Body Mass Prediction Model Using Skeletal Elements and Morphological Grouping

Abstract

Each facet of an organism's function, such as heart rate, metabolism, organ size and function, feeding ecology, and locomotion is influenced by its body size. Using body size to estimate the size and functions of organs in extant mammals is standard and reasonably straightforward. However, estimating the body size and organ function of fossil species, particularly transitional species such as the Eocene whales, is more challenging. The Eocene whales are diverse in body size, structure, locomotion, and habitat, therefore, a single allometric model does not provide the most accurate body size estimation for each fossil whale family. This in turn leads to less accurate organ size and function assessment. The overall goal of my research is to predict the body size of the Eocene whales and use these estimates to predict other organ functions. Grouping representative modern mammals to create body mass prediction models to use for the Eocene whales is the first step in this research project. I measured the bones of the skull, vertebrae, and limbs of select species of modern aquatic, semi-aquatic, and terrestrial mammals. I then calculated an average body mass for each species represented using recorded body weights. Using Principal Component Analysis (PCA), I generated morphological groupings of these mammals from the skeletal measurements. For each PCA analysis completed, I constructed a linear regression equation to predict body size from the variables used in each analysis and the average body weight. I tested the prediction models using the measurements of 12 species representatives that were not included in the data set used to create the model, specifically Odocoileus virginianus, Tragulus napu, Hippopotamus amphibus, Tapirus indicus, Sus scrofa, Canis lupus lycaon, Nyctereutes procyonid, Ehydra lutris, Gulo gulo, Neophoca cinerea, Lobodon carinophagus, and Ursus maritimus. Through PCA I identified five morphological groups from the mammals in the data set: Aquatic, Semi-aquatic, Semi-terrestrial, Large Terrestrial and Small Terrestrial. The test mammals all fit with their species. For the prediction model using all skeletal variables, the standard error of the estimate (SEE) for all 12 test individuals ranged from 0.3359 – 0.7496. For the skull variables, the SEE ranged from 0.1260 – 0.3197. For the limb variables, the SEE ranged from 0.1716 – 0.3964. For the vertebral variables, the SEE ranged from 0.1511 – 0.2851. These small SEE values and the large sample size substantiate the accuracy of these predictive models and the reliability of their respective confidence intervals (CI) and prediction intervals (PI) for the models. The results confirm that these linear regression models will be appropriate to predict the body mass of the Eocene whales in the second stage of this project.