Coalescence-based demographic analyses shed light on complex evolutionary scenarios.
Despite the different analyses that support the presence of three genetic clusters, one of them potentially resulting from the hybridisation of the other two, we could not rule out alternative explanations. To investigate further the underlying processes responsible for the observed pattern, we conducted a D-statistic test to differentiate between gene flow and ILS. The results found no significant support for gene flow between the three genetic clusters. Moreover, we observed symmetrical counts of the ABBA and BABA patterns, indicating no excess of shared alleles. Given the lack of evidence for gene flow, we explored alternative demographic models to understand the evolutionary history of the lineages better. By considering multiple evolutionary processes simultaneously, demographic analyses can disentangle complex scenarios that may produce ambiguous or conflicting signals, providing deeper insights into the species’ evolutionary dynamics (Galià-Camps et al., 2024; Puckett & Munshi-South, 2019). We proposed five scenarios encompassing the observed patterns in genomic structuring and D-statistic analyses. These scenarios suggested: the existence of three recently diverged independent lineages, reflecting the potential effects of incomplete lineage sorting (Models A and B, Figure S1), ongoing gene flow as evidenced by the potential hybrids identified in Porto Santo (Model C, Figure S1), and, finally, we also considered the possibility that H. insularum PS arose from secondary contact betweenH. insularum MBD and H. maderiana , either episodic, without contemporary gene flow (Model D), or with contemporary gene flow (Model E, Figure S1). The scenario of three lineages showing high levels of co-ancestry was preferred over those involving hybridisation. Additionally, the effective population size estimation of theHogna lineages, as inferred by the stairway plot 2, identified an ancestral split on Porto Santo, followed by a similar pattern of population growth between the two resulting lineages. Interestingly, the next split, which corresponds to the origin of theH. insularum MDB lineage, shows a much steeper, continuous population expansion compatible with a bottleneck following the colonisation of new islands (Madeira, Desertas) from the ancestral Porto Santo. This founder event inevitably depleted genetic variation (Cerca, Cotoras, Bieker, et al., 2023) and helps to explain the exclusive mtDNA haplotypes found in Desertas and, especially Madeira.
Results of snapp, however, failed to support the topology selected by the demographic analyses and revealed instead a high level of discordance among alternative tree topologies of Hogna lineages. High levels of gene-tree discordance suggest the occurrence of hybridisation and/or ILS (Mallet et al., 2016). Since our previous analyses discarded gene flow, we attribute that observed discordance to the presence of ILS. Short speciation times and large effective population sizes may result in the retention of ancestral polymorphisms across species (Maddison, 1997). Although they do not directly influence the description of the observed patterns, it is worth noting that our divergence time estimates differed across methods and from those based on mitochondrial data by Crespo et al. (2022) (~1.2 million years ago). These discrepancies could be attributed to the inherent characteristics of the methods: StairwayPlot is more informative for events spanning hundreds of generations, whereas fastsimcoal2 is better suited for detecting more recent demographic changes (Lapierre et al., 2017; Patton et al., 2019).
Regardless of the times, the inferred patterns of ancestral population sizes reconcile well with the higher levels of retained polymorphisms observed between H. maderiana and H. insularum PS, and, as stated above, the reduced polymorphisms in genome-wide data and mitochondrial haplotypes observed in H. insularum MDS.