Synthesizing and improving populations without using male sterility

The method of manual hybridization for rice evolved mostly from discoveries by Taillebois and Castro (1986) who showed that, to produce hybrid seeds, the entire plant need not be used but only the principal stem or tiller with the panicle obtained from the plant of origin. This makes managing male and female parents in the field possible. At hybridization, the best tillers are chosen and their leaves removed. They are then taken to the hybridization site where they must be placed in containers with water. They are then emasculated and pollinated. The hybrid seeds should be left to develop in a protected site, which could be a greenhouse or mesh house. The simplicity of the method reduces labour, increases the rate of success, and makes extensive crossing feasible in improvement programmes (Castro et al., 1999). The use of this new technique has made manual recombination of recurrent selection populations possible.

In 2002, Embrapa Arroz e Feijão incorporated into its population improvement programme population CNA-12, which was developed without using genetic male sterility. The goal was to seek its improvement to obtain cultivars with stable resistance to blast for the tropical and subtropical regions of Brazil. Its case is described below.

Synthesizing population CNA-12

Obtaining cycle 0

Population CNA-12 was formed using:

(1) the sources of resistance to blast Huan-Sen-Goo, 5287, CNAi 9020, CNAi 9029, Oryzica Llanos 4 and Oryzica 1

(2) cultivars or elite lines Diamante, Javaé, BRS Formoso, Jequitibá, Marajó, BRS Taim, IRGA 417, BRS Chuí, CNA-8502 and CNA-8621.

Crosses were carried out among the sources of resistance (male parents) and cultivars or elite lines (female parents), so that every source of resistance participated in three combinations for a total of 18 crosses. A backcross was carried out in the direction of the cultivars or elite lines to reduce the participation of the sources of resistance in the populations. Tables 13 and 14 show, respectively, the crosses and backcrosses that formed the 18 populations (P1 to P18).

Obtaining cycles 1 to 5

To recombine or generate the cycle-1 population, Bearzoti’s proposal [1996, cited by Ramalho (1997)] was used, slightly modified due to the orientation of the cycle 0 crosses where a given source of resistance was combined with a determined cultivar or elite line. In this procedure, each population was always crossed with two more, so that a circle of crosses is formed, where cycle 5 is again the same as cycle 1. This procedure of recombination has some advantages:

• combination is directed; thus, the probability of losing alleles from the original parental materials is less

• A smaller number of crosses are made with regard to the complete or partial diallel

• Facility in identifying crosses as, in each case, the same population is crossed with only two others.

Cycle 1 was formed as follows:

The procedures for forming cycles 2 and 3 are presented in Tables 15 and 16. To form cycles 4 and 5, the same principle was followed as presented in the tables so that, in the end, in cycle 5, the formation follows that of cycle 1, as described in the list above.

Improving populations

From each of the 18 populations (Table 14), and in each cycle, F₂ seeds are obtained. Within each population, about 2000 plants of the F₂ generation are inoculated under controlled conditions with the blast races most commonly found in the region targeted by the programme. Twenty resistant plants per population are selected, totalling 360 plants, which are then transplanted to pots to obtain F_2:3 seeds.

Between harvests, the seeds are multiplied to compose the evaluation trials of the F_2:4 families, which will be carried out at harvest in several sites within the region. The experimental design is that of Federer’s augmented blocks (Federer, 1956), in which 360 families will be evaluated, together with four checks. These experiments will primarily evaluate for resistance to blast under the conditions of the ‘Ou chamber’, for grain quality and yield, and other traits of economic interest.

Based on the joint analyses of data from the trials, from each population five families will be selected, to make a total of 90 families, which will be crossed to form the next cycle. With these procedures, each selection cycle will take 3 years to complete.

Table 13. Crossing strategy to synthesize the rice population CNA-12 for resistance to blast

Table 14. Rice populations formed by backcrossing the crosses listed in Table 13.

The adopted procedure for improvement means that, at any time, a new source of resistance to blast and/or an elite line can be introduced into the crossing cycles. This makes the system highly dynamic.

Line development

Simultaneously with improving the population, line development is initiated. In evaluations of F_2:4 families, the 36 (10%) best families will be selected according to data on the means of the trials carried out in different sites. The F_2:5 families will be planted between harvests to select individual plants. The F_5:6 lines obtained will participate in the National Network of Yield Trials. With this, the programme will have new lines for evaluation in only 3 years.

Table 15. Combinations involving chain crosses that form the populations for the second recombination cycle (cycle 2).

Table 16. Combinations involving chain crosses that form the populations for the third recombination cycle (cycle 3).