Examination of specimens listed in the original description of Nothobranchius lucius revealed that they belong to two species. Nothobranchius lucius is redescribed based on the type series and other specimens collected in the type locality area, the Kibasira Swamp area, Rufiji River basin, Tanzania, at elevations between 250 and 300 m. Nothobranchius insularis sp. n. is described on the basis of specimens collected in the north part of Mafia Island, Tanzania, at about 10–15 m elevation. Nothobranchius lucius and N. insularis are considered closely related species based on the shared presence of long jaws; caudal fin, in males, with a broad dark grey to black band on its posterior margin; black dots over the whole flank, in females; metapterygoid curved, with its middle and dorsal portions separated from the sympletic by a broad interspace; and posterior process of the quadrate shorter than the ventral length of the quadrate without process. Characters useful to distinguish them include premaxillary dentition, caudal fin shape, colour pattern of flank and unpaired fins in females, fin length, and number of neuromasts of the posterior section of the anterior supraorbital series. Both species are members of a group that also includes N. elongatus, N. hengstleri, N. interruptus, N. jubbi, N. krammeri, and N. melanospilus, which are all diagnosed by the presence of two neuromasts in the anterior section of the anterior supraorbital series.

Biodiversity hotspot, East African coastal forests, Nothobranchius melanospilus species group, Rufiji River, systematics, taxonomy

The East Africa region comprising the coastal forests of south-eastern Kenya and eastern Tanzania is an important centre of biological diversity often known as the East Africa biodiversity hotspot (EABH), with high occurrence of endemic species (Myers et al. 2000, Azeria et al. 2007). This region exhibits the greatest species diversity of Nothobranchius Peters, 1868, a killifish genus that occurs in a wide geographical portion of central and eastern Africa (e.g., Parenti 1981). Among the about 60 valid species in the genus, 27 have been recorded in EABH (e.g., Costa 2009, Dorn et al. 2014). Species of Nothobranchius live in temporary pools formed during the rainy season, where resistant eggs undergo diapause during the dry season (e.g., Polacik et al. 2011, Pinceel et al. 2015). In EABH, there are two rainy seasons, the ‘long rains’ (Masika), from March to May, and the ‘short rains’ (Muli), from October to December, which alternate with dry periods when pools and some streams completely disappear, as described first by Fitzgerald (1898). Species of Nothobranchius and other African and South American aplocheiloid killifishes possessing this uncommon life cycle have been equivocally called annual fishes, reflecting the past belief that these fishes have a single generation per year (e.g., Myers 1942). Aplocheiloid killifishes with this life cycle have been alternatively known as seasonal killifishes in more recent studies (e.g., Costa 2002).

Species of Nothobranchius have been recently proposed as model organisms for ageing processes since they are naturally short-lived and easily bred in laboratories (e.g., Genade et al. 2005, Harel et al. 2015). However, they are mainly known as aquarium fishes due to the striking colouration exhibited by males of most species in the genus. Consequently, in the last four decades many collecting trips in EABH have been reported in the aquarium fish literature (e.g., Seegers 1997, Wildekamp 2004; Neumann 2008), which resulted in several new taxonomic records and descriptions of new species.

In a brief revision of the N. melanospilus species group, Wildekamp et al. (2009) described N. lucius Wildekamp, Shidlovskiy & Watters, 2009 on the basis of material collected in the Kilombero River drainage, Tanzania, which had been formerly identified as N. aff. melanospilus in the aquarium fish literature (e.g., Wildekamp 2004). In the original description of N. lucius, specimens from the Mafia Island, in Tanzania, previously identified as N. melanospilus (Pfeffer, 1896) by Wildekamp (2004), were listed as additional non-type material belonging to N. lucius. The N. melanospilus species group was tentatively diagnosed by the presence of dark dots on the flank in females (i.e., black in N. melanospilus, dark brown in N. lucius and pale grey in N. makondorum Wildekamp, Shidlovskiy & Watters, 2009, according to Wildekamp et al. 2009). Subsequently, monophyly of this species group was supported in a multigene phylogeny (Dorn et al. 2014). All specimens listed by Wildekamp et al. (2009) is deposited in the Muséé Royal de l’Afrique Centrale (MRAC), Tervuren, Belgium.

While studying aplocheilid collections deposited in MRAC, discrepancies were found between data obtained by the current author from that listed by Wildekamp et al. (2009) from the same specimens. Consequently, a redescription of N. lucius is warranted and presented here. Comparisons also indicated that the populations from the Mafia Island, an off-shore island in the coast-line of Tanzania, belong to a distinct, new species, which is herein described.

All material examined is deposited in Musée Royal de l’Afrique Central, Tervuren (MRAC). Morphometric and meristic data were taken following Costa (1988), except for the snout length, measured between the anterior margin of the orbit and the anterior extremity of the middle portion of the upper jaw; measurements are presented as percent of standard length (SL), except for those related to head morphology, which are expressed as percent of head length. Specimens with deformed body were not measured to avoid unnecessary error. Fin-ray counts include all elements. Osteological preparations (C&S) were made according to Taylor and Van Dyke (1985). Terminology for frontal squamation follows Hoedeman (1958) and for cephalic neuromast series Costa (2001). Species were delimited using unique combinations of character states (diagnosability criterion; e.g., Davis and Nixon 1992).

Wildekamp et al. (2009) designated the type series of N. lucius containing the holotype and 22 paratypes that were collected in the Kilombero River drainage. An additional 18 non-type specimens were also listed. However, the table of morphometric and meristic data included in that paper indicates a total of 53 specimens of N. lucius examined, not 41 as indicated above. Such discrepancy supports the idea that the number of specimens used in the description was larger than that reported in the type material list, even when adding specimens listed under the non-types section that also includes material herein described as N. insularis. It is remarkable to notice that the type series of N. lucius includes some specimens that are poorly preserved (e.g. part of MRAC A7-02-P-10-25). These latter specimens have deformed body and damaged fins. Interestingly, other non-type specimens that are listed from the same drainage are well preserved (e.g. MRAC A7-02-P-26-27, 28-32).

Herein I speculate that measurements derived from poorly preserved specimens contributes to the variation presented by Wildekamp et al. (2009), which exceeds the variation reported herein based solely on the well preserved specimens reported by these authors. In addition, I have found discrepancies in what they report for some meristic characters, which are not subject to misinterpretation due to poor preservation. For example, Wildekamp et al. (2009) report the following for N. lucius: dorsal-fin rays 14–17, anal-fin rays 15–19, longitudinal series scales 31–35, and transverse series scales 12–14. Herein, however, I document that dorsal-fin rays range from 15–17, never 14; anal-fin rays range from 16–18, never 15 nor 19; longitudinal series scales range from 31–33, never 34 nor 35; and transverse series scales were 9–10, never 12–14. Lower fin-ray counts reported by Wildekamp et al. (2009) may be due to the first ray being often minute and embedded by thick epithelial tissue, but other differences noted above can not be explained.

Wildekamp et al. (2009: 238) distinguished N. lucius from other morphologically similar congeners by it having a more slender body and stronger dentition. When diagnosing N. lucius, Wildekamp et al. (2009: 247) distinguished this species from N. melanospilus and N. makondorum by the longer head, longer snout and more slender body in males. Morphometric values presented in that study, however, were notably overlapped among the three species, indicating that these characters are not useful to effectively distinguish species of the N. melanospilus group. I confirm this conclusion in the present study. The present comparison of jaw dentition in species of the N. melanospilus group also indicates that the dentition of N. lucius and N. insularis is not stronger than in other species, but in fact, N. lucius and N. insularis differs from other species of the group by both having more elongated jaw bones, which often yields a pike-like appearance in lateral view. In N. lucius and N. insularis the length between the anterior border of the dentary and the posterior tip of the angulo-articular is longer than the length between the anterior tip of the quadrate and the posterior margin of the preopercle (Fig. 4A; vs. shorter in the remaining species of the N. melanospilus group, Fig. 4B).

Figure 4. 

Jaws, jaw suspensorium and opercular apparatus, left side, lateral view, of: A. Nothobranchius lucius, paratype, male, MRAC A7-02-P-10-25, 43.0 mm SL; B. N. melanospilus, male, UFRJ 6591, 43.5 mm SL. Larger stippling indicates cartilage. Scale bar = 1 mm.

According to Wildekamp et al. (2009: 238), N. lucius may be distinguished from other species of the N. melanospilus group by two colouration characters: “a denser pattern of dark brown (not black) spots on the body, head and all fins of the females” (vs. black spots on the flank and base of the unpaired fins in N. melanospilus and grey spots on the flank in N. makondorum), and “a wide black margin and orange-red submargin to the caudal fin” in males (vs. caudal fin red with a narrow black margin). However, the first diagnostic character does not correspond to the figure of a live female of N. lucius from the type locality that is included in that same paper. The female in the figure shows that black dots, not brown, cover the whole flank, the side of the head, and the unpaired fins (Wildekamp et al. 2009: 248, fig. 9). In preserved females of N. lucius here examined, dots are dark brown to black, rounded, and distributed on the whole flank, side of the head, a large portion of the unpaired fins, and the basal portion of the pectoral fin; the dots of the flank are mainly arranged in horizontal rows (Fig. 1C). The colour pattern documented here for N. lucius greatly differs from the colour pattern exhibited by preserved females of N. insularis. In this species, dark grey to black dots are vertically elongated and mainly arranged in oblique rows on the flank, being rare or absent on the side of the head, and are restricted to the basal portion of the unpaired fins, while being absent on the pectoral fin (Fig. 2C). The occurrence of this female colour pattern in live exemplars of N. insularis is confirmed in a photograph of a live female, published in an aquarium journal paper that is a report of a collecting trip in the Mafia Island (Nagy 2009: 154).

Wildekamp et al. (2009) first diagnosed the N. melanospilus group, comprising N. lucius, N. makondorum and N. melanospilus, on the basis of the presence of dark dots on the flank of females. However, examination of the material listed by Wildekamp et al. (2009) as belonging to N. makondorum revealed that only female specimens from the type locality area (i.e., Ruvuma River basin, southern Tanzania) have grey dots on the flank, whereas specimens from all other areas to south do not show dark pigmentation. Although this group is corroborated by molecular data (Dorn et al. 2014), no unique morphological character was found to be useful to unambiguously diagnose the N. melanospilus group as delimited by Wildekamp et al. (2009). On the other hand, a more inclusive clade also comprising N. elongatus Wildekamp, 1982, N. hengstleri Valdesalici, 2007, N. insularis, N. interruptus Wildekamp & Berkenkamp, 1979, N. jubbi Wildekamp & Berkenkamp, 1979, and N. krammeri Valdesalici, 2008, in adittion to N. lucius, N. makondorum and N. melanospilus, is here diagnosed by the presence of two neuromasts in the anterior section of the anterior supraorbital series (Fig. 5). In other nothobranchines, the anterior section contains a single neuromast. The N. melanospilus group as herein delimited is also supported by molecular data (Wildekamp et al. 2009, Dorn et al. 2014). Therefore, I propose to include all the eight species listed above in the N. melanospilus group.

Figure 5. 

Diagrammatic representation of the latero-sensory system on the dorsal surface of the head in: A. Nothobranchius lucius, MRAC A7-02-P-28-32, male, 43.6 mm SL; B. Nothobranchius insularis sp. n., MRAC A702-P-38-44, paratype, male, 48.3 mm SL. arn – anterior rostral neuromast; asass – anterior section of the anterior supraorbital series; prn – posterior rostral neuromast; psass – posterior section of the anterior supraorbital series; pss – posterior supraorbital series.

Monophyly of the group comprising N. lucius and N. insularis is supported by derived characters states used to distinguish them from other congeners, including the presence of dark dots on the whole flank in females and the relatively long jaws, as discussed above. The present study also indicates that two character states of the jaw suspensorium are uniquely found in N. lucius and N. insularis: the metapterygoid is curved and its middle and dorsal portions are separated from the sympletic by a broad interspace (vs. nearly straight and in close proximity or in contact with sympletic) and the posterior process of the quadrate is short, its length shorter than the ventral length of the quadrate without process (vs. approximately equal or greater) (Fig. 4). Among species of the N. melanospilus group, N. lucius is unique in having an inner row of teeth directed inside the mouth on the median portion of premaxilla that are longer than the teeth of the external row (Figs 4 and 6).

Figure 6. 

Left isolated premaxilla, dorsal view, ascending process inclined backwards, of: A. Nothobranchius lucius, paratype, male, MRAC A7-02-P-10-25, 43.0 mm SL. B. Nothobranchius insularis sp. n., paratype, male, MRAC A702-P-38-44, 43.7 mm SL. Scale bar = 1 mm.

This study was developed during a six-month stay at Muséé Royal de l’Afrique Centrale, Tervuren, Belgium (MRAC), funded by CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico - Ministério de Ciência e Tecnologia, Brazilian Federal Government, grant process 200627/2015-5). I am especially grateful to Jos Snoecks for providing me the necessary conditions to study African killifishes at MRAC. Thanks are also due to Miguël Parrent, Baudouin Mafuta and Emmanuel Abwe for technical assistance. The manuscript befitted from the valuable criticisms provided by Emmanuel Vreven and Thomas Philbrick, and to Peter Bartsch for editorial support.