The lineage of Thau Lagoon seahorses differs genetically and morphologically from the marine population of the same species (Hippocampus guttulatus), which makes this population a Evolutionary Significant Unit (ESU), endemic to some occitan lagoons (less than 150 km²) .
For several years, the Peau Bleue association has been interested in the monitoring and census of the individuals of this population and is studying it in order to ensure its conservation. To carry out this monitoring, the association relies on a photographic database from which they manually identify the individuals. The Hippo-ATLAS collaborative database now contains nearly 3,000 photographs of 98 individuals (average of 30 photographs per individual). In the same vein, the association has developed a participatory observatory, the Fish Watch Forum, which aims to collect and collect observations on marine fish from Western Europe and the Mediterranean.
This ability to identify individually the seahorses of a population allows to study life history traits (growth, reproductive cycles...), behavioral characteristics (inter-individual relationships, pair fidelity, day/night behavior...) or ecological characteristics (relationship to habitat and topography, mobility and their seasonal variations...).but also to determine more precisely the numbers of the Occitan population.
To meet this need, we would like to design a software for the photographic re-identification of individuals thanks to artificial intelligence. To realize this project, we plan first to train the algorithm to the recognition of seahorse species and then to move on to individual recognition.
* Evolutionary Significant Unit : Population units with significant adaptive variation based on the agreement between data sets obtained by different techniques (life cycle information, morphometric, range and distribution data, and genetic data).
To think together and discover how to create a re-identification algorithm for seahorses in order to refine their census, join us on Tuesday 3 August at 5 pm on https://meet.jit.si/HACK4NATURE for the launch of the challenge!
©P. Louisy / Peau-Bleue