DNA sequencing
We developed the enhanced eDNA assay to detect intraspecific genetic diversity of B. toni by introducing unique molecular tagging systems (see Supplementary Materials). Utilizing stringent ASV methods, this assay minimizes the risk of false positives caused by index hopping and sequencing errors (19 , 20 ). To select a river that hosts B. toniacross Hokkaido for this study, we analyzed eDNA samples collected from 293 rivers. In this step, we prepared five dual-indexed libraries using the fish-universal MiFish primer set (21 ) and performed eDNA metabarcoding on a MiSeq sequencing platform (Illumina, San Diego, CA, USA). Throughout the five MiSeq runs, the eDNA of B. toni was detected in 189 out of 293 river samples (64.5%; fig S1). Among these valid samples, we selected 51 river samples covering the coastline of Hokkaido to implement the Barbatula -specific eDNA assay on a fine scale with the goal of determining the phylogeographic pattern ofB. toni (table S1). Using newly-developed PCR primers amplifying 266 bp of the mitochondrial DNA cyt-b gene of B. toni , we prepared a dual-indexed library comprising four replicates of the 51 river samples, four field blanks, and eight PCR blanks. After trimming the adapters and primers from the result of a single MiSeq sequencing run, 9,193,270 reads were obtained. After read merging, quality filtering, denoising, and chimera removal, 5,981,225 reads and 65 haplotypes of B. toniwere finally obtained. Furthermore, no eDNA was detected in the field and PCR blanks. A conservative decision was then made to accept 50 out of 65 haplotypes based on two criteria: 1) detection in all four sequencing replicates and 2) eDNA concentration > 1.0 copy per liter. The first criterion eliminated 15 haplotypes (one haplotype with 3 out of 4 detections, two with 2 out of 4 detections, and twelve with 1 out of 4 detections) as potential erroneous haplotypes with low detection frequencies (22 ). The eDNA concentrations of all remaining haplotypes exceeded the second criterion (8.3 copies/L at minimum).
To reliably infer the phylogenetic relationship among the B. toniin Hokkaido and Sakhalin (Russia), we obtained tissue-derived DNA sequences through Sanger sequencing (Materials and Methods-2). Using the primer set Le-L4 and Le-H4 (23 ), we amplified 1,055 bp of the cyt-b DNA sequences of five individual tissue samples collected from four river systems: Shiribeshi–Toshibetsu River, Koetoi River, Tokoro River (Nos. 46, 7, and 17 in table S1), and a river in Sakhalin.