Research

Publications
Title: Marine mammal morphometrics: 3D modeling and estimation validation
First author: Zhang, Changqun; Zhou, Haojie; Christiansen, Fredrik; Hao, Yujiang; Wang, Kexiong; Kou, Zhangbing; Chen, Ruipeng; Min, Jie; Davis, Randall; Wang, Ding
Journal: FRONTIERS IN MARINE SCIENCE
Years: 2023
Volume / issue: /
DOI: 10.3389/fmars.2023.1105629
Abstract: Techniques of 3D modeling have earned increasing popularity in scientific studies as they offered unprecedented traits in representing objects. As with all mathematical models, the 3D model will be useful once its accuracy has been validated with direct measurements and the robustness of its predictive capability tested. Although measures of body mass and volume are essential to comprehend the life history of animals, such morphometrics, especially total volume, are challenging to obtain from marine mammals due to their elusive nature, aquatic lifestyle, and large size. In this study, accurate measurements of body volume were directly detected from fresh carcasses of eleven finless porpoises (Neophocaena asiaeorientalis sunameri) and used to validate 3D models recreated from the same animals using Blender 3D graphics. Published models using truncated cones or elliptical shapes, based on 3, 5, 8, or 19 measurements of girths or heights & widths along the body, were also applied to porpoises to verify their accuracy. The Blender-generated 3D model produced the most accurate estimates of body volume compared to conventional truncated models, with a mean error of only 2.5% to the direct volume measurements. When photogrammetric images are available, the model can predict the body volumes based on total length alone. Similar accuracy was possible with the elliptical model using 19 height & width measurements (5% increments in total length). However, significant (p < 0.001) inaccuracy resulted from truncated models with 3, 5, or 8 girth measurements and elliptical models with 3 or 5 height & width measurements, and the accuracy of these models also decreased significantly as the number of measurement sites was reduced. Moreover, Blender 3D models can be extrapolated to animals with images absent, and such predicted volumes were tested to be trustworthy (overall skill = 0.998, r = 0.998, p < 0.01). Therefore, we recommend that researchers use either digital 3D models or elliptical models with 5% increments sectioning the torso to obtain accurate estimates of the body volumes of free-ranging marine mammals.