Comparative phylogeography of four Indo-Pacific scarine labrids: an insight into the evolutionary patterns of reef fish

This thesis examines the genetic structure of four widely distributed Indo–Pacific parrotfish populations using a combination of phylogenetic and phylogeographic analyses. These data were used to identify spatial and temporal patterns of population structure, genetic diversity and the underlying historical processes for each species. The four species Chlorurus sordidus, Scarus ghobban, S. rubroviolaceus and S. psittacus have relatively recent evolutionary origins, ranging between 2, (C. sordidus), 3 (S. rubroviolaceus), and 4 million years (S. ghobban and S. psittacus) (Alfaro et al. 2009). These species also differ in their association with coral reefs; of the four species, S. psittacus is most capable of occupying shallow water habitats devoid of coral reefs, while S. ghobban can extend into much deeper water (up to 250 m) than the other three species. Sequences from the mitochondrial control region for the four species collected from their Indo–Pacific distributions were compared. The sampling data from Indo-Pacific wide collections are: C. sordidus, 354 bp, 351 individuals, 18 locations; S. rubroviolaceus, 378 bp, 292 individuals, 15 locations; S. ghobban, 350bp, 239 individuals, 12 locations; and S. psittacus, 322 bp, 164 individuals, 12 locations. The comparative analyses include data from Bay et al. (2004) on C. sordidus and Winters et al. (2010) on S. psittacus. For each species, we used Bayesian and neighbour-joining analyses to generate the best tree topology and trees were outgroup rooted using relevant sister taxa. The phylogenetic relationships were also represented as a minimum-spanning haplotype tree to visualise genetic diversity and its spatial distribution. To determine the level of genetic differentiation between individuals from different sampling locations, we used pairwise F(ST) comparisons. Pairwise F(ST) comparisons were also made between clades identified in the phylogenetic analyses. Estimates of genetic and geographical distances between populations were used to assess isolation by distance. We analysed the molecular variance to determine the source of greatest genetic variation with the analyses structured as follows: (1) geographic location with populations grouped into their respective ocean basin, (2) populations partitioned into geographic regions and (3) groupings based on clade structure established in the phylogenetic analysis. The coalescence-based program Migrate 2.3 was used to infer migration rates and directions between sampled locations. The timing of divergence between geographically defined populations, as well as phylogenetically determined clades, was determined using coalescence analyses that were calculated in Arlequin 3.01. Phylogenetic analyses revealed evidence of genetic partitioning between western Indian Ocean and Pacific Ocean populations for three of the four species. This was confirmed by significant pairwise F(ST) comparisons for C. sordidus ( F(ST IO–PO) = 0.594–0.762, P < 0.001), S. rubroviolaceus ( F(ST IO–PO) = 0.5290–0.8528, P < 0.0001) and S. ghobban ( F(ST IO–PO) = 0.748–0.803, P ≤ 0.001), but not for S. psittacus ( F(ST IO–PO) = 0.046–0.166, P < 0.01). For all species, individuals from Western Australia were associated with the Pacific Ocean clade, despite geographically located in the Indian Ocean. A genetic break associated with the central Pacific barrier was also found in C. sordidus and evidence of a third genetically distinct population at Cocos Keeling Island was detected in S. ghobban ( F(ST IO–CK) = 0.754–0.762, P < 0.001; F(ST PO–CK) = 0.766–0.851, P ≤ 0.009). We report the location of the genetic break between western Indian Ocean and Pacific Ocean populations to be at Cocos Keeling Island for S. ghobban, Christmas Island for C. sordidus and further west again for S. rubroviolaceus. Localised population structure was identified at peripheral locations for all four species. Specifically, Hawaiian populations were genetically differentiated in C. sordidus, S. rubroviolaceus and S. psittacus; Marquesan/French Polynesian populations were differentiated in C. sordidus, S. rubroviolaceus and S. psittacus; east Pacific populations were differentiated in S. rubroviolaceus and S. ghobban; Western Australian populations were differentiated in S. ghobban; and finally, Arabian Gulf and northern Oman populations were differentiated in C. sordidus. Isolation by distance was only detected in S. rubroviolaceus and was only significant within the ocean basin (Z(IO) = 290.5452, r = 0.6856, R² = 0.470, P < 0.05; Z(PO) = 51568.0352, r = 0.5428, R² = 0.295, P < 0.05). Migration estimates indicate largely uneven gene flow that was predominantly from east to west for all but S. psittacus and could be generally explained by present-day oceanographic currents. Migration analysis was not possible for S. psittacus. Coalescence calculations for the three species indicate that the timing of divergence between extant western Indian Ocean and Pacific Ocean populations of C. sordidus and S. rubroviolaceus both took place during the Pleistocene approximately 1 million years ago (mya), and that the divergence of extant populations of S. ghobban took place prior to that, approximately 2.4 mya. These coalescent ages were an order of magnitude older than the coalescent age of S. psittacus. Aspects of genealogical concordance were demonstrated in three of the four species in this study. Aspect I was evidenced by strong bootstrap support for at least two distinct lineages. Aspect II was evidenced by support for the lineages in other studies employing independent molecular markers. Aspect III of genealogical concordance was evidenced by congruent patterns of phylogenetic structure across the three codistributed species which distinguished all west Indian Ocean individuals from the east Indo-Pacific individuals. Aspect IV was evidenced by the same strongly supported lineages separated at the same biogeographic area: the Cocos Keeling and Christmas Islands. Despite a lack of phylogenetic structure at the largest spatial scale in S. psittacus, we noted that all four species had the following in common: a central population containing individuals from the eastern Indian Ocean and the central Pacific Ocean, and several smaller populations containing individuals from peripheral locations. Therefore, we inferred that isolation at peripheral locations is responsible for lowered genetic diversity and is a major force behind generating the genetic differences that ultimately lead to evolutionary novelty and reduced genetic diversity in widespread scarine labrids. Furthermore, overlap at central locations, such as at Christmas and Cocos Keeling islands, contributes to the increase in genetic diversity, which is the building block for adaptation under selection during times of environmental flux. The contrasting pattern of population structure at the level of ocean basins observed between S. psittacus and the other three species can be explained by smaller ancestral populations during Plio-Pleistocene low sea level stands owing to its higher level of habitat specificity compared with the other three species. Given enough time, and a large enough population, we would expect to see similar levels of partitioning at the Indo – West Pacific Barrier. Finally, we conclude that populations fluctuate substantially over evolutionary timescales, and different species experience these fluctuations at different times, irrespective of the evolutionary ages of the species, as evident from the variable coalescent ages of extant populations. These differences are likely associated with inherent differences in ecology