Mammal Hip Morphology and Function: Coxa Recta and Coxa Rotunda



Using 15 parameters, we provide a systematic description of mammal proximal femoral morphology. We established two types of proximal femoral morphology, termed coxa recta and coxa rotunda, characterized by low versus high concavity of the head–neck junction. Concavity is a measure of the sphericity of the femoral head as it meets the femoral neck that can be quantified by angular measurements. We asked whether the parameter of concavity corresponds with the classification of mammal proximal femoral morphology based on coalesced versus separate ossification patterns and locomotor patterns. Statistical analysis demonstrated a distinction between coxa recta and coxa rotunda with significant differences between the two groups in all but 3 of the 15 parameters examined. We found the most discriminating measurement between mammal hips to be the concavity of the posterior head–neck junction (beta angle). Coxa recta (small concavity) and coxa rotunda (large concavity) relate to the ossification pattern seen in proximal femoral development, and species-specific patterns of locomotion. We interpret the two hip types to reflect optimization for strength (recta) versus mobility (rotunda). Conceptually, both hip types can be recognized in humans, where coxa recta can be related to the development of osteoarthritis. Anat Rec, 2013. © 2012 Wiley Periodicals, Inc.

Mammal hip joint morphology varies markedly between species. Some species have round femoral heads that may allow a large range of motion of the head in the acetabulum, for example, the large apes such as gorilla and chimpanzee. Others species have femoral heads that are not round at all, but oval or globular-shaped, for example, bovids such as bison.

To classify mammal hip morphology, Serrat et al. (2007) described two patterns of proximal femoral ossification. Briefly, the proximal femur starts with one chondroepiphysis, which subsequently separates into a trochanteric and capital physis, or remains as one coalesced epiphysis. These ossification patterns, separate and coalesced, afford only global description of proximal femoral anatomy, and appear to have no clear relationship to body size, phylogeny, or locomotion categories (Serrat et al., 2007). Conversely, when locomotion is assessed on a finer scale by looking at habitat specific patterns of locomotion, Kappelman (1988) found a clear relation between the type of hip loading and movement and femoral head shape within the family of bovids. Bovid species living in open habitats (savanna) depend on speed for survival, and therefore, rely on maximum power development at the hip. Their hips have laterally expanded (less spherical) femoral heads that largely restrict movement to the parasagittal plane, increasing locomotor power. In contrast, forest bovids have more spherical femoral heads allowing more abduction and rotational movement, facilitating movement through dense vegetation (Kappelman, 1988).

Human hip joint morphology can be quantified clinically by radiography, CT or MRI scans, but a comprehensive description of proximal femoral morphology in the normal human population was given only recently by Toogood et al. (2009) using 10 parameters. In this study, femoral head sphericity at the head neck junction was quantified by the parameter “concavity.” A similar, systematic description of mammal proximal femoral morphology is lacking, yet it may help better to describe and understand proximal femoral morphology in relation to ossification and locomotion patterns.

We asked whether the parameter of concavity corresponds with the classification of coalesced versus separate ossification pattern and locomotion patterns.

Specifically, we aim,

  • 1to provide a comprehensive description using 15 morphologic parameters of mammal proximal femoral morphology that can be used for large museum collections.
  • 2to asses which parameter distinguishes best between the two ossification patterns and locomotion patterns.


The selection of species was based on femoral ossification pattern and corresponding locomotion patterns. One group consists of cursorial and saltatorial mammals with a coalesced epiphysis, characterized by an aspherical femoral head at the junction of the head and the femoral neck (low concavity). Often, the head is positioned asymmetrically on the femoral neck and there is a straight section on the head–neck junction (Fig. 1). We use the term coxa recta for this type of proximal femoral morphology (Hogervorst et al., 2009).

Figure 1.

Coxa recta AP and lateral photographs (not to scale): (A) kangaroo (Macropus giganteus), (B) dog (Canis lupus familiaris), (C) bison (Bos bison), (D) horse (Equus caballus). Note oval femoral head and lack of offset. Also the coalesced epiphyseal line is visible.

The other group consists of aquatic/amphibious and climbing mammals. These femora have a spherical femoral head, positioned more symmetrically on the femoral neck. Due to the roundness of the head and symmetrical positioning on the femoral neck, there is high concavity of the femoral neck (Fig. 2). We use the term coxa rotunda for this type of proximal femoral morphology (Hogervorst et al., 2009).

Figure 2.

Coxa rotunda AP and lateral photographs (not to scale): (A) orangutan (Pongo pygmaeus), (B) chimpanzee (Pan troglodytes), (C) beaver (Castor fiber), (D) seal (Phoca Vitrulina). Note round femoral head and large offset.


Specimens studied are listed in Table 1, and form part of the collection of Naturalis, the Dutch Natural History Museum in Leiden, the Netherlands. We examined 76 dried femora of eight mammalian species. The recta group consists of 35 femora (bovids, carnivores, marsupials, equids), the rotunda group 41 femora (hominoids, rodents, pinnipeds). The sample was restricted to adults, assessed by epiphyseal closure. Sex was known for all primate specimens, not for the other mammals.

Table 1. Averages, standard deviation, and one-way ANOVA for each species per parameter
 Orangutan (Pongo pygmaeus) Average (SD)Chimpanzee (Pan) Average (SD)Beaver (Castor fiber) Average (SD)Seal (Phoca) Average (SD)Kangaroo (Macropus) Average (SD)Dog (Canis lupus familiaris) Average (SD)Bison (Bos bison) Average (SD)Horse (Equus) Average (SD)Sig.
  1. Sig., significance; F, F-distribution; Version, negative value denotes retroversion, positive value anteversion of the femoral neck.

CCD137.2 (4.5)126.8 (5.1)130.9 (5.8)118.0 (6.2)110.7 (1.8)127.2 (4.5)119.1 (4.0)118.3 (5.8)F(7,68) = 27.69, P < 0.001
MPFA87.2 (7.2)70.0 (6.5)60.4 (1.9)51.4 (6.1)47.9 (4.1)68.6 (3.1)56.7 (3.2)44.3 (3.2)F(7,58) = 79.05, P < 0.001
Alpha22.3 (2.4)24.8 (5.5)22.3 (8.4)43.8 (15.0)53.4 (7.4)17.8 (11.7)66.2 (10.7)26.0 (6.1)F(7,67) = 33.02, P < 0.001
Beta31.2 (3.4)38.5 (5.1)34.8 (8.5)30.8 (7.6)58.7 (3.3)63.2 (12.5)67.7 (16.0)74.4 (11.4)F(7,67) = 35.05, P < 0.001
Gamma36.3 (7.9)39.4 (2.2)46.4 (9.0)79.1 (18.5)97.3 (13.4)69.5 (12.7)92.7 (9.5)101.1 (12.0)F(7,68) = 51.22, P < 0.001
Delta40.7 (6.2)40.3 (3.3)41.7 (5.1)58.9 (6.4)50.8 (4.3)52.6 (6.9)67.3 (15.6)71.7 (4.8)F(7,68) = 26.09, P < 0.001
Epiphysis AP88.3 (5.6)85.7 (6.0)84.7 (8.8)81.5 (5.1)52.1 (3.3)76.3 (8.3)59.9 (6.9)74.1 (8.2)F(7,64) = 28.71, P < 0.001
Epiphysis LAT94.2 (5.0)95.8 (6.2)98.5 (13.0)98.7 (12.4)92.2 (7.1)104.3 (7.1)97.9 (10.5)88.6 (7.2)F(7,60) = 2.85, P < 0.02
Translation AP1.3 (2.4)0.0 (0.0)−1.5 (2.7)−3.4 (3.0)−12.5 (2.3)−3.7 (3.2)−4.0 (2.4)−5.9 (2.4)F(7,68) = 21.78, P < 0.001
Translation LAT3.9 (2.0)5.2 (3.5)6.1 (3.4)−4.7 (7.3)0.4 (4.0)12.4 (6.7)3.4 (3.4)12.6 (3.4)F(7,67) = 15.37, P < 0.001
Version−1.6 (6.9)0.4 (13.3)15.5 (7.5)−2.4 (5.7)27.6 (5.0)−8.1 (9.9)−0.3 (6.6)15.7 (5.1)F(7,66) = 19.36, P < 0.001
Offset anterior28.8 (3.2)27.6 (4.3)29.5 (7.3)15.1 (9.6)10.8 (3.4)32.7 (10.2)19.8 (4.4)24.7 (5.3)F(7,67) = 11.28, P < 0.001
Offset posterior22.9 (2.7)18.3 (2.4)21.3 (4.9)23.2 (5.7)11.0 (4.1)13.5 (4.3)16.6 (3.8)5.6 (3.2)F(7,67) = 23.31, P < 0.001
Offset inferior19.2 (7.0)16.9 (2.3)16.2 (2.4)6.9 (2.6)8.3 (3.7)9.6 (3.3)3.2 (2.0)3.0 (2.7)F(7,68) = 27.35, P < 0.001
Offset superior21.8 (6.1)19.1 (1.6)12.6 (2.7)2.0 (2.4)0.0 (0.0)4.1 (2.7)0.0 (0.0)0.0 (0.0)F(7,68) = 87.54, P < 0.001


Photographs were taken in two standardized positions. Anteroposterior (AP) photographs were taken with the femoral neck aligned parallel to the examination table, by placing 1-mm-thick cards under one of the femoral condyles. For femora with anteversion, cards were placed under the lateral femoral condyle. For femora with retroversion, cards were placed under the medial femoral condyle. Parallelism of the femoral neck with the examination table was assessed visually.

Lateral (LAT) photographs of the femoral neck were taken with the femoral neck aligned parallel with the edge of the table as seen from above. Both femoral condyles were resting flat on the examination table for this view. The camera was leveled with the examination table to allow measurement of version of the femoral neck (ante- or retro-version).

Reference of Proximal Femoral Morphology

We use the terms anterior, posterior, superior, and inferior to define positions in both mammal and human proximal femora (Toogood et al., 2009: #2889). Superior denotes (toward) the femoral head aspect of the femur, anterior toward the patellar aspect of the femur.


We used the GNU Image Manipulation Program 2.6 ( to measure all specimens. In Figs. 3 and 4, all measurements are outlined. To determine the center of rotation of the femoral head, a circle was drawn over the femoral head. Version of the neck was assessed on the lateral view; the neck-shaft angle was assessed on the AP view. Parameters studied and their results are listed in Tables 1 and 2. One observer measured all of the specimens. These measurements were in part repeated to calculate an intra-observer reliability. Also a second observer repeated these measurements to calculate an inter-observer reliability.

Figure 3.

1: Femoral neck axis (A), mpfa (B), ccd angle (C); 2: superior offset (de), inferior offset (fg); 3: gamma and delta angle; 4: epiphyseal angle (HA).

Figure 4.

1: Neck axis (A); 2: anterior offset (bc), posterior offset (de); 3: alpha and beta angle; 4: epiphysial angle (AF), anteversion angle (AG).

Table 2. Independent t-test for group differences coxa rotunda versus coxa recta
 Coxa rotunda Average (SD)Coxa recta Average (SD)Sig.
CCD129.0 (9.0)119.7 (1.2)t(74) = 4.90, P < 0.001
MPFA70.8 (15.4)55.2 (10.8)t(64) = 4.69, P < 0.001
Alpha angle28.2 (12.6)38.7 (22.2)t(73) = −2.55, P < 0.02
Beta angle33.4 (6.8)66.8 (13.0)t(73) = −14.20, P < 0.001
Gamma angle49.8 (20.3)89.3 (17.3)t(74) = −9.03, P < 0.001
Delta angle45.3 (9.5)61.5 (12.7)t(74) = −6.36, P < 0.001
Epiphysis AP85.4 (6.6)67.3 (11.9)t(70) = 8.04, P < 0.001
Epiphysis LAT96.5 (9.2)95.8 (9.9)t(66) = 0.30, P = 0.766
Translation AP−0.8 (3.0)−5.9 (4.0)t(74) = 6.34, P < 0.001
Trans LAT2.5 (6.0)8.1 (6.9)t(73) = −3.75, P < 0.001
Version2.6 (10.9)6.8 (15.0)t(72) = −1.42, P = 0.161
Offset anterior25.3 (8.6)23.3 (9.9)t(74) = 0.92, P = 0.362
Offset posterior21.6 (4.4)11.6 (5.6)t(64) = 8.64, P < 0.001
Offset inferior15.0 (6.4)5.9 (4.2)t(74) = 7.22, P < 0.001
Offset superior14.2 (8.7)1.2 (2.4)t(74) = 8.57, P < 0.001


Femoral neck axis.

To determine this axis, two or occasionally three circles were drawn in the neck. Then, a line was drawn, connecting the centers of these circles. Notably, in the coxa recta group, the neck axis can be hard to find due to the short and broad trapezoidal shape of the neck in extreme forms, for example, the kangaroo.


Anterior offset is the perpendicular distance (ab) between lines A and B, which are parallel to the femoral neck axis. Line A is tangential to the head; line B is tangential to the neck. Posterior offset is defined as distance (cd) between lines C and D. Similarly, superior and inferior offset can be measured on the AP photograph.

Concavity of the head–neck junction.

Concavity is a measure of the sphericity of the femoral head as it meets the femoral neck (Toogood et al., 2009). It is determined by the angle between two lines. The first line is the femoral neck axis, and the second line is drawn between the center of the femoral head, found by best-fit of a perfect circle over the head, and the point where the cortical surface of the femoral head exits this circle. These angles are referred to as the alpha, beta, gamma, and delta angles for the anterior, posterior, superior, and inferior aspect of the head–neck junction, respectively. These angles now allow easy quantification of the morphology of the head–neck junction and the femoral head itself, a region described mainly qualitatively before. These angles can be measured in large numbers of specimens, using photographs, radiographs, or CT or MRI scans.

Translation angle.

If the neck axis did not cross the center of the head, we measured the translation between these two points as an angle. The angle was measured at the point where the neck axis entered the circle drawn over the head of the femur (Fig. 5).

Figure 5.

Translation angle (black lines) in a horse. The angle is measured between the rotational center and the point where the neck axis enters the circle.

Epiphyseal angle.

When an epiphyseal line was recognized, the angle was measured between this line and the femoral neck axis.

Caput-collum-diaphyseal angle (CCD).

The angle between the longitudinal axes of the femoral neck and shaft.

Medial proximal femoral angle (MPFA).

The angle between the axis of the femoral shaft and the line between the major trochanteric tip and the center of the femoral head.

Statistical Analysis

To establish quantitative measurements to distinguish between the two types of proximal femoral morphology, we applied an independent t-test on each parameter to compare the recta group with the rotunda group. Ranges, means, and standard deviations were determined for each of the parameters.

To enhance the visual examination an additional test was done. We used a one-way ANOVA to compare the differences between the separate species. Post hoc tests with a Bonferonni correction were done to elucidate which parameter distinguished best between coxa recta and coxa rotunda.


Mammal Proximal Femoral Morphology

Using 15 parameters, mammal proximal femoral morphology is described in Table 1.

Differences Between Groups: Coxa Recta Versus Coxa Rotunda

We found the groups of coxa recta and coxa rotunda to be different in 12 of the 15 measured parameters (Table 2). Only anterior offset, lateral physeal angle, and anteversion showed no statistically significant difference between coxa recta and coxa rotunda groups.

Differentiating Parameters Between Coxa Recta and Coxa Rotunda

Each parameter measured showed a significant difference between species (one-way ANOVA, Table 1). Post hoc tests showed that of all parameters examined, the beta angle distinguished best between coxa recta and coxa rotunda (Bonferonni correction, Table 3).

Table 3. Bonferonni correction of the beta angle
Beta angleOrangutan (Pongo pygmaeus)Chimpanzee (Pan)Beaver (Castor fiber)Seal (Phoca)Kangaroo (Macropus)Dog (Canis lupus familiaris)Bison (Bos bison)Horse (Equus)
Orangutan/////////////   (P < 0.001)(P < 0.001)(P < 0.001)(P < 0.001)
Chimpanzee ////////////  (P < 0.01)(P < 0.001)(P < 0.001)(P < 0.001)
Beaver  //////////// (P < 0.001)(P < 0.001)(P < 0.001)(P < 0.001)
Seal   ////////////(P < 0.001)(P < 0.001)(P < 0.001)(P < 0.001)
Kangaroo(P < 0.001)(P < 0.01)(P < 0.001)(P < 0.001)////////////   
Dog(P < 0.001)(P < 0.001)(P < 0.001)(P < 0.001) ////////////  
Bison(P < 0.001)(P < 0.001)(P < 0.001)(P < 0.001)  //////////// 
Horse(P < 0.001)(P < 0.001)(P < 0.001)(P < 0.001)   ////////////

Ossification Pattern and Hip Morphology

In the coxa recta (cursorial and saltatorial mammals: horse, bison, dog, and kangaroo (Fig. 1), the epiphyses of the proximal femur are coalesced, that is, the ossification centers of the femoral head and greater trochanter are merged (Serrat et al., 2007). In the coxa rotunda (ambulatory/suspensory and aquatic/amphibious mammals: orangutan, chimpanzee, beaver, and seal), the epiphysis of trochanter and femoral head have become separated during growth (Serrat et al., 2007) (Fig. 2).

Inter- and Intra-Observer Variability

To quantify the degree of variability between observers and observations, an intraclass correlation coefficient (ICC) was calculated. The most important measurements are shown in Table 4.

Table 4. Inter- and intra-observer interclass correlation coefficient
CCD0.82, P < 0.0050.82, P < 0.005
MPFA0.99, P < 0.0010.98, P < 0.001
Alpha angle0.94, P < 0.0010.91, P < 0.001
Beta angle0.93, P < 0.0010.83, P < 0.005
Gamma angle0.96, P < 0.0010.96, P < 0.001
Delta angle0.84, P < 0.0020.80, P < 0.005
Offset anterior0.82, P < 0.0010.98, P < 0.001
Offset posterior0.91, P < 0.0010.94, P < 0.001
Offset inferior0.97, P < 0.0010.96, P < 0.001
Offset superior0.98, P < 0.0010.99, P < 0.001


We found the best discriminating measurement for the proximal femur to be the beta angle. We use the descriptive terms coxa recta and coxa rotunda to denote femoral head sphericity (or the lack of it), and the morphology of head–neck junction. The combination of femoral head sphericity (measured by α-, β-, γ-, and δ-angle) and the position of the femoral head on the femoral neck (measured by offset and translation angles) determines the concavity of the femoral neck with low and high concavity, respectively, in coxa recta and coxa rotunda.

Clearly, our findings are influenced by the selection of species examined, and we certainly do not intend to establish a dichotomy between two mammal hip types. We do suggest, however, that the coxa recta and rotunda hip types can help in the interpretation of form and function in mammal hips. We interpret these hip types as follows. A coxa recta is a sturdy hip seen predominantly in runners or jumpers. It has the widest possible femoral neck in order to withstand the high impact loading of ground contact (hoof or pawstrike), with jumping or (long-distance) running. It has low concavity and a “straight” (recta) section on the head–neck junction to better withstand the high forces at the postero-superior head–neck junction. We find this represented in a beta angle higher than 45° and a lack of superior offset. We hypothesize that, in quadrupeds, the postero-superior head–neck junction is where highest forces are found, for these mammals load their femur in a flexed position in a horizontally or obliquely moving pelvis.

A coxa rotunda is a “rotation hip” seen in climbers and swimmers. It has a narrower neck and high concavity with a round head (beta angle usually below 45°), to allow optimal axial rotation. This hip type has a large range of motion, specifically abduction and axial rotation, which can greatly benefit climbing, to grab or hold onto a substrate—but also swimming, for it allows rotation in a flipper to increase thrust.

Relating to human hip morphology, a coxa recta morphology is now associated with “cam type” femoroacetabular impingement (FAI) (Ganz et al., 2008). Cam type impingement denotes impingement between the aspherical portion of the femoral head (recta section) and acetabular rim, which may damage the acetabular rim cartilage. FAI is now considered an important mechanical cause for the development of hip joint arthrosis (Ganz et al., 2008).

In contrast to quadrupeds, the human femur is in a markedly more extended position in a vertical pelvis. The highest forces may then be found in the antero-cranial head–neck junction at heel-strike, that is, ∼90° shifted anteriorly on the femoral head–neck junction in comparison with cursorial mammals. Indeed, this is where a recta section can be found in the human hip (Ganz et al., 2008; Gosvig et al., 2010).

In humans, we find both types of proximal femoral morphology. More specifically, we find both recta (aspherical) and rotunda (spherical) femoral head morphology and these morphologies are gender-related. In population studies, recta morphology is found in ∼20% of males and 5% of females (Toogood et al., 2009; Gosvig et al., 2010; Reichenbach et al., 2010).

Given their functional significance, we interpret these mammal proximal femoral morphologies as type I traits, that is, a result largely from a direct interaction between genes expressed in local pattern formation during growth and development and the functional (connective tissue) biology of the adult (Lovejoy, 2005). In other words, they have a distinct genetic basis and are not largely the product of a connective tissue response to loading or other environmental factors.

We acknowledge several shortcomings of this study. First, the morphology addressed in this study is best observed with three-dimensional imaging. However, we believe that overall visual inspection allowed us to inspect the important landmarks, and adequately assess them in AP and lateral photographs. Using measurements from digital photographs as described by Toogood et al., allows quantification of large numbers of (museum) specimens, which may often not be available for analysis in CT or MRI-scanners. Second, sex was indicated in the primate species, whereas, in the other specimens, sex was not specified. We did not find sexual dimorphism of the proximal femur in the primates studied, other than overall size difference. Third, the sample size of the specimens was somewhat limited; however, with the numbers available, it does show different types of hip with significant differences between the two. Fourth, in contrast to humans, during stance, these mammals have their hips in flexed position in the acetabulum. The varying degree of hip flexion between species was not taken into account. We assume the difference in locomotion creates a different loading pattern, thus a difference in development of the hip joint. This is why, we speculate that when a coxa recta morphology is seen in the human hip, it is located in the antero-superior quadrant (vs. postero-superior in quadrupeds).


We thank Naturalis for access to the vertebrate collections and the use of photography equipment.