555 Use of Model Assisted Yield Frameworks for the Analysis of Cotton Cultivar Response to Drought M.J. Lacape1 and J. Wery2 1 CIRAD-CA, Montpellier, France 2 ENSA-INRA, Montpellier, France ABSTRACT Three experiments were conducted under field conditions in Senegal to study the effects of a post-floral drought on seed-cotton yield and yield components of five cotton cultivars. The fraction of transpirable soil water (FTSW) was used to characterize the soil water deficit as experienced by the plants in each elementary plot. Two model-assisted frameworks of analysis of the seed-cotton yield were used: (1) number of bolls and average boll weight, associated with a leaf and flower development model; (2) total aerial biomass produced and harvest index, associated with a light interception model. The effects of irrigation treatments on seed-cotton yield varied from year to year, as rainfall pattern in the pre-flowering period differed. The overall variation in yield between years and experiments was accounted for by the average FTSW during the reproductive period and by the duration of this period. Water deficit induced an earlier termination of plant development (cutout, when NAWF = 5) that reduced the effective flowering period and the number of fruiting sites produced. A reduction of the average boll weight was also observed under terminal drought conditions. Water deficit induced an overall reduction in leaf area index and radiation interception that largely accounted for the effects of water deficit on total aerial biomass production. Duration of the flowering period and harvest index were the most efficient variables to explain genotypic differences in yield under the various conditions of water supply. The two frameworks of analysis of the seed-cotton yield, and the associated phenology and growth models, allowed a common analysis of plant responses to drought as observed under the various years, irrigation conditions, and for the various genotypes. Introduction In most African countries cotton is grown as a rainfed crop and water deficit remains a major limitation to yield (Hearn, 1995). The adoption of better adapted cultivars is one of the strategies proposed for the improvement of crop production in dry areas. As yield performance under drought is the result of many physiological processes, breeders need alternative variables as drought-adaptive traits (Blum, 1988). Different types of analytical models have been proposed to «dissect» yield into a small number of independent physiological components. The final seed-cotton yield can be analyzed as the product of a number of bolls and of an average weight of seed-cotton per boll (ABW). Seed-cotton Yield per m² = Number of plants per m² x Number of bolls per plant x ABW (Eq.1) Number of bolls per plant = Number of flowers x (1 - rate of abscission) Number of flowers = Flowering rate x Duration of flowering period This yield components framework, (YCF) associates final measurements on the plant and the parameters of a simple development model based on the evolution of the number of reproductive phytomers with thermal time (Lacape, 1998). Yield can also be analyzed as the product of above- ground biomass and harvest index (Biomass Production framework, BPF). Daily biomass production can be interpreted as the result of water (Eq. 2a) or radiation (Eq. 2b) used by the canopy. Seed-cotton Yield per m² = ( (Daily Biomass Production per m²)) x HI (Eq.2) Daily Biomass Production per m² = T x WUE (Eq.2a) Daily Biomass Production per m² = PAR0 x RAE x RUE (Eq.2b) Where HI = harvest index, T = water transpired, WUE = water use efficiency, PAR0 = incident photosynthetically active radiation, RAE = radiation absorption efficiency, RUE = radiation use efficiency The objective of the study was to compare the efficiency of these frameworks for the analysis of seed- cotton yield variation of different genotypes under terminal drought in a set of field experiments. The second framework used equation 2b because RAE and RUE can be more easily related to morphological and M.J. Lacape and J. Wery Proceedings of the World Cotton Research Conference-2. Athens, Greece, September 6-12, 1998. pp.555-562.