In the present study, at least one species, including type species of all nominal genera except for Pteroleptus, was selected as representative of each pteroin genus. Two species of closely related genera, Hoplosebastes armatus and Scorpaenodes guamensis (Ishida, 1994; Smith et al., 2018), and four additional species of the Scorpaenoidei, including Neosebastes entaxis, Rhinopias eschmeyeri, Scorpaenopsis ramaraoi, and Thysanichthys crossotus, were selected as outgroups. Fish were purchased from fish landing sites or aquarium shops. The collection information is shown in Table S1 . Additional seven COI sequences of various species, including five species for which we do not have access to sequencing and morphological examinations, were downloaded from GenBank and BOLD Systems ( Table S3 ). In total, 19 out of 30 valid species of the Pteroini and six outgroups were included in the phylogenetic analyses ( Table S3 ). Tissue samples were collected from muscle or fins and preserved in 95% ETOH and stored at −20°C. Voucher specimens were fixed in 10% neutral buffered formalin and transferred to 70% ETOH for permanent preservation. They were deposited in the Department of Oceanography, National Sun Yat-sen University, Kaohsiung (DOS), and Academia Sinica Institute of Zoology, Taiwan (ASIZP). Catalog numbers for voucher specimens are listed in Table S3 .

Meristic counts were generally made on the left side and followed by Motomura (2004b) and Motomura et al. (2005). The standard length (SL) is measured as the direct distance from the tip of the upper lip to the middle of the posterior margin of the hypural plate. The last two dorsal and anal soft rays were counted as a single ray. Pectoral-fin rays are counted beginning with the uppermost ray. The longitudinal scale rows were taken from above the first pored lateral scale to the caudal-fin base; the number of near-vertical to oblique scale rows is above the lateral line. The terminology of head spines followed the diagram from Randall and Eschmeyer (2002). Terminologies of color patterns on the head were generally followed by Matsunuma and Motomura (2014); Matsunuma et al. (2017), and Matsunuma and Motomura (2019). Character mapping of selected characters was performed on the phylogeny using the stochastic mapping approach in Mesquite version 2.75 (Maddison and Maddison, 2011).

Two mitochondrial (COI and cyt b) and five nuclear (gylt, plagl2, Ptr, rhodopsin, and zic1) genetic markers were selected for phylogenetic reconstruction. Genomic DNA was extracted from tissues using a GeneMark Easy Tissue and Cell Genomic DNA Purification Kit following the manufacturer’s protocol. Polymerase chain reactions (PCR) were performed in a 25 μl volume containing 3 μl of 10× Taq Buffer, 2 μl of dNTP mixture at 10 mM, 1 μl each of forward and reverse primers at 5 μM, 0.125 μl of Pro Taq Plus DNA polymerase (Protech Technology Enterprise, Taiwan), 1 μl of template DNA, and the remaining ultrapure water. In some cases when PCR failed, PCR reactions were performed in a 25 μl volume consisting of 12.5 μl of SuperRed PCR Master Mix (2x), 1 μl each of forward and reverse primers at 5 μM, 1 μl of template DNA, and the remaining ultrapure water. COI was amplified using combinations of universal COI primer pairs (Ward et al., 2005), FishF1, FishF2, FishR1, and FishR2. Cyt b was amplified using previously published primer pairs (Schmidt and Gold, 1993; Perdices et al., 2001), L14724 and H15915. Rhodopsin was amplified using combinations of previously published primer pairs (Chen et al., 2003; Chen et al., 2008), RH28F, RH1039R, and RH193F. The remaining nuclear genes (gylt, plagl2, Ptr, and zic1) were amplified using the primers reported by Li et al. (2007). The thermal cycle profiles of COI consisted of an initial denaturation step at 95°C for 4 min, followed by 35 cycles of 94°C for 30 s, 48°C for 30 s, and 72°C for 1 min; 92°C for 3 min, followed by 34 cycles of 92°C for 1 min, 53°C for 90 s, and 72°C for 3 min for cyt b; 94°C for 3 min, followed by 10 cycles of 94°C for 45 s, 58°C for 45 s, and 72°C for 75 s, and then additional 30 cycle of 94°C for 45 s, 55°C for 30 s, and 72°C for 75 s for five nuclear genes. All reactions ended in a final step at 72°C for 4–10 min. All PCR products were verified on 2% agarose gels and reliable enzymatically cleaned with the SAP-Exo Kit (Jena Bioscience, Jena, Germany). PCR products were sequenced in the forward and reverse directions by a biotechnology company (Genomics, Taiwan). Forward and reverse sequences were assembled and edited using BioEdit ver. 7.2.5 (Hall, 1999). Accession numbers for sequences generated in this study and downloaded from databases are listed in Table S3 .

The alignments of multiple sequences were independently performed on each gene using clustalW (Thompson et al., 1994) in BioEdit version 7.2.5 (Hall, 1999). Substitution saturation of all genes was tested using DAMBE version 6.3.17 (Xia, 2013). Models of nucleotide substitution were determined for each genetic marker using PartitionFinder 2 (Lanfear et al., 2017). The optimal partitioning and model scheme identified the data as four partitions, including COI + cyt b (the best fit model as TRN + I + G), Ptr + rhodopsin + plagl2 (HKY + I + G), gylt (K80 + G), and zic1 (HKY + I). The sequences of seven genes were concatenated for phylogenetic analysis. For both datasets of COI and concatenated sequences, Maximum likelihood (ML) analysis was performed using the RAxML Blackbox web server (Kozlov et al., 2019). Bootstrap support was calculated with 100 reiterations. Maximum parsimony (MP) analysis was conducted using PAUP version 4.0 (Phylogenetic Analysis Using Parsimony) (Swofford, 2002) with 1,000 bootstrap reiterations. Bayesian inference (BI) of the combined molecular and selected seven morphological data was run using MrBayes version 3.2.2 (Ronquist et al., 2012). The parameter for the morphological coding was independently set to a gamma-shaped rate variation. The Markov chain Monte Carlo (MCMC) analysis was simultaneously run in four parallel chains for 3,000,000 generations with a sample frequency of every 1,000 generations, after checking that it was sufficient for convergence. A consensus tree was constructed after discarding the burn-in of the first 750,000 generations and evaluating statistical confidence in nodes by Bayesian posterior probabilities. A phylogenetic tree was displayed with FigTree version 1.4.3 (Rambaut, 2007).

Fourteen species of five pteroin genera and six outgroups were sequenced. A total of 245 sequences were generated in the present study. The concatenated sequences consist of two mitochondrial and five nuclear genetic markers, including COI (678 bp), cyt b (896 bp), gylt (732 bp), plagl2 (815 bp), Ptr (751 bp), rhodopsin (790 bp), and zic1 (673 bp), amounting to a total of 5,335 bp.

All genes were not found to have reached saturation. The individual gene trees of ML, BI, and MP are shown in Figure S1 . Topologies among individual gene trees were not highly congruent, but ML, BI, and MP trees were reconstructed based on concatenated sequences as in previous studies (Schonhuth et al., 2015; Tavera et al., 2018; Costa et al., 2020; Sudasinghe et al., 2020; and many more). Only ML topology ( Figure 2 ) was shown because the three phylogenetic results were consistent in most clades. The Pteroini were a monophyletic group with high support of 100 bootstrap values or posterior probabilities of 1 in all analyses ( Figure 2 ). Eight major clades within the Pteroini were consistently recovered in the ML, BI, and MP trees. The two specious genera, Pterois and Dendrochirus, were both polyphyletic with Pterois divided into two groups (Pterois I and II) and Dendrochirus divided into three groups (Dendrochirus I, II, and III; Figure 2 ). Pterois I comprised the type species P. volitans, P. russelii, and P. lunulata. Pterois II included P. antennata, P. radiata, and P. paucispinula. Dendrochirus I was a monospecific clade comprising the type species Dendrochirus zebra. Dendrochirus II consisted of D. bellus and D. brachypterus, while Dendrochirus III was only composed of D. biocellatus.

In the ML and BI trees, Dendrochirus III was at the basal node of the Pteroini, followed by the clade comprising two species of Brachypterois. The remaining taxa were divided into two clades. The first was composed of Dendrochirus II + Ebosia + Parapterois, in which Dendrochirus II was the sister group of the genera Ebosia and Parapterois. The second contained Dendrochirus I + Pterois I + Pterois II with Pterois I sister to the sub-clade of Dendrochirus I + Pterois II. The topology of the MP tree was nearly identical to the ML and BI trees except for that the interrelationships of Pterois I, Dendrochirus I + Pterois II, Dendrochirus II + Ebosia + Parapterois, and Brachypterois were unresolved.

To maximize taxon sampling, the ML tree of the COI gene was reconstructed based on fragments of 655 bps generated in this study and downloaded from databases ( Table S3 ). The additional species from the online database completed the sampling of Brachypterois, and added one more species to Ebosia and Pterois, respectively. The eight major clades were also retrieved from the COI tree with high bootstrap supports ( Figure 3 ), but their interrelationships were different from the concatenated tree ( Figure 2 ). The Dendrochirus III clade was at the basal node, followed by Parapterois. The remaining taxa were divided into two clades. The first was composed of Dendrochirus II and Pterois I. The second comprised Brachypterois, Dendrochirus I, Ebosia, and Pterois II with Brachypterois at the basal node. In the COI tree, Brachypterois and Ebosia were monophyletic while the monophyly of Parapterois could not be determined since there is only one species included in the analysis.

Seven coloration patterns ( Figure 2 ) composed of several conspicuous reddish-brown to dark bands from cheek to postorbital region on head were recognized in seven clades (Ebosia and Parapterois shared the same coloration pattern in one clade). The first pattern was present in the Pt. I clade and was characterized by three narrow bands: the first band extending through the eye, from supraocular spine base to interopercle; the second band from the posteroventral margin of the orbit, reaching obliquely to subopercle; and the third band saddling the nape, reaching the central posterior of the opercle. The second pattern was found in Pt. II clade, characterized by three broad bands and a large black blotch: the first band obliquely crossing the eye, extending to the interopercle; the second band from the posteroventral margin of the orbit, reaching obliquely to the first band, but this band was absent in some individuals; and the third band saddling the nape, reaching the margin of the subopercle. The joining of the first and third bands is mostly connected with a large black blotch on the interopercle (rarely connected without a large black blotch, seen in Pterois cincta and P. radiata) (see Matsunuma and Motomura, 2016b, Figures 2 - 4 , 6). The third pattern was present in the D. I clade and was characterized by two broad bands and a large black blotch: the first band crossing the eye to interopercle, with the lower tip forwardly curved; the second band saddling nape, reaching margin of subopercle, the lower tip backwardly curved. The large black blotch is also present on the interopercle but is not connected with the first and second bands. The fourth pattern was present in D. II clade characterized by three broad bands; the first band extending through eye, from supraocular spine base to the joining between interopercle and mandible; the second band from posterior of orbit and extending to interopercle; the third band saddling nape, reaching margin of the subopercle with the lower tip forwardly curved, but this band is absent in some individuals. The fifth pattern was present in the Ebosia + Parapterois clade and was characterized by three broad bands: the first band extending through the eye, from supraocular spine base to the joining between interopercle and mandible; the second band from posterior of the orbit and extending to the interopercle; the third band saddling the nape, reaching the central posterior of the opercle. The sixth pattern was found in the Brachypterois clade and was characterized by an irregular large black blotch present on the opercle. The seventh pattern was found in the D. III clade and was characterized by three irregular large blotches: the first blotch at the posteroventral margin of the orbit; the second at the interopercle; and the third at the nape. The third one is absent in some individuals, and the first and second blotches together look like a large blotch with a narrow notch on the upper side. In addition to color patterns from cheek to postorbital region on head, six characters were summarized to diagnose all clades except for D. II and D. III clades, including (a) absence of pectoral-fin branched rays ( Figures 4A, B ), (b) upper pectoral rays free from membrane ( Figures 4A–C ), (c) number of pectoral-fin rays, (d) number of scales in the longitudinal series, (e) distance between suborbital ridge to orbit ( Figure 5 ), and (f) number of short barbels on the snout tip ( Figure 2 ). The information on the remaining selected morphological characters is shown in Figure 2 and Table S2 . In addition, the result of the ancestral state reconstruction for six diagnostic characters in the Pteroini using stochastic mapping is shown in Figure S2 and summarized in Figure 2 .

Based on the color patterns on the head, six morphological characters, and phylogenetic analyses, the taxonomy of the two specious genera Dendrochirus and Pterois needed revisions. Pterois is divided into clades I and II. The type species of Pterois is in the Pterois I clade, where the generic name is attached, and the genus Pteropterus Swainson, 1839, is revalidated for the Pterois II clade. Dendrochirus is divided into clades I, II, and III. The type species of Dendrochirus is in the Dendrochirus I clade, where the generic name is attached. We propose a new genus, Neochirus gen. n., for the Dendrochirus II clade. Nemapterois Fowler, 1938, is revalidated for Dendrochirus III. Diagnoses of the new, two revalidated, and two revised genera are described, and keys to eight pteroin genera are provided. To avoid ambiguous readings of taxonomic status in the following sections, the genus name generally follows the taxonomic treatment herein unless otherwise noted.

Neochirus gen. n.

Type species. Dendrochirus brachypterus (Cuvier and Valenciennes, 1829)

Etymology. From the Greek nέας (neo, means new), and χϵίρ (cheír, means finger, hand), a reference to the split from Dendrochirus. Feminine.

Diagnosis. The new genus, Neochirus gen. n., can be distinguished from other pteroin genera by having head color pattern IV ( Figure 2 ), a pair of short barbels on the tip of the snout, soft-rayed dorsal fin without ocellated spots, pectoral-fin rays 17–19, upper pectoral-fin rays not free from membrane ( Figure 4D ), anal-fin III, longitudinal scale rows less than 65, maxilla not covered by scales, mandible without spinous ridges, coronal and parietal spines separated, parietal spine not elongated throughout life.

Included species. Dendrochirus barberi (Steindachner, 1900), D. bellus (Jordan and Hubbs, 1925), D. brachypterus, D. hemprichi Matsunuma, Motomura, and Bogorodsky, 2017, and D. tuamotuensis Matsunuma and Motomura, 2013.

Dendrochirus Swainson, 1839

Type species. Dendrochirus zebra (Cuvier and Valenciennes, 1829)

Diagnosis. Distinguished from other pteroin genera by head color pattern III ( Figure 2 ), three short barbels on the tip of the snout, a soft-rayed portion of the dorsal fin without ocellated spots, upper pectoral-fin rays free from membrane pectoral-fin rays more than 15, anal-fin III, longitudinal scale rows less than 65, maxilla not covered by scales, mandible without spinous ridges, coronal and parietal spines separated, and parietal spine not elongated throughout life.

Included species. Dendrochirus koyo Matsunuma and Motomura, 2019, D. zebra.

Nemapterois Fowler, 1938

Type species. Nemapterois biocellatus Fowler, 1938

Diagnosis. Distinguished from all other Pteroini by head color pattern VII ( Figure 2 ), a pair of short barbels on the tip of the snout, a pair of preorbital barbel extraordinary long, soft-rays of dorsal fin with two large black ocelli, anal-fin III, maxilla not covered by scales, mandible without spinous ridges, coronal and parietal spines separated, parietal spine not elongated throughout life.

Included species. Dendrochirus biocellatus.

Pterois Oken, 1817

Type species. Pterois volitans Linnaeus, 1758

Diagnosis. distinguished from other pteroin genera by head color pattern I ( Figure 2 ), a pair of short barbels on the tip of the snout, soft-rays of the dorsal fin without ocellated spots, pectoral-fin rays less than 15 and upper pectoral-fin rays free from membrane ( Figure 4A ) through life, anal-fin III; longitudinal scale rows more than 65, maxilla not covered by scales, mandible without spinous ridges, coronal and parietal spines separated, parietal spine not elongated throughout life.

Included species. Pterois andover Allen and Erdmann, 2008, P. longicauda Swainson, 1839, P. lunulata Temminck and Schlegel, 1843, P. miles (Bennett, 1828), P. russelii Bennett, 1831 and P. volitans.

Pteropterus Swainson, 1839

Type species. Pterois radiata Cuvier and Valenciennes, 1829

Diagnosis. Distinguished from other pteroin genera by head color pattern II ( Figure 2 ), three short barbel on the tip of the snout, soft-rays of dorsal fin without ocellated spots, pectoral-fin rays more than 15 and upper pectoral-fin rays free from membrane ( Figure 4B ) through life, anal-fin III; longitudinal scale rows less than 65, maxilla not covered by scales, mandible without spinous ridges, coronal and parietal spines separated, parietal spine not elongated throughout life.

Included species. Pterois antennata (Bloch, 1787), P. brevipectoralis (Mandrytsa, 2002), P. cincta Rüppell, 1838, P. mombasae (Smith, 1957), P. paucispinula Matsunuma and Motomura, 2015, P. radiata, P. sphex Jordan and Evermann, 1903.

Keys to eight pteroin genera.

1a. No short barbels on snout tip....................................................2

1b. 2 or 3 short barbels on snout tip..............................................3

2a. Anal-fin III; maxilla covered by scales; mandible with spinous ridge; Head color pattern VI.................................... Brachypterois

2b. Anal-fin II; maxilla not covered by scales; mandible without spinous ridges; head color pattern V......................Parapterois

3a. Soft-rayed portion of terminal dorsal fin without two large black ocelli.......................................................................................4

3b. Soft-rayed portion of terminal dorsal fin with two large black ocelli; a pair of long preorbital barbels; head color pattern VII..................................................................Nemapterois

4a. Pectoral fin ray more than 15; longitudinal scales less than 65.............................................................................................5

4b. Pectoral fin ray less than 15; longitudinal scales more than 65; head color pattern I......................................................Pterois

5a. Coronal and parietal spines separated; parietal spine not elongated throughout life.............................................................6

5b. Coronal and parietal spines continuous; parietal spine elevated as a thin bony crest in adult males; head color pattern V................................................................................Ebosia

6a. Upper pectoral-fin rays not free from membrane.................7

6b. Upper pectoral-fin rays free from membrane throughout life, head color pattern II...........................................Pteropterus

7a. 3 short barbels on snout tip; head color pattern III............................................................................... Dendrochirus

7b. 2 short barbels on snout tip; Head color pattern IV.........................................................................Neochirus gen. n.