Phylogenetic relationships of the neon tetras Paracheirodon spp. (Characiformes: Characidae: Stethaprioninae), including comments on Petitella georgiae and Hemigrammus bleheri

Neon tetras (Paracheirodon spp.) are three colorful characid species with a complicated taxonomic history, and relationships among the species are poorly known. Molecular data resolved the relationships among the three neon tetras, and strongly supported monophyly of the genus and its sister taxon relationship to Brittanichthys. Additionally, the sister-taxon relationship of the rummy-nose tetras Hemigrammus bleheri and Petitella georgiae was strongly supported by molecular and morphological data. Therefore, we propose to transfer the rummy-nose tetras H. bleheri and H. rhodostomus to the genus Petitella. Furthermore, Petitella georgiae is likely to be a species complex comprised of at least two species.


INTRODUCTION
Among characiforms, Characidae is the most diverse Neotropical fish family, with 1,188 valid species, of which 206 were described in the last ten years (Fricke et al., 2020a). Most members of Characidae are small-sized fishes, under < 8 cm standard length (SL), and many are popular aquarium species commonly known as "tetras" (Mirande, 2019).
The genus Paracheirodon Géry, 1960 is comprised of three small, brilliantly colored neon tetra species from South America (Weitzman, Fink, 1983) which are popular in the aquarium trade. Paracheirodon axelrodi (Schultz, 1956) and P. simulans (Géry, 1960) occur in small streams and headwater tributaries of the Negro and Orinoco rivers (Weitzman, Fink, 1983;Marshall et al., 2011), while P. innesi (Myers, 1936) occurs in blackwater and clearwater streams of the Ucayali-Solimões and Purus rivers (Weitzman, Fink, 1983). Paracheirodon, thus, is an emblematic example of a group of Amazonian fishes that are distributed in a biogeographical region known as the "Central Blackwater Amazon" (Dagosta, de Pinna, 2019).
Historically, both P. innesi and P. simulans were described as species of the genus Hyphessobrycon Durbin, 1908(Myers, 1936Géry, 1960), while P. axelrodi was originally described as a species of the genus Cheirodon Girard, 1855(Schultz, 1956. Géry (1960) established the genus Paracheirodon -designating H. innesi as its type species -due to its morphological affinities with Cheirodon axelrodi, but differing from it by the presence of tricuspid uniserial premaxillary teeth. Consequently, until Weitzman, Fink (1983) performed a taxonomic review of the neon tetras, whereby all of the species were placed within the genus Paracheirodon, these three species were in three distinct genera (Cheirodon, Paracheirodon, and Hyphessobrycon) in two different characid subfamilies (Tetragonopterinae and Cheirodontinae).
The taxonomic review of Weitzman, Fink (1983) provided eight morphological synapomorphies to support the monophyly of Paracheirodon. However, relationships among the three species were not well established and the authors did not provide any hypothesis of phylogenetic relationships between Paracheirodon and other characid groups, due to a lack of phylogenetic informativeness of the morphological 3/11 scielo.br/ni | sbi.bio.br/ni data available at the time. They also provided one synapomorphy for the clade P. axelrodi and P. innesi, namely the dorsal placement of the lateral blue body stripe and its posterior termination near the base of the adipose fin, which in P. simulans reaches the caudal fin base.

MATERIAL AND METHODS
Study area and sample collection. Sampling of P. simulans was carried out in palm swamps in an interfluvial region of the middle Negro River, at the headwaters of Igarapé Tulia (0°40'0.12"S, 63°33'51.48"W), during 2009-2010 (see Marshall et al., 2011 for a complete field description). We also sequenced five Brittanichthys axelrodi specimens from Santa Isabel do Rio Negro (0°36'58" S, 64°55'24" W), due to its close phylogenetic relationship with the genus Paracheirodon (Javonillo et al., 2010;Mirande, 2019), as well as one individual of Petitella georgiae from the Purus River (6°22'30" S, 63°16'29"W), since Mirande (2019) suggests a close phylogenetic relationship of Petitella and Paracheirodon. Individuals were collected using small dip nets, then preserved in 95% ethanol while in the field, before being deposited posteriorly in the Universidade Federal do Amazonas animal tissue collection (CTGA) using individual ID tags (Tab. 1).
Molecular data of P. axelrodi, P. innesi, and other characids obtained from GenBank were also included for posterior data analysis (Tab. S1). All sequences generated in this study have been deposited in GenBank (Tab. 1).

Data analysis.
The nucleotide sequences were organized and verified using Geneious 6 (Kearse et al., 2012). The forward and reverse chromatogram reads for each sequenced sample were assembled into contigs and verified visually. The COI nucleotide sequences were also translated into putative amino acids; no internal stop codons were found. Additional COI and 16S rRNA data of P. axelrodi, P. innesi and other characids from the "clade C" of Javonillo et al. (2010) were also included. We also conducted a detailed search in GenBank for additional data of Stethaprioninae sensu Mirande (2019)   The phylogenetic reconstruction using the COI and 16S rRNA concatenated dataset recovered the monophyly of Paracheirodon with high bootstrap support (91%), where P. axelrodi and P. innesi figure as sister species, while P. simulans is sister to this clade (Fig. 1). The results also support Brittanichthys as the sister-group of Paracheirodon (98% bootstrap support). Petitella georgiae from Peru and the Purus River were sister taxa, albeit divergent (p-distance = 9.1%), and formed sister clade to Petitella bleheri (p-distance = 12.6%); this phylogenetic relationship was highly supported (100% bootstrap support). In contrast, we were unable to confirm the monophyly of the clade comprised of Paracheirodon, Brittanichthys, Petitella georgiae + Petitella bleheri, as proposed by Mirande (2019).  Diagnosis. The genus Petitella is readily distinguished from all remaining characid genera by the possession of a distinctively bright red head, the presence of a black horizontal bar that extends from the end of the caudal peduncle to the middle rays of the caudal-fin, and the presence of an oblique black bar in each caudal-fin lobe, separated by white colored bands. Contact between frontals anterior to frontal fontanel present; posterodorsal margin of ethmoid cartilage and lateral ethmoids distant from lateral ethmoids; 17 or fewer branched anal-fin rays; only one or two anal-fin hooks on each ray of adult males; the presence of parallel longitudinal ridges on the posterior field of scales; scales covering one-third of the length of caudal-fin lobes; coloration of the head distinctively red, especially the snout.
Petitella bleheri is distinguished from its congeners by the much more intense and widespread red color of the head, extending up to the humeral region (vs. limited red coloration and not extending to humeral region in P. georgiae and P. rhodostoma); horizontal black bar on the end of the caudal peduncle is never prolonged forward (vs. prolonged up to the anal-fin in P. georgiae and P. rhodostoma); anal-fin hyaline (vs. a black bar on the base of the anterior part of the anal-fin, continuing obliquely on the branched rays in P. georgiae and P. rhodostoma).
Petitella rhodostoma is distinguished from its congeners by the red head color not extending to the humeral region and the presence of a black spot on the lower posterior border of the caudal peduncle (vs. head color not extending to humeral region with only one black spot on caudal peduncle in P. georgiae, and head color extending to humeral with two black spots on caudal peduncle in P. bleheri); dentary with 5-6 teeth, with 5 cuspids, usually followed by 4 conical ones (vs. 9-11 teeth, with 5 cuspids in P. georgiae, and 6 teeth, with 6 or 7 cuspids, followed by 1 or 2 tricuspidate ones in P. bleheri).

DISCUSSION
Our results support the monophyly of Paracheirodon, proposed by Weitzman, Fink (1983) based on morphological data. In addition, the sister-species relationship of P. axelrodi and P. innesi was also confirmed, which had been suggested by the same authors based on the position and termination of the blue lateral body stripe in the vicinity of the adipose fin.
Following Mirande (2019), all of these genera and species share only one morphological synapomorphy: the pelvic-fin bony hooks absent in adult males of species bearing hooks on fins (see Appendix S8, node 988). However, in this study, we were unable to recover phylogenetic relationships other than the sister taxon relationship of Paracheirodon and Brittanichthys, due to a lack of phylogenetic information in the COI and 16S rRNA genes for deep phylogenetic nodes (Javonillo et al., 2010).
The phylogenetic reconstruction (Fig. 1) also confirms Petitella georgiae as a sistergroup of Petitella bleheri. The striking p-distance divergence (9.1%) between the Peruvian P. georgiae and our samples from the Purus River indicates the possibility of Petitella georgiae being a species complex. The monophyly of P. bleheri and P. georgiae was highly supported in our analyses -the first time Petitella georgiae was included in any molecular phylogeny -as well as by the morphological data of Mirande (2019), who reported seven morphological synapomorphies supporting the sister-taxon relationship between these two species (see Appendix S8, node 1129).
Petitella georgiae, Petitella bleheri and Petitella rhodostoma (Ahl, 1924) share a very similar coloration marked by an intensely bright red head and the presence of three conspicuous horizontal black bars on the caudal fin (Mirande, 2010). Although Géry, Mahnert (1986) compared the type material of H. bleheri to both H. rhodostomus and P. georgiae, they chose not to discuss either phylogenetic affinities or the validity of the genus Petitella itself, due to the "new weights being given to certain cranial characters in the tetras as proposed by certain anatomists", citing Weitzman, Fink (1983). Mirande (2010) stated that "Petitella georgiae Géry & Boutière is mainly distinguished from Hemigrammus bleheri by having only one row of premaxillary teeth (vs. two rows)".
Although P. rhodostoma was not included in our analyses nor in Mirande's (2019) matrices, there is little doubt that these three species are very closely related and likely form a monophyletic group. Only the rummy-nose tetras share a distinctive and intensely bright red head, which is of different color and pattern than the rest of the body -a pattern not observed in any other characid species (Mirande, 2019, Appendix S1, character 491). Given that both morphological and molecular data support monophyly of these species, and Hemigrammus Gill, 1858 has been demonstrated as 9/11 scielo.br/ni | sbi.bio.br/ni a non-monophyletic entity in multiple studies (Javonillo et al., 2010;Oliveira et al., 2011;Mirande, 2009Mirande, , 2010Mirande, , 2019 and in our study Hemigrammus unilineatus Gill, 1858, the type species of Hemigrammus, is sister to Moenkhausia hemigrammoides Géry, 1965, we therefore suggested the transfer of both H. bleheri and H. rhodostomus to the genus Petitella Géry & Boutière, 1964 and provided a tentative diagnosis for the genus based on original descriptions of the species and the morphological synapomorphies of Mirande (2019). Compelling evidence for the monophyly of the rummy-nose tetras has been published in the recent years using morphological (Lima, Souza, 2009;Mirande, 2010), molecular (this study) and total-evidence (Mirande, 2019) datasets.
By the inclusion of new taxa into characid phylogenies, it was possible to confirm previous hypotheses and propose new ones. However, it is also important to point out that there is still a huge knowledge gap regarding phylogenetic relationships of Characidae, which is likely to remain for some time. For example, of the 141 genera recognized by Mirande (2019), at least 40 do not have any molecular data available in GenBank (pers. obs.) and many genera and species remain incertae sedis. This is due to a lack of information beyond their original descriptions, which makes it difficult to infer phylogenetic relationships of these species and genera, which, in turn, guide higher-order classification.