The recent study of Morimoto et al. (2011) on proximal femoral muscle attachment patterns in modern great apes contains evolutionarily important observations. However, their article includes specific statements about chimpanzees' proximal femoral muscle attachments that are at present unsupported, as well as unwarranted depictions of our previous study. We here outline the relevant issues and propose alternative observations and interpretations.
Direct associations of muscle distribution/attachment patterns and specific osteological features need to be known for an accurate interpretation of bone morphology. However, as stated in their article, the relationships of the two are not necessarily reliably established. We therefore welcome their innovative study using computed tomography (CT) imagery. In particular, we find it intriguing that chimpanzees and gorillas may superficially share similar osteological patterns (as suggested by Lovejoy et al., 2002), but in fact differ significantly in their actual muscle deployment patterns.
However, as discussed below, we believe that both the utility and the limits of CT imagery (in part in relation to resolution) need to be more carefully considered, with more adequate crosschecking using actual dissections. For example, it is often difficult to distinguish among: 1) muscle tissue superficially juxtaposed on a bony surface but encased within its own fascia that in turn lacks significant connections to the underlying bone or periosteum, 2) muscle fibers and/or their fibrous extensions that attach directly into the periosteum but not the underlying bone, and 3) muscles with fibrous/tendinous attachments (Sharpey's fibers) that actually penetrate the underlying bone. The criteria followed in delineating specific muscle attachment areas are not necessarily spelled out (e.g., Raven, 1950; Sigmon, 1974; Swindler and Wood, 1982), but presumably have been based primarily on 2) and 3).
PROXIMAL FEMORAL MUSCLE ATTACHMENT PATTERNS
There appears to have been some misunderstanding of what was intended in our former descriptions (Lovejoy et al., 2002, 2009a). Morimoto et al. (2011) depicted alternative muscle attachment patterns of the proximal femur and attributed to Lovejoy et al. (2002, 2009a), an unusual pattern in which the vastus lateralis (VL) arises from the posterior-medial aspect of the upper femoral shaft (their Fig. 1C Pan and Gorilla). They then rejected this pattern as being unsupported by their observations. However, in actuality, Lovejoy et al. (2002, 2009a), had based their descriptions on previously published studies (Raven, 1950; Sigmon, 1974; Swindler and Wood, 1982). They did not advocate the muscle attachment pattern reported and then criticized by Morimoto et al. (2011).
In Lovejoy et al. (2002) pages 104–105, the osteological features under consideration were described as follows:
As it descends the shaft, in all specimens examined, this lateral spiral pilaster courses between (and thereby separates) two areas of muscle attachment on the proximolateral aspect of the femur. One, which lies more superiorly and posteromedial to the pilaster, is a rough, reticulated surface whose upper portion is often a continuation of the trochanteric origin of the vastus lateralis. It blends inferiorly with the rugose insertion of the adductor brevis. There is some terminological confusion with respect to the latter. Raven referred to this muscle in the gorilla as the adductor minimus (Gregory, 1950), though Sigmon (1974) stated that no pongids have an adductor minimus. Swindler and Wood (1973) made no comment on the subject, while Woodburne (1978) noted that the adductor minimus of humans is merely the upper transverse fibers of the adductor magnus. As these details do not concern us here, we will simply refer to the adductor that inserts here as the brevis. The lateral spiral pilaster lies between this complex area of rugosity and a second one for insertion of the gluteus maximus (superficialis) that is anterolateral to it. This insertion also runs several centimeters superoinferiorly. As the pilaster courses between (or passes through) these two muscle attachment areas, it divides one from the other, though it itself is rarely roughened in a manner that would imply penetration by Sharpey's fibers.
The above discussion, and other brief descriptive passages in Lovejoy et al. (2002, 2009a), were not intended to be a comprehensive description of femoral muscle attachments. For example, we did not mention the VL attachment of modern apes that runs along much of the femoral shaft (Raven, 1950; Uhlmann, 1968; Sigmon, 1974; Swindler and Wood, 1982), because it does not impact our considerations of proximal femoral shaft morphology. Morimoto et al.'s (2011) Fig. 1C can be taken to suggest that we considered apes to lack such an attachment, which is certainly not the case. More importantly, Morimoto et al. (2011) mistakenly attributed to Lovejoy et al. (2002, 2009a) the opinion that modern apes have VL attaching superomedial to the lateral spiral pilaster (LSP). Again, we did not intend this. The confusion may have stemmed from the following passages of our previous publications:
…a rough, reticulated surface whose upper portion is often a continuation of the trochanteric origin of the vastus lateralis. It blends inferiorly with the rugose insertion of the adductor brevis…
These passages were apparently interpreted by Morimoto et al. (2011) to mean that the VL itself continues posteromedially, whereas our intention was that the adductor complex attaches at this “superior broad roughened area” and that this rugose surface feature is sometimes osteologically continuous with more superolaterally located area where the trochanteric origin of the VL arises.
To summarize, the muscle attachment areas that correspond to the actual (and/or intended) descriptions of Lovejoy et al. (2002, 2009a) are schematically shown in Fig. 1. Although Morimoto et al. (2011) attributed a chimpanzee/gorilla “muscle topography” sequence of VL-LSP-GM to Lovejoy et al. (2002, 2009a), this is simply not the case. The muscle attachment pattern that corresponds to the Lovejoy et al. (2002) description is best schematized as (ADD)-LSP-GM-VL or (ADD)-LSP-GM/VL; the GM/VL notation implies that the two muscles attach adjacent to each other or to a common intermuscular septum.
Our interpretation of proximal femoral muscle attachment areas, as shown in Fig. 1, actually differs little from Morimoto et al.'s (2011) preferred modern chimpanzee pattern, but an important difference is the position of the gluteus maximus (GM) attachment relative to the bony “boss” that Lovejoy et al. (2002) termed as the “lateral spiral pilaster” (LSP). In contradistinction to our interpretations, Morimoto et al. (2011) consider the GM to attach medial to the LSP, and hence an (ADD)-GM-LSP-VL configuration. In short, they assert that the osteologically prominent rugosity anterolateral to the LSP in chimpanzees corresponds to the VL and not to the GM. We regard this interpretation of the CT imagery as unconfirmed (see below).
Also, note that although Morimoto et al. (2011) described the human pattern as GM-LP-VL, this must also be incorrect. Morimoto et al. (2011) depict the human lateral pilaster (LP) as occurring between the GM and the VL attachments (their Fig. 1C and D of Homo). However, in human femora, the smooth bony prominence that Lovejoy et al. (2002) termed the LP must underlie the VL, whose fibrous attachment occurs at or adjacent to the lateral margin of the gluteal tuberosity itself (e.g., Goss, 1973). Then, the correct sequence notation of the human attachment pattern is (ADD)-GM-VL-LP or (ADD)-GM/VL-LP. As described by Lovejoy et al. (2002), the LP is variably present and primarily represents platymeric (i.e., mediolateral) expansion of the diaphysis (Fig. 2)—if the above interpretation is correct, the LP is not homologous to the LSP.
INTERPRETATIONS OF CT IMAGERY
In our previous studies, we considered the proximal femoral muscle attachments of chimpanzees and gorillas to exhibit a common pattern (Lovejoy et al., 2002, 2009a). Although we did not refer to orangutans in our previously published articles, we had observed that orangutan femora also exhibited the same pattern. Specifically, in extant great apes, a variably defined bony prominence or area (the LSP) that apparently lacks Sharpey's fiber penetration occurs between: 1) a supero- and posteromedially situated rugose area of bone, presumably for fibrous attachments of the adductor complex; and 2) a somewhat infero- and anterolaterally located fossa-like rugosity, presumably for a Sharpey's fiber attachment of the GM. We interpreted all three great apes to share this pattern, as it is schematically shown in Fig. 1 for chimpanzees (also, Fig. 5 of Lovejoy et al., 2002).
According to Morimoto et al. (2011), the CT data support the above-suggested muscle pattern in gorillas and orangutans, but not in chimpanzees. In particular, and as noted above, Morimoto et al. (2011) describe the chimpanzee GM as inserting medial to the LSP. However, the examination of two of the same CT data sets used by Morimoto et al. (2011) (available online, courtesy of Primate Research Institute, Kyoto University) suggests that their conclusions are premature. These are the only two nonfixed adults studied, and the examination of successive CT sections perpendicular to the femoral shaft shows that either the GM itself or a broad intermuscular septum continuous with the GM attaches within the rugose “fossa-like” area anterolateral to the LSP (Fig. 3). This is concordant with the GM attachment pattern suggested by Lovejoy et al. (2002, 2009a).
Further superiorly along the LSP, the fossa-like structure disappears, and the prominence of the LSP itself is better defined. In this area, according to Morimoto et al.'s (2011) segmentation of the lateral intermuscular septum (or its superior extension), the latter (and hence the GM) putatively attaches medial to the LSP. On the contrary, Lovejoy et al. (2002) hypothesized that a smooth bony surface such as that of the LSP should be largely devoid of Sharpey's fiber attachments. Our own examination of the CT imagery of this upper LSP region (Fig. 3) suggests that, if the lateral intermuscular septum did, in fact, attach along the LSP, this could have been at (or even at the lateral margin of) the LSP, rather than medial to it. From the examination of successive slice imagery, we do not consider the CT data sufficient to resolve either the presence/absence or the position of actual fibrous attachments at and around the upper LSP. Such judgments involve subtle interpretations of CT data, necessitating validation by combining CT observations with parallel dissections of the same specimens. However, we agree that the chimpanzee GM insertion can be more variable than previously thought, with individuals that exhibit the GM/VL insertion more posteromedial than is “typical,” occupying a location, for example, usually taken by the inferior portion of a LSP (Fig. 4).
To summarize the subtrochanteric topographic relationships of the relevant features, the chimpanzee sequence can be considered potentially different from both gorillas and humans as follows (medial to lateral):
humans: GM-VL-LP or GM/VL-LP
gorillas (and orangutans): LSP-GM/VL
chimpanzees (this study): LSP-GM/VL, in part LSP/GM/VL
chimpanzees (Morimoto et al., 2011): GM-LSP-VL (at least superiorly)
Meanwhile, pending validation studies that combine CT imagery and parallel dissection of the same specimen, the differences of opinion regarding the chimpanzee condition can be considered as alternative hypotheses of inter-relationship between GM/VL muscle attachments and osteological features. It is possible that normal variation spans all of the chimpanzee patterns suggested above.
In any case, the above considerations establish that the chimpanzee pattern is distinct from the human one in the relative positions of the GM/VL and LP (lateral in humans) or LSP (medial at least relative to VL). Furthermore, the gluteal complex involves other significant features (such as the development and structure of the cranial maximus, e.g., Stern, 1972; Sigmon, 1974), which should impose caution on any simple equation of human and chimpanzee patterns. We therefore consider Morimoto et al.'s (2011) evolutionary interpretations that are reliant on their reassessment of chimpanzee GM insertions as yet unsubstantiated.
At the same time, we agree with Morimoto et al. (2011) that the chimpanzee pattern does appear to be characterized by more medial and less anterolateral GM attachments than occurring in gorillas and orangutans, and with variation that includes GM attachment along the LSP (or equivalent location). The magnitude of positional difference between chimpanzees and gorillas should be quantifiable, and such would further strengthen Morimoto et al.'s (2011) observations that the gorilla and chimpanzee in fact differ substantially in muscle deployment patterns. The orangutan condition also appears to differ from that of the gorilla in details. Although orangutans appear to share with the gorilla a strong anterolateral position of the GM insertion, they may differ from gorillas in the supero-inferior location of the superior GM insertion (for details on position of GM tendinous insertion, see Stern, 1972).
Finally, which pattern was primitive, and which derived, and what is the evolutionary significance of all this? Morimoto et al. (2011) proposed three hypotheses: 1) H1, humans, and chimpanzees share the primitive condition; 2) H2, the gorilla/orangutan condition is primitive and that chimpanzees and humans are each derived; and 3) H3, the gorilla/orangutan condition is primitive and chimpanzees and humans represent a shared-derived condition. Although they preferred their H3, by considering their putative human–chimpanzee pattern shared derived, one must note that this was based on their coded “topographic pattern,” which needs substantial revision as discussed above. Our interpretations (discussed above) of the relevant musculoskeletal morphology suggest that the chimpanzee and human conditions are distinct, and probably independently derived (see below).
Morimoto et al. (2011) considered the LSP and LP to be developmentally programmed osteological structures and therefore useful units of analysis. In contrast, these bone features are notable primarily by their lack of surface texture, and as such are more likely a passive consequence of the musculoskeletal developmental program of an entire bony region (e.g., proximal femoral shaft). Although configuration and topographies of fibrous tissues (such as muscle fascia and tendinous attachments) are probably under relatively direct influence of the developmental program (i.e., primarily reliant for expression on positional information established in the anlagen, Lovejoy et al., 2003), we hypothesize that LP and LSP are localized areas of enhanced subperiosteal bone deposition (i.e., “bossing”) that reflect interactive downstream readouts, involving complex inter-relationships of various epigenetic effects. It should be noted in this regard that the disposition of muscle attachments must be constantly modified during growth to maintain positional relationships, and structures such as the LP and LSP are most likely to be relatively passive consequences of such surface (i.e., Sharpey's fiber) rearrangements. Therefore, Lovejoy et al. (2002) used nonconventional bony terminologies, such as the LP and LSP, to avoid the problems of past terminologies that often imply homology. In contradistinction, Morimoto et al. (2011) considered the human LP and the chimpanzee LSP to be homologous, which we consider unlikely based on the above-discussed developmental background and the extensive differences seen between chimpanzees and humans in proximal femoral musculoskeletal patterns.
The GM with a distinct ascending tendon inserting at the uppermost lateral shaft is considered as the primitive primate condition (Stern, 1972). Judging from osseous morphology, the >4.4 Ma hominids, Ardipithecus ramidus and Orrorin, appear to largely retain this primitive pattern. This inference is based on a distinct third trochanteric expression (or homologous strong impression) at the upper lateral shaft, which is widely seen in Miocene apes such as Proconsul (Ward et al., 1993), Nacholapithecus (Rose et al., 1996), and Dryopithecus (Moyà-Solà et al., 2009). On the contrary, the extant great apes, including the chimpanzee, can be considered derived in their distally displaced insertion of the cranial GM. In extant great apes, not only does this insertion occur at a position distinctly lower than the lesser trochanter, there is usually no expression of a third trochanter or a homologous strong bony rugosity.
In light of these observations, we find that Morimoto et al.'s (2011) study is highly insightful in showing that chimpanzees appear to differ from gorillas and orangutans in lacking a strong anterolateral displacement of the GM insertion (and the entire lateral intermuscular septum). Such differences suggest that distal migration of the cranial GM insertion likely occurred independently in chimpanzees and gorillas, in concert with the many other observations that support the hypothesis of considerable parallelism in the modern chimpanzee and gorilla postcranium (detailed in White et al., 2009; Lovejoy et al., 2009b). Relative to that of known basal hominids (Orrorin/A. ramidus) and many Miocene apes, the chimpanzee GM insertion appears derived, but less so than it is in either gorillas or orangutans.
Although Morimoto et al. (2011) considered it unlikely for gorillas and orangutans to have independently acquired similar muscle–bone topographies, such apparent identities in coded character states need not be homologous. In fact, judging from the differences seen in the entire GM structure of gorillas and orangutans (Stern, 1972; Sigmon, 1974), it is probable that the gorilla and orangutan conditions occurred independently. In particular, the extreme gorilla condition might have resulted from an allometric effect that necessitated disproportionate hypertrophy of the adductor/hamstring complex, thereby relocating the lateral intermuscular septum and GM insertions laterally (and anterolaterally). The orangutan condition might have been a part of a larger scale restructuring of the hip and thigh muscles, perhaps in relation to its notably enhanced hip joint mobility (e.g., MacLatchy and Bossert, 1996).
The human pattern is best considered a modified version of the primitive A. ramidus pattern, matched (albeit with some variation) by post-4 Ma Australopithecus, and characterized by hypertrophy of the cranial part of the GM with a strong ascending tendon insertion at the third trochanter/gluteal tuberosity/hypotrochanteric fossa complex. Some of these aspects were summarized in Lovejoy et al. (2009a)'s study as follows:
…the gluteal complex in Ar. ramidus remains anterolaterally displaced as in Proconsul and Orrorin and still unlike its more posteromedial position in most Australopithecus. Such medial translation of the maximus insertion is probably a consequence of hypertrophy of the quadriceps at the expense of the hamstrings. Indeed, the combination of a broad, ape-like, expansive ischial tuberosity and broad proto-linea aspera in Ar. ramidus suggests that the hamstring/quadriceps exchange had not yet achieved its modern proportions, although some expansion of the maximus was probably present given the substantial restructuring of the ilium and trans-iliac space. In contrast, most Australopithecus specimens (such as MAK-VP-1/1, A.L. 333w-40, and A.L. 333-110) had marked elevation and narrowing of a true linea aspera that is typical of later hominids.
Morimoto et al. (2011) discussed the implication of their results in assessing how the hominid pattern of gluteal/thigh muscle “topography” and “allocation” might have evolved. Based on their (in our view unsupported) hypothesis of chimpanzee and human similarities, they argued that the muscle topography characteristic of hominids must have evolved prior to the human–chimpanzee split, and therefore must have been independent of any adaptation to bipedality. Although our revised interpretation of muscle attachment patterns fails to corroborate such propositions, we close this critique by pointing out that the evolutionary polarity of muscle attachment patterns summarized above is consistent with the hypothesis that a significant shift occurred between A. ramidus and Australopithecus, in both “topography” and “allocation” of hip musculature. In addition to the gluteal morphology detailed above, this is inferred from the drastic shortening and morphological change of the lower pelvis (ischium) and posterior expansion of the ilium, as outlined in the studies of White et al. (2009), Lovejoy et al. (2009a), and Haile-Selassie et al. (2010).
The authors thank the authors of the Morimoto et al.'s (2011) paper for a candid exchange of opinion regarding these issues, and for assistance in accessing the Kyoto University Primate Research Institute CT data sets. The authors also thank M. Nakatsukasa and T. Kimura for discussion, and H. Endo for access to comparative specimens.