A Critical Assessment of Sampling Biases in Geometric Morphometric Analysis: The Case of Homo erectusZhang, Chi (2020) A Critical Assessment of Sampling Biases in Geometric Morphometric Analysis: The Case of Homo erectus. Doctoral Dissertation, University of Pittsburgh. (Unpublished)
AbstractThis study primarily explores the potential of GMA in capturing and recognizing specific morphological features of specimens commonly allocated to H. erectus. For these purposes, four surface semilandmark datasets were collected for analyzing the gross morphology of the entire crania, posterior crania, temporal bones, and frontal bones. Results show that though dense surface semilandmarks can potentially capture many morphological features that cannot be included by using discrete landmarks, some shape differences visualized by PCA, include those in lateral cranial profiles, the trajectories of the squamosal sutures and the inclinations of the nuchal planes, do not match observed morphology. Procrustes superimposition is identified as an important, yet usually ignored, factor that may contribute to some incompatibilities between visualized shape differences and observed morphology. This is because Procrustes superimposition rotates configurations to different orientations to minimize their overall differences, while observed morphology is based on aligning specimens in the Frankfurt plane. This study also assesses three sampling issues associated with statistical analyses commonly used for testing whether H. erectus is too variable to be a single species based on landmark data. The first two are whether densities of semilandmarks and focusing on different cranial regions can yield inconsistent results of the same statistical analyses. The third is whether including a large sample of H. sapiens may affect results of PCAs. Results suggest that semilandmarks used in this study are oversampled, because using a much smaller number of semilandmarks can yield nearly identical results. The reason is that GMA places more weight on gross shape differences than details. Furthermore, the analyses of the temporal and frontal bones show results incompatible with those of the entire and posterior crania, likely because the former analyses focus on shapes of individual bones, while the latter analyses focus on gross cranial shapes. In addition, including a large sample of H. sapiens with fossils cause higher-ranked PCs to emphasize variations within H. sapiens. In this case, it is also important to examine lower PCs to understand shape differences between fossils. Share
Details
MetricsMonthly Views for the past 3 yearsPlum AnalyticsActions (login required)
|
![[feed]](/images/feed-icon-32x32.png){kind=icon}