Specimens found between 1985 and 1988 in the Magé River Basin, south-eastern Brazil were misidentified as L. splendens. The recent rediscovery of other specimens in the Estrela River Basin near the type locality of L. splendens has clarified the species’ concept, making it possible to recognise the Magé River Basin specimens as a new species. The new species is herein described as Leptopanchax sanguineus sp. nov. and is distinguished from all other cynopoecilines by a unique colour pattern in males, including red bars with sinuous margins. It was collected in a well-preserved, temporary shallow swampy area within dense moist forest, but since 1990 the species has not been found again. Leptopanchax sanguineus sp. nov. is one of three species of cynopoeciline killifishes living in lowland moist forests of the coastal plains of Rio de Janeiro State, where the greatest diversity of endemic cynopoecilines is concentrated. Each of these species has been recorded a single time in the last 30 years, a surprisingly low record attributable to intense deforestation during the last several decades resulting in small fragmented lowland moist forests of today. This study indicates that seasonal killifishes adapted to uniquely live in this kind of habitat should be regarded with special concern in studies evaluating conservation priorities.

Biodiversity, conservation, moist tropical forest, systematics, taxonomy

The Atlantic Forest of south-eastern Brazil encompasses one of the most species-rich biota in the world, with a high diversity of plants and animals (Myers et al. 2000). Although the greatest part of the original Atlantic Forest was extirpated in the last three centuries and consequently several endemic species became endangered or even extinct, new species are still being recognised in recent years (Tabarelli et al. 2005; Costa 2016a). Aplocheilid killifishes are represented in the Atlantic Forest by 14 genera and over 45 valid species (Costa 2009, 2014, 2016b; Costa and Amorim 2013, 2014; Costa et al. 2014), of which six genera and 18 species belong to the tribe Cynopoecilini (Costa 2008, 2016b; Ferrer et al. 2014). Like several other South American and African aplocheiloids, cynopoecilines are seasonal killifishes, which uniquely have their entire life cycle restricted to temporary pools and swamps formed during rainy seasons (Myers 1942; Costa 2002a, 2009).

The greatest species diversity of cynopoecilines is concentrated in the coastal plains of Rio de Janeiro State, south-eastern Brazil (i.e., eight valid endemic species in three genera, of which two genera are endemic), with most taxa consisting of miniature species not surpassing 25 mm standard length (SL) and exhibiting high diversification of morphological traits (Costa 2008, 2016a). Different kinds of vegetation formations sheltering distinct seasonal killifish habitats are present in this region, including temporary pools in seasonally dry forests and coastal restingas, and seasonal swamps in dense moist forests (Costa 1995, 2009, 2016a). This region also contains the greatest occurrence of cynopoeciline species threatened with extinction in South America, some of them critically endangered or presumably extinct (Costa 2002b, 2009, 2012), with most taxa poorly represented in ichthyological collections and not collected in recent years (Costa 2016a).

Among areas of endemism for seasonal killifishes in the Rio de Janeiro coastal plains is the area encompassing river basins draining the southern flank of the coastal mountain range, Serra do Mar, and flowing into the Baía de Guanabara (i.e. Guanabara Bay area in Costa 2009, 2012; hereafter GBA). Three species of seasonal killifishes have been reported to occur in the GBA: Leptolebias marmoratus (Ladiges, 1934), Leptopanchax opalescens (Myers, 1942) and Leptopanchax splendens (Myers, 1942). This area was formerly occupied by a dense moist forest, but presently, after over 60 years of intense deforestation, the forest is restricted to small enclaves within urban areas. All three species were primarily recorded from the Estrela River Basin, in the western portion of the GBA (Myers 1942). Whereas L. opalescens was found in open vegetation habitats close to the forest border, L. marmoratus and L. splendens were found only within the dense moist forest (Myers 1942; Costa et al. 1988; Costa 2009). All three species were considered extinct after extirpation of known habitats in the 1950s, but L. marmoratus and L. opalescens were rediscovered some decades after, in neighbouring basins of western GBA (Costa 2002c, 2013). Costa and Lacerda (1988) redescribed L. splendens on the basis of old collections deposited in museums. They also provided colour photographs of a male tentatively identified as L. splendens and its habitat, consisting of a well-preserved forest area in the Magé River Basin, eastern portion of the GBA. However, the recent rediscovery of a population of L. splendens near its type locality in the Rio Estrela basin, over 60 years after its last record, has shown that this species is probably endemic to the Estrela River Basin in the western part of the GBA (Costa et al. 2019) and is not conspecific with specimens from the Magé River Basin in the eastern part of the GBA. Specimens of both populations differ greatly in several morphological characters, including fin ray, scale and vertebra counts, extent and relative position of fins, presence of filamentous rays on unpaired fins, presence of contact organs on male pectoral fin, presence of dermosphenotic bone, and male colour pattern. The new species from the Magé River Basin, which was misidentified as L. splendens in the last three decades (Costa and Lacerda 1988; Costa 2002b, 2009), is herein described.

The description of the new species was based on specimens collected over 30 years ago and then preserved for study, deposited in the ichthyological collections of the Museu Nacional, Universidade Federal do Rio de Janeiro, Rio de Janeiro (MNRJ) and Museu de Zoologia, Universidade de São Paulo, São Paulo (MZUSP). Colouration characters were analysed and described based on photographs taken from a male collected in 1985 (published in Costa and Lacerda 1988: fig.1) and notes taken from direct observation in aquaria of live specimens collected in 1985 and 1987; colouration characters were also checked in photographs of live specimens born in aquaria, published in Costa (1995: fig. 113) and Seegers (2000: fig. S31853-4). Since no new photographs are available, a coloured pencil drawing based on available material and rigorously following fish proportions and colours was made to illustrate the new species. Measurements and counts follow Costa (1988). Measurements are presented as percentages of SL, except for those related to head morphology, which are expressed as percentages of head length. Fin ray counts include all elements. Osteological data were taken from cleared and stained specimens prepared following Taylor and Van Dyke’s (1985) protocol; the abbreviation C&S indicates specimens prepared for osteological observation and preserved in glycerine. Terminology for frontal squamation followed Hoedeman (1958), and for the cephalic neuromast series followed Costa (2001). Comparative material is listed in Costa (2016a) and Costa et al. (2019). The map illustrating species distribution was generated using QGIS Geographic Information System, Open Source Geospatial Foundation Project.

Morphological data of L. sanguineus, then identified as L. splendens, were used in a phylogenetic analysis of the Cynopoecilini (Costa 2016a), justifying its inclusion in the genus Leptopanchax (Costa 2016b). Leptopanchax sanguineus shares with other congeners a golden distal stripe on the dorsal fin in males and vermiculate marks on the caudal fin in males (Fig. 2A). These apomorphic character states are present in all species placed by Costa (2016b) in Leptopanchax, including L. citrinipinnis, the type species of the genus (Fig. 2C–F), although the golden stripe on the dorsal fin is not always well delimited and vermiculate marks on the caudal fin may acquire different shapes. A third synapomorphy described by Costa (2016b) to diagnose Leptopanchax is the presence of an elliptical caudal in males, longer than deep (Fig. 2C–F), a condition that is not clearly attributed to L. sanguineus, in which the caudal fin is shorter and its posterior margin is slightly pointed (Fig. 2A). Among species of Leptopanchax, L. sanguineus was considered by Costa (2016b) to be closer to L. aureoguttatus (Cruz, 1974) and L. itanhaensis (Costa, 2008), which also have a horizontal red stripe between the orbit and the preopercle in males (Fig. 2A, C, D). Leptopanchax sanguineus differs from all other cynopoecilines by having red bars on the male flank that are wider than the interspace and have sinuous margins, and the presence of a small red spot on the posterior portion of the iris (Fig. 2A).

The recent rediscovery of L. splendens poses some incongruence in its positioning. This species also has a red stripe between the orbit and the preopercle and shares with L. sanguineus the presence of red bars on the whole flank and blue iris in males (Fig. 2B), but it does not have the character states diagnostic for Leptopanchax. In L. splendens, there is no broad golden distal stripe on the male dorsal, but a narrow bluish white stripe, and there are no vermiculate marks on the male caudal fin, which is pale orange, with a few pale red bars on the basal region (Fig. 2B). On the other hand, the series of narrow straight red bars on the male flank in L. splendens (Fig. 2B) may be non-homologous to the irregular broad red bars in L. sanguineus (Fig. 2A), but similar to the narrow straight bars present on the anterior portion of the male flank in Mucurilebias leitaoi (Cruz & Peixoto, 1991), which also exhibits a narrow white stripe on the dorsal fin (Cruz and Peixoto 1991; Costa 2014). Therefore, based on available data, the phylogenetic position of L. splendens remains unclear, and possibly it is not closely related to L. sanguineus, although both species are endemic to neighbouring river basins (Fig. 3). The uncertain phylogenetic position of L. splendens is also reinforced by the presence of a dermosphenotic bone and male contact organs on the male pectoral fin, which are plesiomorphic conditions for cynopoecilines. These two plesiomorphic character states occur in species of the genera Mucurilebias Costa, 2014 and Notholebias Costa, 2008, but not in species of the clade comprising Leptopanchax, Campellolebias Vaz-Ferreira & Sierra, 1974, Cynopoecilus Regan, 1912, and Leptolebias Myers, 1952, in which the dermosphenotic and pectoral-fin contact organs are always absent (Costa 2016a).

Leptopanchax sanguineus, known from a single locality, is a typical moist-forest species, an ecological adaptation considered to have arisen three times independently among cynopoecilines (Costa 2016a). Presently, only six cynopoeciline species are adapted to live in this kind of habitat: Cynopoecilus notabilis Ferrer, Wingert & Malabarba, 2014, Leptolebias marmoratus, Leptop. aureoguttatus, Leptop. itanhaensis, Leptop. sanguineus, and Leptop. splendens. Records of cynopoeciline forest-dwellers are extremely rare when compared with records of cynopoeciline species found in open vegetation habitats. For example, in southern Brazil, species of Cynopoecilus, subgenus Cynopoecilus, that are always found in open vegetation habitats, are frequently sampled in field studies (Costa 2002c; Costa et al. 2016). Contrastingly, the only species of the subgenus Poecilopanchax Costa, 2016, C. notabilis, is known from a single locality (Ferrer et al. 2014). Similarly in south-eastern Brazil, where a rich species diversity of cynopoecilines is concentrated (Costa 2009), species of the genus Notholebias, which are found in open vegetation habitats, are known from several localities (Costa 1988) and are often recorded in the literature (e.g. Costa and Amorim 2013), whereas each of the three species endemic to the BGA adapted to life within dense forests, Leptol. marmoratus, Leptop. sanguineus and Leptop. splendens, were recorded a single time in the last 30 years of continuous field studies directed to killifish habitats in the region (Costa and Lacerda 1988; Costa 2002c; present study). Actually, this discrepancy in field records is associated with the intense deforestation process in the Atlantic Forest (Costa 2008). Several reserves have protected moist and semi-deciduous forests along the coastal mountain range (Serra do Mar), but lowland moist forests, where killifishes are found, were greatly extirpated during the last two centuries and are presently represented by rare small fragments. Therefore, killifish species uniquely inhabiting lowland moist forests should be regarded with special concern in evaluation studies for conservation priorities.

Thanks are due to M. Britto and C. Moreira for hospitality and assistance during visits to their institution, to J. L. Mattos for photographing L. splendens, P. F. Amorim for assistance in the laboratory, and to M. Moutinho for supporting several field studies in the Campo Escoteiro Geraldo Hugo Nunes. A former version of the manuscript befitted from the criticisms provided by P. Buckup and J. Maldonado. This study was supported by CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico, Ministério de Ciência e Tecnologia, grant number 307349/2015-2) and FAPERJ (Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro, grant number E-26/202.954/2017).