Browsing by Author "Holvast, Emma Jane"

Filter results by typing the first few letters
Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Taxonomic predictions of seabird identity using 3D bone shape : a thesis presented in partial fulfilment of the requirements for the degree of Master of Science in Zoology at Massey University, Albany, New Zealand
    (Massey University, 2021) Holvast, Emma Jane
    Identifying a species from an isolated bone can be an important first step in faunal analysis research. Identifications using molecular methods are not always possible (e.g. fossils), and specialist taxonomic expertise is not always available. Here we have investigated a shape-based (i.e. geometric morphometric) method for identifying taxa from a single bone that doesn’t require specialised taxonomic knowledge. Three-dimensional (3D) shape analyses are routinely used to study variation in bone shape among species, primarily to understand the role of environment and other drivers of shape evolution. Here we used 3D geometric morphometrics to predict taxonomic identities for seabird bones using partial least squares discriminant analysis (PLS-DA). We tested the hypothesis that digitised humeri and femora from seabirds can be confidently classified at order- or family-level using PLS-DA. Our analytical protocol aimed to accurately assign a bird bone to the correct taxonomic family, given that family-level analyses are useful for both modern and fossil bones. This protocol emphasised a minimal knowledge of osteological terminology on the part of the user, and instead used 3D geometric morphometrics and partial least squares discriminant analysis. Datasets of digitised seabird humeri and femora were generated from the collections at Auckland War Memorial Museum, Canterbury Museum and Te Papa Tongarewa Museum of New Zealand. Our novel taxonomic classification protocol used both well-established landmark-based techniques and novel landmark-free methods. Both the bootstrap and repeated cross-validation partial least squares discriminant analysis models as applied to both femur and humerus shape were highly successful in their ability to correctly predict new samples into their respective classes at both order- and family-level. Landmark-based PLS-DA assigned femora from penguins and tubenosed birds to the correct order and family with 100% and 93.5% accuracy, respectively. Pseudolandmark-based PLS-DA assigned femora and humeri from penguins and tubenosed birds to both the correct order and family with 100% accuracy. These results are a promising indication of the future utility of this method which may allow for taxonomic identification using more-automated landmark-free shape analysis techniques. Furthermore, this method could be used to produce finer-level classifications (genus or species), and could be applied to groups beyond seabirds.