Pogonolepis, Steetz

Schmidt-Lebuhn, Alexander N., 2022, Sequence capture data support the taxonomy of Pogonolepis (Asteraceae: Gnaphalieae) and show unexpected genetic structure, Australian Systematic Botany 35 (4), pp. 317-325 : 322-323

Species delimitation in Pogonolepis

As implied by the above, molecular data confirmed Short’s (1986) taxonomy of Pogonolepis , but not his hypothesis of the polyphyletic origin of P. muelleriana suggested by its diversity in chromosome numbers. On current evidence, it appears that extant populations of P. muelleriana and P. stricta constitute two sister lineages, and that the former may have arisen from a single event. However, clade support for P. stricta is considerably lower than for its sister, potentially reflecting a lower degree of lineage sorting since divergence of the two species. Another caveat is that geographic sampling of P. muelleriana in Western Australia was limited to the southern half of its distribution in that state ( Fig. 2 ), so it remains possible that broader sampling may yet show multiple origins. However, it appears highly likely that all its eastern Australian populations form a clade, because they are well represented in the data.

This result intuitively supports the continued recognition of P. muelleriana and P. stricta as species. However, it is desirable to make methodological and taxonomic approaches transparent. In the present case, the existence in the Angianthus clade of other pairs differing only in reproductive system ( Short 1985) and of various other genera with combinations of sexually and asexually reproducing species ( Short 1983, 1989, 1990 a, 1990 b, 2015) means that it would be logical to apply a consistent approach to species delimitation in this clade.

The problem remains that because the two members of each pair have different reproductive systems, naive application of a single species concept is impossible. If, for example, non-outcrossing populations are to be separated from outcrossing ones because there is no gene flow between the two entities (biological species concept), then, logically, each individual of the non-outcrossing populations would have to be its own species. This illustrates the force of arguments that species are better considered as phenomena in need of explanation, rather than a pre-existing concept into which patterns of diversity are forced ( Mishler and Wilkins 2018).

Such an approach provides the conceptual flexibility under which, in this case, P. stricta can be considered a biological species and P. muelleriana an agamospecies ( Zachos 2016). Although it may be argued that the data presented here show the two taxa to be reciprocally monophyletic and, therefore, the concept of monophyletic species to be equally applicable, the statement that a sexually reproducing species such as P. stricta is monophyletic is a category error. Because it is sexually reproducing, there is no phylogenetic structure to be -phyletic in, be it mono-, para- or poly- ( Hennig 1966).

Reproductive systems and gene concordance

To examine the effect of the reproductive strategy on the concordance between gene trees and the concatenated phylogeny, I calculated gCFs. My expectation was they would be higher in non-outcrossing P. muelleriana , because it seemed logical to assume that it would show less recombination between different genes than outcrossing P. strica . However, with the exception of one outlier branch, the opposite was the case, and despite equal and geographically dispersed sampling, the median gCF was much higher in the outcrossing species. Randomisation of terminal names to simulate complete recombination between genes in one species produced gCF values in line with those observed in non-outcrossing P. muelleriana .

Values of sCF were close to 33% across internal nodes in both species, implying that their internal structure is poorly resolved in both cases, as would be expected if there is either no phylogenetic structure because of recombination or too few informative characters. The large number of informative characters in both species suggests that the former is more likely to be the case.

It is nonetheless unclear why the outcrossing species shows unexpectedly high gene concordance and lower heterozygosity than does its non-outcrossing sister. To understand the processes behind the genetic structure of the species, a more detailed understanding of the reproductive system will be required. Short (1986) interpreted P. muelleriana as selfing. Although he cautioned that the possibility of apomixis could not be excluded on available evidence, he associated it with abnormal pollen formation as found in Taraxacum Weber ( Richards 1973) , which he did not observe in the species. However, there is no reason to assume that apomicts show abortive pollen, and pseudogamous apomicts require (potentially self-)pollination to activate the endosperm even as no fertilisation of the egg cell takes place ( Noirot et al. 1997).

There are therefore several possibilities for the reproductive system of P. muelleriana , including selfing, facultative apomixis, and obligate apomixis. In the first two cases, outcrossing is still possible, albeit, presumably, much less likely to happen than in P. stricta , because of the much smaller number of pollen grains. It would be difficult to emasculate the minuscule, sequentially maturing flowers of a capitulum, but flow cytometric examination of fruits would allow selfing and apomixis to be differentiated by the genome-size ratio of embryo and endosperm ( Matzk et al. 2001; Chen et al. 2019). In addition to the reproductive system, dispersal distances would influence the genetic structure of a species ( Hamrick and Loveless 1986), but the two species of Pogonolepis do not differ in their fruit morphology or adaptations to dispersal.

Another possibility is that an allopolyploid origin of at least part of Pogonolepis muelleriana explains the unexpected genetic structure, because the species is known to have either 2 n = 12 or ~20–24 chromosomes, compared with 2 n = 8–10 in P. stricta ( Short 1986) . In this scenario, alleles inherited from divergent parental populations would produce higher heterozygosity and potentially lower gene concordance than expected from a non-outcrossing diploid. HybPiper produced more paralog warnings for samples of P. muelleriana (median 23, mean 25.8) than for P. stricta (median 15, mean 20.0), which could be seen to provide some support for this interpretation under the assumption that autopolyploidy would lead to paralogs too similar to be recognised.

What complicates this interpretation is that gene heterozygosity values in both species of>90% are unexpectedly high and in line with what would be expected of hybrids ( Nauheimer et al. 2021). This raises the possibility that the entire genus may have a polyploidisation event in its recent ancestry, even before the duplication in Pogonlepis muelleriana .

Finally, the high heterozygosity itself suggests another possible explanation. The phylogeny was inferred using data assembled with HybPiper, which returns a single sequence for a locus if variants of that locus are not divergent enough to be flagged as possible paralogs. This means that different alleles of a heterozygous locus may be retrieved effectively randomly from each sample, and this could then reduce gene concordance on the phylogeny. Given that non-outcrossing Pogonolepis muelleriana showed higher gene heterozygosity, this effect may be more pronounced in that species, leading to its lower observed gene concordance.

Utility of sequence capture at the species level

The confirmation as being misidentified after morphological re-examination of all specimens seemingly in the ‘wrong’ position in the molecular phylogeny demonstrates the feasibility of diagnosing species affiliation using Angiosperms353 sequence capture data, assuming sufficient reference data are available, in this case for several specimens each of the two species of Pogonolepis and some outgroups.

In addition, the data resolved geographic structure for both species of Pogonolepis . In P. muelleriana , the three Western Australian specimens formed a grade under the eastern specimens, as expected from a western origin of this species followed by dispersal to the east. Specimens of P. stricta were split approximately evenly into two strongly supported clades, marked A and B in Fig. 3 . Clade A contained samples from south-western Western Australia in a triangle bounded by just south of Shark Bay, Bunbury, and Hyden. Clade B comprised samples from a northern, mostly interior, area approximately bounded by just east of Shark Bay, Geraldton, the Hamersley Lakes and Lake Way.

That this level of resolution can be achieved inside species suggests sequence capture as an attractive approach for phylogeographic studies, not least because it is able to produce data more reliably from herbarium specimens with potentially degraded DNA than are many other molecular methods.

Figures

Fig. 2. Geographic spread of the specimens of Pogonolepis at CANB sampled for molecular analysis (large, pale circles) and ranges of species according to the Australasian Virtual Herbarium (dots; see https://doi.org/10.26197/ala.6556e234- 7160-49de-bf63-5123af624e94, accessed 31 May 2022). Red: P. muelleriana; blue: P. stricta. Note that specimens of P. muel- 150 leriana geocoded in Canberra and Hobart were likely cultivated.

Fig. 3. Likelihood phylogeny of concatenated supermatrix of Pogonolepis and outgroups. Numbers above branches indicate UltraFast Bootstrap values, gene Concordance Factors, and site Concordance Factors. The dashed branch was shortened for the figure. Western Australian specimens of P. muelleriana are marked with (WA), all others are from eastern states. A and B indicate informally named clades inside P. stricta.