THE ANATOMICAL RECORD 298:291–313 (2015)
Exploring Eucladoceros Ecomorphology Using Geometric Morphometrics SABRINA C. CURRAN* Ohio University, Department of Sociology and Anthropology, Athens, OH 45701
ABSTRACT An increasingly common method for reconstructing paleoenvironmental parameters of hominin sites is ecological functional morphology (ecomorphology). This study provides a geometric morphometric study of cervid rearlimb morphology as it relates to phylogeny, size, and ecomorphology. These methods are then applied to an extinct Pleistocene cervid, Eucladoceros, which is found in some of the earliest hominin-occupied sites in Eurasia. Variation in cervid postcranial functional morphology associated with different habitats can be summarized as trade-offs between joint stability versus mobility and rapid movement versus power-generation. Cervids in open habitats emphasize limb stability to avoid joint dislocation during rapid flight from predators. Closed-adapted cervids require more joint mobility to rapidly switch directions in complex habitats. Two skeletal features (of the tibia and calcaneus) have significant phylogenetic signals, while two (the femur and third phalanx) do not. Additionally, morphology of two of these features (tibia and third phalanx) were correlated with body size. For the tibial analysis (but not the third phalanx) this correlation was ameliorated when phylogeny was taken into account. Eucladoceros specimens from France and Romania fall on the more open side of the habitat continuum, a result that is at odds with reconstructions of their diet as browsers, suggesting that they may have had a behavioral regime unlike any extant cervid. Anat Rec, C 2014 Wiley Periodicals, Inc. 298:291–313, 2015. V
Key words: cervidae; locomotion; phylogeny; femur; tibia; calcaneus; third phalanx
The Pleistocene was a time of mammalian megafauna and members of the deer family, Cervidae, were no exception. One of these giants, Eucladoceros or the comb-antlered deer, was widely distributed across Eurasia. Since it was contemporaneous with the earliest appearance of hominins outside of Africa, elucidation of Eucladoceros paleobiology informs our understanding the range of habitats those early hominins would have encountered. This study analyzes Eucladoceros functional morphology to reconstruct its locomotor behavior and habitat preferences. Morphology associated with locomotion is known to vary by habitat (Kappelman, 1988, 1991; Plummer and Bishop, 1994; Kappelman et al., 1997; Scott et al., 1999; DeGusta and Vrba, 2003, 2005; Weinand, 2007; Kovarovic and Andrews, 2007; Plummer et al., 2008) and as such, it can be used as a paleohabitat proxy. However, factors such as phylogenetic relatedness and body size are also known to influC 2014 WILEY PERIODICALS, INC. V
ence skeletal morphology. This study has two goals: (1) to extend the methods established in Curran (2012) to address issues of body size and phylogeny in extant
Grant sponsor: National Science Foundation; Grant number: BCS-0824607.; Grant sponsor: University of Minnesota Doctoral Dissertation Fellowship.; Grant sponsor: University of Minnesota, Herb E. Wright Fellowship in Paleoecology; Grant sponsor: University of Minnesota, Department of Anthropology Block Grants. *Correspondence to: Sabrina Curran, Ohio University, Department of Sociology and Anthropology, Athens OH 45701. E-mail: [email protected] Received 3 October 2014; Accepted 11 October 2014. DOI 10.1002/ar.23066 Published online in Wiley Online Library (wileyonlinelibrary. com).
292
CURRAN
Fig. 1. The four morphological features examined in this study, with their position in the cervid rearlimb (center). Each feature is demonstrated with a photograph of the region of interest and the Procrustesaligned coordinates for the specimens in this study (visualized with rola-Castillo, 2013). A: the Geomorph package for R, Adams, and Ota
Medial margin of the femoral patellar surface (right-side, Alces alces). B: Tibial lateral condyle margin (right-side, Alces alces). C: Calcaneus landmarks (right-side, Alces alces). D: Plantar margin of the third phalanx (side unknown, Odocoileus hemionus).
cervids and (2) apply these methods to Eucladoceros to better understand the locomotor behavior and ecological niche(s) of members of this extinct genus.
of ecomorphological studies such as the one presented here.
CERVID REARLIMB ECOMORPHOLOGY ECOMORPHOLOGY Ecological functional morphology, or ecomorphology, is the study of adaptation at its core (Wainwright, 1994; Plummer et al., 2008). What we are really asking is “how do different phenotypes affect fitness?” For this study, this question becomes “how do variants in skeletal morphology affect locomotion, and especially escape from predators, in the Cervidae?” In answering this question, we are also addressing the structure of the habitat, for variation in vegetative density and structure is one of the main influences on artiodactyl locomotion (Kappelman, 1988; K€ohler, 1993). An individual’s morphology places limits on the types of behaviors that can be performed and the environments in which those behaviors can take place (Wainwright, 1994). For example, a forest-dwelling cervid can often lose a predator by frequently changing direction through a series of powerful leaps, but this behavior would not be very successful in fleeing a cursorial predator in an open landscape (Kappelman, 1988). It is performance [defined by Wainwright (1994) as “an organism’s ability to carry out specific behaviors”] that forms the link between morphology and ecology. Since performance requirements vary between habitats, and morphology varies with performance requirements, we can then link morphology to habitats. This linkage between morphology and performance is the foundation
Just as bovid remains have been shown to be useful as proxies for paleoenvironments in Africa (Plummer and Bishop, 1994; Kappelman et al., 1997; Scott, 2004; Kovarovic and Andrews, 2007) and Southeast Asia (Weinand, 2007), so too are cervid remains useful for reconstructing the paleohabitats of Eurasian sites, where they are often the dominant taxa. Extant cervids inhabit a wide range of environments, from tundra and grasslands to swamps and forests (both temperate and tropical), making them an appropriate proxy for past habitats. Within these various habitats, cervids perform a variety of predator-escape behaviors. Cervids avoid predators mainly by being cryptic, especially in densely vegetated habitats (Bro-Jørgensen, 2008), although some largebodied forms (Alces alces, Cervus unicolor) will confront predators (Geist, 1998), while others (Rangifer tarandus, Cervus elaphus) depend on the “selfish herd” (Hamilton, 1971) for safety. In open environments, cervids depend on speedy flight and thus their morphological adaptations reflect rapid movement and joint stability. Cervids in closed environments must locomote through a more complex three-dimensional landscape and their morphology reflects the need for increased mobility. In this study, geometric morphometrics (GM) is used to quantify variation in four skeletal features from the rearlimb of Cervidae as it relates to locomotion in different habitats. While there are many components that influence the functional morphology of the cervid
EUCLADOCEROS ECOMORPHOLOGY
Fig. 2. Visualizations of medial femoral patellar margin morphology based on data presented in Curran (2012). A single canonical variate was created between representative specimens of these habitat extremes with a multivariate regression of Procrustes coordinates on canonical variates scores for the two-group CVA (McNulty et al., 2006). The coefficients of this regression were then treated as a vector that was added to and subtracted from the consensus landmark configuration (after Frost et al., 2003). The two resulting configurations demonstrate the differences between the habitat extremes as mod-
293
eled in Morpheus (Slice et al., 2002). A: Top 5 example of an openadapted cervid (Hydropotes inermis, AMNH 147434), middle 5 visualization of the mean open-adapted medial femoral patellar margin, and bottom 5 the ellipse formed by continuing the medial margin’s arc; and B: a Top 5 closed-adapted cervid (Pudu mephistophiles, MVZ 122523), middle 5 visualization of the mean closed-adapted medial femoral patellar margin, and the more circular shape formed by continuing the arc of the mean closed-adapted specimens. Scale bar 5 1 cm.
294
CURRAN
Fig. 3. Visualizations of lateral tibial plateau morphology based on data presented in Curran (2012). A: open-adapted morphology with an extant example (Ozotoceros bezoarticus, NMNH 270379) from superior view (top) and posterior view (bottom), with corresponding visualizations of the mean open-adapted lateral tibial plateau margin from the
superior view and the posterior view; B: open-adapted morphology with an extant example (Mazama rufina, MVZ 124095) from superior view (top) and posterior view (bottom) with corresponding visualization of the mean closed-adapted lateral tibial plateau margin from the superior and posterior view.
rearlimb, this study focuses on features involved in power generation and stabilization mechanisms. The four features (see Fig. 1) were selected to sample skeletal morphology important to locomotion: (1) the medial margin of the patellar surface of the femur is hypothesized to indicate torque generation at the knee (Kappelman, 1988); (2) the range and type of motion at the knee is analyzed here using the morphology of the tibia’s lateral condylar margin (Organ and Ward, 2006); (3) the rest angle and range of motion of the calcaneus is hypothesized to be related to the number of times the animal moves through its pace-cycle per minute (Curran, 2012); and (4) plantar morphology of the third phalanx (where the cervid contacts the substrate) reflects the role of the hoof in limb stabilization during locomotion (K€ohler, 1993). Examined in concert, these four features inform our understanding of cervid locomotion and rearlimb ecomorphology. Below I discuss previous studies regarding the functional morphology and biomechanics of these regions with a particular focus on Curran (2012), the study that introduced the GM methods used here, and of which this study is a direct extension.
center of rotation for the knee). Hermanson and MacFadden (1996) argued that it is unclear how this would provide mechanical advantage since the patella tracks the patellar fossa and not the medial ridge. However, Shockley (2001) points out that a portion of the patella does track over the medial ridge, raising the medial portion of the patella during extension. This may increase stability by increasing the tension on the medial patellar ligament and restricting movement at the joint. Further, Shockley (2001) states that an enlarged medial ridge would resist forces from adductors and prevents the patella from dislocating. Janis et al. (2012) suggest that the limb is positioned more laterally in ungulates during a gallop due to the need to avoid their large abdomen and thus increased medial stabilization is necessary. Here, the medial patellar margin is digitized with a series of three-dimensional coordinates along the most prominent ridge (Fig. 2). Prior analysis of extant cervids (Curran, 2012) demonstrated that forest-adapted (closedadapted) cervids have medial margins with a smooth, continuous arc (see Fig. 2B) while cursorial (non-forest) cervids have a distinct anterior projection (Fig. 2A). This is the exact morphology hypothesized by Kappelman (1988) wherein if one were to continue the arc of the medial patellar surface, closed-adapted cervids would appear more circular while cursorial cervids would be more elliptical for both margins (See Fig. 2, bottom images). For cursorial cervids, the anterior bulge creates a progressive increase in the distance of the patella to the center of rotation throughout limb extension (Kappelman, 1988). This, in combination with increased medial patellar ligament tension (Shockley, 2001; Janis, 2012), allows cursorial cervids to achieve speedy flight
Medial Femoral Patellar Margin As is the case for most artiodactyl knees, cervid femora have larger medial than lateral patellar margins. This asymmetry is more remarkable in larger and openadapted species (Janis et al., 2012). Kappelman (1988) hypothesized that a larger medial ridge serves to increase torque at the knee by increasing the length of the moment arm (moving the patella further from the
EUCLADOCEROS ECOMORPHOLOGY
295
Fig. 4. The hock (ankle) joint of an artiodactyl with each element labeled and the visible joint articular surfaces colored (green 5 astragalus-calcaneus, orange 5 os malleolus-calcaneus, and purple 5 cubonavicular-calcaneus).
with reduced risks of patellar dislocation. Interestingly, the contrast here is between forest-adapted (i.e., closed) deer and deer in all other habitat types (i.e., open and
intermediate habitat types) and there is not a continuum of variation like that found in other skeletal features discussed below (calcaneus, third phalanx). This suggests
296
CURRAN
Fig. 5. A: An example of a closed-adapted specimen (Pudu mephistophiles, MVZ 122523; medial view) with landmarks indicated; B: an example of an open-adapted specimen (Ozotoceros bezoarticus, MVZ 122523) with landmarks indicated; C: the aforementioned specimens overlapped in articular position with landmarks indicated; D: visualization of the mean closed-adapted (bottom gray wireframe) and openadapted (upper white wireframe) calcaneus (medial view) based on analyses in Curran (2012).
that there are specific constraints for locomotion in complex forests not found in other habitats, most critical of which is likely the requirement of rapid changes in direction to avoid obstacles (brush, logs, streams, etc.).
Lateral Tibial Condylar Margin Another component of the knee, the femoro-tibial (i.e., stifle) joint, reflects the type and range of movement at the knee (Organ and Ward, 2006; Sylvester, 2013). As with humans, movement at the knee is mainly restricted to the parasagittal plane (flexion and extension). The lateral condyle of the femur in cervids is quite round and sits on the lateral condyle of the tibia, which is convex in the anterior-posterior direction and concave medial-laterally. Sylvester (2013) hypothesized that the medially-expanded morphology of the medial tibial condyle in chimpanzees and gorillas (relative to humans) is related to the greater range of rotation at the knee in apes. Similarly, cervids that locomote through complex habitats such as some forests and woodlands need increased joint mobility (i.e., rotation) to avoid obstacles such as underbrush and downed trees. Previous GM analysis of the extant cervid lateral tibial condyle margin (Curran, 2012) showed that forestadapted cervids do have expanded lateral tibial condylar surfaces (Fig. 3). This expansion is especially evident on the medial side of the condylar surface. It was also found that forest-adapted cervids have more shallow joint surfaces, which again allows for greater mobility (but note that the surface is not completely flat and thus still provides some stability). Forest-adapted cervids rapidly change directions while fleeing predators to avoid obstacles or to place obstacles between themselves and their predators (Kappelman, 1988) and thus a more mobile joint allows for more flexibility in movement during quick changes in posture. Those cervids that must rely on speedy flight in open habitats to escape predators have more concave joint surfaces and narrower condyles (Hamrick, 1996) that function to limit motion to flexion and extension rather than rotation. Since cursorial cervids do not change direction as rapidly as forest-
adapted cervids, stability (and speed) is of primary concern during flight from predators (Hildebrand, 1985). It should be noted that (as with the femoral patellar margin) the variation in this feature does not form a continuum but instead reflects locomotion in complex habitats (closed forests) versus rapid locomotion in more open habitats and in fact, in all other habitat groups besides forests (Curran, 2012).
Calcaneus Despite being an important element of Cetartiodactyla (whales and artiodactyls) evolution (Geisler, 2001; Gingerich et al., 2001; Thewissen et al., 2007) and functional morphology, the calcaneus remains understudied compared to the other elements discussed here. Further, it is a dense bone that has good preservation in fossil contexts and thus warrants investigation due to its frequent recovery. In cervids, the calcaneus is a nonloadbearing element in the rearlimb that articulates primarily with the posterior surface of the astragalus but also with the os-malleolus superiorly and the cubonavicular inferiorly (Fig. 4). This bone’s main function is to act as an extension of the lever arm for the distal portion of the rearlimb (metatarsal and phalanges). As the gastrocnemius muscle (which inserts on the the calcaneal tuber) is contracted, the limb is extended with little excursion away from a parasagittal plane (Schaeffer, 1947) and the cervid moves forward. When the anterior portion of the calcaneus is set onto a plane defined by its rest position on the cubonavicular (see Fig. 5), calcanei of more open-adapted cervids are more vertically oriented (relative to the tibia), while a more posteriorly oriented calcaneus is associated with closed-adapted cervids (Curran, 2012). This means that the open-adapted calcaneus travels a shorter distance with each gastrocnemius contraction (a shorter stride) and power is sacrificed for speed. To escape predators in open habitats, cervids increase their flight speed by increasing their pace cycle (the number of times their foot contacts the substrate per second (Gambaryan, 1974; Geist, 1998). In forest-adapted cervids, which have
EUCLADOCEROS ECOMORPHOLOGY
297
Fig. 6. Visualizations of the plantar margin of the third phalanx based on data presented in Curran (2012). A: example of an extant dry-substrate specimen (Cervus elaphus, MVZ 132244) and visualizations of the mean dry-adapted plantar margin of the third phalanx; B: example of an extant wet-substrate specimen (Blasroceros dichotmus, NMNH 261018) and visualizations of the mean wet-adapted plantar margin of the third phalanx.
many more obstacles to clear during flight from predators, the more oblique rest position of the calcaneus results in a slower but more powerful force. These cervids have fewer pace cycles per second but are able to place obstacles between themselves and predators and to change directions quickly.
Third Phalanx Plantar Margin Since the third phalanx is the place where the cervid interacts directly with the terrain, its morphology is a reflection of the hardness and slope of the substrate. On substrates that are wet or slippery, cervids stabilize their limb by splaying their phalanges at the metatarsophalangeal joint (fetlock) and have long, thin, and tapered third phalanges (K€ohler, 1993). Hard and slopped surfaces, often found in mountainous terrain, also require phalangeal splaying, but only at the second to third phalangeal joint (K€ohler, 1993). These phalanges are usually short and narrow to allow for precise landing during rock-hopping. In open habitats with hard substrates, all phalanges are locked into a line with the metapodial to avoid dislocation during speedy flight and are blocky in appearance. In my prior analysis of third phalanx plantar morphology (Curran, 2012), it was found that cervids closely follow the hypotheses proposed by K€ohler (1993). Specifically, cervids locomoting across hard, dry substrates have wide phalanges with little anterior tapering (Fig. 6A) while those moving across slick substrates have thin, tapered plantar margins (Fig. 6B). Open-adapted
cervids also have flatter plantar surfaces, which provide an even more stable platform for rapid locomotion. By combining the findings discussed above, a picture of cervid locomotor adaptations can be generated. Morphological variation in cervid postcranial anatomy (as summarized in Fig. 7) comes down to a compromise between mobility and stability. Cervids that must escape predators by speedy flight (i.e., those in open habitats) emphasize stability over mobility. These stability adaptations are most evident in the morphology of the lateral tibial condylar surface, which is narrow and somewhat concave to prevent dislocation, and the wide, flat plantar surface of the third phalanx. Speed is generated by moving through a full pace-cycle as quickly as possible. This means that the limb does not move as far with each pace-cycle, though this may incur a reduction in power generation. However, the morphology of the patellar surface serves to increase torque through the pace-cycle and thus may make up for some of the power lost due to a short pace-cycle. Torque increase would occur when the patella is furthest from the center of joint rotation, which is when it moves over the anterior bulge of the patellar surface. This is the point where the knee joint is almost fully extended and thus the foot is moved quickly into position to begin the next pace-cycle. Requirements for locomotion in closed habitats revolve around increased mobility, which is necessary when there are more ground-level obstacles, as found in forests. The round, flat morphology of the lateral tibial condyle demonstrates the increased mobility at the knee and the tapered third phalanges are associated with the
298
CURRAN
Fig. 7. Summarization of the morphological features associated with open- and closed-adapted cervids based on data presented in Curran (2012).
ability to splay the toes (K€ohler, 1993). The posterior orientation of the calcaneus indicates that it moves further with each pace-cycle, which is required with saltatorial locomotion (movement in a series of leaps). An increase in calcaneal movement is slower, but will also increase power generation, which is necessary due to the lack of torque increase over the pace-cycle at the knee. With saltatorial locomotion, the cervid is able to rapidly shift directions and move quite far with each pace-cycle.
Confounding Factors: Phylogeny and Body Size Phylogenetic relatedness constrains the variation possible in morphology (Felsenstein, 1985). If we are interested in morphology associated with different habitats, we must understand how much of the morphological variation detected is due to shared ancestry. The question then becomes whether it is prudent or necessary to remove that signal. In some cases, adaptive change (to environmental conditions) may follow taxonomic differentiation (Polly et al., 2013) and thus by removing the variation in morphology associated with phylogeny we may actually be removing the environmental signal we are interested in examining. Recent studies have demonstrated that traits associated with function [mandibular morphology in rhinoceroses (Piras et al., 2010) and bovids (Raia et al., 2010) and post-cranial morphology in bovids (Barr, 2013; Louys et al., 2013)] are more related to phylogeny than function. Since the variables used in this study were specifically designed to test functional hypotheses, it is predicted that their form will be better explained by function than phylogeny, though this must be tested for the variables included here. Recently, ecomorphology has been criticized because it was found that most of the variation in linear measure-
ments could be attributed to size (Klein et al., 2010). This is likely due mostly to isometric size variation, for no matter how similar in morphology a small deer (for example, a pudu) and a large deer (for example, a moose) are, if raw measurements are used in an analysis, they will always be differentiated due to differences in size. Fortunately, the method employed in this study, GM, removes isometric size as part of its preliminary data transformation (Procrustes superimposition). However, even this transformation will not remove shape differences that are due to different body sizes (allometry). Nor can it adjust for any covariance of body size and morphology that is related to ancestry. For example, almost all of the Muntiacini are of small body size and thus will likely share morphology simply due to this relationship. To account for variation due to allometry and phylogeny this in this study, a phylogenetic partial least square of shape on size will be conducted.
Application to Eucladoceros Ecomorphology The history of cervid evolution is one of dispersal from warm, closed environments of the Miocene into progressively more open, arid, and colder environments (Geist, 1998). Cervids are generally conceptualized as mostly woodland species, and indeed, many are (see Table 1), though there are extant open-adapted species (Ozotoceros bezoarticus, Hydropotes inermis, and Rangifer tarandus). As a result, cervids are often used as proxies for forested to wooded habitats (Lordkipanidze et al., 2007; Petronio et al., 2011). However, adaptation to cold and open habitats is impressively demonstrated with Plio-Pleistocene Libralces, Eucladoceros, and Megaloceros, large-bodied cervids with prodigious antler racks. Though these genera are now extinct, very large deer bearing huge antlers
299
EUCLADOCEROS ECOMORPHOLOGY
TABLE 1. Sample sizes of extant specimens for the four morphological units of analysis, tribal, habitat, and substrate categories for each species Species Alces alces Axis axis Axis porcinus Blastocerus dichotomus Capreolus capreolus Cervus duvaucelii Cervus elaphus Cervus eldii Cervus mariannus Cervus nippon Cervus timorensis Cervus unicolor Dama dama Elaphodus cephalophus Elaphurus davidianus Hippocamelus antisenis Hippocamelus bisculus Hydropotes inermis Mazama chunyi Mazama gouazoubira Mazama rufina Muntiacus atherodes Muntiacus muntjak Muntiacus reevesi Odocoileus hemionus Odocoileus virginianus Ozotoceros bezoarticus Rangifer tarandus Totals
Tribe Alceini Cervini Cervini Odocoileini Capreolini Cervini Cervini Cervini Cervini Cervini Cervini Cervini Cervini Muntiacini Cervini Odocoileini Odocoileini Capreolini Odocoileini Odocoileini Odocoileini Muntiacini Muntiacini Muntiacini Odocoileini Odocoileini Odocoileini Odocoileini
Habitat
Substrate
C IO IO IO IO IO IC IO IO IC IO IC IC C IO O O O C C C C C C O IO O O
D E E W E W D W D D D D D W D Mt Mt W W W W W W W Mt E D D
FemMed
TibLat
Cal LMs
Phal3
6 2 3 3 7 1 8 2 3 4 2 2 5 7 4 2 1 4 0 2 0 0 1 4 17 18 4 11 123
6 2 3 3 7 1 9 2 4 4 2 2 4 6 4 2 1 4 1 3 2 1 3 4 17 19 4 11 131
5 2 2 3 7 1 8 2 4 4 2 2 2 6 4 2 1 4 0 2 2 3 1 4 15 18 3 8 117
3 2 0 2 6 0 4 2 0 2 2 0 4 1 1 1 0 2 0 1 0 0 0 1 7 11 2 4 58
Cal LMs 5 Calcaneus Landmarks, FemMed 5 Femoral Medial Patellar Margin, TibLat 5 Lateral Tibial Plateau Margin, Phal3 5 Third Phalanx Plantar Margin, Habitat: O 5 Open, IO 5 Intermediate Open, IC 5 Intermediate Closed, C 5 Closed, Substrate: D 5 Dry, E 5 Ecotone, W 5 Wet, and Mt 5 Mountain
still exist today in relatively cold habitats (Alces alces, Rangifer tarandus, and Cervus elaphus). First appearing in Europe around 2.5 Ma (Croitor, 2009), Eucladoceros became a major part of the Eurasian Pleistocene fauna. Although the exact phylogenetic position of Eucladoceros is unknown, it was likely in the extant tribe Cervini and not the extinct Megacerini (Lister et al., 2005). According to Croitor (2009) only two species should be recognized in Villafranchian Europe, E. ctenoides and E. dicranios. This study includes specimens attributed to E. ctenoides, which was approximately the size of a modern red deer that had up to six tines on each antler (Kaiser and Croitor, 2004) and specimens attributed more generally to Eucladoceros sp. Adaptations to open habitats are reflected in the diet of Eucladoceros ctenoides, as the mesowear signal of their dentition is similar to that of grazing bovids (Kaiser and Croitor, 2004). Furthermore, their femoral head is wide in its lateral aspect, limiting movement to the parasagittal plane, which is associated with adaptations to cursoriality (Kappelman, 1988; Kaiser and Croitor, 2004). This study examines Eucladoceros specimens from three different sites in Europe that contain at least two different subspecies of E. ctenoides. Two sites (Saint Vallier, France, and Gr aunceanu, Romania, are nearly contemporaneous [MN17: 2.4–1.9 Million years ago (Ma)], while the third, Sene`ze, France is slightly younger (MN18), though located very close geographically to Saint Vallier (Fig. 8).
Saint Vallier, France is an open-air site along the Rhone River and specimens were deposited there in fluvial conditions, though they have little weathering or abrasion (Valli, 2004b). Saint Vallier, containing the remains of Eucladoceros ctenoides vireti, is the reference locality for biozone MN17 (Valli, 2004b). Gr aunceanu, situated in the Oltet¸ River valley in central Romania, represents the most fossiliferous assemblage of the Tetiou I sequence (R adulesco et al., 2003). This site is dated to the Early Pleistocene (MN 17, R adulesco et al., 2003). The specimens from this site are identified only as Eucladoceros sp. Sene`ze is faunally equivalent to Gr aunceanu (R adulesco et al., 2003), although younger. Specifically, Sene`ze must be younger than 2.1 Ma, as it was deposited in a tephra layer with this date and was accumulated in lacustrine (maar) deposits (Delson et al., 2006) with reverse (Matuyama) polarity, except for a normal section (Reunion) at the top (Roger et al., 2000). It is the reference for MN18, the Late Villafranchian (Delson et al., 2006) and contains Eucladoceros ctenoides ctenoides.
MATERIALS AND METHODS Here, cervid skeletal morphology is analyzed using GM. With GM shape is recorded as a series of 3D coordinates, which then [after generalized Procrustes analysis (GPA) and projection into a tangent space] are compared across the sample under investigation
300
CURRAN
Fig. 8. Locations of the three early Pleistocene sites in this study.
TABLE 2. Summary of fossil specimens in this study Site name
Location
Date
Taxa present
Cal LMs
Phalanx 3
Gr aunceanu St. Vallier
Romania France
MN 17 (2.4–1.9 Ma) MN 17 (2.4–1.9 Ma)
Sene`ze
France
MN 18 (2.2–1.5 Ma)
Eucladoceros sp. Eucladoceros ctenoides vireti Cervidae (large size) Eucladoceros ctenoides ctenoides Cervidae (large size)
17 3 6 2 0
24 0 0 6 2
Ma 5 Millions of years, CalLMs 5 Calcaneus landmarks.
(Cooke and Terhune, this volume). The entire shape of an anatomical region is analyzed with GM to find those specimens that group together (or apart) in morphospace due to similar (or dissimilar) morphology. GM provides a good starting place to assess how shape varies and then serves as a good tool to test if shape varies in predictable ways among functional groups. One of the advantages of GM is that morphological differences can be visualized using information about the interrelationships of the 3D coordinates under analysis and thus visualizations can then be used to test hypotheses regarding how shape varies. Quantification of complex shapes, sensitivity to subtle variation, and visualization of morphological variation make GM an important tool in the functional morphologist’s toolkit, and a powerful tool for examining shape variation in cervids.
The comparative (training) sample of extant cervid specimens (n 5 136) studied here was collected from four museums in the United States [American Museum of Natural History (New York), the Museum of Vertebrate Zoology at the University of California, Berkeley (Berkeley, CA), and the National Museum of Natural History (Washington, DC). the Chicago Field Museum (Chicago, IL), see Table 1]. Undamaged, adult, nonzoo, and nonpathological specimens with life history information were preferentially sampled. Every effort was made to sample evenly across the sexes, the four habitat categories, and the five cervid tribes. Data on fossil Eucladoceros specimens were collected on specimens from three early Pleistocene sites in France (Saint Vallier and Sene`ze, collections which are both split between Claude Bernard University of Lyon 1 and Collection du Musee des Confluences, Lyon) and
EUCLADOCEROS ECOMORPHOLOGY
Fig. 9. Phylogenetic tree showing relationships among taxa in this study; downloaded from 10 kTrees.org.
301
302
CURRAN
TABLE 3. Regressions of the natural log of centroid size on the natural log (ln) of several body size estimators in cervids Feature
Estimator
FemMed FemMed FemMed TibLat TibLat TibLat CalLMs CalLMs CalLMs Phal3 Phal3 Phal3
lnBW lnBH lnBL lnBW lnBH lnBL lnBW lnBH lnBL lnBW lnBH lnBL
R2
P-value
0.83 0.83 0.82 0.85 0.82 0.82 0.87 0.81 0.83 0.81 0.68 0.82
Exploring Eucladoceros Ecomorphology Using Geometric Morphometrics SABRINA C. CURRAN* Ohio University, Department of Sociology and Anthropology, Athens, OH 45701
ABSTRACT An increasingly common method for reconstructing paleoenvironmental parameters of hominin sites is ecological functional morphology (ecomorphology). This study provides a geometric morphometric study of cervid rearlimb morphology as it relates to phylogeny, size, and ecomorphology. These methods are then applied to an extinct Pleistocene cervid, Eucladoceros, which is found in some of the earliest hominin-occupied sites in Eurasia. Variation in cervid postcranial functional morphology associated with different habitats can be summarized as trade-offs between joint stability versus mobility and rapid movement versus power-generation. Cervids in open habitats emphasize limb stability to avoid joint dislocation during rapid flight from predators. Closed-adapted cervids require more joint mobility to rapidly switch directions in complex habitats. Two skeletal features (of the tibia and calcaneus) have significant phylogenetic signals, while two (the femur and third phalanx) do not. Additionally, morphology of two of these features (tibia and third phalanx) were correlated with body size. For the tibial analysis (but not the third phalanx) this correlation was ameliorated when phylogeny was taken into account. Eucladoceros specimens from France and Romania fall on the more open side of the habitat continuum, a result that is at odds with reconstructions of their diet as browsers, suggesting that they may have had a behavioral regime unlike any extant cervid. Anat Rec, C 2014 Wiley Periodicals, Inc. 298:291–313, 2015. V
Key words: cervidae; locomotion; phylogeny; femur; tibia; calcaneus; third phalanx
The Pleistocene was a time of mammalian megafauna and members of the deer family, Cervidae, were no exception. One of these giants, Eucladoceros or the comb-antlered deer, was widely distributed across Eurasia. Since it was contemporaneous with the earliest appearance of hominins outside of Africa, elucidation of Eucladoceros paleobiology informs our understanding the range of habitats those early hominins would have encountered. This study analyzes Eucladoceros functional morphology to reconstruct its locomotor behavior and habitat preferences. Morphology associated with locomotion is known to vary by habitat (Kappelman, 1988, 1991; Plummer and Bishop, 1994; Kappelman et al., 1997; Scott et al., 1999; DeGusta and Vrba, 2003, 2005; Weinand, 2007; Kovarovic and Andrews, 2007; Plummer et al., 2008) and as such, it can be used as a paleohabitat proxy. However, factors such as phylogenetic relatedness and body size are also known to influC 2014 WILEY PERIODICALS, INC. V
ence skeletal morphology. This study has two goals: (1) to extend the methods established in Curran (2012) to address issues of body size and phylogeny in extant
Grant sponsor: National Science Foundation; Grant number: BCS-0824607.; Grant sponsor: University of Minnesota Doctoral Dissertation Fellowship.; Grant sponsor: University of Minnesota, Herb E. Wright Fellowship in Paleoecology; Grant sponsor: University of Minnesota, Department of Anthropology Block Grants. *Correspondence to: Sabrina Curran, Ohio University, Department of Sociology and Anthropology, Athens OH 45701. E-mail: [email protected] Received 3 October 2014; Accepted 11 October 2014. DOI 10.1002/ar.23066 Published online in Wiley Online Library (wileyonlinelibrary. com).
292
CURRAN
Fig. 1. The four morphological features examined in this study, with their position in the cervid rearlimb (center). Each feature is demonstrated with a photograph of the region of interest and the Procrustesaligned coordinates for the specimens in this study (visualized with rola-Castillo, 2013). A: the Geomorph package for R, Adams, and Ota
Medial margin of the femoral patellar surface (right-side, Alces alces). B: Tibial lateral condyle margin (right-side, Alces alces). C: Calcaneus landmarks (right-side, Alces alces). D: Plantar margin of the third phalanx (side unknown, Odocoileus hemionus).
cervids and (2) apply these methods to Eucladoceros to better understand the locomotor behavior and ecological niche(s) of members of this extinct genus.
of ecomorphological studies such as the one presented here.
CERVID REARLIMB ECOMORPHOLOGY ECOMORPHOLOGY Ecological functional morphology, or ecomorphology, is the study of adaptation at its core (Wainwright, 1994; Plummer et al., 2008). What we are really asking is “how do different phenotypes affect fitness?” For this study, this question becomes “how do variants in skeletal morphology affect locomotion, and especially escape from predators, in the Cervidae?” In answering this question, we are also addressing the structure of the habitat, for variation in vegetative density and structure is one of the main influences on artiodactyl locomotion (Kappelman, 1988; K€ohler, 1993). An individual’s morphology places limits on the types of behaviors that can be performed and the environments in which those behaviors can take place (Wainwright, 1994). For example, a forest-dwelling cervid can often lose a predator by frequently changing direction through a series of powerful leaps, but this behavior would not be very successful in fleeing a cursorial predator in an open landscape (Kappelman, 1988). It is performance [defined by Wainwright (1994) as “an organism’s ability to carry out specific behaviors”] that forms the link between morphology and ecology. Since performance requirements vary between habitats, and morphology varies with performance requirements, we can then link morphology to habitats. This linkage between morphology and performance is the foundation
Just as bovid remains have been shown to be useful as proxies for paleoenvironments in Africa (Plummer and Bishop, 1994; Kappelman et al., 1997; Scott, 2004; Kovarovic and Andrews, 2007) and Southeast Asia (Weinand, 2007), so too are cervid remains useful for reconstructing the paleohabitats of Eurasian sites, where they are often the dominant taxa. Extant cervids inhabit a wide range of environments, from tundra and grasslands to swamps and forests (both temperate and tropical), making them an appropriate proxy for past habitats. Within these various habitats, cervids perform a variety of predator-escape behaviors. Cervids avoid predators mainly by being cryptic, especially in densely vegetated habitats (Bro-Jørgensen, 2008), although some largebodied forms (Alces alces, Cervus unicolor) will confront predators (Geist, 1998), while others (Rangifer tarandus, Cervus elaphus) depend on the “selfish herd” (Hamilton, 1971) for safety. In open environments, cervids depend on speedy flight and thus their morphological adaptations reflect rapid movement and joint stability. Cervids in closed environments must locomote through a more complex three-dimensional landscape and their morphology reflects the need for increased mobility. In this study, geometric morphometrics (GM) is used to quantify variation in four skeletal features from the rearlimb of Cervidae as it relates to locomotion in different habitats. While there are many components that influence the functional morphology of the cervid
EUCLADOCEROS ECOMORPHOLOGY
Fig. 2. Visualizations of medial femoral patellar margin morphology based on data presented in Curran (2012). A single canonical variate was created between representative specimens of these habitat extremes with a multivariate regression of Procrustes coordinates on canonical variates scores for the two-group CVA (McNulty et al., 2006). The coefficients of this regression were then treated as a vector that was added to and subtracted from the consensus landmark configuration (after Frost et al., 2003). The two resulting configurations demonstrate the differences between the habitat extremes as mod-
293
eled in Morpheus (Slice et al., 2002). A: Top 5 example of an openadapted cervid (Hydropotes inermis, AMNH 147434), middle 5 visualization of the mean open-adapted medial femoral patellar margin, and bottom 5 the ellipse formed by continuing the medial margin’s arc; and B: a Top 5 closed-adapted cervid (Pudu mephistophiles, MVZ 122523), middle 5 visualization of the mean closed-adapted medial femoral patellar margin, and the more circular shape formed by continuing the arc of the mean closed-adapted specimens. Scale bar 5 1 cm.
294
CURRAN
Fig. 3. Visualizations of lateral tibial plateau morphology based on data presented in Curran (2012). A: open-adapted morphology with an extant example (Ozotoceros bezoarticus, NMNH 270379) from superior view (top) and posterior view (bottom), with corresponding visualizations of the mean open-adapted lateral tibial plateau margin from the
superior view and the posterior view; B: open-adapted morphology with an extant example (Mazama rufina, MVZ 124095) from superior view (top) and posterior view (bottom) with corresponding visualization of the mean closed-adapted lateral tibial plateau margin from the superior and posterior view.
rearlimb, this study focuses on features involved in power generation and stabilization mechanisms. The four features (see Fig. 1) were selected to sample skeletal morphology important to locomotion: (1) the medial margin of the patellar surface of the femur is hypothesized to indicate torque generation at the knee (Kappelman, 1988); (2) the range and type of motion at the knee is analyzed here using the morphology of the tibia’s lateral condylar margin (Organ and Ward, 2006); (3) the rest angle and range of motion of the calcaneus is hypothesized to be related to the number of times the animal moves through its pace-cycle per minute (Curran, 2012); and (4) plantar morphology of the third phalanx (where the cervid contacts the substrate) reflects the role of the hoof in limb stabilization during locomotion (K€ohler, 1993). Examined in concert, these four features inform our understanding of cervid locomotion and rearlimb ecomorphology. Below I discuss previous studies regarding the functional morphology and biomechanics of these regions with a particular focus on Curran (2012), the study that introduced the GM methods used here, and of which this study is a direct extension.
center of rotation for the knee). Hermanson and MacFadden (1996) argued that it is unclear how this would provide mechanical advantage since the patella tracks the patellar fossa and not the medial ridge. However, Shockley (2001) points out that a portion of the patella does track over the medial ridge, raising the medial portion of the patella during extension. This may increase stability by increasing the tension on the medial patellar ligament and restricting movement at the joint. Further, Shockley (2001) states that an enlarged medial ridge would resist forces from adductors and prevents the patella from dislocating. Janis et al. (2012) suggest that the limb is positioned more laterally in ungulates during a gallop due to the need to avoid their large abdomen and thus increased medial stabilization is necessary. Here, the medial patellar margin is digitized with a series of three-dimensional coordinates along the most prominent ridge (Fig. 2). Prior analysis of extant cervids (Curran, 2012) demonstrated that forest-adapted (closedadapted) cervids have medial margins with a smooth, continuous arc (see Fig. 2B) while cursorial (non-forest) cervids have a distinct anterior projection (Fig. 2A). This is the exact morphology hypothesized by Kappelman (1988) wherein if one were to continue the arc of the medial patellar surface, closed-adapted cervids would appear more circular while cursorial cervids would be more elliptical for both margins (See Fig. 2, bottom images). For cursorial cervids, the anterior bulge creates a progressive increase in the distance of the patella to the center of rotation throughout limb extension (Kappelman, 1988). This, in combination with increased medial patellar ligament tension (Shockley, 2001; Janis, 2012), allows cursorial cervids to achieve speedy flight
Medial Femoral Patellar Margin As is the case for most artiodactyl knees, cervid femora have larger medial than lateral patellar margins. This asymmetry is more remarkable in larger and openadapted species (Janis et al., 2012). Kappelman (1988) hypothesized that a larger medial ridge serves to increase torque at the knee by increasing the length of the moment arm (moving the patella further from the
EUCLADOCEROS ECOMORPHOLOGY
295
Fig. 4. The hock (ankle) joint of an artiodactyl with each element labeled and the visible joint articular surfaces colored (green 5 astragalus-calcaneus, orange 5 os malleolus-calcaneus, and purple 5 cubonavicular-calcaneus).
with reduced risks of patellar dislocation. Interestingly, the contrast here is between forest-adapted (i.e., closed) deer and deer in all other habitat types (i.e., open and
intermediate habitat types) and there is not a continuum of variation like that found in other skeletal features discussed below (calcaneus, third phalanx). This suggests
296
CURRAN
Fig. 5. A: An example of a closed-adapted specimen (Pudu mephistophiles, MVZ 122523; medial view) with landmarks indicated; B: an example of an open-adapted specimen (Ozotoceros bezoarticus, MVZ 122523) with landmarks indicated; C: the aforementioned specimens overlapped in articular position with landmarks indicated; D: visualization of the mean closed-adapted (bottom gray wireframe) and openadapted (upper white wireframe) calcaneus (medial view) based on analyses in Curran (2012).
that there are specific constraints for locomotion in complex forests not found in other habitats, most critical of which is likely the requirement of rapid changes in direction to avoid obstacles (brush, logs, streams, etc.).
Lateral Tibial Condylar Margin Another component of the knee, the femoro-tibial (i.e., stifle) joint, reflects the type and range of movement at the knee (Organ and Ward, 2006; Sylvester, 2013). As with humans, movement at the knee is mainly restricted to the parasagittal plane (flexion and extension). The lateral condyle of the femur in cervids is quite round and sits on the lateral condyle of the tibia, which is convex in the anterior-posterior direction and concave medial-laterally. Sylvester (2013) hypothesized that the medially-expanded morphology of the medial tibial condyle in chimpanzees and gorillas (relative to humans) is related to the greater range of rotation at the knee in apes. Similarly, cervids that locomote through complex habitats such as some forests and woodlands need increased joint mobility (i.e., rotation) to avoid obstacles such as underbrush and downed trees. Previous GM analysis of the extant cervid lateral tibial condyle margin (Curran, 2012) showed that forestadapted cervids do have expanded lateral tibial condylar surfaces (Fig. 3). This expansion is especially evident on the medial side of the condylar surface. It was also found that forest-adapted cervids have more shallow joint surfaces, which again allows for greater mobility (but note that the surface is not completely flat and thus still provides some stability). Forest-adapted cervids rapidly change directions while fleeing predators to avoid obstacles or to place obstacles between themselves and their predators (Kappelman, 1988) and thus a more mobile joint allows for more flexibility in movement during quick changes in posture. Those cervids that must rely on speedy flight in open habitats to escape predators have more concave joint surfaces and narrower condyles (Hamrick, 1996) that function to limit motion to flexion and extension rather than rotation. Since cursorial cervids do not change direction as rapidly as forest-
adapted cervids, stability (and speed) is of primary concern during flight from predators (Hildebrand, 1985). It should be noted that (as with the femoral patellar margin) the variation in this feature does not form a continuum but instead reflects locomotion in complex habitats (closed forests) versus rapid locomotion in more open habitats and in fact, in all other habitat groups besides forests (Curran, 2012).
Calcaneus Despite being an important element of Cetartiodactyla (whales and artiodactyls) evolution (Geisler, 2001; Gingerich et al., 2001; Thewissen et al., 2007) and functional morphology, the calcaneus remains understudied compared to the other elements discussed here. Further, it is a dense bone that has good preservation in fossil contexts and thus warrants investigation due to its frequent recovery. In cervids, the calcaneus is a nonloadbearing element in the rearlimb that articulates primarily with the posterior surface of the astragalus but also with the os-malleolus superiorly and the cubonavicular inferiorly (Fig. 4). This bone’s main function is to act as an extension of the lever arm for the distal portion of the rearlimb (metatarsal and phalanges). As the gastrocnemius muscle (which inserts on the the calcaneal tuber) is contracted, the limb is extended with little excursion away from a parasagittal plane (Schaeffer, 1947) and the cervid moves forward. When the anterior portion of the calcaneus is set onto a plane defined by its rest position on the cubonavicular (see Fig. 5), calcanei of more open-adapted cervids are more vertically oriented (relative to the tibia), while a more posteriorly oriented calcaneus is associated with closed-adapted cervids (Curran, 2012). This means that the open-adapted calcaneus travels a shorter distance with each gastrocnemius contraction (a shorter stride) and power is sacrificed for speed. To escape predators in open habitats, cervids increase their flight speed by increasing their pace cycle (the number of times their foot contacts the substrate per second (Gambaryan, 1974; Geist, 1998). In forest-adapted cervids, which have
EUCLADOCEROS ECOMORPHOLOGY
297
Fig. 6. Visualizations of the plantar margin of the third phalanx based on data presented in Curran (2012). A: example of an extant dry-substrate specimen (Cervus elaphus, MVZ 132244) and visualizations of the mean dry-adapted plantar margin of the third phalanx; B: example of an extant wet-substrate specimen (Blasroceros dichotmus, NMNH 261018) and visualizations of the mean wet-adapted plantar margin of the third phalanx.
many more obstacles to clear during flight from predators, the more oblique rest position of the calcaneus results in a slower but more powerful force. These cervids have fewer pace cycles per second but are able to place obstacles between themselves and predators and to change directions quickly.
Third Phalanx Plantar Margin Since the third phalanx is the place where the cervid interacts directly with the terrain, its morphology is a reflection of the hardness and slope of the substrate. On substrates that are wet or slippery, cervids stabilize their limb by splaying their phalanges at the metatarsophalangeal joint (fetlock) and have long, thin, and tapered third phalanges (K€ohler, 1993). Hard and slopped surfaces, often found in mountainous terrain, also require phalangeal splaying, but only at the second to third phalangeal joint (K€ohler, 1993). These phalanges are usually short and narrow to allow for precise landing during rock-hopping. In open habitats with hard substrates, all phalanges are locked into a line with the metapodial to avoid dislocation during speedy flight and are blocky in appearance. In my prior analysis of third phalanx plantar morphology (Curran, 2012), it was found that cervids closely follow the hypotheses proposed by K€ohler (1993). Specifically, cervids locomoting across hard, dry substrates have wide phalanges with little anterior tapering (Fig. 6A) while those moving across slick substrates have thin, tapered plantar margins (Fig. 6B). Open-adapted
cervids also have flatter plantar surfaces, which provide an even more stable platform for rapid locomotion. By combining the findings discussed above, a picture of cervid locomotor adaptations can be generated. Morphological variation in cervid postcranial anatomy (as summarized in Fig. 7) comes down to a compromise between mobility and stability. Cervids that must escape predators by speedy flight (i.e., those in open habitats) emphasize stability over mobility. These stability adaptations are most evident in the morphology of the lateral tibial condylar surface, which is narrow and somewhat concave to prevent dislocation, and the wide, flat plantar surface of the third phalanx. Speed is generated by moving through a full pace-cycle as quickly as possible. This means that the limb does not move as far with each pace-cycle, though this may incur a reduction in power generation. However, the morphology of the patellar surface serves to increase torque through the pace-cycle and thus may make up for some of the power lost due to a short pace-cycle. Torque increase would occur when the patella is furthest from the center of joint rotation, which is when it moves over the anterior bulge of the patellar surface. This is the point where the knee joint is almost fully extended and thus the foot is moved quickly into position to begin the next pace-cycle. Requirements for locomotion in closed habitats revolve around increased mobility, which is necessary when there are more ground-level obstacles, as found in forests. The round, flat morphology of the lateral tibial condyle demonstrates the increased mobility at the knee and the tapered third phalanges are associated with the
298
CURRAN
Fig. 7. Summarization of the morphological features associated with open- and closed-adapted cervids based on data presented in Curran (2012).
ability to splay the toes (K€ohler, 1993). The posterior orientation of the calcaneus indicates that it moves further with each pace-cycle, which is required with saltatorial locomotion (movement in a series of leaps). An increase in calcaneal movement is slower, but will also increase power generation, which is necessary due to the lack of torque increase over the pace-cycle at the knee. With saltatorial locomotion, the cervid is able to rapidly shift directions and move quite far with each pace-cycle.
Confounding Factors: Phylogeny and Body Size Phylogenetic relatedness constrains the variation possible in morphology (Felsenstein, 1985). If we are interested in morphology associated with different habitats, we must understand how much of the morphological variation detected is due to shared ancestry. The question then becomes whether it is prudent or necessary to remove that signal. In some cases, adaptive change (to environmental conditions) may follow taxonomic differentiation (Polly et al., 2013) and thus by removing the variation in morphology associated with phylogeny we may actually be removing the environmental signal we are interested in examining. Recent studies have demonstrated that traits associated with function [mandibular morphology in rhinoceroses (Piras et al., 2010) and bovids (Raia et al., 2010) and post-cranial morphology in bovids (Barr, 2013; Louys et al., 2013)] are more related to phylogeny than function. Since the variables used in this study were specifically designed to test functional hypotheses, it is predicted that their form will be better explained by function than phylogeny, though this must be tested for the variables included here. Recently, ecomorphology has been criticized because it was found that most of the variation in linear measure-
ments could be attributed to size (Klein et al., 2010). This is likely due mostly to isometric size variation, for no matter how similar in morphology a small deer (for example, a pudu) and a large deer (for example, a moose) are, if raw measurements are used in an analysis, they will always be differentiated due to differences in size. Fortunately, the method employed in this study, GM, removes isometric size as part of its preliminary data transformation (Procrustes superimposition). However, even this transformation will not remove shape differences that are due to different body sizes (allometry). Nor can it adjust for any covariance of body size and morphology that is related to ancestry. For example, almost all of the Muntiacini are of small body size and thus will likely share morphology simply due to this relationship. To account for variation due to allometry and phylogeny this in this study, a phylogenetic partial least square of shape on size will be conducted.
Application to Eucladoceros Ecomorphology The history of cervid evolution is one of dispersal from warm, closed environments of the Miocene into progressively more open, arid, and colder environments (Geist, 1998). Cervids are generally conceptualized as mostly woodland species, and indeed, many are (see Table 1), though there are extant open-adapted species (Ozotoceros bezoarticus, Hydropotes inermis, and Rangifer tarandus). As a result, cervids are often used as proxies for forested to wooded habitats (Lordkipanidze et al., 2007; Petronio et al., 2011). However, adaptation to cold and open habitats is impressively demonstrated with Plio-Pleistocene Libralces, Eucladoceros, and Megaloceros, large-bodied cervids with prodigious antler racks. Though these genera are now extinct, very large deer bearing huge antlers
299
EUCLADOCEROS ECOMORPHOLOGY
TABLE 1. Sample sizes of extant specimens for the four morphological units of analysis, tribal, habitat, and substrate categories for each species Species Alces alces Axis axis Axis porcinus Blastocerus dichotomus Capreolus capreolus Cervus duvaucelii Cervus elaphus Cervus eldii Cervus mariannus Cervus nippon Cervus timorensis Cervus unicolor Dama dama Elaphodus cephalophus Elaphurus davidianus Hippocamelus antisenis Hippocamelus bisculus Hydropotes inermis Mazama chunyi Mazama gouazoubira Mazama rufina Muntiacus atherodes Muntiacus muntjak Muntiacus reevesi Odocoileus hemionus Odocoileus virginianus Ozotoceros bezoarticus Rangifer tarandus Totals
Tribe Alceini Cervini Cervini Odocoileini Capreolini Cervini Cervini Cervini Cervini Cervini Cervini Cervini Cervini Muntiacini Cervini Odocoileini Odocoileini Capreolini Odocoileini Odocoileini Odocoileini Muntiacini Muntiacini Muntiacini Odocoileini Odocoileini Odocoileini Odocoileini
Habitat
Substrate
C IO IO IO IO IO IC IO IO IC IO IC IC C IO O O O C C C C C C O IO O O
D E E W E W D W D D D D D W D Mt Mt W W W W W W W Mt E D D
FemMed
TibLat
Cal LMs
Phal3
6 2 3 3 7 1 8 2 3 4 2 2 5 7 4 2 1 4 0 2 0 0 1 4 17 18 4 11 123
6 2 3 3 7 1 9 2 4 4 2 2 4 6 4 2 1 4 1 3 2 1 3 4 17 19 4 11 131
5 2 2 3 7 1 8 2 4 4 2 2 2 6 4 2 1 4 0 2 2 3 1 4 15 18 3 8 117
3 2 0 2 6 0 4 2 0 2 2 0 4 1 1 1 0 2 0 1 0 0 0 1 7 11 2 4 58
Cal LMs 5 Calcaneus Landmarks, FemMed 5 Femoral Medial Patellar Margin, TibLat 5 Lateral Tibial Plateau Margin, Phal3 5 Third Phalanx Plantar Margin, Habitat: O 5 Open, IO 5 Intermediate Open, IC 5 Intermediate Closed, C 5 Closed, Substrate: D 5 Dry, E 5 Ecotone, W 5 Wet, and Mt 5 Mountain
still exist today in relatively cold habitats (Alces alces, Rangifer tarandus, and Cervus elaphus). First appearing in Europe around 2.5 Ma (Croitor, 2009), Eucladoceros became a major part of the Eurasian Pleistocene fauna. Although the exact phylogenetic position of Eucladoceros is unknown, it was likely in the extant tribe Cervini and not the extinct Megacerini (Lister et al., 2005). According to Croitor (2009) only two species should be recognized in Villafranchian Europe, E. ctenoides and E. dicranios. This study includes specimens attributed to E. ctenoides, which was approximately the size of a modern red deer that had up to six tines on each antler (Kaiser and Croitor, 2004) and specimens attributed more generally to Eucladoceros sp. Adaptations to open habitats are reflected in the diet of Eucladoceros ctenoides, as the mesowear signal of their dentition is similar to that of grazing bovids (Kaiser and Croitor, 2004). Furthermore, their femoral head is wide in its lateral aspect, limiting movement to the parasagittal plane, which is associated with adaptations to cursoriality (Kappelman, 1988; Kaiser and Croitor, 2004). This study examines Eucladoceros specimens from three different sites in Europe that contain at least two different subspecies of E. ctenoides. Two sites (Saint Vallier, France, and Gr aunceanu, Romania, are nearly contemporaneous [MN17: 2.4–1.9 Million years ago (Ma)], while the third, Sene`ze, France is slightly younger (MN18), though located very close geographically to Saint Vallier (Fig. 8).
Saint Vallier, France is an open-air site along the Rhone River and specimens were deposited there in fluvial conditions, though they have little weathering or abrasion (Valli, 2004b). Saint Vallier, containing the remains of Eucladoceros ctenoides vireti, is the reference locality for biozone MN17 (Valli, 2004b). Gr aunceanu, situated in the Oltet¸ River valley in central Romania, represents the most fossiliferous assemblage of the Tetiou I sequence (R adulesco et al., 2003). This site is dated to the Early Pleistocene (MN 17, R adulesco et al., 2003). The specimens from this site are identified only as Eucladoceros sp. Sene`ze is faunally equivalent to Gr aunceanu (R adulesco et al., 2003), although younger. Specifically, Sene`ze must be younger than 2.1 Ma, as it was deposited in a tephra layer with this date and was accumulated in lacustrine (maar) deposits (Delson et al., 2006) with reverse (Matuyama) polarity, except for a normal section (Reunion) at the top (Roger et al., 2000). It is the reference for MN18, the Late Villafranchian (Delson et al., 2006) and contains Eucladoceros ctenoides ctenoides.
MATERIALS AND METHODS Here, cervid skeletal morphology is analyzed using GM. With GM shape is recorded as a series of 3D coordinates, which then [after generalized Procrustes analysis (GPA) and projection into a tangent space] are compared across the sample under investigation
300
CURRAN
Fig. 8. Locations of the three early Pleistocene sites in this study.
TABLE 2. Summary of fossil specimens in this study Site name
Location
Date
Taxa present
Cal LMs
Phalanx 3
Gr aunceanu St. Vallier
Romania France
MN 17 (2.4–1.9 Ma) MN 17 (2.4–1.9 Ma)
Sene`ze
France
MN 18 (2.2–1.5 Ma)
Eucladoceros sp. Eucladoceros ctenoides vireti Cervidae (large size) Eucladoceros ctenoides ctenoides Cervidae (large size)
17 3 6 2 0
24 0 0 6 2
Ma 5 Millions of years, CalLMs 5 Calcaneus landmarks.
(Cooke and Terhune, this volume). The entire shape of an anatomical region is analyzed with GM to find those specimens that group together (or apart) in morphospace due to similar (or dissimilar) morphology. GM provides a good starting place to assess how shape varies and then serves as a good tool to test if shape varies in predictable ways among functional groups. One of the advantages of GM is that morphological differences can be visualized using information about the interrelationships of the 3D coordinates under analysis and thus visualizations can then be used to test hypotheses regarding how shape varies. Quantification of complex shapes, sensitivity to subtle variation, and visualization of morphological variation make GM an important tool in the functional morphologist’s toolkit, and a powerful tool for examining shape variation in cervids.
The comparative (training) sample of extant cervid specimens (n 5 136) studied here was collected from four museums in the United States [American Museum of Natural History (New York), the Museum of Vertebrate Zoology at the University of California, Berkeley (Berkeley, CA), and the National Museum of Natural History (Washington, DC). the Chicago Field Museum (Chicago, IL), see Table 1]. Undamaged, adult, nonzoo, and nonpathological specimens with life history information were preferentially sampled. Every effort was made to sample evenly across the sexes, the four habitat categories, and the five cervid tribes. Data on fossil Eucladoceros specimens were collected on specimens from three early Pleistocene sites in France (Saint Vallier and Sene`ze, collections which are both split between Claude Bernard University of Lyon 1 and Collection du Musee des Confluences, Lyon) and
EUCLADOCEROS ECOMORPHOLOGY
Fig. 9. Phylogenetic tree showing relationships among taxa in this study; downloaded from 10 kTrees.org.
301
302
CURRAN
TABLE 3. Regressions of the natural log of centroid size on the natural log (ln) of several body size estimators in cervids Feature
Estimator
FemMed FemMed FemMed TibLat TibLat TibLat CalLMs CalLMs CalLMs Phal3 Phal3 Phal3
lnBW lnBH lnBL lnBW lnBH lnBL lnBW lnBH lnBL lnBW lnBH lnBL
R2
P-value
0.83 0.83 0.82 0.85 0.82 0.82 0.87 0.81 0.83 0.81 0.68 0.82
