Two- And Three-Dimensional Photogrammetric Mass Estimation Techniques For Two Phocid Species: Halichoerus Grypus And Phoca Vitulina Concolor

• Main Author: Harris, Jennifer
• Format: Text
• Language: unknown
• Published: DUNE: DigitalUNE 2012
• Subjects: non-intrusive mass determination; photogrammetric analysis; PGA model; seal; marine mammals; Marine Biology; Olympus; harbor seal; North Atlantic; Phoca vitulina
• Online Access: https://dune.une.edu/theses/12; https://dune.une.edu/cgi/viewcontent.cgi?article=1011&context=theses

Collecting mass measurements of seals is a common technique used to determine health. To determine the mass of Western North Atlantic Grey seal (Halichoerus grypus) and Atlantic Harbor seal (Phoca vitulina concolor) pups, a researcher must physically measure each animal. This produces stress for both pups and their mothers. Photogrammetric analysis (PGA) (evaluating photographs to obtain characteristics of a subject) has been used to determine physical measurements in a number of marine mammals. The purpose of this study was to develop a nonintrusive method for determining the mass of grey and harbor seal pups. Through this study we developed two- and three-dimensional PGA multiple regression models for predicting body mass of weanlings of both species. Photographs of grey seal pups were taken in the field and harbor seal photographs were taken in a captive setting. Calibration parameters were determined in Matlab and Olympus software, using an object of known measurement as a scale. Three-dimensional stereo-PGA was the most accurate close-range mass estimation technique. The most accurate grey seal model demonstrated significant agreement (p=0.006, r2=0.913) between predictions and the true population mean at a 95% CI. The harbor seal model with the highest accuracy demonstrated significant agreement as well (p= r2=0.904). Two-dimensional grey and harbor seal PGA models functioned best when used for distance PGA, predicting mass within 4% - 20% accuracy, at distances up to 22 meters. PGA models were validated through results of models created from physical measurements. For instance, a high correlation, Adjusted r2=0.885, was seen in harbor seal physical models, however a strong correlation, Adjusted r2=0.807, was seen in harbor seal PGA models as well. Models built in this study will be useful in future field and captive setting work with both species. Using these models for distance PGA purposes will also limit mother-pup disturbance.