Development of a photo-identification catalogue of hawksbill turtle (Eretmochelys imbricata) nesting population around Mahé (Seychelles) using and evaluating the software I3S Pattern

Individual recognition of hawksbill turtles, as for other sea turtles, is primarily achieved through various forms of tagging which can increase stress and possibly harm individuals. Especially in endangered species such as the hawksbill turtle, invasive techniques are morally and ethically questioned. Non-invasive photographic identification has successfully been used in a number of species and has recently been used for sea turtle research. The present study evaluated the performance of the novel matching software I3S Pattern, specifically designed for turtles, and created a catalogue for the nesting population around Mahé, including opportunistic in-water data. Various variables affecting matching performance were assessed including user skill level, area outlining, and sand and reflection coverage. Across the sand and reflection datasets, the software was shown to reliably identify an average of 95.15% of individuals within the top 50 with 80.72% of correct matches displayed on rank 1. Sand/reflection coverage in the area of interest and area outlining were factors shown to significantly affect performance. It is thus recommended to include only photographs with a sand/reflection cover of ≤20% in the area of interest which would further increase matching performance and to standardise area outlining via user training. The study aimed to provide a threshold for match uncertainty using the Information Criterion developed by Speed et al. (2007) for use with I3S Classic, however due to software differences further adaptation of the algorithm is required to provide a reliable threshold.

Promotor(s) & Supervisor: Joanna Bluemel

ISCED Categories

The highlighted icons, represent the fields of education (in compliance with ISCED Classification) engaged during this course/programme.

0511 - Biology", "0521 - Ecology