This probably goes without saying, but...it's been a year. One for the books. At Piedmont Wildlife Center, some things have changed: we've gone virtual for our Animal Encounter programs, and camps continue to operate under COVID-19 protocols. Some things have remained the same: our box turtle study is going stronger than ever, and the turtles in the park are blissfully unaware of any sort of chaos taking place in the world.

As fall turns into winter and the turtles are heading into brumation (a fancy word for reptilian hibernation), we're hunkering down by the computer to do some data analysis. Thanks to folks who have submitted turtle photos through our citizen science project, Triangle Turtle Trekkers, and photos our Conservation Team has taken of turtles at Leigh Farm Park, we have a large set of photos to explore with FIT software, developed by our research partner, WildTrack.

Normally used for analyzing animal footprints, FIT has been specially modified for us to use for analyzing box turtle shells. The software is designed to perform hundreds of measurements with each shell photo, and groups photos together by similarity. The goal of FIT is to see how well the software can identify individual turtles from photographs of the shell.

We've performed FIT on all of our resident turtles at PWC, and the software grouped them correctly with no problem. This time around, we tried using hundreds of photos of wild turtles we have marked during our study. We know which turtle is which because of the markings, but the software doesn't use that; it's just taking measurements.

The results were very interesting--I used photos from 16 individuals, but the software predicted there were 21. Some groupings contained multiple individuals, incorrectly identifying them as one; other individuals were predicted to be multiple.

So why was the prediction so far off? I have several sneaking suspicions. For one, this was my first time performing this analysis; it may be that I need some more practice with the software to improve the accuracy of the predictions. The other possible explanation is that this was the first time the software was presented photos of the same individual over a span of several years. The photos of our resident turtles were all taken on the same day. Could it be that age plays a factor in its ability to accurately make predictions?

The only way to find out is to keep collecting and analyzing data, and your support is what enables us to do so! Without it, we would not have the tools necessary to collect and analyze this invaluable data on box turtles and utilize cutting-edge monitoring methods.

Thank you for your investment in the future of the eastern box turtle, and have a happy hibernation (er, holiday)!