A survey of fishes of Hombolo Lake, Dodoma, Tanzania, with evidence for local extinction of a native tilapia as a consequence of stocking

The fish community of the Hombolo Lake, an impoundment on the Wami catchment near Dodoma, Tanzania, was surveyed in 2014 and 2017. The lake contains a relatively low diversity community dominated by two Oreochromis species introduced from outside the Wami catchment, O. niloticus and O. esculentus. Evidence from historical collections suggests that the native O. urolepis was formerly present, and its current absence is likely to be the result of competitive exclusion or genetic swamping by non-native species introduced for fishery enhancement. Four other fish species were also recorded.


Introduction
Tanzania is a global hotspot of tropical fish biodiversity and, in particular, houses the largest diversity of native tilapia species of the most important farmed genus Oreochromis of any country (Lind et al. 2012). Recent studies have shown that stocking of exotic strains of tilapia can have detrimental effects on native species, through competitive exclusion and genetic swamping following hybridisation (Angienda et al. 2011;D'Amato et al. 2007, Deins et al. 2014, Ndiwa et al. 2014. Until recently, little information has been available about the current distributions of native and exotic species in Tanzania. Here, we focus on Lake Hombolo, a small (7x 1.4km) water body in central Tanzania, located in the Dodoma region; S05˚57'02.8"; E035˚58'07.6" at an elevation of approximately 1040m (Figures 1-4). It is an impoundment of a small tributary stream on the Upper Wami River catchment, reportedly in the 1950s, originally for the irrigation of vinyards. According to local officials, this use continued at the time of our visits. The water is rather saline and alkaline, with conductivities of 3,300-4,000 μScm -1 and pH 8-8.5 (Shemsanga et al. 2017). No published information on its fish fauna could be found, although we were able to locate some material in the collection of the Natural History Museum in London, and to examine the specimens.

Methods
The lake was sampled during January 2014 (BPN, GFT) and January 2017 (GFT, A.G.P. Ford, A.Shechonge, A.M. Tyers). Specimen identifications were checked with type material in international museum collections. Voucher specimens, photographs and fin clips for DNA analysis have been archived at respective institutions.

Results
When visited, the water was very turbid, with some remaining standing dead trees and fringed with banks of reeds/papyrus. At the shoreline sites, the water remained very shallow for a long distance offshore. Active fisheries with gillnets, beach seines and mosquito seines were seen (Figures 5,6). 2014 Wami Site 6: Hombolo Lake. S05˚57'08.2"; E035˚56'34.6". This site was accessible by driving and then walking across the sun-baked mud of the dry lake bed. Samples were bought from seine netters camped at the shoreline several hundred metres from the permanent bank. All specimens seen here were Oreochromis esculentus, including large numbers of relatively small males in breeding dress (Figure 8), suggesting that there was a breeding aggregation nearby.  Figure 13), with a few Clarias cf. gariepinus (Burchell 1822) also seen. There was no sign of O. esculentus being landed. The area to the north of Site 1 was very muddy and judged impassible by car. We approached the lake on foot (at 2017 Site 2: Figure 12), but found no fishing in progress apart from 2 people who seemed to be catching shrimps with some kind of makeshift net. We hired them to use a length of mosquito netting as a seine in shallow water. This yielded a large number of Astatotilapia bloyeti (Sauvage 1883) (Figure 15), including many rather small males showing both blue and yellow background colour. Other species landed included a few Enteromius cf. paludinosus and a single Coptodon cf rendalli (Boulenger 1897) (Figure 14). With the assistance of a local official, we were able to obtain 3 specimens of O. esculentus from a trader, from an unknown location on the lake. It was not clear whether the relative rarity of this species was due to declining populations or targetting of the fisheries in operation at the time of visiting.  The smallest three were badly bent. All specimens were pale coloured with conspicuous thin dark bars and a 'tilapia mark' on the dorsal fin. In some specimens, the caudal fins had a few prominent stripes. They were photographed but not measured (Figure 16 A-D). Specimen C shows jagged, more vertical stripes, that are fainter nearer the caudal peduncle, which fits better with the fresh O. urolepis specimen (F) than the O. niloticus specimen (E).
The London specimens (Fig 16A-D) appear to be Oreochromis urolepis (Norman 1922), a species native to the Wami system, but not recorded in the 2014 or 2017 sampling. They certainly are not Oreochromis esculentus (Figs. 8,9). According to Trewavas (1983, page 211), diagnostic features of O. esculentus include a slender caudal peduncle and absence of stripes on the caudal fin, which are consistent with the 2014 specimens illustrated above, but not with these 1972 specimens. The other Oreochromis species presently found in the lake is O. niloticus, which also has striped caudal fins (Figure 7, 13). There are no clearly diagnostic morphological differences between this species and O. urolepis at such small sizes (Trewavas 1983). However, O. niloticus usually has more consistently prominent and regular stripes in the caudal fin than O. urolepis. Comparison of similarly-sized strongly-striped individuals of the two species (e.g Figure  16 E-F) indicates that the stripes of O. niloticus tend to be regular, gently curved and are appear dark close to the caudal peduncle, while those of O. urolepis tend to be straight, but rather jagged and fade closer to the body. The London specimens also have rather deep caudal peduncles, which is characteristic of O. urolepis.
In any case, it would be surprising if O. niloticus was in Hombolo Lake in 1972. The species was not widely stocked in Tanzania at that time, and when interviewed in 2014, local fishermen claimed O. esculentus to be native but O. niloticus to have been accidentally introduced in the 1990s when flooding led to overflow of nearby aquaculture ponds. Thus, it seems likely that the 1972 specimens are correctly identified as 'Tilapia hornorum Trewavas 1966', a junior synonym of O. urolepis. Oreochromis urolepis is native to the Wami system, and O. hornorum (sometimes considered a distinct subspecies) was originally described from specimens collected on this system near Kilosa on the Mkondowa River. Thus, it appears that the lake was almost certainly originally populated by the native Oreochromis urolepis.

Discussion
At the time of sampling in 2014-17, the fish fauna of the lake was relatively depauperate, with only 6 species recorded. Of these, three have almost certainly been introduced: Oreochromis niloticus, O. esculentus and Coptodon cf. rendalli. Within Tanzania, the Nile Tilapia, O. niloticus, is native only to Lake Tanganyika and is catchment, but it was stocked into Lake Victoria in the 1950s, initially probably from Lake Albert in Uganda (Trewavas 1983). The Nile Tilapia has been collected from Lake Victoria for production in hatcheries and distribution to fish farmers throughout Tanzania since the 1990s. It has also been widely stocked in natural lakes and impoundments of natural river systems (Shechonge et al. 2018a).
Oreochromis esculentus is endemic to the catchment of Lake Victoria (Trewavas 1983), where it has largely been replaced by O. niloticus (Shechonge et al. 2018a). It was widely stocked in Tanzania during the 1950s and 60s, most notably in Lakes Singida, Nyumba ya Mungu (Trewavas 1983) and Rukwa (Seegers 1996).
Our surveys since 2011 have indicated its continued occurrence in these lakes (Shechonge et al. 2018a).
Coptodon rendalli is native to Tanzania, occurring naturally in the catchments of Lakes Tanganyika (Eccles 1992) and Malawi (Nyasa), the type locality being the Lake Malawi outflow, the Upper Shire River (Boulenger 1897). Eccles reports that it has been widely distributed through stocking and fish farming. In our recent surveys, Coptodon was found in almost all water bodies surveyed. Identification is slightly complicated by the fact that Coptodon rendalli and the very similar-looking C. zillii (Gervase 1848) (not native to Tanzania) have both been stocked in Lake Victoria. Coptodon have been seen in hatcheries in Tanzania where it was claimed that Nile Tilapia, originally sourced from Lake Victoria, were being bred for distribution to fish farmers: if the Coptodon were also from Lake Victoria, these could be either or both species or indeed hybrids.
Two of the other species, provisionally identified as Clarias gariepinus and Enteromius paludinosus are widespread and very abundant in most water bodies in the region (Eccles 1992), so it is probable that they are native. Work in progress indicates that E. paludinosus represents a complex of species. The final species observed, the small haplochromine cichlid Astatotilapia bloyeti, is indigenous to the Wami system (type locality near Kilosa) and to other river systems to the north and west and so is very likely to be naturally occurring.
The absence of Oreochromis urolepis from our recent collections is noteworthy. As this species is native to the Wami system and appears to have been present in the lake in the 1970s, the likeliest explanation is that it has been exterminated through competitive exclusion and/or genetic swamping by the introduced O. esculentus and/or O. niloticus. This does not bode well for the long-term survival of this species, given that O. niloticus is established in all major catchments where this species is native (Shechonge et al. 2018a) and that the two species are known to hybridise (Shechonge et al. 2018b).