Morphological-Based Variation of the Fish Populations Using Groupwise Registration; Applied to Microscopic Images of Fish Otolith Using Aphanius dispar as a Model

Abstract

The shape of hard structures such as scales and otoliths is often used to study fish population variations, palaeo systematics, and to discuss evolutionary changes. To date, different semi- and fully automated techniques have been developed to study shape variations, while none of them provided textural information from image samples. They can also be tedious to human error. Therefore, new methods are always required to provide this information precisely. Here we used a new framework as groupwise registration to evaluate if it is suitable to study the inter- and intra-population variations. To do this objective, we used the otoliths of a killifish, Aphanius dispar (Ruppell, 1829), as a model. This approach relies on visual comparison of the all otolith images from six isolated populations from Persian Gulf and Arabian Peninsula. The examined framework provides a pair of representative image and perturbation map from each population dataset. The representative image is optimally close to all input samples and the perturbation map indicates the local variance over the populations. Furthermore, this framework allows for measuring morphometric indices by finding corresponding points on the input samples, which eliminates the labor of repeating manual measurement and reduces human interaction error. As a result, the representative image and the perturbation maps were found to be considerably useful to study differentiation among A. dispar populations. This method is not only sufficient to investigate morphology of the otoliths, but also can widely be used to study other hard structures in fish and fisheries studies and paleontological investigations.