View - American Journal of Botany

1298 AMERICAN JOURNAL OF BOTANY
[Vol. 86
Fig. 6. Single most parsimonious tree obtained from the combined
analysis of morphology, rbcL, and matK data. L 1565, CI 0.61,
RI 0.59. Decay values are below lines, and bootstrap values are above
lines. Taxa traditionally placed in the tribe Andromedeae are in boldface.
the Andromedeae, as traditionally circumscribed (see Stevens,
1969, 1971, 1995), cannot be maintained, as genera
of the Gaultheria group are more closely related to the
Vaccinieae than they are to those of the Lyonia group.
The strong support for the Lyonia group in the combined
analysis indicates that the group should be given formal
recognition, and it is here proposed as the tribe Lyonieae—Lyonieae
Kron & Judd, tribus nov. Epidermis foliorum
lignea; filamenta floris plerumque S-formis; testa
cellulis plerumque elongatis. Type genus: Lyonia Nutt.
Other nomenclatural proposals relating to these genera
will be presented in connection with a new classification
of the Ericaceae (Kron et al., unpublished data).
The monophyly of the Gaultheria group is not supported
in most of the analyses and only weakly supported
in the molecular morphology analysis. In the combined
rbcL and matK analysis the Gaultheria group is split into
two clades that are not resolved with respect to Vaccinieae.
Inclusion of the morphological data results in only
a weakly supported Gaultheria group. This group needs
further study, especially a more comprehensive sampling
of Gaultheria, Pernettya, and Diplycosia. In addition,
Leucothoë appears as paraphyletic or polyphyletic in
most of the trees. Exceptions are in some of the morphological
trees and in the rbcL analysis. In both cases the
relationship of Leucothoë racemosa to L. fontanesiana is
weakly supported.
Andromeda, Chamaedaphne, and Oxydendrum often
have been considered taxonomically isolated (Palser,
1951, 1952; Stevens, 1969, 1971), but our results support
an isolated position for only Oxydendrum (Fig. 6).
Andromeda has several distinctive features, which
probably represent autapomorphies, e.g., seeds with a
strongly multilayered testa, campylotropous ovules (Palser,
1952), the lack of calyx and corolla stomata, and lack
of multicellular hairs. The linking of this genus with Zenobia
was not anticipated and is not reinforced by any
morphological characters included in our analyses. The
base chromosome number of Andromeda, i.e., 12, also is
aberrant given its placement with genera of the Gaultheria
group, which are mainly x 11. However, it may be
significant that Zenobia nearly lacks multicellular hairs,
i.e., they are only associated with the leaf serrations, and
if these were lost (as presumably occurred in the ancestors
of Andromeda) then this genus, like Andromeda,
would lack such hairs.
Chamaedaphne is distinctive in its floral anatomy, embryology,
campylotropous ovules (Palser, 1951, 1952),
anthers with tubules, and distinctive lepidote indumentum
(composed of peltate scales that are different in form
from those of Lyonia sect. Lyonia; see Judd, 1979). Both
molecular and morphological analyses place Chamaedaphne
with members of the Gaultheria group. It is
placed sister to Leucothoë racemosa in our combined
analysis. Both have stamens with smooth filaments (no.
36-2; see Fig. 4, where feature is homoplasious). Stevens
(1969) noted that ‘‘in stomata, bracteole, seed and general
anatomical characters it shows some similarity to the
Gaultheria group.’’ For example, its stomata are positionally
paracytic and testa cells isodiametric.
Oxydendrum also is phenetically distinctive, mainly
because of its floral anatomy (all traces to the floral organs
leave an elongated axis separately; Palser, 1952),
deeply hooked leaf midrib vascular bundle (Stevens,
1969, 1971), anthers with terminal tubules, and terminal
inflorescence that fruits in the same year in which the
shoot on which it is borne was initiated (Lems, 1962).
Cox (1948) considered Oxydendrum in its own tribe because
its wood has a distinctive medullary ray type and
a high percentage of porous vessel perforation plates. Our
analyses suggest that these features represent a combination
of autapomorphies and retained plesiomorphies.
Our results are in general agreement with those of Stevens
(1969, 1971), Judd (1979), and Middleton (1991) in
supporting the recognition of two major clades within the
genera traditionally placed within Andromedeae. Clearly,
they also support the hypothesis (see Stevens, 1995) that
the Andromedeae are not monophyletic, since Vaccinieae
are closely related to members of the Gaultheria group.
The close relationship of Vaccinieae to some genera traditionally
placed within Andromedeae was first noted by
Stevens (1969, 1971), and their phenetic similarity was
evident to Watson, Williams, and Lance (1967).
Higher level phylogenetic analyses of Ericaceae based
on both nuclear and chloroplast DNA (see Kron and
Chase, 1993; Kron, 1996, 1997; Kron et al., 1998) support
the existence of a clade comprising Oxydendrum, the
Gaultheria group, the Lyonia group, and the Vaccinieae,
i.e., the ingroup taxa of the analyses herein presented.
This monophyletic group, which in large part corresponds
to the Vaccinioideae of Stevens (1969), probably should
be given subfamilial rank within an expanded Ericaceae