A new genus for the tiny hawk Accipiter superciliosus and semicollared hawk A. collaris (Aves: Accipitridae), with comments on the generic name for the crested goshawk A. trivirgatus and Sulawesi goshawk A. griseiceps

Multiple molecular phylogenetic studies have demonstrated that two Neotropical raptors, tiny hawk Accipiter superciliosus and its sister species semicollared hawk A. collaris, are not closely related to core Accipiter, and that A. superciliosus, at least, possesses osteological characters not replicated in the remainder of the genus. Based on these data, there is a need to recognise their distinctiveness at generic level. However, as recently noted in two global bird checklists, no name is available to accommodate them, so we provide a new nomen here. Furthermore, two Asian accipitrids, crested goshawk A. trivirgatus and its presumed closest relative Sulawesi goshawk A. griseiceps, are also phylogenetically distinctive; in this case the genus-group name Lophospiza is applicable. We also designate type species for two genus-group names (Hieraspiza and Eusparvius) currently in the synonymy of Accipiter, and, as an aid to future workers, we provide a synonymy of the genus Accipiter and a list of species currently included in Accipiter for which published molecular phylogenetic data are apparently lacking.


Introduction
Accipiter Brisson, 1760, is a large, virtually cosmopolitan and morphologically variable genus. Recent studies based on mitochondrial and nuclear DNA sequences have shown that Accipiter (sensu Dickinson and Remsen 2013) is not monophyletic. These studies provide strong evidence that the harriers (genus Circus Lacépède, 1799) form part of the Accipiter clade (Kocum 2006(Kocum , 2008Griffiths et al. 2007;Lerner et al. 2008;Hughall and Stuart-Fox 2012;Nagy and Tökölyi 2014;Jiang et al. 2015;Oatley et al. 2015;Mindell et al. 2018). The genera Erythrotriorchis Sharpe, 1875, andMegatriorchis Salvadori &D'Albertis, 1876, have also been recovered within the Accipiter clade (Mindell et al. 2018). Conversely, several studies have shown that the Neotropical tiny hawk A. superciliosus (Linnaeus, 1766) occupies a position outside the Accipiter clade (Kocum 2006(Kocum , 2008Hughall and Stuart-Fox 2012;Oatley et al. 2015;Mindell et al. 2018). Another Neotropical species, semicollared hawk A. collaris P. L. Sclater, 1860, was found to be its sister species (Mindell et al. 2018). In all studies, the precise position of A. superciliosus and A. collaris proved unclear, due to low nodal support, but these species are clearly not part of the Accipiter mega-clade (including Circus + Erythrotriorchis + Megatriorchis), which itself was supported by high posterior probability (Kocum 2006(Kocum , 2008Hughall and Stuart-Fox 2012;Oatley et al. 2015;Mindell et al. 2018). Even more distantly related to Accipiter is crested goshawk A. trivirgatus (and presumably Sulawesi goshawk A. griseiceps), which is sister to a clade comprising all other species of Accipitrinae and Buteoninae (Mindell et al. 2018;Choi et al. 2021). For an overview of these relationships, see Fig. 1. Olson (2006) pointed out that A. superciliosus shows several unique osteological characters in the context of Accipiter and recommended that it be placed in a separate genus, Hieraspiza Kaup, 1844. Olson (2006 based himself on Hellmayr and Conover (1949: 48, footnote), who reported that 'Hieraspiza Kaup was originally created for several "East Indian species" to which, the author says, virgatus might possibly belong. A few years later (in Oken's Isis, 1847, col. 169) Kaup specifically listed A. tinus, A. minullus, and A. virgatus as pertaining to the genus, among which Gray, in 1855, selected Falco tinus as genotype.' Falco tinus Latham, 1790, is a junior synonym of A. superciliosus (Stresemann and Amadon 1979). Subsequently, Mindell et al. (2018) considered that genetic and morphological data indeed support the recognition of a separate genus for A. superciliosus and A. collaris, and followed Olson (2006) by using Hieraspiza.

Genus-group name of A. superciliosus and A. collaris
In contrast, Mathews (1920: 67) (2014) stated that A. virgatus is the type species of Hieraspiza and that either a new genus must be erected for A. superciliosus or an existing genus name must be discovered.
The name Hieraspiza was first used by Kaup (1844: 116), where it is a nomen nudum, because there is no description and no valid 'indication' (ICZN 1999, Art. 12).
The relevant text, in its entirety, reads: 'Falkenweihesperber, die ich Hieraspiza nenne, scheinen einige ostindische Arten zu bilden, zu welchen vielleicht virgatus gehört.' In other words, Kaup stated that some East Indian species seem [our emphasis] to form a group, to which virgatus perhaps [our emphasis] belongs, and he applied the name Hieraspiza to this group. As the inclusion of virgatus is only tentative, this species is deemed not to have been originally included (Art. 67.2.5) in 1844, contra Friedmann (1950) and others, so it cannot serve as an indication. Subsequently, Kaup (1845) listed this genus again, this time explicitly including just two species, virgatus and "Dussumieri" [= Accipiter badius dussumieri], thus providing a valid indication (Art. 12.2.5) 1 . As a result of its exclusion from the originally included species, the possible case for superciliosus as the type species is unsustainable (Art. 67.2) because Kaup did not link it with this genus until 1847. This makes G. R. Gray's (1855) subsequent selection of "Falco tinus" as the type for Hiera spiza irrelevant. To stabilize this name, we hereby select Falco virgatus Temminck, 1822 (= Accipiter virgatus) as type species of the genus-group name Hieraspiza.
In searching for an existing name to accommodate A. collaris and A. superciliosus, we assembled a list of synonyms of Accipiter (sensu Dickinson and Remsen 2013), which is based on multiple sources but principally Friedmann (1950) (see Appendix 1). Although only 33 of the 49 species of Accipiter, Erythrotriorchis and Megatriorchis (sensu Dickinson and Remsen 2013) were included in the most comprehensive phylogenetic study to date (Mindell et al. 2018) 2 , these represent all but one of the type species of the available genus-group names current-1 For those who might seek to recognise Hieraspiza as having been validly introduced in 1844, invoking Art. 12.2.5 to suggest that use of the species name virgatus under the genus name is sufficiently clear, despite Kaup's evident double uncertainty, it nevertheless remains the case that the type species of Hieraspiza is virgatus. In other words, the case against Hieraspiza being an available genus name for Accipiter superciliosus is unambiguous. ly listed within the synonymy of Accipiter. The exception is spot-tailed goshawk Accipiter trinotatus (Bonaparte, 1850), the type species of Erythrospiza Kaup, 1867, a name unavailable due to its being a junior homonym (see Appendix 1), but also, automatically, the type of its three derivatives: Chirospizias Sundevall, 1874, a replacement name 3 ; Erythrospizias Gurney, 1875, an unjustified emendation but nonetheless an available genus-group name (Art. 33.2.3); and Spilospiza Salvadori, 1875, another replacement name. Nevertheless, the earlier name Eusparvius Bonaparte, 1854, is also available and can be used for this species, Bonaparte (1854: 538) having included A. trinotatus, A. griseiceps, A. hiogaster, and A. rufitorques within the original grouping (note that none of these has been screened molecularly; see Appendix 2). All four of these available genus-group names (Eusparvius 4 , Chirospizias, Erythrospizias and Spilospiza) might be considered nomina oblita as they have barely been mentioned, never mind used, since being introduced. No type species for Eusparvius Bonaparte, 1854, was originally designated or appears to have been subsequently fixed, and Art. 68 cannot be used to select one, so we hereby fix the type of Eusparvius as A. trinotatus, the first of the four species originally listed by Bonaparte for the oldest available genus-group name applicable to this species.
A. trinotatus is endemic to Sulawesi. It shares no diagnostic character state with A. superciliosus and A. collaris, and differs strongly from these two Neotropical species in various characters, including: tail pattern (large spots on the central rectrices in A. trinotatus, barred in A. superciliosus and A. collaris); tertial pattern (large white spots in A. trinotatus, plain in A. superciliosus and A. collaris); and the pattern of the underparts (plain in A. trinotatus, barred in A. superciliosus and A. collaris) (del Hoyo et al. 1994;Ferguson-Lees and Christie 2001). Despite the lack of phylogenetic evidence for the placement of A. trinotatus, we believe its morphology and biogeography provide no reason to suspect a close relationship with A. superciliosus and A. collaris.
We conclude that A. superciliosus and A. collaris cannot be isolated in any existing genus and that there are no 3 Sundevall specifically named as his type species 'Erythrospiza griseigularis' (= Accipiter hiogaster griseogularis), but because this is a replacement name it has the same type species as the name it replaced, i.e. Accipiter trinotatus (Art. 67.8, '… the same type species … despite any statement to the contrary'). 4 This name was introduced in a manner identical to that in which Bonaparte (1854) introduced the genus-group name Eunisus, just seven lines earlier. Eunisus has been accepted as valid by authorities such as Richmond (1917) and Friedmann (1950), although both of them overlooked Eusparvius. Richmond (1917: 590) interpreted Eunisus as a replacement name for Bonaparte's own genus Nisus, preoccupied by Lacépède's Nisus. It is reasonable to treat Eusparvius identically, as it appears to be a replacement name for Sparvius Bonaparte, preoccupied by Sparvius Vieillot. Bonaparte's names Nisus and Sparvius do not appear to have been published elsewhere. available genus-group names applicable to these species. Thus, it is necessary to provide a new genus-group name. We propose: Microspizias gen. nov.

Diagnosis.
Microspizias differs from all species of Accipiter by a combination of (i) small size (total length <30 cm), (ii) white vent barred grey (in adult M. collaris) or chocolate-brown (in adult M. superciliosus), and (iii) juveniles dimorphic, rufous morph with distinct rufous fringes to feathers of upperparts (Ferguson-Lees and Christie 2001). In addition, Olson (2006) noted that in M. superciliosus the procoracoid process has a very distinct foramen. This foramen is invariably absent in Accipiter (Olson 1987). Olson (2006) further pointed out that the configuration of the skull, sternum and pelvis of M. superciliosus are very different from Accipiter, and that the hind-limb bones of M. superciliosus are much more robust than the extremely gracile elements of Accipiter.
Microspizias differs from Kaupifalco monogrammicus by (i) absence of black and white throat stripes (black central throat stripe bordered on each side by a white stripe in Kaupifalco), (ii) absence of a solid grey breastband (present in Kaupifalco), (iii) presence of three grey tail bands (one white band in Kaupifalco), and (iv) yellow cere, tibia and toes (orange in Kaupifalco) (Ferguson-Lees and Christie 2001).
Microspizias differs from Harpagus in (i) presence of three grey tail bands (two or three white or grey bands in Harpagus), (ii) adult male without dark mesial throat stripe (present in Harpagus), (iii) greyish-barred underparts (plain grey or rufous, or rufous-barred in adult Harpagus), and (iv) juveniles dimorphic, rufous morph with rufous feather fringes on upperparts (monomorphic, with brown to black-brown upperparts in Harpagus) (Friedmann 1950;Ferguson-Lees and Christie 2001).
Etymology. The name is derived from the Greek words μικρος (small, tiny) and σπιζιας (hawk). Its gender is masculine. The name refers to the small size of both species, especially that of M. superciliosus.

Genus-group name of A. trivirgatus and A. griseiceps
Placement of crested goshawk A. trivirgatus in a separate genus from Accipiter is warranted based on its phylogenetic distinctiveness (Mindell et al. 2018;Choi et al. 2021). Lophospiza Kaup, 1844, is available as a genus-group name for this species and its presumed closest relative, Sulawesi goshawk A. griseiceps. Its gender is feminine. Crested goshawk and Sulawesi goshawk thus become Lophospiza trivirgata and Lophospiza griseiceps, respectively. Morphologically, they differ from all species of Accipiter by their well-developed crest (Ferguson-Lees and Christie 2001).