https://www.nature.com/articles/srep06042

Abstract
The homology of the ‘semilunate’ carpal, an important structure linking
non-avian and avian dinosaurs, has been controversial. Here we describe
the morphology of some theropod wrists, demonstrating that the
‘semilunate’ carpal is not formed by the same carpal elements in all
theropods possessing this feature and that the involvement of the
lateralmost distal carpal in forming the ‘semilunate’ carpal of birds is
an inheritance from their non-avian theropod ancestors. Optimization of
relevant morphological features indicates that these features evolved in
an incremental way and the ‘semilunate’ structure underwent a lateral
shift in position during theropod evolution, possibly as a result of
selection for foldable wings in birds and their close theropod
relatives. We propose that homeotic transformation was involved in the
evolution of the ‘semilunate’ carpal. In combination with developmental
data on avian wing digits, this suggests that homeosis played a
significant role in theropod hand evolution in general.

Introduction
Avian wings are highly modified, fully foldable tetrapod forelimbs that
typically function in flight. One part of the avian wing with an
important role in both wing-folding and flight-related movements is the
wrist, which is composed of two separate proximal carpals and two distal
carpals that become fused to the metacarpals in early ontogenetic
development1,2,3. Not only is the small number of carpal elements an
evolutionary inheritance from ancestral theropods, but the unique shapes
of these elements were gradually established in theropod evolution. From
a functional perspective, a very important morphological feature of the
avian wrist joint is a transversely trochlear articular facet (hereafter
trochlear facet) on the lateral portion of the proximal surface of the
carpometacarpus (Fig. 1c), which is inferred from ontogenetic data to be
formed by the two lateral distal carpals (see electronic supplementary
material)2,3. The trochlear facet plays a key role in folding the hand
and flapping wing such as keeping the wing in place and preventing the
manus from supinating during forward flight4.

Figure 1

Diagram showing the position and general morphology of the transversely
trochlear proximal articular facet of the carpometacarpus in selected
theropod hands with the phalanges omitted (upper: proximal view; lower:
dorsal view; medial side of hand to left).

(a) The basal coelurosaurian condition (based on Guanlong). (b) The
basal paravian condition (based on Sinovenator). (c) The neornithine
condition (based on Crossoptilon). Yellow indicates the ‘semilunate’
carpal; grey-yellow indicates the transverse groove; green indicates the
metacarpals.

Full size image
In adult non-avian theropods, a trochlear morphology occurs on a
separate distal carpal, called the ‘semilunate’ carpal (Fig. 1a,b). The
‘semilunate’ carpal was first identified by Ostrom5, who described this
structure in the dromaeosaurid Deinonychus and listed it as one of the
most significant features supporting the theropod hypothesis of avian
origins based on the presence of a nearly identical element in
Archaeopteryx6. The ‘semilunate’ carpal was subsequently identified in
various other non-avian theropods, though with considerable variations
in its shape, size and position (e.g., ref. 7). The presence of the
‘semilunate’ carpal in non-avian theropods indicates that some
morphological modifications that ultimately proved important for flight
evolved early in theropod evolution and that an avian-like mechanism for
folding the wrist joint evolved before the origin of birds.

It should be noted that in most published literature the ‘semilunate’
carpal is homologized only with the medial portion of the trochlear
facet of the carpometacarpus in living birds7,8, without including the
lateralmost portion which is formed by distal carpal 4 (Fig. 1c). Given
that the ‘semilunate’ carpal is defined by a proximally transversely
trochlear morphology, it is more appropriate to view the ‘semilunate’
carpal of extinct theropods as equivalent to the whole trochlear facet
of the carpometacarpus in living birds.

The ‘semilunate’ carpal is clearly homologous to one or more of the
small distal carpals in the wrists of primitive theropods, but the
details are controversial. In Deinonychus and several other
maniraptorans, the ‘semilunate’ carpal is an enlarged element covering
the proximal ends of the two medialmost metacarpals5,7. Furthermore, a
large distal carpal occupying the same position in the basal
neotheropods Syntarsus and Coelophysis has been identified as a compound
bone formed by fusion of distal carpals 1 and 2, leading Gauthier to
suggest that these distal carpals were also homologous to the
‘semilunate’ carpal of non-avian maniraptorans9. However, this
hypothesis is in conflict with ontogenetic data from both Mesozoic
birds10 and living birds2,3, which show that the distal carpal proximal
to the medialmost metacarpal is absent in birds and the lateralmost
carpal is involved in the formation of the transversely convex and
trochlear proximal articular surface. This conflict has been repeatedly
cited as evidence against the theropod hypothesis of avian origins
(e.g., ref. 11,12).

In the present study, we describe the detailed morphology of the
‘semilunate’ carpal in several non-avian theropods, discuss changes in
‘semilunate’ carpal morphology during theropod evolution, comment on the
conflicting primary homology hypotheses that have been proposed for this
element and propose a new scenario for the evolutionary history of the
‘semilunate’ carpal.