Tamanho de coleção original, métodos de agrupamento e amostragem para obtenção de coleção nuclear de germoplasma

Abstract

The objective of this study was to verify the effect of original germplasm collection size on the effectiveness of different clustering methods and sampling, used to obtain core collections, and to compare these methods in the establishment of core collections. Simulations were made for seven sizes of original collections, and seven sampling strategies were used for the establishment of core collections using morpho-agronomic traits. Eight strategies for obtaining core collections were used, six of them by grouping techniques (Tocher and sequential Tocher), and two without grouping (random and Tocher with inverted criteria grouping). For grouping strategies, constant, logarithmic and proportional sampling were used. The explained variance, the value of the determination coefficient, the coincidence coefficient, and the coincidence standard deviation coefficient between the core and the original collections were determined. The constant and logarithmic sampling strategies generate core collections which are representative of the original collections. The sequential Tocher grouping is more efficient in representing the original collection by the core collection than the Tocher's. Among the evaluated strategies the sampling by Tocher's inversed grouping criterion is the most efficient to generate representative core collections of the original collections.