Introduction

In recent years, developmental biology has made enormous strides in understanding the growth and modification of bones, and the constraints on bone growth as well (Hall 2005; Currey 2006). Most paleontologists no longer view bones as static entities, but as dynamic three-dimensional objects that can vary in shape not only during ontogeny, but also due to changed biomechanical forces. Because of both ontogenetic and ecophenotypic factors, bone shapes and sizes can vary quite widely in a population (Yablokov 1974). Shape and size of fossil bones are important factors in making taxonomic decisions, and assessing the variability of a single population is critical in deciding how much variability in a fossil sample can be attributed to a single species, or requires other explanations.

In particular, variability due to different styles of growth between endochondral bones (which ossify directly from an embryonic cartilaginous precursor, often constrained by joints and articular surfaces) and less constrained intermembranous bones, is highly relevant to these issues. The topic of intermembranous and endochondral bone growth, size and variability is one that is not commonly touched upon, except briefly in passing, in paleontological literature. Generally, intermembranous bones are measured and discussed as only a slightly relevant topic in regards to larger studies of species or interspecific variation and sexual size dimorphism.

Intermembranous bones form directly from the connective tissue late in embryological development and after birth through intramembranous ossification. Some intermembranous bones, such as the kneecap (patella), are almost always ossified in adult mammals (with minor exceptions). Other intermembranous bones, known as sesamoids, occur only in areas where a tendon passes over a joint, and ossify in irregular and unpredictable patterns (Vickaryous and Olson 2007). The number and shape of intermembranous bones vary greatly within the Mammalia, and are highly taxon-dependent. Humans have only one sesamoid (the pisiform) in the carpus. In many mammals, such bones include the patella and large sesamoids in the manus and pes. In ungulates, on the other hand, the only large sesamoid element is the patella. The sesamoids in the manus or pes are small nodular ossifications in the digital flexor tendons, both at the metapodial-phalangeal joint and the distal interphalangeal joint; suids have as many as 13 sesamoids in the manus alone.

Vickaryous and Olson (2007) point out that although sesamoids receive little attention in paleontological literature, the majority of tetrapod lineages develop at least one sesamoid. "As a group, sesamoids and their ilk represent something of an anatomical enigma, with an enormous degree of variability in size, shape, and position both within and between taxa. Consequently, most skeletal descriptions relegate these elements to passages that summarize... bones and cartilages, predisposing them to continued marginalization" (Vickaryous and Olson 2007). This scientific neglect is largely because sesamoids are not as commonly preserved as more massive and larger bones of skeleton, or sometimes cannot be reliably associated with a known species.

Recent anthropological literature has commented on patellar variability in humans of both recent and Pleistocene age. Trinkaus (2000) concludes that all of his samples "exhibit considerable variability in these patellar proportions." Trinkaus and Rhoads (1999) and Ward et al. (1995) also discussed the variability of fossil hominid patellae, but in the context of functional morphological interpretations, rather than comparative variability. The literature cited above suggests that because modern lineages show patellar variability, the Pleistocene fossil record may also provide data to suggest that sesamoids have been variable throughout the history of life.

Walmsley (1940), Bland and Ashhurst (1997), and Bongers et al. (2005) discuss the development of the patella and its ossification from connective tissues and hyaline cartilage. Sarin et al. (1999) and Goldberg and Nathan (2004) analyzed the variability of human sesamoid bones, but without comparing this variability to that of endochondral bones or using it in a systematic context. Our own impetus for this research was stimulated when we noticed a similar high level of variability in ground sloth patellae (Figure 1), and the second author has seen many similar instances in the large collections of fossils he has examined over the past 40 years. Prothero (2005, tables 5.1-5.9) documented some of this variability in North American rhinocerotid patellae.

Based on these considerations, intermembranous bones are predicted to show a higher level of variability than endochondral bones because they have limited articulation with other bones and are formed through intramembranous ossification. Many intermembranous bones are referred to as 'free-floating." By contrast, endochondral bones are more constrained from unusual growth by articulations with other bones. Apparent variability may also be the result of ossification into the tendons because animals tend to replace minor tissue damage with bone at the intersection of bone and tendon as an inflammatory response. Therefore, older individuals or individuals who have suffered tendon or joint injury would display larger or oddly shaped sesamoids (Andrew Clifford, personal communication, 2007).