Understanding the structure of hybrid zones provides valuable insights about species boundaries and speciation, such as the evolution of barriers to gene flow and the strength of selection. In river networks, studying evolutionary processes in hybrid zones can be especially challenging, given the influence of past and current river properties along with biological species-specific traits. Here, we suggest that a natural hybrid zone between two divergent lineages of the sexually dimorphic Neotropical fish Nematocharax venustus was probably established by secondary contact as a result of a river capture event between the Contas and Pardo river basins. This putative river capture is supported by hydrogeological evidence of elbows of capture, wind gaps, and geological faults. The morphological (color pattern) and genetic (mtDNA and RADseq) variation reveal a clinal transition between parental lineages along the main river, with predominance of F2 hybrids at the centre of the hybrid zone, absence of early generation backcrosses, and different levels of hybridization in the tributaries. We highlight that different sources of information are crucial for understanding how the riverscape spatial history influences the connectivity between and within rivers systems and, consequently, the dynamics of gene flow between freshwater lineages/species.

Four folders:

(1) faststructure and gghybrid: input files for fastSTRUCTURE (.str) and gghybrid (.csv) analyses. The gghybrid file contains two columns to the left of the marker columns, which correspond to the individual references and the sample population (see Figure 1 in Barreto et al. 2020).

(2) parallelnewhybrid: contains a file in the NewHybrids format and its associated individuals file. Two other copies of the genotype file were used to run parallelnewhybrid.

(3) hzar morphological data: input files for HZAR cline analyses using the caudal peduncle and eye coloration marks.

(4) hzar genomic data: input files for HZAR cline analyses using the mtDNA haplotypes, the ancestry coefficients estimated in fastSTRUCTURE (K = 2), and the 99 putatively unlinked SNPs (i.e., one SNP per locus) from which we estimated cline parameter values independently.