International partnership for rice improvement in Latin America: CIRAD, a case study

In Latin America and the Caribbean (LAC), rice is an important staple food. A significant increase in rice production has been achieved mainly through the release and adoption of high yielding varieties. During the last 40 years, about 300 varieties were released. They were developed by national and international rice research institutions. The Centre for International Cooperation in Agricultural Research for Development, Food Crop Department (CIRAD-CA), in Montpellier, France, worked in collaboration with Embrapa Rice and Beans, Goiânia, Brazil, and the International Centre for Tropical Agriculture (CIAT), Cali, Colombia, in the development of varieties for LAC. This study looks at the contribution made by CIRAD’s international partnership to rice improvement, with emphasis in Latin America. The presence of CIRAD’s lines was detected in the breeding programs in Brazil, Bolivia, Colombia, Mexico, Guatemala and Peru. The following results were obtained: a) 61 unique lines developed by CIRAD were used as parents in more than 2,000 crosses in the region; b) five CIRAD’s lines were directly released as commercial cultivars; c) 25 varieties released to farmers in the region have at least one line from CIRAD participating on its crosses; and d) six breeding lines are in the pipeline for release in Brazil, Colombia, Nicaragua and Peru.


INTRODUCTION
Rice is the most cultivated food crop in the world, right after wheat. Currently, the total production is approximately 592 million tons, coming from 151 million hectares (FAOSTAT, 2001). Rice is consumed by two-thirds of the world's population and is the staple food for most of it. Significant progresses have been made to increase rice production worldwide, and genetic improvement has had a major role in it. Looking at the prediction for the years ahead, in 2020 the world's population will be around 7,5 billion (Maclean et al., 2002); this will require a rice production increase of almost 50% to meet the demand (IRRI, 1996). Therefore, an increase in production by raising the yield potential is the main goal of almost all breeding programs.
Due to rice importance and distribution in the world, the International Agricultural Research Centres (IARC) have designated three of its 16 units to work on rice research. They are the International Rice Research Institute (IRRI), Los Baños, Philippines; the International Centre for Tropical Agriculture (CIAT), Cali, Colombia; and the Africa Rice Centre (former West Africa Rice Development Association In 1981, CIRAD started a joint project on rice improvement with Embrapa Rice and Beans. At that time the first institution was known as IRAT (Institute for Research in Tropical Agriculture) and the second as CNPAF (Rice and Beans National Research Centre). One of the objectives was to bring and use genetic resources developed in Africa to the Brazilian national breeding program (Châtel et al., 1982). In 1992 CIRAD started a collaborative project with CIAT rice project in order to foster and strengthen breeding activities for the Latin America's upland ecosystems (Châtel et al., 1996).
This paper analyses the contribution made by the germplasm developed by CIRAD's international partnership to rice improvement, with an emphasis in Latin America.

MATERIAL AND METHODS
The methodology used in this study to produce data considered all breeding lines and varieties developed by CIRAD (or former IRAT) in its rice breeding programs in Africa and in Latin America. The information was gathered based upon literature review and database surveys. All lines generated by CIRAD, which were used in crosses or as direct releases as commercial varieties in Latin America, were considered as CIRAD's contribution to rice improvement in the region.
Crossing program records were analyzed based on the information published by INGER-América Latina (1991) and Martínez and Cuevas-Pérez (1995). In Brazil, at Embrapa Rice and Beans, "Instituto Agrônomico de Campinas" (IAC), and Embrapa Temperate Climate; in Colombia, at the "Instituto Colombiano Agropecuario" (ICA) and CIAT; in Mexico, at the "Instituto Nacional de Investigaciones Forestales y Agropecuarias" (INIFAP); and in Peru, at the "Instituto Nacional de Investigaciones Agropecuarias y Agroindustriales" (INIAA). To develop these analysis, three sources of contributions were accounted for: a) participation as parental material in any combination made by national breeding programs; b) participation as parents of commercial varieties developed by national programs; and c) participation as breeding lines released as varieties.
To access the contribution in the crosses, the following information was recorded for a given breeding program and specific period: total number of crosses made by the institution, total number of crosses where CIRAD's line was involved in the combinations as female or as male parent, and the number of lines from CIRAD used by the national program.
Variety release database was consulted to determine the number of lines from CIRAD that made directly to farmers' field or were involved in one of the crosses that produced such varieties. To attribute the contribution to CIRAD, we only looked at the parents and the grandparents of each variety released in the region.

RESULTS AND DISCUSSION
The results will be presented and discussed considering the three types of contribution mentioned before: as a source of parental lines in crosses, as part of the progenitors in released varieties and as commercially released breeding lines.

Source of parental lines for crossing programs
The "International Network for Genetic Enhancement of Rice" (INGER) was a powerful mechanism to exchange germplasm between international and national programs; and also for the exchange between national programs (INGER-América Latina, 1990). Most of CIRAD's germplasm came to Latin America through this mechanism. Based on the information collected from INGER-América Latina (1991) and Martínez and Cuevas-Pérez (1995), summarized in Table 1, it was possible to show the number of lines used by national programs as parental materials in their crosses. Brazil, after Colombia, used a larger number of lines taking more advantage of the germplasm developed by CIRAD, than the others, as shown in Table 1. To a certain extent, this result could be expected because upland ecosystem is priority for the Embrapa Rice and Beans breeding program, when   Breeding and Applied Biotechnology, v. 2, n. 4, p. 639-648, 2002 compared to other programs and countries in the region, as well as with CIAT in Colombia. Most CIRAD's work targeted this ecosystem. Table 1 shows that more than 16 thousand croses were made by the four countries using CIRAD's germplasm in their crossing programs. At least one line developed by CIRAD was present in 12.3% of the total combinations. Almost half of the crosses made in Colombia involved CIRAD's lines, and that is because the local and CIAT's breeding programs were counted together, and also because CIAT has the regional mandate to carry out activities for the tropical savannas ecosystem.  (EMBRAPA, 1981) it was no surprise to see these two varieties involved in the crosses of the most frequently used CIRAD's lines.
IAC only used eight lines, seven of which are among the ones used by Embrapa Rice and Beans. The difference is that they were used in crosses involving others local progenitors. They were used in 77 combinations (10 times as female and 67 as male). Similar to what was observed for Embrapa Rice and Beans breeding program, the early maturating line IRAT 112 with 29 presences was the most used germplasm. Other important lines were IRAT 10 (11), IRAT 116 (10), IRAT 140 (10), and IRAT 146 (10), and none of them were among the most frequent ones in Embrapa's program.
Embrapa Temperate Climate only used three CIRAD's varieties in its breeding program, two (IRAT 104 and IRAT 135) were different from the other 58. Fifteen crosses were made with these three lines and they were always used as a male parent. This low presence in this breeding program is not surprising because the target ecosystem for this institution is the irrigated one and most of CIRAD's germplasm introduced in the country came from the upland ecosystem, as mentioned before.
The prevalence of lines such as IRAT 112, IRAT 195, and IRAT 13 in these breeding programs is justified by the fact that they have high yield potential, modern plant type, and earliness (CIRAD, 1993), priority traits for upland varieties in Brazil (EMBRAPA, 1981 (Table 6). CIAT identified this series as a source of resistance to Rice Hoja Blanca Virus (RHBV), an important disease in the region, and to Crop Breeding and Applied Biotechnology, v. 2, n. 4, p. 639-648, 2002 the insect vector Tagosodes orizicola (Muir) (Pardey et al., 1996). Even though the disease and its insect vector were never observed in that country, Makalioka was identified as the resistant gene source to Tagosodes orizicola (Muir) (Martínez et al., 1994). Sources of resistance to the RHBV were found in the IRAT 120 series coming from the japonica parent Chianan 8 (CIAT, 1989).
In Mexico, rice improvement started in 1949 and was responsible for the release of 42 varieties during the period 1956(Ortega Arreola et al., 1992. The savannas located between the States of Campeche and Tabasco have favourable ecological conditions for rice production under both irrigated and upland ecosystems (Rogriguez Avila and Orono Castro, 1992 In Peru, where 95% of rice production is irrigated, the rice-breeding program used CIRAD's lines as parents. In fact the most used lines belong to the IRAT 120 series. As it was previously mentioned, this germplasm was developed in Madagascar for the irrigated rice ecosystem and was identified by CIAT as a source of resistance to the RHBV and Tagosodes orizicola (Muir). The Peruvian breeding objective was to have improved lines which were resistant to the RHBV as a preventive measure, since the disease is not important in the country as it is in Colombia or other tropical Latin American countries (Olaya Viera, 1983Viera, , 1984.

Parental lines in developed varieties or released directly
The use of CIRAD's germplasm in the national breeding programs allowed the countries to release, between 1982 and 2002, a total of 30 varieties. Brazil is the country where more varieties were released (17), followed by Colombia (5), Bolivia (5) and Guatemala (3). In Mexico and Peru there was not any commercial variety that came from selected progenies of crosses involving CIRAD's lines or direct releases. It is an interesting fact to see Guatemala listed among the countries where CIRAD's lines made a contribution to the farmers. In fact, this result is a spill-over from CIAT-Colombia's breeding project from where progenies were introduced and locally selected under favourable upland rice ecosystem with no acid soils and good rainfall distribution, present in Guatemala (Ramírez, 1992).
Since Brazil had the national program that took more advantage of the genetic resource managed by CIRAD, it is expected to concentrate more attention. For the last twenty years  there were 22 rice varieties released for the upland ecosystem in Brazil. Out of these 22 varieties, 17 have CIRAD's contribution, seven came from the joint program with Embrapa, six were developed jointly with CIAT, three were CIRAD's lines directly released to farmers (IREM 16B, Cabaçu, and Rio Verde) and one was developed only by Embrapa (Table 2). These results stressed the importance of CIRAD's germplasm for Embrapa's upland breeding program; the lines from CIRAD were present in 77% of the upland rice varieties released in Brazil in the last two decades.
Among the parents that generated the 17 rice varieties we found nine lines that came from CIRAD. The most common one was IRAT 216, which was also directly released as a variety for the state of Mato Grosso (EMBRAPA, 1992). This line was used because of its modern plant and long grain type (CIRAD, 1993). Table 6 shows that IRAT 124, which also contributed strongly to the released varieties, is a line coming from Makalioka and Chianan 8 cross, resistant to the RHBV and Tagosodes orizicola (Muir) modern plant type and with a good adaptation to favourable upland conditions (present in the Mato Grosso State).
In Colombia, the exploitation of upland acid soil ecosystem in the Altillanura is new; rice germplasm has been seen as an alternative to stimulate it (Sarkarung and Zeigler, 1991). Currently the only upland rice varieties available to farmers are the ones released between 1989 and 2002 by CIAT (Table 3). All varieties came from crosses where there was at least one line from CIRAD involved. Similar to what has been observed for the Brazilian program, IRAT 216 is the most common parent.
Oryzica Llanos 4 is a variety which was released for irrigated and favourable upland ecosystems (Leal et al., 1988), without acidity or severe soil stresses, such as the ones observed in the Altillanura. The line IRAT 122 contributes to this variety because it is a source of tolerance to RHBV.
In Bolivia, favourable upland rice cultivation is based on two different systems: the mechanised and the manual system performed by small farmers (Manchego, 1992;Guzmán, 1994). During the period 1993-2002, five varieties were released in Bolivia attending both systems (Table 4). Three varieties were released for the mechanised system, which came from progenies developed from CIAT's crosses and have IRAT 124 and/or IRAT 216 as parents. Two CIRAD lines bred in Africa were directly released as commercial varieties for the smallholder's system. They are rustic, show a taller plant type that make them better adapted to manual harvesting, and have a medium grain type praised by the producer/ consumer.
For Guatemala there were three releases (Table 5). They all came from a local evaluation of selected introduced progenies of crosses developed by CIAT in Colombia, and distributed by INGER-Latin America (Ramírez, 1992).

Origin of the CIRAD's parents in commercial varieties released in Latin America
Fourteen CIRAD's lines have contributed to rice enhancement and further release of commercial varieties in Latin America (Table 6.). As it can be seen, they originated from CIRAD's breeding projects in West Africa (Côte d'Ivoire and Burkina Faso), Madagascar, and Latin America (Brazil and French Guyana).

Promising lines to be released
The contribution of CIRAD's line to national breeding programs will still be important in the near future. Table 7 shows that there are six upland breeding lines performing well in Brazil, Colombia, Nicaragua and Peru. In Brazil, the line CNA 8812 (CT13226-11-1-M-BR1) was identified as promising for the upland ecosystem. The line shows good drought tolerance, earliness, and excellent grain quality (O. P. de Morais, personal communication).
In Colombia, CIRAD 445 was identified as a potential variety for the coffee growing region with an altitude less than 1450 m (Moreno et al., 2002). The uplandhillside lines CIRAD 446 and CIRAD 447 were identified for the small holders in the Andeans (M.    Valès, personal communication).
In Nicaragua, farmers of the central region for upland and rainfed-lowland ecosystems adopted CIRAD 301. This line is praised for its good adaptation to different stress, for its high yield potential and as well as grain quality (G. Trouche, personal communication). In Peru, CIRAD 409 -selected and released in Colombia (Table 3), is seen as an excellent alternative for the slash and burn rice system in the Pucallpa region (D. White, personal communication).

CONCLUSIONS
The analysis made leads to the following conclusions: • A total of 61 CIRAD's lines were used for crossing in Latin America, and some of them were used in high frequency mainly in Brazil and Colombia, where CIRAD had direct international rice breeding collaboration; • There were more than 2,000 crosses made in by four Latin American Countries where at least one Crop Breeding and Applied Biotechnology, v. 2, n. 4, p. 639-648, 2002 parent was from CIRAD; • Fourteen different CIRAD's lines participated in the genetic background of the commercial varieties released; • Twenty-five commercial varieties released in Brazil, Bolivia, Colombia and Guatemala had at least one CIRAD's line participating in the cross of origin; • Five IRAT lines bred by CIRAD were directly Table 6. List of CIRAD's lines used as parents of the commercial varieties released in Latin America.