Neolamprologus

Members of the genus Neolamprologus (49 described species, all but one endemic to Lake Tanganyika) are highly diverse, and the genus is probably not monophyletic (Poll, 1986; Schelly et al., 2003; Aibara et al., 2005). Within the genus Neolamprologus , 10 described species are characterised by a lunate tail with long filaments, uncommon among lamprologines. These closely related species form the stenotopic lithophilic N. savoryi complex (Poll, 1978; Brichard, 1989), of which several members are popular with the aquarium trade (Konings, 1998) and which are of interest to behavioural biologists (Balshine-Earn et al., 1998). The 10 members of the N. savoryi complex are, in chronological order of description: N. savoryi (Poll, 1949) , N. brichardi (Poll, 1974) , N. pulcher (Poll, 1974) , the latter two first described as subspecies of N. savoryi (Trewavas & Poll, 1952) , N. splendens (Brichard, 1989) , N. olivaceous (Brichard, 1989) , N. gracilis (Brichard, 1989) , N. falcicula (Brichard, 1989) , N. crassus (Brichard, 1989) , N. marunguensis Büscher , 1989 and N. helianthus Büscher , 1997 . The species complex has been referred to as the N. brichardi complex (Konings, 1998), but because N. savoryi was described first (Poll, 1949), while N. brichardi was described later as a subspecies of N. savoryi (Trewavas & Poll, 1952) , we here refer to the complex as the N. savoryi species complex. Here we report on two new species belonging to the complex, both of which occur at the Bangwe peninsula on the east coast of Lake Tanganyika, 5 km south of Kigoma (Tanzania, Fig. 1).

Most species in the N. savoryi complex, apart from N. savoryi and N. brichardi , have been reported only from small areas within the lake. Neolamprologus splendens , N. gracilis , N. helianthus , N. marunguensis , N. olivaceous , and N. crassus were reported from small areas (between 7°15'S and 7°50'S) on the west coast of the lake (Fig. 1). The type localities given in the species descriptions are Cape Zongwe for N. splendens , Cape Kapampa for N. gracilis , the coast stretching 20 km north from the Lunangwa river for N. helianthus and Kapampa for N. marunguensis . The type localities of N. crassus and N. olivaceous are not fully clear. Brichard(1989) gives in his description “in and around the Bay of Luhanga” as the locality for both species, and mentions their sympatric occurrence. Luhanga is in the very north west of the lake, near Uvira, and is likely a misspelling for Lunangwa Bay. The type locality written on the holotype label of N. olivaceous is Lunangwa Bay, in the south west of the lake, and on those of the holotypes of N. crassus and N. gracilis , Masanza is given as the locality. Masanza is near Cape Kapampa, about 60 km north of Lunangwa Bay.

The type locality of N. falcicula is Magara, Burundi, in the north east of the lake. Neolamprologus savoryi and N. brichardi are the only species in the complex that occur almost lake-wide. Their type localities are Kigoma and Kisoje respectively, both on the Tanzanian east coast. The type locality of N. pulcher is Kasanga, in the south east ( Maréchal & Poll, 1991). Much of the coast line has not been extensively investigated, and there may be more undiscovered species within the complex.

The genus Neolamprologus is defined by the combination of the following characters. The presence of scales on chest, abdomen and anterior to origin of dorsal fin, first pelvic ray longest, 6-8 canines on upper jaw, dorsal fin 18-20 spines and 7-10 soft rays, anal fin 5-8 spines and 5-8 soft rays, 29-33 vertebrae, infraorbital bones absent (Poll, 1986, with adjustments as to fin ray and vertebrae counts).

The N. savoryi complex is defined by the combination of 30-40 scales in lateral line, less than 19 gill rakers, no molariform pharyngeal teeth, tail lunate, dorsal and anal fins and lobes of the caudal produced into long filaments, no scales on cheek, D.XVIII -XX, 7-10, A.V -VIII, 5-8 (Poll, 1978, with adjustments).

Key to the species of the Neolamprologus savoryi complex

1 Marks on operculum .................................................................................................................................... 2

- Opercular marks absent .............................................................................................................................. 8

2 V-shaped mark with bright spot on operculum,>35 scales in the longitudinal line .................................... 3

- Opercular mark not V-shaped, <35 scales in the longitudinal line ..............................................................4

3 Opercular spot red, scales on paired fins, ctenoid scales on anal and dorsal fins ...................... N. splendens

- Opercular spot yellow, no scales on paired fins, cycloid scales on anal and dorsal fins .......... N. helianthus

4 Twelve soft rays in pectoral fin, anal fin filament longer than of dorsal fin, headlength> 34 %SL, 3 scales between lateral lines, bars on body ................................................................................................ N. savoryi

- Thirteen soft rays in pectoral fin, anal fin filament shorter than of dorsal fin, headlength <34 %SL, 2 scales between lateral lines, no bars ............................................................................................................ 5

5 Cephalic pits, cheek depth> 28 % HL, opercular mark vague ..................................................... N. crassus

- Cephalic pits absent, cheek depth <28 % HL, opercular marks distinct .................................................... 6

6 No conspicuous spots on scales, no yellow border under eye, caudal peduncle length>18 %SL ................ .................................................................................................................................................... N. brichardi

- Conspicuous spots on scales, yellow border under eye (white in preserved specimens), caudal peduncle length <18 %SL ........................................................................................................................................... 7

7 Scales between pectoral fin and pelvic fin, and between dorsal fin origin and lateral line clearly visible, rows of spots on scales regular ..................................................................................................... N. pulcher

- Same scales deeply embedded and not always clearly visible, rows of spots on scales irregular, with interruptions..................................................................................................................................... N. olivaceous

8 No scales on occiput, few scales on nape .................................................................................................... 9

- Few scales on occiput, many scales on nape.............................................................................................10

9 Distinct black and white marginal bands on dorsal and caudal fin and striped pattern on unpaired fins, preorbital depth <17 % HL, eight to fifteen scales on operculum.................................................... N. walteri

- Faint markings and no distinct black and white marginal bands on unpaired fins, preorbital depth> 17 % HL, 12-30 scales on operculum.................................................................................................. N. falcicula

10 Scales around caudal peduncle> 17, gill rakers> 15, 3 scales between lateral lines.................. N. gracilis

- Scales around caudal peduncle <17, gill rakers <10, 2 scales between lateral lines...............................11

11 Body depth <29 % SL, headlength> 30 % SL, inter orbital width <25 % HL, caudal peduncle length/ depth ratio>1.25, no scales on unpaired fins and cycloid scales on dorsal and anal fins................... N. chitamwebwai

- Body depth> 29 % SL, headlength <30 % SL, inter orbital width> 25 % HL, caudal peduncle length/ depth ratio <1.10, scales on unpaired fins and ctenoid scales on dorsal and anal fins................. N. marunguensis

Discussion

Seehausen et al. (1998) noted in cichlids of the rocky shores of Lake Victoria that while sympatric congenerics differed in male coloration, less closely related sympatric cichlids often did not differ in male coloration. Seehausen et al. (1998) suggested that the differences in male coloration are needed to maintain reproductive isolation between close relatives. There is no sexual dimorphism in the species of the N. savoryi complex, and reproductive isolation between these sympatric close relatives is apparently maintained by a different mechanism. The fewer and less obvious differences between N. walteri and N. chitamwebwai compared with differences with N. savoryi and N. brichardi suggest that the former couple is more closely related. It seems plausible that in N. walteri and N. chitamwebwai ecological differentiation is key to the maintenance of reproductive isolation.

The two new species although closely related, show clear ecological differentiation. Due to habitat isolation, N. walteri and N. chitamwebwai did not occur in exactly the same areas, but were separated by no more than 4 (on the south side of Cape Bangwe) to several hundred metres (on the north side of Cape Bangwe where sandy beaches interrupt the rocky shores and neither species occur). The two new species can therefore be considered sympatric. Stable isotope data (13C, 15N) showed no overlap of diet between N. walteri , N. chitamwebwai and N. brichardi and little overlap of N. chitamwebwai with N. savoryi (Verburg & Hecky, unpublished data). 13C data of N. walteri were intermediate between the high values of N. chitamwebwai and N. savoryi (high 13C indicating a more benthic diet) and the low values of N. brichardi (low 13C indicating a more planktivore diet, Verburg & Hecky, unpublished data).

In sympatric fish species pairs often one will be a benthic feeder while the other specializes on pelagic food sources (Schliewen et al., 1994; Schluter, 2000). Ecological differentiation may be related to morphological differences between species in the complex. Differences in head measures and their allometric coefficients as found between N. walteri and N. chitamwebwai (Fig. 5B) may be related to the method of preyhandling (Wilhelm, 1984). As shown by Wilhelm (1984; compare Fig. 5B), it is not only variation in body size together with allometry that accounts for variation in head morphology between cichlid species. Both N. chitamwebwai and N. savoryi , the two species with the most benthic diet according to stable isotope data, have smaller cheek depths than the other species in the complex (Fig. 5B). The teeth in the posterior row on the lower pharyngeal bone of N. chitamwebwai (Fig. 7) while resembling those of N. savoryi (Poll, 1949) are slightly larger than of N. walteri , and especially N. brichardi (Trewavas & Poll, 1952) , possibly related to a more benthic diet of N. chitamwebwai and N. savoryi while N. brichardi feeds on zooplankton (Konings, 1998). In addition, N. brichardi has more gill rakers (11-16) than does N. walteri (6-11) and N. chitamwebwai (6-9), which agrees with a more pelagic diet of N. brichardi . However, several morphological differences found between N. chitamwebwai , N. walteri and N. brichardi contrast with what would be expected from a more benthic diet of N. chitamwebwai compared with N. walteri and N. brichardi . The number of gill rakers did not differ between N. chitamwebwai and N. walteri , and the small body depth and the longer and less steep ascending process of the premaxilla (Fig 6) of N. chitamwebwai are considered to be more typical for pelagic planktivorous fish (Day et al., 1994; Meyer, 1987). However, while the members of the N. savoryi complex may differ in the benthic versus pelagic proportions of their diets, they are all littoral species that do not migrate from the rocky shores, including the planktivore N. brichardi , and are rarely seen more than 1 m away from the bottom. Therefore interspecific morphological differences related to the way of feeding are probably less evident than those often seen between other species pairs of which one is benthic and the other truly pelagic (Robinson & Wilson, 1994).

Evidence of ecological differentiation between closely related sympatric cichlids is rare in literature. Closely related sympatric cichlids in the large African lakes (with literature mainly referring to Lake Victoria and Lake Malawi) have been considered to differ little in diet and habitat use, encouraging the idea that cichlids can coexist without niche partitioning (Coyne & Orr, 2004). Coyne & Orr (2004) suggested that studies of ecological differentiation in closely related sympatric cichlid species are badly needed to test this idea. While there have been several studies that showed ecological differentiation between closely related sympatric cichlids in Lake Victoria (Goldschmidt et al., 1990; Goldschmidt & Witte, 1990; Witte, 1984), our paper suggests a new opportunity to examine ecological differentiation in two new closely related sympatric cichlid species from Lake Tanganyika.

Neolamprologus is a genus of cichlids endemic to eastern Africa with all but one species, Neolamprologus devosi from the Malagarasi River, occurring in Lake Tanganyika. It is the largest genus of cichlids in Lake Tanganyika and also the largest genus in the tribe Lamprologini, which includes Altolamprologus, Chalinochromis, Julidochromis, Lamprologus, Lepidiolamprologus, Telmatochromis and Variabilichromis. The latter is a monotypic genus doubtfully distinct from Neolamprologus.

It is already known for some time that according to mtDNA sequence analysis, this genus is very probably polyphyletic. It is likely that it will be revised eventually; if Variabilichromis is split off, at least some of the more ancient lineages currently placed in Neolamprologus are probably worthy of separation also. However, the morphological similarity and numerous undescribed species do not make analyses easier, and as with many cichlids, recent speciation and abundant hybridization seriously confound molecular studies to the point where single-gene studies or those using only mtDNA or nDNA are essentially worthless for resolving Lamprologini phylogeny.

While lineages are clearly different in their morphology, habits and ecology, gene flow between genera and species is common enough due to extremely low postzygotic isolation. Males of Neolamprologus apparently have always readily and successfully mated with females of other Lamprologini they found ready to spawn: mtDNA lineages similar to other Lamprologini genera are widely encountered in species placed in Neolamprologus. And not only do such hybrids seem to be fertile at least to a limited extent in many cases, new species often appear to originate from such interbreeding.

Neolamprologus cylindricus Neolamprologus multifasciatus