CHAPTER 13
CIRAD/CIAT Rice Project:
Population Improvement
and Obtaining
Rice Lines for the
Savannah Ecosystem

Marc Châtel^[25]
Yolima Ospina^[26]
Francisco Rodríguez^[26]
Victor Hugo Lozano^[26]

Marc Châtel

Abstract

Since 1996, following recommendations by CIAT, the collaborative project on rice improvement between CIAT and CIRAD is phasing out conventional breeding through intraspecific crosses. It will now concentrate on broadening the crop’s genetic base. For upland rice, new breeding strategies include developing populations with broad genetic bases and their improvement through recurrent selection. Use of a recessive male-sterility gene facilitates developing rice populations for different ecosystems and specific sites. In Colombia, upland rice populations were improved through two methods of recurrent selection, with fertile plants being selected at each stage to develop lines through the pedigree method. The most advanced lines are being evaluated for yield in cooperation with the Colombian Corporation of Agricultural Research (CORPOICA, the Spanish acronym). The project expects to continue developing and improving populations, to develop and evaluate lines from these populations, and to share them with national upland rice programmes in Latin America.

Resumen

Desde 1996, en respuesta a las recomendaciones del CIAT, el proyecto de cooperación en arroz entre el CIAT y el CIRAD ha disminuido las actividades de mejoramiento convencional por cruces intra-específicos y se ha concentrado en la ampliación de la base genética del cultivo. Para el arroz de secano el desarrollo de poblaciones con amplia base genética y el mejoramiento de las mismas mediante selección recurrente son las nuevas estrategias. La creación de poblaciones se facilitó por la utilización del gen recesivo de androesterilidad que permitió crearlas para diferentes condiciones y sitios. En Colombia se mejoraron poblaciones de arroz de secano utilizando dos métodos de selección recurrente. Además, en cada etapa del proceso se seleccionaron plantas fértiles para desarrollar líneas a través del método de pedigrí. Las líneas más avanzadas se están evaluando en ensayos de rendimiento en cooperación con la Corporación Colombiana de Investigación Agropecuaria (CORPOICA). En el futuro el proyecto seguirá creando y mejorando poblaciones desarrollando y evaluando líneas obtenidas de esas fuentes, y compartiendo éstas con los programas de mejoramiento de arroz de secano de América Latina.

Introduction

The classic methods of genetic improvement were, and still are, the main ones for developing and releasing innumerable varieties of upland rice in South American countries such as Bolivia, Brazil and Colombia (INGER, 1991). These results, however, are notorious in having consequences such as the narrowing of the genetic bases of the products made available to farmers (Cuevas-Pérez et al., 1992; Rangel et al., 1996; Montalván et al., 1998). Hence, the task of organizations such as CIAT (Cali, Colombia) and CIRAD (Montpellier, France) is to seek new methods of improvement that would permit the continued production of varieties while broadening the variability of germplasm available to farmers.

Since 1996, the strategy for efforts executed within the framework of the agreement between the CIAT Rice Project and CIRAD has been to reduce the production of fixed lines directly related to classic improvement by crossing lines and/or varieties within the japonica group. In contrast, the resources directed towards broadening the genetic base of upland rice have increased, as also has the use of the population improvement method (Châtel et al., 2001).

As a starting point for this project’s objectives, researchers, by exploiting rice genetic resources, have created populations with broad genetic bases (Châtel and Guimarães, 1998) and which have been improved through recurrent selection. To facilitate the construction and random recombination of populations, the recessive male-sterility gene (ms) of the mutant indica ‘IR36’ was used (Singh and Ikehashi, 1981).

To obtain varieties with distinct and broader genetic bases than those currently available in the Latin American market, the broad genetic variability present in these populations must be exploited. This result must occur through plant selection and use of methods such as the pedigree, mass selection, modified mass selection or combinations of them in all the various stages of the population improvement process. In this context, the Project has developed a series of segregating lines, of which the most advanced are being evaluated in yield trials in the savannah ecosystem at the ‘La Libertad’ Experiment Station (LLES), Villavicencio, Colombia.

As part of the CIRAD/CIAT Rice Project’s strategy, the number of lines developed, using population improvement, has grown slowly since 1997, compared with the number of lines obtained through crosses. In 2002, more than 90% of fixed lines evaluated and selected by the scientists working in the Project came from the populations that are being improved through recurrent selection.

As part of the strategy to support the region’s countries, the Project not only created and sent segregating lines, but also different populations to national programmes (Châtel and Guimarães, 1998). Partners in this strategy include Bolivia, Brazil and Colombia. In Asia, because China is interested in the methodology and in upland materials, the Project has also shipped several lines and some populations to the Food Crops Research Institute, Province of Yunnan (Tao et al., 2000). As well as this global effort, some countries such as Cuba, request specific populations for local problems (Pérez Polanco et al., 2000; Chapter 11, this volume).

This chapter aims to present, as an example, the use of the population improvement strategy for some populations managed in Colombia, the results obtained with line development and the advances achieved by the international collaborative project between CIRAD and CIAT from the viewpoint of improving rice for the savannah ecosystem.

Basic concepts

Before discussing the advances of this Project, some basic concepts used in population improvement through recurrent selection are presented. The method is well known to be efficient for improving those traits that are quantitative and of low heritability. For other types of traits and simpler inheritance systems, other methodologies are more efficient, even though population improvement can also function with these, as Borrero et al. (2000) indicated in their work on resistance to the rice ‘hoja blanca’ virus (RHBV; causal agent of white leaf disease). One advantage of the method is that, on the one hand, it permits the breaking of genetic linkage blocks (Hanson, 1959) that, in rice, have been linked through generations of selection and self-pollination and, on the other hand, releases genetic variability by means of successive cycles of crosses (recombinations).

The continuous selection of traits of interest, generation after generation, permits the accumulation of favourable alleles for the traits being improved (increase of allelic frequency). Hence, this method is said to be slow, with results being obtained on a medium-to-long-term basis, but that it produces genetic improvement in populations. Innumerable examples demonstrate the efficiency of the method, principally for open-pollinated crops such as maize (Jenkins, 1940) and certain autogamous crops such as soybean (Guimarães, 1985; Piper and Fehr, 1987), wheat (Altman and Busch, 1984), cotton (Miller and Rawlings, 1967; Meredith and Bridge, 1971) and barley (Bajaj et al., 1990).

Rice composite populations with a recessive gene for male sterility, perform, in principle, as an open pollinated crop. At flowering, fertile plants self-pollinate their own flowers but, at the same time, the pollen they produce also naturally pollinates the male-sterile plants in their midst. This process permits the release of a large quantity of genetic variability while enabling the combination of genes of various parental materials in the same segregating population. Usually, the initial stages of improving these populations present a high level of segregation for a large number of traits and, on advancing the improvement process, some traits - those with the simplest genetic control - are quickly fixed and the population becomes more uniform.

To develop varieties and take advantage of the genetic variability available in each segregating generation within the population improvement process, plant breeders select fertile plants to initiate the process of developing lines through the classic improvement methods of pedigree, mass selection or modified mass selection. This chapter presents advances in population evaluation, and information on lines obtained this way.

One of the Project’s tasks is to register and preserve the populations existing in the region. Each year, some populations are taken to the field only to harvest the male-sterile plants recombined at random (maintaining the population). Currently, 30 populations are registered in the catalogue (Châtel and Guimarães, 2000), and are available from CIAT to any interested person or institution.

Evolving studies

The Project’s principal objective in the population improvement of upland rice for the savannah ecosystem, carried out in collaboration between CIRAD and CIAT, is to develop, adapt and improve populations that are principally from the japonica group.

According to Taillebois and Guimarães (1989), the first upland rice populations in Latin America were created within the framework of the project between Embrapa Arroz e Feijão (Goiânia, Brazil) and CIRAD (at that time, IRAT) during 1984-1990. In 1992, the project between CIRAD and CIAT, in Colombia, was created, introducing into this country from Brazil, populations CNA-IRAT 5 and CNA-IRAT A.

The two introduced populations were planted at the LLES under acid-soil conditions to determine the potential and adaptation of these germplasm materials to Colombian conditions. According to Guimarães et al. (1995), CNA-IRAT A performed better than expected. However, the evaluation indicated that, despite having potential, this population did not possess the genetic variability desired for all the traits regarded as priority by the Project.

The next step was to create a population for a specific site, that is, a population that, besides possessing a broader genetic base, would also have genes from elite lines selected locally and with traits of interest. Châtel et al. (1997a) describe the constitution of and strategy for creating this new population (PCT-4), which originated from introducing variability into CNA-IRAT A. In its turn, PCT-4 was used as the base germplasm for developing population PCT-11 (Ospina et al., 2000).

In parallel with the work of creating populations, executing the strategy of improving the populations of introduced germplasm was also carried out. The two priority traits for selection in these populations were resistance to leaf blast and tolerance of the insect pest Tagosodes orizicolus Muir (Châtel et al., 1997b). Recently, population improvement through recurrent selection focused on PCT-4 and PCT-11.

In addition to this work, and with the idea of generating fixed lines for their distribution to national programmes in the region - one of the objectives of this collaborative project on population improvement - the two populations were treated as sources of genetic variability, generating segregating lines that were selected, using the pedigree method.

Strategies for population improvement

As is well known, the recurrent selection method is a cyclic continuous process that involves three basic stages: selection of plants or families (selection units); evaluation of the selections; and recombination of the best plants or families (recombination units). To improve the populations, this Project used two units of selection: S₀ plants and S_0:2 progeny or families.

The first strategy was based on phenotypic recurrent selection or mass selection in both sexes. The S₀ plants of each recurrent cycle are the selection units and, at the same time, of recombination. Hence, each cycle needed only one planting of the population in the site where pressure exists for the traits being sought.

When the strategy requiring the evaluation of families was used, fertile S₀ plants were selected during the normal cropping season in Colombia, that is, from March to September at the LLES. The S_0:1 generation was advanced during October to February at the Palmira Experiment Station (PES). The S_0:2 seed was harvested at PES and planted at the LLES during the following cropping season. The S_0:2 lines were evaluated and compared with three checks. Overall, the experimental design of Federer’s augmented blocks was used (1956; FAB). Thus, each recurrent cycle was completed in 2 years.

Phenotypic recurrent selection in both sexes

Populations PCT-4, PCT-5 and PCT-A were subjected to three phenotypic recurrent selection cycles in both sexes for the traits resistance to leaf blast and resistance to the RHBV. The methodology used was based on evaluation and selection throughout the vegetative development (from germination to flowering) of those plants that phenotypically performed as resistant to the two diseases. In the field, any plant that appeared susceptible during any stage of its development was immediately eliminated.

So that the process of population improvement could be broadened to include other traits, and be of more interest to national programmes, when harvesting the male-sterile plants, only those fulfilling the minimum agronomic requisites were chosen. Given that only healthy plants reached harvest, selection was based on the two sexes, that is, only healthy fertile plants pollinated the neighbouring, also healthy, male-sterile plants. Ospina et al. (2000) describe the results they obtained by using this strategy, and concluded that only one selection cycle was sufficient to significantly reduce the number of diseased plants in the populations.

Table 1. Evaluation for resistance to the rice ‘hoja blanca’ virus (causal agent of white leaf disease) in the S₂ lines of populations PCT-4, PCT-5 and PCT-A, conducted at the Palmira Experiment Station, Colombia, 1999. A scale of 1 to 9 was used, where 1 is highly resistant and 9 is highly susceptible.

a. Acronyms are FEDEARROZ for Federación Nacional de Arroceros, Colombia; ICA for Instituto Colombiano Agropecuario; and IRRI for International Rice Research Institute.

With respect to line development and comparisons of selection strategies after three cycles of mass recurrent selection for RHBV, 107 S₂ lines from the three populations were evaluated at PES in 1999. Results of these evaluations showed that 54.2% of lines, originating from the populations with broad genetic bases and being improved by recurrent selection, presented resistance to RHBV (Table 1). These results are comparable with those obtained by other improvement programmes that use classic methods for autogamous plants. The difference and the advantage of lines generated by the Project lie in their broader genetic bases. However, this affirmation is based on the number of and origin of the parental materials involved in the crosses that produced the populations. At this stage of work, results are still not available for use with more precise tools such as genetic markers to compare the genetic bases of different materials.

These materials, with their newly created genetic variability, are being subjected to population improvement in the framework of the CIRAD/CIAT Rice Project. They, and the lines derived from them, are available to national genetic improvement programmes to use for the two priority traits of the rice crop in Latin America. During the 2001 cropping season, the three improved populations were planted at the LLES, and S₀ plants were selected for improved line development by the pedigree method.

Recurrent selection based on S_0:2 lines

Population PCT-4 is site-specific for the Colombian acid-soil savannah conditions. For this population, the strategy used was that of improvement through recurrent selection based on the evaluation of S_0:2 progeny or families. Since 1995, population PCT-4 was subjected to three recurrent cycles, using this alternative and giving origin to the population identified as PCT-4\SA\1\1,SA\1,SA\1 (where ‘SA’ is the Spanish abbreviation of suelos ácidos, that is, ‘acid soils’ (Ospina et al., 2000).

Taking advantage of this population, some variants were made to develop basic knowledge and better understanding of the effect of the recombination process on rice populations. Ospina (2002) developed her thesis on comparing two alternatives: one cycle of selection and two and three recombinations after selection.

Another study of the process attempted to discover the possible effect of successive recombination after one selection. To do this, after the first cycle of recurrent selection for acid soils (\SA) population PCT-4 was recombined three times (\SA\3). The S₀ seeds of the resulting population (PCT-4\SA\3\1) were planted at the LLES in year 2000. The best 240 fertile S₀ plants were selected and advanced to S_0:1 generation (S_0:2 seeds) in year 2001 at PES. In 2002, these 240 S_0:2 lines were planted at the LLES, using the FAB design with three checks. The result of this selection will be compared with population PCT-4\SA\0\1 and the original.

For each recurrent cycle after the evaluation of the S_0:2 lines, 30% of the best performers were selected, based on the trial’s results. The best lines were recombined, using the remnant seeds of the S₀ plants, that is, the S_0:1 seeds. Thus, three cycles of recurrent selection were completed for acid soils.

Once improved populations were obtained by the two methods (one selection, followed by three recombinations; and three cycles of recurrent selection, one after the other), a comparison of performance of the lines extracted from both was made to determine the genetic gains obtained (Ospina et al., Chapter 17, this volume).

Developing lines from the improved populations

Extracting segregating lines

Extracting segregating lines from the populations improved through recurrent selection in the context of the Project is an important part of the work’s general strategy. To fulfil this aspect, fertile plants were chosen from all possible sources of variability that appeared during the stages of population improvement. These genotypes comprise the starting point for developing promising lines, future varieties and/or potential parental materials for national programmes of genetic improvement.

Table 2. List of all the segregating lines evaluated. These originated from different upland rice populations in the collaborative CIRAD/CIAT Rice Project, ‘La Libertad’ Experiment Station, Villavicencio, Colombia, 2002.

As a result of work over the previous 3 years, in the 2001 planting season at the LLES, 291 lines were chosen, and advanced and selected through the classic pedigree method. These materials originated from different populations and generations (Table 2).

Advanced generations represent fixed lines that have passed through the entire agronomic process of selection and evaluation. Recently, the best-performing lines were selected at the LLES and PES. Currently, some of these materials are in preliminary or advanced yield trials at the LLES.

Yield trials

Generally, these trials are planted under acid-soil conditions, with fertilizer applications of 300 kg ha^-1 of dolomitic lime applied 30 days before planting; 177 kg ha^-1 of nitrogen (split as 59 kg ha^-1 at 20, 35 and 45 days after planting); 155 kg ha^-1 of phosphorus at planting; and 116 kg ha^-1 of potassium (split as 58 kg ha^-1 at planting and 29 kg ha^-1 at 20 and 35 days after planting). Where not strictly necessary, no kind of chemical fumigation was applied to control diseases or insect pests. The experimental design was randomized complete blocks with three replications. Evaluations were carried out on the principle agronomic traits, and all plots were harvested to estimate the yield of the lines

Table 3. Performance of the best line developed from population PCT-4 in yield trials conducted in 2000 and 2001 at the ‘La Libertad’ Experiment Station, Villavicencio, Colombia.

a. PCT-4\SA\1\1 is the nomenclature for a selection for acid soils after one recombination, which corresponds to one cycle of recurrent selection (Châtel and Guimarães, 1997).

During the 2000 and 2001 cropping seasons, yield trials were planted at the LLES. The most promising lines were compared with three commercial checks developed through traditional breeding methods (Oryzica Sabana 6, released in 1992; Oryzica Sabana 10, released in 1994; and Line 30 or CIRAD 409, released in 2002). At the same time, 24 advanced lines, selected from the first cycle of improvement through recurrent selection of population PCT-4, were evaluated. The results of the 2 years showed a yield that ranged from 2000 to 3488 kg ha^-1. The checks Oryzica Sabana 10, Oryzica Sabana 6 and Line 30 (CIRAD 409) yielded, on average, 2000, 2633 and 2931 kg ha^-1, respectively.

The combined analysis of the 2 years showed that the three best-performing lines identified in 2000 and line PCT-4\SA\1\1>975-M-2-M-3, con- firmed their excellent performance, with an average yield of 19%, 32% and 74% more than CIRAD 409, Oryzica Sabana 6 and Oryzica Sabana 10, respectively. Their cycle is equal to that of the earliest maturing check, CIRAD 409. One conclusion from the 2000 trial is that it is possible to break the correlation that exists between early maturity and yield potential, which was confirmed in the 2001 trial. The data for line PCT-4\SA\1\1>975-M-2-M-3 and the checks are presented in Table 3. During 2002, the trial was repeated with the collaboration of CORPOICA at five different sites: two at the LLES and three in the Altillanura, a region of the Colombian Eastern Plains.

Table 4. Number and percentages of lines selected by plant breeders during the First International Workshop on Selecting Upland Rice carried out at the ‘La Libertad’ Experiment Station, Villavicencio, Colombia, 7-11 August 2000.

a. The lines selected by Argentina are identical to those selected by Bolivia. Bolivian breeders trained those of Argentina, who are initiating work with upland rice improvement.

Selecting lines for national programmes

One major characteristic of the CIRAD/CIAT Rice Project is to provide plant breeders in the region with opportunities to select from populations that are undergoing improvement. National programmes are demanding varieties adapted not only to savannahs, but also to other types of upland environments. In 2000, with the collaboration of Embrapa Arroz e Feijão, the ‘First International Workshop on Selecting Upland Rice’ was organized in Villavicencio, Colombia. In 2002, the Second Workshop was carried out in Santa Cruz de la Sierra, Bolivia, organized by CIAT-Bolivia. The objectives of the events were:

• Promote the integration of upland-rice breeders of the region
• Share experiences in managing segregating populations and developing fixed lines for the savannah ecosystem
• Select, in the experimental field, lines for later introduction to the respective countries
• Train plant breeders in managing the recurrent selection method.

Workshop participants came from six countries: Argentina, Bolivia, Brazil, Colombia, Cuba and Venezuela.

Table 4 shows the selections of lines made by the participants at the First Workshop. The participants selected between 8% and 20.6% of all lines, with Colombia and Brazil selecting the most. The principle traits of the selected lines were early maturity, modern plant type (small number of tillers and erect architecture), long slender grains (of especial interest to Brazil), resistance to blast and high yield potential.

This type of event should be repeated as part of the Project’s strategy. It is a key element in maintaining the group’s motivation, improving the participants’ technical knowledge, keeping the Project up to date with the demands of each national programme, and selecting materials for introducing into different countries. The ‘Third International Workshop on Selecting Upland Rice’ was held at Villavicencio, Colombia, in 2003.

Evaluating genetic diversity

The exploitation of the broad genetic variability in populations from recurrent selection should result in varieties with distinct and broader genetic bases than those currently available on the Latin American markets. As we have seen, this result should occur through plant selection and use of methods such as those of pedigree, mass selection and modified mass selection, or combinations of these in all stages of population improvement.

In this context, the Project has developed a series of segregating lines, of which the most advanced are being evaluated in yield trials in the Colombian savannah ecosystem. As already described, the results showed that it is possible to break the correlation between early maturity and yield potential.

The use of biotechnological tools such as molecular markers should permit:

• Evaluation of the genetic diversity of lines developed from populations, comparing them with the commercial varieties already existing in Latin America for the upland ecosystem

• Understanding of the evolution of genetic diversity through different crossing cycles, together with the determination of the adequate number of recombinations in the parental alleles, during both the populations’ development and their improvement through recurrent selection.

To use an example of the constitution and genetic difference expected from the new materials, when these were compared with the varieties currently planted in the region: line PCT-4\SA\1\1>975-M-2-M-3 originates from a mixture of genes from at least 40 direct parental materials (father and mother) produced by improvement programmes in different parts of the world. In Colombia, the commercial varieties used as checks in the yield trials carry genes from a maximum of four direct parental materials.

Final comments

The CIRAD/CIAT Rice Project has achieved one of the goals incorporated into its initial strategy: to develop and make available to plant breeders in Latin America populations with broad genetic bases. To better understand the functioning of this methodology in the rice crop and to better train personnel in the region, the Project worked with three populations, using two alternatives of recurrent selection, one through mass selection and the other based on families. The improved germplasm from the application of these two alternatives is available as a source of genotypes with an increased frequency of favourable genes for the two traits subjected to selection pressure (leaf blast and RHBV).

Even though the Project must work with the possibility of generating fixed lines with genetic bases that are distinct from those present in the region’s commercial varieties, that is part of its original idea, to seek to offer national programmes alternatives that will lead to the release of new and improved varieties. This material will be delivered in the near future through new nurseries to be set up by CIAT (i.e. CIAT observation nurseries).

To develop these lines, as already mentioned, advantage is taken at all stages of recurrent selection to extract fertile plants and improve them through the pedigree method.

Currently, advanced lines of population PCT-4 that had, over the last 3 years, yield and agronomic performance superior to the checks are in the final stage of evaluation. One is surpassing the best check in yield by almost 20% and may be released by the Colombian programme as an alternative for increasing yields and broadening the genetic base.

In the next few years, the Project will continue working on the following activities:

Create populations of recurrent selection, according to the demands of national programmes, which are extending from varieties for savannahs to others for upland ecosystems

• Continue improving populations that today are being managed by the Project
• Develop and evaluate lines obtained from these sources
• Share advanced lines with upland rice improvement programmes in different countries of the region.

References

Altman, D.W. & Busch, R.H. 1984. Random intermating before selection in spring wheat. Crop Sci., 24: 1085-1089.

Bajaj, R.K.; Bains, K.S.; Chahal, G.S. & Khbhra, A.S. 1990. Effect of intermating and selection in barley. Crop Improv., 17: 54-58.

Borrero, J.; Châtel, M. & Triana, M. 2000. Mejoramiento poblacional del arroz irrigado con énfasis en el virus de la hoja blanca. In E.P. Guimarães, ed. Avances en el mejoramiento poblacional en arroz, pp. 105-118. Santo Antônio de Goiás, Brazil, Embrapa Arroz e Feijão.

Châtel, M. & Guimarães, E.P. 1997. Recurrent selection in rice, using a male-sterile gene. Cali, Colombia, CIRAD-CA & CIAT. 70 pp.

Châtel, M. & Guimarães, E.P. 1998. Catalogue registration to manage rice gene pool and population improvement. 1st ed. Cali, Colombia, CIRAD/CIAT Rice Project, CIRAD-CA & CIAT. 62 pp.

Châtel, M. & Guimarães, E.P. 2000. Catalogue registration to manage rice gene pools and populations improvement. 2nd ed. Cali, Colombia, CIRAD/CIAT Rice Project, CIRAD-CA & CIAT. 75 pp.

Châtel, M.; Guimarães, E.P.; Ospina, Y. & Borrero, J. 1997a. Utilización de acervos genéticos y poblaciones de arroz de secano que segregan para un gen de androesterilidad. In E.P. Guimarães, ed. Selección recurrente en arroz, pp. 125-138. Cali, Colombia, CIAT.

Châtel, M.; Ospina, Y. & Borrero, J. 1997b. Recurrent selection breeding, using gene pools and populations with recessive malesterile gene and conventional breeding; annual report of the collaborative project. Cali, Colombia, CIRAD, CIAT & FLAR. 65 pp.

Châtel, M.; Ospina, Y.; Rodríguez, F. & Lozano, V.H. 2001. Composite population breeding for upland savannas and lowland rice ecosystems; annual report 2001. Cali, Colombia, CIRAD/CIAT Rice Project, CIRAD-CA & CIAT. 33 pp.

Cuevas-Pérez, F.E.; Guimarães, E.P.; Berrío, L.E. & González, D.I. 1992. Genetic base of irrigated rice in Latin America and Caribbean, 1971 to 1989. Crop Sci., 32(4): 1054-1059.

Federer, W.T. 1956. Augmented (or hoonuiaku) designs. Hawaïan Planter’s Rec., 55: 191-208.

Guimarães, E.P. 1985. Genetic improvement of soybean from populations developed by alternative strategies of recurrent selection strategies. Ames, IA, Iowa State University. 116 p. (PhD dissertation)

Guimarães, E.P.; Châtel, M.; Ospina, Y. & Borrero, J. 1995. Mejoramiento de arroz para suelos ácidos; Informe annual 1993A-1994B. Cali, Colombia, CIRAD/CIAT Rice Project, CIRADCA & CIAT. 184 pp.

Hanson, W.D. 1959. The breakup of initial linkage blocks under selection mating systems. Genetics, 44: 857-868.

INGER (International Network for Genetic Evaluation of Rice). 1991. Cruzamentos de arroz América Latina. Vol. 1. Cali, Colombia, CIAT. 426 pp.

Jenkins, M.T. 1940. The segregation of genes affecting yield of grain in maize. J. Am. Soc. Agron., 32: 55-63.

Meredith, W.R., Jr. & Bridge, R.R. 1971. Breakup of linkage blocks in cotton, Gossypium hirsutum L. Crop Sci., 11: 695-698.

Miller, P.A. & Rawlings, J.O. 1967. Breakup of initial linkage blocks through intermating in a cotton breeding program. Crop. Sci., 7: 199-204.

Montalván, R.; Destro, D.; Silva, E.F. da & Montaño, J.C. 1998. Genetic base of Brazilian upland rice cultivars. J. Genet. Breed., 52: 203-209.

Ospina, Y. 2002. Respuesta a la selección y a ciclos de recombinación en la población de arroz (Oryza sativa L.) de secano PCT-4. Palmira, Facultad de Ciencias Agropecuarias, Universidad Nacional de Colombia. 76 pp. (MSc thesis)

Ospina, Y.; Châtel, M. & Guimarães, E.P. 2000. Mejoramiento poblacional del arroz de sabanas. In E.P. Guimarães, ed. Avances en el mejoramiento poblacional en arroz, pp. 241-254. Santo Antônio de Goiás, Brazil, Embrapa Arroz e Feijão.

Pérez Polanco, R.; Châtel, M. & Guimarães, E.P. 2000. Mejoramiento poblacional del arroz en Cuba: situación actual y perspectivas. In E.P. Guimarães, ed. Avances en el mejoramiento poblacional en arroz, pp. 131-144. Santo Antônio de Goiás, Brazil, Embrapa Arroz e Feijão.

Piper, T.E. & Fehr, W.R. 1987. Yield improvement in a soybean population by utilizing alternative strategies of recurrent selection. Crop Sci., 27: 172-178.

Rangel, P.H.N.; Guimarães, E.P. & Neves, P. de C.F. 1996. Base genética das cultivares de arroz (Oryza sativa L.) irrigado do Brasil. Pesq. Agropecu. Brasil., 31(5): 349-357.

Singh, R.J. & Ikehashi, H.I. 1981. Monogenic male sterility in rice: induction identification and inheritance. Crop Sci., 21(2): 286-289.

Taillebois, J. & Guimarães, E.P. 1989. CNA-IRAT 5 upland rice population. Intl Rice Res. Newsl., 14(3): 8.

Tao, D.; Hu, F.; Yang, Y.; Xu, P. & Li, J. 2000. Yunnan, China: Mejoramiento poblacional de arroz para rendimiento de granos, resistencia a Piricularia, tolerancia a frío y calidad de granos. In E.P. Guimarães, ed. Avances en el mejoramiento poblacional en arroz, pp. 145- 154. Santo Antônio de Goiás, Brazil, Embrapa Arroz e Feijão.