Within-taxon phylogeographic structure
Both our SNP and mtDNA datasets contain sufficient genetic insights to explore phylogeographic trends within the four primary taxa. In addition to the well-supported dichotomy between KW and KWm, the SNPs identified a shallow split in KW between Bulloo and Cooper populations (sites 111–114 versus 115–120; Fig. 3), They also revealed additional albeit more complicated phylogeographic structure for the other three taxa (Figs. 3, 4), herein further explored using taxon-specific PCoAs (Fig. 7).
For KN, there was a primary dichotomy between the three northern and four southern sites which reflects latitude rather than river basin membership (Fig. 7a). Most notably, the Burdekin KN were closely allied with those in the Fitzroy River (Figs. 3, 7) despite being somewhat of a northern outlier (Fig. 1).
A more complex pattern is evident within KE+. All relevant PCoAs (Figs. 2b, 2d, 4c, S1) consistently found that Condamine fish (KEm) were genetically most similar to those in the Burnett, while two geographic outliers (sites 1 and 16) and a number of regional population clusters were present for pure KE in the RAxML tree (Fig. 3b). A KE-specific PCoA revealed a similar pattern of diversity, including the site 1 and 16 outliers (discussed separately), but was also able to detect a primary phylogeographic split between sites from the Maroochy River northwards and west to the Burnett (sites 10–29, excluding site 16) versus sites south of and including the Pine and Brisbane Rivers (sites 28–48). The northern outlier (site 1, Barron River) clearly clusters with sites 19–24 (Burrum and Mary Rivers), supporting its likely status as an introduced population from that part of KE’s range. Intriguingly, both individuals from site 16 (Burnett River) are anomalously placed, one intermediate between the two primary clusters (and showing elevated heterozygosity levels) and the other clustering with the southern phylogroup. This same pattern is displayed in the allozyme data (PCoA not shown) and in the mtDNA tree (Fig. S3), with one individual from site 16 clustering with Brisbane River haplotypes (southern phylogroup) and the other with Burnett River haplotypes (northern phylogroup).
With respect to KS, most sites are relatively homogeneous, with only modest structure relating to geographic outlying populations in two of the Border Rivers (Severn and McIntyre Rivers; sites 60, 61), the Macleay River (site 49), and the Shoalhaven River (site 52), the latter clustering with one of the sites in the adjacent drainage divide (site 83, Murrumbidgee River) and therefore inferring a source population. Although the mtDNA data are also relatively homogeneous for KS, the Border Rivers harbored two distinctive cytb lineages at high frequency that were absent elsewhere in pure KS (Fig. 6).
Discussion Building on the work of Thacker et al. (2007), we present three comprehensive molecular datasets that together identify four primary taxa (KN, KE+, KS, and KW+) plus several examples of historic (KEm, KWm) and relatively recent admixture (KSxKE) within the parent species H. klunzingeri . All primary taxa, lineages and admixed zones are fully diagnosable by numerous independent genetic markers and our intensive sampling provides an ideal starting point for future field surveys to plug apparent distributional gaps, ecological assessments of the H. klunzingericomplex, or formal taxonomic revision. Regarding the latter, allocating the nominal form of H. klunzingeri s.s. (Ogilby, 1898) to a specific taxon may prove problematic, given the type locality (the Murray River in South Australia) is apparently, at least the time of collecting for genetic evaluation, a hybrid zone (KSxKE).