Bottleneck or founder effect
The haplotype network for ND2 showed the predicted mtDNA starburst pattern for a population that had experienced a bottleneck or founder effect, i.e. a star-like shape where core haplotype is shared by most populations and then many less-common peripheral haplotypes (Ferreri, Qu, & Han, 2011; inset Figure 3). This result was upheld byTajima’s D , where a negative result indicated a population expansion after a bottleneck event (Ramakrishnan, Hadly, & Mountain, 2005); for this study we were unable to determine whether this pattern was due to a bottleneck versus a founder effect. Either instance may be due to changes in habitat distribution during the last Pleistocene glacial period —- a possible extrapolation of the Pleistocene refugia hypothesis (Grubb, 1982). The Pleistocene refugia hypothesis (sometimes referred to as Holocene Refugia) suggests that repeatedly expanding glaciers toward the end of the Pleistocene epoch resulted in a large proportion of speciation events, especially among sister taxa (Avise, 2009; Avise & Walker, 1998). While the Pleistocene refugia hypothesis has previously undergone criticism stating many of these divergence events preceded the Pleistocene (Knapp & Mallet, 2003; Whitten, 1979), there remains a large number of studies across taxa where speciation or isolation events correlated with major changes in climate ~37 KYA (e.g. Foltz, Nguyen, Kiger, & Mah, 2008; Johnson & Cicero, 2004; Quick et al., 2016; Ribera & Vogler, 2004; Vörös et al., 2017).
However, Africa did not have glaciers in the Pleistocene, and so changes in distribution of C. frenatus were likely due not to glaciers specifically, but to shifts in vegetation structure from changing weather patterns during the last glacial period. As the fynbos experienced a major shift wherein ericaceous fynbos contracted >96 KYA, expanded ~96–37 KYA, and again retreated ~37 KYA (Quick et al., 2016), we suggest two possibilities in relation to this hypothesis: (1) C. frenatus may have been confined to one mountain range when fynbos retreated >96 KYA, then radiated out from this bottleneck population when fynbos expanded ~96–37 KYA; or (2) C. frenatus was extirpated (or not initially present) in the Cape Fold Belt, but the expansion of ericaceous Fynbos ~96–37 KYA, along with shrinking surrounding Afromontane forests, leading to colonisation by a founder population of the sister C. aurantiuslineage. In the second scenario, speciation may have been promoted by geographic isolation resulting in the current differences between the two species. However, the bottleneck or founder population of C. frenatus may also have originated much earlier during the Miocene, as Cape Fold uplift during this time period has been suggested as the main source of speciation among fynbos plant communities (Cowling, Procheş, & Partridge, 2009; Pirie et al., 2016).
From a purely genetic perspective, our results suggest C. frenatus and C. aurantius have low species-level diversity based on uncorrected p -distance values. Our recorded cytbp- distance result for C. frenatus -C. aurantius (1.5 %) was below most p -distance values for species-level distances of 3.3–12.8 % in passerines (Liu et al., 2016; Luo et al., 2014; Martens, Tietze, & Sun, 2005), although (Martens, Tietze, & Päckert, 2011) suggested sister-species can be <3 %. Similarly, our recorded ND2 p -distance result for C. frenatus -C. aurantius (4.3 %) was at the lower range of p -distance values for previous species-level distances of 2.0–13.7 % of passerines (Aliabadian et al., 2012; Luo et al., 2014; Zuccon & Ericson, 2012). However, this is not entirely surprising, as (Martens et al., 2011) indicate one should use caution in applying previous speciesp -distance values to apparent sister species.
Edge extinction
The mountain Fynbos of South Africa extends from the coast of the Cape Peninsula (in the very southwest corner of South Africa), to ~300 km north into the Cederberg mountains, and ~600 km east ending near Port Elizabeth. Thus, while the Cape Peninsula consists of habitat which is suitable forC. frenatus , there have been no records of birds on these mountains in recorded history (Cohen and Frauenecht 1993); indeed, there are also no records from the eastern coastal edges of mountain fynbos habitat (Lee & Barnard, 2016). Species extinction is often preceded by initial habitat fragmentation and increasing distance between suitable patches, resulting in local population extinction on the edges of a species distribution (Ceballos & Ehrlich, 2002; Woodroffe & Ginsberg, 1998). Conceivably C. frenatus were once located in this area, but have since been extirpated. This may have been due to human encroachment (or human-associated species such as domestic cats and dogs, or invasive rats) or the area became unsuitable due to too many years between fires. In the first instance, it may be that C. frenatus do occasionally recolonize the mountains only to be again driven off by anthropogenic disturbance, and in both instances it may be that once C. frenatus edge territory populations become extinct they are unlikely to recolonize [as suggested by Ceballos and Ehrlich (2002)]. Such edge extinction may have occurred as recently as the last three decades as despite seemingly suitable habitat on the Lady Slipper mountain range near Port Elizabeth, Eastern Cape, where birds were recorded by SABAP1 (1987 - 1991), there are no records from the more recent SABAP2 (2007 to present).
Increased variation at the edge of species’ distributions has been attributed to reduced dispersal and gene flow (Sittenthaler et al., 2018), fragmentation, isolation, genetic drift, and small population size (Arnaud‐Haond et al., 2006; Böhme, Schneeweiß, Fritz, Schlegel, & Berendonk, 2007), or human conflict (Woodroffe & Ginsberg, 1998). In addition, while samples collected from the furthest east known occurrence of C. frenatus in 2017 (i.e. Kleinrivier), none were recorded on return in 2019 (pers. obs. KNO). As in Black GrouseTetrao tetrix (Sittenthaler et al., 2018), the greatest genetic variation for C. frenatus was found in edge populations (i.e. Kogelberg; inset Figure 3). In all cases, these edge populations are placed at increased risk of extinction (Ceballos & Ehrlich, 2002; Woodroffe & Ginsberg, 1998).
Conclusions
Genetic theory predicts reduced effective population size will result in a substantial loss of genetic variation, corresponding to a reduction in allele number and heterozygosity at polymorphic loci (Nei, 1978; Varvio, Chakraborty, & Nei, 1986). It thus seems possible that decliningC. frenatus populations are resulting in an overall loss of genetic diversity, although low diversity may also be due to selective pressure. While we did not detect any geographical genetic structure inC. frenatus populations, their overall low genetic diversity suggests further research is needed to conserve remaining populations and dispersal avenues. Moreover, as C. frenatus habitat continues to warm, and suitable alpine Fynbos retreats upslope, it seems likely climate change will impact gene flow in the future. Our finding of low genetic variation and no genetic structuring may be complicated by the specialized niche inhabited by C. frenatus ; that being said, a lack of genetic diversity may indicate an inability to adapt to changing environments. The signature of a bottleneck/founder effect withinC. frenatus suggests an interesting link between paleoclimate, the potential for climate refugia, and current species distributions in the Fynbos. While this is the first genetic structuring study on an avian endemic of the Cape Fold Belt sky islands, our study provides insights into processes that may have impacted speciation and evolution within this unique study system.