resulted in additional labor, interest, thresher use, and transport costs. The low-input system either with or without fertilizers is vastly more profitable than traditional shifting cultivation (Table 4).
Transition to Other Systems
This low-input system, therefore, is a transitional technology in both agronomic and economic terms. After 3 years, the field is devoid of
felled logs and most of the remaining tree stumps are sufficiently decomposed to be destroyed with a good kick. The land clearing process is thus
complete, providing several options to the farmer. One is to put the land into a managed fallow and then start a second cropping cycle. A second is to plow, lime, fertilize, and rotate crops intensively; a third is pastures, and a fourth is agroforestry.
The experiment described above was modified to address some of these options after the seventh crop harvest in July 1985. The 1-ha field was
divided into eight 1250 m2 plots, providing four treatments in a randomized

complete block design with two replications. The replicates were located on the previously fertilized and not fertilized treatments in order to block the residual effects. Two treatments were designed to test the weed control factor by continuing the low-input system (cowpea-rice-cowpea), a third was planted to kudzu fallow, and the fourth was a high-input system.
Continuing the Low-Input System
Prior crop yields suggest that the system collapsed after the seventh crop. This observation was confirmed by growing three more crops with a weed-control variable: full weed removal at economically unrealistic levels vs. the conventional treatment as previously described. The full treatment consisted of eliminating weeds by a pre-emergence application of 2.25 kg/ha active ingredient of metolachlor plus 2.5 L/ ha of paraquat, followed by 0.28 kg ha active ingredient of sethoxydim supplemented by handweeding was needed. The actual cost of this treatment was US $225/ha per crop, a totally unrealistic level. The conventional treatment was the pre-emergence application of 1.5 L/ha of 2,4D followed by 2.5 L/ha of paraquat 5 days later and no handweeding, with a total cost of $25/ha per crop. Both weed control plots received the same application of NPK fertilizers to rice as stated previously, in order to eliminate P and K deficiencies.
Grain yields of the eighth, ninth, and tenth consecutive crops under conventional weed control were low with cowpea, and practically zero with rice (Table 5). When weeds were totally removed, yields of the ninth (rice) and tenth (cowpea) crops reached acceptable levels. Consequently, it seems reasonable to assume that the collapse of the system is directly related to weed-control problems.
Kudzu Fallow and a Second Crop Cycle
Traditional shifting cultivation involves a secondary forest fallow period of 4 to 20 years, supposedly to replenish soil nutrient availability and control weeds, although the processes involved are not well understood. Farmer experience around Yurimaguas indicates that a minimum desired age of fallow is about 12 years, but population pressures effectively reduce this period to an average of 4 years. Slashing and burning young forest fallows results in faster grass weed invasion than would occur in older fallows because the weed seed pool declines with age. Considering the limited

likelihood of long secondary fallow periods in developing humid tropical areas, the need for an improved fallow is apparent.
Following a farmer's suggestion, we studied the use of tropical kudzu (Pueraria phaseoloides) as a managed fallow. Unlike its temperate-region counterpart (Pueraria lobata), tropical kudzu does not produce storage roots and therefore is easy to eradicate by slash-and-burn. Kudzu fallows were grown in previously cultivated fields for different durations. In the most infertile and compacted soils of Yurimaguas, kudzu is slow to establish and initially shows several classic nutrient deficiency symptoms, but within 3 months a complete canopy is attained, the kudzu leaves become dark green, and weeds are smothered. Aboveground dry matter and ash biomass accumulation by kudzu peaks at about 2 years. We observed increases in exchangeable Ca and Mg and decreases in Al saturation on the topsoil of kudzu fallow plots that had a lime and fertilizer application history. But no improvements in these topsoil chemical properties were recorded in the kudzu fallow plots that had never been limed or fertilized during a previous cropping period. Consequently, the subsoils must have some nutrients available for recycling if significant recycling by a managed fallow is to take place in such acid soils.
The same kudzu ecotype was seeded in this low-input experiment, on August 28, 1985, after harvesting the seventh rice crop, which was heavily infested with Rottboelia excelsa and other weeds. No fertilizers were added to the kudzu plots, but one handweeding was used to pull tall Rottboelia plants. As before, kudzu was slow in establishing, but within 3 months it had developed a complete ground cover and a surface litter layer. Kudzu was slashed with machetes on September 13, 1986; after 10 days of dry weather, it was burned in a total time of 4 minutes for the 1250 m2 plots. Ash sampled 1 day after the burn contained significant accounts of nutrients which were incorporated to the soil by the first rains (Table 6). A clear residual effect of the previous NPK fertilization is evident in the ash composition, particularly in P and K contents.
A crop of Africano rice was planted 3 days after the kudzu burn and harvested on January 22, 1987. It received the 30-22-40 kg/ha of NPK as did all previous rice crops. Grain yields were the highest obtained to date at this site (Table 7). This is partly due to very favorable rainfall distribution for rice growth as evidenced by similar rice yields obtained in

other experiments at that time, but also due to the absence of significant weed pressure. The 1-year kudzu fallow, therefore, effectively suppressed weed growth in a way far superior to the herbicide combinations attempted to date. The plots are now growing a subsequent crop of upland rice.
Changes in topsoil chemical properties in the kudzu fallow plots are shown in Table 8 at the end of the first cropping cycle, after I year of kudzu fallow (I day prior to burning it), and after the first harvest of the second cropping cycle. The effect of the kudzu fallow on topsoil chemical properties includes a significant decrease in exchangeable Ca and K, presumably due to plant uptake with no changes in acidity, Al saturation, or available P and K (Table 8). Differences in the last four properties are significant in terms of prior fertilization treatment, which the kudzu fallow maintains.
Topsoil chemical properties after the first harvest of the second cropping cycle show fewer differences due to previous fertilization than at the end of the first cropping cycle, partly because the entire area was
fertilized with 30-22-48 NPK formula. Nevertheless, there is an overall trend of increasing available P, K, Ca, and Mg that may be related to the nutrient content of the ash and prior fertilization during the first cropping
cycle. Topsoil properties at 54 months after burning, representing seven crop harvests, 1 year of kudzu fallow, and one crop harvest afterwards are about as good or better than 3 months after burning the original forest (see previous technical report). Topsoil total organic matter contents have
increased, probably as a result of the kudzu fallow litter inputs (Table 8). It appears reasonable to speculate that organic-matter contents will decrease slightly with subsequent cropping as it did during the first cropping cycle.
The second cropping cycle continues in order to determine how long it will last, except that no fertilizer is being applied and weed control will only be at the conventional level.
High-Input Crop Production
Another option is for the low-input system to serve as a 3-year transitional period to intensive, fertilizer-based, continuous cropping systems for areas that have developed a sufficient road, credit, and market infrastructure to make this possibility attractive. The fields are certainly ready for mechanized tillage, provided slopes are suitable, because most of

the felled vegetation has decomposed. One treatment of this field experiment
was tilled to 25 cm with a 50-HP tractor, limed with 3 t/ha of dolomitic lime, fertilized with 25 kg/ha P as triple superphosphate, 25 kg/ha Mg as MgS04, I kg Zn as ZnS04, 1 kg Cu as CuSO4, and 1 kg B as borax. Lime and fertilizers were incorporated with tillage. "Marginal 28," an adapted corn variety, was planted in ridges at a population of 56,000 plants/ha. This crop then received 100 kg N/ha as urea and 100 kg/ha of K as KCl in three split applications. Weeds remaining after tillage were controlled by 2.25 kg/ha active ingredient of metalochlor. A second crop of corn was then planted after mechanically incorporating corn stover. It received an
application of 100-25-100 kg/ha NPK and metalochlor, at the same rate. The corn was followed by soybean (cv. Jupiter), which received an application of 30-25-100 kg NPK/ha. Insecticides were used in corn and soybean as needed to control mild insect attacks to these crops.
Corn yields were normal for high-input systems for the planting season (3-4 t/ha), while soybean yields were somewhat lower than normal (2.0 t/ha) partly because of unusually heavy rains (Table 9). The system appears stable and of similar productivity to long-term high-input systems grown at Yurimaguas. A combined total of 8.6 t/ha of high value grain (corn and soybean) was produced in approximately 15 months. Total productivity of the entire sequence was 22.4 t/ha of grain in 4 years and 4 months, or 5.2 t/ha per year. In order to decrease weed infestations, using the kudzu fallow prior to shifting to high input cropping may be advisable.
Legume-Based Pasture
The low-input system can also serve as a precursor to establishing improved, acid-tolerant pastures, beginning with the clearing of secondary forests. Income-generating food crops can be grown and the pasture species may be planted either vegetatively or by seed under a rice canopy. Several combinations of persistent, acid-tolerant grasses and legumes produced high and sustained liveweight grains in Yurimaguas for 6 years, as reported in the legume-based section. The kudzu fallow itself could be used as a pasture in rotation with grass-based pastures. Although we have not shifted from lowinput cropping to pastures in an actual experiment, the possibility appears feasible. Since weed encroachment is a major limiting factor in pasture establishment, it may be advantageous to limit the number of crops in order

to minimize the weed build-up that occurred during the sixth and seventh crops. Planting kudzu fallow, burning it after 1 year, and then establishing the pastures may be a better approach.
The low-input cropping system is a good way of providing cash income and ground cover during the establishment phase of tree plantations. The
decision, however, has to be made early in order to transplant or seed the tree crops at adequate spacing shortly after clearing and burning the second forest. Unless liming is contemplated, the choice of tree crops should be limited to acid-tolerant ones. Examples of acid-tolerant crops for industrial purposes are rubber (Hevea brasiliensis), oil palm (Elaeis guineensis) and guarana (Paulinia cupana); for food production, peach palm (Gulielma gasipaes); for alleycropping, perhaps Inga edulis. Woody species known to be sensitive to soil acidity such as Theobroma cacao or Leucaena leucocephala should be avoided.
The low-input system has been used successfully in nearby experiments for the establishment of peach palm and multipurpose tree production systems that include fast-growing (Inga edulis) and slow-growing (Cedrelinga cataeniformis) species. For peach palm, seedlings are transplanted with the first rice crop. Within 18 months, the peach palm produces too much shade for further crop growth; kudzu is then planted as an understory.
The low-input system has several potentially positive environmental impacts. It provides a low-cost alternative for shifting cultivation in highly acid soils. In order to produce the grain yields reported for the first cropping period, a shifting cultivator would need to clear about 14 ha in 3 years, in comparison to 1 ha in this low-input system. Furthermore, the use of secondary forest fallows instead of primary forests is emphasized, although the system should work well starting from primary forest.
Erosion hazards are largely eliminated by the absence of tillage and the presence of a plant canopy on the soil surface, be it slash-and-burn debris, crop canopies, crop residue mulch, or a managed fallow. Nutrient recycling is maximized, but nutrients exported as grain must be replenished by outside inputs in soils so low in nutrient reserves. Perhaps just as importantly,

the low-input system does not lead the farmer into a corner; it provides a wide range of options after the first cropping cycle is complete.
There are many unanswered questions about the technology just described.
Although its feasibility during the first cropping cycle followed by a managed fallow period has been demonstrated, information about the second cropping cycle is limited to one crop harvest. It cannot be stated at this point that a modified form of shifting cultivation with a 3:1 crop to managed fallow ratio is feasible on a long-term basis.
More in-depth knowledge of weed population shifts and fertility dynamics is needed. Zero tillage poses a major constraint to long-term weed control. Soil data have been confined to readily determined chemical parameters; soil
physical and biological dynamics are now being intensively studied. Tropical kudzu is but one of several promising species for managed allows, and others are being investigated in terms of above- and belowground biomass accumulation, nutrient cycling, and weed suppression. A fresh look at the management of organic inputs and soil organic matter is being researched.
The effect of age of fallow needs to be studied in greater detail. What are the trade-offs with a longer fallow? Also, can the weed problem be reduced by having a shorter time period between harvesting and planting? Time will tell.

Table 1. Nutrient concentrations of rice (cv. Africano desconocido)
and cowpea (cv. Vita 7) grain and straw at harvest, and fine
roots at anthesis, under low-input systems in Yurimaguas.
Mean values of seven rice harvests and three cowpea
harvests for aboveground parts, and two crop harvests each
for roots.
Crop part N P K Ca Mg
------------------ %-----------------Rice Grain 1.40 0.23 0.36 0.03 0.11
Straw 1.01 0.07 2.85 0.28 0.18
Roots 1.19 0.09 0.93 0.21 0.05
Cowpea Grain 3.86 0.35 1.36 0.06 0.21
Straw 1.86 0.13 3.97 0.95 0.23
Pods 0.70 0.06 2.17 0.17 0.22
Roots 1.23 0.12 0.71 0.59 0.19
Table 2. Total dry matter and nutrient accumulation by five rice
and two cowpea crops harvested in 34 months without
fertilization, and amounts returned to the soil.
Plant part mattera N P K Ca Mg
t/ha ------------------ kg/ha----------------Grain + pods 14.2 256 34 87 65 18
Straw 18.1 232 15 565 80 35
Roots 5.2 63 5 32 18 5
Total 37.5 551 54 684 104 58
% returned 62 54 37 87 94 59
to soil
a. Based on mean grain:straw ratios of 0.84 for rice and 0.52 for
cowpea; mean cowpea pod weight of 0.32 t/ha per crop, and fine
root biomass of 0.65 and 0.97 t/ha per crop for rice and cowpea
in the top 30 cm of soil.

Table 3. Labor input, production costs, and revenues incurred in
the low-input system with seven crops without
Crop sequence
1 2 3 4 5 6 7
Input or output Rice Rice Cowpea Rice Cowpea Rice Rice
Labor (man day/ha) 172 79 99 79 99 79 79
Cost (U.S. $/ha):
Labor 380 140 113 134 167 130 95
Herbicides 21 21 25 26 25 24 25
Insecticides 0 11 14 14 13 0 0
Seed 19 17 75 18 51 16 17
Bags 16 18 8 20 7 18 50
Thresher rent 0 34 0 38 0 34 80
Transport to market 12 12 14 14 14 12 14
Loan interest and fees 135 80 86 105 108 111 225
Total cost 583 333 335 369 385 345 506
Grain produced (t/ha) 2.44 2.99 1.10 2.77 1.19 1.84 1.52
Price (U.S. $/t) 321 281 1420 305 1127 265 274
Gross revenue (U.S. $/ha) 783 840 1562 845 1341 488 416
Net return (U.S. $/ha) 200 507 1227 476 956 143 -90
Net return/Cost (%) 34 152 366 129 248 41 -18

Table 4. Cumulative production costs and returns actually incurred with
seven crops in 3 years with and without fertilization and under
shifting cultivation.
Low-input system
Inputs and outputs Not fertilized Fertilized cultivation
U.S. $/ha % U.S. $/ha % U.S. $/ha %
Labor inputs 1159 41 1185 35 380 65
Chemical inputs:
Fertilizers 0 0 292 9 0 0
Herbicides 167 6 167 5 21 4
Insecticides 52 2 52 2 0 0
Other inputs:
Seeds 213 7 213 6 19 3
Bags 137 5 140 4 16 3
Thresher use 186 7 189 6 0 0
Transport to market 92 3 96 3 12 2
Loan interest and fees 850 30 1073 38 135 23
Total costs 2843 100 3351 100 583 100
Gross revenues: 6275 6688 783
Net returns: 3432 3377 200
% Returns/cost 121 100 30

Table 5. Grain yields of three additional crop harvests in the low-input
system at two levels of weed control.
Weed-control level
Planting Harvest
Crop sequence date date Conventional Full LSDO.05
---------------------------------------------------------- ------------- t/ha-----8. Cowpea cv. Vita 7 Aug. 19, '85 Oct. 31, '85 0.58 0.58 ns
9. Rice cv. Africanoa Jan. 9, '86 May 9, '86 0.09 1.60 0.19
10. Cowpea cv. Vita 7 July 15, '86 Sept. 25, '86 0.43 0.82 0.26
a. Fertilized with 30-22-48 kg/ha of NPK as previously.
Table 6. Dry matter and nutrient content of 1-year-old kudzu fallow ash 1
day after burning. Mean of 12 observations in the previously not
fertilized treatments and 11 in the previously fertilized
treatment. Date: September 24, 1986, 51 months after clearing.
fertilizer Dry
treatments matter N P K Ca Mg Zn Cu Fe Mn
t/ha ----------------------kg/ha--------------------Not fertilized 1.45 21 7 42 58 14 0.17 0.08 1.07 0.82
Fertilized 2.29 26 21 103 92 20 0.29 0.12 0.99 1.34
LSDO.05 0.85 ns 9 51 ns ns 0.12 ns ns ns

Table 7. Dates of planting and harvesting kudzu fallow and subsequent crops
and crop yield as affected by a fertilizer differential prior to
planting kudzu.
Grain yield
fertilizer level
Planting Harvest
Crop sequence date date None Yes LSDO.05
---- -------------------------------------------------------------t/ha-8. Kudzu fallow Aug. 28, '85 Sep. 13, '86 -- -- -9. Rice cv. Africano Sep. 26, '86 Jan. 22, '87 3.86 3.98 0.41
10. Rice cv. Africano Feb. 13, '87 in progress
------------------------------------------------------------------Table 8. Changes in selected topsoil (0-15 cm) chemical properties prior to
and after 1-year fallow, and after harvest of a subsequent crop
in the kudzu fallow plots.
-------------------------------------------------------------------------Months Prior Exchangeable
Plot after fertili- pH Al Avail. Organic
status burning zation (H20) -Al Ca Mg K sat. P matter
---- --------------------- ---------------------------------------------------- cmol/L------- % mg/dm3 %
End of first 35 No 4.5 1.9 0.98 0.10 0.26 50 4 1.92
cropping cycle Yes 4.5 1.2 0.98 0.16 0.19 46 14 1.77
(seven crops)
After 1 year 52 No 4.5 1.8 0.60 0.09 0.13 68 7
in kuzdu fallow Yes 4.6 1.1 0.57 0.13 0.11 57 14
(before burning)
At first 54 No 4.8 1.6 1.05 0.18 0.23 52 14 2.44
harvest of second Yes 4.5 1.1 0.76 0.20 0.14 49 25 2.71
cropping cycle
LSDO.05 0.2 0.3 0.25 0.05 0.05 11 6 ns
CV 4 25 35 32 39 21 48 15

Table 9. Grain yields of the first three intensively managed
continuous crops after shifting from a low- to highinput system.
Planting Harvest Grain
Crop sequence date date yield
8. Corn cv. Marginal 28 Sept. 9, '85 Jan. 19, 86 3.86
9. Corn cv. Marginal 28 March 22, '86 July 24, 86 2.90
10. Soybean cv. Jupiter Sept. 11, '86 Dec. 23, 86 1.87
Totals 15 months 8.63

Weed Control in Low-Input Cropping Systems
Jane Mt. Pleasant, N. C. State University, Raleigh Robert E. McCollum, N. C. State University, Raleigh
Weed control in low-input cropping systems must rely on cultural practices to increase the crop's ability to compete against weeds and thereby reduce the amount of herbicide and manual weeding inputs that are needed to maintain crop yields. Mulching, tillage, timely fertilization, increasing crop density, and the use of competitive cultivars are all examples of cultural practices that can aid in controlling weeds.
To identify cultural practices that can form the basis of a weed management program for a continuously cropped rice-cowpea rotation under
low-input conditions.
The experiment was established after cutting and burning a 5- to 10year-old secondary forest at Yurimaguas. Five consecutive crops, identified as Cycles 1 through 5 (rice-rice-cowpea-rice-cowpea), were planted. The experimental design was a split-plot, with tillage and residue management as main plots. There were three main-plot treatments: (1) rototill with previous crop residues incorporated, (2) rototill with residues mulched and,
(3) no-till with residues mulched. A factorial arrangement of two crop densities (high and low) and three weed-control practices (handweeding, herbicide, and no control) comprised the subplot treatments. Oxadiazon and propanil were the herbicides used on rice, and metolachlor was applied to cowpea.
Data from Cycles 2 through 5 were combined and analyzed as a single experiment. Cycle 1 was not included in this combined analysis because there was no residue management variable in the first crop that succeeded the forest fallow. The remaining four cycles were consistent in treatment through the duration of the experiment. Although the emphasis in this report is on the combined analysis, some data from individual crop cycles are also reported.

Results and Discussion
Crop Species
Rice had more weeds and lower relative yields than cowpea (Figure 1), and failure to control weeds had a much greater effect on rice than on cowpea yields (Figure 2). Because cowpea establishes quickly, we hypothesize that it covers the row and shades out emerging weed seedlings. Rice, in contrast, is much slower in forming a canopy. With the ground left unshaded, weed seedlings in rice quickly become competitors.
Weed infestation was also more severe the longer the field was cropped (Figure 1). Weed levels in Cycle-4 rice were much higher than in Cycle-2 rice. The same pattern was seen in cowpea: weed infestation was higher in Cycle 5 than in Cycle 3.
Residue Management
Mulching was ineffective as a weed control practice, but relative crop yields were higher when residue was incorporated (Figure 3). In this
experiment the primary role of residue management appears to be related to nutrient release rather than weed control. We theorize that rapid
decomposition of residues following incorporation releases nutrients more quickly for immediate recycling than when they are allowed to decompose on the surface.
No-till plots had more weeds and lower yields than rototilled plots in the combined analysis, but the effect of tillage on weed growth changed over the course of the experiment (Figure 4). In the first cycle, rototilled plots had many more weeds than no-till plots. By Cycle 5, however, the notill treatment had the greater weed infestion.
Rototilling in Cycle 1 provided an ideal seedbed for weed seed germination. With tillage, weed seeds were brought to the surface where they germinated in a flush. In contrast, weed infestation in Cycle-i no-till plots was low. Fire destroyed both standing vegetation and surface seeds. Without soil disturbance to bring buried weed seeds to the surface, weed infestation in the first crop was minimal.
When the field is cropped continuously, however, lack of tillage through several cropping cycles brings increased weed problems. Soil disturbance has

a positive effect in a continuous cropping system because existing vegetation can be completely eliminated between crops. While weeds in no-till plots were burned back with a pre-plant application of paraquat between crops, they quickly regrew. Consequently, after five consecutive no-till cropping cycles, untilled plots had a much larger weed infestation than rototilled treatments.
Crop Density
Increasing crop density was an effective weed control measure (Figure 5). In most cycles there were fewer weeds and higher product yields when the crops were planted at the higher density. But yields, particularly rice, increased at the higher density even when weed growth was not affected.
Apparently closer-spaced rice was more efficient in intercepting sunlight for photosynthetic production, and the higher yields were independent of the effect of crop density on weed growth.
Weed-Control Practice
Failure to control weeds reduced product yields in both rice and cowpea, but the weed-control method (manual or chemical) had little effect on yield (Figure 5).
Management and Research Implications
If a viable low-input continuous cropping system is to evolve in the Peruvian Amazon, it will depend on upland rice as the central cash crop, and the rice or any associated crop will be planted without tillage. It is now apparent that we cannot control weeds through repeated cycles of rice-cowpea rotations without large and unprofitable inputs for weed control. For this reason, such a cropping system must be considered transitional. It may form
a bridge between shifting cultivation and a more permanent agriculture, but it is not a stable, long-term alternative to shifting cultivation.
Lack of tillage and the poor competitive ability of upland rice are the primary causes of the weed-control problem. Weed infestation in rice
increases dramatically with time when a field is cropped continuously without tillage, even when rice is rotated with cowpea. Because upland rice is such a poor competitor, grasses invade vigorously and crop yields decline sharply with successive rice crops.
It is possible to control weeds (either manually or with herbicides) and maintain yields, but the cost of control is prohibitive for a low-input

system within the present price-profit structure in Peru. With rice yields of 2 to 3 t/ha and cowpea yields of 1 to 2 t/ha, weed-control costs represent 25 to 40% of the value of the crop.
Based on our present knowledge, a realistic "lifespan" for the low-input system is probably five or six crop cycles, after which the cropping system must be interrupted by a fallow period or tillage in order to disrupt and displace the weed community. Research should now focus on developing
effective and economic weed-control strategies for this transitional lowinput system comprised of no more than six consecutive crop cycles. Our work
suggests several avenues that may be productive.
Weed-control inputs are unnecessary or minimal in the first rice crop after a forest fallow and in all cycles with cowpea. The majority of weedcontrol costs will be concentrated in the second, third, and fourth rice crops.
Increasing the planting density of rice is an effective and cheap form of weed control. Work is needed to establish optimum planting densities for rice cultivars in this environment. As demonstrated with cowpea, an
aggressive, fast-growing crop is another form of inexpensive weed control. Rice cultivars used in the low-input system should be selected for their competitive abilities. Early canopy formation, to shade out weed seedlings, is probably a critical characteristic for rice cultivars in this management system.
We have shown that herbicides can provide excellent weed control in upland rice. Furthermore, we suggest that this control method may become economic if herbicide use can be integrated with other control practices so that the rate and number of applications can be reduced. A practical and effective weed-management program for a low-input system will combine cultural practices with chemical and manual methods of control.

N E Rice
\ EZ Cowpea
e 100
0 0
R CP C2 C4 C3 C5
Figure 1. Effect of crop species and number of cropping cycles on weed infestation in four consecutively-planted
short-season crops. Yurimaguas 1984-1985.
R = rice; CP = cowpea; C2 = Cycle 2; C3 = Cycle 3;
C4 = Cycle 4; C5 = Cycle 5.

, 400 100
NE H Rice
H- Cowpea
S300 75
S200- > 50a)
"O >
S100- 25
0 *--a Cowpea
0 0 *- Rice
No control All control No control All control
Figure 2. Effect of weed control on weed infestation and relative yield in rice and cowpea in four
consecutively-planted short-season crops. Yurimaguas 1984-1985. NO CTRL = no weed control;
ALL CTRL = treated with herbicide or handweeded.

150- 1o0
E EZ Mulch E Mulch
Incorp E Incorp
4' 100
> 5055 0
ii 250 ,0
Figure 3. Effect of residue management on weed infestation and relative yield in four consecutivelyplanted short-season crops. Yurimaguas 1984-1985.