Identifying individual animals over a period of time can provide information on population size, and individual survival amongst other things and as such is a key tool for conservation. But how do you identify animals that basically look alike? In turtle conservation throughout the world, much effort is focused on the use of marker tags and in these species the easily accessible part of the population is nesting females.
In some places, such as the Aldabra Atoll World Heritage site in Seychelles, juveniles are caught in the reef flats for tagging and weighing. Mature adult males, however, are pretty much inaccessible not only because they occur in deeper waters, but also swim much faster and weigh significantly more, making it hard to “throw” them into a boat for weighing, measuring and tagging.
Juvenile turtles are easy to catch in shallow lagoons for mark and recapture studies, photo Pierre André Adam.
Given these limitations, a number of organisations around the world have looked at ways of using photographs to identify individuals and this has now been tried with turtles. In Hawaii, for example, researchers looked for patterns in the scales and have been able to track turtles for over 15 years.
Clothahump, was first identified in 1988 (left) off Honokowai, West Maui in Hawaii and re-sighted in 1993 (right), photo Turtle Trax.
This method of photo identification, however, is subjective. Claire Jean, Project Officer at Kelonia, in association with the Information Technologies department at the University of Reunion have devised a new method which uses the number, location and shape of scales from the left and right side of the turtle's head to identify individuals in a population. So far, this method has been tested and validated on Green turtles and Claire is being assisted by Cyrille Lebon, an MSc student from the University of St Etiennes in France, in this task .
With the recent submission of over 70 images by MCSS , Cyrille is doing further testing to see if the same method can be applied to Hawksbill turtles. Despite having fewer scales to code, preliminary results show that the method should work on this smaller species, provided that the image is complete i.e. nesting turtles are usually covered in sand and often, their neck is slightly retracted during laying thus obscuring outlying scales.
Images of nesting turtles are not always usable because they are partly obscured by sand, photo Uli Enfont.
In addition to testing the coding method on turtles, Claire and Cyrille are also testing humans to see how easy or hard it is to teach others to use this method. Using a standard protocol (written in French unfortunately!), volunteers were asked to code 16 photographs of turtles after reading the protocol. Despite her limited French, Elke also volunteered for this test.
Early results show that while the protocol is well annotated and generally well understood, a few exceptions stipulated in the photo-ID programme require additional clarification, meaning proper training is indispensable. However, as Cyrille is doing his thesis on validating the Kelonia photo-ID protocol, he needs figures to back up this claim, so at the recent photo ID mini-symposium held at the University of Reunion (see coming blog), he had a captive audience on which to re-test the protocol, this time with training.
Cyrille presenting the turtle photo-ID protocol for re-testing (with training) on University students, photo Claire Jean.
Once again, Elke was roped into the test. Her results for test 1 (without training) were; 71 errors; 17 missing scales and 11 added scales, while for test 2 (with training); 64 errors, 10 missing scales and 9 added scales … and according to Cyrille, these were the best results from all participants for the two tests!!
HELP US KEEP TRACK OF OUR TURTLES:
If you would like to assist with this programme, please feel free to send us your photographs of turtles. Simply include: your name, the date and location of the photograph and the turtle’s behaviour at the time (i.e. feeding, resting, swimming etc.) and we will include it in the rapidly growing database.