Penguins are wing-propelled diving homeotherms that hunt in austral oceans. Penguins will forage in waters as warm as 28°C (Wellington et al., 2001) and as cold as −0.6°C (Stonehouse, 1967), while maintaining a core body temperature of 38.5°C (Prinzinger et al., 1991). Adult penguins weigh as much as 46 kg (prefasting maximum) or as little as 1 kg and hunt between 0° and 78° angle south latitude (Stonehouse, 1975). Hot-blooded penguins can forage successfully in water cooler than core body temperature because of a suite of thermal adaptations. Penguins may maintain subdermal fat stores (Cherel et al., 1994), which retards the transmission of heat through to the dermis. The hydrodynamic requirements for marine foraging have shaped penguins into relatively large and rotund birds, reducing the surface area (relative to volume) through which heat may be lost. The feathers of penguins are modified into short, bladed structures that tightly overlap to insulate the epidermis from direct cold water exposure (Dawson et al., 1999). Penguins also rely on specialized arteriovenous mechanisms for heat retention (e.g., Scholander, 1955; Trawa, 1970). Three sets of arteriovenous heat retaining structures have been described in penguins, located in the neck, wings, and legs (Frost et al., 1975). Wing-borne arteriovenous systems have been the focus of several anatomical studies (e.g., Watson, 1883; Trawa, 1970), with increasing focus toward thermoregulation (e.g., Frost et al., 1975; Thomas and Fordyce, 2007), whereas neck and leg systems await the benefits of a comparative study (Frost et al., 1975).
Underwater flight has shaped penguin wings into thin blades with comparatively high surface areas (e.g., Stonehouse, 1967). The wings of penguins have consequently become conduits for heat loss and are well-supplied with arteries despite limited musculature beyond the shoulder. Wings perform a dual-purpose from the perspective of thermoregulation: wings are heat shedders during bouts of high exercise and heat conservers when facing hypothermia. Heat conservation is achieved by counter-current flow across a vascular network, where outgoing arterial blood warms incoming venous blood, with heat returned to the body core instead of continuing out to the periphery (e.g., Scholander and Schevill, 1955). Trawa (1970) first recognized the adaptive significance of closely associated veins and arteries in the wings of penguins and proposed that the “humeral arterial plexus” (sensu Frost et al., 1975) was responsible for a 25°C temperature gradient between the shoulder and wingtip of emperor penguins (Prévost and Sapin-Jaloustre, 1964). Humeral arterial plexi have now been described for five species of penguins in four crown genera: Adélie (Pygoscelis adeliae Hombron and Jacquinot, 1841), African (Spheniscus demersus Linnaeus, 1758), emperor (Aptenodytes forsteri Gray, 1844), king (Aptenodytes patagonicus Miller, 1778), and little blue (Eudyptula minor Forster, 1781) (Watson, 1883; Trawa, 1970; Frost et al., 1975; Louw, 1992; Thomas and Fordyce, 2007). Further, plexi have been observed in two other species (rockhopper, Eudyptes chrysocome Forster, 1781; yellow-eyed, Megadyptes antipodes Hombron and Jacquinot, 1841) but until now have not been anatomically described (general comment on wing anatomy was provided by Filhol, 1882). The humeral arterial plexus is plesiomorphic in crown penguins: an osteological correlate of the plexus (humeral arterial sulcus) has been found in fossil stem penguins (Thomas et al., 2011). Thus, all living genera of penguins have humeral arterial plexi with fossil evidence supporting evolution of the plexus at least 49 million years ago (Thomas et al., 2011).
The number of arteries forming the penguin humeral arterial plexus varies among species. Emperor penguin plexi feature 12–15 arteries, whereas king penguins have 7–8 arteries, Adélie penguins have five arteries, African penguins have three arteries, and little blue penguins have two arteries. Increasing the number of arteries would increase the surface area available for heat exchange and create a more efficient heat-retention mechanism. The number of plexus arteries increases with increasing body size (Stonehouse, 1967), paradoxically implying that larger penguins with better insulation require more efficient counter-current heat exchange systems. Larger penguins have relatively larger wings (Stonehouse, 1967), however, and hence have larger conduits for heat to be lost to the ocean. Alternatively, humeral artery number may reflect ancestral physiology independent of environmental adaptation, given that the larger emperor and king penguins tend to occupy extreme branches in crown penguin phylogenies (i.e., apical or basal in phylogenetic trees; Baker et al., 2006; Ksepka et al., 2006). If the humeral arterial plexus is an adaptation to thermal regime, then we would anticipate that the number of plexus arteries in a penguin species would correlate closely with body or flipper size irrespective of phylogenetic position. Humeral arterial plexi have previously been described for only four of the six extant penguin genera; here, we describe the plexus in the remaining two genera, and thus complete the survey of the plexus for the penguin crown group. We studied the humeral arterial plexi of yellow-eyed and erect-crested (Eudyptes sclateri Buller, 1888) penguins to characterize the relationship between the number of plexus arteries, penguin size, and phylogeny. Future studies should consider arteriovenous structures in the neck and feet as well; these systems have been reported only for S. demersus: interspecific comparisons of counter-current heat exchange systems across all living penguin genera are only possible for wing vasculature at this time.
All specimens dissected in this study died naturally and were found as carcasses in the wild. Dissections of yellow-eyed penguins (M. antipodes) were conducted on a desiccated specimen (OU21985, 55 cm from bill tip to end of tail length) recovered as a beach-stranded carcass in the 1980s from the Otago Peninsula, New Zealand by C.M. Jones (before taonga status was imparted by the Ngai Tahu Claims Settlement Act, 1998) and a frozen specimen (OM1645, 54 cm bill tip to end of tail length) loaned for dissection from Otago Museum. A previously dissected M. antipodes wing [specimen TA46, from PhD studies by T. Ando, Department of Geology, University of Otago; material supplied by Dr A.G. Hocken of Oamaru after necropsy for Department of Conservation (DOC) Oamaru] was used as a guide for the dissections. A frozen specimen of an erect-crested penguin (E. sclateri; OM891, 49 cm from bill tip to end of tail length), also provided by Otago Museum, was examined. The erect-crested penguin carcass was collected from New Brighton, New Zealand and stored at Otago Museum since 24 April 1987. Permission to dissect the Otago Museum specimens (OM891 and OM1645) was arranged with the Dunedin DOC office and Otago Museum. The rarity of naturally dead yellow-eyed and erect-crested penguins restricted the availability of specimens for study, and the rarity and condition of material precluded histological studies or vascular casting to support macroanatomical observations.
Vessel topology was directly observed using a Zeiss Stemi 2000-C binocular microscope. Vessels were identified as arteries by tracing their direct connection with the axillary artery, which was initially identified using injected dye. Veins were identified either as vessels connected to the marginal vein or as vessels associated with but not connected to the branches of the axillary artery. Vessel identification was made using the anatomical studies of Watson (1883), Trawa (1970), Frost et al. (1975), and Louw (1992) and described according to the orientations as depicted in Fig. 1, and the osteological characteristics were depicted in Fig. 2. Note that ventral surface here refers to the underside of the penguin while swimming (white feathered surface), the equivalent of “anterior” in Ksepka et al. (2006). Vessel names follow Frost et al. (1975) and Baumel (1993) unless otherwise indicated.
Arteries of the ventral surface
The axillary artery (A. axillaris) of M. antipodes is the single common artery that supplies blood to the entire limb of the yellow-eyed penguin (Fig. 3). The segment of axillary artery present within the penguin limb falls short of the humerus; instead, the axillary artery diverges into five smaller arteries within the subdermal tissues of the shoulder joint. The first, most proximal branching produces the profunda artery (A. profunda brachii), the arterial source to the dorsal surface of the wing. The second branch produces the marginal artery (e.g., Frost et al., 1975), which extends along the posterior edge of the humerus, crosses the elbow against the trochlear process, sits in against the posterior border of the ulna, and continues on to the ventral surface of the ulnare (where it was too small to be traced).
The remaining three divisions of A. axillaris, in order of increasing distance from the shoulder, produce the anterior, central, and posterior humeral arteries of the humeral plexus. All three humeral arteries approach the posterior edge of the humerus toward the uppermost third of the diaphysis. The anterior and medial humeral arteries traverse the ventral surface of the diaphysis in proximity and pass over or around the radial condyle toward the proximal end of the radius. A smaller artery (A. anterior anastomotic) branches off from the anterior humeral artery to supply the brachial muscle (M. brachialis). The posterior humeral artery traverses the center of the ventral face of the humerus, passes between the radial and ulnar condyles, and continues toward the anterior edge of the ulna. The medial and posterior humeral arteries anastomose at the proximal end of the radioulnar interosseous space becoming the cubital artery (A. cubiti, e.g., Frost et al., 1975; Thomas and Fordyce, 2007). The anterior and posterior humeral arteries enter the radioulnar interosseous space, where the former becomes the primitive radial artery (A. radialis primitiva) and the latter becomes the cubital artery (A. cubiti; e.g., Frost et al., 1975; Thomas and Fordyce, 2007). The cubital artery lies embedded within tissue along most of the length of the radioulnar interosseous space but emerges toward the distal end of the ulna where it crosses the ulnare and approaches the posterior edge of the os metacarpale minus element of the carpometacarpus. The primitive radial artery remains in proximity to the cubital artery within the forearm and terminates into the radius-radiale joint.
The cubital artery diverges into two vessels within the wrist of the yellow-eyed penguin: the more posterior carpometacarpal artery and the more anterior carpal artery (A. carpometacarpale and A. carpale, respectively: Frost et al., 1975). The carpometacarpal artery runs the length of the intermetacarpal foramen into which it terminates. The carpal artery extends toward the anterior edge of the carpometacarpus and crosses onto the proximal phalanx of the second digit, before continuing on to, and terminating at the distal end of, the distal phalanx of the second digit.
Veins of the ventral surface
Veins were thin walled and less turgid than arteries (Fig. 3). A single, large vein approaches the wing amidst shoulder muscles and branches into several smaller vessels. These smaller veins approach the posterior edge of the humerus surrounded by a cluster of arteries. The largest of the veins entering the wing travels posterior to the full length of the humerus and ulna and is equivalent to the marginal vein observed in other penguins (e.g., Frost et al., 1975). The smaller veins remain in direct contact with the entire arterial network: arteries of the humeral plexus contact at least two veins for the entire distance, whereas posthumeral arteries are associated with only a single vein each.
Arteries and veins of the dorsal surface
Vessels along the dorsal surface of the yellow-eyed penguin wing were indistinct and consequently difficult to detail using standard dissection techniques (Fig. 3). The profunda artery (A. profunda brachii) traverses the diaphysis along the posterior edge of the humerus and then diverges. The more anterior artery from the division of the profunda artery trends toward the anterior edge of the humerus and crosses on to the radius to become the radial collateral artery (equivalent to A. collateralis radialis of Baumel, 1993), but could not be traced after entering the radioulnar interosseous space. The more posterior division of the profunda artery continues toward the distal end of the humerus, but could not be traced beyond the proximal end of the ulna (equivalent to A. collateralis ulnaris of Baumel, 1993). The ulnar collateral artery gives rise to a smaller artery, which supplies tissues associated with the trochlear process. Veins were observed in association with each collateral artery and along the proximal extent of the profunda artery and are assumed to be associated with the entire arterial network of the dorsal surface (but these veins were too small to be traced).
Arteries and veins of the ventral surface
The vascular architecture of the ventral surface of the erect-crested penguin (E. sclateri) is essentially identical to that of the yellow-eyed penguin (Fig. 4). The only notable difference between the two species is the arrangement of the humeral arteries. As with the yellow-eyed penguin, the anterior humeral artery approaches the posterior edge of the humerus and continues along the ventral face before crossing the distal end of the humerus and entering the radioulnar interosseous space. The medial artery follows a similar trajectory but does not contact the anterior artery as in M. antipodes. In addition, the medial humeral artery anastomoses with the posterior humeral artery to form the cubital artery, rather than anastomosing with the anterior artery to form the primitive radial artery (cf.M. antipodes). Furthermore, the posterior humeral and medial humeral arteries are directly associated, whereas the posterior humeral artery of the yellow-eyed penguin is isolated from the rest of the plexus. Apart from differences within the plexus itself, no other significant differences were observed between the vascular anatomies of the yellow-eyed and crested penguins.
Arteries and veins of the dorsal surface
As with the ventral surface, the vasculature of the E. sclateri dorsal surface mimics that of the yellow-eyed penguin, although more detail could be seen in the erect-crested penguin dissection (Fig. 4). Like penguins described elsewhere (e.g., Frost et al., 1975; Thomas and Fordyce, 2007), the profunda artery approaches the posterior edge of the humerus in the more-proximal half of the diaphysis, continues across the face of the humerus to the more distal portion of the diaphysis, and diverges to form two vessels: A. collateralis radialis and A. collateralis ulnaris. The more anterior radial collateral artery trends toward M. brachialis, where it provides a small offshoot vessel to supply the muscle, before traversing over to the proximal end of the radius and entering the radioulnar interosseous space. The more posterior divergence of the profunda artery, equivalent to the ulnar collateral artery of Baumel (1993), continues toward the trochlear process before approaching the posterior edge of the ulna. Two smaller arteries diverge from the ulnar collateral artery: a more proximal split supplies the tissues associated with the trochlear process, with a more distal split feeding into the humero-ulnar depression. Arteries were not traced beyond these terminations, and veins were found in association with every artery on the dorsal surface.
The wings of yellow-eyed (M. antipodes) and erect-crested (E. sclateri) penguins each contain an axillary artery divided into three branches (humeral plexus), where each branch is associated with at least two veins. There was no variation in vascular topology observed between the two yellow-eyed penguin specimens. The plexus topologies of each penguin differed, but the preplexus and post-plexus topologies were essentially identical. The common arterial source to the wing of each penguin is A. axillaries, with A. brachialis replaced by a humeral plexus. The first division of A. axillaris is always A. profunda brachii, which supplies the dorsal surface of the wing. Vessels were traced on both the ventral and dorsal surfaces of each penguin wing, and we noted that the ventral surfaces were extensively vascularized compared with the dorsal surfaces. The reduced vascularization of the dorsal surface may benefit heat retention on land or reduce the severity of any laceration, as the extensively vascularized ventral surface can be pressed against the body (T. Ando, personal communication, 2011).
Origin of Humeral Plexus Variation
The number of humeral arteries ranges from two in little blue penguins to 15 in emperor penguins and includes some intraspecific variation (Trawa, 1970; Frost et al., 1975; Louw, 1992; Thomas and Fordyce, 2007). Heat is transferred where humeral arteries contact adjacent veins (Frost et al., 1975); increasing vascular contact area by increasing the number of arteries would elevate the efficiency of heat transfer. Heat-retention efficiency should therefore differ among penguins and should scale to compensate for heat lost through flippers (e.g., Chappell et al., 1989). The hydrodynamic function of flippers requires a svelte profile (Hui, 1988), and larger penguins need larger flippers for optimal stroke rates (albeit with negative allometry; Sato et al., 2010). Increased heat-retention efficiency is therefore expected for penguins with larger flippers to compensate for the increased surface area exposed to cold water. Indeed, flipper surface area correlates well with the number of humeral arteries (exponential model, r2 = 0.91, P-value < 0.001; Fig. 5), and the correlation improves substantially when the Adélie penguin is removed (r2 = 0.99, P-value < 0.001). The number of plexus arteries correlates well with body size (exponential model, r2 = 0.88, P-value = 0.004) but less well with sea surface temperature at breeding sites (exponential model, r2 = 0.68, P-value = 0.022; Fig. 5). Again, the Adélie penguin is an outlier, and correlations with body mass and sea surface temperature improve with its removal (exponential model, r2 = 0.95, P-value = 0.001 and r2 = 0.84, P-value = 0.010). Humeral arteries may also be heat shedders. Larger birds burdened with lower surface to volume ratios may benefit from plexi that can broadly distribute blood close to the wing surface. The origination of humeral arterial plexus has been proposed as an adaptation for foraging in deeper, cooler water (Thomas et al., 2011). If the plexus is used for both heat retention and heat loss (e.g., Frost et al., 1975), then we would anticipate that the latter function is a secondary use of a structure primarily used to keep penguins warm.
Adélie penguins are logical outliers when both flipper surface area and water temperature are considered in parallel. Adélie penguins are comparatively small Antarctic endotherms and have a wing surface area that is typical for a 5 kg or 70 cm long penguin (e.g., proportional to penguins in warmer waters; Stonehouse, 1967). The large gradient between body core (38.5°C; Prinzinger et al., 1991) and ocean water (−0.6°C; Stonehouse, 1967) temperatures requires Adélie penguins to have more efficient heat retention than equivalent sized penguins (e.g., yellow-eyed, erect-crested) that live at more temperate latitudes; such attributes may explain why Adélie penguins have comparatively more humeral arteries. Conversely, Adélie penguins have much smaller flippers than emperor penguins (Stonehouse, 1967), providing a smaller conduit for heat loss. Adélie penguins consequently need fewer arteries than emperor penguins to compensate for the heat lost to Antarctic water (Trawa, 1970). Alternatively, the relatively high number of humeral arteries in Adélie penguins may reflect both ancestry and adaptation, given the potential close relationship between Pygoscelis and Aptenodytes (T. Ando, personal communication, 2011).
We interpret the number of humeral arteries as a biologically plastic response to the thermal demands on individual penguin species and observe dramatic variation across the penguin crown group. The most recent common ancestor (MRCA) to all extant penguins lived at least 11 million years ago (Ksepka and Clarke, 2010). Molecular evidence (Baker et al., 2006; Ksepka et al., 2006) suggests that Aptenodytes (e.g., emperor and king penguins) is the most basal crown genus (i.e., most closely related to the MRCA), whereas morphological evidence has instead identified Spheniscus + Eudyptula (e.g., African and little blue penguins) as the most basal crown genera (Bertelli and Giannini, 2005). Note that Ksepka et al. (2006) found morphological support for a basal divergence of Spheniscus + Eudyptula as well as support for a basal divergence of Aptenodytes in a combined morphological + molecular analysis: topological differences reflect rooting preferences of the morphological and molecular data (see discussions by Bertelli and Giannini, 2005; Ksepka and Ando, 2011). The identity of the most basal crown penguin has dramatic implications for the evolution of the humeral plexus. Given Aptenodytes as basal, then humeral arteries would have been lost in more apical crown genera, this would imply an evolutionary reversal, as humeral arteries were originally derived by splitting the single brachial artery as seen in most birds (Baumel, 1993). Conversely, a basal position for Spheniscus + Eudyptula would imply that sequential splits of the original humeral arteries occurred during the radiation of crown penguins. Irrespective of the exact crown clade topology, it is evident that the humeral arterial plexus is a dynamic structure that has undergone significant evolution to accommodate the wide range of thermal regimes experienced by modern penguins.
The authors thank Otago Museum for arranging permission and samples for this study. The authors also thank Dallas Hyde, Tatsuro Ando, and Dan Ksepka for constructive suggestions for improvements.