Effectiveness of early-generation testing applied to upland rice breeding

A plant breeding program is a long-term investment. Therefore, periodic assessment of the effectiveness of a breeding strategy is essential to maximize genetic gains per unit of time and resource invested. In this work, we assessed the effectiveness of the early-generation testing (EGT) approach used in the upland rice (Oryza sativa L.) breeding program at Embrapa (Brazilian Agricultural Research Corporation), Brazil, estimating the genetic progress achieved for three traits in two distinct phases, spanning 15 years. In the first phase (from 2003 to 2010), it was used the bulk method within F₃ progenies with prior testing of F₂ crosses, while in the second phase (from 2010 to 2017), it was used the bulk method within F₂ progenies. The dataset comprised 70 yield trials, involving 1884 F_3:5 progenies (phase I) and 925 F_2:4 progenies (phase II) from an elite population, and 10 check cultivars, evaluated for grain yield (GY), plant height (PH) and days to flowering (DTF). For estimating the genetic gain, we adapted a generalized linear regression method to compute bi-segmented linear regression coefficients. Desirable genetic gains were achieved only for GY in both phases of the breeding program, with 78.75 kg ha^?1 year^?1 (2.68%) in the first phase, and 53.78 kg ha^?1 year^?1 (1.54%) in the second phase. These results show the effectiveness of EGT, especially via bulk method within F₃ progenies with prior testing of F₂ crosses, applied to upland rice breeding. Some refinements are discussed in the method to make it more cost-effective and more efficient in achieving genetic gains.