Con los artículos científicos se presenta la nota de tapa que, en esta edición, abre el debate sobre un INTA como plataforma de políticas públicas. Referentes de ciencia y tecnología de organismos públicos y privados se refieren a su vinculación con el organismo.
Trabajos en Prensa
Evaluation of methods for the statistical analysis of comparative trials of sunflower (Helianthus annuus L.) yield
The oil yield per unit area is the main criterion for selection of most sunflower breeding programs. Comparative performance testing by a network installed throughout the sunflower region of Argentina contributes to both the decision-making of farmers and their advisers, and to the selection of superior genotypes by plant breeders. The aim of this study is to quantify the differences in predictive ability of different experimental designs and methods of statistical analysis in the estimation of the genotypic effects oil yield (kg ha-1) and its impact on the selection of superior hybrids. Forty-three multi-environmental sunflower trials (261 local trials) were evaluated for five seasons. The relative efficiency of the most common breeding program design (Randomized Complete Blocks) was compared with two alternative models for design and analysis of trials. A ranking of hybrids was obtained for each model and, with a selection pressure of 20%, the degree of overlap was evaluated in the selection of peer models using a coefficient of coincidence. The spatial model was the best fit between actual and expected oil yields, and also showed an increase in the efficiency of selection of superior genotypes in relation to the block design through a higher coefficient of coincidence with the Ideal Model (that of highest relative efficiency). According to these results, the precision of estimates from official networks of comparative
crop yield tests could be strongly improved using available statistical packages without increasing the need for investment of additional resources in terms of the number of plots.
Keywords: experimental design, relative efficiency, official trial networks, sunflower, mixed linear models, spatial analysis.
1Dto. Métodos Cuantitativos y Sistemas de Información, Facultad de Agronomía, UBA.
Correo electrónico: firstname.lastname@example.org
2Dto. Métodos Cuantitativos y Sistemas de Información, Facultad de Agronomía, UBA. Ifeva, UBA-CONICET. Correo electrónico:
3DuPont Pioneer, 41309 La Rinconada, Sevilla, España. Correo electrónico: email@example.com