4.4. Incongruent mitochondrial/nuclear pattern
An interesting pattern was observed when species delimitation was performed with both markers: two individuals were not assigned to the same group for both genes (we re-sequenced both genes and obtained unchanged results for both individuals). On one hand individual 291, was assigned to C. verrucosa sp. A in 18S species delimitation, but to C. verrucosa sp. B for COI; on the other hand, individual 116, was assigned to C. verrucosasp. B in 18S species delimitation, but to C. verrucosasp. A for COI. This means that only for these two specimens the nuclear and mitochondrial genes were addressing different speciation patterns. Apparent incongruence between phylogenies from 18S and COI sequences was addressed already in ascidians by other authors (Pérez-Portela, Bishop, Davis & Turon, 2009; Stach & Turbeville, 2002). Even though in 18S sequences 70 polymorphic sites were found, only a single site was dividing the sequences into two groups that were congruent with mitochondrial and morphological inference, then other 69 variable sites were explaining intraspecific variability, or difference among the species (A and B) with four basal groups (see results section 3.3. and discussion section 4.1.). While C. verrucosa sp. A was characterized by a thymine in the unique variable site of 18S sequences, C. verrucosa sp. B was characterized by a cytosine. Specimen 291 presented in this variable site two overlapped peaks (thymine and cytosine). This can be explained under two different assumptions, on the one hand, it can be expected that C. verrucosa sp. A is characterized by a thymine in the unique variable site, but at the same time presents a low-frequency second allele; this allele carries a cytosine in the variable site, which would be predominant in the C. verrucosa sp. B . Thus, the individual 291 would be heterozygous. On top of this, it is also possible that with the primers that we used to sequence 18S we sampled by PCR only a proportion of the allele population, and missed a group of molecules (null allele) that represents, for example, a heterozygote group of individuals. On the other hand, specimen 291 can be interpreted as a hybrid among C. verrucosa sp. A and sp. B .
In order to explain the crossed pattern of these two specimens, some hypotheses can be considered: a) introgression, b) ancestral polymorphism with incomplete lineage sorting or c) convergent mutation. Hybridization and introgression patterns were already proposed in ascidians (Bouchemousse, Liautard-Haag, Bierne & Viard, 2016; Nydam et al., 2017). Broadcast spawners present a reproductive system that has a higher probability of introgression/hybridization than others. Whilst artificial cross-fertilization among Ciona intestinalis Type A and B was demonstrated, hybrids remained infertile (Caputi et al., 2007; Sato et al., 2014). Once hybridization may occur, backcrossing with the parental types is more likely since genomes would be more similar than between species (Edmands, 2002). In this study, specimen 291 may represent the small population of hybrids among C. verrucosasp. A and sp. B , while specimen 116 could be the result of an event of introgression, in which it might inherit allele A at COI from one parent and allele B at 18S from the other. In the same way, there is evidence of introgression within C. intestinalis Type A and B, species that show high genetic divergence (12.4%) at COI (Nydam & Harrison, 2011; Nydam et al., 2017). These cryptic species, newly elevated to species status: C. intestinalis and C. robusta(Brunetti et al., 2015) show historical rates of gene flow, with almost complete absence of contemporary hybridization (Bouchemousse et al., 2016; Roux, Tsagkogeorga, Bierne, & Galtier, 2013). This evidence supports the hypothesis of introgression within C. verrucosa sp. A and sp. B which presented, in this work, 10% genetic divergence at COI. Under those circumstances, it would be of interest to experimentally test hybridization among C. verrucosa sp. Aand sp. B . Ancestral polymorphism with incomplete lineage sorting (ILS) is another hypothesis that must be discussed under these results, it is difficult to distinguish this evolutionary process from introgression since both produce similar patterns. Discrepancies between trees inferred with mitochondrial and nuclear genes can be explained by these non-exclusive evolutionary hypotheses (Zhou et al., 2017). Nevertheless, bearing in mind the high genetic divergence among C. verrucosa sp. A and sp. B , ILS is unlikely under neutral evolution (Pamilo & Nei, 1988), since the loss of polymorphism and the fixation of an allele are positively correlated with time since divergence (Maddison, 1997; Wendel & Doyle, 1998). There is circumstantial evidence that ascidians may be characterized by an elevated rate of molecular evolution (Delsuc et al., 2006). Therefore, a convergent mutation can also explain that individual 116 presents allele B in 18S, but being assigned to C. verrucosa sp. A for COI, given that the thymine mutates to a cytosine in the unique variable site.