We tested the power of quantitative morphology to distinguish Amelanch
ier taxa at a Maine site where most individuals come from polyploid ag
amospermous species and taxonomic boundaries are potentially obscured
by hybridization. We sampled 92 individuals from seven species and one
putative hybrid, 13 from a microspecies we call A. ''rubra,'' and 15
that fit no published description (unknowns). We performed principal c
oordinates analysis on 20 quantitative and qualitative characters, six
recognized taxa, and the unknowns. Three groups are suggested: 1) A.
intermedia, A. laevis, A. ''rubra,'' four A. stolonifera, and four unk
nowns; 2) A. nantucketensis and three A. stolonifera, and 3) A. canade
nsis and 11 unknowns. In the seat terplot of the first two principal c
oordinates axes, the only two species that overlap are A. intermedia a
nd A. laevis. Discriminant function analyses indicate that A. ''rubra'
' could be a hybrid of A. laevis and either A. nan tucketensis or A. s
tolonifera and do not support the hypothesis that A. intermedia is A.
canadensis X laevis. The unknowns group with A. canadensis, A. laevis,
or A. ''rubra'' and mostly do not appear to be F(1)s. Agamospermy and
/or insufficient time for hybrids to establish after disturbance may e
xplain the absence of an extensive hybrid swarm.