4.3 Inferring key drivers of diversification in the small Puffinus
Based on our inferred date of the Indian –Atlantic lineage split, gene exchange between Indian and Atlantic birds have apparently occurred after the closure of the Isthmus of Panama: we suggest that this happened through individuals going around South Africa, since African continent is an insurmountable barrier for Procellariformes (Silva et al. 2015). Indeed, these shearwaters are tropical or subtropical species (at least currently). Off South Africa, until 1 My ago, sea surface temperatures (SST) were approximately 2°C higher than today (Bell et al. 2015), suggesting that migration between Atlantic and Indian oceans may have remained possible for such birds. From 2.0, a strong decrease of SST occurred in both oceans (Bell et al. 2015), and gene flow between Indian and Atlantic may thus have ceased, in agreement with our estimated time of divergence (1.76 My ago). Once the Atlantic birds were isolated from Indo-Pacific populations, differentiation started to occur among Atlantic lineages 1.38 to 1.90 My ago (respectively for the 9 markers or only the mtDNA markers). This period also corresponds to a further decrease of the SST in the Atlantic (Bell et al. 2015), a southward shift of the subtropical front and warmer waters in the Southern Ocean (Maiorano et al. 2009), and sea ice development in the North extending southwards from the Arctic to the current Great lakes in the USA (Webb & Bartlein 1992). Cold temperatures, preventing the colonization of potential northern breeding sites such as the Azores, may have forced shearwaters to spread over the two sides of the Atlantic (Fig. 2d). The third step in the colonization of Atlantic breeding grounds concerns the divergence of baroli from boydi,which would have occurred around 0.85-1.26 My ago, a period that corresponds to a stabilization of the SST at the current level in the North Atlantic. The ice melt may then have allowed northward colonization on both sides of the Atlantic, from Cape Verde to the Canaries, Madeira and the Azores, and from the lesser Antilles to the Bahamas and Bermuda. Similar timing of divergence has been suggested between Calonectris edwardsii (Cape Verde) and C. diomedea(North Atlantic and Mediterranean) 0.7-0.9 My ago (Gómez-Díaz et al. 2006) and Puffinus olsoni (Canaries) and Puffinus puffinus(North Atlantic) 0.2-1.0 My (Ramirez et al. 2010). In the Indian Ocean, a reversed pattern of southward colonization occurred, shearwaters colonizing from Seychelles, partly continental in origin, to Réunion 0.70-1.01 My ago, precisely at the time strong volcanic activity ended on Réunion (Gillot & Nativel 1989). Mauritius was probably colonized long before (Mauritius age: 8My; McDougall & Chamalaun 1969), then Rodrigues and Réunion (both about the same age, 2 My; McDougall 1971), but shearwaters are now extinct on Mauritius and Rodrigues, following human colonization since the 17th century, so no sample is available. The southward movement was however not necessarily related to changes in SST, but rather to availability of volcanic islands that eventually emerged in a southward direction.
Mantel tests showed a strong correlation between geographic and genetic distance between lineages, i.e. at large scale. Indeed Little shearwaters are poor flyers compared to other Procellariiformes. However, isolation-by-distance within each lineage was visually not detected, therefore distance alone could not be a factor of population divergence at this smaller scale; we argue that sea temperature could rather be the most important factor of divergence in our case at this scale. Seabirds indeed depend on both sea and the islands where they breed, thus sea temperature has strong impacts on seabird phenology, breeding, survival and abundance (see Sydeman et al. 2012 for a review). We suggest that foraging ecology, which strongly depends on SST, is an important process shaping divergence among lineages. The segregation of foraging areas among populations is an important factor of differentiation among seabirds (Friesen 2015; Friesen et al. 2007), as shown for other shearwaters (Genovart et al. 2007; Gómez-Díaz et al. 2006), petrels (Gangloff et al. 2013; Welch et al. 2012), storm-petrels (Deane 2013; Smith et al. 2007) and albatrosses (Alderman et al. 2005; Burg & Croxall 2001). Assessment of non-breeding and breeding distributions at sea of the little shearwaters complex revealed that all three Atlantic taxa show rather separated foraging and wintering areas (Ramos et al. in review), and further suggest that boydi rather than lherminieri was ancestral in the North Atlantic. Indeed,boydi is more flexible in its foraging ecological niche, suggesting ancestral behavior, with a far larger potential distribution at sea covering all central North Atlantic (Zajková et al. 2017, see map in Ramos et al. in review). In addition, over the last My, SST oscillations gradually increased in amplitude with up to five degree difference in range (Bell et al. 2015; Herbert et al. 2011). These oscillations were showed to correlate with changes in marine productivity (Martı et al. 2009), prey species diversity (Yasuhara & Cronin 2008), atmospheric circulation (Chang et al. 2000) and sea level (e.g. in Atlantic Nascimento et al. 2011, Zazo et al. 2010). It is likely that these oscillations also contributed to divergence in the North Atlantic, and we suggest that significant Tajima’s D tests found in almost half of the populations studied represent traces of past bottlenecks and population expansions as it has already been shown in other taxa (Ramakrishnan et al. 2005; Weber et al. 2004; Zhu et al. 2006).
A last interesting side effect of our study is that it suggests that the small black and white shearwaters have shown a very recent radiation speciation event, with not less than 13 species radiating in just 1.46 million years since P. puffinus , P. assimilis and P. newelli clades are either embedded in lherminieri orbailloni clades. All these species are rather coastal shearwaters (compared to more pelagic species such as the larger shearwaters), and such high rate or speciation may be the result of the high climatic oscillation that occurred over the last 2 million years which may have favored high rates of colonization and extinction on coastal islands.