^a Through the use of DNA sequences from 10 nuclear genes, previously never employed in phylogenetic analyses, Percophis brasiliensis was resolved, with strong support, as the sister lineage of notothenioids (Near et al. Reference Near, Dornburg, Harrington, Oliveira, Pietsch and Thacker2015). The support from morphology is minimal but not conflicting. Given the absence of any persuasive prior or alternative hypotheses for a sister group, this is definitely an advance.

^b Bravo et al. (Reference Bravo, Lloris, Pequeño and Rucabado1999) have placed Bovichtus elongatus and B. argentinus in the synonymy of Bovichtus chilensis. This paper appeared too late to be included in our 2000 list.

^c Although Balushkin (Reference Balushkin2016) has placed Bovichtus oculus in the synonymy of B. psychrolutes, Stewart (Reference Stewart and Roberts2015) continues to recognize B. oculus as valid, as do we. Bovichtus oculus is known from the holotype in the Museum of New Zealand Te Papa Tongarewa (NMNZ P. 018510), and from two other specimens in the same collection tentatively identified as B. oculus. Although Balushkin (Reference Balushkin2016, pp. 632 and 637) indicates that he examined a radiograph of this holotype and measured its eye diameter, it is doubtful that the specimen he examined was the holotype. Records held by the NMNZ provide no evidence that the specimen was ever radiographed or that it was ever sent out on loan. Bovichtus oculus has eyes with a horizontal eye diameter < 9.0 times in SL, the most important character in distinguishing it from other species of bovichtids including B. psychrolutes (Hardy Reference Hardy1988). Balushkin reports that in his radiograph this value was 9.2–9.4 times in SL, not the 8.2 in SL reported by Hardy in his table 1. He also states that his measurements place the holotype in the range of eye diameters for B. psychrolutes and therefore that it is this latter species. However, re-measurement of the NMNZ type specimen of B. oculus using two different calipers showed that the eye was a little smaller than recorded by Hardy, but still < 9 times in SL. The actual measurements were: bony eye diameter = 13.04 mm; SL = 109.15 mm = 8.37 in SL or 11.94% of SL (Andrew Stewart, Collection Manager, Fishes, NMNZ, personal communication to J.T. Eastman, 12 December 2016.) Given this information, we continue to recognize B. oculus as a valid species. It should also be noted that, based on information in Eschmeyer's Catalog of Fishes, the correct date for this species is 1989, not 1988 as indicated in our 2000 list, because the article was not published until 1989.

^d We follow Balushkin (Reference Balushkin1992, Reference Balushkin2000) in considering Cottoperca gobio (Günther 1861) as a junior synonym of C. trigloides (Forster 1801). Some South American ichthyologists continue to recognize the South American population as Cottoperca gobio (Günther 1861).

^e The spelling of this name has gone back and forth between Eleginopidae/Eleginopsidae. Sheiko (Reference Sheiko2019) recently resolved this matter by determining that the correct stem is Eleginop-, not Eleginops-, and therefore the correct name for the family is Eleginopidae.

^f Although Cryonotothenioidea bears the suffix of a superfamily, Near et al. (Reference Near, Dornburg, Harrington, Oliveira, Pietsch and Thacker2015, Reference Near, Macguigan, Parker, Struthers, Jones and Dornburg2018) have employed this and several other rank-free names for various notothenioid clades. Although rank-free names do not comply with the International Code of Zoological Nomenclature (Sheiko Reference Sheiko2019), cryonotothenioids has begun to appear in the literature as a convenient way of referring to the clade that radiated in Antarctica and that encompasses 91% (128/140) of notothenioid species. We think it is acceptable to use it informally with the understanding that it is a non-Code compliant, colloquial name for the five families of the Antarctic clade. Use of left («) and right (») guillemets or double chevrons (as «cryonotothenioids») indicates that the name is non-Code compliant (Sheiko Reference Sheiko2019).

^g Duhamel et al. (Reference Duhamel, Gasco and Davaine2005, pp. 327 and 334) are not convinced that there is sufficient evidence to prove that Gobionotothen barsukovi is distinct from G. acuta. Although Balushkin (Reference Balushkin2014) has supplemented his original description (Balushkin Reference Balushkin1991) of G. barsukovi with meristic data from additional specimens, there are no genetic data and we also remain skeptical about the validity of this species. A recent molecular phylogenetic study supported the validity of G. acuta and G. marionensis, but did not include specimens of G. barsukovi (Miya et al. Reference Miya, Gon, Mwale and Poulin2016).

^h Voskoboinikova & Kellermann (Reference Voskoboinikova and Kellermann1993) have described another species of Gvozdarus, G. balushkini, from the Weddell Sea based on a single 30 mm SL specimen. This seems premature to us given the dearth of information about G. svetovidovi, a species known from only two specimens.

ⁱ Using a combination of morphological and molecular genetic data, Dornburg et al. (Reference Dornburg, Federman, Eytan and Near2016a) determined that Lepidonotothen nudifrons consists of two cryptic species: L. nudifrons around South Georgia and the South Sandwich Islands, and L. cf. nudifrons in the area of the Antarctic Peninsula. Obviously, the cryptic species will not be recognized until it has been formally described and named. The genus of these species is now Nototheniops.

^j Patagonotothen supposedly contains cryptic species that have not been named (Ceballos et al. Reference Ceballos, Roesti, Matschiner, Fernández, Damerau, Hanel and Salzburger2019).

^k The correct name is now back to Pleuragramma antarcticum, not P. antarctica (Sheiko Reference Sheiko2019).

^l Until recently, there had never been any question about the validity of the sister species Trematomus lepidorhinus and T. loennbergii. However there have been reports of investigators being unable to distinguish specimens of these species morphologically or through the use of mitochondrial COI gene barcodes or nuclear gene DNA sequences (Lautrédou et al. Reference Lautrédou, Bonillo, Denys, Cruaud, Ozouf-Costaz, Lecointre and Dettai2010, Reference Lautrédou, Hisinger, Gallut, Cheng, Berkani and Ozouf-Costaz2012, Causse et al. Reference Causse, Ozouf-Costaz, Koubbi, Lamy, Eleaume, Dettai, Duhamel, Busson, Pruvost, Post, Beaman and Riddle2011, Dettai et al. Reference Dettai, Adamowizc, Allcock and Arango2011a, Reference Dettai, Lautrédou, Bonillo, Goimbault, Busson and Causse2011b, Smith et al. Reference Smith, Steinke, Dettai, Mcmillan, Welsford, Stewart and Ward2012). However, these species are distinguished by next-generation DNA sequencing (RADseq) (Near et al. Reference Near, Macguigan, Parker, Struthers, Jones and Dornburg2018). Karyological studies also support the existence of two species (Ghigliotti et al. Reference Ghigliotti, Cheng, Ozouf-Costaz, Vacchi and Pisano2015), as does morphological divergence of the sense organs and brains, and their preferred habitat depths (Eastman & La Mesa Reference Eastman and La Mesa2021). Therefore we continue to recognize these two species as valid and suggest that the difficulty in identifying at least some of the Indian Ocean specimens is attributable to phenotypic plasticity in the dimensions of the head in T. lepidorhinus (Eastman & La Mesa Reference Eastman and La Mesa2021).

^m Since 2000 five new species of Harpagifer have been described on the basis of being either the “soft” littoral form or deeper-dwelling “spiny” form at a given island (Prirodina Reference Prirodina2004). The primary morphological difference between the species in each pair is in the degree of head spination, especially the extent of development of the supraorbital ridge and its protuberances. Duhamel et al. (Reference Duhamel, Gasco and Davaine2005, p. 328, 358) question the validity of Harpagifer andriashevi Prirodina Reference Prirodina2000, H. nybelini Prirodina Reference Prirodina2002, H. crozetensis Prirodina Reference Prirodina2004 and H. macquariensis Prirodina Reference Prirodina2000. Although it was described after the publication of their book, Harpagifer permitini Neyelov & Prirodina Reference Neyelov and Prirodina2006 should probably added to this group. Duhamel et al. (Reference Duhamel, Gasco and Davaine2005) consider the species in this genus to be “extremely polymorphic” and caution that, based on their samples from the MNHN collections, meristic data do not discriminate the species. They also state that the diagnoses of the species of Harpagifer are also heavily reliant on the degree of development of the supraorbital protuberance, a highly subjective character showing intermediate degrees of development in species from coastal areas. This situation would obviously benefit from the perspective that molecular genetic data could provide.

ⁿ This is a recent change in the spelling of the name of this species. It incorporates the corrections of Koerber (Reference Koerber2009) with respect to the use of diacritical vowels and diphthongs in the scientific names of certain notothenioid species dedicated to the Swedish zoologist Einar Lönnberg. Koerber's corrections are accepted and included in Eschmeyer's Catalog of Fishes.

^o We view two additional species of Pogonophryne, P. minor Balushkin & Spodareva Reference Balushkin and Spodareva2013a and P. pallida Balushkin & Spodareva Reference Balushkin and Spodareva2015, as doubtful and they are not included in the list. See Eakin's critiques in Supplemental material 1 concerning these two species.

^p There are widely differing opinions about the number of species of Channichthys, a genus endemic to the Kerguelen Plateau. Iwami & Kock (Reference Iwami, Kock, Gon and Heemstra1990) and Kock (Reference Kock2005) recognize only C. rhinoceratus. Balushkin (Reference Balushkin2000) recognizes four species. Duhamel et al. (Reference Duhamel, Gasco and Davaine2005) recognize C. rhinoceratus, C. velifer and suggest the existence of a single undescribed species. In considering the species level taxonomy of Channichthys, Duhamel et al. (Reference Duhamel, Gasco and Davaine2005) state that (page 368, French to English translation using Google Translate, with [bracketed material] added by us):

”The systematics and biology of Channichthys require a complete overhaul because the original descriptions were very imprecise. Following the original description of C. rhinoceratus from the coastal region, another species, C. rugosus, [Channichthys rugosus Regan 1913, based on a single 400 mm specimen, with no drawing provided] was minimally described from the same area and then later placed in synonymy. The specimens subsequently captured on the plateau have led to the description of a profusion of new species that need to be reexamined to draw conclusions on their status and to correctly redefine C. rhinoceratus (as well as other species, valid nor not). In addition, the measurements carried out on individuals greatly depend on the condition of the specimen at the time of death (often gaping opercula and mouth open …) and only meristic characters can then be used. The report of C. rhinoceratus at Marion from an otolith in a gentoo penguin (P. papua) [stomach presumably?] has never been validated by the capture of other complete specimens. The initial descriptions of the larvae of C. rhinoceratus and C. velifer, then those of C. rhinoceratus and C. rugosus, add to the confusion because no series is complete and valid.”

Adding to the uncertainty about the number of species of Channichthys, Shandikov (Reference Shandikov2011) described Channichthys richardsoni and implied that, based on his previous work, there are a total of nine species of Channichthys around Kerguelen. Eschmeyer's Catalog of Fishes also indicates that there are nine valid names for species of Channichthys. Therefore, given the confusion over the number of species and the possible existence of phenotypic plasticity in C. rhinoceratus, we think it is best to maintain a conservative approach by recognizing only C. rhinoceratus until a comprehensive analysis, including genetic data, lends some clarity to this situation.

^q These two species were formerly in the genus Cryodraco. Sheiko (Reference Sheiko2019, p. 66) has determined that, according to the Principle of Priority, the name Pagetodes Richardson 1844 stands as the valid senior isonym of Cryodraco Dollo, 1900. Dornburg et al. (Reference Dornburg, Eytan, Federman, Pennington, Stewart, Jones and Near2016b) provide molecular data in support of previous morphological data (La Mesa et al. Reference La Mesa, Vacchi, Iwami and Eastman2002) documenting the validity of the two species of Pagetodes (formerly Cryodraco).

^r Sheiko (Reference Sheiko2019) has determined that the correct spelling is Pagetopsis maculata, not P. maculatus.