To make efficient use of large germplasm collections, it is advisable to as
semble a representative core collection and to evaluate the relationships a
mong the traits studied. However, the assemblage of a core collection from
very large germplasm collections is problematic The computing resources nee
ded to carry out genetic distance calculations and comparisons with commonl
y available programs is prohibitively large. The objects of this study were
(i) to develop a method which assembles a core collection by maximizing th
e diversity (measured as mean Euclidean distance) from within groups of acc
essions defined by species, subspecies, and geographic origin and (ii) to t
est the effectiveness of the method on a collection of 20 997 annual Medica
go accessions from the Australian Medicago Resource Center in Adelaide, Sou
th Australia, that bad been evaluated for 27 agronomic characteristics. The
method resulted in a core collection of 1705 accessions that represented 7
4% of the extremes of the 27 characters, indicating that the entire range o
f the characters was represented in most cases. Accessions representing the
extremes easily could be added to the core collection. The method used req
uires relatively minor computing resources and should be useful to curators
of large germplasm collections. To assess the relationships among the 27 m
easured traits, correlation coefficients of all possible combinations of tr
aits were calculated. The most strongly associated traits were, as expected
, such traits as grams of seed per plant and grams of pods per plant and in
dicated that some traits could be omitted from future evaluations with litt
le loss of information, thereby increasing the efficiency with which germpl
asm evaluations can be carried out.