Fig 1.
Maps of Lake Turkana (left) and the Nile basin (right) showing the locations of our sampling localities and the outlines of the separate basins (based on [28,29] and original observations).
The asterisk indicates the current endorheic Lotikipi Plain, which is being temporarily flooded during exceptionally rainy years. During the wet Pleistocene climatic phases and the African humid period of the Holocene, the Lotikipi Plain repeatedly posed an aquatic link between the Nile and Turkana basins.
Table 1.
Sampling localities, coordinates and sample sizes.
Table 2.
Primer combinations used for AFLP.
Table 3.
Substitution models for nucleotide data partitions of the Wenyonia spp. coxI dataset selected using the AIC in PartitionFinder for BI and ML runs.
Fig 2.
Migration hypotheses tested by software MIGRATE, datasets represented here are a) host; b) parasite.
Arrows depict direction of historical gene flow.
Fig 3.
Phylogenetic tree of the cox I Wenyonia spp. computed by MrBayes for a) the model of molecular evolution; b) the best-fit partitioning schemes.
Branches are collapsed into groups, which correspond to sampling locations. Statistical support for each group was obtained from the Bayesian posterior probability.
Fig 4.
Haplotype network for the cox I Wenyonia spp. dataset (a) and for the cox I Synodontis spp. dataset (b) computed by PopART v1.7.
Most groups were separated from each other by a maximum of six mutations. The exception was group 4, which was separated by ten mutations from group2 and eleven mutations from group3. Samples of S. nigrita clustered together and created a separate network. The sizes of haplotypic nodes are relative to the sample size.
Table 4.
Results of population genetic statistics of mtDNA for both parasites and hosts.
Table 5.
Results of the Synodontis spp. and Wenyonia spp. migration hypotheses tested with the coalescent based software MIGRATE, showing comparisons of marginal likelihoods and model probabilities for each dataset.
Fig 5.
Results of PCoA for (a, b) Wenyonia spp. and (c, d) Synodontis spp. based on the genetic distances generated by GenAlEx software.
Results are presented for the first three axes, because the second and the third axis explained similar levels of variability. In PCoA, all individuals were labelled with different colours based on their geographical or species origin. Asterisks (a, b) indicate clearly separated clusters comprising solely Wenyonia Turkana samples. Synodontis samples (c, d) from all localities clustered together without forming distinguishable groups. S. nigrita individuals clustered together and even showed a tendency to create a separate group (d). Turkana—medium 1–3 refers to three different localities: 1, Central Island; 2, Kerio River delta; 3, Kalokol.
Fig 6.
Individual-based cluster representation of all sampled Wenyonia spp. (a) and Synodontis spp. (b) as revealed by the Bayesian inference of population structure.
Each colour represents one assumed population cluster K. Multiple coloured bars display an individual’s estimated membership proportion in more than one population (q), i.e. admixture. Turkana—medium 1–3 refers to three different localities: 1, Central Island; 2, Kerio River delta; 3, Kalokol.
Table 6.
Results of the AMOVA based on FST for Wenyonia virilis and Synodontis spp.
AFLP data from Lake Turkana and the Nile; Turkana freshwater and saline part. Estimations are based on 999 permutations and the significant at P<0.001.