Field-based screens for genetic variation in reproductive-stage drought tolerance are often confounded by genetic variation in root depth, flowering date and biomass at flowering. To reduce these confounding effects and to impose drought stress more precisely, we grew contrasting genotypes of rice (Oryza sativa L.) in shallow containers of flooded soil. When water was withheld from the shallow-rooted indica genotype IR64 for 6 days starting at −17, −7, 0, 10 and 20 days after heading (DAH), the impact on grain yield was greatest at −7 and 0DAH. The most sensitive yield component was filled grain percentage (spikelet fertility). Data on yield and spikelet fertility were also obtained for another shallow-rooted lowland indica (BRRI Dhan 31) and two deep-rooted upland tropical japonicas (Azucena and Moroberekan). Compared with well-watered controls, withholding of water for 6 days reduced spikelet fertility by 80% in IR64, BRRI Dhan 31 and Azucena but by 22% in Moroberekan. In a detailed comparison, water was withheld from IR64 for 5 days and from Moroberekan for 6 days, starting 3 days before the heading date of well-watered controls. Under these stress conditions, heading was delayed for 4 days in IR64 and 6 days in Moroberekan. Spikelet fertility in the top four rachis branches was reduced by 80% in IR64 and 16% in Moroberekan, a difference attributable principally to reduced anther dehiscence and lower stigma pollen density in IR64. Two properties of Moroberekan may contribute to high anther dehiscence after stress/re-watering: (i) constitutively superior development of fibrous structures in the endothecium at the anther apex and base and (ii) better maintenance of pollen size. All ovaries of Moroberekan received pollen—an average of 60 grains in well-watered plants and 42 grains in stressed/re-watered plants. In well-watered IR64, all ovaries received an average of 31 grains, whereas in stressed/re-watered IR64, 67% of ovaries received no pollen and the remainder received an average of eight grains. Moroberekan may, therefore, be a source of reproductive-stage drought tolerance through genes that maintain anther dehiscence during recovery from low water status.
Field Crops Research (2006) 97 (1) [doi:10.1016/j.fcr.2005.08.019]