Please use this identifier to cite or link to this item: http://hdl.handle.net/20.500.12136/2087
|Title:||Evolutionary novelties and geometric morphometrics: a case study of lower second molar morphology|
Bermúdez de Castro, José María
Olejniczak, Anthony J.
|Publisher:||Institut Català de Paleontologia|
|Citation:||I Iberian Symposium on Geometric Morphometrics, 2009, 71-73|
|Series/Report no.:||Paleontologia i evolució. Memòria especial;3|
|Abstract:||The study of evolutionary novelties, especially structures that arise or disappear during evolution, are diffcult to approach since geometric morphometric methods are based on comparisons of homologous landmarks. Such novelties often have been excluded from morphological studies, although their presence is sometimes recorded insofar as they affect the location of adjacent landmarks (e.g. MARCUS et al.,2000). Nonetheless, a variety of homology-free methods for the study of outlines (e.g. BOOKSTEIN, 1997) and surfaces (GUNZ et al., 2005; POLLY, 2008, among others) have been proposed recently, with the aim of explicitly including novel structures in geometric morphometric analyses. These methods have been criticized, however, because they rely on mathematical criteria of correspondence among points rather than on the biological criterion of homology (KLINGENBERG, 2008). The use of land-marks that delimitate the novel structure when present, and fall on the same point when absent, has been suggested to perform better (KLINGEN-BER, 2008). The aim of this study is to explore and compare different approaches to the study of evolutionary novelties through the analysis of the human lower second molar morphology. This tooth has five main cusps in all Australopithecus and Early Homo specimens, whereas the fifth cusp (c5) is typically missing in more recent Homo species (i.e., H. heidelbergensis, H. neanderthalensis and H. sapiens). A sample of 131 molars of several species of genera Australopithecus, Paranthropus and Homo was analyzed using three approaches: 1. only 40 semilandmarks were recorded, as an example of a homology-free approach (“Semi” in table 1); 2. only 17 landmarks were recorded, including type III landmarks and those which delimitates the fifth cusp as proposed in KLINGEN-BERG (2008) (“Land” in table 1); and 3. both landmarks and semilandmarks were recorded, excluding those related to the c5, as in MARCUS et al. (2000) (“Semi+Land no c5” in table 1). Additionally, three combined approaches were tested: 4. all the landmarks (including those related to the c5) and 40 semilandmarks, using as a sliding criterion the minimization of Procrustes distance (“c5 Proc Dist” in table 1); 5. the same landmarks and semilandmarks as in #4, above, but using the minimization of bending energy as the criterion to slide semilandmarks (“c5 BE” in table 1); and 6. forty semilandmarks and all the landmarks not related to the fifth cusp plus the landmark located at the point where the central developmental groove is intersected by the disto-buccal developmental groove, (BIGGERSTAFF, 1969), which is slid towards the molar periphery as the c5 is reduced in size or disappears (“Beginning c5” in table 1). In each case, the sample was superimposed using a Generalized Procrustes Analysis (ROHLF & Slice, 1990). A relative warp analysis (BOOKSTEIN, 1991) was also carried out in each case, testing the correlation among the PC1 and PC2 scores. A canonical variates analysis, together with an assingment test, was performed to determine which method yields the highest percentage of correct classification. The results of these analyses are summarized in table 1. The correlation among PCs is higher in the analyses which include the c5, since all of them have a leading PC (explaining more than 50% of the morphological variance) related to the loss of that cusp. However, the assignment test based on the CVA shows that the highest percentage of correctly assigned specimens corresponds to the analyses that do not include landmarks related to the c5, and to the analysis including only the single landmark corresponding to the beginning of the fifth cusp. Repetition of this analysis of human lower second molar morphology using different landmark and semilandmark conformations reveals that the best results (in terms of higher percentage of correct classification) correspond to evaluations which explicitly do not include the studied evolutionary novelty (the loss of the c5). The homology-free method based on the analysis of semi-landmarks is not particularly informative when applied to the human molar outline, although the percentage of correct classification are close to those attained when including only landmarks. Since the presence or absence of c5 corresponds to the leading PC and CV, and has intraspecific variability in the later Homo groups, the analyses which include this cusp are largely showing the intraspecific variability of these groups, which have the largest sample sizes. We can hypothesize that the inclusion of this evolutionary novelty in the landmark configuration following KLINGENBERG (2008) might provide a clear separation of taxonomic groups if the evolutionary novelty were a distinguishing characteristic of some species. In this study, the best way to recover interspecific differences seems to be the combination of a set of semilandmarks describing the shape of the outline and a group of landmarks that do not include any point related to the evolutionary novelty, or which include just one point that shows its size, when present.|
|Description:||Comunicación presentada en: I Iberian Symposium on Geometric Morphometrics, 23-25 de julio 2009, Sabadell (Barcelona)|
|Appears in Collections:||Congresos, encuentros científicos y estancias de investigación|
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.