Animal density can be accurately estimated from camera trap photos. That is what our intern Lennart Suselbeek argues in his MSc thesis. Lennart took all 862 paca visits that our camera traps recorded in ten different 1-ha plots on BCI ; over 6000 camera traps nights in total. Lennart went through all photos to try individually recognize these animals by the unique spot patterns on their sides (see picture). He found 54 different individuals.
Then, Lennart estimated paca densities for each plot in two ways. First, he used a generally accepted Capture-Recapture model (SECR) that uses the individual identities. Then, he used a new “Random Encounter Model” (REM) that uses visitation rates, movement speeds and activity levels but not individual identities. The two estimates nicely correlated across the ten sites, which suggests that the REM, developed by our collaborators Marcus Rowcliffe and Chris Carbone, does a good job. This means that we can use camera traps to estimate densities of mammals that we cannot recognize.
And so did Lennart. He just received his MSc degree from Wageningen University and immediately started a PhD on seed dispersal by rodents with Patrick Jansen. Congratulations Lennart!