During investigations carried out between October 1973 and January 1977 in the Caprivi Strip, Van der Waal discovered in the vicinity of Gunkwe and near Lusese three pools with an undefined Nothobranchius species. This species was only found in these three localities and already considered rare and endangered (Van der Waal & Skelton, 1984).

However, until today the description and exact location of the original collecting sites and biotopes remain rather vague. Amongst the 39 regularly visited collecting sites in the eastern Caprivi Strip only three appeared to harbor the new Nothobranchius species. "One was a small isolated pool in a small temporary stream draining part of a floodplain into the large permanent Mutualwize Channel and the other two locations were in two small forest pans on the extreme edge of the floodplain of the Zambezi" (Van der Waal & Skelton, 1984).

• The first location [most probably corresponding to location 10 of Van der Waal & Skelton (1984)] was situated in the Eastern Floodplain and Chobe River area, east of Lusese village. This general sampling location was described as "Mutualwize, Lusese. Very Large permanent drainage channel, deep water with beds of Nymphaea caerulea, Echinochloa sp., Ceratophyllum dermersum, Lagarosiphon major and Najas pectinata." Fish prospecting equipment used included gill-nets, seine-nets and hand-nets; the site was visited and collected 5 times". The more specific location where the Nothobranchius sp. was found corresponded to a small isolated pool in a small temporary stream draining part of a floodplain into the large permanent Mutualwize Channel (Van der Waal & Skelton, 1984).
• The second location [corresponding to location 38 of Van der Waal & Skelton (1984)] was described as: "Small pan near Gunkwe. Rain-fed pan full of Cyperus sp. and Ottelia sp.". Fish prospecting equipment used included a hand-net. The site was collected and visited 2 times" (Van der Waal & Skelton, 1984). One of these pans was called the Sakamanduna Pan [Van der Waal & Skelton (1984); Plate 10:310].

Nothobranchius species Van der Waal & Skelton (1984)
Nothobranchius brieni, Jubb (1976)
Nothobranchius species "Chobe River", Wildekamp (1978)
Caprivi Killifish, Skelton (1993)
Nothobranchius species "Caprivi 95/1", Watters & Wood (1996a)

Rosenstock (1995: 166) recognises that this species closely resembles N.kafuensis, particularly its Kayuni State Farm variety, although the color pattern is quite different. Presently, it is considered to represent a different species but it remains to be seen whether it differs sufficiently from N.kafuensis to maintain this status (Rosenstock, 1995).

Wild caught males reached 65 mm; females 55 mm (Watters & Wood, 1996a). In the aquarium, male's reach on the average 42-46 mm, while females remain somewhat smaller, 33 to 34 mm long.

Citing van der Waal & Skelton (1984), "Caprivi is a finger-like extension [11 655 km²] to the north eastern corner of Namibia and represents a territorial relict of the aspirations of the German Empire to provide an east-west bridge between German East West Africa [presently Tanzania] and German West Africa [Namibia]. Caprivi was occupied for the first time in 1908 and named by Von Streitwolf in honour of the Chancellor, Count Von Caprivi. The strategic value of this relatively small area to Namibia lies in the rich sup-ply of fresh water, of which there is a general shortage elsewhere in the territory". The Caprivi Strip lies roughly between 17°30' and 18°30'S and 23°15' and 25°15'E [Figure 3] and has an average altitude of 930-m. The Zambezi River borders it in the northeast. To the west and the south it is bordered by the Kwando River, which after swinging eastwards forms the Linyanti swamp, Lake Liambezi and the Chobe River, eventually draining into the Zambezi (van der Waal & Skelton, 1984).

van der Waal & Skelton (1984) gave a fair insight of the hydrological dynamics of the Caprivi rivers system. "The terrain in Caprivi is remarkably level so that hydrological factors have a large influence on the area available to fish as well as on terrestrial life and govern the lives of the Eastern Floodplain living Basubia people. The 284 000 km² catchment of the Zambezi River above Katima Mulilo covers most of western Zambia. Differences of up to 8.2 m have been recorded between low water and flood in the river above the floodplain, but the average rise over a period of 23 years is 5.2 m at Katima Mulilo [data provided to van der Waal by the Department of Water Affairs, Windhoek]. The flood inundates most of the Eastern Floodplain, leaving only the higher sandy ridges as islands. Low water level occurs in October, with a sharp rise in January reaching one or more peaks in February to April before a decline in May-June. The floodplain is thus annually inundated from February to June. During low water from September to December most of the area is totally dry and is covered by terrestrial grasses."

"The Kwando (or Mashi) River originates further west in Angola and has a catchment area of 57 000 km² (Colquhoun, 1968). The seasonal difference in water level is only 1 m on average in the 2-10 km broad river valley. Water level differences are even less in the 10-30 km wide Linyanti Swamp at the southern end of the Kwando River. The floods in this system are retarded as a result of the meandering course of the main river and the cushioning effect of the vast Phragmites - Typha - Cyperus swamps that largely fill the river valley. The flood peak therefore only reaches Kongola near the northern border of Caprivi during May - June and Lake Liambezi at the end of the Linyanti Swamp by August. This flood is only able to stabilise the receding water level of Lake Liambezi, the rest being lost to evaporation. The lake connects with the Zambezi via the Chobe River, which presently is completely blocked by vegetation, including Salvinia molesta. Most of the water of the Chobe River originates in the Zambezi floods enabling the Chobe to push back as far as and beyond Ngoma [Figure 3]. When the floods recede, the flow changes direction and drains the southern part of the Eastern Floodplain. During low water the Chobe does however drain part of the water of Lake Liambezi if the lake is full enough. Under such circumstances, flow in the Chobe changes direction three times a year: August - January eastwards towards the Zambezi [Kwando flood and rainwater drainage], February - March westwards towards Lake Liambezi [without reaching it!] and April - June eastwards - drainage of floodplain. In high flood years, when the gauge at Katima Mulilo registers 7.0 m or more for a period of more than two weeks, there is a direct shallow link between the Zambezi floodplain and Lake Liambezi at Bukalo [Figure 3]. Available water level data from 1952 to 1975 shows that a possible temporary link may have been established during ten high flood years.

There is a similar shallow temporary connection between the Okavango River and the Kwando via the Magwegana or Selinda Spillway. This always brings water from the Okavango towards the Kwando/Linvanti, as was observed during 1975, an exceptionally high flood year. During average or below average years the Selinda may be totally dry" van der Waal & Skelton (1984).

Jackson (1961) recognizes that the upper Zambezi River is a typical "sand-bank" river, mainly with a sandy bottom. Citing van der Waal & Skelton (1984), "rocky substrates occur only at Katima Mulilo and at Impalila Island in Caprivi. At Katima Mulilo, the Zambezi main stream crosses silicified riverine ferricretes to form rapids. The rocky outcrop at Impalila Island is a basalt dyke, which forms the lip of the Eastern Floodplain. The whole of Caprivi lies within the Kalahari Sand Basin and consequently soils are mainly sandy with clay soils occurring especially around Lake Liambezi and the Chobe River. The character of the Zambezi and the Kwando Rivers differs markedly in Caprivi. The Zambezi has a seasonally inundated floodplain divided by many large permanent channels and low-lying semi-permanent grass swamps. The meandering Kwando River is bordered by a vast, permanent Phragmites mauritianus and Cyperus papyrus swamp, filling all of its wide river valley with the exception of bays, backwaters, forest-covered islands and a strip of seasonally covered grassland. The Linyanti Swamp is similar to the Kwando but is generally more uniform. Lake Liambezi lies at the end of the Linyanti swamp, has an open water surface of 10 000 ha and is bordered by a swamp of at least 20 000 ha".

According to van der Waal & Skelton (1984), the fish fauna of Caprivi is characterised by swamp and floodplain-loving fish species to which one can count Nothobranchius sp., Aplocheilichthys johnstonii, A. katangae and A. hutereaui.

As a final note, van der Waal & Skelton (1984) reported that "Caprivi has a unique fish fauna and this survey has shown that some species deserve special conservation attention. In the first place those species with narrow habitat requirements where the habitat itself is vulnerable. Nothobranchius sp. is a good example of this category. A second category is those species with limited distribution that have special appeal to the aquarium industry e.g. Nothobranchius sp. It is suggested that special conservation status be given to this species and measures be taken to safeguard their environment and to prevent their exploitation. Controlled cropping of larger and medium fish species for consumption and of small fish species for the aquarium trade just after the Zambezi floods recede in May to July can, however, be justified".

The collections recorded by van der Waal & Skelton (1984) were made from October 1973 to January 1977 whilst one of them [van der Waal] was employed by the Caprivi Government Service to investigate the fisheries of Lake Liambezi and their development. "Systematic collections and surveys of the fish life in the whole of Caprivi were made but as the area was relatively undeveloped and roads hardly did exist, sledge tracks were used with four-wheel drive vehicles to reach many of the collection sites. It was possible to reach certain localities, especially in the Linyanti Swamp and Eastern Floodplain, only during certain seasons or even only once in the four years" (van der Waal & Skelton, 1984).

Amongst the 39 regularly visited collecting sites in the Caprivi Strip only three appeared to harbour a new Nothobranchius species. "One was a small isolated pool in a small temporary stream draining part of a floodplain into the large permanent Mutualwize Channel and the other two locations were in two small forest pans on the extreme edge of the floodplain of the Zambezi" (van der Waal & Skelton, 1984):

• The first location [most probably corresponding to location 10 of van der Waal & Skelton (1984)] was situated in the Eastern Floodplain and Chobe River area. This general location was described as "Mutualwize, Lusese. Very Large permanent drainage channel, deep water with beds of Nymphaea caerulea, Echinochloa sp., Ceratophyllum dermersum, Lagarosiphon major and Najas pectinata. Gill-nets, seine-nets and hand-nets, collected 5 times" (van der Waal & Skelton, 1984).
• The second location [corresponding to location 38 of van der Waal & Skelton (1984)] was described as: "Small pan near Gunkwe. Rain-fed pan full of Cyperus sp. and Ottelia sp.. Hand-net. Collected 2 times" (van der Waal & Skelton, 1984). One of these pan was called the Sakamanduna Pan [van der Waal & Skelton (1984); Plate 10:310]

van der Waal & Skelton (1984) further report that "despite an intensive search the Nothobranchius sp. was not found in any other similar pans. Both these small pans lie adjacent to main gravel roads and the water is being used by local farmers and their stock and also by road-building contractors which therefore imposes a threat to these habitats".

The authors noted that most habitats were collected intensively but that it remained possible that fish species could have been missed out as a result of two factors: (i) the temporary limited distribution and or (ii) the selective habitat preference of a fish where it was difficult to collect. Examples of the first-mentioned group were Nothobranchius sp., which was collected only from three localities.

In mid-April 1995, Watters, Wood and Ipple only found in the Gunkwe area, south-westerly from Katima Mulilo, in the direction of Mpacha, in a remnant pool of about 5-m in diameter and about 20 to 30-cm deep their first 3 males of the "Caprivi-Nothobranchius". The estimated size of the pan when full was about 150 m x 40 m. These fish were in poor condition [water very murky with a high content of fine mud in suspension (water 19°C [66°F] at 8 am on 11 April 1995; total dissolved solids 160 ppm and pH:7.18 - the substrate consisted of very fine, black and sticky mud. The extreme softness of the substrate is regarded as being due to agitation by drinking cattle] (Watters & Wood, 1996a).

The nicest specimens were however caught in another pool about 300-metres away "but this one set quite far back from the road among the trees and, consequently, almost hidden from sight. Even from a distance it could be discerned that this was a substantial pool [50x30 meters with a maximum depth of 1-m at time of collection; water murky with a moderately high content of fine mud in suspension] with sedges and grasses growing in the water. There were large numbers of Nothobranchius in the pool, mainly occupying the shallows among the grasses and around clumps of sedge". The water conditions here were: pH: 7.59; 28°C [82°F] at 11 am on 11 April 1995; vegetation consisting of sedges and grasses; Total dissolved solids: 70 ppm; substrate: very fine, black and sticky mud (Watters & Wood, 1996a).

These two recent locations [locations 1995 1A and 1B] are apparently situated in the same elongated and shallow depression; At the time of collection, they formed distinct pans, but it is likely that previously, at the height of the rainy season, a temporary connection might have existed between them. The road has however been build between both pans, suggesting that such connections are rare these days. The Nothobranchius specimens of both pools were however identical (Watters & Wood, 1996a).

Nothobranchius "Chobe River" was discovered in 1973 by ichthyologist Ben van der Waal during an exhaustive investigation of the freshwater fishes of the eastern Caprivi area, carried out in collaboration with the Caprivi government during the period from October 1973 to January 1977 (van der Waal & Skelton, 1984). The collections made by van der Waal & Skelton (1984) included a few undescribed species or species of uncertain identification. The status of the new Nothobranchius species was being investigated by Jubb (pers. comm. to van der Waal & Skelton) and was likely to be undescribed (van der Waal & Skelton, 1984).

At the time of its discovery (1973-77), the species was already considered rare and endangered by van der Waal & Skelton (1984), and recommendations had been formulated to the Government of Namibia to protect the species and its habitat. Twenty years after its discovery, the species was still undescribed and its status in the wild was still unknown.

Because of this uncertainty, Watters, Wood and Ippel organised a collecting trip to Namibia in April 1995. From the Chobe River crossing at Ngoma [or Ngoma Bridge], the journey brought them northwestwards towards Katima Mulilo on an earthen road. They traveled for about 30-km on this road before turning eastwards towards the village of Lusese [Figure 3], one of van der Waal's collecting sites situated in the Eastern Floodplains area. On this sandy road track to Lusese, one small pool was investigated but only some shrimp were caught (Watters & Wood, 1996a:63-64). Apparently on arrival at Lusese village, some photographs were taken but this appeared to be an unpopular move and the village chief refused the team to further fish the area. It only could retreat without inquiring or reaching van der Waal's location at the Mutualwize channel, further east of Lusese.

On rejoining the main road from Ngoma Bridge to Katima Mulilo, the team soon reached the village of Bukalo. Watters & Wood (1996a & b) report that in the vicinity of this village one could find one of the Nothobranchius biotopes discovered by van der Waal. One pool located a few kilometers [7-km] south of Bukalo looked particularly promising to the 1995 team and they felt sure it was the locality of van der Waal. However, no Nothobranchius were found in this and in several other pools and pans that they encountered further along the road to Katima Mulilo. According to published information in van der Waal & Skelton (1984; Plate 10:310) [Figure 3], location 38 of van der Waal and more particularly the Sakamanduna Pan is not situated near Bukalo as thought by Watters and Wood (1996a & b) but is situated more towards the northwest and closer to Gunkwe and Mpacha, away from the Ngoma Bridge - Katima Mulilo road.

In the direction of Mpacha, near Gunkwe, they finally found along the road an almost dried up pool of about 5-m in diameter and about 20-30-cm depth. In the murky water of the pool Watters caught one male Nothobranchius. However, it got eaten by a water scorpion which was simultaneously netted. Further efforts did not yield any additional Nothobranchius. The next morning, two more males were captured but they were in very poor conditions (Watters & Wood, 1996 a). This locality was situated in a forested area along the extreme northern edge of the floodplain.

In the immediate vicinity of the previous pool, they found another but larger pool, set quite far back from the road hidden among the trees and hidden from sight. It was a pool with sedges and grasses growing in the water. Several specimens were collected from this pool (Watters & Wood, 1996 a).

The fish was misidentified as Nothobranchius brieni by Jubb (1976) and was then referred to as Nothobranchius species "Chobe river" by Wildekamp (1978), and as "Caprivi Killifish" by Skelton (1993). Watters & Wood (1996 a) named the Nothobranchius population they found near Gunkwe Nothobranchius species Caprivi 95/1 and considered it to be identical to that from van der Waal's Bukalo [rather from the Sakamanduna Pan or from the Mutualwize channel] location.

According to Watters & Wood (1996a), wild caught males displayed an extreme aggressive behaviour and it was necessary to keep males separated from each another. Even the wild females appeared to be aggressive toward one another. The males constantly drive the females to spawn and, if not well fed and given periods of rest apart from the males, they will rapidly lose condition. The wild specimens proved to be extremely prolific spawner and in one instance where a rough count was made, a single trio produced at least 650 eggs over a 10 day period (Watters & Wood, 1996).

Maintenance of the Caprivi 95/1 population does not present any special problems a part from the above-mentioned aggressiveness. In the F1 and later generations this aggressiveness seemed to have waned away. Watters & Wood's (1996) experience and also those to whom eggs from the wild fish were distributed, indicate an incubation time for the eggs from the wild fish ranging from 2 to 41/2 months. On hatching the fry are relatively large and can easily consume freshly hatched Artemia-nauplii.

Watters & Wood (1996) consider this population of Nothobranchius to be extremely rare in the wild and, therefore, susceptible to habitat destruction by human activities.

Kladec (1997) mentions that on hatching the fry measure 4.1 to 4.2 mm. They take easily freshly hatched brine shrimp and later fine Cyclops. Growth is rapid, and at five weeks of age they are about 2 cm long. At that size, males have already begun to colour up. At seven weeks the first spawning attempts are being made. According to his experience, maintenance and breeding of the Caprivi 95/1 population presents no major problems. This population is as easy as the well-known N.guentheri and N.foerschi. The population is very prolific. A single female is capable of producing 250 eggs in a month. Hatching time depends on the temperature during storage: 10 to 14 weeks; 10 weeks at 26-29°C, and 14 weeks at 21-23°C.

• BELL-CROSS. G. (1972): "The fish fauna of the Zambezi River System"; Arnoldia, 5(2a):1-19
• Colquhoun, B. et al. (1968): "Report on bridge across the Chobe River at Ngoma, Eastern Caprivi; report to the Department of Bantu Administration and Development"; 26 pages
• Kladec, J. (1997): "A New Nothobranchius population from Namibia: Nothobranchius spec. Caprivi 95/1"; Killi-News; Journal of the British Killifish Association; April 1997 N°379, pp.59-60; 1 fig.
• Holtzhausen, J. A. (1990): "Freshwater fishes of Namibian wetlands - A review"; Madoqua 17 (2):189-191.
• Jackson, P. B. N. (1961): "Ichthyology. Kariba Studies. The fish of the Middle Zambezi"; Trustees of the National Museums of Southern Rhodesia; Manchester University Press; 36 pgs.
• Jubb, R.A. (1976): "Three interesting species of Nothobranchius"; B.K.A. Killi-News N° 127:1-8.
• Rosenstock, J. (1995): "Nothobranchius from North to South"; J. British Killifish Association, Killi News, 362:155-166, 6 figs.
• Skelton, P.H. (1987): "South African Red Data Book - Fishes"; South African National Scientific Programmes Report N° 137: 199 pgs.
• Skelton, P.H. (1993): "A complete Guide to the freshwater fishes of Southern Africa"; Southern Book Publishers, Zimbabwe: 388 pgs.
• van der Waal, B.C.W. & P.H. Skelton (1984): "Check list of fishes of Caprivi"; Madoqua 13(4):303-320, 4 tabs, 1 fig, 12 plates.
• Watters, B. & T. Wood (1996a): "A search for Nothobranchius species in central southern Africa"; Killi News - Journal of British Killifish Association; May 1996; N° 368, pgs 62-71, 1 map; 5 figs.
• Watters, B. & T. Wood (1996b): "A search for Nothobranchius species in Central Southern Africa"; Journal of the American Killifish Association; May-June; 29(3):71-81; 5 figs.
• Wildekamp, R. H. (1978) "Die Nothobranchius Arten der Zaire Provinz Shaba"; Deutsche Killifish Gemeinschaft Journal 10(8): 121-129.