Traditionally, the species of Calonectris were usually placed in the same genus as the other Puffinus shearwaters; the former genus came to be widely recognized only about 50 years after its description, based mostly on the work of Kuroda (1954) (e.g. Alexander et al. 1965). In the final phylogenetic hypothesis of our study, Calonectris is supported by eight unambiguous synapomorphies, such as the condylus occipitalis nonvisible in lateral view of the skull; the presence of a thin osseous lamina in apertura nasalis ossea, which is shared with Puffinus nativitatis and P. huttoni; and the elliptical shaft of humerus, not flattened as in other species of Puffinus (sensu lato) (characters 24, 34, and 69, respectively); all three of these were also cited by Kuroda (1954). The monophyly of the group, corroborated herein, has also been recovered by previous molecular-based studies, despite the different taxonomic treatments given to the terminal taxa (e.g. Nunn & Stanley 1998; Austin et al. 2004; Penhallurick & Wink 2004; Welch et al. 2014).
In our phylogenetic analysis, relationships within the genus Calonectris were almost entirely established by synapomorphies from continuous characters, such as the lengths of wing bones (e.g. characters 9, 10, 11, and 12), and received high support (RBS ≥ 70 for all nodes). Calonectris diomedea and C. borealis are sister taxa to each other, sharing seven synapomorphies from continuous characters. Most molecular phylogenies have included only one or two species of Calonectris as terminal taxa. Ramirez et al. (2010) and Tennyson & Shepherd (2017) included only C. diomedea, whereas some other studies also sequenced C. leucomelas (e.g. Nunn & Stanley 1998; Austin et al. 2004; Pyle et al. 2011; Welch et al. 2014). Heidrich et al. (1998), when assessing relationships of Mediterranean and North Atlantic shearwaters through mtDNA, used both C. diomedea and C. borealis, although borealis was considered a subspecies of C. diomedea. In the present study, C. leucomelas appeared as sister to (C. diomedea + C. borealis), forming a clade supported by eleven synapomorphies (i.e., nine from continuous characters, plus one unambiguous and one ambiguous). Penhallurick & Wink (2004), who included three of four species of Calonectris, found C. leucomelas to be sister to (C. diomedea + C. borealis), the same relationship we present herein. However, these authors suggested that these two taxa should continue to be treated as subspecies, whereas most of the recent literature considers C. diomedea and C. borealis to be valid species (i.e. Sangster et al. 2012; Gill & Donsker 2019).
Together with C. leucomelas, C. diomedea, and C. borealis, the present study has also included C. edwardsii, which was found to be sister to (C. leucomelas (C. diomedea + C. borealis)). This result contrasts with the relationship proposed by Gómez-Díaz et al. (2006, 2009), who also included all four species in their phylogeographic analyses based on the cytochrome b gene. According to these authors, C. diomedea and C. edwardsii are more closely related to each other than either is to C. borealis, and C. leucomelas is sister to all other Calonectris species. Nonetheless, when discussing only the biometric structure of Calonectris species, Gómez-Díaz et al. (2006) acknowledged the similarity of C. diomedea and C. borealis, the largest species that they included. According to them, C. leucomelas and C. edwardsii are, respectively, the medium-sized and the smallest species of the genus. As mentioned before, the relationship proposed herein was mostly based on synapomorphies from continuous characters, which can explain the apparent clustering by body size measurements.