In article ,
[Iris Cohen] wrote...

It seems that the DNA data places Celtis and allies, hence Celtidacee, as
distinct from Ulmaceae,

Who are these allies?

Well,

http://www.mobot.org/MOBOT/Research/APweb/

gives:

"List of Genera in CANNABACEAE

Cannabis L.
Celtis L.
Gironniera Gaudich.
Helminthospermum Thwaites = Gironniera Gaudich.
Humulopsis Grudz. = Humulus L.
Humulus L.
Mirandaceltis Sharp = Aphananthe Planch.
Nematostigma Planch. = Gironniera Gaudich.
Parasponia Miq.
Pteroceltis Maxim.
Sparrea Hunz. & Dottori = Celtis L.
Sponia Comm. ex Decne. = Trema Lour.
Trema Lour."

A quick lit search find the following [given in order of increasing
age]. Celtidaceae was apparently increasingly accepted as a separate
family from Ulmaceae by various workers after this was argued for by
researchers in the 1960s and 1970s on mainly mophological/anatomical
grounds [see below]; the more recent DNA evidence calls for including
Cannabaceae within the "Celtidaceae" clade, and "Cannabaceae" will be
the older name for the combined family.

TITLE: Urticalean rosids: Circumscription, rosid ancestry, and
phylogenetics based on rbcL, trnL-F, and ndhF sequences
AUTHOR, EDITOR, INVENTOR: Sytsma,-Kenneth-J [Reprint-author]; Morawetz,
-Jeffery [Author]; Pires,-J-Chris [Author]; Nepokroeff,-Molly [Author];
Conti,-Elena [Author]; Zjhra,-Michelle [Author]; Hall,-Jocelyn-C
[Author]; Chase,-Mark-W [Author]
SOURCE: American-Journal-of-Botany. 2002; 89(9): 1531-1546
ABSTRACT: To address the composition of the urticalean rosids, the
relationships of the component families (maximally Cannabaceae,
Cecropiaceae, Celtidaceae, Moraceae, Ulmaceae, and Urticaceae) and
analyze evolution of morphological characters, we analyzed sequence
variation for a large sampling of these families and various rosid
outgroups using rbcL, trnL-F, and ndhF plastid regions. Urticalean
rosids are derived out of a lineage including Barbeyaceae,Dirachmaceae,
Elaeagnaceae, and Rhamnaceae, with Rosaceae less closely related; thus,
they are imbedded within Rosales. Ulmaceae are the sister to all
remaining families. Cannabaceae are derived out of a subclade of
Celtidaceae; this expanded family should be called Cannabaceae.
Cecropiaceae are derived within Urticaceae and are polyphyletic with
Poikilospermum derived elsewhere within Urticaceae; this expanded
family should be called Urticaceae. Monophyletic Moraceae are sister
to this expanded Urticaceae. Support for these relationships comes
from a number of morphological characters (floral sexuality, presence
or absence of hypanthium, stamen type and dehiscence, pollen pore
number, ovule position, and embryo alignment) and chromosome numbers.
Most fruit types, in terms of ecological dispersal, are derived
independently multiple times and are strongly correlated with habitat.

TITLE: Further evidence for paraphyly of the Celtidaceae from the
chloroplast gene matK
AUTHOR, EDITOR, INVENTOR: Song,-B-H [Author]; Wang,-X-Q [Reprint-
author]; Li,-F-Z [Author]; Hong,-D-Y [Author]
SOURCE: Plant-Systematics-and-Evolution. 2001; 228(1-2): 107-115
ABSTRACT: Based on the chloroplast matK gene sequence, a phylogenetic
analysis of the Urticales in its traditional circumscription and its
putative affinities produced three equally most parsimonious trees
with tree length = 1527 steps, CI = 0.6863 and RI = 0.6352, indicating
that the Ulmaceae s. 1. are polyphyletic while the Celtidaceae are
paraphyletic, and particularly, Cannabis and Humulus in the Cannabaceae
are consistently nested within the Celtidaceae. Therefore, the present
data strongly suggest that the Cannabaceae should be merged with the
Celtidaceae to form a monophyletic group. According to the present
study, the Celtidaceae including Cannabaceae are more closely related
to the Moraceae and Urticaceae than to the Ulmaceae s. str.. Gironniera
and Alphananthe are both basal clades of the Celtidaceae rather than
members of Ulmaceae s. str..The Rhamnaceae and Rosaceae are the closest
relatives of the traditional Urticales, which is very congruent with
the newest system of flowering plants put forward by APG.

TITLE: The Ulmaceae, one family or two? Evidence from chloroplast DNA
restriction site mapping
AUTHOR, EDITOR, INVENTOR: Wiegrefe,-Susan-J [Reprint-author]; Sytsma,
-Kenneth-J [Author]; Guries,-Raymond-P [Author]
SOURCE: Plant-Systematics-and-Evolution. 1998; 210(3-4): 249-270
ABSTRACT: The Ulmaceae is usually split into two subgroups, referred
to as either tribes or more commonly subfamilies (Ulmoideae and
Celtidoideae). The two groups are separated, with some exceptions,
on the basis of leaf venation, fruit type, seed morphology, wood
anatomy, palynology, chemistry, and chromosome number. Propositions
to separate the two groups as distinct families have never gained
general acceptance. Recent morphological and anatomical data have
suggested, however, that not only is family status warranted but that
Celtidaceae are more closely related to Moraceae and other Urticales
than to Ulmaceae. In order to test these alternative sets of
relationships, restriction site mapping of the entire cpDNA was done
with nine rare cutting enzymes using 11 genera of Ulmaceae s. l.,
three other families of the Urticales, and an outgroup family from
the Hamamelidae. Cladistic analysis of the data indicates that Ulmaceae
s. l. is not monophyletic and that distinct families (Ulmaceae and
Celtidaceae) are warranted; that Ulmaceae is the sister group to
Celtidaceae plus all other families in the order; and that Cannabaceae
might be nested within Celtidaceae. Familial placements of various
problematic genera (e.g. Ampelocera, Aphananthe) are resolved and
character evolution of key morphological, anatomical, chemical, and
chromosomal features are discussed.

TITLE: A molecular phylogeny of Celtidaceae and Ulmaceae (Urticales)
based on rbcL nucleotide sequences
AUTHOR, EDITOR, INVENTOR: Ueda,-Kunihiko [Reprint-author]; Kosuge,-
Keiko [Author]; Tobe,-Hiroshi [Author]
SOURCE: Journal-of-Plant-Research. 1997; 110(1098): 171-178
ABSTRACT: On the basis of 1,290 bp sequences of the chloroplast gene
rbcL, a molecular phylogeny of seven of nine genera of the Celtidaceae
and four of six genera of the Ulmaceae was produced. These data were
analyzed together with some other urticalean genera using three methods
(i.e., maximum parsimony, maximum likelihood, and neighbor joining
methods). Maximum likelihood topology among 18 trees obtained
indicated that the Urticales are monophyletic with its common clade
splitting basally into two: one leading to a line comprising Ampelocera
(traditionally placed in Celtidaceae) and Ulmaceae, and the other
leading to a line comprising the remaining genera of Celtidaceae,
Moraceae, and other Urticales. Ulmaceae, to which Ampelocera is a
sister group, are monophyletic, as supported by many lines of
morphological evidence. In contrast to Ulmaceae, the monophyly of
Celtidaceae (excluding Ampelocera) was not supported, and resolution of
relationships of Celtidaceae with other Urticales, as well as of those
within the family, is left for future study.

TITLE: Phylogenetic analysis of Ulmaceae
AUTHOR, EDITOR, INVENTOR: Zavada,-Michael-S [Reprint-author]; Kim,-
Muyeol [Author]
SOURCE: Plant-Systematics-and-Evolution. 1996; 200(1-2): 13-20
ABSTRACT: A phylogenetic analysis of the Ulmaceae, Cannabaceae,
Barbeyaceae, and Broussonetia of the Moraceae produced nine equally
parsimonious trees with 127 steps. The Ulmoideae (Ulmaceae, sensu
GRUDZINSKAYA) are a monophyletic group and distinct from the
Celtidoideae. The genus Ampelocera occupies an isolated taxonomic
position among the celtidoids. The similarity of Ampelocera to the
fossil celtidoid flower Eoceltis of North America suggests that
Ampelocera possesses an archaic suite of characters, and occupies a
primitive position among the celtidoids, the Cannabaceae and the
Moraceae. The relationships among the other celtidoid taxa,
Cannabaceae, and Broussonetia are problematic. The Cannabaceae and
Broussonetia of the Moraceae are nested within the celtidoids
suggesting that this is a paraphyletic group. The close, but unresolved,
relationship of the celtidoids to the Moraceae and Cannabaceae observed
in this analysis, and the appearance of the celtidoids in the fossil
record prior to the ulmoids suggests that the evolutionary relationship
of the ulmoids and celtidoids may be more distant than current
taxonomic treatments reflect.

TITLE: Gynoecial vascular anatomy and its systematic implications in
Celtidaceae and Ulmaceae (Urticales)
AUTHOR, EDITOR, INVENTOR: Omori,-Yuji [Reprint-author]; Terabayashi,-
Susumu [Author]
SOURCE: Journal-of-Plant-Research. 1993; 106(1083): 249-258
ABSTRACT: Vasculature in the bicarpellate, pseudomonomerous gynoecium
with two distinct styles is examined and compared in all of 15 genera
of Celtidaceae and Ulmaceae (Urticales). Gynoecial vasculature is
diversified in the families but consistent in a genus or a group of
genera. Our observations corroborate the earlier suggestion that
Ulmaceae (six genera) basically have three-bundled styles, while
Celtidaceae (nine genera) always have one-bundled styles. Comparisons
with other Urticales and with Eucommiaceae as an outgroup of Urticales
indicate that Celtidaceae are more closely related to Moraceae in
sharing one-bundled style (a synapomorphy), rather than to Ulmaceae.
This supports a separation of Celtidaceae as a distinct family from
Ulmaceae sensu lato. Based on the degree of fusion of major vascular
bundles running in the gynoecium, we further distinguish three types
of gynoecial vasculature in Ulmaceae and, likewise, three types in
Celtidaceae, and discuss evolutionary trends of gynoecial vasculature
as well as some generic relationships within the families.

TITLE: VERNATION PATTERNS IN CELTIDACEAE AND ULMACEAE URTICALES AND
THEIR EVOLUTIONARY AND SYSTEMATIC IMPLICATIONS
AUTHOR, EDITOR, INVENTOR: TERABAYASHI-S [Reprint-author]
SOURCE: Botanical-Magazine-Tokyo. 1991; 104(1073): 1-14
ABSTRACT: In two associated families, Celtidaceae and Ulmaceae,
vernation pattern (represented by spatial relationships between leaf
lamina and stipules, the presence or absence of stipular fusion, lamina
orientation, and lamina folding pattern) is consistent within a genus
but shows a significant diversity within a family. Six vernation types
are distinguished and tentatively named: 1) Celtis type (Aphanathe,
Celtis, Lozanella, Parasponia, Pteroceltis, Trema), 2) Chaetachme type
(Chaetachme), 3) Gironniera type (Gironniera), 4) Holoptelea type
(Ampelocera, Holoptelea, Phyllostylon), 5) Zelkova type (Hemiptelea,
Planera, Zelkova), and 6) Ulmus type (Ulmus). The former three types
(found in most of celtidaceous genera) possess free or fused stipules
inside of the lamina; in contrast, the latter three types (found in all
six ulmaceous genera and Ampelocera) are characterized by having the
free stipules outside of the lamina. Within Celtidaceae, Celtis type is
probably primitive in having free stipules and an ordinarily oriented
lamina: Chaetachme type (with fused, convolute stipules and obliquely
oriented laminas) and Gironniera type (with laterally oriented laminas)
are the derived. Likewise, within Ulmaceae, both Zelkova and Ulmus types
(with laterally oriented laminas) are the derived, while Holoptelea type
(with ordinary oriented laminas) is primitive. Comparisons in vernation
pattern suggest the distinctness of Celtidaceae from Ulmaceae and the
isolated position of Ampelocera.

TITLE: KARYOMORPHOLOGY AND RELATIONSHIPS OF CELTIDACEAE AND ULMACEAE
URTICALES
AUTHOR, EDITOR, INVENTOR: OGINUMA-K [Reprint-author]; RAVEN-P-H [Author];
TOBE-H [Author]
SOURCE: Botanical-Magazine-Tokyo. 1990; 103(1070): 113-132
ABSTRACT: Based on karyomorphological features, six (examined in this
study) of nine genera of Celtidaceae are divided into three groups: 1)
Celtis, parasponia, Pteroceltis and Trema; 2) Aphananthe; 3) Giromniera,
and six genera of Ulmaceae into two: 1) Holoptelea and Phyllostylon;
2) Hemiptelea, Planera, Ulmus and Zelkova. The first four genera share
the simple chromocenter types at the resting stage and .vkappa. = 10,
with all chromosomes with submedian or median centromeres (frequency
of chromosomes with subterminal or terminal centromeres 0%, although
uncertain in Trema). Aphananthe has .vkappa. = 13, but resembles the
above four genera in other features. Gironniera is distinct from all
other Celtidaceae in having the diffuse-complex chromocenter type and
..vkappa. = 14, features which occur in Ulmaceae. In Gironniera the
frequency of chromosomes with subterminal or terminal centromeres is
43%, a proportion similar to those in Holoptelea (36%) and Phyllostylon
(58%) of Ulmaceae. All six genera of Ulmaceae have .vkappa. = 14, yet
Hemiptelea, Planera, Ulmus and Zelkova are distinct from Holoptelea
and Phyllostylon (with the simple chromocenter type) in having the
diffuse-complex chromocenter type and in predominantly possessing
chromosomes with subterminal or terminal centromeres (93%). Evidence
from karyomorphology, as well as from other sources, suggests 1) that
Aphananthe (.vkappa. = 13) is most distantly related to all genera with
..vkappa. = 10 within Celtidaceae, 2) that Gironniera may have a key
role for understanding evolutionary relationships between Celtidaceae
and Ulmaceae, and 3) that Holoptelea and Phyllostylon represent
derivatives of a line that diverged early from a common ancestor of
all Ulmaceae. On the basis of comparisons with other Urticales and the
putative outgroups of the order, it is also suggested that the
chromosome morphology of Ulmaceae represents the more derived state
in Urticales.

TITLE: CHARACTERISTICS OF THE STRUCTURE AND DEVELOPMENT OF THE PERICARP
OF REPRESENTATIVES OF THE FAMILIES ULMACEAE AND CELTIDACEAE
AUTHOR, EDITOR, INVENTOR: CHERNIK-V-V [Reprint-author]
SOURCE: Botanicheskii-Zhurnal-(St.-Petersburg). 1980; 65(4): 521-531
LANGUAGE: RUSSIAN
ABSTRACT: Twenty species were studied: Ulmus glabra Huds., U. laevis
Pall., U. campestris L., Holoptelea integrifolia (Roxb.) Planch.,
Hemiptelea davidii (Hance) Planch. Zelkova carpinifolia (Pall.) C. Koch,
Phyllostylon brasiliense Capanema, Celtis glabrata Stev., C. caucasica
Willd., Pteroceltis tatarinowii Maxim., Trema lamarckiana (Roem. et
Schult.) Blume, T. micrantha (L.) Blume T. orientalis (L). Blume,
Parasponia andersonii (Planch.) Planch., Lozanella enanthiophylla
(Donn.-Sm.) Killip, Morton., Aphananthe aspera Planch., A. Curranii
(Merr.) Grudz., A. philippinensis Planch., Gironniera nervosa Planch.
and G. subaequalis Planch. Study of the fruit at various stages of
development made it possible to identify 2 types of pericarp structure
corresponding to the 2 families. In Ulmaceae (Ulmus, Holoptelea,
Hemiptelea, Zelkova, Phyllostylon) the pericarp consisted of 4 layers
of cells (outer epidermis, middle layer, layer of mechanical tissue,
inner epidermis). The pericarp of a majority of genera of Celtidaceae
(Trema, Parasponia, Lozanella, Aphananthe, Gironniera) consisted of 3
layers of cells (outer epidermis, middle layer, layer of sclerenchyma
tissue). Only in Celtis and Pteroceltis did the pericarp have 4 layers
due to the presence of 2 layers of sclerenchyma tissue.

TITLE: SEED COAT MORPHOLOGY AND EVOLUTION IN CELTIDACEAE AND ULMACEAE
URTICALES
AUTHOR, EDITOR, INVENTOR: TAKASO-T [Reprint-author]; TOBE-H [Author]
SOURCE: Botanical-Magazine-Tokyo. 1990; 103(1069): 25-42
ABSTRACT: The seed coat surface morphology of Celtidaceae and Ulmaceae
(Urticales) indicates a significant evolutionary diversity. Celtis,
Chaetachme and Pteroceltis (Celtidaceae) have a unique sculpturing
with many crateriform holes; such holes occasionally sparsely occur in
seeds of Aphananthe, Gironniera (Celtidaceae) and Planera (Ulmaceae),
but not in those of the nine remaining genera of the two families. The
perforated seed coat further occurs in at least some genera of all other
urticalean families. A pattern of its occurrence in families and genera
suggest that the perforation represents a common archaic feature of all
Urticales, rather than a feature derived many times independently within
the order. The seed coat of Celtidaceae and Ulmaceae seems to have
lately lost the holes probably by a neotenic evolution: one or more
times within Celtidaceae, and one time in an ancestral line leading to
all Ulmaceae. The derived reticulate seed coat surface sculpturing,
which is shared by Gironniera (Celtidaceae) and some Ulmaceae, is
probably the result of parallel evolution. On the basis of evidence from
seed coat morphology and other sources, close relationships of Lozanella,
Parasponia and Trema within Celtidaceae, as well as variously distinct
positions of Ampelocera, Aphananthe and Gironniera, are also discussed.

TITLE: GENERIC RELATIONSHIPS IN THE ULMACEAE BASED ON FLAVONOID
CHEMISTRY
AUTHOR, EDITOR, INVENTOR: GIANNASI-D-E [Reprint-author]
SOURCE: Taxon-. 1978; 27(4): 331-344
ABSTRACT: The family Ulmaceae is most often treated as a single family
with 2 subfamilies: the Ulmeae (Ulmoideae) and Celteae (Celtidoideae)
or, more recently, as two separate families: the Ulmaceae and the
Celtidaceae (sensu Grudzinskaya). A flavonoid survey of 80 spp. of
Ulmaceae shows that each of the 19 genera [Ampelocera, Aphananthe,
Celtis, Chaetachme, Chaetoptelea, Gironniera, Hemiptelea, Holoptelea,
Mirandaceltis, Lozanella, Parasponia, Planera, Plagioceltis,
Phyllostylon, Pteroceltis, Trema, Ulmus, Zelkova, Barbeya] is
characterized by the production of flavonols (Ulmoid) or glycoflavones
(Celtoid), but not both. Further, the arrangement of genera based on
this flavonoid dichtomy is remarkably compatible with the generic
assignments in Grudzinskaya's bifamilial concept of the Ulmaceae. The
only exceptions are Ampelocera, Aphananthe, and Gironniera (in part),
which are normally considered Celtoid but possess Ulmoid (flavonols)
chemistry. However, recent anatomical and morphological studies of
these 3 genera indicate that their relationship to the Celtoid line
may not be as direct as was supposed, a point also suggested by the
flavonoid chemistry.

TITLE: ARRANGEMENT AND REDUCTION OF PERIANTH AND ANDROECIUM PARTS IN
REPRESENTATIVES OF ULMACEAE AND CELTIDACEAE
AUTHOR, EDITOR, INVENTOR: CHERNIK-V-V [Author]
SOURCE: Botanicheskii-Zhurnal-(St.-Petersburg). 1975; 60(11): 1561-1573
ABSTRACT: The vascular bundles of appendicular organs in all the
species studied [Ulmus glabra, U. laevis, Zelkova carpinifolia,
Hemiptelea davidii, Celtis caucasica, C. glabrata, Trema lamarckiana,
T. orientalis, Ampelocera cubensis, Parasponia andersonii have cyclic
disposition in the place where they leave the flower's stele: this
shows the whorl arrangement of flower's parts on the receptacle. The
vascular bundles of abortive corolla have preserved in the flowers of
Ulmus; in H. davidii these bundles form a distinct circle and alternate
with the vascular bundles of calyx. The presence of reduced bundles of
abortive corolla is characteristic of Z. carpinifolia. In the order of
separation of stamens from the flower tube in Ulmus, spiral features of
secondary origin can be traced. In representatives of Celtidaceae
(Celtis, Trema, Parasponia, Ampelocera) the process of reduction of
vascular system of perianth, androecium and gynoecium has developed
much further than in Ulmaceae. Their flowers usually possess a fixed
number of parts. Cyclicity is characteristic of flowers from the genus
Ampelocera, standing somewhat separately among Celtidaceae. The data
on the flower structure and the anatomy of its vascular system suggest
a higher evolutionary development of Celidaceae representatives
compared with representatives of Ulmaceae, which have preserved some
primitive features.

TITLE: THE ULMACEAE-D AND REASONS FOR DISTINGUISHING THE CELTIDOIDEAE-D
AS A SEPARATE FAMILY CELTIDACEAE-D
AUTHOR, EDITOR, INVENTOR: GRUDZINSKAYA-I-A [Author]
SOURCE: Botanicheskii-Zhurnal-(St.-Petersburg). 1967; 52(12): 1723-1749
CONCEPT CODES: 00504- (General-biology-Taxonomy-nomenclature-and-
terminology)
11108- (Anatomy-and-Histology-Microscopic-and-ultramicroscopic-anatomy)
50526- (Botany-general-and-systematic-Dicotyledones)
51000- (Morphology-anatomy-and-embryology-of-plants)

cheers