The non-Haplochromis fish fauna in Uganda: an update on the distribution and a review of data gaps

Freshwater fishes are the second most threatened group of vertebrates after amphibians. In most developing countries, the conservation of freshwater fishes is largely hampered by limited information and data. The Red List assessments by the International Union for Conservation of Nature (IUCN) provide a benchmark for monitoring and mitigating species extinction risk, but these assessments require, inter alia, quantitative information on the species range in the wild. This information is largely missing for many species that face extinction threats. Here, we combine species occurrence data, expert knowledge, and literature to (i) review and update the distribution of non-Haplochromis fish species native to Uganda and (ii) determine the national geographical range of these fish species relative to their global range. For the latter, we use the IUCN’s standard protocol for mapping distribution of freshwater species from which we derive both the national and global extent of occurrence (EOO) and area of occupancy (AOO). We show that at least 110 non-Haplochromis species occur in Uganda, with the majority species having a wide distribution outside the national boundaries. However, substantial gaps exist in data on presence of the species in their native range, where more than 60% of the species are designated as “possibly extant.” We elaborate on fish species previously believed to occur in Uganda, e.g., Amphilius kivuensis Pellegrin, 1933; Bagrus degeni Boulenger, 1906; Marcusenius macrolepidotus (Peters, 1852); Petrocephalus catostoma (Günther, 1866); and Lacustricola pumilus (Boulenger, 1906), but where recent studies suggest they lack a native distribution within the country. In addition, we highlight fishes with occurrence records that site them in areas beyond their previously known range, requiring further investigations to ascertain their present status. This work has potential to (i) trigger evidence-driven policies aimed at site-based conservation and rethinking of extent of protected areas, (ii) stimulate data collection, especially in areas where fishes are designated as “possibly extant” and “presence uncertain,” and (iii) aid the IUCN Red List assessments, for which conservation status of the majority non-Haplochromis species is outdated and many others remain unevaluated due to lack of quantitative information on their range.


Introduction
Freshwater fishes are the second most threatened group of vertebrates after amphibians (Darwall & Freyhof, 2016). The threats to freshwater fishes are multi-faceted and multi-dimensional, but the major ones include over exploitation (both fishing and overenthusiastic collecting), habitat degradation and pollution (driven largely by human activities in the catchments), invasive species, climate change, and infrastructure development (e.g., hydro-electric power dams, oil and gas development) (Magurran, 2009;Barbarossa et al., 2021;Costa et al., 2021). Fishes are being overexploited due to the perception that they are renewable "agricultural products" with no intrinsic biodiversity value (Wadewitz, 2011). Native fisheries have been greatly impacted by invasive species through predation, competition, hybridization, and habitat alteration (e.g., Kaufman, 1992;Ogutu-Ohwayo, 1990). Furthermore, freshwater water bodies have been polluted by dumping harmful substances in the form of insecticides, pesticides, heavy metals, mills waste, and crude oil, causing fish kills and affecting species diversity (Malik et al., 2020). Efforts by governments globally to reduce threats and halt or reverse biodiversity loss have not been successful, and the magnitude of these threats is likely to worsen in future as demand for aquatic food and animal protein continues to increase (Darwall et al., 2018;Dudgeon et al., 2006).
Conservation of freshwater fishes is partly being hampered by limited information and data that are needed to guide conservation planning. Ex situ conservation, for instance, requires information on their current status in the wild and the condition of the natural habitat. This information is largely missing for many species that face extinction risk. For instance, the International Union for Conservation of Nature (IUCN) Red List database of threatened species, which is the world's widest benchmark for conservation and development planning processes (Hoffmann et al., 2008), shows that the majority of the freshwater fish species have remained unassessed due to lack of data (IUCN, 2017).
Uganda is endowed with a number of freshwater bodies: up to 18% of the surface area is covered by lakes, rivers, and swamps. There are five major lakes and about 160 small lakes and numerous rivers. These water bodies lie in two ichthyofaunal provinces, i.e., the East Coast ichthyofaunal province (Lakes Victoria, Kyoga, and Edward systems) and the Nilo-Sudan ichthyofaunal province (Lake Albert system) (Fig. 1). These water bodies harbor over 500 fish species , many of which are endemic (Greenwood, 1966).
To date, Greenwood (1966) is still the most widely used reference for the distribution of fishes in Uganda. Excluding the Haplochromis cichlids (commonly referred to as "haplochromines"), a total of 94 fish species in 17 families are documented in Greenwood (1966). Yet, taxonomic discoveries as well as revisions of previously described species have been on the rise since the work of Greenwood (1966). For example, between 2003 and 2022, approximately 250 new fish species were described globally per year (Fricke et al., 2022a). This observation suggests that new fish species description is a remarkably active area of species discovery, and therefore, Greenwood (1966) is likely out of date, requiring more recent reviews, based on numerous studies and surveys that have been conducted since then.
Recently, online databases such as FishBase (https:// www. fishb ase. org), the Global Biodiversity Information Facility (GBIF) (https:// www. gbif. org/), and the Freshwater Biodiversity Portal for Uganda (FWB) (https:// fresh water biodi versi ty. go. ug/) have served as key reference tools for more recent data and information. FishBase is the most comprehensive of these reference tools and has served as an alternative to Greenwood (1966) for the updates on the fishes of Uganda. However, FishBase also has limitations. First, FishBase derives most of the distributional information on fishes in Uganda, excluding haplochromines, from Greenwood (1966). Second, Fish-Base tends to be ambiguous when there is no precise information on the species' range. For example, a species such as Garra hindii (Boulenger, 1905) in family Cyprinidae is documented in FishBase to occur in "Uganda, Kenya, Cameroon, and the Congo basin"; yet, Garra species are mostly endemic (Jos Snoeks, personal communication), and such a large country distribution, spanning more than four ichthyofaunal provinces, is unlikely. Third, FishBase collates distribution information from published literature, leaving out enormous information from unpublished surveys and gray literature. Other reference databases such as GBIF and FWB, which host occurrence data from surveys, require analytical skills to deduce distributional information. Such skills are not common with Page 3 of 24 412 Vol.: (0123456789) most potential data users, especially policymakers, hence the need to collate this information from various sources into user-friendly format.
The study by Decru et al. (2020) is the most-recent attempt to collate information on the distribution of fishes in Uganda. The authors evaluated the occurrence of the non-Haplochromis species found in the Lake Edward system by examining occurrences in three ichthyofaunal provinces: Nilo-Sudan, East Coast, and Congo ichthyofaunal provinces. These authors found that the Lake Edward system alone hosted 34 non-Haplochromis species, belonging to 10 families and 21 genera, registering six new species that were not listed by Greenwood (1966) and online databases such as FishBase. Similarly, there were several species in Greenwood (1966) that were no longer valid due to taxonomic revisions. These findings further underscore the need to review and update the distribution of fishes in other systems within the country.
Besides, both the online databases (such as FishBase, GBIF, and FWB) and systematic reviews such as Decru et al. (2020) only provide qualitative information on the species distribution, which is insufficient, for instance, in assessing the conservation status of the species. The IUCN Red List of Threatened Species, which is the global benchmark for the conservation status of species, currently relies on quantitative information on either population trends or the geographical extent to designate a species as "Threatened" (Critically Endangered, Endangered, or Vulnerable), "Near Threatened", "Least Concern", or "Data Deficient" (IUCN, 2019a). Because information on population trends is scanty, most assessments in the IUCN Red List rely on the geographical extent of the species. The Extent of Occurrence (EOO) and Area of Occupancy (AOO) are by far the most commonly used metrics to infer the geographical extent of species used in IUCN assessments (IUCN, 2019a). When estimated at a scale that is biologically relevant to a species, EOO and AOO can provide insights into the species' habitat requirements, threats, and limitations, and if data are available over time, valuable trend information (IUCN, 2019a). However, this information is largely missing for many freshwater fishes, including those in Uganda. In the case of Uganda, the absence of this information could be one of the reasons (i) the Red List assessments for over 200 fish species have remained out-of-date and (ii) the National Red List assessments, which have recently become popular globally (Brito et al., 2010) and exist for the terrestrial taxa in the country (WCS, 2016), are still lacking for fishes. If this gap in the knowledge of species range is not addressed, it will also be hard to appreciate the loss of diversity and the need to develop evidence-driven policies aimed at sitebased conservation and rethinking of extent of protected areas. Consequently, species losses may continue and the opportunity to conserve much of the remaining rich biodiversity will be lost.
The purpose of this study, therefore, is twofold: first; to provide an update on the current status of non-Haplochromis fishes in Uganda, building on Decru et al. (2020); second, to estimate the extent and range for the non-Haplochromis fish species in Uganda in relation to the global range. We considered only the non-Haplochromis fishes because haplochromines are largely understudied and the majority species are undescribed. Consequently, identification of these fishes in the field is challenging and data on occurrence is scanty.

Study area and scope
In this study, we reviewed distribution of non-Haplochromis fish species in all water bodies across Uganda. The major water bodies can be classified into five major lakes (Victoria, Kyoga, Albert, Edward, and George) and fiver major rivers (Nile, Katonga, Kagera, Sio, and Aswa). In addition, there are over 160 small lakes, numerous swamps, and wetland spread across the country. All these water bodies cover up to 18% of the total country surface area (Nsubuga et al., 2014) in six major drainage basins (Fig. 2). The fishes in these water bodies mainly live in two major ichthyofaunal provinces: the Nilo-Sudan and the East-Coast provinces. However, these two ichthyofaunal provinces span many countries, i.e., from the Indian Ocean (East Africa) to the Atlantic Ocean (West Africa) to the Mediterranean Sea (North Africa). This large extent suggests that fishes that are not endemic to a specific water body could have a very wide distribution and range (in terms of EOO and AOO). Consequently, one would expect the proportion of national EOO and AOO to the global range to be substantially smaller (probably less than 10% for the fishes in the bigger Nilo-Sudan ichthyofaunal province, and less than 50% for the fishes in the relatively smaller East Coast ichthyofaunal province). To investigate this hypothesis, one needs to map both global and national range of every fish species under consideration to estimate the values of EOO and AOO at global and national levels. Whereas fish faunas in the Lake Edward system (which is at the intersection of Congo and East Coast ichthyofaunal provinces) are also included in this review to give a complete picture, we mainly focused on the water bodies not included in Decru et al. (2020), which all lie in the Nilo-Sudan and East Coast provinces.
Data sources Decru et al. (2020) relied on information from literature, distribution notes in FishBase, museum collections, and field expeditions. In this study, field expeditions and access to museum specimens were not possible. Therefore, we relied on the approach used by IUCN, where mapping of the species range is based on past observation records, knowledge of habitat preferences, and remaining suitable habitats (IUCN, 2019b). This approach is also powerful in identifying priority conservation areas, identifying data gaps (which can inform further detailed investigations and data collection), and informing management decisions (Jetz et al., 2012). The data was derived from the literature (both published and unpublished), expert knowledge, online databases such as FishBase (Froese & Pauly, 2022), Eschmeyer's Catalog of Fishes (Fricke et al., 2022b), the GBIF (GBIF.org, 2022), the FWB (Musinguzi et al., 2023;, and institutional archives, especially from the National Fisheries Resources Research Institute (NaFIRRI).

Updating the species list
A preliminary list of non-Haplochromis fishes in Uganda was downloaded from FishBase, a global online database of fishes, from the country-specific page (Froese & Pauly, 2022). The taxonomic status of the names was checked using Eschmeyer's Catalog of Fishes (Fricke et al., 2022b), which is also an on-line database that references work on the scientific names of fish species and genera.
Fish species occurrence data were downloaded from the GBIF (filtered by country) (GBIF.org, 2022) and the FWB. Data from these databases follow Darwin Core attributes, which facilitate easy use (Wieczorek et al., 2012). Therefore, all the occurrence data acquired from non-standard Darwin Core formats such as those of NaFIRRI archives was first mobilized into similar formats. These records were rigorously compared with the species list from FishBase to identify any inconsistencies in the distribution of known species and to ascertain where new species may have been recorded in recent surveys. Also, the records were checked against a host of literature sources, especially FishBase (Froese & Pauly, 2022), Eschmeyer's Catalog of Fishes (Fricke et al., 2022b), Greenwood (1966), and Decru et al. (2020), for inconsistencies in taxonomy as well as taxonomic updates, uncertainty over the identity or authenticity of the records, and the accuracy of the locations. Whenever the distribution of a species was in doubt, appropriate species occurrence designation was applied following the IUCN standard attributes for spatial data (Table 1).
Before use, occurrence data was cleaned to make it fit for purpose by removing or correcting inaccurate, unreasonable, or incomplete data (García-Roselló et al., 2014). These included records with localities that were not in Uganda and all the records without specific epithets. Fields with data that were not relevant to the study were deleted to simplify data cleaning. Occurrences that had no coordinates were georeferenced where possible, based on information in the recorded field notes, locality, water body, and location remarks, using Google Earth Pro. These coordinates were in decimal degrees.
Other occurrence data from the institution archives were mobilized following the Darwin Core attributes, where higher taxonomy for taxa in the data was generated using the GBIF species name matching tool (https:// www. gbif. org/ tools/ speci es-lookup). This tool is used to normalize species names from a.csv file against the GBIF backbone. The Canadensys coordinate conversion tool (http:// data. canad ensys. net/ tools/ coord inates) was used to convert the geographic coordinates from other formats to decimal degrees.
All the occurrence datasets were changed from GBIF standards to IUCN standards using IUCN Environ Monit Assess (2023) 195:412 standard attributes for spatial data, which contain the core and optional fields (IUCN, 2019a). The core fields are the required or recommended attributes such as binomial, origin, presence, compiler, citation, and year compiled, whereas the optional fields are attributes that may or may not be included in the dataset such as record number, recorder, country code, verbatim latitude, and verbatim longitude. Among the core spatial fields, species presence is the most important attribute as it determines whether the species range can be quantitatively ascertained or not (IUCN, 2019a). This attribute has five fields ( Table 1), but only three fields were used: "extant," "possibly extant," and "presence uncertain." Most of the species had occurrence records, but these records were too old to ascertain the present status of the species in the area (i.e., more than 30 years). These species with older records were presumed possibly extant on assumption that not enough field searches for these species have been done (Jos Snoeks, personal communication). Other fishes had records that extended beyond their known and expected range, based on various literature sources listed above. These fishes were designated as presence uncertain (Table 1) until specimens are obtained to confirm their validity. For the attributes such as presence, origin, season, and basis of the record, records were filled in using IUCN lookup codes (Table 1; IUCN, 2019a). The codes for the origin of each taxon were recorded based on information from FishBase (Froese & Pauly, 2022), Greenwood (1966), and Decru et al. (2020).

Modeling the species range
An updated species list (Table 2) was sent to the IUCN Freshwater Unit to create a Species Information Service (SIS) account. The SIS is the central database used by IUCN to store and manage species accounts and assessments for publication on the IUCN Red List database. In SIS, a new working dataset was created, comprising all the fish species to be mapped. Creation of the SIS account enabled access to the IUCN Freshwater Mapping Application (FWMA) used for mapping species distributions and calculating EOO and AOO (IUCN, 2019b). FWMA is a web-based mapping application that is widely used, as part of assessments for the IUCN Red List of Threatened Species, to produce distribution maps of freshwater species based on hydro-basin layers (Lehner & Grill, 2013;IUCN, 2019b). This application also provides an online platform to produce new distribution maps or update existing distribution maps for species published on the IUCN Red List.
The FWMA uses point data on species' location to create a map, encompassing all possible hydro-basins where a species can be found. Species were mapped in the FWMA using hydro-basin level 8; after, the Possibly extant There is no record of the species in the area, but the species may possibly occur, based on the distribution of potentially suitable habitat at appropriate altitudes, or the records exist but are more than 30 years old and the present status cannot easily be ascertained. Identifying Possibly Extant areas is useful to flag up areas where the taxon should be searched for. 4 Possibly extinct The species was formerly known or thought very likely to occur in the area (post 1500 AD), but it is most likely now extirpated from the area due to habitat loss and/or other threats, and there have been no confirmed recent records despite searches. 5 Extinct The species was formerly known or thought very likely to occur in the area (post 1500 AD), but it has been confirmed that the species no longer occurs because exhaustive searches have failed to produce recent records, and the intensity and timing of threats could plausibly have extirpated the taxon. 6 Presence uncertain A record exists of the species' presence in the area, but this record requires verification or is rendered questionable owing to uncertainty over the identity or authenticity of the record, or the accuracy of the location.
Page 7 of 24 412 Vol.: (0123456789) Originally known from Albert Nile and Aswa River (Greenwood, 1966), but no recent data to ascertain the current status. Recent records suggests the species is present in Lake Albert (Wandera & Balirwa, 2010), but this needs verification M. rheni (Fowler, 1936) Lake Victoria. There are no recent records for this species, but possibly extant M. victoriae (Worthington, 1929)  Lake Albert and the delta of Murchison Nile; also possibly extant in Albert Nile (Greenwood, 1966   L. alluaudi (Pellegrin, 1909) Lake Edward system (Rivers Mubuku, Sibwe, and Ruimi), but it has not been observed recently L. huloti (Banister, 1976) Possibly extant in Lake Albert. There is insufficient information on distribution as it only known from the type locality (Zega on the Vuda River) L. somereni (Boulenger, 1911) Rivers in the Rwenzori mountains (i.e., Sibwe, Mubuku, Rwimi, and Kirimia). No recent records from these areas, which could be attributed to limited sampling of the area ) E. pellegrini (Poll, 1939 Lake Edward system 25,708.9 58.1 18,818 55.3 E. nyanzae (Whitehead, 1960) Lake Victoria basin. Current status unknown due to lack of recent data. Records from Lake Kawi are uncertain and likely to be misidentifications E. jacksonii (Günther, 1889) Lake Victoria and Kyoga basins 74,350.487 10-20 38,182 10-20 E. profundus (Greenwood, 1970) (Whitehead, 1960) Aswa river and Lake Kyoga affluent rivers (Greenwood, 1966) and affluent rivers of Lake Victoria in the North East (Whitehead, 1960 (Greenwood, 1966). Records from Lake Edward system are uncertain as they are likely to be E. pellegrin   Aswa river. There are no recent records, but the species is possibly extant (Greenwood, 1966 (Greenwood, 1956) Lake Victoria and Victoria Nile, and Lake Kyoga. The current status is unknown due to lack of recent data, but is possibly extant Clarias alluaudi (Boulenger, 1906)   Lake Albert and Albert Nile. Present status unknown due to lack of recent data S. macrops (Greenwood, 1963) Aswa river. Current status unknown due to lack of data, but possibly extant S. nigrita (Valenciennes, 1840) Lakes   Affluent rivers of Lake Victoria and Aswa river drainage (Greenwood, 1966) N. robustus (Ahl, 1935 Drainage basins of Lake Victoria, Albert, and Kyoga (Nagy & Watters, 2018) 145,188.97 87.5 77,405 > 80 N. ugandensis (Wildekamp, 1994) Drainages of Lakes Victoria and Kyoga and Aswa river (Nagy et al., 2020) 210,998.14 > 80 111,760 > 80 N. elucens (Nagy, 2021) Aringa system, Aswa drainage (Nagy, 2021 EOO, which is the smallest area that encompasses all the known, inferred or projected sites where a species is extant, was estimated by the minimum convex polygon method, while the AOO, which is the area inside the EOO occupied by the species, was estimated using the grid cell method. The two methods are described in detail in IUCN (2019b). Only hydro-basins where the species was extant (Table 1) were used in the estimation of EOO and AOO. Initially, all mapping was done in the FWMA. The FWMA uses standardized base layers and follows the standard IUCN methods for mapping freshwater species and automatically calculates EOO and AOO. While the FWMA is quick and straightforward,  (Ahl, 1924) Lake Albert, including the delta of the Semliki River and the Victoria Nile River below Murchison Falls L. centralis (Seegers, 1996) Lake Victoria and Kyoga basins (Wildekamp, 1995;Seegers, 1997) 143,376.75 20.9 82,282 28.6 L. vitschumbaensis (Ahl, 1924) Lakes Edward and George; Kazinga Channel, and affluent rivers; northern parts of Lake Victoria, and Lake Kyoga drainage Wildekamp, 1995)

Diversity and distribution of non-Haplochromis fishes
A review of the 13,656 non-Haplochromis occurrence records from GBIF and FWB, combined with literature, resulted in a total of 110 non-Haplochromis fish species belonging to 11 orders, 21 families, and 48 genera (Table 2). This number includes all species that were recognized as "extant" and "possibly extant." Out of this total, 21 species were found only in the rivers and streams, which is likely to be an underestimate as rivers are not as widely sampled as large lakes, and therefore, more sampling could reveal more species. Generally, among the major lakes, species richness was highest in Lake Victoria followed by Albert and least in the Lake Edward system (Fig. 3). A total of 12 non-Haplochromis fishes were found to be endemic: five in Lake Victoria; one in Lake Albert; one in Aswa river, two in Lake Edward; and three in the Rwenzori rivers (Sibwe, Rwimi, and Mubuku) ( Table 2). Approximately 60% of all fishes were recognized as "possibly extant," either in parts or their entire native range (Table 2). These records were more than 30 years old but were considered "possibly extant" on the assumption that there has been limited sampling in their native range. However, we also found that some species previously believed to be in Uganda had no native locus within the country. Also, recent occurrence records sited some species in areas beyond their known native range, where their presence in those areas remains uncertain (Table 2). Here, we present a summary (by family) mainly for species whose presence is uncertain or outside Uganda. A discussion for considering species "possibly extant" is included in Table 2.
Mormyridae Bonaparte 1831: Greenwood (1966) recorded 13 species under this family, while Witte and van Densen (1995) added Marcusenius rheni to the list of Lake Victoria Mormyridae, except there are no recent records for this species. In this list, Petrocephalus catostoma was believed to occur in Lake Victoria and Nabugabo and the Victoria Nile (East Coast ichthyofaunal province). Other recent occurrence records from the GBIF extended the distribution to Lakes Albert, Kyoga, George, Nabugabo, Kagera, and Nyaguo, spanning two ichthyofaunal provinces (GBIF.org, 2022). However, recent studies have shown that this species is native to the Zambezi ichthyofaunal province (Kramer et al., 2012). Being a riverine species, this new evidence suggests that presence of this species in Uganda is unlikely. In this paper, we have considered records previously attributed to this species as P. degeni Boulenger, 1906(sensu Kramer et al., 2012. Another species with uncertain distribution is Marcusenius macrolepidotus with records from Lakes Kyoga and Nakuwa. This species also has a native distribution in the Zambezi ichthyofaunal province (Froese & Pauly, 2022). Whereas fish species can be shared by two ichthyofaunal provinces, e.g., Microctenopoma damasi (Poll & Damas, 1939), Pseudocrenilabrus multicolor (Schöller, 1903), and Oreochromis leucostictus (Trewavas, 1933), which occur in both the East Coast and Nile-Sudan, possibly via River Semliki , M. macrolepidotus may not be shared with fishes in the East Coast province (Kyoga and Nakuwa) and, therefore, its presence in Uganda is unlikely. Other species, e.g., Hippopotamyrus grahami (Norman, 1928) and Mormyrus caschive Linnaeus, 1758, have records that extend their distribution beyond the known native range and need verification (Table 2).
Cyprinidae Rafinesque 1815: Greenwood (1966) listed several species under the genus Barbus Cuvier and Cloquet 1816 before it was split into the large hexaploid barbs in the genus Labeobarbus Rüppell, 1835 (Vreven et al., 2016) and small diploid barbs in the genus Enteromius Cope, 1867 (van Ginneken et al., 2017). Our review showed that records from surveys previously attributed to Enteromius cercops and E. perince need to be reconciled in line with the recent taxonomic updates. The former was previously recorded in the affluents of Lakes Victoria, Kyoga, and Edward (Greenwood, 1966). However, the work of Decru et al. (2020) showed that records previously attributed to this species in the Lake Edward system were E. perince, which were latter re-identified as E. cf. mimus, while E. cercops was synonymized with E. alberti (Poll, 1939) (Maetens et al., 2020). However, it is still not known whether the records from Lake Kyoga basin previously attributed to E. cercops conform to E. alberti. Other species, e.g., Labeo coubie Rüppell, 1832, Labeobarbus ruwenzorii (Pellegrin, 1909), and most species within the genus Enteromius, were found outside their known range and their presence in those areas remains uncertain (Table 2).
Bagridae Bleeker, 1858: Greenwood (1966) listed only two species under this family: Bagrus docmak (Fabricius, 1775) and B. bayad (Fabricius, 1775), but also noted the likelihood of a third species, B. degeni, in Lake Victoria, except that it closely resembled B. docmak and considered the two species to be one. Recently, Ferraris (2007) listed B. degeni as a distinct species occurring in Lake Victoria, which was subsequently adopted by FishBase (Froese & Pauly, 2022), but with no new observations or additional specimens. Given that there is no single record of this species from Lake Victoria, apart from the type specimens, and considering the numerous surveys conducted in the lake, its presence is unlikely and the specimens are likely to be those of B. docmak.
Amphiliidae Regan 1911: One Amphilid species, Amphilius jacksonii Boulenger, 1912, was recorded in Uganda by Greenwood (1966), but three other species have subsequently been recorded: Amphilius kivuensis, A. lujani Page, 2015, andZaireichthys rotundiceps (Hilgendorf, 1905) (see Seegers et al., 2003;Thomson & Page, 2010;Thomson et al., 2015). However, the presence of A. kivuensis in Uganda has not been confirmed. Morphologically, A. kivuensis, which is more of a Congo ichthyofaunal province species, although also reported in upper Kagera drainage in Rwanda (Froese & Pauly, 2022), shares a close resemblance with A. uranoscopus (Pfeffer, 1889), which is reported to be widespread in East and Central Africa (Froese & Pauly, 2022). Whereas A. uranoscopus too has not been confirmed in Uganda, with its earlier specimens considered by Decru et al. (2020) as A. cf. kivuensis, in terms of biogeography, it is more likely than A. kivuensis. The specimens in Uganda previously attributed A. kivuensis might turn out to be A. uranoscopus , but this needs further investigation. Another species that needs further investigation is A. jacksonii Boulenger, 1912. This species is known from the Lake Edward drainage. Several occurrence records from the GBIF (GBIF.org, 2022) suggest that the distribution extends to the Kyoga drainage, but this is unlikely and these records could be Amphilius lujani (Froese & Pauly, 2022). Meanwhile, two new possible Amphilids from the Lake Edward system are considered in this study: A. sp. Bwindi and A. cf. kivuensis , bringing the total possible number of Amphilids in Uganda to five species.
Mastacembelidae Swainson 1839: This family has one representative species in Uganda, Mastacembelus frenatus Boulenger, 1901, which is known from the Lakes Victoria and Kyoga basins and the Aswa river system (Greenwood, 1966). There is, however, one recent record from the Murchison Nile delta, which places this species in the Lake Albert system. This is the first record of this species in the Lake Albert system, although there is another specimen from the system that is currently under investigation to establish whether it is a true Mastacembelus record (Jos Snoeks, personal communication). Unlike other records whose presence was designated as uncertain, we reasoned that the species cannot be confused with any other in the field and must therefore be a true record of Mastacembelus. Nonetheless, no specimen was kept for this record, hence the need for resampling the area to confirm the record.
Procatopodidae Fowler 1916: This family represents the riverine killifishes, commonly known as the African lampeyes, originally attributed in old literature to family Cyprinodontidae (see Greenwood, 1966). In this family, Lacustricola pumilus was originally (and is still wrongly) identified in many surveys as Aplocheilichthys pumilus (Boulenger, 1906). A revision of the genus Aplocheilichthys Bleeker, 1863 and its synonymization with the genus Lacustricola Myers, 1924(Huber, 1999 implies that A. pumilus lacks a native locus in Uganda because L. pumilus is endemic to Lake Tanganyika drainage (Froese & Pauly, 2022). In the Lake Edward system, most of the records previously attributed to A. pumilus are considered here as L. bukobanus (Ahl, 1924)  , while those in the Lakes Victoria and Kyoga systems are considered as L. margaritatus Nagy & Watters, 2022(see Nagy & Watters, 2022.

Extent and range of the non-Haplochromis species
The estimates of EOO and AOO for the non-Haplochromis fishes are shown in Table 2. Generally, the EOO and AOO for most species (59) were less than 50% of the global range; in fact, 29 species had EOO and AOO values less than 10% of the global range, implying that most of the species have a relatively wider distribution outside Uganda. According to the IUCN red listing criterion B, the minimum EOO for a species to trigger extinction threat category is 20,000 km 2 (IUCN 2019a). Table 2 shows that only 29 fish species (26%) had EOO values less than 20,000 km 2 . The EOO and AOO for 25 fish species (about 22% of the total non-Haplochromis fishes) could not be quantitatively ascertained because their presence was designated as possibly extant, requiring intense surveys to ascertain the current status of these species in their native habitats.

Discussion
The purpose of this study was to review and update the literature on the distribution and range of non-Haplochromis fishes in Uganda, building on the recent work by Decru et al. (2020) for the Lake Edward system. Although no new field survey was conducted, this study was possible because of the large amounts of data that have recently been made available through GBIF and the Freshwater Biodiversity Portal for Uganda. This study recorded a total of 110 non-Haplochromis species to occur in Uganda in different hydro-basins. This number implies 20 additional species to the original list by Greenwood (1966) that comprised 90 non-Haplochromis species with a valid native distribution in Uganda. The study also revealed a total of 17 non-Haplochromis species that occur within Uganda, but extending to areas outside their known native range documented in Greenwood (1966), FishBase (Froese & Pauly, 2022), and Eschmeyer's Catalog of Fishes (Fricke et al., 2022b). This finding is not necessarily surprising given the recent (especially within the last 2 decades) numerous studies and surveys, which largely remained unpublished until the work of Natugonza and Musinguzi (2022) and Musinguzi et al. (2023). Given that these new records have not been affirmed by examination of their specimens, their occurrence in the new localities is still uncertain. Generally, the specimens for these new occurrences are not available. The NaFIRRI has conducted most of the surveys that recorded most of the fishes outside their known native range, but without systematic preservation of specimens. This was a huge omission. A systematic search and collection of specimens from these new localities is urgent. This study provides a foundation from which future surveys and field expeditions can be prioritized.
In terms of diversity, more species were recorded in large water bodies or hydro-basins, which is also expected. Distribution and diversity, among others, are influenced by the size of the water body, surface area of the drainage basin, and topographical features (Franklin, 2010;Pelayo-Villamil et al., 2015;Trigal & Degerman, 2015). The higher species richness observed in Lake Victoria, for example, is expected because of its size and habitat heterogeneity, with fish being able to evolve and adapt to the different habitats and ecological niches (Wagner et al., 2012). However, it is also important to note that these large water bodies have been the major focus for numerous fisheries investigations and surveys, with limited attention to small water bodies such as swamps, rivers, streams, and minor lakes. These small ecosystems have been shown to harbor large assemblages of fish (Tibihika et al., 2015), and therefore, it is possible that with extended sampling and targeted surveys covering these systems, more species could be recorded beyond what is listed in this study.
Aside from species diversity, limited investigations on small ecosystems can also be seen in the estimated geographical range for some species. Table 2 shows many species without values of EOO and AOO because these species were designated as "possibly extant" due to a lack of recent data. IUCN guidelines for mapping species suggest that a species should only be designated as "extant" if it has been observed or recorded within the last 30 years (Table 1; IUCN, 2019a). Some non-Haplochromis species were described from their type localities and have not been recorded since then. Nevertheless, these fishes were designated as possibly extant as fish may not just disappear or move to another area, but instead, there have not been targeted searches for these species (Jos Snoeks, personal communication). This delimitation will have implications for the red listing as these fishes whose range could not be quantitatively ascertained can only be listed as Data Deficient (IUCN, 2019a). This means, even with the large amounts of data in GBIF and FWB, more than 50% of non-Haplochromis fishes are generally still data-deficient. Still, this designation is important as it points researchers to areas that are in need of data collection (IUCN, 2019a).
The national EOO and AOO for most fishes were less than 50% of their global range, a finding that was consistent with expectation. It was generally expected that since Uganda shares a small portion of both the Nile-Sudan and East Coast ichthyofaunal provinces (Fig. 1), the proportion of the species' national range to the global range would be substantially low. This expectation was consistent with fishes especially from the Lake Albert system, i.e., Lake Albert, Murchison, and Albert Nile, which, apart from the few endemic species, shares similar fish fauna with Western and Northern Africa (Nile-Sudan ichthyofaunal province). For the East Coast fauna, the proportion of national range to global range partly deviated from expectation, where majority of the fishes had national EOO and AOO greater than 50% of the global range. This finding was unexpected, but also not entirely surprising because the fish fauna of Lake Victoria resembles that of the Lakes Kyoga and Edward systems. For instance, all fish families occurring in the Lake Edward system occur in Lake Victoria, although four families that are native to Lake Victoria (i.e., Alestidae, Mastacembelidae, Mochokidae, and Schilbeidae) are absent in the Edward system . The three systems are often collectively referred to as the Lake Victoria region (e.g., Greenwood, 1966;Natugonza et al., 2021). The connection between Lake Victoria and Kyoga is clear: the two systems are directly connected by Victoria Nile. However, the connection between Lake Victoria and the Lake Edward system is not straightforward, although it is suggested that the two systems could be connected through the marshy areas on the Katonga and Rwizi rivers . Therefore, the large extent of the Lake Victoria region, coupled with a high species endemism, may explain why the Ugandan non-Haplochromis fishes of the East Coast province have a big national distribution and range approaching the global range.

Conclusions and recommendations
This study presents the first quantitative assessment of the range (in terms of geographical extent of occurrence and area of occupancy) of the non-Haplochromis fishes in Uganda. This information has potential application in the IUCN Red List assessments, for which conservation status of the majority species is outdated and many others remain unevaluated due to lack of quantitative information on their range. By comparing the values of the national range with the global range, this study also provides valuable information that can be used in the assessment of conservation status of fishes at a national scale, which, in Uganda, is still lacking. Furthermore, the data underscores the importance of actions to halt degradation of fish habitats and overexploitation as the geographical restrictedness, shown for the majority of the non-Haplochromis species native to systems in the East Coast ichthyofaunal province, means that the fishes have a limited chance to be conserved elsewhere.
However, this review also shows that immense data gaps still exist on the species occurrence, despite the huge effort by the National Fisheries Resources Research Institute to mobilize and publicly share data, with over 30 datasets shared via the GBIF to date. This gap can easily be addressed by targeted collections, starting with the locations where fishes were designated as "possibly extant" or "presence uncertain." The information presented here will help in prioritizing resources during future surveys to fill this data gap.
While data accessibility is fairly being addressed, especially for the fishes of Uganda, through numerous efforts such as the GBIF, FishBase, and Freshwater Biodiversity Portal for Uganda, issues still remain on the quality of the data and authenticity of the records. Many records suggest occurrence of fish species beyond their known native range, yet there are no voucher specimens to ascertain the status of the species in the new localities. For species such as Nannocharax niloticus (Joannis, 1835), even when no recent specimens exist, we were able to ascertain their presence based on the photographs of live fish taken during recent surveys (see https:// fresh water biodi versi ty. go. ug/ speci es/? code= SP84Q L7). With voucher materials and photographs lacking for most of the recent records, it is impossible to verify the range for the species designated in new localities. We recommend that in future, where voucher specimens cannot be preserved and deposited at a museum, at the minimum, photographs of live fish should be taken and properly curated (with date and location details).
Another gap emanates from the bias in the choice of localities for fish diversity surveys. Figure 2, for instance, shows species richness to be substantially higher in hydrological basins with large water bodies (with established commercial fisheries) than small water bodies (small lakes, rivers, streams, and swamps). While it is possible that large water bodies host large assemblages of fish species, it is also possible that the poor species richness in small water bodies could be due to sampling bias, where minor aquatic systems-in terms of commercial fisheries-are left out of diversity surveys. Yet, these minor systems have been shown to have higher species richness relative to their size, harbor large assemblages of endemic species, are refugia for species that are threatened in larger water bodies, and rank highest in terms of the need for site-based conservation (Basooma et al.,  2022). The implication is that the total number of non-Haplochromis species in Uganda could be much higher than presented here if these small water bodies were sampled with same rigor as large systems. Future surveys, therefore, should consider these small (fragile) ecosystems, which are also the most vulnerable to habitat and environmental changes (Canning & Waltham, 2021).
Finally, the gaps above may be best addressed with dedicated field biologists being trained to become the next-generation biodiversity experts in the country/ region. The surveys that have collected most of the data used in this review have been conducted through collaborations with international fish taxonomy experts, but over the years, these collaborative efforts have diminished, with limited capacity of local scientists to continue the work on fish taxonomy. While there have been efforts to improve capacity of local scientists in fish taxonomy, such as the training sessions offered by FishBase, there is a need to intensify these efforts, with local scientists and technicians being trained on fish species collection and documentation, taxonomic identification, and monitoring and collection management. These trainings should preferably be conducted in the region to benefit a bigger audience.