Candidate species
The genetic resolution of candidate species has often relied on identifying lineages using either gene genealogies (e.g., mtDNA or nDNA gene trees) or multi-locus population trees (for allozymes or genomic data). However, as such tree-only approaches detect genetic structure rather than candidate species per se(Sukumaran & Knowles, 2017; Unmack et al., 2022), lineages delineated in this manner need not directly equate to biological or evolutionary species but may instead reflect major phylogeographic breaks within a species or a composite of two or more species plus admixed individuals.
While there is no simple formula for deciding whether two genetically distinctive allopatric populations are conspecific or represent different species, we have recently advocated a six-step approach to assist in this task (Unmack et al., 2022). These steps are: identify lineages, hybrids, and introgressed populations using a combination of ordination of individuals (step 1) plus phylogenetic methods (step 2), followed by pairwise assessments of lineage diagnosability (step 3), comparative geographic distribution (step 4), and sampling intensity (step 5), and concluding with a review of any other biological information that might indicate that lineages are not conspecific (step 6). Unfortunately, observations relevant to this final step are largely unavailable in the literature, since many ecological studies ofHypseleotris in eastern Australia have not attempted to reliably distinguish H. klunzingeri from a suite of congeneric and often co-occurring taxa (e.g., Meredith, Matveev & Mayes, 2003; Lintermans, 2007), now known to comprise a complex of sexual species and ‘unisexual’ (hybridogenetic) lineages (Unmack et al., 2019; Thacker et al., 2022b). We hope that a recent taxonomic revision by Thacker et al. (2022a) for this hemiclonal species complex, which includes five sexual species and multiple unisexual combinations, will help establish a more robust taxonomic framework for identifying individuals to their correct sexual group and hence facilitate the documentation of comparative biological information for all sexual forms of Hypseleotris , including those referrable to the H. klunzingeri complex.
Table 4 summarizes the outcomes of applying steps 3–5 to the primary taxa identified for H. klunzingeri using steps 1 and 2. As shown, there is strong evidence that KN, KE+, and KS are all valid candidate species, being unequivocally or effectively diagnosable from each other at hundreds of unlinked genes and displaying distributional patterns that are inconsistent with being phylogeographic lineages within a single species (Table 4, Fig. S5). Given their comparatively low number of diagnostic differences, the decision as to whether the allopatric taxa KS and KW+ are conspecific or represent distinct evolutionary species remains the only taxonomic question not fully resolved by our stand-alone genetic datasets. However, as the number of molecular characters that diagnose KS from KW+ greatly exceeds the nine partially-diagnostic morphological characters that delineate other co-occurring species of Hypseleotris (Thacker et al., 2022a), we have concluded that KW+ ‘probably’ represents a fourth candidate species. A full resolution of its taxonomic status will require additional targeted assessments of any morphological and other biological differences between KS and KW+, and must include exemplars of pure KS, pure KW, and KWm. The scenario of sister Cooper versus MDB candidate taxa is also evident in another co-occurring freshwater fish (Australian smelt, Retropinna spp.; Unmack et al., 2022).