Genetic divergence in the Northern Fulmar
Both mitochondrial markers indicated strong divergence between Atlantic and Pacific populations of Northern Fulmar, supporting the pattern initially suggested by DNA barcoding results (Kerr et al. 2007; Johnsen et al. 2010) but differing from the current taxonomic recognition (American Ornithologists' Union 1998). However, not even the fast-evolving control region indicated any clear divergence within the Atlantic population, thus failing to support the presence of two separate evolutionary significant units in that region. Results from nuclear markers were less clear. Haplotype sharing occurred in each of the three markers but to varying degrees. Sharing was most prominent for CHD1-Z, which featured almost no variation at all. Consequently, haplotype networks were much more informative than the concatenated nuclear tree. The networks demonstrate that rare haplotypes were not shared between the Atlantic and Pacific population, suggestive of early-stage divergence.
The lack of diversity at the CHD1-Z locus is slightly unusual. The Z chromosome has a smaller effective population size than its autosomal counterparts on account of its single-copy occurrence in females. Additionally, sexual selection is also thought to result in faster divergence of Z-linked genes (Borge et al. 2005). The low diversity observed within and between populations might seem to implicate male-biased gene flow and question the integrity of any postzygotic boundaries between the populations since the Z chromosome has recently been regarded as a “hotspot” for speciation genes (Sæther et al. 2007). However, CHD1-Z did not just demonstrate low variation between populations but also between other Fulmarine species. The CHD1-Z sequence in the focal taxa only differed from its congener, F. glacialoides, by two substitutions. Additionally, F. glacialoides sequences only differed from those of the genus Macronectes by a single substitution, and the two Macronectes species showed no interspecific variation at all. Further to this point, female Procellariiformes are known to disperse more than males (Burg et al. 2003). These results suggest “intron A” of the Z chromosome may simply be largely conserved in the Fulmarini clade.
In general, the discrepancy between the mitochondrial and nuclear markers is not surprising and is consistent with many similar studies (Welch et al. 2011; Gangloff et al. 2013). Time to reciprocal monophyly is generally correlated with generation time and effective population size (Avise et al. 1984; Hudson and Coyne 2002). Both of these measures are relatively large in the Northern Fulmar, with generation time generally estimated around 20 years (Koons et al. 2006; Jones et al. 2008) and census estimates suggesting both populations include millions of breeding pairs (Brooke 2004). These values collectively would predict a disproportionately long time to reciprocal monophyly relative to the divergence data, even for mitochondrial markers, though it should be borne in mind that modern population estimates might not be reflective of population trends at the time of divergence. Using values from Table 1 in Hudson and Coyne (2002) and our own divergence date estimates, it can be inferred that the long-term effective population size that would allow the observed mitochondrial divergence between Atlantic and Pacific birds would range from roughly 17,000 to 75,000 individuals/population. In turn, even the low end of this range would potentially require 2.9 million years to observe with high probability the reciprocal monophyly of a nuclear marker. This is consistent with our data, where nuclear markers appear to display reciprocal monophyly between F. glacialis and F. glacialoides, which our data suggests diverged ~3 Mya.
The above estimates depend on the accuracy of the generation time estimates for Northern Fulmars. For example, if generation time were in fact closer to the average age at first reproduction (8–12 years, Hatch and Nettleship 1998) then the above population size estimates would double, but the values would still remain much below the current census population estimates. The settlement of the northern hemisphere is hypothesized to have followed a founder event from a southern ancestor traveling up the Pacific coastline (Voous 1949), which would likely result in a smaller initial population size in the north and subsequently a shorter predicted time to reciprocal monophyly. The division between the Atlantic and Pacific colonies on the other hand is more likely to have followed a vicariant event (i.e., the extension of Arctic ice), which theoretically would cleave the northern species into two more evenly sized populations. This too would help explain why monophyly is observed between F. glacialis and F. glacialoides, but not between F. glacialis populations. Detracting from this logic is the rapid population growth observed in the North Atlantic, where newly colonized locations such as the Faeroe Islands have seen populations surpass 500,000 breeding pairs in under 200 years (Fisher 1952). Alternatively, selective sweeps might also help explain the degree of mitochondrial divergence in F. glacialis, but evidence for such events is still largely lacking (Kerr 2011). The smaller effective population size of the mitochondrial genome appears to remain the most parsimonious explanation for the mito-nuclear discrepancy.
Northern Fulmars in the Pacific tend toward smaller body size and slimmer bills, whereas Atlantic birds are larger on average and bill size varies clinally (being smaller in the north), but measurements for both features overlap between these two regions (Brooke 2004; Pyle 2008; Howell 2012). Morphological differences are difficult to encapsulate due to the variable plumage, but Pacific birds are noted for having a darker tail that contrasts against a paler rump, whereas Atlantic birds are more uniformly colored across the tail and rump (Sibley 2000). Unfortunately, this feature is largely obscured or completely lost in Pacific dark morphs. Avian taxonomic decisions are often skewed toward characters that are diagnosable in the field, but this sometimes obscures distinct evolutionary lineages (Watson 2005).
One of the inherent challenges of studying divergent populations in allopatry is the inability to test the criteria of the biological species concept under natural conditions, i.e., assortative mating in sympatry (Mayr 1963). A common approach to taxonomic decisions for allopatric populations is to make comparisons to closely allied species pairs that do occur in sympatry (Helbig et al. 2002). Some authors have suggested that differences between Northern Fulmars from the Atlantic and Pacific are on par with that observed between other Procellariid species (Howell 2012). A notable example is Macronectes giganteus and M. halli, which differ only subtly by primarily qualitative morphological characters and exhibit far less genetic differentiation than the Northern Fulmar populations (Techow et al. 2010). Other examples, such as Pterodroma phaeopygia and P. sandwichensis echo the case of Macronectes, but these species breed on separate islands and may not be considered sympatric per se (Welch et al. 2011). It is hypothesized that high natal philopatry reduces selection for plumage divergence in Procellariids, which makes species recognition within this group very challenging in general (Brooke 2004).
The Northern Fulmar shares a common phylogeographic distribution with a few other seabird species and allospecies pairs, for example, Uria aalge (Morris-Pocock et al. 2008), U. lomvia (Birt-Friesen et al. 1992), and Fratercula arctica/corniculata (Friesen et al. 1996). Interestingly, divergence dates for these others pairs are variable, with the youngest only being 56,000–226,000 years ago (Morris-Pocock et al. 2008). Land and ice are common barriers to dispersal for all of these species (Friesen et al. 2007), so it is curious that they don't share a common divergence date. These dates are primarily predicted based on molecular clock estimates, so it could be that rate heterogeneity is underappreciated. However, the Northern Fulmar's exceptional generation time would predict a slower evolutionary rate compared to the other species, yet it yields the oldest divergence date for the group. If the other species were able to cross the Arctic passage at later dates, it would be reasonable to surmise that there would be no barrier to prevent the Northern Fulmar from completing similar movement, but the integrity of the populations remains intact.
Two historical features remain at odds with our findings: the Fulmarus fossil record and the Pacific coast origin hypothesis. Two prehistoric fossils from Kern County, California are attributed to the genus, including the proximal end of a carpometacarpus identified as F. hammeri (Howard 1969) and a complete humerus identified as F. miocaenus (Howard 1984). These fossils were said to belong to the Barstovian and Clarendonian stages, respectively, which roughly range from 10 to 16 Mya. This timeline vastly predates our estimates for the time to most recent common ancestor for even Fulmarus-Macronectes, and instead is more in line with the origin of the entire Fulmarini clade. Given that we used well-accepted fossil calibrations to obtain our date estimates, we propose that the Howard fossils likely belonged to either an ancestral lineage or to independent lineages with no modern descendents, but in either case should not be included in the genus Fulmarus. Wetmore (1926) describes a fossilized fragment of a left humerus collected in Maryland that is so similar to the modern day species that it “cannot be distinguished from [Fulmarus] glacialis”, but while he speculated that this fossil dated back to the Miocene (on account of its dark coloration and degree of fossilization), he also stated that the stratum from which it was collected is more consistent with a Pleistocene origin. The latter scenario would be consistent with our data.
Voous (1949) proposed that the Northern Hemisphere was colonized by ancestral fulmars traveling up the cooler water of the west coast of the Americas. He used bill shape to support this hypothesis, indicating that the slender bill of F. glacialoides was the ancestral condition and the broader bill of the North Atlantic F. glacialis was the most derived (the relatively slim bill of Pacific birds would thus reflect an intermediate state). This hypothesis has perpetuated unchallenged (Hatch and Nettleship 1998), though before Voous (1949) it was thought that the Atlantic was colonized prior to the Pacific (Fisher and Waterston 1941). Contrary to Voous' hypothesis, our molecular data reveal more derived characters in the sequences of Pacific birds for four out of five markers (CHD1-Z was the exception because the two populations predominantly shared a single haplotype). This could detract from the Pacific coast origin hypothesis, though it is possible that the increased divergence in the Pacific is reflective of that region's smaller population size, which can have a positive effect on the molecular evolutionary rate (Ohta 1973).
MC1R and color morphs in the Northern Fulmar
The association between MC1R mutants and melanism was first deduced in model laboratory species such as mouse and subsequently chicken, but a common role was later confirmed in such disparate avian taxa as skuas, geese, and passerines (see Mundy 2005 for review). MC1R has since been regularly employed as a candidate gene in avian plumage studies, though often arguably in vain. Although MC1R-melanism correlations have been found in additional avian taxa (Baiao et al. 2007; Pointer and Mundy 2008; Uy et al. 2009; Vidal et al. 2010), a near equal number of studies have unsurprisingly yielded negative results (MacDougall-Shackleton et al. 2003; Cheviron et al. 2006; Haas et al. 2009; Hull et al. 2010; Cadena et al. 2011; Dobson et al. 2012). Several of the latter studies examined species that exhibited finely scaled patterns, not melanism per se (MacDougall-Shackleton et al. 2003; Cadena et al. 2011; Dobson et al. 2012). Others examined species where melanic plumage was limited to a single sex and age class (Cheviron et al. 2006).
In the absence of a functional analysis, there is an intrinsic risk of falsely attributing cause to MC1R mutations when it could simply correlate with plumage variation because of population divergence. Doucet et al. (2004) described an association between MC1R and melanism in the island subspecies of Malurus leucopterus, which features melanic plumage in males, but Mundy (2005) correctly speculated that this association might be an artifact of population divergence, as was later demonstrated by Driskell et al. (2010). In his review, Mundy (2005) offers strong evidence for how demographic history can be ruled out as an explanation for the association in at least three iconic species, which bolsters the role of MC1R in at least some cases. Interestingly, shared amino acid substitutions are often recorded in unrelated species (Mundy 2005), which seemingly gives further credence to causation, but these substitutions are also found in nonmelanic species (Pointer and Mundy 2008). If MC1R mutations do in fact play a role in determining color morph in the Northern Fulmar then a strenuous explanation would be required to explain the mechanism.