(Figs. 2 and 3. See also photographs in Brichard [1989, pp. 344 and 377] and Konings [1988, p. 93])
Diagnosis. The species is very similar to N. falcicula , and N. chitamwebwaisp. nov. Most differences were found with N. savoryi . Neolamprologus walteri can be distinguished from N. brichardi , N. crassus , N. gracilis , N. splendens , N. pulcher , N. helianthus and N. marunguensis by a small interorbital width (14.8-22.1 % vs. 22.5-29.1 % HL). Neolamprologus walteri can be distinguished from N. falcicula , N. crassus , N. gracilis , N. splendens , N. pulcher and N. marunguensis by a small preorbital depth (11.8-16.9 % vs. 17.0-22.3 % HL). Neolamprologus walteri differs from N. olivaceous , N. savoryi , N. splendens , N. pulcher , N. brichardi and N. helianthus by the absence of markings on the operculum. With 33-35 scales in the longitudinal series, N. walteri is distinguished from N. gracilis , N. splendens and N. helianthus (36-38 scales). With 6-9 gill rakers, N. walteri is distinguished from N. brichardi and N. gracilis (10-18 gill rakers). Neolamprologus walteri differs from all species in the N. savoryi complex except N. falcicula by the absence of scales on the occiput; from N. splendens , N. marunguensis and N. savoryi by the absence of scales on the paired fins and by the absence of ctenoid scales on the dorsal and anal fins; from N. savoryi , N. brichardi , N. olivaceous and N. pulcher by the presence of cephalic pits; from N. olivaceous and N. pulcher by the absence of conspicuous spots on scales and from N. brichardi , N. pulcher , N. helianthus , N. crassus , N. splendens and N. marunguensis by a smaller caudal peduncle depth (11.3-13.2 % vs. 13.4-15.3 % SL). Neolamprologus walteri differs from N. savoryi and N. gracilis by having less scales between the upper and lower lateral lines (2 vs. 3) and by having less scales around the caudal peduncle (16 vs. 19). Neolamprologus walteri further differs from N. savoryi by the absence of bars on the body, by a smaller predorsal fin length (29.2-34.0 % vs. 34.6-36.8 % SL) and by having more pectoral fin rays (13 vs. 12). Neolamprologus walteri differs from N. chitamwebwaisp. nov. in the linear relations between cheek depth and head length and between body depth and standard length (significantly different intercepts, Ancova, P <0.002, Fig. 5) with larger cheek depth and larger body depth for specific specimen size. Neolamprologus walteri further differs from N. chitamwebwaisp. nov. in habitat preference (rubble substrate with fine sediment vs. large boulders with sand) and by having more pronounced markings on the dorsal and caudal fins (Fig. 2). Neolamprologus walteri differs from N. chitamwebwaisp. nov. by fewer canal scales in the lower lateral line (median 6, 54 % of all scales in the lower lateral line vs. median 9.5, 77 %, P <0.003, t test). Preorbital depth (POD) was significantly smaller (Ancova, P = 0.002 for POD as % of head length and P <0.001 for POD as % of standard length, Fig. 5) in N. walteri (11.8-16.9 % HL) than in all four examined type specimens of N. falcicula (17.0-18.5 % HL. The four examined type specimens include all type specimens of N. falcicula that are not suspected to be non-specific with the holotype; pers. comm. Jos Snoeks). Neolamprologus walteri differs further from N. falcicula by a smaller maximum standard length (57 mm vs 66 mm) and other associated size measures (two of the four type specimens of N. falcicula , including the holotype, are larger in body depth and head length than the largest N. walteri type specimen), by having drab grey-brown juveniles with bluish dorsal and anal fins (compared with bright yellow-orange fins of the juveniles of N. falcicula ; Konings, 1998), by having a distinct pattern on the unpaired fins (compared with no or vague markings in N. falcicula . Compare photographs of N. walteri [pp. 344 and 377 in Brichard, 1989] with N. falcicula [page 339 in Brichard, 1989] and the description by Brichard [1989, p. 343]: “the unpaired fins [of N. falcicula ], which are devoid of any markings”), and perhaps by having more dorsal soft rays (8-10 vs. 7 for holotype of N. falcicula ). The juveniles of N. walteri with bluish fins and greybrown body differ strongly from juveniles of the undescribed form N. sp. ‘cygnus’ (found at the south east end of the lake) which are bright blue and yellow (Konings, 1998). Juveniles of N. falcicula and N. sp. ‘cygnus’ do not have the vertical bars on the body of the juveniles of N. walteri . Neolamprologus walteri had on average more canines in the upper (8) and lower jaws (6), than found in all other species, except N. chitamwebwaisp. nov. , N. marunguensis and N. falcicula (in 2 of the 4 investigated type specimens). There were fewer canines in the holotype of N. falcicula , 3 and 6 (in upper and lower jaw respectively) and 6 and 4 respectively in paratype 80-31-P7 (the holotype and paratype of N. falcicula may have lost canines).
Description. Counts and measurements for the type series are given in Table 1. Range of standard length (SL) of investigated specimens: males 34.1-56.7 mm (18 specimens), females 32.5-49.6 mm (11 specimens). No sexual dimorphism. Meristics, morphometric measurements and colour patterns showed no differences between the populations of Tembo Rock and Muzungu beach.
Four-8 canines in upper jaw (mode 8, 69%) and 4-7 canines in lower jaw (mode 6, 72 % of specimens, Table 1) on anterior part of the dental arcade, with two outer canines on both jaws enlarged and the middle two smallest (Fig. 6). Dentigerous arm of premaxilla straight. Upper jaw tooth rows 5-6 anteriorly reducing to a single row half way along the jaw, outer row teeth (including the canines) 32-38. Lower jaw with 5-6 rows of teeth reducing to a single row approximately 2/3 along jaw, outer row teeth 27. Mouth terminal, isognath.
Six-11 gill rakers along lower limb of first gill arch (mode 7, Fig. 6). Gill rakers were slender, concentrated near the posterior end of the lower arch and decreasing in size towards anterior end of the arch. Marked with small dark spots, sometimes 1 or 2 reduced at lower end of arch, sometimes one or several bifid.
Scales absent on cheek, preoperculum and between eye and dorsal margin of preoperculum. Opercular scales 8-15, cycloid, variable in size but smaller than flank scales. Scales absent on occiput. Scales on nape cycloid, deeply embedded, varying from a few around the origin of dorsal fin to fully covered. Small scales between pectoral and pelvic fins 5-10, cycloid, difficult to distinguish. Small scales between upper lateral line and origin of dorsal fin cycloid, often covered by skin. Small cycloid scales on chest and belly. Urogenital area small ctenoid scales. Chest scales smallest. Scales on flank ctenoid. Dorsal fin with scaly sheath, few cycloid scales between soft rays, less between spines. Anal fin with scaly sheath, few cycloid scales between rays, more than on dorsal fin. No scales on pectoral and pelvic fins. Caudal fin small ctenoid scales, mostly on rays, covering about on average the anterior 75 % of tail. Larger specimens tended to have a larger part of tail covered with scales. Scales on outer rays range in adults up to past posterior end of middle rays.
Thirty-three-35 scales in the longitudinal series. In the upper lateral line 13-26 canal scales (76-100 % of all scales in the upper lateral line, median 20, 91%). In the lower lateral line 2-11 canal scales (13-100 % of all scales in the lower lateral line, median 6, 54 %). Scales without canal segments in between canal scales were pitted. Anterior to canal scales in lower lateral line, pitted scales generally extended almost up to the operculum. A positive relationship between standard length and number of scales in upper lateral line, and number of canal scales in lower and upper lateral lines. Lower and upper lateral lines overlapped in specimens>51 mm.
First pelvic fin ray longest. Caudal fin lunate. First spines of dorsal and anal fins small. Anal filament on average 2.5 % longer than dorsal fin filament. Dorsal fin margin posterior to sixth spine straight.
Lower pharyngeal jaw about as wide as long (Fig. 7). Dentigerous area covered with slender, pointed, unicuspid teeth, anteriorly slightly recurved, posterior rows straight and enlarged.
Post-lachrymal infraorbital bone series absent (Fig. 8). Free neuromast papillae on infraorbital lateral line system. Suborbital dermal depression forming a thin but distinct groove. A row of more or less contiguous cephalic pits with papillae runs from a pore posterodorsal of orbit to a pore anteroventral of orbit. Papillae clearly visible in some specimens, tiny or absent in others.
Many epidermal papillae around the pectoral, behind and on the pelvic fin, around the genital pore and posterior edges of operculum and preoperculum in specimens of Tembo Rock. Slightly or not at all present in specimens from Muzungu Beach.
Coloration. Body dark, grey to almost black, darker than N. chitamwebwaisp. nov. Larger specimens are lighter. Caudal margins of scales brown, produces a checkered pattern on the body. Between first 4 spines of dorsal fin (these increasing in size from rostral to dorsally) the margin of membrane yellow (lappets). Blue eye and blue stripe under the eye. On rare occasions stressed individuals show bands on body as N. savoryi .
No trace of opercular markings. Conspicuous black and white bands along dorsal margin of tail and posterior margin of dorsal fin (submarginal white or transparent). Vertical bands on dorsal and caudal fins. In preserved specimens vertical bands on fins less visible, no colour on iris or suborbital stripe.
Juveniles. Vertical bands on fins less clear or not visible in juveniles, especially on tail. Smallest young (up to ± 20 mm length) observed in natural habitat bluish fins and light vertical bars, with dark and light bars. In specimens of 14-25 mm body grey-brown. Emarginate tail, no filaments on caudal, dorsal and anal fins. Pelvic fin blackish, margin dorsal fin black. Tail 10 - 40 % covered with scales, no scales on other fins, operculum scaled, no scales on nape and occiput.
Distribution. Neolamprologus walteri was found from near the Burundi border with Tanzania in the north, to Cape Kabogo in the south ( 4°30'S - 5°25'S ). No observation dives were made in Tanzanian waters beyond these limits, except near Kisoje (Fig. 1) where N. walteri was absent. In Zambian waters, at the south end of the lake, N. walteri is absent. Around Kigoma, N. walteri is by far the most common fish species in the littoral.
Habitat and biology. Fig. 9 shows N. walteri in its habitat. Neolamprologus walteri was observed syntopically with N. brichardi and with N. savoryi . Neolamprologus walteri and N. chitamwebwaisp. nov. were not found in exactly the same area, but were associated with alternative habitat types. Neolamprologus walteri was found in bays and sheltered sites, exposed to weaker currents than the (only) site in which N. chitamwebwaisp. nov. was found. At the Bangwe peninsula south of Kigoma, N. walteri occurs on the north and south shores of the peninsula, while N. chitamwebwaisp. nov. is found on a more exposed site between these two localities on the peninsula (Fig. 4). The habitat of N. walteri features gentle slopes, with small stones or rubble, cemented by calcite, with fine sediment in spaces between rubble, and stones with stromatolite growth (Cohen et al., 1997). Neolamprologus walteri is often associated with bivalve shell accumulations covering large areas (always Pleiodon spekii [Pelecypoda: Mutelidae]). The shells (10 to 15 cm long) are cemented together and provide many small spaces for refuge and breeding.
The two sites where the paratypes were collected differed in topography and in species community composition. At Muzungu Beach N. walteri occurred in dense aggregates on patches of small encrusted rubble on relatively steep slopes. The patches were generally several tens of square metres, and separated from each other by large rocks. N. brichardi was primarily associated with the larger rocks, and generally did not occur on the rubble patches. The fish species assemblage in this habitat was relatively diverse, due to habitat heterogeneity with large boulders and deep water near the shore. At Tembo Rock rubble patches were larger,>100 m2 and sloped less steeply and large boulders were absent. Neolamprologus brichardi occurred here in large feeding aggregations, in the same area as N. walteri . Such large feeding aggregations of N. brichardi have been reported by Brichard (1989) and can consist of 100,000's of individuals. In this habitat the fish community consisted virtually only of N. walteri and N. brichardi , with a sharp segregation of the space occupied in the water column between the two species. Neolamprologus walteri occupied the lower water layer, near the bottom at an average 20 to 30 cm height (max c. 50 cm), with a high population density, and the less sedentary N. brichardi foraged in about equal numbers immediately above, at an average c. 60 cm off the bottom, over the same area.
In sheltered areas around Kigoma with rubble on a sandy bottom, N. walteri is the most common fish species and can occur in high densities. Neolamprologus walteri is a substrate spawner that forms large congregations and maintains small contiguous territories of about 1 m2 or less, each occupied by a number of fish including two or more adults, several subadults and juveniles. Other species of similar size are chased from their territory. Neolamprologus walteri colonies can cover hundreds of square metres at depths of 5 to 20 m with densities reaching 20 to 30 individuals per square metre. They excavate sediment next to and between rocks, thereby creating numerous crevices that function as refuges in the territory in which they hide during the day from approaching predators and spend the night. Crevices are shared by several individuals. Eggs were never observed and are probably deposited within these crevices.
Every individual N. walteri is strongly restricted to a small area and is rarely seen to travel more than 2 metres (>2 m seen only once in about 60 observation dives). The home range of an individual generally has a diameter of 20 to 200 cm, the larger the individual the larger the area. Although locally very abundant, N. walteri is probably a poor disperser, in view of its extreme philopatry.
Neolamprologus walteri either forages on the substrate, ingesting sediment, or hovers at 10 to 50 cm off the bottom, while taking plankton. When N. walteri forages in the water column it stays closer to the bottom than N. brichardi . The larger fish move higher from the substrate than the younger ones and are probably more dependent on a planktonic diet. The juveniles stay within 5 cm of the substrate. Neolamprologus walteri takes occasionally bites from the substrate surface for feeding, and sediment is egested through gills and mouth. When feeding on plankton, N. walteri has a lower feeding rate than N. brichardi .
Neolamprologus walteri is relatively wary, as N. savoryi , and stays close to the bottom and dives for the nearest cover immediately when approached. Fry are generally very few, as expected in lamprologine substrate breeders. Young <1 cm in length are usually visibly present in the colonies, though always very near the substrate and are first to vanish into the bottom.
Neolamprologus walteri were observed at a depth range of 2 to 30 m, but can probably occur deeper as well. They are most common at 7 to 15 m. There is a large depth overlap with N. brichardi , but N. walteri has lower densities in shallower depths than N. brichardi . Neolamprologus savoryi at the Bangwe peninsula is most often found at depths of 15 to 30 m, but also occurs shallower. Within its distribution in the lake N. walteri is present almost everywhere where there is some rocky cover, even in areas with an otherwise flat sandy bottom at 3 m depth, except at the site where N. chitamwebwai occurs. While territories of N. walteri occurred as close as 4 m from the southern boundary of the N. chitamwebwai community (its northern boundary is sandy and contains neither species), and the species can therefore be considered sympatric, their distributions were never found to overlap and it appears that their distributions are determined by the type of substrate.
Etymology. Named after Walter Dieckhoff, who first recognised it as an undescribed species (Konings, 1988). It has been referred to as ‘walteri’ in the aquarium trade for two decades and by Konings (1988).
