4. Population structure, genetic diversity, and genetic differentiation
Haplotype frequencies based on relative read counts (eDNA) or the number of individuals (tissue) within each site were plotted and compared. Additionally, the spatial patterns of haplotype frequencies were compared with the population structure inferred from genome-wide SNP data.
To identify overall trends in intrapopulation genetic diversity, haplotype diversity (h S;\(h_{S}=1-\sum_{i}{p_{i}}^{2}\), where \(p_{i}\) is the haplotype frequency of the i -th haplotype) and haplotype richness (hr ; number of haplotypes corrected for differences in the number of individuals or total read counts among populations; haplotype version of allelic richness (El Mousadik and Petit 1996)) of each population were calculated. For interpopulation genetic differentiation, we first calculated Nei’s F ST (G ST; Nei 1973), as haplotype differentiation in Hudson et al. (1992) (i.e., it does not matter whether two haplotypes differ by one nucleotide or by tens). Other than G ST, we also used the allele frequency distance (AFD; Berner 2019) with the modification of using haplotypes instead of alleles. AFD is considered less susceptible to differences in sample size between populations and more sensitive in the range of weak differentiation that is of interest in local scale studies (Berner 2019). In addition, this metric is identical toD PS (1 − proportion of shared alleles; Bowcock et al. 1994), which is known to reflect recent gene flow (~10 generations) particularly relevant to management and the purpose of landscape genetics (Leroy et al. 2018; Savary et al. 2021). Hereafter, we refer to this metric as D PS, although this notation has not yet been used for haplotype-based analysis. The calculation process of D PS for two arbitrary populations Pop A and Pop B is\(D_{\text{PS}}=(\ \sum_{i}{\left|p_{\text{iPop}A}-p_{\text{iPop}B}\right|\ )/2}\), where \(p_{\text{iPop}A}\) and \(p_{\text{iPop}B}\) are the haplotype frequencies of the i -th haplotype in Pop A and Pop B, respectively. As this study focuses on local genetic differentiation reflecting recent gene flow, subsequent analyses and discussions are primarily based on D PS, while analyses using more common and well-known G ST statistics were also conducted.
To clarify the spatial scale at which gene flow is more dominant than genetic drift for genetic differentiation, Mantel correlograms displaying the spatial correlation of D PS (as simple genetic distance available in eDNA) and geographic distance for each 7.5 km distance class until 60 km and then from 60 km to 128 km (maximum value) were generated. Within the distance classes where the correlation coefficient is significantly positive, gene flow among populations is considered to be particularly active (Diniz‐Filho and De Campos Telles 2002). The correlogram in each distance class was assessed with 9999 permutations using the package ecodist 2.0.9 (Goslee and Urban 2007) in R 4.2.1 (R Core Team 2022).
Genetic differentiation statistics were also calculated by SNP data (Text S1). Generally, genetic diversity and genetic drift pressure differ between nuclear DNA and mtDNA (Toews and Brelsford 2012; Morin et al. 2018; Saitoh 2021). However, at the spatial scale at which gene flow is dominant, because the theoretical pattern of being inversely proportional to migration is common (Allendorf et al. 2022), genetic differentiation should be synchronized regardless of markers in the absence of sex differences. On the spatial scale explicitly stated, the more informative SNP data were used as the known truth.