Description: Cases of convergent evolution, particularly within ecomorphological contexts, are instructive in identifying universally adaptive morphological features across clades. Tracing of evolutionary pathways by which ecomorphological convergence takes place can further reveal mechanisms of adaptation, which may be strongly influenced by phylogeny. Ecomorphologies of carnivorous mammals represent some of the most outstanding cases of convergent evolution in the Cenozoic radiation of mammals. This study examined patterns of cranial shape change in the dog (Canidae) and hyena (Hyaenidae) families, in order to compare the evolutionary pathways that led to the independent specialization of bone-cracking hypercarnivores within each clade. Geometric morphometrics analyses of cranial shape in fossil hyaenids and borophagine canids provided evidence for deep-time convergence in morphological pathways toward the independent evolution of derived bone-crackers. Both clades contained stem members with plesiomorphic generalist/omnivore cranial shapes, which evolved into doglike species along parallel pathways of shape change. The evolution of specialized bone-crackers from these doglike forms, however, continued under the constraint of a full cheek dentition and restriction on rostrum length reduction in canids, but not hyaenids. Functionally, phylogenetic constraint may have limited borophagine canids to crack bones principally with their carnassial instead of the third premolar as in hyaenids, but other cranial shape changes associated with durophagy nevertheless evolved in parallel in the two lineages. Size allometry was not a major factor in cranial shape evolution in either lineage, supporting the interpretation of functional demands as drivers for the observed convergence. The comparison between borophagines and hyaenids showed that differential effects of alternative functional "solutions" that arise during morphological evolution may be multiplied with processes of the "macroevolutionary ratchet" already in place to further limit the evolutionary pathways available to specialized lineages.

Description: This paper presents a hybrid model for early onset prediction of driver fatigue, which is the major reason of severe traffic accidents. The proposed method divides the prediction problem into three stages, that is, SVM-based model for predicting the early onset driver fatigue state, GA-based model for optimizing the parameters in the SVM, and PCA-based model for reducing the dimensionality of the complex features datasets. The model and algorithm are illustrated with driving experiment data and comparison results also show that the hybrid method can generally provide a better performance for driver fatigue state prediction.