Taxonomic revision and phylogenetic analysis of the flightless Mancallinae (Aves, Pan-Alcidae)

Abstract Although flightless alcids from the Miocene and Pliocene of the eastern Pacific Ocean have been known for over 100 years, there is no detailed evaluation of diversity and systematic placement of these taxa. This is the first combined analysis of morphological and molecular data to include all extant alcids, the recently extinct Great Auk Pinguinus impennis, the mancalline auks, and a large outgroup sampling of 29 additional non-alcid charadriiforms. Based on the systematic placement of Mancallinae outside of crown clade Alcidae, the clade name Pan-Alcidae is proposed to include all known alcids. An extensive review of the Mancallinae fossil record resulted in taxonomic revision of the clade, and identification of three new species. In addition to positing the first hypothesis of inter-relationships between Mancallinae species, phylogenetic results support placement of Mancallinae as the sister taxon to all other Alcidae, indicating that flightlessness evolved at least twice in the alcid lineage. Convergent osteological characteristics of Mancallinae, the flightless Great Auk, and Spheniscidae are summarized, and implications of Mancallinae diversity, radiation, and extinction in the context of paleoclimatic changes are discussed.


Introduction
Alcidae Leach 1820 is a clade of pelagic wing-propelled-diving Charadriiformes Huxley 1867 including 23 extant species with an exclusively northern hemisphere distribution (del Hoyo et al. 1996). Th e fossil record indicates that alcid diversity during the Late Miocene (11.6-5.3 mya) and Early Pliocene (5.3-3.6 mya) equaled or exceeded extant alcid diversity (Olson 1985;Olson and Rasmussen 2001;Dyke and Walker 2005;Smith et al. 2007), although systematic evaluation of fossils referred to Alcidae is needed to refi ne estimates of paleodiversity in the clade. Additionally, the systematic position of most extinct species referred to Alcidae have yet to be evaluated in a phylogenetic analysis.
Primarily owing to the penguin-like characteristics of the fl ightless Great Auk Pinguinus impennis (Linnaeus 1758), alcids were grouped systematically with penguins and other 'waterbirds' including loons, grebes, and ducks by many ornithologists in the 18 th and 19 th centuries (Linnaeus 1758;Vigors 1825;Brandt 1837;Swainson 1837;Coues 1868), and this misconception lingered well into the 20 th century (Verheyen 1958). However, there is consensus among modern classifi cations with regard to the placement of Alcidae in a monophyletic Charadriiformes (Ridgway 1919;Storer 1960;American Ornithologists' Union 1998). Analyses of morphological (Strauch 1978;Björklund 1994;Chu 1995;Zusi 2006, 2007;Livezey 2009Livezey , 2010Mayr 2011) and molecular data (Sibley and Ahlquist 1972;Sibley and Ahlquist 1990;Ericson et al. 2003;Paton et al. 2003l;Th omas et al. 2004;Cracraft 2004;Paton and Baker 2006;Fain and Houde 2007) support the charadriiform affi nities of Alcidae. Furthermore, phylogenetic analyses of molecular data with dense taxonomic sampling for Alcidae support the monophyly of an extant alcid clade (Th omas et al. 2004;Baker et al. 2007;Pereira and Baker 2008). Previous morphology based analyses of alcid relationships have been limited with respect to taxon sampling. Th e compatability analysis of Strauch (1978) included only three alcid species and the subsequent analysis of alcid relationships (Strauch1985) did not include any outgroup taxa. Th e parsimony based analysis of alcid relationships by Chandler (1990a) was limited to a hypothetical outgroup terminal. Th e recent morphology based analyses of Zusi (2006, 2007) Livezey (2009Livezey ( , 2010 and Mayr (2011) included Alcidae as a single, taxon level terminal.
Although all extant alcids are volant, two lineages of extinct fl ightless auks are known. Th ese fl ightless auks superfi cially resemble penguins, and share many morphological features convergent with those southern hemisphere wing-propelled divers such as an elongated fi rst metacarpal and humeri with anteriorly rotated humeral heads (Miller and Howard 1949; see Appendix 1). During the Miocene and Pliocene a diverse assemblage of alcids including the fl ightless Great Auk Pinguinus Bonnaterre, 1790, and other volant auks such as Alca Linnaeus, 1758 andMiocepphus Wetmore, 1940 were present in the Atlantic Ocean (Olson and Rasmussen 2001;Wijnker and Olson 2009). Similarly, during the Miocene and Pliocene the Pacifi c was inhabited by a lineage of fl ightless alcids known as the Mancallinae Brodkorb 1967. Although Mancallinae (contents = Mancalla Lucas, 1901+ Praemancalla Howard, 1966sensu Brodkorb 1967;Ol-son 1985) and Pinguinus share several morphological characteristics related to extreme adaptation for wing-propelled diving and the subsequent loss of aerial fl ight (Fig. 1), phylogenetic results indicate that these taxa are not closely related within Alcidae. Pinguinus is consistently recovered as the sister taxon to Alca (Chandler 1990a;Moum et. al. 2002;Baker et. al. 2007;Pereira and Baker 2008), and previous phylogenetic analyses place Mancallinae as the sister taxon to all other Alcidae (Chandler 1990a;Smith 2008), suggesting that fl ightlessness evolved separately in Mancallinae and Pinguinus.
Fossil records of Mancallinae are restricted to the northern Pacifi c Ocean basin. Miocene and Pleistocene aged fossils have been reported from Japan (Hasegawa et al. 1988;Kohno 1997;Fig. 2), although these remains have not been systematically described or fi gured in publication. In contrast to the sparse record of the clade from the western Pacifi c Ocean, thousands of mostly isolated remains are known from California, USA and northern Baja California, Mexico (Miller and Howard 1949;Howard 1966Howard , 1968Howard , 1970Howard , 1971Howard , 1976Howard , 1978Howard , 1981Howard , 1982Chandler 1990b;Fig. 2), and range in age from Late Miocene to Late Pleistocene (Table 1). Th e northernmost occurrence is in Humboldt County California (H oward 1970;Kohl 1974) and the southernmost occurrence is in Baja California, Mexico (Howard 1971).
Discovery of an articulated partial skeleton referable to Mancallinae (SDSNH 68312) from the Early Pliocene Capistrano Formation of Orange County California prompted a re-examination of diversity and morphological variation within this clade. Previously reported Mancallinae remains are reviewed (Appendix 1), and the results of an extensive survey of Mancallinae remains are reported. Th ree new species of Mancallinae are described, and the systematic placement of Mancallinae within Alcidae, as well as the inter-relationships of Mancallinae species is evaluated in a combined phylogenetic analyses of morphological and molecular sequence data. Th is study represents the fi rst time that relationships among all 23 extant alcids and 29 other charadriiform outgroup taxa have been assessed in the context of a combined phylogenetic analysis.

Anatomical terminology and taxonomic conventions
Description of anatomical features primarily follows the English equivalents of the Latin osteological nomenclature summarized by Baumel and Witmer (1993). Th e terminology of Howard (1929) is followed for features not treated by Baumel and Witmer (1993). Measurements follow those proposed by Von den Driesch (1976). All measurements were taken using digital calipers and rounded to the nearest tenth of a millimeter. Ages of geologic time intervals are based on the International Geologic Timescale (Gradstein et al. 2004;Ogg et al. 2008).
W ith the exception of species names (e.g., Fratercula arctica), which follow the 7th edition of the Checklist of North American Birds (American Ornithologists' Union 1998) for extant species, all taxonomic designations (e.g., Fratercula) are intended as clade names as defi ned by the International Code of Phylogenetic Nomenclature (i.e., Th e PhyloCode v.4c; Cantino and de Queiroz 2010), regardless of use of italics or previous rank recognized by other authors, and are not intended to convey rank under the Linnaean system of nomenclature. Th e PhyloCode recommendation that all scientifi c names be italicized (Recommendation 6.1A) was not followed here. Only species names are italicized herein. Pursuant to Article 21.2 of the PhyloCode, the fi rst word of species names are considered prenomen, not genus names (see also Dryat et al. 2008).

Taxon and character sampling
All extinct taxa were evaluated by direct observation of holotype and referred specimens. Whenever available, a total of fi ve or more specimens of each extant species (Appendix 2) including both sexes were evaluated to account for intraspecifi c character variation and sexual dimorphism respectively. Only adult specimens, assessed based upon degree of ossifi cation (Chapman 1965), were evaluated for osteological characters, and when available, specimens from multiple locations within the geographic range of extant species (i.e., subspecies) were examined to account for geographic variation within species. Reproductive, chick integument, dietary, and some myological characters were scored from published sources (Appendix 3). Descriptions of anatomical characteristics are followed by character numbers and character state symbols from Appendix 3 (e.g., 23:0 = character number 23, character state 0).

Smith 2011
Mancalla vegrandis 11), and Dove (2000;n = 34). Only 34 of the 38 characters used by Dove (2000) varied in the taxa examined in this study. Of the 34 used in this analysis, eighteen were modifi ed (i.e., split into 2 separate characters) according to the philosophy of character independence proposed by Hawkins et al. (1997), resulting in a total of 52 microfeather characters. Th e cladistic matrix also includes a molecular sequence alignment of 11,601 base pairs from eight DNA sequence types (including gaps). See Appendix 5 for details of sequence availability, inclusion for each species, and sequence authorship. Molecular sequence data (mitochondrial: ND2, ND5, ND6, CO1, CYTB; ribosomal RNA: 12S, 16S; and nuclear: RAG1) were downloaded from GenBank. Preliminary sequence alignments for each gene were obtained using the program ClustalX v2.0.6 (Th ompson et al. 1997), and then manually adjusted using the program Se-Al v2.0A11 (Rambaut 2002).

Phylogenetic analyses
A combined approach of phylogeny estimation was used to evaluate the systematic position of Mancallinae species. Simulations show that the combination of molecular and morphological data often provides a more accurate estimate of phylogeny with respect tics: quadrate apneumatic (38:1); reduced pneumatic foramen of anterior sternum (59:0); omal extremity of furcula angled sharply rather than gently curving as in other charadriiforms (75:1); coracoidal tuberosity of furcula, positioned anterior to coracoidal facet (77:1); dorsoventral compression of humeral shaft exceeds that of all other charadriiforms (141:1/2); bicipital tubercle of radius distally elongated rather than in the form of a rounded tubercle as in other charadriiforms (162:1). Apomorphies of the alcid crown clade, Alcidae, are provided in Table 4.

Mancallinae Brodkorb, 1967
MANCALLINAE (contents = Mancalla + Miomancalla gen. n.) is referable to Pan-Alcidae based upon dorsoventral compression of the humeral shaft (141:2). Th e humeral shafts of Pan-Alcidae are more dorsoventrally compressed than in all other Charadriiformes. Mancallinae is diff erentiated from all other alcids on the basis of the following unambiguously optimized humeral apomorphies: deltopectoral crest extends past the midway point of the humeral shaft rather than restricted to the proximal half of the humeral shaft (104:2); presence of a 'mancalline muscle scar' extending distally from the primary pneumotricipital fossa (discussed below; 120:1); capital groove communicates with transverse ligament sulcus resulting a notched rather than rounded appearance of ventral margin of the humeral head in anterior view (136:2); humeral head rotated anterodorsally rather than in-line with humeral shaft (139:1); humeral shaft arced rather than sigmoidal (140:1); presence of fossae in tricipital sulci (150:1); anterior surface of the ventral condyle rounded rather than fl attened (153:0). Additional proposed apomorphies of Mancallinae include distal elongation (184:1) and anterior fl attening of the fi rst metacarpal (185:1). Th ese characteristics are present in Mancalla cedrosensis Howard, 1971, Miomancalla howardi sp. n., and two additional associated specimens referable to Mancallinae (SDSNH 77966 and LACM 107028). Although these two characters are also diagnostic for Alcini Storer, 1960, the clade composed of Alca, Pinguinus, Alle Link, 1806, andUria Brisson, 1760, the degree of distal elongation and anterior fl attening in Mancalla exceeds that observed in Alcini.

Mancalla Lucas, 1901
Original diagnosis (sensu Lucas, 1901)-Referable to Alcidae based upon dorsoventral compression of the humeral shaft. Diff ers from other Alcidae in the following characteristics: humerus short, with arced rather than sigmoid lengthwise curvature; anterior rotation of the humeral head; ventral margin of m. brachialis scar a distinct ridge.
Amended diagnosis. Mancalla is diff erentiated from Miomancalla on the basis of the following humeral characteristics: supracoracoidial crest does not broaden proximally (113:2); distal margin of the primary pneumotricipital fossa convex rather than concave (126:0); ventral margin of the ventral tubercle narrow and ventrally expanded (i.e., convex) rather than wide and deeply grooved (134:0); capital groove constricted rather than wide (137:1). Additional proposed apomorphies which are present in Mancalla cedrosensis and two additional associated specimens (SDSNH 77966 and LACM 128870) referable to Mancalla but not to species include: ulna shorter than carpometacarpus (180:1); ulna and radius more dorsoventrally compressed than other alcids; extension of the dorsal ulnar condyle farther distally to the ventral ulnar condyle than in other alcids (182:0); pisiform process of carpometacarpus reduced or absent (188:1).
Locality and horizon. Late Pliocene or Early Pleistocene (Zanclean or Calabrian) Niguel Formation of Orange County, California. Latitude, longitude, and elevation data are on fi le at SDSNH (locality 3202). Details of the geologic setting are provided in Appendix 6.
Diff erential diagnosis. Scar extending into primary pneumotricipital fossa is raised in relief to the fl oor of the primary pneumotricipital fossa and the humeral shaft as in Mancalla cedrosensis, rather than an excavated pit as in Mancalla vegrandis sp. n. and Mancalla californiensis Lucas 1901 (121:1;Fig. 5); dorsal and ventral edges of scar extending into primary pneumotricipital fossa taper to a point as in Mancalla vegrandis, rather than remaining parallel as in Mancalla californiensis and Mancalla cedrosensis (123:1); humerus longer than Mancalla cedrosensis, Mancalla californiensis, and Mancalla vegrandis (Tables 2, 3).
Anatomical description. Both scapulae are preserved (Fig. 3G, H). As in all Alcidae, the scapular shaft is mediolaterally compressed throughout its entire length. Th e proximal end of the scapular shaft is more rounded in other Charadriiformes. As in Mancalla vegrandis, the acromion projects farther anteriorly than that of Mancalla cedrosensis and other alcids (e.g., Uria, Aethia). As in Mancalla cedrosensis, the coracoidal tubercle is less pronounced than in Mancalla vegrandis. As in Mancalla vegrandis and Mancalla cedrosensis, a scapulotricipital tubercle is present just distal to the glenoid process on the ventral margin of the scapular shaft. Th is feature is also present in other fl ightless wing-propelled divers such as Spheniscidae and Pinguinus, but is not known in any volant alcid. As in Mancalla vegrandis, the scapular shaft, including the caudal extremity, is slightly more robust than in other alcids (e.g., Alca, Aethia). Th e caudal extremity is less dorsoventrally expanded than in Mancalla vegrandis. Th e caudal extremity is not known for Mancalla cedrosensis.
Fragments of the sternum preserve the sternal rostrum, coracoidal sulci, and the carinal apex (Fig. 3A, B). Th ese features are not preserved in Miomancalla howardi and comparisons are therefore limited to extant alcids and specimens of Mancallinae that are not presently referable to species. Th e morphology of the sternal rostrum is consistent with that of all other Alcidae. Although no coracoid is preserved in the holotype specimen of Mancalla lucasi, the shape of the coracoidal sulci of the sternum is consistent with the ~150° angle of the sternal articulation of the coracoid in Mancalla cedrosensis and Mancalla vegrandis. Th e sternal articulation of the coracoid, and the coracoidal sulci of the sternum in other alcids curves more acutely (e.g., ~90° in Alca torda; Fig. 6).
Complete right and left humeri are preserved (Fig. 3C, D, E and F). Based upon humeral proportions, M. lucasi represents the largest known species of Mancalla (Table 2). As in other Mancalla species, the ventral margin of the ventral tubercle is convex, and the capital groove is relatively narrower than other Alcidae. Th e ventral tubercle does not project as far ventrally as in Mancalla californiensis (Fig. 5). Th e distal end of the deltopectoral crest transitions to the shaft more abruptly than in Mancalla vegrandis. As in other Mancalla, the humeral head is rotated anteriorly, and the supracoracoideus muscle scar does not broaden proximally. Mancallinae is characterized by a scar of unknown function that is positioned adjacent to the primary pneumotricipital fossa (hereafter referred to as the 'mancalline scar ';Fig. 5). Th e position of the 'mancalline scar' suggests an accessory insertion of m. humerotriceps (Howard 1949), which can be divided into as many as four separate heads in some birds (Baumel and Witmer 1993). Other potentially homologous muscle scars include m. coracobrachialis cranialis, which is well developed in penguins (Ksepka et al. 2008), or m. scapulocranialis caudalis (see Matsuoka and Hasegawa 2007). However, the exact function of this feature is unknown because it is not present in any other charadriiform. Th e shape, position, and development of this scar is variable in Mancallinae (Fig. 5). Th e 'mancalline scar' of Mancalla lucasi is raised in relief like that of Mancalla cedrosensis, rather than excavated as in Mancalla californiensis and Mancalla vegrandis (Fig. 5). As in Mancalla vegrandis, the scar extends from a point just proximal to the junction of the bicipital crest with the humeral shaft, tapers to a point, and extends into the primary pneumotricipital fossa (Fig. 5). Th e dorsal and ventral margins of the 'mancalline scar' remain approximately parallel in Mancalla californiensis and Mancalla cedrosensis (Fig. 5). As in all Mancallinae, the humeral shaft is arced rather than sigmoidal or straight. As in other Mancalla, the dorsal supracondylar process is separated from the dorsal epicondyle by a small notch. A tubercle or papilla is present on the posterior side of the distal end of the humerus adjacent to the dorsal condyle (Howard 1976). As with all Mancallinae, the anterior surface of the ventral condyle is rounded, rather than fl attened as in all other alcids. Rounded fossae are present at the proximal ends of  Th e left femur is preserved (Fig. 3I) and is smaller (~15%; Table 2) than in Miomancalla howardi sp. n. (Table 3), and larger (~19%) than in Mancalla cedrosensis (Howard, 1971). Extant alcids do not display statistically signifi cant degrees of sexual dimorphism in their size, plumage, or osteological morphology (Storer 1952; Diff erential diagnosis. Dorsal and ventral edges of the mancalline scar extending into primary pneumotricipital fossa taper to a point as in Mancalla lucasi, rather than remaining parallel as in Mancalla californiensis and Mancalla cedrosensis (123:1; Fig. 5); mancalline scar extending into primary pneumotricipital fossa is an excavated pit as in Mancalla californiensis rather than raised in relief to the fl oor of the primary pneumotricipital fossa and the humeral shaft as in Mancalla cedrosensis and Mancalla lucasi (121:0); humerus shorter than other known Mancalla (Tables 2 and 3).
Anatomical description. Two cervical vertebrae are preserved ( Fig. 7A and B). Comparisons with Miomancalla howardi are limited to generalities regarding shape in dorsal view, for which the morphology of Mancalla vegrandis is consistent with that of Miomancalla howardi. Only thoracic vertebrae are known for Mancalla cedrosensis. One of the vertebrae (Fig. 7A) is mediolaterally narrower than the other (Fig. 7B). Although the width of cervical vertebrae other than the axis and atlas do not vary considerably in extant Alcidae, the 3 rd and 4 th cervical vertebrae of some charadriiforms (e.g., Larosterna inca Lesson, 1827) are mediolaterally narrower than cervical vertebra posterior to the 4 th (i.e., C5, C6, C7). Th e dorsal surface of the broader vertebra ( Fig. 7B) is perforated by a small foramen (i.e., perforation of laminae arcocostales). In extant alcids, only the third and fourth cervical vertebrae are perforated. Typically in extant Alcidae, the third cervical vertebra is punctured by a small foramina, whereas the foramina in the fourth cervical vertebra is much larger, leaving only a thin strut of bone bordering it laterally. Th e morphology of the preserved vertebrae is suggestive of C3 and C4; however, defi nitive assignment cannot be made at this time.
One complete cervical rib and one complete costal rib ( Fig. 7D and E) are preserved along with several other rib fragments (not fi gured). No morphological differences were evident between the ribs of Mancalla vegrandis, Mancallinae specimen SDSNH 25236, and other alcids for which the ribs are known.   All but the omal extremities of the furcula are preserved (Fig. 8D). Th e furcular rami are mediolaterally compressed as in all other Alcidae. Th e anterior surface of the furcular rami dorsal to the apophysis is rounded or convex as in Uria, rather than grooved as in Cepphus. Th e furcular apophysis does not bear the ventrally expanded, bladelike interclavicular process characteristic of extant Alcidae. However, the possibility that this feature was lost to damage cannot be ruled out. No additional morphological diff erences were evident between the preserved portions of the furcula of Mancalla vegrandis and other alcids for which the furcula is known.
Th e left coracoid is complete except for a small portion of the medial margin of the sternal facet (Fig. 8E). A fragment of the right coracoid preserves the medial margin of the sternal facet and the sternal portion of the coracoidal shaft (not fi gured). As in Mancalla cedrosensis the furcular facet is rounded, rather than oval as in Aethia and Fratercula. Th e head of the coracoid is apneumatic as in all Alcidae, but the brachial tuberosity is deeply undercut as in Alca and Pinguinus. Th e humeral articulation is more rounded than in extant Alcidae. As in Cepphus, the scar marking the position of m. supracoracoideus is less distinct than in other Alcidae. As in Mancalla cedrosensis, Aethia, and Alle, the procoracoidal process is not punctured by a foramen for passage of the tendon of m. supracoracoideus. Th e procoracoid process points dorsomedially as in all Alcidae except Aethia, in which the procoracoid points more ventromedially. As in Mancalla cedrosensis, Brachyramphus, Uria, Aethia, and Ptychoramphus Brandt, 1837, the sternal margin of the procoracoid process is concave, rather than convex as in Cerorhinca, Fratercula, and Pinguinus. As in many alcids (e.g., Alca, Brachyramphus) a single, distinct, straight ridge, which extends from the lateral angle of the sternal facet towards the humeral facet is present. Th is ridge does not extend sternally in Synthliboramphus, Cepphus, Fratercula, Aethia, Ptychoramphus, and Cerorhinca. Th is ridge is less pronounced and positioned farther laterally in Mancalla cedrosensis. A well-developed lateral process is present. Th is feature is absent in Mancalla cedrosensis. Th e dorsal mar- gin of the medial sternal process is notched as in most alcids (e.g., Alca torda). As in Mancalla cedrosensis, the posterior surface of the sternal end of the coracoid is more excavated than in extant Alcidae and the sternal facet is curved ~150°.
Right and left scapulae are preserved ( Fig. 8A and B). As in all Alcidae, the scapular shaft is mediolaterally compressed throughout its entire length. As in Mancalla lucasi, the acromion projects farther anteriorly than that of other alcids (e.g., Uria, Aethia). Th e acromion of Mancalla cedrosensis does not project as far anteriorly as that of Mancalla vegrandis. Th e coracoidal tubercle is more pronounced than in Mancalla lucasi and Mancalla cedrosensis. As in Mancalla lucasi and Mancalla cedrosensis, a scapulotricipital tubercle is present just distal to the glenoid process on the ventral margin of the scapular shaft. As in Mancalla lucasi, the scapular shaft, including the caudal extremity, is slightly more robust than in other alcids (e.g., Alca, Aethia). Th e caudal extremity is more dorsoventrally expanded than in Mancalla lucasi. Th e caudal extremity is not known for Mancalla cedrosensis.
Parts of the left distal end of the sternum including the distal end of the carina, and the left lateral process are preserved (Fig. 8C). Mancalla lucasi and Miomancalla howardi do not preserve the same portions of the sternum so comparisons cannot presently be made between the sterni of Mancallinae. As a result of the deep incisure of the lateral notches the lateral processes of Mancalla vegrandis are more elongate that any other alcids for which the sternum is known. In other Charadriiformes this condition is present only in the Glareolidae and Scolpacidae, and resembles the sternum in Spheniscidae (Fig. 9).
Th e left humerus is preserved (Fig. 7C). Based upon humeral proportions, Mancalla vegrandis represents the smallest known species of Mancalla (Table 2). As in other species of Mancalla, the ventral margin of the ventral tubercle is convex, and the capital groove is relatively narrower than other Alcidae. Th e ventral tubercle does not project as far ventrally as in Mancalla californiensis. Th e distal end of the deltopectoral crest transitions to the shaft less abruptly than in Mancalla lucasi. As in other Mancallinae, the humeral head is rotated anteriorly and the supracoracoideus muscle scar does not broaden proximally. Th e 'mancalline scar' is excavated as in Mancalla californiensis, rather than raised in relief like that of Mancalla cedrosensis and Mancalla lucasi (Fig. 5). As in Mancalla lucasi, the 'mancalline scar' extends from a point just proximal to the junction of the bicipital crest with the humeral shaft and tapers to a point, and extends into the primary pneumotricipital fossa. Th e margins of this scar remain parallel in Mancalla californiensis and Mancalla cedrosensis. As in all Mancallinae, the humeral shaft is arced rather than sigmoidal or straight. As in other Mancalla, the dorsal supracondylar tubercle is separated from the dorsal epicondyle by a small notch. A tubercle or papilla is present on the posterior side of the distal end of the humerus adjacent to the dorsal condyle (Howard, 1966). As with all Mancallinae, the anterior surface of the ventral condyle is rounded, rather than fl attened as in all other Alcidae. Rounded fossae are present at the proximal ends of the humerotricipital and scapulotricipital grooves. Th e fl exor process extends distal to the ventral condyle as in all Mancallinae and Pinguinus.
Th e pelvis is preserved in dorsal view (Fig. 7F). Comparisons of pelves within Mancallinae are limited to Miomancalla howardi. As in all alcids the anteroposterior length of the pelvis is greater than two times the mediolateral width across the antitrochanters.
Th e relative length of the pelves of other charadriiforms is anteroposteriorly shorter. Th e proximal end of the preacetabular ilium is wide as in Miomancalla howardi and most alcids (e.g., Brachyramphus). Th e distal end of the preacetabular ilium is relatively broader than in Miomancalla howardi. As in Miomancalla howardi the antitrochanteral sulcus does not extend proximally to contact the antitrochanter. As in most Alcidae (e.g., Brachyramphus), the post-acetabular dorsal ilium narrows, rather than broadens as in Uria, Cepphus, and some Fraterculinae. Th e iliosynsacral suture is perforated as in Uria, Alca, Pinguinus, and Synthliboramphus, rather than fused along its entire length as in Cepphus, Brachyramphus, and Fraterculinae. Th e dorsal iliac spine has a pointed tip as in all alcids other than Aethia and Ptychoramphus, in which the end of the spine is blunt.
Remarks. Mancalla vegrandis corresponds in size and humeral characteristics with some material previously referred to Mancalla milleri Howard, 1970. However, Mancalla milleri is considered Alcidae incertae sedis (see Appendix 1 for details of the taxonomic revision).
Etymology. Mio to refl ect Miocene occurrences of known species within the taxon, and mancalla to refl ect the sister group relationship with Mancalla Lucas, 1901.
Etymology. Th is new species is named in honor of Hildegarde Howard in recognition of her many contributions to the systematics of extinct Alcidae.  Referred specimen. SDSNH 24584, a left humerus (Fig. 12) from the Late Miocene lower member (Messinian) of the San Mateo Formation of San Diego County, California (SDSNH locality 3177). Th is specimen was noted but not named or described by Chandler (1985) and Livezey (1988).
Anatomical description. Th e holotype specimen is preserved in a matrix of dark grey, highly indurated, siltstone (Fig. 10). Some elements are slightly crushed and many cortical bone surfaces are considerably abraded, obscuring fi ne morphological details in many portions of the specimen.
Elements of the skull are exposed in oblique right lateral view (Figs 10,11). Th e premaxilla, maxilla, nasal, lacrimal, jugal, frontal, and squamosal are present. Additional fragments of bone adjacent to the posterior frontal may represent a portion of the parietal. An unidentifi ed fragment of bone protrudes from the external narial opening. Th e premaxilla is relatively shorter and mediolaterally compressed in comparison with the only other known premaxillae referable to Mancallinae (LACM 103940; SDSNH 25236; Fig. 13), which resemble the more terete bills of some other Alcidae (e.g., Uria). Th e maxilla, which broadens anteriorly before fusion with the premaxilla, is complete but broken at approximately its midpoint. As in many alcids (e.g., Cepphus, Alca) the nasal contacts the maxilla at ~45° angle. Th is angle is ~60° in the puffi ns and auklets (i.e., Fratercula, Cerorhinca, Aethia, and Ptychoramphus). As in Pinguinus, and in contrast to other alcids, the lacrimal appears to be directed ventrally rather than posteroventrally. However, crushing of the skull may have changed the relative orientation of elements and it is possible that distortion is responsible for this condition. Th e jugal is preserved in contact with the mandible. Fusion between the jugal and the jugal process of the premaxilla is visible. Th e frontal is distorted by crushing and most morphological details obscured in this element. Th e outline of the right orbit is visible, but is deformed by ventrolateral displacement of the lateral margin of the frontal. Th e frontal bears a robust orbital rim as in Uria, Miocepphus, Alle, Alca, and Pinguinus.
Th e mandible is preserved in right lateral view (Figs 10,11). Th e mandibular symphysis is elongate as in Uria and Fratercula. Th e mandibular rami are fused along a relatively shorter distance in some alcids (e.g., Alle). Th e proximal and distal ends of the mandible are dorsoventrally expanded, similar to the condition in Alca and Pinguinus. A pair of small posterior mandibular fenestrae is present as in other known Mancallinae mandibles (LACM 103940;SDSNH 25236;Fig. 13), Fraterculini Storer, 1960, andsome charadriiforms (e.g., Stercorarius longicaudus Vieillot, 1819).
At least two cervical vertebrae are partially exposed on the surface of the slab (Fig. 10). Fine morphological details are obscured by matrix and the poor preservation of the vertebrae. One vertebra resembles the axis, but positive identifi cation is hindered by matrix and damage to the element. Th e other is a cervical vertebra exposed in dorsal view. Mancallinae vertebrae are known only from the holotype specimens of Mancalla cedrosensis and Mancalla vegrandis. Comparisons with Mancalla cedrosensis are not possible because only a single thoracic vertebra is preserved in the holotype specimen. Th e shape of the dorsal surface of the cervical vertebrae of Miomancalla howardi is consist- ent with that of Mancalla vegrandis. Further preparation of the holotype specimen of Miomancalla howardi, or discovery of additional material referable to this species is necessary before more details of vertebral anatomy can be described for this species.
Fragments of the sternum are preserved adjacent to the humerus in what appears to be ventral view (Fig. 10). Th e craniolateral process appears to point dorsally, rather than anteriorly as in Mancalla lucasi, although the possibility that crushing of this element altered the relative orientation of that feature cannot be ruled out. Other morphological details are obscured by matrix and the poor preservation of the sternum.
Th e holotype specimen preserves the proximal end of the right humerus in posterior view (Fig. 10). In addition to the head of the humerus, which is slightly crushed, the outline of the proximal half of the humeral shaft is visible as an impression in matrix. A complete left humerus (SDSNH 24584; Fig. 12) is referable to Miomancalla howardi based upon its similar proportions (i.e., larger than any other known Mancallinae; Table 2), and the fact that the ventral surface of ventral tubercle is more deeply grooved than in any other alcid. Th e ventral surface of the ventral tubercle is also grooved in Pinguinus and Miomancalla wetmorei, but the degree of excavation of this groove is more pronounced in Miomancalla howardi. Th e ventral margin of the ventral tubercle of Mancalla is convex. Th e capital groove is relatively wider than that of other species of Mancallinae, and it is incised more deeply into the transverse ligament sulcus in anterior view than in Miomancalla wetmorei. Th e proximal end of the deltopectoral crest is less pronounced than in Miomancalla wetmorei. Th e distal end of the deltopectoral crest transitions to the shaft less abruptly than in Mancalla. Th e humeral head is rotated more anteriorly than in Miomancalla wetmorei, and is more similar to the condition in Mancalla. As in Miomancalla wetmorei and Fratercula, and in contrast to the condition in Mancalla species, the supracoracoideus muscle scar broadens proximally. In Miomancalla howardi and Miomancalla wetmorei the 'mancalline scar' extends from a point just proximal to the junction of the bicipital crest with the humeral shaft and tapers to a point that meets the dorsal border of the primary pneumotricipital fossa (i.e., crus dorsale fossae of Baumel and Witmer 1993:99). Th e scar is relatively smaller in Miomancalla and Mancalla lucasi than in comparison with other Mancallinae. Th e scar is an excavation in all Mancallinae except Mancalla cedrosensis and Mancalla lucasi, in which the scar is raised in relief to the fl oor of the primary pneumotricipital fossa and the humeral shaft. Th e shaft of the humerus is arced more so than in Miomancalla wetmorei or any other known alcid, and is less dorsoventrally compressed than in Pinguinus. As in all alcids other than Mancalla, the dorsal supracondylar process is continuous with the dorsal epicondyle, rather than separated from it by a small notch. Th e dorsal supracondylar process is less pronounced than in Miomancalla wetmorei. A tubercle or papilla on the posterior side of the distal end of the humerus adjacent to the dorsal condyle was described by Howard (1966), who used that characteristic to diff erentiate between species of Mancalla that possessed the tubercle, and species of Miomancalla (Praemancalla sensu Howard, 1966) that did not posses it. Th e tubercle is present in Miomancalla howardi. As with all Mancallinae, the anterior surface of the ventral condyle is rounded, rather than fl attened as in all other Alcidae. Rounded fossae are present at the proximal ends of the humerotricipital and scapulotricipital grooves. Th at character cannot be evaluated in Miomancalla wetmorei or Mancalla californiensis owing to damage to the holotype specimens of those species and current lack of referable specimens. Th e fl exor process extends distal to the ventral condyle as in all Mancallinae and Pinguinus.
Th e left carpometacarpus is preserved in dorsal view (Fig. 10). Although hundreds of Mancallinae carpometacarpi are known from Pliocene marine deposits in California, the holotype specimens of Miomancalla howardi and Mancalla cedrosensis are the only associated specimens that allow for species-level referral of carpometacarpi. Th e carpometacarpus of Miomancalla howardi is larger than that of Mancalla cedrosensis (~23%; Table 3; Howard 1971), and displays the distal elongation of metacarpal I that is characteristic of Mancallinae. Th e abraded preservation of this element limits further comparisons.
Th e pelvis is exposed in dorsal view (Fig. 10). Comparisons within Mancallinae are limited to Mancalla vegrandis. As in all alcids the anteroposterior length of the pelvis is greater than two times the mediolateral width across the antitrochanters. Th e relative length of the pelves of other charadriiforms is anteroposteriorly shorter. Th e proximal end of the preacetabular ilium is wide as in Mancalla vegrandis and most alcids (e.g., Brachyramphus). Th e distal end of the preacetabular ilium narrows more so than in Mancalla vegrandis. As with Mancalla vegrandis the antitrochanteral sulcus does not extend proximally to contact the antitrochanter. Th e dorsal iliac spine has a pointed tip as in all alcids other than Aethia and Ptychoramphus, in which the end of the spine is blunt.
Th e distal ends of both tibiotarsi are missing or embedded in matrix (Fig. 10). Th e poor preservation of these elements limits comparisons with the smaller holotype tibiotarsi of Mancalla cedrosensis to size (~26% larger; Table 3; Howard 1971).
Th e right femur is exposed in posterolateral view along the edge of the block but is severely abraded: however, the left femur is well-preserved and exposed in anterior view (Fig. 10). Th e femur is robust and less sigmoidal in shape in comparison with the femora of extant alcids such as Alle or Uria, resembling the condition in Mancalla lucasi and Mancalla cedrosensis, the only other Mancallinae from which the femur is known. Th e intercondylar sulcus is relatively broader and more well-defi ned proximally than that of Mancalla lucasi and Mancalla cedrosensis. As in Cepphus, Brachyramphus, and Synthliboramphus, the distally extending and anteriorly projected crest of the femoral trochanter is convex in shape. Th is feature is fl attened (e.g., Alca and Uria) or concave (e.g., Fratercula and Cerorhinca) in other alcids. Th e femoral head appears relatively smaller in comparison with this element in Mancalla cedrosensis and Mancalla lucasi. Th e length of the femur is greater than in Mancalla cedrosensis and Mancalla lucasi (Table 3; Howard 1971).
Th e left tarsometatarsus is preserved in anterior view (Fig. 10). Th e anterior surface of the shaft is deeply grooved as in Mancalla cedrosensis and Fratercula. Associated specimens with tarsometatarsi that would allow for referral of isolated tarsometatarsi to species are not currently known from other Mancallinae. Th e outlines of trochlea are visible but the distal end of the element is too badly abraded to discern fi ne morphological details.

Phylogenetic results
Owing to the incomplete and fragmentary preservation of most Mancallinae specimens referable to species, preliminary analysis of the systematic relationships of Mancalla resulted in an unresolved polytomy among Alcidae sub-clades (i.e., relationships between Mancallinae, Cepphus, Brachyramphus, Synthliboramphus, Alcini, andFraterculinae (contents = Fraterculini Storer, 1960 + Aethiini Storer, 1960) unresolved at the base of a monophyletic alcid clade (results not shown). Two additional phylogenetic analyses were performed to investigate the position of Mancallinae within Charadriiformes, and the interrelationships of Mancallinae species. Th e primary phylogenetic analysis included a Mancallinae supraspecifi c terminal (SST) constructed by combining scorings from 19 Mancallinae specimens (including all holotype material; Appendix 4). Th e referral of all Mancallinae specimens used to construct the SST was evaluated based upon the unambiguously optimized apomorphies listed in the diagnosis section for Mancallinae above. Note that due to damage or missing elements in Mancallinae holotype specimens, fi ve of the specimens used to construct the Mancallinae supraspecifi c terminal preserve morphological data not preserved by the holotype specimens, thus providing a more compete picture of morphological variation in Mancallinae than if only the holotype specimens were analyzed. Th e results of the fi rst analysis were used to constrain the topology of trees accepted during a secondary tree search in which the species-level relationships of Mancallinae were evaluated.
Th e primary combined phylogenetic analysis of the cladistic matrix including a Mancallinae SST resulted in two most parsimonious trees (MPT's) of 15,974 steps (Fig. 14;CI: 0.38;RI: O.50;RCI: 0.19). Additional analyses performed with all characters unordered did not result in topological diff erences, or an increase in the number of MPT's recovered. Pan-Alcidae is recovered as the sister to Stercorariidae, a result that is congruent with the results of previous molecular based analyses (Ericson et al. 2003;Paton et al. 2003;Th omas et al. 2004;Paton and Baker 2006;Baker et al. 2007), but confl icts with previous morphology-based analyses (Strauch 1978;Björklund 1994;Chu 1995Chu , 1998Chu et al. 2009;Livezey 2009Livezey , 2010Mayr 2011). Alcidae and Stercorariidae have not been recovered as sister taxa in any previous morphology based analysis, suggesting that molecular sequence data is solely responsible for this hypothesis. Th ere is, however, morphological character support for this clade (Table 4). Th e combined analysis estimate of relationships among the Alcidae crown clade is congruent with the results of recent analyses of molecular sequence data (Th omas et al. 2004;Paton et al. 2003;Baker et al. 2007;Pereira and Baker 2008), except that Synthliboramphus is placed at the base of Alcinae, rather than as the sister to Alcini (Fig. 14). However, the parsimony analysis by Pereira and Baker (2008) also recovered Synthliboramphus at the base of Alcinae in one of two most-parsimonious topologies. Th e positions of other species (e.g., Alca + Pinguinus), and sub-clades in Alcidae (e.g., Fraterculinae + Alcinae) are consistent with the results of recent molecular-based analyses (Baker et al. 2007;Pereira and Baker 2008) with dense taxonomic sampling for Alcidae. Th e only prior morphology-based analyses with suffi cient taxonomic sampling for comparison to these results, those by Strauch (1985) and Chandler (1990a), resulted in topologies that strongly confl ict with more recent hypotheses of alcid relationships in that they do not support a traditional Fraterculinae (i.e., monophyly of Fraterculini + Aethiini).
Th e Aethiini (i.e., Ptychoramphus + Aethia) are placed basal to the Alcinae (Alca + Pinguinus + Cepphus + Brachyramphus + Synthliboramphus), rather than as sister to the Fraterculini (i.e., Cerorhinca + Fratercula) in the topology of Strauch (1985). Although the work by Chandler (1990a) represented an increase in the number of characters scored for Alcidae, the results of that analysis placed Alle alle and Cepphus in a clade with the Fraterculini, rather than in Alcinae. Th e combined analysis, as well as previous analyses (Watada 1987;Moum et al. 1994;Friesen et al. 1996;Th omas et al. 2004;Baker et al. 2007;Pereira and Baker 2008) strongly support monophyly of a Fraterculinae clade consisting of Ptychoramphus, Aethia, Cerorhinca, and Fratercula, and the sister-group relationship between Fraterculinae and Alcinae as defi ned here (Fig. 14).
Only the systematic position of Alle alle Link, 1806 remains unresolved within Alcini (Fig. 14). Th e systematic position of Alle alle is potentially the most contentious issue within alcid systematics, as it has been recovered as the sister to Alca + Pinguinus (Moum et al. 1994(Moum et al. , 2002Baker et al. 2007), sister to Alca + Pinguinus + Uria (Strauch 1985), sister to Uria (Th omas et al. 2004;Pereira and Baker 2008), sister to Fraterculinae (Chandler 1990a), and sister to Cepphus + Aethia + Brachyramphus (Chu 1998). Resolution of this issue will likely require a comprehensive analysis of alcid relationships including dense taxonomic sampling of extinct Alcidae.
Mancallinae is placed as the sister taxon to all other Alcidae (i.e., placed outside of crown clade Alcidae; Fig. 14). Th is result is consistent with the only previous analysis that included Mancallinae (Chandler 1990a). Th e clade composed of crown Alcidae + Mancallinae is therefore designated Pan-Alcidae. Th e monophyly of Pan-Alcidae is supported by fi ve unambiguously optimized morphological characters with a CI = 1.0 (UOMC; Table 4).
Th e combined analysis recovered relationships among the 29 charadriiform outgroup taxa that are largely congruent with prior molecular-based analyses of the clade, Table 4. Unambiguously optimized morphological characters with a CI of 1.0 supporting alcid clades in the resultant phylogenetic tree (Fig. 15). Character numbers from Appendix 3 are followed by character state symbols (e.g., 23:0 = character number 23, state 0). '*' indicates selected locally optimized apomorphies with a CI of < 1.0.

Clade
Character numbers and states that support monophyly Pan-Alcidae + Stercorariidae *63:0; *124:1; but do not support previous morphology-based results. Larus and Hydrophasianus (i.e., gulls and jacanas) are recovered as more closely related to one another than either are to Charadriius (i.e., plovers), as in the results obtained by Hackett et al. (2008). Also consistent with the results of prior molecular analyses (Ericson et al. 2003;Paton et al. 2003;Paton and Baker 2006;Baker et al. 2007), Alcidae + Stercorariidae is placed as the sister to Laridae + Sternidae + Rynchopidae. In contrast to the combined analysis results presented herein and recent molecular based results, the results of the phylogenetic analyses of morphological data by Zusi (2006, 2007) and Livezey (2009Livezey ( , 2010 place Alcidae as the sister taxon to Stercorariidae + Rynchopidae + Laridae. However, taxon sampling for Alcidae was limited to Uria in the analysis of Zusi (2006, 2007), and Alcidae was included as a single, taxon level terminal in the analyses of Livezey (2009Livezey ( , 2010 and Mayr (2011). Th e morphology based phylogeny of Mayr (2011) placed Alcidae in a polytomy with Dromadidae, Stercorariidae, and a clade comprising Laridae + Sternidae + Rynchopidae. Th e results of the combined analysis are congruent with recent molecular-based analyses, which place Lari (e.g., alcids, gulls, and pratincoles) as the sister to Scolpaci (e.g., sandpipers and curlews), and place Charadri (e.g., plovers), at the base of Charadriiformes. Th is hypothesis contrasts with morphology-based results (Björklund 1994;Chu 1995), which were the result of parsimony-based re-analyses of the compatibility analysis of Strauch (1978). In the topology recovered by Björklund (1994) the Charadri and Scolpaci are placed in an unresolved polytomy basal to the Lari, whereas the Lari and Charadri are placed in an unresolved polytomy basal to the Scolpaci in the topology recovered by Chu (1995). Th e morphology based analyses of Livezey (2009Livezey ( , 2010 and Mayr (2011) recover Scolpaci as an outgroup to a Charadri + Lari clade. Th e contents of Charadri, Scolpaci, and Lari estimated by the combined analysis are consistent with the composition of those clades recovered in prior molecular-based phylogenetic analyses (Sibley and Ahlquist 1990;Paton et al. 2003;Ericson et al. 2003;Paton and Baker 2006;Baker et al. 2007), supporting the monophyly of Charadri, Lari, and Scolpaci. An additional combined analysis was performed in which the tree was a priori rooted with the Scolpaci clade (i.e., Hydrophasianus, Tryngites, Numenius, and Bartramia) to mimic the phylogenetic results of Livezey (2009Livezey ( , 2010 and Mayr (2011). Th is alternative rooting scheme did not aff ect relationships recovered among Alcidae or Lari species and clades, between Alcidae and Stercorariidae, or between Mancallinae and other alcids (results not shown). Also of interest is the placement of Rynchops (i.e., skimmers). Recent molecular analyses recovered Rynchops as the sister to Laridae (Paton et al. 2003;Baker et al. 2007) or sister to Sternidae (Paton and Baker 2006). Th e morphology-based analyses by Chu (1995Chu ( , 1998 placed Rynchops as the sister to Sternidae + Laridae + Stercorariidae. Th e results of the combined analysis place the Black Skimmer Rynchops niger Linnaeus, 1758 as the sister taxon to the White Tern Gygis alba Sparrman, 1786. Considering the accepted placement of Gygis alba in Sternidae (American Ornithologists ' Union 1998, Brigde et al., 2005, this result would suggest Sternidae paraphyly. Although, this result is not entirely novel because an alternative hypothesis also places Gygis outside Sternidae, as the sister to Laridae + Sternidae (Baker et al. 2007). However, denser taxonomic sampling of Rynchopidae, Sterndidae, and other Charadriiformes may resolve this issue in the future.
Anous (i.e., noddies) was recovered as the sister to Sternidae + Laridae + Rynchopidae in the combined analysis, a placement consistent with the molecular-based results reported by Baker et al. (2007), and in confl ict with the morphology-based results obtained by Chu (1998), which place Anous as the sister to Stercorariidae. Th e only study with dense taxonomic sampling of terns and noddies (Bridge et al. 2005) included a single larid (Larus delawarensis Ord, 1815) as an outgroup taxon, but placed Anous basally in Sternidae. Resolution of the systematic affi nities of Anous will likely require denser taxonomic and character sampling across Laridae, Sternidae, Rynchopidae, Anous, and other non-Lari charadriiforms.

Discussion
Th e taxonomic revision and description of new Mancallinae species herein confi rms previous estimates of high diversity in Mancallinae (Howard 1970;Olson 1981;Chandler 1990a), and in combination with the phylogenetic results of the combined analysis, provide a new context for the interpretation of the evolutionary success of this lineage of fl ightless wing-propelled divers. Similar to the hypothesized independent evolution of fl ightlessness in penguins and plotopterids (Smith 2010), the placement of Mancallinae as the sister taxon to crown Alcidae suggests that fl ightlessness evolved independently in the Mancallinae and Pinguinus lineages, making the many osteological characteristics shared between these taxa an even more compelling example of morphological convergence. Phylogenetic support for the monophyly of Miomancalla and Mancalla also provides further contextualization for the interpretation of morphological diff erences between these sister taxa. Although known diversity is higher for Mancalla, there is an apparent trend towards decrease in size for more derived members of the clade, with the larger Miomancalla and Mancalla lucasi placed basally in the resultant topology (Fig. 15). Although it is tempting to infer large body-mass as the ancestral state for Pan-Alcidae, the reconstruction of this character is ambiguous according to the phylogenetic results, and there is an ~25Ma gap in the fossil record between the oldest known fossil alcid and the oldest Mancallinae fossils. Th e most important contributing factor regarding the ambiguity of ancestral states within Pan-Alcidae is the incompleteness of the early alcid fossil record. Although an abundance of taxa are known from the Miocene and Pliocene, only a single fragmentary alcid fossil is known form the Eocene (Chandler and Parmley 2002). Th e only Oligocene fossils that are currently referred to Alcidae are two fragmentary and isolated specimens from the Iwaki Formation in Japan (Ono and Hasegawa 1991). Eocene and Oligocene localities and collections should be targeted to increase knowledge of early diversity and ancestral states within Pan-Alcidae.
Although impressive with regard to the quantity of taxa sampled (n = 242) and the number of morphological characters scored for those taxa (n = 1107), comparisons with the results of a recent morphology based analysis of Charadriiformes (Livezey 2009(Livezey , 2010 and the results of this study are limited to relationships among outgroup charadriiforms because Alcidae was included only as a suprageneric taxon. With respect to relationships among major charadriiform clades, some of the results of Livezey (2010) are admittedly in confl ict with a growing consensus of molecular results based upon a variety of methods (e.g., parsimony, Bayesian) and sampling schemes (mitochondrial and nuclear DNA sequences). For example, although the placement of Charadri in a derived position within Charadriiformes to the exclusion of other clades (Livezey 2010) is in agreement with some previous hypotheses (Strauch 1978;Sibley and Ahlquist 1990;Christian et al. 1992;Björklund 1994;Chu 1995;Th omas et al. 2004;Livezey and Zusi 2007), these hypotheses are in contrast with the results of more recent multigene molecular based hypotheses that recovered Charadri in a more basal position. (Ericson et al. 2003;Paton et al. 2003;Paton and Baker 2006;Baker et al. 2007;Fain and Houde 2007;Hackett et al. 2008). Th ere exists no metric with which to choose between the contrasting results of those many analyses, and thus systematic relationships between major clades of Charadiirformes remain somewhat uncertain. However, the combined analysis results reported herein represent the most inclusive analysis to date with respect to variety of phylogenetically informative data sampled.

Referral of fossils to species level
Referral of specimens to named species, or recognition of new species, based solely upon size, or provenience, or age, or any combination of those three criteria, run the risk of incorrectly assigning specimens to species, or incorrectly assessing species diversity (Norell 1989;Stewart 2002b;Nesbitt and Stocker 2008;Bell et al. 2010). To avoid the possibility of recognizing two or more fossil species based upon diff erent skeletal elements of the same species, recognition of new species must be predicated upon diagnoses or diff erentiation from previously named species within a taxon (Appendix 1). Occurrence within the same deposit or deposits of similar age is not considered strong evidence that fossils represent the same taxon. Similarly, a lack of recorded occurrences of a fossil taxon within a deposit or deposits of a particular age does not preclude the possibility that a taxon may have been extant during the time of deposition. For example, if the holotype specimen of a species is an isolated humerus, then only associated specimens with humeri consistent with that of the holotype specimen allow for initial referral of additional skeletal elements. When previously recognized holotype specimens consist of isolated elements, isolated material consisting of elements other than the holotype element cannot be referred to the species level until associated specimens are discovered that facilitate such referral. Although these criteria do not preclude the possibility that cryptic species may lead to underestimation of species diversity (see Stewart 2002bStewart , 2007, these criteria do avoid overestimation of diversity and incorrect assignment of specimens that can result from less rigorous methods (i.e., size, provenience, or age based methods) of specimen referral and species recognition. In the case of Mancallinae remains, there is little doubt that hundreds of isolated fossils are referable to that clade; however, to avoid future taxonomic confusion, referrals should only be made based upon the criteria outlined above. Th e morphological diff erences between Mancallinae holotype and referred specimens described and phylogenetically optimized herein provide a basis for the potential apomorphy-based referral of hundreds of additional isolated Mancallinae remains, which will facilitate future detailed study of interspecies morphologic and size variation in Mancallinae.

Flightlessness and convergence
Th e etymology of Mancalla (mancus-from the Latin for crippled or lame, and ala from the Latin for wing ;Brown 1956) refl ects an antiquated view of fl ightlessness. Th e fl ightless condition observed in ostriches and some rails for example, in which the pectoral elements are diminished in size, has been attributed to lack of predatory pressures and energy conservation strategies (Livezey and Humphrey 1986;McNab 1994). Th e fl ightless condition observed in penguins, plotopterids and some auks (i.e., Mancallinae and Pinguinus) refl ects specialization for wing-propelled diving in the form of a functional 'trade-off ' between aerial and sub-aqueous fl ight (Storer 1960;Olson and Hasegawa 1979;Bengston 1984;Livezey 1988;Habib 2010). Th is extreme specialization for wing-propelled diving results in characteristics that are shared not only among fl ightless alcids, but also with penguins and plotopterids. It was the outward resemblance of Spheniscidae to the familiar Great Auk Pinguinus impennis of the northern Atlantic Ocean that prompted sailors who fi rst encountered Spheniscidae in the southern hemisphere to call them penguins (Olson and Lund 2007). Osteological characteristics shared between fl ightless alcids and penguins include decrease in range of motion and shortening of the distal wing elements in comparison with volant alcids (Raikow et al. 1988; Fig. 16), distal elongation of metacarpal one (Fig. 17), arced or curved wing elements ( Fig. 1), an increase in the size of the tricipital crests of the distal humerus ( Fig. 1), and a deeply grooved ventral margin of the ventral tubercle ( Fig. 1). Mancallinae share additional convergent characteristics with Spheniscidae such as dorsoventral expansion of the omal extremity of the furcula, and deeply incised lateral sternal notches (Fig. 9). Although the functional signifi cance of these modifi cations is not precisely known, the demands of wing-propelled pursuit diving for fi sh involving powered up-strokes and down-strokes likely played a role in the evolution of the convergent morphological characteristics shared by fl ightless alcids and penguins.
One characteristic that is unique to Mancallinae among all known fl ightless birds, is the shorter length of the ulna compared with that of the carpometacarpus (180:1). In most birds these proportions are opposite of that observed in Mancallinae, with the ulna being longer than the carpometacarpus. Th ree associated Mancallinae specimens (LACM 107028; SDSNH 77966), including the holotype specimen of Mancalla cedrosensis (LACM 15373) display this characteristic. Statistical analysis of osteological proportions of fl ightless alcids quantifi ed the dorsoventral compression of wing elements and shortening of distal wing elements, but surprisingly, Livezey (1988) did not mention the unique relationship between the lengths of the ulna and carpometacarpus. A survey of the proportions of distal wing elements among extinct and extant birds was conducted to assess the distribution of this character state. Th e only other birds that are known to share this characteristic are several species of hummingbirds (e.g., Phaethornis pretrei; see Mayr 2004, Table 1). Th e precise functional signifi cance of having a longer carpometacarpus than ulna would require detailed functional morphological study, but given the extreme pectoral specialization of both Mancallinae and Trochilidae, and the need of both of these taxa to produce thrust on both up-strokes and down-strokes, it seems reasonable to postulate that the increased dependence on thrust generated from primary feathers attached to the carpometacarpus (Chai 1997) may play a role in this osteological modifi cation. Although the primaries of Mancallinae would likey have been much shorter than those of Trochilidae, water is a considerably more dense fl ight medium with diff erent functional requirements than those for aerial fl ight (Habib 2010). Interestingly, this characteristic is not known in any extinct or extant penguin (J. Clarke, pers. com.).
Th e relatively large size of Pinguinus and some Mancallinae as compared to other alcids (Livezey 1988) may be linked with fl ightlessness, because the decreased buoyancy of large body size confers an advantage to piscivorous predators (Sparks and Soper 1987). Additionally, because these diving birds likely spent the majority of their time in the water (i.e., fl ightless, and came ashore only to breed), the thermal constraints imposed on them are decreased by large body size (Furness and Burger 1988). Furthermore, because Mancallinae were fl ightless, weight constraints related to maintaining the ability for aerial fl ight no longer restricted increases in body mass (Simpson 1946, Ksepka et al. 2006. Estimates of body mass in Mancallinae (excluding Miomancalla howardi) range from 1 kg in the smallest species (i.e., Mancalla californiensis) to 4kg in larger species (i.e., Mancalla lucasi; Livezey 1988). Although smaller than the 5kg mass estimated for Pinguinus, the estimated body mass of Mancallinae is greater than volant extant alcids (Livezey 1988). Miomancalla howardi is the largest known Mancallinae, and given the increased shortening and dorsoventral compression of wing elements of Mancallinae as compared to Pinguinus, it may have approached the mass of Pinguinus. Several Pliocene species of Alca are known to have exceeded the size of extant Alca torda (Olson and Rasmussen 2001;Smith and Clarke in review), and estimates based on fossils from Belgium indicate that at least one Pliocene Atlantic species, Alca stewarti Martin et al., 2001, was approaching the wing-loading threshold for fl apping-fl ight (Martin et al. 2001;Dyke and Walker 2005). Th is apparent trend towards increased size in two separate alcid lineages, known from separate ocean basins during the Miocene and Pliocene is in stark contrast to the smaller body size of most extant alcids. Th e largest extant alcids are the Murres (Uria aalge and Uria lomvia), with an average body mass of 800-1000g, but the most speciose clade of extant alcids, the auklets Aethia and Ptychoramphus, are among the smallest of extant alcids with a body mass of 85-297g (del Hoyo et al. 1996). Additionally, the Mancallinae lineage and the Alca + Pinguinus lineage are considered the dominant seabirds in their respective oceans during the Pliocene (Olson 1985;Olson and Rasmussen 2001). Th is temporal disparity in size suggests that the conditions that led to radiations of large alcids in the Pacifi c and Atlantic Oceans are no longer in place, and that small to moderate size may have played a role in diff erential survival of alcid species since the Pliocene. However, the largest known alcid, the Great Auk, was not driven to extinction by competition from smaller species or lack of ability to adapt to a changing environment, but rather was exterminated through human exploitation (Bengston 1984;Fuller 1999).
Body size in extant alcids has been correlated with dive depth and feeding ecology (Piat and Nettleship 1985;Prince and Harris 1988;Watanuki and Burger 1999), and larger body size in extant alcids is associated with piscivory (Bradstreet and Brown 1985). Foraging ranges, dive depths, and prey selection are similar in extant alcids and penguins (Prince and Harris 1988). Little is known about the feeding strategies of Pinguinus (Olson 1977), and there is no direct evidence of feeding strategies in Man-callinae; however, the large size of many Mancallinae and morphological comparisons with extant piscivorous alcids suggest that Mancallinae were specialized for piscivory. For example, the terete bill of Mancalla (e.g., LACM 103940) may be evidence of piscivory, because this characteristic in alcids has been linked with that feeding strategy (Bédard 1969).

Geological and phylogenetic context for Pan-Alcidae
Th e oldest unequivocal fossil alcid (GCVP 5690) is from Late Eocene deposits of the Hardie Mine, Gordon, Georgia, USA (Chandler and Parmley 2002). Likely because of the incompleteness of the specimen, phylogenetic results (not shown) place this specimen at the base of Alcinae in an unresolved polytomy with other Alcinae clades. However, this placement is based upon a single shared character (equal width of the tricipital sulci) and the possibility that characteristics shared with Alcinae are pleisiomorphic for Alcidae should be considered. Accordingly, this fossil is considered Alcidae incertae sedis, rather than Alcinae insertae sedis. Th e presence of alcids in Late Eocene (Chandler and Parmley 2002) is congruent with divergence estimates placing the origin of Alcidae in the Paleocene (Baker et al. 2007;Pereira and Baker 2008). Although, as pointed out by Wijnker and Olson (2009), those divergence estimates suff er from serious fl aws with respect to the taxonomic status and ages assigned to fossils used as calibrations.
Th e taxonomic status of all but one earlier (i.e., Mesozoic, Paleocene, and Earlymid Eocene) fossil referred to Charadriiformes (Olson and Paris 1987;Harrison and Walker 1977) consists of unassociated, undiagnosable fragments (Hope 2002;Mayr 2005Mayr , 2009. Th e earliest known defi nitive charadriiform fossil is a humerus that is tentatively referred to the Charadri (Hou and Ericson 2002). Although no radiometric-based date is known for this fossil, the age of Jiliniornis huadianensis Hou & Ericson, 2002 (IVPP V.8323) is estimated to be Middle Eocene based on biostratigraphic correlation (Hou and Ericson 2002). A minimum age of divergence between Alcidae and other charadriiforms in the Eocene suggests that the charadriiform fossil record is quite incomplete (i.e., extensive ghost lineages inferred based upon the fossil record).
Th e fossil record of Mancallinae ranges in age from Middle Miocene through Late Pleistocene (i.e., Turtonian-Calabrian; Becker 1987). Th e oldest record of Mancallinae may be the holotype specimen of Miomancalla wetmorei (LACM 42653) from the Mid-Late Miocene Monterey Formation exposed in Laguna Niguel, California; although, the precise stratigraphic position of the holotype locality is unknown. Deposition of the Monterey Fm. spans ~10ma from 17.9-7.4 Ma (i.e., Turtonian; DePaolo and Finger 1991). Th e holotype specimen is from the upper part of the formation (Domning and Deméré 1984), and would therefore be ~12-7.4 Ma. Miomancalla howardi is known from the Late Miocene San Mateo Formation, which ranges in age from 8.7-4.9 Ma (Zanclean-Messinian; Deméré and Berta 2005). Miomancalla is replaced in Pliocene sediments by Mancalla, with four species known from the Capistrano, San Diego, San Mateo, Niguel, Almejas, and Purisima Formations. Th e San Mateo Fm. records the highest diversity of Mancallinae, with Miomancalla howardi found in the lower unit, and Mancalla cedrosensis, Mancalla lucasi, and Mancalla vegrandis from the upper unit. Th e Capistrano Fm., which may be correlative with the San Mateo Fm. , has produced remains of Miomancalla howardi from the lower unit and Mancalla californiensis from the upper unit. Th e most geographically widespread and chronologically long-lived species (~5.0 Ma -470 ka) is Mancalla lucasi, known from the Pliocene San Mateo, San Diego, and Niguel formations, and also from the Pleistocene Hookton Formation (Howard 1970;Kohl 1974;Domning and Deméré 1984).
Just as coldwater upwelling is linked to biological productivity in modern seabird communities (Hyrenbach and Viet 2003;Briggs et al. 1987) the Miocene appearance of Miocepphus in the Atlantic Ocean and Miomancalla in the Pacifi c Ocean coincides with the formation of permanent Antarctic icecaps and shallowing of the Central American Seaway (CAS) that resulted in steeper latitudinal thermal gradients. Th is resulted in intensifi ed gyral circulation of surface waters, and strengthened coastal and trade winds that promote upwelling (Ford and Golonka 2003). Th e Early Pliocene (~5 -3.5 Ma) was a time of relative climate stability and high sea level that coincides with the appearance of speciose alcid lineages in the Atlantic and Pacifi c Oceans (Warheit 1992). High Mancallinae diversity in the Pacifi c Ocean, and high Alca diversity in the Atlantic Ocean (Olson and Rasmussen 2001; Smith and Clarke in review) coincides with documented cooling during the Late Miocene and Early Pliocene (~14-3.6 Ma), and establishment of the California current system in the Pacifi c (Zachos et al. 2008;Lariviere et al. 2009). Although the geology of eastern Pacifi c marine units is more complex than that of coeval geologic formations from the passive Atlantic margin, sea-level fl uctuation records indicate that the same Early Pliocene cycles of transgression and regression are recorded on western Atlantic and eastern Pacifi c coasts (Haq et al. 1988). Th e Middle Pliocene (~3.5-3.0 Ma) was characterized by continued global cooling, continued shallowing of the CAS, and the beginning of northern hemisphere glaciation cycles which led to increased coldwater upwelling in the Pacifi c (Bartoli et al. 2005;Lawrence et al. 2006). Th e emergence of the Panamanian Isthmus and the fi nal closure of the CAS at ~2.7 Ma resulted in increased northern hemisphere glaciation, which is associated with a severe drop in sea-level (~45m) and the establishment of the modern profi le of the California ocean-current system on which Pacifi c alcids rely today (Hyrenbach and Viet 2003;Bartoli et al. 2005). Th e microfaunal record documents a southward shift in Atlantic and Pacifi c cold-adapted foraminiferal faunal regimes (Bartoli et al. 2005), separation of Pacifi c and Caribbean cocolith assemblages at 2.74 Ma in response to fi nal closure of the Isthmus, and an increase in thermohaline circulation as a result of separation of the Atlantic and Pacifi c Ocean basins. By ~2.5 Ma the modern climate regime was in place, involving small (i.e., meter scale) fl uctuations in sea level associated with Late Pliocene and Pleistocene glacial cycles (Bartoli et al. 2005). Th e apparent response of seabirds to these climate-related changes in the environment was a signifi cant decrease in diversity (Warheit 1992; Olson and Rasmussen 2001;Dyke and Walker 2005), because only a single species of Alca survives today in the Atlantic, and only a single specimen of Mancalla is known from the Pleistocene (Howard 1970;Kohl 1974,). Confi rmation of causal links between these climatic shifts and decreased seabird diversity will require more intense sampling of Late Pliocene and Pleistocene seabird fossils and evaluation of other proposed factors such as competition for nesting grounds with pinnipeds (Warheit and Lindberg 1988).
Known diversity of extinct Atlantic alcids now approaches that of extinct Pacifi c alcids (~16-19 species ranging from Miocene-Pleistocene age; Smith and Clarke in review). Th e diff erential extinction of Atlantic alcids, compared with that of Pacifi c lineages, may be linked to climatic changes that eff ected the Atlantic and Pacifi c Oceans in diff erent ways. Th e alcid Pacifi c Ocean origin hypothesis is based primarily on higher extant diversity in the Pacifi c Ocean; however, higher extant diversity in the Pacifi c is not evidence of origination area, and the two oldest known alcid fossils are both from Atlantic deposits (Wetmore 1940;Chandler and Parmley 2002;Wijnker and Olson 2009). Although the lack of older fossils from the Pacifi c may simply refl ect a gap in the fossil record, hypotheses concerning Pacifi c ancestral origination of alcids based upon proposed greater extant Pacifi c species diversity should accordingly be re-evaluated. However, the basal position of Mancallinae and their restriction to the Pacifi c basin may be viewed as support for the Pacifi c origin hypothesis for Pan-Alcidae (Storer 1952;Kozlova 1957;Olson 1985;Konyukhov 2002;Pereira and Baker 2008).
Regardless of the ancestral area of the clade (i.e., Atlantic or Pacifi c), hypotheses regarding the spread of alcids from one ocean basin to another include dispersal by icefree northern passage through the Bering Strait and Arctic Ocean, and southern dispersal across the submerged Isthmus of Panama (Olson 1985;Konyukhov 2002;Pereira and Baker 2008). Th ese hypotheses are based upon the assumption of dispersal across water, and the fi rst occurrence datum (FAD) for alcid clades, which until the discovery of an auk from the Eocene of Georgia, USA (Chandler and Parmley 2002), included Miocene examples of Mancallinae (Howard 1976), Cepphus (Howard 1982), and Uria (Howard 1981) from Pacifi c deposits, and Miocepphus (Wetmore 1940) and Alca (Wijnker and Olson 2009) from Atlantic deposits. Th e ornithological literature is replete with records of occurrences of alcids hundreds or even thousands of miles from their normal ranges (see Konyukhov 2002) and records of alcid 'wrecks', sometimes composed of thousands of individuals, that were blown many kilometers inland from the sea by storms (Fisher and Lockley 1954;Stewart 2002a). Given the expanse of geologic time being considered (Paleocene-Recent), the possibility that such events may have led to the dispersal of populations from one ocean basin to another ocean basin must be considered.
As suggested by Bédard (1985), the presence of Atlantic alcids in the Eocene (Chandler and Parmley 2002) confi rms that the cold adapted lifestyle of some alcids (e.g., Uria) evolved from ancestors that were adapted to warmer (i.e., Eocene) climates. Th e development of basically modern ocean circulation patterns was not achieved until ~24-20 Ma when opening of Drake Passage initiated dramatic cooling of Antarctica and formation of a strong Antarctic current that resulted in a switch from high productivity in equatorial regions, to more northern coastal regions (Lear et al. 2000;Ford and Golonka 2003;Liu et al. 2009). Although the southern location of the earliest alcid fossil locality (Georgia, USA) cannot necessarily be interpreted as support for a southern route of dispersal, warm-adapted alcids in the Eocene likely were not restricted to a northern dispersal route.

Conclusions
Rigorous taxonomic evaluation of alcid fossil material resulted in a more refi ned picture of diversity within Mancallinae, and facilitated phylogenetic analysis of specieslevel relationships within the clade. Th e combined analysis and total evidence approaches adopted herein resulted in a well-resolved and strongly supported hypothesis of the position of Mancallinae with respect to other Charadriiformes, and the interrelationships of Mancallinae species. Th e phylogenetic position of Mancallinae as the sister taxon to all other Alcidae (i.e., crown clade Alcidae) suggests extensive ghost lineages in Pan-Alcidae, provides further evidence that the charadriiform fossil record is quite incomplete, and demonstrates that fl ight was lost independently in at least two lineages of alcids. Th e stem-lineage position of Mancallinae recovered in this analysis is consistent with previous phylogenetic placement of this clade (Chandler 1990a), but contrasts with previous hypotheses of close relationship between Mancallinae and Alcinae (Olson 1985). Although extremely derived morphologically as a result of modifi cations related to fl ightlessness, Mancallinae do possess a unique suite of characters, some of which are otherwise found exclusively in Alcinae or Fraterculinae, and some of which are otherwise known only from non-alcid charadriiforms. Although it would not aff ect the number of inferred origins of fl ightlessness in Alcidae, the placement of Mancallinae at the base of Alcinae, or at the base of Fraterculinae, would only require an additional 2 steps of tree length (manually calculated in MacClade; Maddison and Maddison 2005), and thus the position of Mancallinae recovered here may be sensitive to the inclusion of additional fossil taxa with morphologies representing ancestral states for Pan-Alcidae. Th e hypothesized split between the lineages leading to Mancallinae and crown clade Alcidae raises questions about the evolution of fl ightlessness in charadriiforms, and the biological factors that may have led to the split between Alcidae and their proposed sister taxon, Stercorariidae.
Miomancalla howardi is placed as the sister taxon of Miomancalla wetmorei, and is the largest known species of Mancallinae. Th e large size and resemblance of the bill of Miomancalla howardi to that of the Great Auk Pinguinus impennis provides an example of within-lineage convergence between two species separated by time and geography. Th e independent acquisition of morphological characteristics in both lineages of fl ightless alcids (i.e., Mancallinae and Pinguinus), and the similarity of these modifi cations to those of penguins and plotopterids, strongly suggests correlation between these morphologies and mode of locomotion. Th e study of convergence within Alcidae may provide insights about the evolution of fl ightlessness in penguins, in which there are no known volant species.
Similarly diverse lineages of alcids inhabited the eastern and western coasts of North America during the Miocene and Pliocene. Approximately coeval Early Pliocene deposits in California and North Carolina record the replacement of Miocepphus by Alca in the Pliocene of the Atlantic, and the replacement of Miomancalla by Mancalla in the Pliocene of the Pacifi c. Global-scale environmental perturbations such as increased cooling following the MMCO, may have contributed to similar scenarios involving species turnover in Pan-Alcidae in both ocean basins.

Appendix 1. Review of the Mancallinae fossil record
Owing to the recognition of several Mancallinae species based upon non-diagnostic material, the systematics of Mancallinae required extensive revision. Th e following review of the Mancallinae fossil record is presented to clarify the systematic position of previously named species and referred fossil material, and to justify the exclusion of some previously named species from the phylogenetic analysis. Although more than 100,000 avian fossils are now known from sediments in California (Miller 1946;Brodkorb 1967;Olson, 1985), the fi rst avian fossil from this state was not reported until 1901 when F. A. Lucas described a nearly complete left humerus from what were thought to be Late Miocene sediments of Los Angeles. Th at specimen (USNM 4976; Fig. 2) represented the fi rst of approximately 4000 fossils that are now referred to the fl ightless alcid taxon Mancalla (Smith, pers. obs). Mancalla californiensis Lucas 1901 was the fi rst of seven fl ightless alcid species recognized between 1901 and 1981 (Table 1).
Th e second report of Mancalla remains (humerus; catalog # uncertain) came from the Early Pliocene San Diego Formation exposed in San Diego, California (Miller 1933). Th e Early Pliocene age of that specimen was congruent with the revised age estimate for the holotype locality of Mancalla californiensis in Los Angeles (Arnold 1906). An additional specimen, a complete right femur (UCMP 33409) from the San Diego Fm., was reported by Miller in 1937. Based on characteristics of that specimen Miller (1937) considered it a Pliocene example of a puffi n (i.e., 'Lunda', Fratercula, and Cerorhinca), and designated the specimen as the holotype of a new taxon, Pliolunda diegense Miller, 1937. Additional Mancalla remains (LM 2218) were reported by Miller (1946), who discussed the possibility that Pliolunda was a synonym of Mancalla, and erected the Family Mancallidae, separating Mancalla from Alcidae. Th e rank of Mancallidae later became subfamily Mancallinae (sensu Brodkorb 1967), systematically reuniting Mancalla and Praemancalla with other Alcidae.
Mounting evidence that more than one species of Mancalla was present during the Early Pliocene came from Howard in 1949. At that time approximately 118 specimens representing Mancalla were known, including two size classes of carpometacarpi from localities in Los Angeles, San Diego, and Corona del Mar, California. Although no associated remains were known, carpometacarpi were referred to Mancalla based upon characters such as an elongated fi rst metacarpal, a morphology considered convergent with that of penguins (i.e., Spheniscidae) by Howard (1949). Humeri of Mancalla were well known, and also display characteristics related to extreme specialization for wing-propelled diving considered by Miller (1946) to be convergent with those of penguins, and prompting the referral of carpometacarpi exhibiting 'penguin-like' features.
Th e growing number of specimens from the San Diego Fm. prompted a review known remains of Mancalla (Miller and Howard 1949), which resulted in the recognition of Pliolunda as a junior synonym of Mancalla. However, additional remains other than humeri were referred solely on the basis of size, provenience, and osteological characteristics correlated by those authors with fl ightlessness. No associated Mancalla remains were known at the time that would allow for referral of femora to Mancalla californiensis, nor to facilitate comparisons between Mancalla californiensis and the holotype specimen of Mancalla diegense. Th e species name Mancalla diegense was emended to Mancalla diegensis by Olson (1981) to refl ect correct latinization of the place name San Diego. Although my recent re-examination of Mancalla material in the collections of UCMP, LACM, and SDSNH identifi ed several associated specimens within the size range of Mancalla diegensis as reported by Howard (1970), and that correspond with characters described for that taxon by Howard (1970), no associated specimens referable to Mancalla californiensis that preserved femora were identifi ed. Th e holotype femur of Mancalla diegensis is, therefore, not presently comparable to Mancalla californiensis. Furthermore, my survey of the femora of all known alcid species revealed that the morphology of the femur is remarkably conserved across alcid taxa, potentially explaining Miller's (1937) original proposal, that UCMP 33409 represented an extinct species of puffi n. No characteristics were identifi ed that would allow for confi dent referral of isolated femora to Mancalla, and Mancalla diegensis is, therefore, considered Pan-Alcidae incertae sedis.
In 1966 Howard described a new Mancallinae taxon from the Late Miocene based upon isolated elements including a distal humerus, carpometacarpi, a partial coracoid, the proximal end of a scapula, and the articular portion of a mandible. Praemancalla lagunenesis Howard, 1966 was considered by that author to be less specialized with respect to features associated with loss of aerial fl ight, and the possibility that Praemancalla might represent a less derived ancestor of Mancalla was proposed. All elements referred to Praemancalla lagunenesis were isolated, so only the holotype distal humerus (LACM 15288) can be compared with previously recognized taxa to evaluate the taxonomic validity of this species. Th e holotype specimen of Praemancalla lagunensis is weathered smooth, obscuring many fi ne morphological details. Although LACM 15288 is referable to Mancallinae based upon the rounded anterior surface of the ventral condyle (153:0), all of the characteristics that Howard (1966) proposed as diagnostic for this species may be an artifact of weathering, or also are found in Mancalla. Praemancalla lagunensis is, therefore, considered Mancallinae incertae sedis.
Another species of alcid with characteristics interpreted as "progressing towards fl ightlessness" (Howard 1968:19), was described by Howard in 1968 from presumed Miocene sediments of Laguna Hills, California. Alcodes ulnulus Howard 1968 was described based upon isolated elements including a complete left ulna, additional ulnar fragments, and a partial carpometacarpus (Howard 1968). Additional material representing this taxon was recovered from the Middle Miocene Topanga Formation along Oso Creek in Orange County, California (Howard and Barnes 1987), confi rming the Miocene age of this species. Ulnae of Alcodes are diff erentiated from those of Mancalla by their more gracile and rounded shafts, and projection of the olecranon farther posteriorly. Although associated humeri and ulnae of Mancalla (e.g., holotype specimen of Mancalla cedrosensis LACM 15373) demonstrate that Alcodes is distinct from Mancalla, the lack of associated "Praemancalla" specimens with ulnae raises the possibility that Alcodes is congeneric with "Praemancalla". Until additional material is recovered that would allow comparison with other recognized alcid taxa, the systematic affi nities of Alcodes in Pan-Alcidae remain uncertain, and Alcodes is, therefore, considered Pan-Alcidae incertae sedis.
Although the review by Howard (1970) expanded the known geographic range of Mancalla, and greatly increased knowledge of character-and size-related diff erences in the taxon, the description of Mancalla milleri Howard, 1970 based upon isolated material further complicated the taxonomy of the clade. Comparisons between the holotype femur of Mancalla diegensis and the holotype femur of Mancalla milleri (LACM 2185) provide limited information because neither of those elements is directly comparable to the isolated holotype humerus of Mancalla californiensis. Additionally, my recent re-examination of the ~4000 fossils referred to Mancalla indicates that femoral characters cited by Howard (1970) are more variable within proposed size classes of Mancalla than previously recognized (Smith pers. obs.). Furthermore, as stated above, femoral morphology is remarkably conserved in Alcidae. Although characteristics of humeri indicate that multiple species of Mancalla are represented by Mancalla material from the San Diego Fm., the species to which the holotype femora of Mancalla diegensis and Mancalla milleri belong will likely never be determined. Mancalla milleri is, therefore, considered Pan-Alcidae incertae sedis.
Mancalla cedrosensis Howard, 1971 was the fi rst species of Mancalla described from associated remains, and also the fi rst that was directly comparable to Mancalla californiensis (Howard, 1971). Th e holotype specimen (LACM 15373; Fig. 2) and additional referred specimens were recovered from Early Pliocene deposits on Cedros Island off the coast of Baja California, Mexico (Howard 1971). Th e associated remains of Mancalla cedrosensis provided the fi rst reliable assessment of inter-element osteological proportions for Mancalla, proportions that supported earlier size-based estimates of diversity among material from the San Diego Fm. proposed by Howard (1970).
Praemancalla wetmorei Howard, 1976 was described based upon a nearly complete humerus (LACM 42653; Fig. 2) from Late Miocene sediments in Laguna Niguel, California. Several features distinguish this species from other Mancalla (see diagnoses below). An associated specimen (LACM 107028) was tentatively referred to Praemancalla wetmorei by Howard (1982) on the basis of overall resemblance between the ulna of that specimen and the paratype ulna of Praemancalla wetmorei. Because the paratype ulna (LACM 32429) is not associated with the holotype humerus, and was referred only on the basis of its occurrence within the same deposit, the affi nities of that specimen remain uncertain. Likewise, the affi nities of additional non-humeral material (e.g., LACM 53907, 37637, 52216) referred to Praemancalla wetmorei by Howard (1976) are uncertain, because those specimens are not comparable to the holotype, and therefore not referable to species at this time. As stated above, the name-bearing specimen of Praemancalla (i.e., Praemancalla lagunensis Howard, 1966) is Mancallinae incertae sedis. Based upon phylogenetic results and apomorphies shared with Miomancalla howardi sp. n., Praemancalla wetmorei is referred to Miomancalla, and becomes Miomancalla wetmorei (Howard 1976).
Mancalla emlongi was described based upon a complete ulna from Early Pliocene San Diego Fm. sediments in San Diego, California (Olson 1981). In the original description, comparisons were made between the holotype specimen of Mancalla emlongi (USNM 243765) and ulnae referred to Mancalla californiensis, Mancalla diegensis, Mancalla milleri, and Mancalla cedrosensis. As stated above, size and provenience alone do not in my opinion constitute strong evidence that material is referable to a taxon previously known from a particular locality or geologic formation. Mancalla diegensis and Mancalla milleri are Alcidae incertae sedis, and there are no known associated specimens that would allow for referral of ulnae to Mancalla californiensis. Although the holotype ulna of Mancalla emlongi can be diff erentiated from ulnae of Mancalla cedrosensis, the possibility exists that Mancalla emlongi is synonymous with another species of Mancallinae (e.g., Mancalla californiensis). Mancalla emlongi is, therefore, considered Mancallinae incertae sedis.
Additional material from the San Diego Fm. including a well-preserved skull and mandible (SDSNH 25236; Fig. 3) was tentatively referred to Mancalla emlongi by Chandler (1990b) on the basis of size. SDSNH 25236 and an additional skull (SDSNH 23753) are referable to Alcidae based upon the strongly protruding cerebellar prominence (35:0), and deeply incised temporal (31:1) and nasal fossae (20:1). Although no cranial apomorphies of Mancallinae have thus far been identifi ed, the cranium of Mancallinae can be diff erentiated from the skulls of all other known Alcidae: diff erentiated from Fraterculinae (Aethia, Ptychoramphus, Cerorhinca, and Fratercula) by the dorsal position and extension of the temporal fossae; diff erentiated from Brachyramphus, Synthliboramphus, Alle, Miocepphus, Alca, Pinguinus, Cerorhinca and Aethia by the lack of supraoccipital foramina; diff erentiated from Cepphus by protrusion of the cerebellar prominence farther posteriorly (condition resembles that in Uria), and deeper interhemispherical furrow along midline of skull; diff erentiated from Uria by depth of nasal fossae (deeper, distinctly bordered posteriorly, and laterally incised in Uria). Although these specimens cannot be referred to species at this time, two associated specimens comprising associated cranial and postcranial material (LACM 103940 Mancalla sp. and SDSNH 68312 Miomancalla howardi sp. n.) allow for comparison of SDSNH 25236 with known cranial morphology of Mancallinae. All three of the aforementioned specimens possess two small caudal mandibular fenestrae (46:1; Fig. 3D), a characteristic known only in the Fraterculini (i.e., Fratercula and Cerorhinca) among crown Alcidae, and also in the proposed sister taxon of Pan-Alcidae, the Stercorariidae. SDSNH 25236 is diff erentiated from Fraterculini by the lack of dorsoventral expansion of the premaxilla and mandible (2:0), and by the less acute angle formed between the jugal and the proximo-ventrally descending bar of the nasal (6:0). SDSNH 25236 is consistent in size and morphological characteristics with the skull and mandible of LACM 103940, which is the only known Mancalla specimen with both cranial and postcranial elements preserved. Additionally, SDSNH 25236 lacks the dorsoventrally expanded mandible of Miomancalla howardi, suggesting systematic placement within Mancalla.
Although Mancalla remains were reported from Pleistocene sediments in Shiriya, Japan (Hasegawa et al. 1988), that material was never described, fi gured, nor systematically evaluated. My recent reexamination of the material confi rms its referral to Mancallinae. Th e presence of Mancalla in Japan provides a considerable range extension, and based upon the age of the material, also confi rms that Mancalla survived into the Pleistocene in the eastern and western Pacifi c Ocean (Howard 1970;Kohl 1974 (1) hooked. Th e anterior tip of the premaxilla is hooked ventrally in a raptorial fashion in some alcids (e.g., Alca torda). Th e anterior tip of the premaxilla in other alcids (e.g., Brachyramphus marmoratus) is decurved slightly ventrally but does not possess a hooked tip.
3. Maxilla, fenestra adjacent to junction of maxilla and palatine: (0) absent; (1) present. Th e ventral surface of the distal end of the maxilla is fenestrated in some alcids (e.g., Cerorhinca monocerata). Th is characteristic is absent in many other alcids (e.g., Cepphus grylle). In life the fenestra is covered by a thin membrane. Because the fenestra does not serve as a passageway for muscle, tendon, or nerves, its purpose may be related to fl exion or weight reduction. 4. Nasal, anterior projection along the ventral surface of the premaxilla (Chandler, 1990b, character 9): (0) contacting; (1) separated. Th e nasals converge beneath the premaxilla in some species (e.g., Uria aalge), while in other species (e.g., Fratercula cirrhata) the lateral nasal bars merge with the ventral premaxilla but remain separated.
11. Palatine,ventral extent of the medial margin of the ventral crest relative to the palatine shelf (crista ventralis medialis, Baumel and Witmer 1993;Strauch 1985, character 3): (0) does not extend beyond ventral margin of palatine shelf; (1) extends beyond ventral margin of palatine shelf. Th e ventral end of the ventral crest of the palatine does not extend as far ventrally as the ventral edge of the palatine plate in most charadriiforms (e.g., Alca torda). In the auklets (e.g., Aethia psittacula) however, it extends beyond the end of the palatine shelf (Strauch 1985).
18. Lamina dorsalis, segmentation: (0) not segmented; (1) segmented. Th e lamina dorsalis is an extension of the mesethmoid that lies against the ventral side of the frontal (Baumel and Witmer 1993). Th is osseous feature is segmented and may play some part in bill kinesis in alcids (e.g., Uria lomvia). Th e lamina dorsalis is fused to the rest of the mesethmoid in many charadriiform species (e.g., Rynchops niger).
26. Lacrimal, articulation with ectethmoid (Chu 1998, character 26): (0) occupies entire lateral margin of ectethmiod; (1) occupies only the ventral half of the lateral margin of the ectethmoid. In the Alcidae (e.g., Alca torda), the lateral margin of the ectethmoid is dorsoventrally expanded and anteroposteriorly fl attened, giving this element a square shape when viewed anteriorly. Th e lacrimal articulates with the ectethmoid along its entire lateral margin. In many other charadriiforms (e.g., Sterna maxima) the ectethmoid tapers laterally to a point. In these taxa the lacrimal extends dorsally from the medially extending ectethmoid.
27. Lacrimal, position in lateral view: (0) posteroventrally directed; (1) ventrally directed. With the exception of Pinguinus impennis and Rynchops niger, the lacrimal of all taxa examined in this study are directed posteroventrally. In contrast, the lacrimal of P. impennis extends ventrally. Th e condition shared by P. impennis and R. niger is not considered homologous here, as the cranium of R. niger is extremely derived (with respect to other charadriiforms).
32. Squamosal, temporal fossa, medial extent: (0) not medially extended; (1) separated by a thin fl at space; (2) separated only by a thin crest. In many alcids (e.g., Aethia psittacula) the temporal fossa is not expressed on the dorsal surface of the skull, although, in some species (e.g., Alca torda) the temporal fossa nearly converge on the dorsal surface of the skull. In Pinguinus impennis the temporal fossa are very deep and separated only by a thin crest. Ordered 33. Squamosal, temporal fossa, shape of medial margin: (0) narrow; (1) broad. In species that possess medially expanded temporal fossa (see character 32) the medialmost extent of the temporal fossa varies in alcids from a broad, relatively 'U-shaped' curve (e.g., Alca torda) to a more pointed, medially narrowing groove (e.g., Uria aalge).
Mandible 39. Mandible, length of symphysis (modifi ed from Chandler 1990b, character 22): (0) short; (1) long. Th e left and right rami of the mandible fuse at the anterior end of the mandible. Th e length of area fused can be either short (i.e., <15% of the total length of the mandible; e.g., Alca torda) or long (i.e., >15% of the total length of the mandible; e.g., Uria aalge).
52. Articular, retroarticular process length (modifi ed from Chandler 1990b, character 23): (0) short; (1) long. In some species (e.g., Aethia pygmaea) the dorsoposteriorly projecting process of the cotyla lateralis is long (i.e., as long or longer than the dorsoventral height from the articular facet to the ventral margin of the mandible in lateral view), while in other species (e.g., Alca torda) this process is shorter.

Atlas, fl ange on the lateral margins of the arcus atlanticus in dorsal view:
(0) straight; (1) laterally angled. Th e posteriorly-projecting processes for articulation with the axis project posteriorly in most species of alcids (e.g., Uria aalge). In some species of alcids (e.g., Fratercula arctica) the zygapophyses angle laterally.
54. Axis, dorsal extension of neural spine: (0) short; (1) long. In posterior view, the neural spine of the axis in most alcids (e.g., Uria aalge) is short (i.e., less than half of the length of the neural spine extends above the level of the anapophyses), although in some alcids (e.g., Alca torda) this projection of the axis is lengthened and extends to a point well above the anapophyses (i.e., more than half of the length of the neural spine extends above the level of the anapophyses).
55. Th oracic vertebrae, hypapohyses: (alae cristae ventralis; Baumel and Witmer 1993; modifi ed from Strauch 1985, character 14): (0) not present on any thoracic vertebrae; (1) present on some thoracic vertebrae. Th e Lari and other non-alcid charadriiforms (e.g., Larus marinus) have poorly developed hypapophyses on their thoracic vertebrae. Well-developed hypapophyses, most with bilateral fl anged wings, are found in all alcids, but the number of vertebrae on which they occur varies among the species. Th ese structures serve as increased area for attachment of M. longus colli ventralis, are functionally correlated with the strength needed by diving birds (Kuroda 1954). It is hypothesized that a greater number of vertebrae with well-developed hypapophyses is a more derived condition. Similar structures are found in other diving birds such as loons, grebes, penguins, and some anseriforms (Beddard 1898in Strauch 1985. (1) sulci separated by manubrium. Th e sternal articular surface of the coracoid is a continuous, smooth, depression in many charadriiforms (e.g., Larus marinus), while in Alcidae (e.g., Alca torda) the sulci are separated by the manubrium (i.e., rostrum sterni; Baumel and Witmer, 1993).
63. Sternum, medial notch (Strauch 1985, character 8): (0) absent; (1) present. Most charadriiforms (e.g., Larus marinus) have a medial sternal notch, but several, including members of the Lari and Alcidae (e.g., Aethia pusilla), do not. Distribution of the states among other charadriiforms thus does not indicate which state is primitive in the alcids. Only the puffi ns (Fraterculini) retain the remnant of the medial sternal notch as a medial sternal fenestra.
64. Sternum, medial notch, shape: (0) a notch; (1) a fenestra. Among alcids, only the puffi ns (Fraterculini) retain the remnant of the medial sternal notch as a medial sternal fenestra. (Strauch 1985, character 9;Chandler 1990b, character 38): (0) a notch; (1) a fenestra. Almost all charadriiforms (including all Lari) have a lateral sternal notch. In the auklets it is reduced to a fenestra, a condition assumed to be a derived state in the Alcidae. Shufeldt (1888Shufeldt ( , 1889 and Lucas (1890) reported that in the auks the lateral sternal notch tends to become ossifi ed with age. Th is condition clearly diff ers from that in the auklets; it is hypothesized to represent merely a variant of the state with the notch present. Kuroda (1954Kuroda ( , 1955 illustrated the variation with age of the sternal notching of some alcids (Strauch 1985).
67. Sternum, posterior extension of carina relative to lateral sternal notches/ fenestrae (modifi ed from Chandler 1990b, character 40): (0) carina extends to distal ends of notches/fenestrae; (1) lateral sternal notches/fenestrae extend posteriorly beyond posterior extent of carina. Th e length of the carina relative to the posterior extent of the lateral sternal notches/fenestrae of alcids varies from extending to a point about equally posterior to the posterior margins of the lateral sternal notches/fenestrae in some alcids (e.g., Alca torda), to a condition in which the lateral sternal notches/ fenestrae extend posterior to the carina (e.g., Aethia cristatella).
68. Sternum, supracoracoideus scar, position: (0) angled medially; (1) straight. In contrast to the condition observed many charadriiforms (e.g., Sterna maxima) in which the scar for the supracoracoideus muscle on the ventral surface of the sternum angles medially from the coracoidal sulcus towards the carina, in Alcidae this scar extends posteriorly for almost the entire length of the carina. Th is feature is correlated with the increased resistance during the upstroke experienced by alcids while fl ying underwater (Kozlova 1957).
70. Sternum, length of area between distal extent of medial fenestra and posterior margin (modifi ed from Chandler 1990b, character 43): (0) short; (1) long. Th e ossifi ed area of sternum posterior to the termination of the carina (i.e., the xiphoid) is short (i.e., wider than long) in some alcids (Alle alle), while in others (e.g., Cepphus grylle) this feature is much longer (i.e., nearly as long or longer than it is wide). 71. Sternum, length: (0) short; (1) long. When compared to their immediate outgroup, the Stercorariidae, alcids have an elongated sternum (i.e., sternum >2× long than wide), a character which has been associated with diving (Storer 1960). Th e greatest length of the sternum (i.e., from the anterior manubrium to the distal xiphoid) is more than two times the width of the sternum across the sternocoracoidal processes in all alcids (e.g. Alca torda). (1) small. A medially oriented crest-like projection characterizes the furcular symphysis of alcids. Th is crest can be either small (i.e., projects less than the width of individual clavicles at symphysis; Aethia psittacula) or large (i.e., projection as wide or wider than that of individual clavicles at symphysis; (e.g., Brachyramphus marmoratus).
75. Furcula, curvature of omal extremity (Chu 1998, character 76): (0) smoothly curving; (1) sharply curved or angled at posterior extremity. Th e transition from the dorsally extending shaft of the clavicles to the omal extremity of the clavicles in Alcidae (e.g., Brachyramphus perdix) is characterized by a distinctly angular bend. Th e furculae of all other charadriiforms (e.g., Larus marinus) examined during the course of this study exhibited a more gently sloping furcular curvature.
77. Furcula, coracoidal tuberosity, position relative to coracoidal facet (Chandler 1990b, character 33): (0) medially adjacent to coracoidal facet; (1) separate and anterior to facet. Th e coracoidal tuberosity contacts the medial margin of the coracoidal facet in alcids (e.g., Brachyramphus perdix), while in many other charadriiforms (e.g., Gygis alba) this tuberosity is more robust, separate from, and anterior to the coracoidal facet. (1) laterally oriented scar. In some alcids (e.g., Uria aalge), the attachment of the acrocoracoacromiale ligament is an anteriorly oriented excavation of the ventral surface of the acromium process bordered medially by a crest. Th is same attachment point in other alcids (e.g., Aethia cristatella) is rotated laterally and is characterized by a relatively smooth attachment surface.
82. Scapula, shape of distal extremity: (0) curved; (1) angled. In contrast to the gently ventrally curving distal extremity of many charadriiforms (e.g., Larus argentatus), the scapulae of all known alcids are characterized by a ventrally directed angular bend proximal to the distal most extremity.
84. Coracoid, furcular facet, notch posterior to bicipital tubercle: (0) absent; (1) present. Th e ventral margin of the furcular facet is curves dorsally just posterior to the process for the attachment of the bicipital muscle in some species of alcids (e.g., Uria aalge). Th e ventral margin of this feature in other alcids (e.g., Alca torda) is gently curved but not notched.
85. Coracoid, supracoracoidal sulcus: (0) pneumatic; (1) apneumatic, but deeply undercut; (2) not deeply undercut. Th e medial side of the distal end or head of the coracoid of some charadriiforms (e.g., Anous tenuirostris) are characterized by a pneumatic excavation. Th e coracoids of all alcids are apneumatic, although the brachial crest is deeply undercut for the passage of the supracoracoideus muscle in some species of alcids (e.g. Cepphus grille), while in some alcids (e.g., Cerorhinca monocerata) the brachial crest is not deeply undercut (i.e., ventrally concave). Ordered 86. Coracoid, brachial tuberosity, shape: (0) a tubercle; (1) a crest. Th e brachial tuberosity is developed as an anteroposteriorly oriented crest in some alcids (e.g., Cepphus grille), while in other alcids (e.g., Aethia psittacula) the brachial tuberosity is devel-oped simply as a small rounded tubercle positioned roughly at the midpoint on the neck of the coracoid. Th e term brachial crest is used here to describe the latter condition.
88. Coracoid, neck in dorsal view (Chandler 1990b, character 29): (0) short; (1) long. Th e neck of alcid coracoids (defi ned here as the head of the coracoid distal to the distal-most extent of the glenoid facet), which extends medially to articulate with the furcula, is elongate (i.e., considerably longer than wide) in some species (e.g., Uria aalge) and gives the neck of the coracoid a rectangular appearance in dorsal view. In other species (e.g., Fratercula cirrhata) this neck is shorter (i.e., roughly as wide as it is long) and results in a rather square coracoidal neck.
89. Coracoid, supracoracoideus scar development: (0) a distinct ridge; (1) ridge reduced or absent. Contact with the supracoracoideus creates a distinct, medially oriented ridge/scar in most alcids (e.g., Alca torda) that gives the shaft of the coracoid a distinctly angular cross-section, while in Cepphus this structure is greatly reduced or absent and the cross-section of the coracoid element is more rounded.
90. Coracoid, supracoracoidal nerve foramen (Strauch 1985, character 13;Chandler 1990b, character 25): (0) absent; (1) present. Th e Lari and most other charadriiforms have a coracoidal foramen (e.g., Pinguinus impennis); it is absent in some species of alcids (e.g., Aethia pusilla; Strauch, 1985). 91. Coracoid, position of supracoracoidal nerve foramen (0) distal; (1) proximal. In alcids that possess a coracoidal foramen, the position of this feature is typically near the midpoint of the of the procoracoid process near the shaft of the coracoid (e.g., Pinguinus impennis), although in Cepphus this foramen is positioned on the extreme anteroproximal edge of the procoracoid process leaving only a very thin strut of bone which forms the dorsal margin of the procoracoid process.
93. Coracoid, tip of procoracoid: (0) straight; (1) hooked. In Brachyramphus, the tip of the procoracoid is hooked anteriorly. Th is feature is absent in all other Alcidae for which the coracoid is known.
96. Coracoid, scar on anterior face of lateral edge of coracoid: (0) absent; (1) present. Alcids (e.g., Cepphus grylle) possess a distinct scar along the anterior surface of the lateral process that is lacking in other charadriiforms (e.g., Bartramia longicauda). Th e exact origin of this scar is unclear, although Fürbringer (1888) discusses several accessory ligaments that attach in this area.
97. Coracoid, scar extension along anterior surface of lateral process: (0) extends to sternal articulation; (1) bordered sternally by crest. Th is scar is less medially and sternally extended and more excavated in the auklets and puffi ns (e.g., Fratercula cirrhata) than in other alcids (e.g., Alca torda) in which this scar is less excavated and extends to the sternal margin of the coracoid (i.e., not bordered sternally by a crest).
98. Coracoid, crest along sternal edge of lateral process: (0) absent; (1) present. In anterior view the sternal edge of the lateral process of some alcids (e.g., Uria aalge) is characterized by a crest or thickening of the sternal margin. Th is characteristic is absent in some alcids (e.g., Alca torda).
100. Coracoid, medial sternal process, notch in dorsal margin: (0) absent; (1) present. Th e posteromedial margin of the proximal coracoidal shaft just distal to the medial sternal articulation (angulus medialis; Baumel and Witmer, 1993) is characterized by a small dorsoanterior oriented projection, giving the shaft a notched appearance in medial view at this point in most alcids (e.g., Aethia cristatella). Th e medial angle is more pointed in other alcids (e.g., Alca torda).
104. Humerus, deltopectoral crest, distal extension (modifi ed from Chandler 1990b, character 53): (0) does not extend to midpoint of shaft; (1) extends distally to the midway point of shaft; (2) extends beyond the midpoint of the humeral shaft. Th e deltopectoral crest extends distally along the anterodorsal margin of the humeral shaft to a point roughly ⅓ to one half of the distance towards the distal end of the shaft in most species of alcids (e.g., Fratercula arctica), while in Pinguinus the deltopectoral crest extends to the halfway point along the shaft. In Mancalla this crest extends distally beyond the midpoint of the shaft.
106. Humerus, deltopectoral crest, dorsal curvature: (0) concave; (1) fl at. In dorsal view, the area between the dorsal surface of the deltopectoral crest and the dorsal tubercle (i.e., the dorsal shaft distal to the head) is concave in many charadriiforms (e.g., Creagrus furcatus). In all alcids except Alca stewarti this space is fl at or slightly convex in some cases (e.g., Brachyramphus marmoratus).
107. Humerus, impressio coracobrachialis scar, depth (Chandler 1990b, character 60): (0) very deep; (1) deep; (2) shallow. Th e scar for attachment of the impressio coracobrachialis muscle in alcids (e.g., Aethia psittacula) is a shallow (i.e., smoothly transitions to anterior surface of humeral head), usually rounded impression (e.g., Alca torda). Th is is in contrast to the condition in most other charadriiforms, in which this muscle scar is a very deeply excavated, usually triangular pit. Ordered 108. Humerus, distal edge of bicipital crest, angle with respect to long axis of shaft: (0) not perpindicular; (1) nearly perpindicular. Th e ventral edge of the bicipital crest forms a nearly perpindicular angle to the shaft in some species (e.g., Pinguinus impennis) while in other species (e.g., Alca torda) the bicipital crest is positioned at an obtuse angle with respect to the long axis of the humeral shaft.
109. Humerus, biciptal crest, transition to shaft: (0) smooth; (1) notched. Th is character, noted by Olson and Winker, (2009), varies from a condition where (in anterior view) the bicipital crest transitions smoothly onto the humeral shaft (e.g., Aethia pusilla) to a condition in which there is a distinct notch or separation between these structures (e.g., Alle alle).
111. Humerus, coracobrachial sulcus, curvature: (0) dorsal; (1) ventral. Th e distal most point of the bicipital surface, as defi ned by the curvature of the coracobrachial sulcus, which curves or angles dorsal to the bicipital crest on the anterior surface of the humerus in some alcids (e.g., Pinguinus impennis), while in other alcids (e.g., Alle alle) the coracobrachial sulcus and the distal edge of the bicipital surface extend ventrally to terminate where the bicipital crest contacts the ventral surface of the humeral shaft.
113. Humerus, supracoracoideus scar, shape: (0) round; (1) long, proximally broadening; (2) long, does not broaden proximally. Th e attachment of the supracoracoideus muscle on the proximal humerus of most charadriiforms (e.g., Larus marinus) is a rounded scar, while in alcids this scar is distally elongated (Crista m. supracoracoidei; Baumel and Witmer 1993). In some alcids (e.g., Fratercula arctica) the proximal end of the scar is much broader than the distal end, while in other alcids (e.g., Alca torda) the scar is relatively the same width throughout its length.
114. Humerus, supracoracoideus scar, transition into the secondary pneumatic fossa (pf2): (0) pf2 borders scar; (1) scar separated from pf2; (2) margo caudalis widely separates pf2 and scar. In most alcids (e.g., Alca torda) the dorsal extent of the excavation of the second pneumatic fossa parallels the ventral margin of the supracoracoideus scar. In some species (e.g., Cerorhinca monocerata) the excavation for pneumatic fossa 2 is separated from the supracoracoideus scar by a thin, fl at, lateromedially oriented projection of the humeral shaft (which is most like the very reduced remains of the margo caudalis). Th e supracoracoideus attachment point in many other charadriiforms (e.g., Larus marinus) is widely separated from the medial portion of the humeral shaft by the margo caudalis and does not extend as far distally as the condition seen in alcids. Ordered 115. Humerus, medial crest between pneumatic fossae, extension relative to the bicipital crest (modifi ed from Chandler 1990b, character 51): (0) ends proximal to distal-most extension of bicipital crest; (1) crest extends to distal extant bicipital crest; Th e crest which divides the pneumatic fossae varies in the distance it extends distally towards the distal margin of the bicipital crest. In some species (e.g., Pinguinus impennis) this crest terminates proximal to the distal edge of the bicipital crest. In some species (e.g., Alle alle) this crest extends to the distal edge of the bicipital crest.
121. Humerus, mancalline scar on posterior side of proximal humerus, conformation: (0) excavated; (1) raised. Th e dorsal and ventral borders of the scar on the posterior side of the proximal humerus of Mancalla extend parallel to one another in some species (e.g., Mancalla californiensis). In other species (e.g., Miomancalla wetmorei) these borders converge proximally, giving this scar a more triangular shape.
122. Humerus, mancalline scar on posterior side of proximal humerus, proximal extension relative to the fi rst pneumatic fossa: (0) extends within the fi rst pneumatic fossa; (1) scar terminates near the distal margin of the fi rst pneumatic fossa. Th e proximal extent of this scar varies from a condition in which the scar extends well within the fi rst pneumatic fossa (e.g., Mancalla californiensis) to a condition in which this scar terminates near the distal margin of the fi rst pneumatic fossa (e.g., Miomancalla wetmorei).
123. Humerus, mancalline scar on posterior side of proximal humerus, shape: (0) ridges parallel; (1) ridges converge proximally. Th e dorsal and ventral borders of the scar on the posterior side of the proximal humerus of Mancalla extend parallel to one another in some species (e.g., Mancalla californiensis). In other species (e.g., Miomancalla wetmorei) these borders converge proximally, giving this scar a more triangular shape.

Humerus, scapulohumeralis caudalis attachment scar, depth:
(0) fl at or slightly concave; (1) a deep pit. As noted by Olson and Rasmussen (2001), in alcids the attachment point of the scapulohumeralis caudalis muscle on the margin of the fi rst pneumatic fossa varies from a basically fl at or slightly concave surface (e.g., Alca torda) to a deeply excavated pit (e.g., Fratercula arctica).
130. Humerus, ridge between ventral tubercle and secondary pneumatic fossa: (0) absent; (1) present. On the posterior side of the humerus in Brachyramphus a slight ridge extends distally from underneath the distally overturned head of the humerus and contacts dorsal margin of the ventral tubercle, thus dividing the second pneumatic fossa from the capital groove.
131. Humerus, ventral tubercle, shape: (0) long and thin; (1) short and thick. In ventral view the ventral tubercle of some species of alcids (e.g., Fratercula arctica) is fairly thin and extends posteriorly to a point roughly level with the posterior extent of the caput. In other alcids (e.g., Alca torda) this feature does not extend as far posteriorly, and is more robust.
132. Humerus, ventral tubercle, lateral margin curvature: (0) single concavity; (1) double concavity. When viewed ventrally the lateral margin of the ventral tubercle of all alcid species other than Cerorhinca monocerata is a single concave curve. Th is feature in Cerorhinca monocerata is characterized by two concave curves. Th is character is the result of the crus ventrale fossae of Cerorhinca monocerata being divided into two sections.
133. Humerus, ventral tubercle, shape of posterior tip: (0) rounded or oval; (1) elongate. In Brachyramphus the posterior-most extension/point of the ventral tubercle is dorsally expanded into an elongate shape. In other alcids (e.g., Fratercula arctica) this feature is rounded or oval in shape.
134. Humerus, ventral tubercle, ventral margin curvature: (0) not deeply grooved; (1) deeply grooved. In anterior or posterior view the point at which the ven-tral tubercle and the ventral margin of the fi rst pneumatic fossa merge varies in its shape from ventrally convex or fl at (e.g., Fratercula corniculata) to ventrally concave (e.g., Pinguinus impennis).
136. Humerus, capital groove, anterior expression (modifi ed from Chandler 1990b, character 52): (0) curved; (1) notched; (2) deep groove. In anterior view the capital groove of most alcids (e.g., Alca torda) is visible as a notch on the lateromedial side of the humeral head. In the aukets (e.g., Ptychoramphus aleuticus) the capital groove is not expressed anteriorly, resulting in a convexly curved shaped lateromedial side of the humeral head. In the Mancallinae alcids (e.g., Mancalla cedrosensis) the capital groove communicates with the ligamental furrow, and is expressed as a deep groove in the ventral margin of the anterior humeral head. Ordered 137. Humerus, capital groove, width: (0) wide; (1) constricted. In all alcids (e.g., Fratercula arctica) except Mancalla the capital groove is an open 'U' shaped groove. Only in Mancalla does the caput overhang the capital groove, giving the proximal wall of the capital groove a convex shape, and constricting this passageway.
139. Humerus, orientation of head relative to shaft: (0) in line with shaft; (1) rotated anteriorly. As noted by Miller (1933), the humeral head of most alcids (e.g., Alca torda) is in-line with the shaft of the humerus, while the ventral portion of the humeral head of mancalline alcids (e.g., Mancalla cedrosensis) is rotated anteriorly.
144. Humerus, dorsal supracondylar process, length: (0) short; (1) long. Th e dorsal supracondylar process of most alcids (e.g., Aethia pygmaea) is short (i.e., the proximodistal length measured from the distal end of the humerus to the proximal termination of the crest on the humeral shaft is shorter than the greatest distal width of the humerus measured from the entepicondyle to the dorsal condyle). Th e dorsal supracondylar process of some alcids (e.g., Mancalla lucasi) extends further proximally onto the humeral shaft.
145. Humerus, dorsal sulcus: (0) continuous; (1) divided. Th e sulcus for passage of extensor metacarpi radialis, which runs between the dorsal supracondylar process and the dorsal condyle is continuous in all alcids (e.g., Pinguinus impennis) except the Fraterculini (e.g., Fratercula arctica), in which this sulcus is divided by a bony crest, forming a round pit on the posterior edge of the dorsal condyle.
146. Humerus, ventral epicondyle, orientation relative to shaft: (0) fl ared ventrally; (1) nearly straight. As noted by Olson and Rasmussen (2001), in anterior view the ventral margin of the ventral epicondyle is fl ared ventrally in Fratercula arctica, but is nearly straight in Cepphus grylle.
150. Humerus, tricipital fossae: (0) absent; (1) present. Th e scapulotricipital and humerotricipital sulci of Mancallinae are characterized by fossae positioned at the proximal end of the sulci. Th e sulci of other alcids transition smoothly onto the posterior face of the humeral shaft. (Chandler 1990b, character 61): (0) level; (1) distal extent of dorsal condyle proximal to distal extent of ventral condyle. Th e dorsal condyles of all extant alcids (e.g., Alca torda) are situated slightly proximal to the ventral condyle. Th e condyles of most other charadriiforms (e.g.,Gygis alba) extend distally an equal distance.

Humerus, relative distal extension of condyles
152. Humerus, ventral condyle in distal view, posterior trochlear process: (0) absent; (1) present. As noted by Marsh (1870) in the original description of Cataractes antiquus a posterodorsally-projecting tubercle is present on the ventral condyle; projecting into the sulcus between the ventral condyle and the saddle that defi nes the distal extent of the humerotricipital sulcus. Th is characteristic is also present in Pinguinus, but is lacking in all other alcids (e.g., Alca torda). Th is character has also been noted in penguins and plotopterids (Marples 1952;Ksepka et al. 2006).
154. Humerus, separation of humeral condyles: (0) absent; (1) present. In distal view the humeral condyles of Brachyramphus are separated, whereas the ventral margin of the dorsal condyle and the dorsal margin of the ventral condyle of other alcids (e.g., Synthliboramphus antiquus) contact one another.
156. Humerus, tubercles dorsal to scapulotricipital groove: (0) absent; (1) present. Many alcids (e.g., Brachyramphus perdix) possess a tubercle along the dorsal border of the scapulotricipital sulcus. In alcids this tubercle is located distal to paired fossae that lye between the raised dorsal margin of the scapulotricpital sulcus and the dorsal sulcus.
160. Humerus, position of pit adjacent to anterior ligament scar: (0) proximal; (1) ventral; (2) detached. Th e position of the small pit, which marks the origination point of the M. pronator sublimis varies in its position. In some species (e.g., Aethia pygmaea) this feature is located at the proximal tip of the anterior ligament scar, while in other species of alcids (e.g., Aethia psittacula) it is located along the dorsal margin of this scar. In some other charadriiforms (e.g., Phaetusa simplex), this scar is detached from the anterior ligament scar.
164. Radius, sulcus tendinosus (Chu 1998, character 102): (0) not divided; (1) divided lengthwise by a crest. Th e tendinal groove located on the dorsal side of the distal radius is divided by a crest in some species of alcids (e.g., Synthliboramphus antiquus). Th is feature is lacking in the charadriiform outgroup taxa examined and also in the auklets (e.g., Aethia pusilla).
165. Radius, notch in distal end: (0) absent; (1) present. In anterior view, the crest associated with the scapho-lunar facet of some alcids (e.g., Aethia pygmaea) extends far enough distally so that a notch is formed between that crest and the ventralmost articulation surface of the distal radius with the radiale. In other alcids (e.g., Alle alle) the crest on the anterior surface of the radius transitions smoothly into the distal end of the radius.
169. Ulna, crest extending from the ventral cotyla to the anterior margin of the ventral collateral ligament tubercle (modifi ed from Chandler 1990b, character 63): (0) absent; (1) present. Th e ventral cotyla of the proximal ulna is separate from the scar for the attachment of the ventral collateral ligament in some species of alcids (e.g., Cepphus grylle). In other alcids (e.g., Alle alle) a crest extends laterally from the ventral cotyla and contacts the anterior margin of the collateral ligament scar.
170. Ulna, crest extending from the ventral cotyla to the posterior margin of the ventral collateral ligament tubercle: (0) absent; (1) present. Although most alcids (e.g., Alca torda) lack a crest, which extends from the ventral cotyla to contact the posterior margin of the ventral collateral ligament scar, several alcids (e.g., Brachyramphus brevirostris) possess this character.
171. Ulna, dorsal cotylar process, anterior margin shape: (0) rounded; (1) straight. Th e dorsal condyle of alcids is bordered on the posterior margin by a posteriorly projecting bladelike process for attachment of the scapulotriceps muscle. Th e anterior margin of this feature in dorsal view can be either rounded (e.g., Alca torda) or straight (e.g., Fratercula cirrhata).
173. Ulna, proximal radial depression, shape: (0) a round pit; (1) a triangular pit; (2) broad and fl at. In contrast to the distinctly triangular shape of the proximal radial depression of most charadriiforms (e.g., Larus marinus), the proximal radial depression of all extant alcids (e.g., Uria aalge) is a round pit situated distal to the ulnar cotylae. In some charadriiforms not closely related to Alcidae (e.g., Bartramia longicauda) the radial depression is broad and fl at. In Miomancalla wetmorei the proximal radial depression is a broad fl at space bordered dorsally and ventrally by distinct crests that occupies the entire anterior surface of the ulna (Howard 1982). Ordered 174. Ulna, brachial impression, breadth: (0) thin; (1) broad. As noted by Howard (1982) the brachial impression on the proximal ulna of some alcids (e.g., Alca torda) is a relatively thin scar (i.e., does not comprise more than half the width of the ulnar shaft), while in some species (e.g., Fratercula arctica) this feature is broader (i.e., comprises more than half of the width of this proximal portion of the ulna).
175. Ulna, intramuscular line: (0) non-distinct; (1) distinct, raised ridge. As noted by Olson (1981), an inter-muscular line runs between the proximal radial depression and the nutrient foramen. Th is inter-muscular line is non-distinct and often barely visible in many species of alcids (e.g., Alle alle), while in others this feature is distinct and raised (e.g., Pinguinus impennis).
178. Ulna, carpal tubercle, shape: (0) fl at or angled distally; (1) concave. Th e distal margin of the carpal tubercle of some alcids (e.g., Alca torda) is fl at or angles slightly distally in some specimens. In Cepphus this surface is concave, giving the distal surface of the carpal tubercle a hooked appearance.
188. Carpometacarpus, pisiform process, development: (0) distinct; (1) reduced. Th e Pisiform process of most alcids (e.g., Miomancalla wetmorei) is a distinct ventral projection. Th e Pisiform process of Mancalla cedrosensis is reduced to a small scar. Similar to the condition observed in penguins, the reduction of this feature in Mancalla may be related to the stiff ening of the wing that is associated with the lack of these highly specialized wing propelled divers need to fl ex the manus.
192. Manual digit II, phalanx 1, shape of process on dorsal surface of the distal end: (0) rounded; (1) rectangular. A bladelike process projects posteriorly from the distal end of the fi rst phalange of the second digit. In dorsal view this process varies from rounded (e.g., Aethia pusilla) to rectangular (e.g., Alca torda).
193. Manual digit II, phalanx 1, length: (0) <1/2 length of carpometacarpus; (1) >1/2 length of carpometacarpus. Th e greatest length of the major phalanx of some charadriiforms (e.g., Sterna maxima) is >1/2 the greatest length of the carpometacarpus. Th e length of the major phalanx is <1/2 the length of the carpometacarpus in all alcids except Pinguinus impennis, in which the relative length of the wing elements has been reduced in association with fl ightlessness.
198. Iliosynsacral suture: (0) fused; (1) perforated. Th e contact between the lateral processes of the sacral vertebrae and the ilium, termed the iliosynsacral suture (sutura iliosynsacralis, Baumel and Witmer 1993), is fused along its entire margin in some alcids (e.g., Cepphus columba), while in other species (e.g., Alca torda) this suture is non-continuous (i.e., perforated by spaces between the lateral processes of the sacral vertebrae). Th is feature is distinct from the foramina intertransversaria of Baumel and Witmer (1993), which are located medially to the iliosynsacral suture.
202. Ischium, relative length of ischial angle and posterior projection (Strauch 1985, character 16;Chandler 1990b, character 46): (0) ischial angle much longer; (1) both structures about the same length. In the Lari and most other charadriiforms (e.g., Alca torda) the ischial angle is much longer than the posterior projection of the ilium; in the auklets (e.g., Aethia pusilla) the length of the ischial angle is much reduced, and the structures are almost the same length. Th ese diff erences also are indicated by Storer's (1945a) measurements of alcid skeletons (Strauch 1985).
205. Femur, trochanteric ridge, length: (0) long; (1) short. As noted by Miller (1937) the trochanteric crest of the femur varies in the extent to which it extends distally down the lateral shaft of the proximal femur from short (i.e., extends distally <2× width of the lateral surface of femoral head; e.g., Synthliboramphus hypoleucus) to long (i.e., extends distally >=2× width of the lateral surface of femoral head; e.g., Alca torda).
210. Tibiotarsus, lateral projection of crest lateral to the groove for peroneus profundus tendon, posterior view: (0) a distinct projection; (1) not visible in dorsal view. Th e lateral edge of the groove for the peroneus profundus tendon projects far enough laterally in some species of alcids (e.g., Alca torda) to be visible in posterior view.
212. Tibiotarsus, length: (0) <2× greatest length of tarsometatarsus; (1) >2× greatest length of tarsometatarsus. Th e greatest length of the tibiotarsus is greater than two times the greatest length of the tarsometatarsus in most alcids (e.g., Alca torda), but in some species of alcids (e.g., Synthliboramphus antiquus) the tibiotarsus is less than twice the length of the tarsometatarsus. (1) bony canal. Th e pattern of the canals in the hypotarsus of charadriiforms is discussed by Strauch (1978). In most charadriiforms canal No. 1 is a bony canal; in the Lari it is either a bony canal or a deep channel. In the Alcidae it may be a bony canal (e.g., Aethia pusilla), or a deep channel (e.g., Alca torda). Th e bony canal in charadriiforms is hypothesized to be primitive (Strauch 1978). More open canals in the hypotarsus have been linked with greater specialization and probably represent derived states (Harrison 1976in Strauch 1985 (1) mostly or completely enclosed bony channel. In the taxa examined, the third tendinal canal of the hypotarsus varies from a shallow groove (e.g., Alca torda) to a partially or fully enclosed bony canal (e.g., Cepphus grylle). In charadriiforms, the tendinal canal No. 3 provides passage for m. fl exor hallicus longus (Strauch 1978).
Integument 224. Maxillary rhamphotheca, color of tip (modifi ed from Chandler 1990b, character 95): (0) black or very dark brown; (1) red, orange or yellow; (2) white. In some alcids the tip of the beak varies in color from the rest of the rhamphothecum. Th e distal tip of the maxillary rhamphothecum varies in color from black or very dark brown (e.g., Alca torda), to shades of red, orange and yellow (e.g., Aethia psittacula), to white or light colored (e.g., Aethia pusilla).
227. Maxillary rhamphotheca, horn at base of maxilla: (0) absent; (1) present. A dorsally projecting horn is present on the base of the posterior maxillary rhamphotheca of Cerorhinca monocerata and Aethia pusilla. Th is feature is absent in all other alcids (e.g., Alle alle) and the nearest outgroups to Alcidae.

Nostril feathering
233. Eye scales (Strauch 1985, character 24;Chandler 1990b, character 90): (0) absent; (1) present. Th e Lari and most other charadriiforms have no eye scales (e.g., Aethia psittacula); they are present in some puffi ns (e.g., Fratercula corniculata;Strauch, 1985). Th ese dermal structures, although they change color during the breeding season and are undoubtedly used for mating display purposes, are also present year round and even on nestlings. Th is suggests that these 'horns' may have another purpose, possibly hydrodynamic. Th e evolution of a hardened horn solely for mating purposes seems unlikely given that a simple mating display could be achieved via feather coloration.
245. Juvenile plumage: (0) resembles winter adults; (1) resembles summer adults. Th e juvenile plumage of all acids (e.g., Uria lomvia) except Alca torda and Alle alle resembles the winter plumage of adults, in which the juvenile plumage resembles the summer plumage of adults (Kozlova 1957).
253. White face color in breeding plumage (Chandler 1990b, character 91): (0) absent; (1) present. During the breeding season the face color of some alcids (e.g., Fratercula arctica) changes to white in color. Th e face color of most alcids (e.g., Alca torda) is not white during the reproductive phase.
265. Color of downy chicks: (0) variable; (1) primarily brown; (2) primarily black; (3) primarily grey; (4) primarily buff or white. Th e down feathering of charadriiform chicks is predictably colored in most species (e.g., black in Cepphus grylle), although the color of the down feathers in some terns (e.g., Sterna maxima) is variable (i.e., sometimes black, sometimes buff  (Sealy 1973). Th e pattern for Pinguinus is unknown. Bengtson (1984), in a review of the literature on Pinguinus, estimated that chicks leave the nest at about 10 days old, which would agree with an intermediate pattern. In the Lari the pattern is semi-precocial; it is hypothesized that shortening of the nestling period in alcids represents a derived condition (Strauch 1985). Ordered Diet 267. Adult prey preference: (0) primarily invertebrates; (1) primarily vertebrates; (2) signifi cant amounts of invertebrates and vertebrates. Many of the smaller alcids (e.g., Alle alle) are specialized feeders on small invertebrates, while some larger alcids (i.e., Fratercula arctica) subsist on a diet of mostly fi sh (del Hoyo et al. 1996).

Rejected characters
Th e following characters from the dataset of Chandler (1990) were rejected due to intraspecies variability: 15, 32, 48, 55, 72, 73, 77, 78; or because they were parsimony uninformative (i.e., they did not vary among taxa examined): 27, 31, 34, 45, 54, 56, 57, 66, 68, 76. Th e following characters of Dove (2000) were not included the matrix because they did not vary in any taxa examined in this study: 9, 10, 28, 29. Several characters of Dove (2000) were split into two separate characters following the philosophy of character independence with respect to absence of character states outlined by Hawkins et al. (1997).
All the characters of Strauch (1985) and Chandler (1990) were rescored for this analysis using multiple specimens (see Appendix 1,comparative material). Many of the characters of Strauch (1985) and Chandler (1990) were modifi ed to describe variability not originally noted by those authors (see notations in character list).

Appendix 6. Geologic setting
Mancallinae material described herein comes from four Miocene and Pliocene aged marine deposits (Domning and Deméré 1984;Ingle 1979;Wagner et al. 2001). Congruent with the habitat of extant alcids (del Hoyo et al 1996), three of these deposits (San Mateo Formation, Niguel Formation, San Diego Formation) are interpreted as the result of shallow to moderate depth marine facies (Vedder 1960;Kern and Wicander 1974;Vedder 1972;Ingle 1979;Wagner et al. 2001) associated with cold-water upwelling ocean systems. Th e upper siltstone facies of the Capistrano Formation, from which Mancallinae fossils have been recovered, contains transported remains of neritic mollusks and microfossils that are mixed with the remains of bathyal species (Kern and Wicander 1974), suggesting a shallow water origin for Mancallinae fossils from the Capistrano Formation. As with other vertebrate fossil assemblages from nutrient-rich cold-water systems (e.g., Pliocene Yorktown Formation assemblage; Ray 1987; Ray and Bohaska 2001), a diverse assemblage of vertebrates including marine mammals and seabirds are documented from marine deposits such as the Pliocene San Diego formation in southern California (Barnes et al. 1981).

San Mateo Formation:
Th e San Mateo Formation is composed of sandstones, siltstones, and conglomerates that interfi nger with the latest Miocene and earliest Pliocene aged member of the Capistrano Formation (Tan and Kennedy 1996), and is interpreted as the result of shallow marine deposition (Vedder 1972). Th e San Mateo Formation is exposed in natural and quarry exposures near Lawrence Canyon in San Diego County, California, and has yielded two distinct vertebrate assemblages including sharks, fi sh, birds, and marine and terrestrial mammals (Barnes et al. 1981;Domning and Deméré 1984;Howard 1982).
Th e vertebrate assemblages of the San Mateo Fm. were discussed by Barnes et al. (1981), who designated the lower assemblage the San Luis Rey River Local Fauna (SLRRLF), and the upper assemblage the Lawrence Canyon Local Fauna (LCLF). Based on marine vertebrates and terrestrial mammals, the age of the younger LCLF has been proposed to be latest Miocene or earliest Pliocene (~5.0 Ma), and correlative with the Late Hemphillian North American Land Mammal Age (NALMA; Domning and Deméré 1984). Mancallinae fossils, including the humerus (SDSNH 24584) referred to M. howardi, have been recovered from the older SLRRLF. Age estimates for the SLRRLF based upon terrestrial mammal and marine bird fossils range from approximately 6.7-10.0 Ma (i.e., Late Miocene or Turtonian equivalent; Barnes et al. 1981, Domning andDeméré 1984).
Capistrano Formation: Th e Capistrano Fm. is composed of sandstones and siltstones that have been correlated with upper portions of the San Mateo Fm. in northern San Diego County (Elliot 1975;Domning and Deméré 1984), and interpreted as the result of marine deep-sea fan deposition on the basis of microfaunal analysis and abundant turbidites (Ingle 1979;Vedder 1972). Th e Capistrano Fm. spans the