PART III: THESES

CHAPTER 15
Genetic Progress
towards Grain Quality
in Rice (Oryza sativa L.)
through Recurrent Selection

César Pompilio Martínez
Silvio James Carabalí
Jaime Borrero
Myriam Cristina Duque
James Silva

César Pompilio Martínez

Rice Project, CIAT, A.A. 6713, Cali, Colombia.
E-mail: [email protected]

Abstract

A population improvement programme, aimed at studying the effect of recurrent selection cycles on rice grain quality - specifically, white belly, grain length and gelatinization temperature - was begun in 2000, at CIAT’s Palmira Experiment Station, Valle del Cauca, Colombia. We used two populations, PCT-6 and PCT-8, which possessed distinctly different genetic backgrounds, but which were both developed for the irrigated conditions of tropical Latin America. Selection criteria took into account the preferences of the region’s consumers. Preliminary results at the end of the first recurrent selection cycle indicated significant differences in selection response, both within and between populations. Although selection increased the proportion of plants with desirable traits, the two original populations responded differently, with PCT-6 responding more favourably than PCT-8. Statistical analysis is under way to estimate genetic gain among selection cycles. Concepts of stability and acceptability were introduced to improve selection criteria for gelatinization temperature.

Resumen

En el año 2000 se inició un programa de mejoramiento poblacional en la Estación Experimental del CIAT en Palmira, Valle, Colombia, con el fin de estudiar el efecto de ciclos de selección recurrente en la calidad de grano del arroz (centro blanco, longitud del grano y temperatura de gelatinización). Se utilizaron dos poblaciones (PCT-6 y PCT-8) con diferentes antecedente genéticos desarrolladas para las condiciones de riego de América Latina Tropical. Los criterios de selección tuvieron en cuenta las preferencias de los consumidores de la región. Los resultados preliminares al término del primer ciclo de selección recurrente indicaron diferencias significativas en la respuesta a la selección impuesta dentro de cada población y entre poblaciones. Aunque la selección impuesta aumentó la proporción de plantas con las características deseadas, las dos poblaciones originales respondieron en forma distinta: la PCT-6 respondió más favorablemente que la PCT-8. Las ganancias genéticas entre ciclos de selección se estimarán al final del segundo ciclo. Los conceptos de estabilidad y aceptabilidad se introdujeron para mejorar la selección para temperatura de gelatinización.

Importance of rice and its quality

Rice is an essential food in the diet of one third of the world’s population. Its production and consumption is concentrated in Asia where more than 90% of the world’s rice is grown and consumed (David, 1991). The 148 million hectares planted throughout the world produce about 590 million metric tonnes of paddy rice per year. Rice, which is grown under a wide diversity of climates, soils and production systems, is subjected to many biotic and abiotic stresses that vary according to site (IRRI, 1993). Consumption per capita and consumer preferences for a given rice type also vary from region to region (Webb, 1991).

The dynamics of rice production and the factors that affect it are highly changeable. The adoption of improved varieties and better crop management practices has brought self-sufficiency to some regions, but not yet to others. Hossain (1996) considers that the impressive growth in rice production over the last 25 years is very difficult to sustain. According to Pingali et al. (1997), the annual growth of rice production in Asia has decelerated from 3.0% before 1975 and 3.2% in 1975-1985 to 1.7% in 1985-1993. The factors responsible for this deceleration include rice prices and the problems caused by intensifying production. In some Asian regions, consumption per capita is dropping, even though, globally, the offer must increase by 50% to 60% in the next 25 years to satisfy demand (Pingali et al., 1997).

According to Sanint and Wood (1998), in Latin America, significant advances in rice production have been obtained in the last 3 decades, making the region almost self-sufficient. Consumers have benefited from reduced prices; consumption per capita has increased from 10 kg in the last 20 years to 30 kg during the 1990s. Rice has become the most important source of calories and protein for 20% of the population with the lowest incomes, besides supplying more calories to the diet than wheat, maize, cassava or potato.

Despite the development of varieties with good grain quality having been an important objective of improvement programmes, today, its relevance is much greater because of (Pingali et al., 1997):

• Changes observed in the technology available for rice production
• Improvement in rural infrastructure
• Economic growth
• Changes in consumers’ acquisitive powers
• Diversification of food patterns.

Another factor is that, under diverse circumstances and scenarios, the consumer is willing to pay higher prices for a given quality and type of rice. Hossain (1996) considers that demand by consumers for rice of better quality can also influence the drop in production.

Different characteristics of ‘grain quality’ in rice largely determine the product’s market price and acceptability. If the consumer does not like the flavour, texture, aroma, appearance or ease of cooking and processing in a new variety, whatever other outstanding trait it may possess loses its value (IRRI, 1985).

The significance of grain quality varies from region to region, depending on the requisites established by the international market, ethnic customs, uses, etc. A given community may demand a certain rice quality that would be unacceptable to another. For example, the preferences and tastes of Asian communities differ from those of Latin America (Martínez and Cuevas-Pérez, 1989). Even within the same country, distinct appreciations of quality may exist.

For the rice market, appearance, translucence and degree of whiteness are very important. Industries and mills pay special attention to the percentage of whole grains, and food processors emphasize characteristics associated with food processing. For nutritionists, nutritional quality is the most relevant attribute and, finally, consumers demand a diversity of rice types (Webb, 1991).

Genetic base and heritability of quality traits

With the increase in the rice offer, economic globalization, and changes in acquisitive power, dietary patterns and working conditions, consumers have become more demanding with respect to grain quality, especially to its appearance and culinary properties. However, plant breeders have made crosses between varieties that differ markedly in their grain characteristics, making selection for quality less predictable and more difficult (Pingali et al., 1997).

Rice quality comes from a polygenic group of traits that are affected by environmental factors, crop management and the resulting interactions among these (Wrigley and Morris, 1997). Grain quality can be considered from several viewpoints: the grain’s external appearance, milling quality, cooking qualities, nutritional quality and quality related to food processing. In this work, by quality, we refer to grain length (GL), presence of white belly (WB) and gelatinization temperature (GELT), taking into account the preferences of consumers in Latin America who prefer long, and slender, translucent rice grains that have an intermediate amylose content and are dry and loose after cooking.

Grain length in rice is measured in millimetres; its size is highly heritable in most environments. Grain length and shape are fixed exceptionally early in segregating generations, and are inherited in a quantitative manner. According to Jennings et al. (1981), GL in F₁ is intermediate among its progenitors, and transgressive segregations for longer and shorter grains are common in F₂ generations.

White belly in the grain refers to the presence of opaque zones within the endosperm that occur in non-glutinous rice when starch and protein particles in the cells are not compacted (IRRI, 1985). Starch grains in the affected areas are spherical with little compaction, in contrast with the compact polyhedral grains that are characteristic of translucent areas (Martínez and Cuevas-Pérez, 1989).

Studies on WB in India and USA interpret this trait as recessive monogenic (wc wb), whereas, in other research, WB appears as a dominant trait. Later research suggests that a multigenic system interacting with environmental factors appears more probable (Chang and Somrith, 1979).

Appearance and degree of WB in the grain are under genetic control, although certain environmental factors partially affect their expression. Grains from the same panicle can differ in opacity. Some varieties such as IR22 do not present WB in any environment, whereas others such as CICA 4 have clear endosperm in some environments and considerable WB in others. Variety IR8 and others present WB in almost all environments (Jennings et al., 1981).

Torres (2001) evaluated 13 commercial varieties released in Colombia, Costa Rica and Venezuela, together with 37 advanced lines, in the Colombian sites of Palmira, Santa Rosa, Ibagué, Saldaña and Montería. He found that Monteria had the most capacity to discriminate between genotypes and exercised the heaviest pressure for WB. In contrast, Saldaña and Santa Rosa showed intermediate pressure and a capacity to discriminate in a similar manner for this trait. In Ibagué and Palmira, WB was low, indicating that these sites were unsuitable for evaluating the effects of selection for WB.

Choi (1979) found that the endosperm without WB was dominant over opaque endosperm and that, in the F₁ generation, the ovary had a greater effect on opacity than did pollen. Guo et al. (1982) found that WB in some crosses performed as a trait that is controlled by polygenes, with a dominant effect.

The gelatinization temperature is the temperature at which starch grains begin swelling, or that temperature at which the granules will swell irreversibly in hot water (Khush et al., 1979). It is classified as high, low or intermediate. On analysing the work of different authors who used distinct populations, the genetic panorama behind GELT in rice starch is seen as complicated.

Rices with high GELT have resulted from crosses between varieties with intermediate and low GELT; crossing varieties with intermediate and high GELT produces materials with intermediate and high GELT; some crosses between varieties with high and low GELT have segregated for high, intermediate and low GELTs (Beachell and Stansel, 1963).

Hsieh and Wang (1988) found that the inheritance of GELT differs considerably according to the progenitors used in the cross. A cross between parents with high GELT with those of low GELT resulted in progeny with high GELT, whereas a cross between intermediate and low classes segregated. Moreover, they found that high GELT was controlled by dominant genes, with additive effects. Crosses among those with a parent possessing high GELT always produced F₂ segregating for high GELT (Jennings et al., 1981).

Masajo (1971), studying GELT in rice, found that heritability estimated by regression of the mean in the F₃ generation on the average index for F₂ was between 76% and 96%. Estimates for the F₃ generation obtained through analysis of variance were between 94% and 97%.

Recurrent selection for improving quality

Plant genetic resources constitute the essential source of genes for improving crops. By recombining favourable alleles in improved varieties, plant breeders have been able to improve crop productivity and quality, and reduce production costs.

Recurrent selection is a method of population improvement that permits the increase of frequency of favourable alleles. It favours multiple recombinations among progenitors that are genetically distant, thus permitting the broadening of the genetic base of the improved germplasm (Châtel and Guimarães 1997).

One reason that justifies population improvement is the difficulty of developing germplasm with high yield potential, resistance to or tolerance of biotic and abiotic problems, good adaptation and excellent grain quality. Unimproved progenitors are almost always poorly adapted to local conditions and, when involved in crosses, usually produce inferior progenies that must be eliminated during selection. Population improvement permits the use of the same sources, minimizing these problems as their genes are already mixed in the populations. The process of gradual accumulation of favourable alleles permits using them without having the negative effects associated with descendance (Guimarães, 2000).

Gradual accumulation of favourable alleles also appears when improved progenitors are used as alternatives in a crossing program. This strategy is the most commonly used by rice genetic improvement programmes to deal with the constant pressure they face to release superior varieties.

Using the strategy of population improvement will help form recombinations of favourable alleles that will fix and obtain as quickly as possible significant genetic gains in grain quality for populations such as PCT-6 and PCT-8.

This chapter aims to present the evaluation of two original populations (PCT-6 and PCT-8) and a recurrent selection cycle for grain quality for each. Genetic progress is observed and the effect of the cycles in terms of genetic gain for grain quality compared.

Materials and methods

The trials were conducted at CIAT’s Palmira Experiment Station (PES), Department of Valle del Cauca, Colombia (33°0'N; 76°30'W; 965 m above sea level; average temperature of 24°C; annual rainfall of 1000 mm).

In January 2000, a preliminary evaluation of populations PCT-6, PCT-7, PCT-8 and GPCT-9 was carried out, using samples of 1000 to 1500 plants per population. The four populations were observed for their agronomic traits and genetic variability, and populations PCT-6 and PCT-8 were selected for presenting the most variability and best plant type in terms of height, tillering, stem strength and vigour. At this stage, the traits for grain quality were not evaluated.

The materials were planted in seedbeds. Between 25 and 30 days, the seedlings were transplanted to the definitive site in a plot that was previously ploughed and prepared in water. Transplanting distance was 0.3 m between rows and between plants. Weed control was both chemical and manual. In accordance with chemical soil analyses, fertilizers were applied as follows: 60 kg ha^-1 of P₂O₅, using triple superphosphate (TSP) as the source; 60 kg ha^-1 of K₂O from potassium chloride (KCl); 120 kg ha^-1 of N from urea; and 4 kg ha^-1 of Zn from ZnSO₄.

The selected populations were designated according to the nomenclature proposed by Châtel and Guimarães (1995), as follows: the prefix ‘P’ stands for ‘population’, followed, without spacing, by ‘CT’, which stands for ‘CIAT’ (i.e., the institution); a hyphen ‘-’; and a consecutive number. The traits for which the population was selected are designated with the acronyms of IRRI’s evaluation system. These and the recombination cycles (before and after selection) are separated with a back slash ‘\’. In our case, the initials CG for the populations signify the selection for calidad de grano (Spanish for ‘grain quality’) and the ‘F’ at the end of the nomenclature indicates the selection of fertile plants in a population.

Population PCT-6\0\0\0 was created at CIAT in 1995 for the tropical irrigated ecosystem by incorporating various materials into population IRAT-MANA 9. The centre had not used some of these materials before in an irrigated rice improvement programme. Population PCT-6 has good yield potential and good plant type. It is early maturing and possesses good genetic variability in terms of long to extra long grains (Martínez et al., 1997).

PCT-8\0\0\0, a tropical irrigated rice population, was developed at CIAT in 1995 by incorporating six lines of different origins into population CNA-IRAT 4\2\1. Population PCT-8 presents good agronomic variability, plants with a good grain type (medium to long) and good yield potential (Martínez et al., 1997).

Table 1. Example of obtaining scores for white belly (WB). The score ‘0’ indicates grains with no WB or are transluscent, and the score ‘5’ opaque grains.

Weighted average = (SN_iS_i)/SN_i

Table 2. Scoring for gelatinization temperature (GELT), using the method according to Martínez and Cuevas-Pérez (1989).

Evaluating grain quality

Grain length (GL)

The methodology used to evaluate GL in rice (Martínez and Cuevas-Pérez, 1989) involves dehusking 5 g of seeds from each plant and polishing them for 15 min. They are then sieved and placed on a dark background. Five grains are chosen at random and measured with a ruler calibrated in millimetres and an average taken for GL. The evaluation scale classifies grains measuring an average length of less than 5.5 mm as short (S); between 5.6 and 6.5 mm, medium (M); between 6.6 and 7.5 mm, long (L); and more than 7.5 mm as extra long (EL).

White belly (WB)

The methodology for evaluating WB (Martínez and Cuevas-Pérez, 1989) involves placing a sample of 3 to 5 g of polished rice from each plant on a dark background. Five grains are taken at random and evaluated against a scale of 0 to 5 by which a weighted average is obtained to indicate the degree of WB in the sample. Table 1 illustrates how the WB values are obtained.

Gelatinization temperature (GELT)

The indirect method of estimating GELT (Martínez and Cuevas-Pérez, 1989) is used, that is, the degree of alkaline dispersion is determined. A sample of 10 polished grains is taken and distributed uniformly in a small plastic box containing 10 ml of a solution of KOH at 1.7% and left for 2 h in an incubator at 30ºC. The grains with low GELT completely dissolve; the endosperm of the intermediate class partially disperses; and those with a high GELT are not affected by the alkali. The degree of alkaline dispersion is determined against an ordinal scale that ranges from 1 to 7 (Table 2).

Recurrent selection cycles

Recurrent selection cycles are based on the selection of fertile S₀ plants. This evaluation is for quality and the later recombination of plants selected according to the criteria established for the traits GL, WB and GELT.

First recurrent selection cycle

Population PCT-6\0\0\0 was planted at PES in the first semester of 2000. About 300 male-sterile plants were marked during flowering and, at harvest, 216 fertile S₀ plants were sampled at random and all male-sterile plants marked. The same was done for population PCT-8\0\0\0, of which 257 fertile S₀ plants and all male-sterile plants were harvested. For each population, seeds from the male-sterile plants were mixed in equal proportions to represent cycle zero (C₀) and the seed stored for later evaluation.

Evaluation

A sample of S₁ seeds harvested from fertile S₀ plants in each population was sent to the grain quality laboratory to evaluate for GL, WB and GELT. The remnant seed was stored for later recombination, once the evaluation cycle was finished.

Selection

Once the S₁ seeds were evaluated in the laboratory, materials were selected according to the parameters described above and using a hierarchical strategy. The first criterion for selection was the variable WB, accepting materials that scored WB < 1. Those with higher scores were rejected, regardless of their scores for GL or GELT. The second criterion was GL, accepting materials with long (L) or extra long (EL) grains, regardless of their GELT. The final criterion was GELT, where materials with high GELT and no segregation were discarded. With this selection scheme, 106 S₁ materials were selected from population PCT-6\0\0\0 and 111 S₁ materials from population PCT-8\0\0\0, bringing together the desired traits of grain quality.

Recombination

For each population, the remnant S₁ seed from each selected S₀ plant after evaluation in the quality laboratory was mixed in equal proportions to form new populations without recombination (PCT-6\CG\0\0F and PCT-8\CG\0\0F). The balanced mixture of each population was planted in the second semester of 2000 at PES. For recombination, about 1500 plants were planted over two planting seasons with an interval of 15 days. During flowering, male-sterile plants were marked and, at the end of the cycle, harvested to carry out a balanced mixture of seed and complete the first recurrent selection cycle (C₁) for grain quality and to produce populations PCT-6\CG\1\0F and PCT-8\CG\1\0F. Some of the seed was stored for later evaluation and the rest of the seed planted again to begin the second recurrent cycle.

Figure 1. Percentages of individuals with extra long (light grey shading), long (dark grey shading) and medium-length (white) grains in two selection cycles of rice populations PCT-6 and PCT-8

Second recurrent selection cycle

In the first semester of 2001, the second recurrent selection cycle was initiated for both populations, following the same steps as for C₁. Seed corresponding to C₁ from the selection in populations PCT-6\CG\1\0F and PCT-8\CG\1\0F were planted and 300 fertile S₀ plants were harvested at random from each population.

Evaluation

The S₁ seeds were taken to the grain quality laboratory at CIAT for analysis for GL, WB and GELT.

Selection

Once the materials were evaluated, selection was carried out, using the hierarchical strategy with the same criteria as used in the previous selection cycle. This resulted in the identification of 136 S₁ materials from population PCT-6\CG\1\0F and 68 S₁ from PCT-8\CG\1\0F.

Recombination

Once the selected materials were identified, a balanced mixture was carried out with the remnant seed to constitute populations PCT-6\CG\1\0F,CG\0 and PCT-8\CG\1\0F,CG\0 without recombining. Each population was planted in the first semester of 2002 at PES for their recombination. About 1500 plants from each population were planted. The same procedure as used for the previous cycle was followed to constitute the second recurrent selection cycle (C₂) for grain quality in both populations. These were designated PCT-6\CG\1\0F,CG\1 and PCT-8\CG\1\0F,CG\1, respectively.

Preliminary results and discussion

Grain length (GL) Rice consumers in Latin America prefer long and slender grains. Hence, one criterion used for selecting individuals was that the grains be either long (L) or extra long (EL).

In population PCT-6\0\0\0, 13% of materials presented EL grain; 78%, L; and 9%, medium length (M). After one selection cycle, population PCT-6\CG\1\0F was found to have 28.4% of individuals with EL grain, 68.2% with L and 3.4% with M. After selection in C₁, the percentages of M and L grains dropped significantly, increasing the proportion of EL, compared with the first cycle (Figure 1).

In population PCT-8\0\0\0, 10.6% of individuals possessed EL grain; 68.6%, L; and 20.8%, M. When conducting C₁ of selection, the population was found to have 3.8% of individuals with EL grain; 83% with L and 13.2% with M. One recurrent selection cycle increased the percentage of L grains, thus reducing the percentages of EL and M grains (Figure 1).

The chi-squared test indicated that differences were highly significant at 0.001% between these two populations for GL. This means that the two populations PCT-6 and PCT-8 responded differently to selection, with population PCT-6 achieving considerable progress. Such a result could be attributed to the different genetic constitution of the two populations, as indicated by Martínez et al. (1997).

White belly (WB)

In population PCT-6\0\0\0, the values for WB ranged between 0 and 3.8, with a mean of 0.8, and in PCT-6\CG\1\0F between 0 and 3.6 with a mean of 0.4. In population PCT-8\0\0\0, the values for WB varied between 0 and 4, with a mean of 1, and in PCT-8\CG\1\0F between 0.2 and 3, with a mean of 0.8. In all four cases, most values were less than the mean of their respective populations.

To analyse the populations, the materials were classified according to a calibrated scale for WB, whereby only materials scoring £ 1 were accepted. Against this scale, population PCT-6\0\0\0 was observed to have 63% of its materials scoring £ 1, whereas, in PCT-6\CG\1\0F, 85% scored £ 1. The chi-squared analysis indicated that differences for WB £ 1 versus WB > 1 were highly significant at 0.001%, thus confirming that an increase in the proportion of materials with WB £ 1 was achieved in C₁. This represents a gain in selection for this trait in population PCT-6 (Figure 2).

Population PCT-8\0\0\0 had 56% of materials with WB £ 1, whereas PCT-8\CG\1\0F had 66%. The chi squared analysis indicated that differences for WB £ 1 versus WB > 1 were significant at 0.005%, indicating that the proportion of individuals with WB £ 1 increased in C₁. Genetic gains for WB occurred in this population, but at a lesser degree than in PCT-6 (Figure 2).

Figure 2. Percentages of individuals scoring less than or equal to 1 (grey shading) or more than 1 (white) for white belly in the two selection cycles of rice populations PCT-6 and PCT-8.

Figure 3. Percentages of grains in different grades of alkaline dispersion in two selection cycles of rice populations PCT-6 and PCT-8.

When comparing the initial cycles (C₀) of populations PCT-6\0\0\0 and PCT-8\0\0\0, the percentage of individuals with WB £ 1 was found to be similar, with the first being 7% more. However, on comparing the two populations at the end of C₁ (PCT-6\CG\1\0F and PCT-8\CG\1\0F), the difference in improved gain was 19%, with WB £ 1 in PCT-6 improving by 22% versus 10% for PCT-8 (Figure 2). This suggests that the progress obtained depended largely on the population’s genetic constitution.

Gelatinization temperature (GELT)

We remind readers that GELT is estimated indirectly by measuring the degree of alkaline dispersion against a scale of 1 to 7. In population PCT-6\0\0\0, the degrees of dispersion ranged from 2 to 7, with most grains concentrating at 5 (30%) and 7 (47%). Thus, of the initial population, 54% of grains had a low GELT, 35% had intermediate GELT and 11% had high GELT (Figure 3A).

After a selection cycle, alkaline dispersion in PCT-6\CG\1\0F was found to be mainly distributed between grades 2 and 7, with most grains concentrating at 5 (17%) and 7 (56%). One selection cycle in population PCT-6 increased the percentage of grains in grade 7 and reduced the percentage in grade 5, which indicates a selection gain for GELT for this population (Figure 3B).

For population PCT-8\0\0\0 (Figure 3C), dispersion values were also distributed between 2 and 7, but with most concentrating at grades 5 (50%) and 4 (15%). This indicates that 65% of the population’s grains had intermediate GELT, and that the rest were distributed at the extreme grades 2 (10%) and 7 (13%). As the first selection cycle finished, grain distribution in population PCT-8\CG\1\0F changed drastically to concentrate at 2 (49%; high GELT) and 5 (18%; intermediate GELT) (Figure 3D).

Data indicate that populations PCT-6 and PCT-8 responded differently to selection for GELT. If we take into account that consumers prefer rice with intermediate or low GELT (grades 4 to 7), we now know that PCT-8 responded negatively as it passed from concentrating in grade 5 (intermediate GELT) in the original population to grade 2 (high) in C₁.

Environmental and genetic factors, and their interaction, are well known to influence the expression of GELT (Wrigley and Morris, 1997). Although populations PCT-6 and PCT-8 have distinct genetic constitutions (Martínez et al., 1997), they also underwent imposed hierarchical selection, where GELT was the third criterion of importance in choosing individuals. This may have restricted the selection of individuals, thus influencing the response observed in the populations.

Table 3. Criteria of stability proposed for selecting plants previously scoring as acceptable for gelatinization temperature (given on a scale of 1 to 7). Ten grains of each of 31 materials (Mat.) were evaluated. Values refer to number of grains. S=stable, U=unstable.

Environmental temperature plays an important role. Rice grains with inter mediate GELT are subject to change in their expression as a function of environmental temperature during flowering and grain filling (Beachell and Stansel, 1963; Martínez and Cuevas- Pérez, 1989). Furthermore, these last two authors suggest that grains with intermediate GELT are more susceptible to changes in temperature during this critical period. For this reason, the selection of fertile plants in the populations under study presented distinct patterns of segregation for GELT for the 10 grains evaluated in the progeny derived from these fertile plants (Table 3).

To select materials according to GELT, the ordinal scale of 1 to 7 (Table 2) was insufficient to clearly explain and classify the materials according to their performance. Together with these considerations, the criteria of acceptability and stability were included in a simulation analysis of all possible values of this variable to help select materials in each separate cycle. This should be regarded as a methodological support for work on grain quality in rice so to use constant and defined parameters in the analysis.

Figure 4. Percentages of individuals accepted (grey shading) and rejected (white) according to the criterion of acceptability for gelatinization temperature in two selection cycles of rice populations PCT-6 and PCT-8.

Acceptable materials are those with low GELT. Hence, we propose accepting a material when 7 or more of the 10 grains have values in the low (6-7) or intermediate (4-5) GELT categories.

Stable materials are those whose two least-frequent categories (smallest quantity of seed in these categories) each has a maximum value of 2. Otherwise, a material is considered unstable. An exception is Mat. 1 (Table 3), which has a distribution of ‘7’ in the low GELT group, ‘0’ in intermediate GELT and ‘3’ in high GELT. The criterion of stability is applied when the material has previously fulfilled the condition of acceptability. The fulfilment of the two conditions (acceptability and stability) leads to the selection of a genotype (Table 3). With this new strategy, the two populations and their recurrent selection cycles can be analysed.

In population PCT-6\0\0\0, 81.5% of the materials were within the acceptable criterion (Figure 4). Once the first selection cycle was carried out, population PCT-6\CG\1\0F had 84.4% of its materials as acceptable. The chi-squared analysis indicated that no significant differences existed between the selection cycles for acceptability.

In population PCT-8\0\0\0, 70.3% of materials were acceptable. On completion of the selection cycle, the new population PCT-8\CG\1\0F had 21.9% of its materials as acceptable. The chi-squared analysis indicated highly significant differences at 0.001% between cycles, showing a clear regression in selection. That is, we ended up with a population that had a greater number of individuals in the ‘rejected’ category (Figure 4).

This may have happened because the selection process allowed several materials that did not comply with the proposed norm to pass into the next cycle.

Figure 5. Percentages of stable (grey shading) and unstable (white) individuals according the criterion of stability for gelatinization temperature in two selection cycles of rice populations PCT-6 and PCT-8.

Another possible explanation of this result is that of Beachell and Stansel (1963), who found that, in crosses between materials with intermediate and low GELTs, they had obtained materials with high GELT. Their data also indicated that populations with genetically distinct bases respond differently to selection for GELT.

When the analysis was carried out for the criterion of stability (Figure 5), we found that, in population PCT-6\0\0\0, 66.7% of individuals fell in the category of ‘stable’, whereas, in the following cycle, PCT-6\CG\1\0F had 84.8% of its materials stable. The chi-squared analysis indicated highly significant differences between the cycles for this trait in population PCT-6. The application of criteria for stability showed gain in terms of a greater number of stable materials.

The application of the criterion of stability to population PCT-8\0\0\0 resulted in 64.4% of individuals being in the category of stable. After a new selection cycle in population PCT-8, 63.5% of individuals were classified as such (Figure 5). The chi-squared analysis did not find significant differences between the cycles of populations, which suggests that the criteria of stability and acceptability were valuable for explaining what happened before in this population.

When the results for the trait acceptability were compared for the two populations in C₁ (Figure 4), we saw highly significant differences (84.4% versus 21.9%) between populations PCT-6 and PCT-8. However, when the concept of stability was applied (Figure 5), highly significant differences were still observed between the two populations in C₁, but to a smaller degree (84.4% versus 63.5%).

When the results for populations PCT-6 and PCT-8 were compared in C₀, population PCT-6\0\0\0 showed significant differences with respect to population PCT-8\0\0\0 for the category ‘acceptable’ (81.5% versus 70.3%), and PCT-6 had a higher percentage of individuals (Figure 4). In terms of stability, no significant differences existed in C₀ (Figure 5).

Although the data are still preliminary and are subject to environmental effects in different semesters of selection, results indicate the importance of carrying out selection in the first cycles for traits of high heritability in an appropriate environment (Torres, 2001).

The evaluations of S₀ plants will permit selecting the best individuals for developing populations that are each time better. Such populations will be used as sources for extracting lines or for male sterility to develop other populations.

Final comments, future plans

For most traits, we achieved significant improvements in grain quality. In the original populations (C₀), genetic variability of the quality traits studied was sufficient to allow the new C₁ populations to present a larger percentage of individuals with desirable grain quality traits. However, populations PCT-6 and PCT-8 responded differently to the imposed selection, and population PCT-6 stood out for its response for WB. With the intensity of selection used, an average of 40% of materials were selected for recombination in each cycle.

Results are very encouraging. However, further work is needed to con- firm preliminary findings, especially in terms of quantifying genetic gains among selection cycles.

We also need to look at additional populations with diverse genetic backgrounds to find out whether the variables studied in this experiment behave similarly, especially in terms of genotype-by-environment interactions. We had studied only the breeding behaviour of a few quality traits. We need to know how economic traits such as yield potential, tolerance of major insect pests and diseases, plant height, and flowering are affected when the main emphasis is devoted to grain quality.

References

Beachell, H.M. & Stansel, J.W. 1963. Selecting rice for specific cooking characteristics in a breeding program. Int. Rice Commun. Newsl., Special issue: 25-40.

Chang, T.T. & Somrith, B. 1979. Genetic studies on the grain quality of rice. In Proc. workshop on Chemical Aspects of Rice Grain Quality, pp. 49-58. Los Baños, Philippines, International Rice Research Institute (IRRI).

Châtel, M. & Guimarães, E.P. 1995. Selección recurrente con androesterilidad en arroz. Cali, Colombia, CIRAD-CA & CIAT. 70 pp.

Châtel, M.; Guimarães, E.P.; Ospina, Y. & Borrero, J. 1997. 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.

Choi, S.J. 1979. Studies on inheritance and selection efficiency of endosperm chalkiness of rice kernels. In Research report of the Office of Rural Development No. 21, pp. 209-234. Suwon, Korea. Rural Development Administration.

David, C.C. 1991. The world rice economy: challenges ahead. In G.S. Khush & H.T. Gary, eds. Rice biotechnology, pp. 1-18. Los Baños, Philippines, International Rice Research Institute (IRRI). (Biotechnology in Agriculture No. 6)

Guimarães, E.P. 2000. Mejoramiento poblacional del arroz en América Latina: Dónde estamos y para dónde vamos. In E.P. Guimarães, ed. Avances en el mejoramiento poblacional en arroz, pp. 299-311. Santo Antônio de Goiás, Brazil, Embrapa Arroz e Feijão.

Guo, E.-N.; Pan, Z.; Wang, S.L. & Lu, S.H. 1982. Studies on inheritance of white belly kernels of keng rice (Oryza sativa subsp. Keng). Jiangsu Acad. Agric. Sci., 6: 611.

Hossain, M. 1996. Recent developments in the Asian rice economy: challenges for rice research. In R.E. Evenson; R.W. Herdt & M. Hossain, eds. Rice research in Asia, pp. 17-34. Los Baños, Philippines, International Rice Research Institute (IRRI).

Hsieh, S.C. & Wang, L.H. 1988. Genetic studies on grain quality in rice. In Rice Grain Quality, proc. symposium, pp. 117-136. Taiwan, Rep. China, Taichung District Agricultural Improvement Station.

IRRI (International Rice Research Institute). 1985. In Annual report for 1984, Los Baños, Philippines.

IRRI (International Rice Research Institute). 1993. Rice almanac. Los Baños, Philippines. 142 pp.

Jennings, P.R.; Coffman, W.R. & Kauffman, H.E. 1981. Mejoramiento de arroz, pp. 127-152. Cali, Colombia, CIAT.

Khush, C.S.; Paule, C.M. & De la Cruz, N.M. 1979. Rice grain quality evaluation and improvement at IRRI. In Proc. workshop on Chemical Aspects of Rice Grain Quality, pp. 21-31. Los Baños, Philippines, International Rice Research Institute (IRRI).

Martínez, C.P. & Cuevas-Pérez, F.E. 1989. Evaluación de la calidad culinaria y molinera del arroz: Guía de estudio. Cali, Colombia, CIAT. 73 pp.

Martínez, C.P.; Lentini, Z.; Châtel, M.; González, D. & Mojica, D. 1997. Uso de selección recurrente en combinación con cultivo de anteras en el programa de arroz de riego del CIAT. In E.P. Guimarães, ed. Selección recurrente en arroz, pp. 139-149. Cali, Colombia, CIAT.

Masajo, T.M. 1971. Breeding behavior and relationship of white belly and gelatinization temperature to amylose content of the rice endosperm. Los Baños, Laguna, University of Philippines. 127 pp. (MSc thesis)

Pingali, P.L., Hossain, M. & Gerpacio, R.V. 1997. Asian rice bowls: the returning crisis? Los Baños, Philippines, International Rice Research Institute (IRRI). 341 pp.

Sanint, L.R. & Wood, S. 1998. Impact of rice research in Latin America and the Caribbean during the past three decades. In P. Pingali & M. Hossain, eds. Impact of rice research, proc. international conference, 3-5 June 1996, Bangkok, Thailand, pp. 405-428. Bangkok, Thailand Development Research Institute & International Rice Research Institute IRRI).

Torres, T.E. 2001. Análisis de estabilidad y criterios de selección para centro blanco en arroz (Oryza sativa L.). Palmira, Universidad Nacional de Colombia. 78 pp. MSc thesis)

Webb, B.D. 1991. Rice quality and grade. In B.S. Luh, ed. Rice utilization, pp. 89-119. New York, Van Nostrand Reinhold.

Wrigley, C.W. & Morris, C.F. 1997. Breeding cereals for quality improvement. In R.J. Henry & P.S. Kettlewell, eds. Cereal grain quality, pp. 321-350. London, Chapman Hall.