Received: 24 January 2020 | Revised: 19 November 2020 | Accepted: 28 November 2020

DOI: 10.1111/jai.14167

ORIGINAL ARTICLE

Genetic diversity and population structure of selected

lacustrine and riverine populations of African catfish,
Clarias gariepinus (Burchell, 1822), in Kenya

George W. Alal1 | James E. Barasa1 | Emily J. Chemoiwa2 | Boaz Kaunda-­Arara1 |

Peter Akoll3 | Charles Masembe3

1
Department of Fisheries and Aquatic
Sciences, University of Eldoret, Eldoret, Abstract
Kenya Determining the genetic characteristics of natural fish stocks is useful for conserva-
2
Department of Biological Sciences,
tion and aquaculture programs. For African catfish, Clarias gariepinus, genetic char-
University of Eldoret, Eldoret, Kenya
3
Department of Zoology, Entomology
acterization could help identify populations suitable as brood stock for culture, and
and Fisheries Sciences, College of Natural those in need of conservation. This study determined the genetic diversity, popu-
Sciences, Makerere University, Kampala,
Uganda
lation structure, and demographic history of C. gariepinus from Lakes Victoria (LV),
Kenyatta (LKE), Kamnarok (LKA), and Rivers Nyando (NR), Tana (TR) and Sosiani (SR)
Correspondence
James E. Barasa, Department of Fisheries
in Kenya. Using 128 DNA sequences of D-­loop control region, 34 haplotypes were
and Aquatic Sciences, University of Eldoret, recovered, of which 79.4% were singletons. Only 7 haplotypes were shared between
P.O Box 1125-­3 0100, Eldoret, Kenya.
Email: [email protected]
sites, implying little gene flow between sites. Number of haplotypes was highest in
LKE and NR populations and lowest in SR. Haplotype diversity was highest in LV,
Funding information
National Commission for Science,
and lowest in SR, while, nucleotide diversity was highest in LKA and lowest in LV.
Technology and Innovation; National Phylogenetic analyses revealed five clusters: Lakes Victoria, Kamnarok and Kenyatta,
Research Fund, Kenya; Austrian Partnership
Programme in Higher Education and
and Rivers Tana and Nyando, from both maximum likelihood tree and minimum span-
Research for Development (APPEAR) ning network. This, together with significant FST values among the sites imply popula-
STRECAFISH Project; National Research
Fund
tion differentiation. Mismatch distributions were multi-­modal in LKA, LKE, NR and
TR, signifying demographic equilibria. Neutrality tests Tajima`s D values for the sam-
pled populations were negative and significantly different, suggesting stable popu-
lations. These results show the existence of genetically distinct populations of C.
gariepinus that require spatially explicit management actions such as reducing fishing
pressure, pollution, minimizing habitat destruction and fragmentation for sustainable
utilisation of stocks.

KEYWORDS

Clarias gariepinus, genetic differentiation, haplotypes, mitochondrial DNA control region,

multi-­modal distributions

1 | I NTRO D U C TI O N Clarias gariepinus inhabits both riverine and lacustrine environ-

ments, apart from streams and swamps, where it plays an important
The African sharp-­tooth catfish, Clarias gariepinus (Burchell 1822), ecological role as a predator (Corbet, 1961). The species grows to
has a Pan-­African distribution, ranging from the Nile River Basin to total lengths of up to 130 cm and weighs over 30 kg (Bruton, 1979;
West Africa and from Algeria to Southern Africa (Cambray, 2003). Cambray, 2003). Its wide distribution is partly due to an omnivorous

J Appl Ichthyol. 2021;00:1–12. wileyonlinelibrary.com/journal/jai© 2021 Wiley-­VCH GmbH | 1

2 | ALAL et al.

feeding habit and ability to tolerate a wide range of environmen- important C. gariepinus fishery, exploited for food and as source of
tal variables (Bruton, 1979). In this regard, the species is especially brood stock for artificial propagation at hatcheries. Sampling of C.
successful in large rivers, where it swims in inundated areas for gariepinus from the sites therefore provided an extensive coverage
breeding, but also inhabits different habitats of varying sizes and of the geographical spread of the species in Kenya, in addition to
conditions. The species has high fecundity (Owiti & Dadzie, 1989) the fact that the sites have varying types and levels of anthropo-
and fast growth rates (Bruton, 1979), which make it an excellent genic influences on the fishery. It was therefore possible to deci-
aquaculture species, cultured for food and nutrition security and pher the effect of these anthropogenic factors on genetic diversity
livelihoods. and population structure of C. gariepinus in different habitats in
In East Africa, juveniles of Clarias are also used as live bait for Kenya. The sampling station(s) within sites were selected randomly
Nile perch, Lates niloticus using long line and hooks in Lake Victoria and fish collected using a combination of gill netting, hook and line,
(Chitamwebwa et al., 2009; Ngugi et al., 2005). In this regard, there and traps and seine nets. Fish samples were identified using a field
is indiscriminate collection of Clarias juveniles from natural habitats identification guide (Witte & van Densen, 1995). Pectoral fins were
for use as live bait by fishermen, a practice which increases fishing clipped from 128 individuals of C. gariepinus (Table 1) as source of
pressure on natural populations, endangers the fishermen's health DNA using sterilized surgical blade and immediately preserved in
and often fails to meet the numbers of live bait required. In order to 95% ethanol in clean labelled cryovial tubes and kept in the freezer
reduce this overexploitation, artificial propagation of C. gariepinus at −20°C before DNA extraction.
at hatcheries has been recommended (Kaufman & Ochumba, 1993),
as a source of Clarias juveniles, which also generates income and
livelihood for farmers (Barasa et al., 2017). However, the efficiency 2.2 | DNA extraction
of artificial propagation of the species at hatcheries is reduced by
poor survival of C. gariepinus larvae (Sulem et al., 2006). Recent Fin clips were thawed, macerated, tissue lysed and incubated at 40°C
studies to address this challenge have focused on genetic charac- overnight for digestion in a shaking water bath in preparation for DNA
terization of C. gariepinus populations in order to identify suitable extraction. Total genomic DNA was extracted from approximately
populations for use as brood stock (Barasa et al., 2014, 2016; 2017; 25 mg of fin clip tissue, using the Qiagen DNeasy Tissue Kit (Qiagen
Ojiambo, 2015) and those populations in need of conservation GmbH, Germany); following the manufacturer`s protocol with minor
(Barasa et al., 2017). modifications. Centrifugation for spinning digested content was set
Preservation of genetic diversity is an important goal of at 10,000 revolutions per minute (rpm) except for the final, which
conservation programmes, since genetic diversity influences was done at 14,000 rpm. Elution was done with 150 μl of AE elution
adaptation and persistence of a species in the environment buffer repeated twice for maximum yield. Presence and quality of the
(Lande, 1988). Furthermore, for C. gariepinus, genetic diversity extracted genomic DNA were visualized using 2% agarose gel elec-
correlates with some fitness traits (Barasa, 2018), and therefore trophoresis. The DNA was stored at −20°C prior to further analysis.
identifying populations of high genetic diversity is crucial in the
sustainable utilization of the species. As a follow up to previous
studies on the genetic characterization of C. gariepinus in Kenya, 2.3 | PCR amplification
this study increased the geographical coverage of C. gariepinus
populations, for improved culture and conservation of the spe- The hypervariable mitochondrial DNA control region (D-­
loop)
cies in East Africa. The study determined genetic diversity and was Polymerase Chain Reaction (PCR) amplified in a thermal cy-
population structure of C. gariepinus from selected lacustrine and cler (ABI 9700) using the following primers: forward primer LN20
riverine habitats in Kenya, using mitochondrial DNA of the D-­l oop (5′-­ACCACTAGCACCCAAAGCTA-­3′) and reverse primer HN20
control region. (5′-­GTGTTATGCTTTAGTTAAGC-­3′) (Benatchez & Danzmann,
1993). PCR reactions were carried out in a 25 μl volume micro-­
centrifuge vials. The mixtures contained 0.5 μl of 10 pmol of each
2 | M ATE R I A L S A N D M E TH O DS primer (HN20 and LN20), 12.5 μl of AmpliTaq® Gold 360 Master
Mix (Applied Biosystems, USA), 9.5 μl of double distilled water and
2.1 | Fish samples 2 μl of genomic DNA template.
The PCR thermal cycling conditions were as follows: initial dena-
Samples of C. gariepinus were collected from six different sites in turation at 95°C for 1 min, 35 cycles of denaturation at 94°C for 30 s,
Kenya from August, 2016 to May, 2017: three lacustrine sites in- primer annealing at 49.3°C for 1 min, initial extension at 72°C for 50 s,
cluding Lakes Victoria (LV), Kamnarok (LKA) and Kenyatta (LKE), followed by a final extension of 72°C for 10 min. All PCR reactions
and three riverine sites including Rivers Nyando (NR), Sosiani were performed in a Thermal Mastercycler (Eppendorf, Germany)
(SR), and Tana (TR)) (Table 1 and Figure 1). The six sites were se- PCR system 9700. In every PCR, a DNA free template was included as
lected based on considerations such as habitat size, climatic and a negative control. The PCR products were visualized on gel electro-
hydrographic conditions, and the fact that all the sites constituted phoresis (100v, 30 min) in 2% agarose gel, and size compared against
ALAL et al. | 3

lambda DNA ladder. The Amplicons were purified using QIAquick PCR Akaike Information Criterion (AIC) using JModelTest v2.1.10 (Darriba
purification kit (Qiagen GmbH, Germany) following manufacturer`s et al., 2012). Initial tree(s) for the heuristic search were obtained au-
instructions. The products were visualized in 2% Nusieve agarose gel tomatically by applying Neighbor-­Joining and BioNJ algorithms to a
stained. The purified products were cycle-­sequenced and analysed on matrix of pairwise distances estimated using the ML approach, and
ABI 3,100 automated-­sequencer. One hundred and twenty-­eight PCR then selecting the topology with superior log-­likelihood value (Darriba
products were selected with correct band, good quality and sent to et al., 2012) performed in PAUP* v4.0a167 (Swofford, 2002). The tree
Macrogen Laboratory (Netherlands) for sequencing. was visualized in Interactive tree of life (iTOL) (Letunic and Bork, 2016).
Minimum spanning network showing the phylogenetic relation-
ships between haplotypes of C. gariepinus from lacustrine popula-
2.4 | Data analysis tions (LKA, LKE and LV) and riverine populations (NR, TR and SR)
was drawn using MEGA X (Kumar et al., 2018) borrowing labels
The chromatograms for the forward and reverse DNA strands were from the ML tree, with a median-­joining approach. The circle size
refined and consensus sequences generated from contigs in the is proportional to the haplotype frequency. Moreover, haplotypes
Sequencher program v5.4.6 (Gene Codes, 2016). The consensus were identified by the number of each haplotype and their resultant
sequences were edited in BioEdit v7.1.11 (Hall, 2005) and aligned branch length represent the number of mutation steps.
using the complete alignment application tool in Clustal X v 2.0 The account of demographic history of C. gariepinus population
(Larkin et al., 2007). The sequences were compared with nucleotide expansion and decline within and between sampling sites was simu-
sequences in the GenBank using the Basic Local Alignment Search lated in (coalescent simulations) DnaSP v6.11.01 (Rozas et al., 2017).
Tool (BLAST) to confirm species identity. Thereafter, the aligned se- Additionally, mismatch distribution analysis, neutrality tests; Fu`s Fs
quences were used for phylogenetic and genetic diversity analyses (Fu, 1997) and Tajima`s D (Tajima, 1989) were estimated in DnaSP
in MEGA X (Kumar et al., 2018). v6.11.01 (Rozas et al., 2017), to discern demographic changes
Genetic diversity within C. gariepinus populations was quantified amongst the C. gariepinus populations.
as the number of haplotypes, haplotype diversity (h), nucleotide di-
versity (π), number of singletons (with percentages), number of shared
haplotypes (with percentages), polymorphic sites in DnaSP v6.11.01 3 | R E S U LT S
(Rozas et al., 2017). Consequently, the distribution and identification
of shared haplotypes was performed in DnaSP, according to segregat- 3.1 | Genetic diversity of C. gariepinus inferred from
ing sites. JModelTest v2.1.10 (Darriba et al., 2012) was used to deter- mtDNA D-­loop
mine the most likely model of evolution for the mtDNA sequences.
Hierarchical distribution of genetic structure of C. gariepinus from Sequences were submitted to the GenBank database and are pub-
the six sites was analysed by standard analysis of molecular variance licly available under their respective accession numbers (Table 1).
(AMOVA) computations (Excoffier, 2004) using haplotype pairwise Genetic diversity of C. gariepinus based on Haplotype diversity (h)
differences performed to partition genetic variation in Arlequin v3.5 and Nucleotide diversity (π) was generally higher in lacustrine popula-
(Excoffier & Lischer, 2010). The AMOVA partitioned total variance into tions than riverine populations (Table 2). The lakes’ and River Nyando
covariance components due to within populations, among populations fish populations exhibited high h, which were not significantly dif-
and among group variances. The covariance computed fixation indices ferent from each other (LV = 0.911 ± 0.042; LKA = 0.774 ± 0.071
included FSC, FST and, FCT respectively as defined by Wright (1965). LKE = 0.866 ± 0.059 and NR = 0.830 ± 0.068) compared to SR
Phylogenetic relationships among lacustrine and riverine popula- and TR populations (SR = 0.518 ± 0.122 and TR = 0.620 ± 0.099).
tions of C. gariepinus mtDNA control region sequence haplotypes were Consequently, Nucleotide diversity values were higher in LKA and
calculated using Maximum likelihood (ML) methods. The best-­fit model LKE populations (LKA = 0.118 ± 0.056 and LKE = 0.109 ± 0.059), than
for DNA evolution was General Time Reversible (GTR + G) selected by NR, SR and TR populations (NR = 0.060 ± 0.041; SR = 0.031 ± 0.027

TA B L E 1 Sampling sites, Population codes, coordinates of sampling sites, sample sizes and GenBank accession numbers of sequences for
128 samples of C. gariepinus collected from different lacustrine and riverine habitats in Kenya

GenBank Sequence
Sites Population code Coordinates Sample size Altitude accession numbers

Lake Victoria LV 1°00′S, 33°00′E 20 1,135m a.s.l MK014577-­MK014596

Lake Kamnarok LKA 0°37′N, 35°37′E 20 1,058m a.s.l MK014597-­MK014616
Lake Kenyatta LKE 2°24′S, 40°40′E 23 10m a.s.l MK014617-­MK014639
River Nyando NR 0°09′S, 34°55′E 23 1,156m a.s.l MK014640-­MK014662
River Sosiani SR 0º32´N, 35º13´E 23 2,084 a.s.l MK014663-­MK014685
River Tana TR 1°30′S, 40°0′E 19 1,975m a.s.l MK014686-­MK014704
4 | ALAL et al.

F I G U R E 1 Map of field sampling locations for Clarias gariepinus from six different sites in Kenya. The three lakes sampled included Lakes
Victoria (LV), Kamnarok (LKA), and Kenyatta (LKE). The three rivers included Rivers Nyando (NR), Sosiani (SR) and Tana (TR)

TA B L E 2 Genetic diversity estimates of C. gariepinus from six aquatic systems in Kenya from 570 bp of mtDNA D-­loop control region
sequences: LV = Lake Victoria, NR = River Nyando, LKA = Lake Kamnarok, SR = River Sosiani, LKE = Lake Kenyatta and TR = River
Tana; h ± S.D (the haplotype diversity ± Standard Deviation), π ± S.D (the nucleotide diversity ± Standard Deviation), Singletons, shared
haplotypes, Theta S (ϴS), and Theta Pi (ϴπ). Numbers in parentheses are percentage values of the respective variables

Sites

Parameters LV LKA LKE NR SR TR

Sample Size 20 20 23 23 23 19
No. of Haplotypes 10(20.41) 8(16.33) 11(22.45) 11(22.45) 4(8.16) 5(10.20)
h ± S.D 0.911 ± 0.042 0.774 ± 0.071 0.866 ± 0.059 0.830 ± 0.068 0.391 ± 0.125 0.620 ± 0.099
π ± S.D 0.008 ± 0.003 0.118 ± 0.056 0.109 ± 0.059 0.060 ± 0.041 0.031 ± 0.027 0.099 ± 0.056
Singletons 3(11.11) 6(22.22) 9(33.33) 5(18.52) 2(7.41) 2(7.41)
Shared Haplotypes 7(31.82) 2(9.09) 2(9.09) 6(27.27) 2(9.09) 3(13.64)
Polymorphic Sites 24 331 336 277 158 315
Theta S (ϴS) 8.174 132.761 125.447 87.515 52.292 126.176
Theta Pi (ϴπ) 0.015 0.261 0.247 0.168 0.097 0.246
ALAL et al. | 5

and TR = 0.099 ± 0.056). Lake Victoria population had the lowest C. gariepinus had 34 haplotypes and 5 clusters (Lake Victoria, River
nucleotide diversity (LV = 0.008 ± 0.003) (Table 2). Nyando, Lake Kamnarok, Lake Kenyatta and River Tana clusters;
In the 570 bp of sequences of mtDNA control region, 423 segre- Figure 3). The most abundant haplotype, from which all other haplo-
gating (polymorphic) sites defined 34 haplotypes. Up to 79.41% (27 types radiated was haplotype 1, consisting of 45 samples (Figure 3).
out of 34) of the total number of haplotypes occurred as singletons, A total of 29 mutational steps were observed between the haplo-
while 20.59% were shared between populations. Lacustrine and types. Analysis of molecular variance (AMOVA), revealed significant
Nyando river populations (LV n = 10, LKA n = 8, LKE n = 11 and NR variation of 17.22% occurring among populations within groups and
n = 11 respectively) revealed a higher number of haplotypes than SR 86.69% occurred within populations, while non-­significant variation
(n = 4) and TR (n = 5) (Table 2). of −3.91% was noted among groups (Table 4).
There was a significant genetic differentiation (nine out of fif-
teen pairwise comparisons) among populations (FST Pairwise com-
3.2 | Population structure of C. gariepinus inferred parisons, Table 4). The LV population was significantly different from
from mtDNA D-­loop LKA, LKE and SR (p < .05), while both LKA and LKE were signifi-
cantly different from riverine populations (p < .05) (Table 4).
The Maximum Likelihood tree revealed five clades: Lake Victoria,
River Nyando, Lake Kamnarok, Lake Kenyatta and River Tana clus-
ters (Figure 2). Most of the samples from SR clustered with sam- 3.3 | Demographic history of C. gariepinus inferred
ples of LV, with haplotype 1 having the highest number of samples from mtDNA D-­loop
(n = 45). Lake Victoria cluster had the highest number of haplo-
types and comprised of all the lacustrine and riverine samples. Lake Mismatch distribution of observed number of pairwise differences
Kamnarok cluster consisted of haplotypes from LKA, LV, LKE, NR between haplotypes of C. gariepinus are presented for individual
and TR, while Lake Kenyatta cluster had samples from LKA, LKE and populations. Lake Victoria (LV) population showed uni-­modal shaped
NR (Figure 2). River Nyando cluster had haplotype 34 as the sole distribution (Figure 4). The reconstructed demographic history
sample delineating this population (Figure 2; Table 3). of Lake Kamnarok (LKA) revealed a multi-­modal shape (Figure 5).
Consistent with the maximum likelihood tree (Figure 2), the min- On the other hand, the reconstructed demographic history of
imum spanning network for the relationship between haplotypes of Lake Kenyatta (LKE) C. gariepinus population revealed multi-­modal

F I G U R E 2 Maximum likelihood
tree for 34 haplotypes of 128 samples,
based on General Time Reversible
(GTR + G) model for the Lacustrine and
Riverine populations of C. gariepinus
from six different sampling sites in
Kenya. The labelling is same as for the
median network (Figure 3). Numbers in
parentheses are frequency of individuals
in LV, LKA, LKE, NR, SR and TR samples.
Numbers on the nodes represent
percent bootstrap values, based on
1,000 bootstrap iterations. Bootstrap
values below 70% were excluded. Clarias
macrocephalus was the out-­group
6 | ALAL et al.

TA B L E 3 Analysis of Molecular Variance (AMOVA) among Lacustrine and Riverine C. gariepinus haplotypes

Source of Variation df Sum of Squares Variance Components % Variation Fixation Index p -­values

Among Groups 2 107.099 −0.723 −3.91 FCT = −0.039 .603ns

Among Populations within 3 250.681 3.185 17.22 FSC = 0.166 .001***
groups
Within Populations 122 1956.369 16.036 86.69 FST = 0.133 .001***
Total 127 2,314.148 18.498

Note: Not Significant = ns, *p ≤ .05, **p ≤ .01, ***p ≤ .001.

F I G U R E 3 Minimum spanning network showing the relationship between haplotypes of Clarias gariepinus from lacustrine populations
(LKA, LKE and LV) and riverine populations (NR, TR and SR). Size of the circle is proportional to the haplotype frequency. Haplotypes are
identified by the number of each haplotype. Branches represent the number of mutational steps

distribution shape (Figure 8). Finally, River Tana showed a uni-­modal

TA B L E 4 FST Pair wise comparisons (below diagonal) and
associated P value (above diagonal) of 128 samples of C. gariepinus mismatch distribution shape (Figure 9).
(Burchell, 1822) from six different sampling locations in Kenya, Among lacustrine populations, LKA and LKE had higher substi-
based on 570 bp of mitochondrial D-­loop control region sequences tutions (471 and 463 respectively) than LV population (29). Riverine
populations showed high substitutions with TR and NR carrying
Associated p values
higher values (441 and 323 respectively) than SR population (193). In
LV LKA LKE NR SR TR comparison, more substitutions were recorded in fish samples from la-
FST LV .000 .000 .775 .018 .135 custrine ecosystems (963) than in riverine ecosystems (957) (Table 5).
LKA .251* .144 .000 .000 .000 Raggedness indices in riverine populations were higher in SR
LKE .285* .038 .000 .000 .000 and TR, exhibiting values of 0.360 and 0.399, respectively than NR

NR −.008 .152* .179* .405 .739 with a value of 0.030. The lacustrine populations had a higher rag-
gedness index in LKA population (0.207) while LV and LKE revealed
SR .011* .252* .280* −.003 .207
lower values of 0.042 and 0.038 respectively. Tajima`s D values for
TR .016 .124* .152* −.021 .007
all the populations was negative and statistically significant (p < .05).
Note: Values with asterisk are significantly different (p < .05).
Riverine populations had lower values (SR = −2.734, NR = −2.627
and TR = −2.521; p = .001) than lacustrine populations (LV = −1.871,
mismatch distribution pattern (Figure 6). Riverine populations exhib- LKE = −2.270; p = .01 and LKA = −2.315; p = .05) (Table 5). Fu`s Fs
ited unimodal shaped mismatch distributions. Nonetheless, a distinct value for Lake Victoria population was negative and not significant
multi-­modal shape was revealed in River Nyando (NR) (Figure 7), (−2.336, p > .05; Table 5), but significant positive for Lakes Kamnarok
while River Sosiani (SR) C. gariepinus revealed a uni-­modal mismatch and Kenyatta (17.413 and 10.566, p < .001). Consequently, rivers
ALAL et al. | 7

F I G U R E 4 The mismatch distribution

of individual Clarias gariepinus mtDNA
control region from the Lake Victoria
populations. The X axis shows the number
of pairwise differences, and the Y axis
shows the frequency of the pairwise
comparisons. The plots were based on
constant population size change model

F I G U R E 5 The mismatch distribution

of individual Clarias gariepinus mtDNA
control region from the Lake Kamnarok
populations. The X axis shows the number
of pairwise differences, and the Y axis
shows the frequency of the pairwise
comparisons. The plots were based on
constant population size change model

Nyando, Sosiani and Tana, revealed positive significant values of Fu`s haplotypes and haplotype diversity of C. gariepinus reported in this
Fs (7.632, p < .01; 14.287 and 23.961, p < .001) respectively (Table 5). study are comparable to values reported for the species from other
lakes in Kenya (Barasa et al., 2017; Nyunja et al., 2017), as well as in
the region (Ojiambo, 2015). Although Lake Kenyatta is much smaller
4 | D I S CU S S I O N than Lake Victoria, the C. gariepinus population in Lake Kenyatta had
a higher number of haplotypes. This could be attributed to stock
4.1 | Genetic diversity of C. gariepinus inferred from augmentation of C. gariepinus fishery of Lake Kenyatta by conserva-
the mtDNA D-­loop tion scientists, to boost dwindling fish population due to overfish-
ing and periodic drying of the lake. Indeed, while Lake Kenyatta has
One way of improving conservation of natural fish populations while been restocked with C. gariepinus in the past, these activities were
simultaneously increasing opportunities for higher aquaculture pro- not documented. Previous studies have reported a high number of
duction is the determination of genetic diversity and population struc- haplotypes in C. gariepinus populations consisting of fish stocks from
ture of commercially important fish species. Overall, the number of multiple sources (Barasa et al., 2017; Grobler et al., 1997; Van De
8 | ALAL et al.

F I G U R E 6 The mismatch distribution

of individual Clarias gariepinus mtDNA
control region from the Lake Kenyatta
populations. The X axis shows the number
of pairwise differences, and the Y axis
shows the frequency of the pairwise
comparisons. The plots were based on
constant population size change model

F I G U R E 7 The mismatch distribution

of individual Clarias gariepinus mtDNA
control region from the River Nyando
populations. The X axis shows the number
of pairwise differences, and the Y axis
shows the frequency of the pairwise
comparisons. The plots were based on
constant population size change model

Bank et al., 1992). The higher number of haplotypes in the Nyando the Tana River population could be attributed to overfishing which
River population compared to populations from other rivers may be increases fish mortality, ultimately lowering the number and diver-
due to its larger drainage basin of 3,587 km2 (Gathenya et al., 2011), sity of haplotypes in C. gariepinus (Barasa et al., 2017) and other
and the fact that the river flows throughout the year. These charac- fishes (Chemoiwa et al., 2013; Muwanika et al., 2012). On the other
teristics are likely to reduce mortality in the population, and help to hand, River Sosiani had the lowest number and diversity of haplo-
maintain higher genetic diversity (Barasa et al., 2016). On the other types among all the sites. This could be attributed to population
hand, Tana River which is equally large and perennial had lower num- bottleneck effects (Frankham et al., 2002) associated with small size
ber and diversity of haplotypes than Nyando River, and this could be rivers (Sosiani River is 25 km long), water abstraction and urbaniza-
attributed to habitat fragmentation through damming of the river tion effects.
for electricity generation (Vogl et al., 2016). Hydroelectric dams The study recorded a high number of singletons for the cumula-
along rivers lead to fish population fragmentation, limiting gene flow tive sampled C. gariepinus populations (27 out of 34) or 79.41%, with
(Barocca et al., 2012). In addition, the low genetic diversity indices of only 20.59% carrying shared haplotypes. As the largest water mass
ALAL et al. | 9

F I G U R E 8 The mismatch distribution

of individual Clarias gariepinus mtDNA
control region from the River Sosiani
populations. The X axis shows the number
of pairwise differences, and the Y axis
shows the frequency of the pairwise
comparisons. The plots were based on
constant population size change model

F I G U R E 9 The mismatch distribution

of individual Clarias gariepinus mtDNA
control region from the River Tana
populations. The X axis shows the number
of pairwise differences, and the Y axis
shows the frequency of the pairwise
comparisons. The plots were based on
constant population size change model

of the sampled sites, Lake Victoria possibly harboured a higher num- genetic drift (Frankham et al., 2002), leading to a random emergence
ber of singletons than any of the populations due to a larger popu- of singletons. This, coupled with restocking of Lakes Kenyatta and
lation of C. gariepinus. However, the lower number of haplotypes (n Kamnarok may account for the higher number of singletons in these
= 3) compared to LKE (n = 6) and LKA (n = 9) could be attributed to populations.
historical predation by Nile perch, Lates niloticus, on C. gariepinus in
the lake (Njiru et al., 2004; Ogutu-­Ohwayo, 1990). Predation by L.
niloticus is reported to have caused a loss of genetic diversity in C. 4.2 | Population structure of C. gariepinus inferred
gariepinus of Lake Victoria (Barasa et al., 2014). Meanwhile, the ab- from the mtDNA D-­loop
sence of L. niloticus in the Nyando River, coupled with the large water
mass and perennial nature could have facilitated the emergence The Maximum Likelihood tree, Minimum Spanning Network and
and conservation of singletons in this population of C. gariepinus. In FST values showed that C. gariepinus populations were differenti-
contrast, small isolated habitats such as LKE and LKA often suffer ated and structured. The differentiation of C. gariepinus could be
10 | ALAL et al.

TA B L E 5 Demographic statistics
Populations Mutations SSD Raggedness Tajima`s D Fu`s Fs
for the combined 570 bp dataset of
Overall Populations 787 0.042ns 0.019ns −2.511*** 12.553*** Lacustrine and Riverine C. gariepinus
Lacustrine 619 0.018 ns
0.013 ns
−2.346** 10.972*** populations. Statistics include mutations,
Raggedness, Tajima`s D and Fu`s Fs
Riverine 617 0.000 ns 0.063ns −2.780*** 17.329***
neutrality tests
ns ns
LV 25 0.010 0.060 −1.957* −2.155ns
LKA 465 0.200*** 0.205*** −2.340** 17.138***
ns ns
LKE 481 0.042 0.061 −2.242** 11.027***
ns ns
NR 306 0.030 0.066 −2.646*** 7.123**
SR 179 0.210*** 0.193*** −2.737*** 13.435***
TR 417 0.122ns 0.137*** −2.534*** 23.177***

Note: Asterisk indicates significant test (*p ≤ .05, **p ≤ .01, ***p ≤ .001), while ns denotes not
significant.

attributed to limited gene flow caused by geographical separation of of Lake Kamnarok by E. crassipes reduced fishing pressure as well as
the sites. Consistently, previous studies recorded significant differ- provided ample feeding, breeding and nursery grounds for C. garie-
entiation among C. gariepinus populations from geographically sepa- pinus, leading to population expansion.
rated lakes in Kenya (Barasa et al., 2017; Nyunja et al., 2017). Indeed, The riverine C. gariepinus populations exhibited a uni-­
modal
geographical separation and isolation is one of the factors govern- distribution suggesting a recent demographic expansion. This was
ing genetic divergence among populations (Lande, 1988; Lehmann consistent with the negative value of Tajima's D, indicative of a re-
et al., 1998). These distinct populations therefore constitute impor- cent population size expansion after a bottleneck event. Indeed, the
tant genetic resources that should be conserved by avoiding inter-­ riverine C. gariepinus populations are prone to disturbances includ-
basin translocation that would result into genetic homogenization ing decline in water levels due to drought, leaving individual fish in
(Barasa et al., 2017). The existence of distinct population differentia- shallow pools, exposed to heavy fishing by human and predation by
tion in the samples was further supported by results of the analysis birds and anthropogenic activities such as sand harvesting (Masese
of molecular variance (AMOVA), which showed that genetic varia- et al., 2012), which destroy habitats for fish. However, recent epi-
tion of the C. gariepinus samples was mainly accounted for by vari- sodes of flooding in the rivers (Masese et al., 2017) provided suitable
ation within populations and also because of significant differences conditions for the growth of the populations. Generally, flood condi-
among populations. tions support proliferation of food items that support breeding and
nursing of catfishes (Halwart & Gupta, 2004).

4.3 | Demographic history of C. gariepinus inferred

from the mtDNA D-­loop 5 | CO N C LU S I O N S

The C. gariepinus populations from lakes had multi-­

modal Lakes Victoria, Kamnarok, Kenyatta and River Nyando C. gariepi-
shaped distributions, suggesting stable populations (Rogers & nus populations are reservoirs of higher genetic variation, while
Harpending, 1992). This distribution conformed to significant value Rivers Tana and Sosiani C. gariepinus have lower genetic variation.
for Tajima's D neutrality test. However, the negative neutrality test All populations of C. gariepinus had singletons with Lakes Kenyatta,
suggests that the population expansion of lacustrine C. gariepinus Kamnarok and River Nyando having higher number of singletons
is recent implying that the equilibrium exhibited by these popula- than Lake Victoria and Rivers Tana and Sosiani. Overall, these popu-
tions was achieved after a population expansion. The stability of lations are important genetic resources, which should be conserved,
Lake Victoria C. gariepinus population after an expansion could be by controlling factors likely to affect their genetic diversity such
attributed to the decline in the stock biomass of the predatory L. as overfishing, pollution, habitat destruction and fragmentation.
niloticus (Kayanda et al., 2009). Additionally, the exotic water hya- Furthermore, more robust analysis by means of nuclear markers to
cinth (Eichhornia crassipes) that invaded Lake Victoria creates anoxic supplement this work is recommended to validate these findings.
conditions restrictive to L. niloticus (Njiru et al., 2002) reducing pop-
ulation growth of L. niloticus. Meanwhile, water hyacinth provided AC K N OW L E D G E M E N T S
refugia and favourable conditions for C. gariepinus to flourish (Barasa The National Commission for Science, Technology and Innovation
et al., 2014; Njiru et al., 2002). Therefore, reduced predation pres- (NACOSTI) (NACOSTI/RCD/ST&I 6th CALL MSc/078) and the
sure and proliferation of C. gariepinus led to expansion and conse- National Research Fund (NRF) of Kenya (NRF/MSc/1/061) research
quently stabilized the population. On the other hand, expansion of C. grant awards to A.G.W funded the field sampling activities and part
gariepinus population in Lakes Kamnarok and Kenyatta could be at- of the laboratory work for this study. The Austrian Partnership
tributed to stock augmentation. In addition, the extensive coverage Programme in Higher Education and Research for Development
ALAL et al. | 11

(APPEAR) for the STRECAFISH Project supported laboratory work Chemoiwa, E. J., Abila, R., Macdonald, A., Lamb, J., Njenga, E., & Barasa,
J. E. (2013). Genetic diversity and population structure of the en-
at Makerere University's Molecular Biology Laboratory (MAKMBL)
dangered ripon barbell, Barbus altianalis in Lake Victoria catchment,
in Uganda by A.G.W. We thank Mr. J. Mayega of MAKMBL for Kenya, based on mitochondrial DNA sequences. Journal of Applied
technical support during laboratory work. Mr. G. Kasoso and Mr. J. Ichthyology, 29, 1225–­1233. https://doi.org/10.1111/jai.12313
Bett helped with field sampling in Lakes Kenyatta and Kamnarok, Chitamwebwa, D., Kamanyi, J., Kayungi, J., Nabbongo, H., Ogolla, A.,
respectively. & Ojuok, J. (2009). The present status of the hook fishery and its
impact on fish stocks of Lake Victoria. African Journal of Tropical
Hydrobiology and Fisheries, 12, 78–­82.
C O N FL I C T O F I N T E R E S T Corbet, P. S. (1961). The food of non-­cichlid fishes in the Lake Victoria
Authors declare no conflict of interest in this manuscript. Basin, with remarks on their evolution and adaptation to lacustrine
conditions. Proceedings of the Zoological Society of London, 136, 1–­
101. https://doi.org/10.1111/j.1469-­7998.1961.tb060​8 0.x
DATA AVA I L A B I L I T Y S TAT E M E N T
Darriba, D., Taboada, G. L., Doallo, R., & Posada, D. (2012). JModelTest:
Data for this work has been submitted to the GenBank database Phylogenomics, Universidade de Vigo, Vigo, Spain (c) Grupo de
and is publicly available under the accession numbers MK014577 Arquitectura de Computadores, UDC, A Coruna, Spain Project.
-­ MK014704. Excoffier, L. (2004). Analysis of population subdivision. Handbook of
Statistical Genetics, https://doi.org/10.1002/04700​22620.bbc25
Excoffier, L., & Lischer, H. E. (2010). Arlequin suite version 3.5: A new
ORCID series of programs to perform population genetics analyses under
George W. Alal https://orcid.org/0000-0002-9788-1144 Linux and Windows. Molecular Ecology Resources, 10, 564–­567.
James E. Barasa https://orcid.org/0000-0001-7221-6618 https://doi.org/10.1111/j.1755-­0998.2010.02847.x
Frankham, R., Ballou, J. D., & Briscoe, D. A. (2002). Introduction to
Conservation Genetics. Cambridge University Press. https://doi.
REFERENCES org/10.1017/CBO97​8 0511​8 08999
Barasa, J. E. (2018). Genetic diversity, population structure and influence Fu, Y. X. (1997). Statistical tests of neutrality of mutations against pop-
on life-­history traits of the African catfish, Clarias gariepinus in Kenya. ulation growth, hitchhiking and background selection. Genetics, 147,
PhD thesis, University of Eldoret, Kenya. 214 pages. 915–­925.
Barasa, J. E., Abila, R., Grobler, J. P., Agaba, M., Chemoiwa, E. J., & Gathenya, M., Mwangi, H., Coe, R., & Sang, J. (2011). Climate and
Kaunda-­Arara, B. (2016). High genetic diversity and population dif- land use induced risks to watershed services in the Nyando River
ferentiation in Clarias gariepinus of Yala Swamp: Evidence from mi- Basin. Kenya. Exploratory Agriculture, 47(2), 339–­ 356. https://doi.
tochondrial DNA sequences. Journal of Fish Biology, 89, 2557–­2570. org/10.1017/S0014​47971​100007X
https://doi.org/10.1111/jfb.13150 Gene Codes Corporation. (2016). Sequencher 5 Series. www.genec​odes.
Barasa, J. E., Abila, R., Grobler, J. P., Dangasuk, O. G., Njahira, M. N., & com
Kaunda-­Arara, B. (2014). Genetic diversity and gene flow in Clarias Grobler, J. P., Hoffman, L. C., & Prinsloo, J. F. (1997). A Comparison of
gariepinus from Lakes Victoria and Kanyaboli, Kenya. African Journal Allozyme Heterozygosity and Life History variables in four strains
of Aquatic Science, 39, 287–­ 293. https://doi.org/10.2989/16085​ of African Catfish (Clarias gariepinus). Southern African Journal of
914.2014.933734 Aquatic Sciences, 23(1), 31–­ 41. https://doi.org/10.1080/10183​
Barasa, J. E., Mdyogolo, S., Abila, R., Grobler, J. P., Skilton, R. A., 469.1997.9631386
Bindeman, H., Ndotono, N. M., Chemoiwa, E. J., Dangasuk, O. Hall, T. (2005). BioEdit v7. 0.5. Biological sequence alignment editor.
G., Kaunda-­ A rara, B., & Verheyen, E. (2017). Genetic diversity Department of Microbiology, North Carolina State University.
and population structure of the African catfish, Clarias gariepi- (Online) Available at: http://www.mbio.ncsu.edu/BioEd​ it/Bioed​
nus (Burchell, 1822) in Kenya: Implication for conservation and it.html
aquaculture. Belgian Journal of Zoology, 147, 105–­127. https://doi. M. Halwart, & M. V. Gupta (Eds.) (2004). Culture of fish in rice fields. FAO
org/10.26496/​bjz.2017.14 and The WorldFish Center, 83 pages.
Barocca, T. M., Santos, G. B., Duarte, N. V. R., & Kalapothakis, E. (2012). Kaufman, L., & Ochumba, P. (1993). Evolutionary and conservation biol-
Evaluation of genetic diversity and population structure in a commer- ogy of cichlid fishes as revealed by faunal remnants in northern Lake
cially important freshwater fish Prochilodus costatus (Characiformes, Victoria. Conservation Biology, 7, 719–­730. https://doi.org/10.1046/
Prochilodantidae) using complex hypervariable repeats. Genetics and j.1523-­1739.1993.07030​719.x
Molecular Research, 11, 4456–­4 467. https://doi.org/10.4238/2012. Kayanda, R., Taabu, A. M., Tumwebaze, R., Muhoozi, L., Jembe, T.,
Septe​mber.27.4 Mlaponi, E., & Nzungi, P. (2009). Status of the major commercial fish
Bernatchez, L., & Danzmann, G. R. (1993). Congruence in control re- stocks and proposed species specific management plans for Lake
gion sequence and restriction-­site variation in mitochondrial DNA Victoria. African Journal of Tropical Hydrobiology and Fisheries, 12, 15–­
of Brook charr (Salvelinus fontinalis Mitchill). Molecular Biology and 21. https://doi.org/10.4314/ajthf.v12i1.57366
Evolution, 10, 1002–­1014. https://doi.org/10.1093/oxfor​djour​nals. Kumar, S., Stecher, G., Li, M., Knyaz, C., & Tamura, K. (2018). MEGA X:
molbev.a040062 Molecular evolutionary genetics analysis across computing plat-
Bruton, M. N. (1979). The breeding biology and early development of forms. Molecular Biology and Evolution, 35(6), 1547–­1549. https://doi.
Clarias gariepinus (Pisces, Clariidae) in Lake Sibaya, South Africa, with org/10.1093/molbe​v/msy096
a review of breeding species of the subgenus Clarias. Transactions Lande, R. (1988). Genetics and demography in biological conservation.
of the Zoology Society of London, 35, 1–­45. https://doi.org/10.1111/ Science, 241, 1455–­1460. https://doi.org/10.1126/scien​ce.3420403
j.1096-­3642.1979.tb000​56.x Larkin, M. A., Blackshields, G., Brown, N. P., Chenna, R., McGettigan, P.
Cambray, J. A. (2003). The need for research and monitoring on the im- A., McWilliam, H., Valentin, F., Wallace, I. M., Wilm, A., Lopez, R.,
pacts of translocated Sharp-­tooth catfish, Clarias gariepinus, in South Thompson, J. D., Gibson, T. J., & Higgins, D. G. (2007). Clustal W and
Africa. African Journal Aquatic Science, 28, 191–­ 195. https://doi. Clustal X Version 2.0. Bioinformatics, 23, 2947–­ 2948. https://doi.
org/10.2989/16085​91030​9503786 org/10.1093/bioin​forma​tics/btm404
12 | ALAL et al.

Letunic, I., & Bork, P. (2016). Interactive tree of life (iTOL v3): An online Owiti, D. O., & Dadzie, S. (1989). Maturity, fecundity and the effect of
tool for the display and annotation of phylogenetic and oher relation- reduced rainfall on the spawning rhythm of a siluroid catfish, Clarias
ships. Nucleic acids Research, 44, 242–­245. mossambicus (Peters). Aquaculture and Fisheries Management, 20,
Masese, F. O., Raburu, P. O., & Kwena, F. (2012). Threats to the Nyando 355–­368. https://doi.org/10.1111/j.1365-­2109.1989.tb003​63.x
Wetland. In Community based Approach to the Management of Nyando Rogers, A. R., & Harpending, H. (1992). Population growth makes waves
Wetland (pp. 68–­8 0). Lake Victoria basin. in the distribution of pairwise genetic differences. Molecular Biology
Masese, F. O., Salcedo-­Borda, J. S., Gettel, G. M., Irvine, K., & McClain, and Evolution, 9, 552–­569.
M. E. (2017). Influence of catchment land use and seasonality on Rozas, J., Ferrer-­Mata, A., Sánchez-­DelBarrio, J. C., Guirao-­Rico, S., Librado,
dissolved organic matter composition and ecosystem metabolism in P., Ramos-­Onsins, S. E., & Sánchez-­Gracia, A. (2017). DNAsP 6: DNA se-
headwater streams of a Kenyan river. Biogeochemistry, 132(1–­2), 1–­ quence polymorphism analysis of large data sets. Molecular Biology and
22. https://doi.org/10.1007/s1053​3-­016-­0269-­6 Evolution, 34, 3299–­3302. https://doi.org/10.1093/molbe​v/msx248
Muwanika, V., Nakamya, M., Rutaisire, J., Sivan, B., & Masembe, C. Sulem, Y. S., Tomedi, E. T., Mounchili, S., Tekeng, S., & Brummett, R. S.
(2012). Low genetic differentiation among morphologically distinct (2006). Survival of Clarias gariepinus fry in earthen ponds: Effects
Labeobarbus species (Teleostei: Cyprinidae) in the Lake Victoria of composts and leaks. Aquaculture, 260, 139–­ 144. https://doi.
and Albertine basins, Uganda: Insights from mitochondrial DNA. org/10.1016/j.aquac​ulture.2006.06.008
African Journal of Aquatic Science, 37(2), 143–­ 153. https://doi. Swofford, D. L. (2002). PAUP*. Phylogenetic Analysis Using Parsimony
org/10.2989/16085​914.2012.668850 (and Other Methods). Version 4. Sinauer Associates. Sunderland,
Ngugi, C. C., Omolo, B., Langdon, C., & Bowman, J. (2005). Studies on Massachusetts.
strategies for increasing the growth and survival of African catfish, Tajima, F. (1989). The effect of change in population size on DNA poly-
Clarias gariepinus juveniles reared for stocking or for use as live bait. morphism. Genetics, 123, 597–­601.
In: 25th Annual Technical report-­Vol II. Oregon State University, Snell Van der Bank, F. H., Grobler, P. J., & Do Preez, H. H. (1992). A compar-
Hall 400 Corvalis, Oregon 97331-­1641, USA. ative biochemical genetic study of three populations of domes-
Njiru, M., Okeyo-­ Owuor, J. B., Muchiri, M., & Cowx, I. G. (2004). ticated and wild African catfish (Clarias gariepinus). Comparative
Shifts in the food of Nile tilapia, Oreochromis niloticus (L.) in Lake Biochemistry and Physiology, 101(B), 387–­ 390. https://doi.
Victoria, Kenya. African Journal of Ecology, 42, 163–­170. https://doi. org/10.1016/0305-­0 491(92)90017​-­L
org/10.1111/j.1365-­2028.2004.00503.x Vogl, A. L., Bryant, B. P., Hunink, J. E., Wolny, S., Apse, C., & Droogers, P.
Njiru, M., Othina, A. N., Getabu, A., Twiddle, D., & Cowx, I. G. (2002). Is (2016). Valuing investments in sustainable land management in the
the invasion of water hyacinth, Eichhornia crassipes solms (mart.), a Upper Tana River basin, Kenya. Journal of Environmental Management,
blessing to Lake Victoria fisheries? In I. G. Cowx (Ed.), Management 30, 1–­14. https://doi.org/10.1016/j.jenvm​an.2016.10.013
and ecology of lake and reservoir fisheries (pp. 255–­263). Fishing News Witte, F., & van Densen, W. L. T. (1995). Fish stocks and fisheries of Lake
Books. https://doi.org/10.1002/97804​70995​679.ch21 Victoria: A handbook for field observations. Samara Publishing.
Nyunja, C., Maina, J., Amimo, J., Kibegwa, F., Harper, D., & Jung'a, J. Wright, S. (1965). The interpretation of population structure by F-­
(2017). Stock structure delineation of the African Catfish (Clarias statistics with special regard to systems of mating. Evolution, 19,
gariepinus) in selected populations in Kenya using mitochondrial DNA 395–­420. https://doi.org/10.1111/j.1558-­5646.1965.tb017​31.x
(Dloop) variability. Journal of Aquaculture Research and Development,
8, 485. https://doi.org/10.4172/2155-­9546.1000485
Ogutu-­ Ohwayo, R. (1990). The decline of the native fishes of lakes
How to cite this article: Alal GW, Barasa JE, Chemoiwa EJ,
Victoria and Kyoga (East Africa) and the impact of introduced spe-
cies, especially the Nile perch, Lates niloticus, and the Nile tilapia,
Kaunda-­Arara B, Akoll P, Masembe C. Genetic diversity and
Oreochromis niloticus. Environmental Biology and Fisheries, 27, 81–­96. population structure of selected lacustrine and riverine
https://doi.org/10.1007/BF000​01938 populations of African catfish, Clarias gariepinus (Burchell,
Ojiambo, D. (2015). Population genetic structure of wild and domesti- 1822), in Kenya. J Appl Ichthyol. 2021;00:1–­12. https://doi.
cated African catfish (Clarias gariepinus) in Victoria and Albertine
org/10.1111/jai.14167
Drainage Basins. MSc. thesis, Makerere University, Kampala,
Uganda. 53 pages.