Automated Taxon Identification (ATI) systems that use a database to identify species or anatomical structures of species from different taxonomical groups have recently been developed. However, few of these works have been applied to marine organisms. In this paper we develop an ATI system for identifying Actynopterigian species from their otolith contour; this is important information for the palaeontological, ictiological and ecological sciences, especially in food web studies. The AFORO website comprises the first web-based automated species identification system based exclusively on otolith shape outline. Species are identified iteratively based on a multiscale representation of the otolith contour. The ATI system is very easy to use as it only requires uploading a suitably oriented otolith image on a black background. Two tests were carried out with a database of 1480 images of left sacullar otoliths (sagittae) from 420 species. The first test analysed 50 different otoliths, 10 per species, from 5 different species. The second test identified 50 otoliths, each from a different species. The two tests obtained similar results (percentage of correct identifications): 72% of specimens were correctly identified at species level, and the percentage increased to 90% at genus level and reached 94% at family level. The best results are obtained for species which have an otolith contour with a very specific morphology, such as the mackerel Scomber colias, and the less efficient results for species with common shaped otoliths with unclear specific characteristics, such as the Mediterranean horse-mackerel Trachurus mediterraneus, or species with large morphological variability between individuals, such as Lophius budegassa or Synodus saurus.

Peer Reviewed