Results

Species composition and rarity

A total of 288 species were recorded from 81 water bodies (i.e., 7 large lakes, 37, small lakes, and 37 rivers) (Fig 1 & Appendix S2). Of the 288 species, 163 were haplochromine cichlids and 125 non-haplochromine. The species accumulation curve increased at a low rate after 20 water bodies (Fig 2). All haplochromine cichlids were recorded in 45 water bodies compared with 75 for non-haplochromines. Species richness was highest in large lakes dominated by Lake Victoria with 175 species followed by Albert, Kyoga, and George (Fig 1, Appendix S2). Kruskal Wallis test showed that waterbody categories differed significantly in mean species richness, and the IUCN status categories (Table 1). However, the nature of a waterbody had a weak effect on the variation in species richness and IUCN categories (Table 1). Post hoc multiple comparisons showed that large lakes were significantly different from rivers and small lakes for all IUCN categories and species richness. However, except for the vulnerable category, rivers and small lakes were not significantly different (Table 1).
Of the 163 haplochromine cichlids, 85 species were found in utmost one waterbody compared with 31 non-haplochromine species (Fig 3). Lake Victoria had 71 rare species, where 11 were critically endangered haplochromine cichlids and one non-haplochromine species (Xenoclarias eupogon ) (Appendix S3). Two near threatened species (H. labiatus and H. oregosoma ) were only found in the interconnected system of lakes George, Edward, and Kazinga Channel.Clarias gariepinus was found in 42 water bodies followed byEnteromius kerstenii , O. niloticus (33), andProtopterus aethiopicus (32). Astatoreochromis alluaudiwas recorded in 30 water bodies, H. nubilus (28), H. lividus (16), H. argenteus (14), H. phytophagus (13),H. obesus (12), and H. parvidens in (11).Oreochromis esculentus was observed in 19 water bodies, O. variabilis (15), and L. victorianus (8). Based on IUCN categories, no vulnerable non-haplochromines was found in Lake Victoria; however, 8 of the 25 vulnerable haplochromine cichlids were only recorded in the lake (Appendix S3). Labeobarbus ruwenzorii andL. alluaudi were only found to Lake Edward and rivers Ivi, Rutushuru, Rwimi, Sebwe, Rwenzori, and Mubuku). Nothobranchius taeniopygus and Synodontis macrops were only found in Aswa River. H. melanopterus was recorded in Lake Kasodo whereasH. commutabilis and H. ampullarostratus were only found in Lake Kachiira (Appendix S3). Of the endangered fish species,Astatotilapia desfontainii was only recorded in lakes Bisina, Victoria, and Victoria Nile; H. simpsoni was only found in lakes Kyoga, Nabugabo, and Kayanja, while Lates macrophthalmus was only recorded in Lake Albert. H. beadlei , a critically endangered species, was found in lakes Nabugabo and Victoria, and H. grantiin Lake Victoria and River Kagera.
Among the 3 waterbody categories, large lakes recorded the highest number of non-shared species (110), followed by rivers (19), and the small lakes (6) (Fig 4). The species richness in waterbody categories differed significantly (Kruskal Wallis: χ2= 18.6, df =2, p=0.001, η2 = 0.2). The nMDS had a stress level of 0.14, with large and small lakes closely clustered except rivers (Appendix S4). Haplochromine cichlids were closely clustered, exceptH. latifasciatus, H. victoriae, A. alluaudi, H. nubilus, and H. schubotzi , while non-haplochromine species were widely distributed (Appendix S5). Analysis of similarity (ANOSIM) among waterbody categories showed significant differences (R=0.24, p<0.001). Similarity percentage (simper) analysis showed that large lakes differed from small lakes (93.5%) and rivers (97.5%), while small lakes differed from rivers by 96.2%. Lates niloticus and O. niloticus contributed the highest average percentage variation between large lakes and rivers. Similarly, L. niloticus had the highest contribution between large and small lakes (Appendix S6). In contrast,A. alluaudi and C. gariepinus had the highest contribution to the variation between rivers and small lakes.

Conservation Priority Index (CPIw)

CPIw was computed for 38 lakes, where 7 were large lakes and 31 small lakes. Small lakes had a mean CPIw of 2.4 (SD 4.6) compared with 0.05 (SD 0.04) for large lakes. A significant difference between the mean CPIw value for large and small lakes was observed (Welch 2-Sample t-test: t = -2.8, df = 30, p-value = 0.008, d=0.7). The surface area of the lakes had a large effect on the CPIw values. Highest CPIw values were recorded for lakes Manywa followed by Kayanja, Gigate, Agu, Naragaga, Kawi, and Nabugabo (Fig 5 & Appendix S7). Low values were mostly recorded for large lakes and highest for small lakes.