Unenlagiid theropods : are they members of the Dromaeosauridae ( Theropoda , Maniraptora ) ?

In the present paper we analyze the phylogenetic position of the derived Gondwanan theropod clade Unenlagiidae. Although this group has been frequently considered as deeply nested within Deinonychosauria and Dromaeosauridae, most of the features supporting this interpretation are conflictive, at least. Modification of integrative databases, such as that recently published by Hu et al. (2009), produces significant changes in the topological distribution of taxa within Deinonychosauria, depicting unenlagiids outside this clade. Our analysis retrieves, in contrast, a monophyletic Avialae formed by Unenlagiidae plus Aves.


INTRODUCTION
Until recently, the deinonychosaurian fossil record has been geographically restricted to the Northern Hemisphere (Norell and Makovicky 2004), but recent discoveries demonstrated that they were also present and highly diversified in the Southern landmasses, suggesting that an important adaptive radiation took place in Gondwana during the Cretaceous.
The term Unenlagiidae was coined by Bonaparte (1999) to include the South American Unenlagia and the Malagasy Rahonavis, as well as the poorly known Early Cretaceous Australian genus Timimus (Rich and Vickers-Rich 1993).Later, Makovicky et al. (2005) coined the subfamily term Unenlagiinae, to fit this clade within the family Dromaeosauridae.
In the current usage, the subfamily Unenlagiinae is applied to the clade formed by Unenlagia, Neuquenraptor, Buitreraptor In its original description, Unenlagia was interpreted as closer to birds than the remaining dromaeosaurids (Novas and Puerta 1997), a point of view that was followed by some authors (Forster et al. 1998, Xu et al. 1999, Rauhut 2003, Novas 2004).Nevertheless, most researchers concluded that Unenlagia and its kin were deeply nested within Dromaeosauridae and Deinonychosauria ( The goal of the present paper is to make a brief overview about deinonychosaurian and dromaeosaurid synapomorphic features that have been cited for unenlagiids with the aim to evaluate the higherlevel phylogenetic relationships of this family of theropods. Most previous phylogenetic analyses were based on the Theropod Working Group matrix (TwiG), which corresponds to the most integrative analysis of coelurosaurian theropods currently available (Norell et al. 2001, Hu et al. 2009; see also Xu et al. 2008).We reanalyze the most recent version of this data matrix with the aim to corroborate the phylogenetic position of Unenlagiidae.We offer an updated codification of derived features present in unenlagiids, but also review some characteristics that were originally interpreted as diagnostic of Deinonychosauria and Dromaeosauridae.After the recoding of several traits in the matrix by Hu et al. (2009), some changes in the tree topology become apparent, and they will be considered in the following pages (see Appendix IV).ABBREVIATIONS HMN -Museum für Naturkunde, Berlin, Germany.IVPP V -The Institute of Vertebrate Paleontology and Paleoanthropology, Beijin, China.MCF PVPH -Museo Carmen Funes, Plaza Huincul, Neuquén, Argentina.MML -Museo Municipal de Lamarque, Lamarque, Río Negro, Argentina.MPCA -Museo Provincial Carlos Ameghino, Cippolletti, Río Negro, Argentina.

MATERIALS AND METHODS
With the aim to discuss the phylogenetic relationships of unenlagiids with respect to the remaining paravians, we performed a phylogenetic analysis using the most recent version of the TwiG, presented by Hu et al. (2009).We have partially modified this dataset, consisting of 88 taxa scored for 366 characters (see Appendices I-II).Definitions of characters 1 through 363 follow Hu et al. (2009); characters 364 through 365 have been added from Novas et al. (2009); character 366 is from Gianechini et al. (2009).We have modified character 199 with the addition of a new character-state (2).In addition, the taxon list used by Phylogenetic analysis was performed using TNT 1.0 (Goloboff et al. 2003).All characters were equally weighted and treated as unordered.A heuristic search was performed with 10,000 random addition sequence replicates to find the most parsimonious trees for the data matrix.Tree bisection and reconnection (TBR) was utilized as the branch-swapping algorithm for the heuristic search.Heuristic searches were performed on 1,000 pseudoreplicate datasets, with 10 random addition sequence replicates for each search.The maximum number of trees saved for each random addition sequence replicate was set to 100.Ratchet searches were performed on 2,000 pseudoreplicate datasets.The phylogenetic analysis resulted in the recovery of 145 Most Parsimonious Trees (MPTs), which yield a Strict Consensus Tree of 1440 steps, with a consistency index of 0.31 and a retention index of 0.71 (Fig. 1B).

CHARACTER ANALYSIS
We discuss below all the traits that have been considered as synapomorphic of Deinonychosauria and/or Dromaeosauridae by previous authors (e.g.Xu et al. 2000Xu et al. , 2008

REVIEW OF PUTATIVE SYNAPOMORPHIES OF DEINONYCHOSAURIA
1 -Lacrimal with anterodorsal process much longer than posterior process (ch.H40).A welldeveloped anterodorsal process of the lacrimal is present in most deinonychosaurians, and was considered as diagnostic for this group (Novas et al. 2009, Hu et al. 2009).In fact, an elongate anterodorsal process of the lacrimal, rivaling the size of the ventral process is a trait seen in most dromaeosaurids, and appears to be especially long in derived troodontids (Makovicky and Norell 2004).Among unenlagiids, the lacrimal is preserved in Austroraptor, in which the anterodorsal process is slightly curved ventrally, being subequal in length to the ventral process.Notably, Austroraptor resembles troodontids, although in the latter ones the anterodorsal process is even larger than the ventral one (e.g.Byronosaurus, Saurornithoides; Barsbold 1974, Makovicky et al. 2003).The presence of a lacrimal with an anterodorsal process much longer than the posterior process is, however, not unique for deinoychosaurians, as it is also present in Archaeopteryx (Novas et al. 2009) as well as in several basal birds, including Jeholornis and most Enantiornithes (Chiappe and Walker 2002).Unfortunately, the preserved specimens of Confusiusornis (Chiappe et al. 1999) and Zhongornis (Gao et al. 2008) precludes us to known if such morphology of lacrimal also applies to these early birds.
In summary, the present analysis supports that the presence of a large anterodorsal process on the lacrimal is diagnostic of the clade Paraves.

-Nutrient foramina on external surface of the dentary lie within a deep groove (ch. H72).
The presence of a well-delimited, narrow, and deep longitudinal groove along the alveolar margin of the dentary, carrying a series of small nutrient foramina was previously considered as diagnostic of Troodontidae (Sues 1977, Currie 1987).More recently, Xu et al. (2008) have suggested that this condition may be diagnostic of Deinonychosauria, because it is also present in members of the clade Microraptoria (e.g.Microraptor, Sinornithosaurus; Xu et al. 1999Xu et al. , 2000Xu et al. , 2003)), as well as in unenlagiids, (Buitreraptor, Austroraptor, Makovicky et al. 2005, Novas et al. 2009; Fig. 2).Moreover, its presence in the latter taxa prompted Novas et al. (2009) to suggest that it may constitute a diagnostic feature of Unenlagiidae.However, the presence of a similar groove among early avian taxa (e.g., Archaeopteryx, Enantiornithes, Ichthyornis; Elzanowski andWellnhofer 1996, Chiappe andWalker 2002) suggest that this condition may be more widespread than previously thought.
The recoding of this character in the present analysis weakens its status as a deinonychosaurian synapomorphy.Moreover, in the context of the present analysis, this feature is not retrieved as diagnostic of any coelurosaurian clade.
3 -Anterior dentary teeth smaller, more numerous, and more closely appressed than those of the middle tooth row (ch.H90).This condition was originally considered as diagnostic of Troodontidae (Sues 1977).Moreover, Sues (1977) indicated that anterior dentary teeth not only were closely appressed, but also lacked separated alveoli, as all the anterior teeth are set in an open groove.Recently, and based on the discoveries of Microraptor and Sinornithosaurus, Xu et al. (2008) suggested that this condition may be better considered as diagnostic of Deinonychosauria.Preserved portions of the dentary in Austroraptor and Buitreraptor clearly indicate that in the latter two taxa the dentary teeth are inserted into separate alveoli, and that they are similar in size to posterior teeth (Gianechini et al. 2009; Fig. 2B).Hu et al. (2009) considered this morphology as diagnostic of Deinonychosauria.However, such a condition is difficult to quantify in basal deinonychosaurians (e.g.Mei, Sinovenator, Microraptor, Graciliraptor) and have been consequently coded by Hu et al. (2009) as indeterminate in their data matrix.However, preserved remains in Anchiornis, Mei, Sinovenator, and Microraptor suggest that the roots of dentary and maxillary teeth may be subcircular in cross-section (Xu 2002 In conclusion, the degree of compression of tooth roots is a trait with an ambiguous distribution among Paraves and, consequently, it is not phylogenetically informative in the context of the present cladistic analysis.

-Roots of dentary and maxillary teeth mediolaterally compressed (ch. H228).
5 -Dentaries lack distinct interdental plates (ch.H90).(Fig. 2C).Currie (1995) has noted that fused interdental plates in dentary and maxillary bones constitutes a diagnostic character for Dromaeosauridae, and that the absence of such plates in Troodontidae represents an apomorphic condition of this theropod group.Later, Makovicky et al. (2005) emphasized on the difficulty to discern between the absence or fusion of dental plates in paravian theropods, and suggested that both conditions may be treated as a single one.They concluded that both refer to the absence of clearly defined interdental plates.Following this proposal, Xu et al. (2008) considered the absence of distinct interdental plates as a derived condition characteristic of Deinonychosauria.Among Unenlagiidae, the dentary is known in Buitreraptor and Austroraptor.In Buitreraptor the dentaries are still embeded in matrix, so the morphology of interdental plates remains unknown.On the other hand, in Austroraptor the dentaries still preserve clear and well-separated interdental plates in the rostral portion of the bone, at least.These unfused plates are subtriangular in shape, being similar to those observed in Archaeopteryx (Elzanowski and Wellnhofer 1996).In addition, in most basal birds, including Sapeornis, Jeholornis, Confusiusornis, and ornithothoracines, the interdental plates are absent (Chiappe et al. 1999, Zhou andZhang 2003a, b).
In the present analysis, the presence of distinct interdental plates is considered as an ambiguous synapomorphic trait of the clade formed by Aves + Unenlagiidae (i.e.Avialae).

-Splenial exposed as a broad triangle between dentary and angular, on lateral surface of mandible (ch. H76
). Currie (1995) proposed a widely exposed splenial subtriangular in shape as diagnostic of Dromaeosauridae, a criterion followed by subsequent authors (Makovicky et al. 2005, Xu et al. 2008, Hu et al. 2009).Although no single splenial nor angular is completely known in unenlagiids, the preserved portion of the dentary in Austroraptor suggests that the splenial was not exposed in lateral view (Fig. 2D).In fact, in troodontids (e.g.Byronosaurus, Saurornithoides; Barsbold 1974, Makovicky et al. 2003) and dromaeosaurids (e.g.Dromaeosaurus, Sinornithosaurus; Currie 1995, Xu and Wu 2001) the caudoventral process of the dentary is reduced and shows an extensive oblique and posterodorsally oriented articular surface for the splenial, a derived condition unknown in remaining theropods, including birds.In contrast, in Austroraptor the caudal end of the dentary is dorsoventrally deep and the ventral margin curves downwards.This morphology suggests that the splenial was hidden in lateral view (Fig. 2D).In the same way, although the jaws of Buitreraptor are caudally damaged, the splenial appears to lack a lateral exposure, as can be seen in the preserved portion of this bone.and Buitreraptor the antitrochanter is present, but only represented by a slightly raised, craniolaterally oriented surface, which lacks well-defined perimeters (Novas 2004; Fig. 3).This condition resembles the one present in Archaeopteryx.
In this way, the recent analysis by Hu et al. ( 2009) considers the presence of a well-developed antitrochanter forming a "hood" over the femoral head as a widespread trait among paravians and, consequently, excludes it from the diagnosis of Deinonychosauria.Xu et al. (2008) suggested this peculiar condition as diagnostic of Deinonychosauria.In fact, in Archaeopteryx and the remaining birds, there are no signs of a prominent depression for the biceps muscle at the distal end of the deltopectoral crest.On the other hand, in Buitreraptor and Unenlagia there exists a prominent depression bounded by a ridge that defines such biceps insertion (Makovicky et al. 2005), as it occurs in dromaeosaurids.In Austroraptor the humerus is damaged and that the condition of the crest remains uncertain.

-Humerus with anterior surface of deltopectoral crest with a distinct muscle scar near the lateral edge along distal end of crest for insertion of biceps muscle (ch. H141). Makovicky et al. (2005) and
This character is retrieved as one of the derived traits that are shared by Unenlagiidae and Deinonychosauria, with the exclusion of Aves. 9 -Ulna with a thick ridge along the anterior margin of the proximal third of the shaft (ch.X257).Xu et al. (2008) considered this feature as apomorphic of Deinonychosauria.In fact, the presence of such ridge may be corroborated in both Buitreraptor and Rahonavis.In birds, including Archaeopteryx and Enantiornithes, the anterior margin of the ulna is nearly flat (Wellnhofer 1992, Chiappe andWalker 2002).
This character is retrieved as one of the derived traits that are shared by Unenlagiidae and Deinonychosauria, with the exclusion of Aves.
10 -Ulna with little anteroposterior flattening, and with its distal end having a transverse width to anteroposterior length ratio significantly less than 2 (ch.X262).Xu et al. (2008) indicated this feature FEDERICO L. AGNOLIN and FERNANDO E. NOVAS as a synapomorphic trait of Deinonychosauria.Among unenlagiids, the ulna is known in Buitreraptor and Rahonavis.In Buitreraptor this bone is partially preserved, with the distal end damaged.However, the available portion of the ulnar shaft indicates that this genus probably retained the plesiomorphic condition of little anteroposterior compression in the distal end, a morphology that is different from that of Deinonychosauria.The incomplete nature of the ulna in Buitreraptor precludes us to identify with certainty its condition and, in consequence, it is coded as a question mark.Besides, the ulna of Rahonavis does not exhibit an anteroposteriorly compressed distal end (Xu et al. 2008), thus constituting a derived condition shared with most avialans.
In consequence, although the anteroposterior flattening of the ulna may be considered as diagnostic of Deinonychosauria, this condition appears to be absent in known Unenlagiidae.Gauthier (1986), this condition was suggested as apomorphic of Deinonychosauria by Hu et al. (2009), and it can be clearly observed in the preserved specimens of Buitreraptor and Unenlagia (Novas andPol 2005, Makovicky et al. 2005).However, a mound-like posterior trochanter is also present in basal birds, including Archaeopteryx (Fig. 3B), Wellnhoferia, and Enantiornithes (Elzanowski 2002, Chiappe andWalker 2002).

-Femur with posterior trochanter distinctly raised from shaft, mound-like (ch. H183). First considered as diagnostic of Coelurosauria by
In summary, the presence of a mound-like posterior trochanter is retrived in the present analysis as diagnostic of the node Paraves.4) and Jeholornis (IVPP-V13350; Fig. 4) exhibit the same trait.Unfortunately, in most of the remaining basal birds the morphology of the plantar surface of the metatarsals cannot be properly observed due to defficient preservation of available specimens.However, in Enantiornithes, a similar ridge is described on well preserved tarsometatarsi (Chiappe 1993), suggesting that this condition may be widespread among basal Avialae.
In the present analysis, the presence of a caudal flange along metatarsal IV is considered diagnostic of Paraves.

-Metatarsal III distally ginglymous (ch. H199).
The presence of a well-developed ginglymoid distal end of metatarsal III is an apomorphic condition that was firstly suggested by Gauthier (1986) as diagnostic of Deinonychosauria, an interpretation that was followed by most recent authors.In fact, in typical dromaeosaurids, such as Velociraptor and Deinonychus, the ginglymoid is clearly seen in posterior and anterior views, and is well extended proximally when viewed anteriorly (Norell and Makovicky 2004; Fig. 4).In addition, in the dromaeosaurid Deinonychus, for example, the ginglymous distal articulation is proximally delimited by a transverse osseous ridge (Ostrom 1969; Fig. 4).In microraptorians (e.g., Microraptor, Graciliraptor, Sinornithosaurus; Xu 2002) the distal ginglymoid is poorly developed, and no excavation of its distal trochlea nor a proximal transverse ridge are evident in anterior view (Hwang et al. 2002).The latter condition is also present in Buitreraptor and Rahonavis, in which the ginglymous is poorly developed, nearly lacking an anterior excavation (Forster et al. 1998 In most derived Avialae, including Confusiusornis, Ichthyornis, Baptornis, and Neornithes, the distal end of metatarsal III conforms a widely ginglymoid distal trochlea, a condition reminiscent to that seen in typical dromaeosaurids (Martin and Tate 1976, Chiappe et al. 1999, Clarke 2004).
To character 202 we have added a new character-state (state 2), "poorly developed ginglymous, weakly extended proximally", a state not recognized previously (e.g., Hu et al. 2009).Thus, in the present analysis, the presence of a poorly excavated and weakly extended ginglymoid is better interpreted as a paravian feature.

-Pedal phalanx II-2 with transversely wide posteroventral lip (ch. H322). Hu et al. (2009)
considered the presence of a wide posteroventral lip at the base of phalanx II-2 as diagnostic of Deinonychosauria.However, this condition seems to be absent in Unenlagiidae, in which the posteroventral heel of pedal phalanx II-2 is transverselly narrow, and medially offset, as can be observed in Rahonavis, Unenlagia In summary, although the presence of a wide posteroventral heel or lip in pedal phalanx II-2 is here considered as diagnostic of Deinonychosauria, this condition is absent in Unenlagiidae.In this way, the presence of pedal phalanx II-2 with condyles dorsally displaced has a wider distribution among paravians than previously thought.

-Trenchant claw on digit II (ch. H201).
A large and trenchant ungual on pedal digit II was mentioned for the first time as diagnostic of Deinonychosauria by Gauthier (1986), and accepted as such by most recent authors (e.g., Hu et al. 2009).However, recent discoveries demonstrated that this trenchant ungual is present also in some basal avialans, including Archaeopteryx and Jeholornis (Paul 2002, Zhou and Zhang 2003a, b, Mayr et al. 2005, Xu andZhang 2005).In this context, the presence of an enlarged ungual on digit II was considered as synapomorphic of Paraves by Xu and Zhang (2005), an interpretation that is accepted here.Although the morphology of phalanges of digit II is not identical in basal birds and deinonychosaurians, the ungual phalanx in both clades is more curved and larger than the remaining phalanges of the foot.Some authors suggested that the development of a raptorial ungual on pedal digit II was convergently developed in Troodontidae, Dromaeosauridae, and Rahonavis (Forster et al. 1998).Under the present phylogeny, a large ungual on pedal digit II was almost probably acquired early in paravian evolution and was secondarily lost in birds more derived than Jeholornis.
There are some additional traits cited by several authors as diagnostic of Deinonychosauria that cannot be checked among unenlagiids, due to defficient preservation.These traits mostly refer to skull anatomy and include: well developed palatal flange on pterygoid, tetrarradiate palatine, presence of a foramen on the lateral surface of the surangular rostral to the mandibular articulation, coronoid reduced to a thin splint, and manual phalanx III-3 subequal in length to the combined lengths of phalanges III-1 and III-2 (Currie 1995 In consequence, unenlagiids and basal birds lack the dorsally displaced maxillary fenestra diagnostic of Droameosauridae.

-Lacrimal "T"-shaped in lateral view (ch. H40
).The morphology of the lacrimal is peculiar to dromaeosaurids, in which this bone has a "T"-shaped morphology due to its nearly straight dorsal margin, with well-developed and subequal anterior and posterior processes, and a larger, columnar, and vertically oriented jugal process (Norell and Makovicky 2004).In the remaining theropods, the lacrimal is usually "L"-shaped due to the poor development of its posterior process (Currie 1995).Among Unenlagiidae, the only genus in which the lacrimal is preserved is Austroraptor, which lacks the "T"-shaped condition seen in most dromaeosaurids.On the contrary, it has a short and cranially curved jugal process, a shortened posterior process, and an enlarged anterior process, thus conferring an "L"-shaped lacrimal in side view.In this regard, unenlagiids resemble troodontids and basal birds (e.g., Archaeopteryx, Jeholornis; Zhou and Zhang 2003b, Mayr et al. 2007), in which the lacrimal is "L"-shaped.

-Lateral process of the quadrate that touches squamosal and quadratojugal above an enlarged quadrate foramen (ch. H54).
The quadrate bone in most dromaeosaurids and Buitreraptor exhibits, in addition to its medial pterygoid process, a well developed lateral bony wing formed by an extension of the anterior margin of the quadrate shaft for contact with the squamosal (Currie 1995, Norell and Makovicky 2004).This morphology, considered to be unique of dromaeosaurids, is also present in Archaeopteryx, as it was first noticed by Walker (1985).In the same way, a lateral process appears to be also present in the basal avialan Confusiusornis (Chiappe et al. 1999), although in this genus it is highly reduced, being represented only by a narrow flange located near the proximal articular head of the bone (Chiappe et al. 1999).The presence of such process in other basal avialans (e.g.Jeholornis, Sapeornis) is difficult to determine due to poor preservation of most of the specimens.
In summary, the presence of a lateral flange in the quadrate of Archaeopteryx suggests that this trait is more widely distributed among paravians than previously thought.Novas et al. (2009) proposed that this condition may be diagnostic of Dromaeosauridae, including unenlagiids.However, a dorsally excavated frontal process of postorbital is also present in some troodontids (Currie 1985(Currie , 1987) )  ). Regrettably, the dorsal margin of the postorbital is unknown in some basal avialans such as Archaeopteryx, Jeholornis, and Sapeornis.In this way, this condition is here considered as widespread among paravians.

-Anterior emargination of supratemporal fossa on frontal sinusoidal (ch. H43). Xu et al. (2000)
considered a sinusoidal anterior ridge defining the supratemporal fossae as diagnostic of Dromaeosauridae, a morphology discussed in detail by Xu (2002) and Norell et al. (2006).These authors demonstrated that in most eudromaeosaurids (e.g.Bambiraptor, Saurornitholestes, Deinonychus) and microraptorians (e.g.Sinornithosaurus), with the sole exception of Tsaagan and Dromaeosaurus, the anterior ridge that delimitates the supratemporal fossa was sinusoidal (i.e., convex on its lateral side, concave at mid-length, and becoming convex again on its medial side; Xu 2002).In contrast, in the unenlagiids

-Dentary with subparallel dorsal and ventral edges (ch. H71
).This feature was originally noticed by Currie (1995) and later discussed by Xu and Wu (2001), who affirmed that a parallel-sided dentary is a feature diagnostic of Dromaeosauridae.In fact, along most of its length, the dentary of most dromaeosaurids has subparallel dorsal and ventral margins, usually with the dorsal margin being sligthly concave, and the ventral one convex.On the other hand, in Archaeopteryx the dentary exhibits both dorsal and ventral margins slightly divergent caudally, a condition that may be verified in a variety of other basal avialans, including Jeholornis, Ichthyornis, and Enantiornithes (Zhou and Zhang 2003a, Chiappe and Walker 2002, Clarke 2004).Unfortunately, the dentary in Unenlagiidae is poorly known, being distorted and incompletely preserved in Buitreraptor, and distorted in Austroraptor.However, in both Buitreraptor and Austroraptor, the caudal end of the dentary is clearly deeper than the cranial end, showing a subtriangular contour in lateral view, a condition reminiscent to that of basal birds (Fig. 2A).

-Unconstricted teeth (ch. H88
).Among derived features interpreted as diagnostic of dromaeosaurids is the lack of constrictions between the crown and the root of the teeth, in contrast with the well-developed constriction present in troodontids and dentate birds (Currie et al. 1990, Martin et al. 1980).However, a detailed analysis of available specimens that were equivocally distributed among dromaeosaurids, suggests that this morphology is not diagnostic of Dromaeosauridae.Although in most dromaeosaurids, including unenlagiids the teeth are unconstricted (Makovicky and Norell 2004, Makovicky et al. 2005, Gianechini et al. 2009), in several dromaeosaurid genera, such as Atrociraptor, Sauror-FEDERICO L. AGNOLIN and FERNANDO E. NOVAS nitholestes, Velociraptor, Sinornithosaurus, and Microraptor, a constriction between the crown and root is present at least in some of their teeth (Currie et al. 1990, Xu et al. 2000, Currie and Varricchio 2004).On the other hand, in Archaeopteryx a constriction between the root and the crown is not always observable, being only seen at the base of some teeth (Elzanowski and Wellnhofer 1996, Martin and Stewart 1999, Martin et al. 1980).In other basal birds, such as Sapeornis and Jeholornis, the teeth appear to lack signs of crown root constriction (Zhou and Zhang 2003a, b).In conclusion, the presence or absence of a basal constriction is a condition difficult to quantify among paravians, and thus is doubtfully diagnostic of any paravian clade.

-Epipophyses of cervical vertebrae placed distally on postzygapophyses, above postzygopophyseal facets (ch. H95).
Epipophyses of cervical vertebrae in dromaeosaurids are relatively long and slender, and usually surpass the caudal margin of postzygapophyses (Ostrom 1969).On the contrary, available cervicals of Buitreraptor  In conclusion, the elongate epipophyses present in typical dromaeosaurids (e.g.Ostrom 1969) are absent in unenlagiids.

-Anterior cervical centra level with of shorter than posterior extent of neural arch (ch. H96).
Although considered by Novas et al. (2009) as diagnostic of Dromaeosauridae, including unenlagiids, the morphology of the posterior margin of the neural arch is variable among paravians.Although in most theropods the neural arch does not approach the posterior level of the vertebral centrum, in dromaeosaurids, the posterior margin of the neural arch approaches the caudal end of the centrum, a condition that may be also seen in several avialans, including Archaeopteryx (HMN 1880), the enantiornithines Iberomesornis and Gobipteryx, the ornithurine Ichthyornis, and some troodontids, such as Troodon and Sinovenator (Kurochkin 1996, Sereno 2000, Xu 2002, Clarke 2004, Makovicky and Norell 2004).
In summary, this trait appears to be more variable and more widespread than previously thought among theropods and may not be considered as diagnostic of Dromaeosauridae.
An Acad Bras Cienc (2011) 83 (1) THE PHYLOGENETIC POSITION OF UNENLAGIIDAE 131 29 -Middle and posterior dorsal vertebrae pneumatic (ch.H106).This trait was considered by Hu et al. (2009) as apomorphic of Dromaeosauridae, including unenlagiids.However, the presence of pneumatization on mid and posterior dorsal vertebrae appears to be more widespread than suggested by Hu et al. (2009).In fact, in Archaeopteryx, although dorsal vertebrae are poorly preserved, a real pleurocoel foramen is present behind the diapophyses in the two last dorsal vertebrae (Britt et al. 1998, Elzanowski 2002).A similar pleurocoel is also present in basal birds, such as Jeholornis, Confusiusornis, and Ichthyornis (Chiappe et al. 1999, Zhou and Zhang 2003b, Clarke 2004), suggesting a wide distribution of pneumatic vertebrae among basal birds.
In conclusion, the presence of pneumatic middle and posterior dorsal vertebrae may be considered as diagnostic of Paraves, rather than Dromaeosauridae.

-Distal caudal chevrons bifurcate at both ends (ch. H123). Although Makovicky et al. (2005)
proposed that this feature united Unenlagiidae with the remaining dromaeosaurids, this condition is also present in basal avialans and troodontids.In fact, cranially and caudally forked chevrons are seen among avialans in Jeholornis and Archaeopteryx (Wellnhofer 1992, Forster et al. 1998, Zhou and Zhang 2003a), and in troodontids they are present in the basal forms Anchiornis and Mei (Xu et al. 2008, Xu andNorell 2004).In summary, we consider this feature as diagnostic of the node Paraves.Zheng et al. (2009), this condition is widespread among birds with the exclusion of Wellnhoferia and Archeopteryx (Hu et al. 2009).Moreover, the same morphology is seen in the basalmost troodontid Anchiornis (Hu et al. 2009).This may suggest a wider distribution of this trait than previously thought.

-Metatarsal II distally ginglymoid (ch. H198).
The distal end of metatarsal II has a deep and well-excavated distal trochlea forming a ginglymous articular surface, which was considered as diagnostic of Deinonychosauria by Gauthier (1986).More recently, this condition was suggested as apomorphic of Dromaeosauridae, including Unenlagiidae (Novas andPol 2005, Makovicky et al. 2005).Curiously, in the basal dromaeosaurids Sinornithosaurus and Microraptor the distal end of metatarsal II lacks a deep excavation on its cranial surface (Xu 2002, Hwang et al. 2002).The same is also true for Unenlagia, Rahonavis, and Buitreraptor, in which the distal end of metatarsal II is poorly excavated in anterior view (Forster et al. 1998 ).Thus, the equivocal distribution of this condition, suggests that the excavated metatarsal II trochlea may not be informative to diagnose Dromaeosauridae.
33 -Pedal phalanx II-2 strongly constricted dorsoventrally at mid-shaft.This condition, first noticed as diagnostic of Dromaeosauridae by Rauhut and Werner (1995) was considered as present in Unenlagia (= Neuquenraptor) by Novas and Pol (2005).This morphology is probably correlated with the special-FEDERICO L. AGNOLIN and FERNANDO E. NOVAS ized articulation for the enlarged and trenchant ungual II-3, as early noted by Ostrom (1969).Although this character is clearly present in Dromaeosauridae and unenlagiids, more recent findings indicate that similar phalangeal morphology is present in basal avialans, such as Jeholornis and Archaeopteryx (Paul 2002, Zhou andZhang 2003a), as well as troodontids (Norell and Makovicky 2004), suggesting that this trait may be diagnostic of paravians rather than of Dromaeosauridae.
34 -Pedal phalanx II-1 shorter than pedal phalanx IV-1 (ch.H342).Hu et al. (2009) suggested that this condition may be considered as an apomorphy shared by all dromaeosaurids.In troodontids, as well as in basal birds (e.g., Archaeopteryx, Jeholornis, Confusiusornis; Chiappe et al. 1999, Paul 2002, Zhou and Zhang 2003b), the pedal phalanx II-1 is much longer than phalanx IV-1, a morphology that is clearly different from the abbreviated condition of phalanx II-1 seen in dromaeosaurids, including Rahonavis (Forster et al. 1998).In the unenlagiids Buitreraptor, Unenlagia, and Austroraptor the phalanx IV-1 remains unknown, and thus the present character cannot be evaluated properly.

-Hallucal ungual strongly curved (ch. H319).
Although a strongly curved hallucal ungual was considered as diagnostic of Dromaeosauridae by Hu et al. (2009), this condition is widespread among basal birds, including Archaeopteryx, Wellnhoferia, Sapeornis, Jeholornis and Enantiornithes, among others (Elzanowski 2002, Zhou and Zhang 2003a, Martin 1995).In these birds the presence of a large and curved hallucal ungual, in association with a reversed hallux, was considered as indicative of arboreal capabilities (Zhou and Zhang 2003b).This indicates that the curvature of pedal ungual of digit I is more widely distributed among paravians than suggested by Hu et al. (2009).
As is the case of Deinonychosauria, several traits considered as diagnostic of Dromaeosauridae by previous authors are not preserved in any unenlagiid specimen.Among them are: posterior process of premaxilla excluding the maxilla from naris, caudal margin of nares farther rostral than the rostral border of the antorbital fossa, large spike on the medial inflected process of the articular, absence of an accesory tympanic recess to the crista interfenestralis, and elongate paroccipital process with parallel dorsal and ventral edges that twist rostrolaterally distally (Norell et al. 2006, Turner et al. 2007b, Hu et al. 2009).

CONCLUSIONS
After recoding several traits in the data matrix of the most integrative and recent analysis of coelurosaurian phylogeny (Hu et al. 2009), several changes in the tree topology are observed.
Our phylogeny resembles that of Hu et al. (2009) in depicting a resolved Paraves, which is diagnosed by an extensive list of features, including acromial margin of the scapula with anterior edge laterally everted, proximal end of ulna divided into two fossae, large and trenchant pedal ungual claw on  Hu et al. 2009) in the exclusion of Unenlagiidae from both Dromaeosauridae and Deinonychosauria.In fact, unenlagiids appear to be more nearly related to Aves than to other theropods.Both unenlagiids and birds share, among other character-states, dentaries with well-defined interdental plates, preacetabular part of ilium markedly longer than postacetabular portion, widely separated distal ends of ischia, and coracoid with acrocoracoid process (see Appendix III).In the same way, some authors noted additional similarities between unenlagiids and birds, as for example elongate forelimbs, vertically oriented pubes, and numerous details of scapular anatomy (Paul 2002, Senter 2007).
Unenlagiidae is here considered monophyletic and includes the genera Rahonavis, Buitreraptor, Unenlagia, and Austroraptor, which are diagnosed by pedal phalanx II-2 with transversely narrow posteroventral keel, dorsal margin of postacetabular blade of ilium concave, and ischium with posterior end deeply concave (see Appendix III).Within Unenlagiidae, Unenlagia and Buitreraptor + Austroraptor share, with the exclusion of Rahonavis, the presence of a tuber along the dorsal edge of ilium and obturator process of ischium forming an acute angle in lateral view.Moreover, Buitreraptor and Austroraptor share non-pneumatic middle and posterior dorsal vertebral centra.
Rahonavis was included within Unenlagiidae in the present analysis, a result that is in agreement with the proposal of Makovicky et al.In conclusion, the phylogenetic position of Unenlagia and its kin, as sister group of Aves, allows us to further investigate in detail the peculiar anatomy of this theropod group, which will raise questions about phylogenetic and functional significances, leading us to consider with better information poorly known aspects of the early evolution of avian flight (e.g., postural activities, flapping abilities).

RESUMO
No presente trabalho analisou-se a posição filogenética do clado derivado de terópodas gondwânicos Unenlagiidae.96.Anterior cervical centra level with or shorter than posterior extent of neural arch (0).or centra extending beyond posterior limit of neural arch (1).97.Carotid process on posterior cervical vertebrae absent (0) or present (1).98. Anterior cervical centra subcircular or square in anterior view (0) or distinctly wider than high, kidney shaped (1).99.Cervical neural spines anteroposteriorly long and dorsoventrally tall (0) or anteroposteriorly short, dorsoventrally low and centred on neural arch, giving arch an "X" shape in dorsal view (1) or anteroposteriorly short and dorsoventrally tall (2) or anteroposteriorly long and dorsoventrally short (3).148.Distal carpals 1+2 well developed, covering all of proximal ends of metacarpals I and II (0) or small, cover about half of base of metacarpals I and II (1) or cover bases of all metacarpals (2).149.Metacarpal I half or less than half the length of metacarpal II, and longer proximodistally than wide transversely (0) or subequal in length to metacarpal II (1) or very short and wider transversely than long proximodistally (2).150.Third manual digit present, phalanges present (0) or reduced to no more than metacarpal splint (1).
152.Unguals on all digits generally similar in size (0) or digit I bearing large ungual and unguals of other digits distinctly smaller (1).153.Proximodorsal "lip" on first manual ungual -a transverse ridge immediately dorsal to the articulating surface -absent (0) or present (1).
155.Preacetabular part of ilium roughly as long as postacetabular part of ilium (0) or preacetabular portion of ilium markedly longer (more than 2/3 of total ilium length) than postacetabular part (1).

THE PHYLOGENETIC POSITION OF UNENLAGIIDAE 119 Fig. 1 -
Fig. 1 -Alternative interpretations about the phylogenetic relationships of Unenlagiidae.A, phylogeny proposed by Hu et al. (2009), depicting unenlagiids as nested within Deinonychosauria and Dromaeosauridae; B, present analysis, depicting unenlagiids as the sister group of Aves.
, Makovicky et al. 2005, Novas and Pol 2005, Norell et al. 2001, 2006, Turner et al. 2007a, b, Novas et al. 2009).Characters whose definitions were taken from Hu et al. (2009) are preceded with the letter H; characters taken from Xu et al. (2008) are preceded with the letter X.
, Novas and Pol 2005, Makovicky et al. 2005).This morphology is very similar to that present in the basal birds Jeholornis and Archaeopteryx, as noted by previous authors (Mayr et al. 2005, Novas and Pol 2005, Xu et al. 2008).
Hu et al. (2009) considered the distal condyles of pedal phalanx II-2 dorsally displaced as a diagnostic trait of Deinonychosauria.In An Acad Bras Cienc (2011) 83(1) THE PHYLOGENETIC POSITION OF UNENLAGIIDAE 127 fact, the dorsally extended condyles of this phalanx are related to the hyperextensible pedal second toe and its corresponding sickle-like claw, a condition traditionally regarded as diagnostic of Deinonychosauria(Gauthier 1986).This derived morphology is present in most troodontids and dromaeosaurids, including Rahonavis, Unenlagia, Buitreraptor, and Austroraptor(Forster et al. 1998, Makovicky et al. 2005, Novas et al. 2009).However, dorsally extended condyles on distal end of phalanx II-2 have been also reported for a wide variety of basal birds, including Archaeopteryx, Jeholornis, Sapeornis, Confusiusornis, and Enantiornithes (Chiappe et al. 1999, Sereno 2000, Paul 2002, Zhou and Zhang 2003a, b).
, Makovicky et al. 2005, Currie and Varricchio 2004, Xu et al. 2000, 2008, Novas et al. 2009, Hu et al. 2009).REVIEW OF PUTATIVE SYNAPOMORPHIES OF DROMAEOSAURIDAESeveral features suggested as diagnostic of Dromaeosauridae by previous authors are proven to be conflictive or more widely distributed than previously thought.These traits are the following:19 -Maxillary fenestra represented by a small and dorsally displaced opening (ch.H240).Turner et al. (2007b) indicated that a dorsally located maxillary fenestra is diagnostic of Dromaeosauridae.However, in both Autroraptor and Buitreraptor the maxillary fenestra is not dorsally displaced, a condition that was considered diagnostic of Unenlagiidae byNovas et al. (2009).In basal birds (e.g.Archaeopteryx, Confusiusornis;Chiappe et al. 1999, Mayr et al. 2007) and in Buitreraptor and Austroraptor the maxillary fenestra is very large and round, a plesiomorphic condition widespread among theropods(Hu et al. 2009; ch.H240, state 0).
, Novas and Pol 2005, Makovicky et al. 2005; Fig. 4).Although a distal ginglymoid is clearly absent in several basal avialans, including Archaeopteryx, Jeholornis and Zhongornis (Mayr et al. 2007, Zhou and Zhang 2003a, Gao et al. 2008), in other early birds, such as Enantiornithes, Ichthyornis, Apsaravis, and Sapeornis a distal ginglymoid is present and well developed, conforming a deep distal trochlea on metatarsal II (Chiappe and Walker 2002, Clarke and Norell 2002, Zhou and Zhang 2003b, Clarke 2004, Xu et al. 2008 (2005)  and in contrast with some phylogenies that posit Rahonavis well internested among Aves(Zhou and Zhang 2003a, Hwang et al. 2002, Turner et al. 2007b, Xu et al. 2008).However, the incomplete nature of the known specimens, together with several similarities shared between Rahonavis and birds noted by previous authors(Forster et al. 1998, Novas 2004, Xu et al. 2008), indicate that more detailed studies may relocate Rahonavis within Aves.
, Xu and Norell 2004, Hu et al. 2009), as also occurs in basal birds (e.g.Archaeopteryx, Enantiornithes; Elzanowski 2002, Chiappe and Walker 2002).Within unenlagiids, the morphology of tooth roots is variable: in Buitreraptor the roots are transversely compressed, but in Austroraptor the roots are subcircular in cross section (Gianechini et al. 2009).
among other outstanding characters (see Appendix III).Our results also indicate that Dromaeosauridae and Deinonychosauria are both probably monophyletic (see Appendix III for a complete list of synapomorphies).However, our phylogeny clearly departs from previous analyses (Norell et al. 2001, Xu et al. 2002, 2008, Xu and Norell 2004, Novas and Pol 2005, Novas et al. 2009,