Abstract
Recent work [1] has established Anatomical Network Analysis as a new method for generating hypotheses regarding anatomical
modules. In short, connections between differing skeletal components are visualised as branching networks, and hypothetical
anatomical modules are extracted depending on the density of connections between components. This is done by enumerating all
muscular or ligamentous attachments between bones, and bony articulations themselves. Networks for bony articulations themselves,
or networks which also include this soft tissue data are then generated. This method has the potential to be a robust way of
testing traditional ideas about modularity of different skeletal components by grounding it in actual anatomical data. Here, I extend
this idea through the interrogation of these hypothesised modules using geometric morphometrics and standard statistical tests for
modularity, including the RV coefficient and the newer technique of the covariance ratio. I apply this new integrated methodology
to two areas of interest to evolutionary anthropologists: the hominin upper body, including the thorax and shoulder. Using a modern
patient dataset from the National Cancer Imaging Archive (NCIA), and standardised anatomical descriptions of musculature
[2,3] it is demonstrated with the analysis of the thorax that common hypotheses of functional units are not born out by either
network analysis or subsequent geometric morphometric analysis. The costal skeleton in particular partitions into four modules in
network analysis. With regards the shoulder area, this is either integrated more with the thorax when muscle attachment data is
taken into account, or as a discrete unit if one only includes bony articulations. When the landmark data is analysed using the RV
coefficient and covariance ratio, overall morphological integration overrides this. I then apply the same technique to our closest living
primate relative, Pan troglodytes, and to hypothetical reconstructions of the fossil homini species Homo erectus, Australopithecus
afarensis and Australopithecus sediba. The results are discussed in the light of hypothesised arborealism in Australopithecus sp. and
obligate bipedalism in Homo erectus.
Acknowledgements:
Primate data is from KUPRI’s collection, Homo sapiens is from NCIA, Homo erectus scans are from research quality casts courtesy of R. Foley and M. Lahr (LCHES Cambridge); Au. afarensis is from a newreconstruction by the author; Au. sediba scans are from Morphosource.
References:[1] ] Esteve-Altava B, Boughner JC, Diogo R, Villmoare BA, Rasskin-Gutman D. 2015. Anatomical network analysis shows decoupling of modular lability and complexity in the evolution of the primate skull. PLoS ONE 10(5): e0127653 [2] Gray, H. 1985. Gray’s Anatomy. 15th ed. London: Chancellor press (reprint)[3] Musculino, J. 2005. The Muscular System Manual: The Skeletal Muscles of the Human Body. London:Elsevier.
modules. In short, connections between differing skeletal components are visualised as branching networks, and hypothetical
anatomical modules are extracted depending on the density of connections between components. This is done by enumerating all
muscular or ligamentous attachments between bones, and bony articulations themselves. Networks for bony articulations themselves,
or networks which also include this soft tissue data are then generated. This method has the potential to be a robust way of
testing traditional ideas about modularity of different skeletal components by grounding it in actual anatomical data. Here, I extend
this idea through the interrogation of these hypothesised modules using geometric morphometrics and standard statistical tests for
modularity, including the RV coefficient and the newer technique of the covariance ratio. I apply this new integrated methodology
to two areas of interest to evolutionary anthropologists: the hominin upper body, including the thorax and shoulder. Using a modern
patient dataset from the National Cancer Imaging Archive (NCIA), and standardised anatomical descriptions of musculature
[2,3] it is demonstrated with the analysis of the thorax that common hypotheses of functional units are not born out by either
network analysis or subsequent geometric morphometric analysis. The costal skeleton in particular partitions into four modules in
network analysis. With regards the shoulder area, this is either integrated more with the thorax when muscle attachment data is
taken into account, or as a discrete unit if one only includes bony articulations. When the landmark data is analysed using the RV
coefficient and covariance ratio, overall morphological integration overrides this. I then apply the same technique to our closest living
primate relative, Pan troglodytes, and to hypothetical reconstructions of the fossil homini species Homo erectus, Australopithecus
afarensis and Australopithecus sediba. The results are discussed in the light of hypothesised arborealism in Australopithecus sp. and
obligate bipedalism in Homo erectus.
Acknowledgements:
Primate data is from KUPRI’s collection, Homo sapiens is from NCIA, Homo erectus scans are from research quality casts courtesy of R. Foley and M. Lahr (LCHES Cambridge); Au. afarensis is from a newreconstruction by the author; Au. sediba scans are from Morphosource.
References:[1] ] Esteve-Altava B, Boughner JC, Diogo R, Villmoare BA, Rasskin-Gutman D. 2015. Anatomical network analysis shows decoupling of modular lability and complexity in the evolution of the primate skull. PLoS ONE 10(5): e0127653 [2] Gray, H. 1985. Gray’s Anatomy. 15th ed. London: Chancellor press (reprint)[3] Musculino, J. 2005. The Muscular System Manual: The Skeletal Muscles of the Human Body. London:Elsevier.
| Original language | English |
|---|---|
| Publication status | Published - 2016 |