M. Malosetti et T. Abadie, Sampling strategy to develop a core collection of Uruguayan maize landraces based on morphological traits, GEN RESOUR, 48(4), 2001, pp. 381-390
Core collections were suggested to improve germplasm utilization. A core co
llection is a subset chosen to represent the diversity of a collection with
a minimum of redundancies. Because diversity is distributed between and wi
thin groups with different degrees of organization, an adequate classificat
ion of accessions into related groups should be performed prior to the sele
ction of a core collection. Different classification strategies for the Uru
guayan Maize Collection were compared, and the best one was used to select
a core collection. The following classification strategies were compared fo
llowing a multivariate approach using the available maize data base: i) rac
ial classification, ii) geographic origin (south and north of the country),
and iii) a combination of kernel type and geographic origin. The third opt
ion was considered the best classification rule, since it takes into accoun
t two points which are closely related to the distribution of diversity: ge
notypic composition and geographic origin. The following five groups were i
dentified in the collection: a) pop, b) floury, c) dent, d) southern flint-
semiflints, and e) northern flint-semiflints. Eight core collections, each
of 90 accessions, were selected, using different strategies to weight the g
roups in the core and to select the accessions from the groups. The P, C, a
nd L strategies were used and combined with either random selection within
the group or the Relative Diversity method. Two samples of 90 accessions we
re obtained at random without considering the classification. The Relative
Diversity method combined with the L strategy produced the best core collec
tion, as it retained the highest percentage of the ranges for the 17 variab
les included in the analysis. On average, 91% of the ranges were retained i
n the core, confirming its representativeness.