Skip to main content
Dryad logo

Data from: Convergence analysis of a finite element skull model of Herpestes javanicus (Carnivora, Mammalia): implications for robust comparative inferences of biomechanical function

Citation

Tseng, Zhijie Jack; Flynn, John J. (2015), Data from: Convergence analysis of a finite element skull model of Herpestes javanicus (Carnivora, Mammalia): implications for robust comparative inferences of biomechanical function, Dryad, Dataset, https://doi.org/10.5061/dryad.90005

Abstract

Predictions of skull biomechanical capability based on virtual models constitute a valuable data source for testing hypotheses about craniodental form and feeding behavior. Such comparative analyses also inform dietary reconstruction in extinct species. 3D modeling using Finite Element (FE) methods is a common technique applied to the comparative analysis of craniodental function in extinct and extant vertebrates. However, taxonomically diverse skull models in the literature often are not directly comparable to each other, in part because of distinctions in how boundary conditions are defined, but also because of substantial differences in the number of FEs composing the models. In this study, we test whether a conventional convergence test is adequate in identifying the minimum number of FEs needed to achieve internally stable results for a single species. We constructed a series of skull models of Herpestes javanicus, and simulated unilateral biting across the dentition; the models differed in the number of FEs, degrees of freedom at the joint and bite point constraints, and type of tetrahedral FEs used. We found that convergence patterns differed across constraint types, FE quantities, and bite position simulated. Four-noded tetrahedral (tet-4) FE models with relaxed constraints produced the most stable measurements compared to over-constrained tet-4 models and to relaxed tet-10 models. In absence of an optimal FE quantity from convergence testing, we propose a broadly applicable sub-sampling protocol, whereby average measurement values across multiple models per specimen are used for among-species comparisons. A regime of sampling three low FE quantity models produced the closest estimates of mean measurement values relative to larger model sets, being within the 95% bootstrap estimated confidence intervals. Future studies should focus on identifying sources of variation associated with other FE modeling protocols, so that they can be accounted for before biomechanical attributes from these simulations are used to infer form–function linkage.

Usage Notes