Our abundance estimation work is focused on the challenge on translating capture rates into reliable measures of population density, even for species that cannot be recognized individually. For this, we use Rowcliffe’s Random Encounter Model (REM), a method that controls for sampling biases that influence how often a species is photographed, such as variation in the distance over which cameras detect animals. Our three test locations for this work are a homogenous forest island in Panama, a heterogeneous fenced park in the Netherlands, and a heterogeneous open system in The Netherlands.
Our animal ecology work focuses on habitat utilization, diel and annual activity patterns, and behavioural responses of wildlife. Here, we consider camera trapping complementary to animal tracking. Rather than tags that record how individuals move through all habitats, we use cameras to record how habitat patches are used by all animals. For example, we study how ungulates in National Park De Hoge Veluwe use habitat types of different quality and how this varies in space and time in response to disturbance. And we compared diel activity patterns of agoutis along a gradient of food availability in Panama. There, we also use camera traps to measure the effect of experimental liana removal on the use of tropical forest understory by terrestrial mammals.
In many projects, we use camera traps placed in a grid or at random points to measure the composition of the terrestrial mammal community in a standardized manner, so that it can be compared between sites. The largest project involves the survey of mammals across the CTFS-ForestGEO network of the Center of Tropical Forest Science. Morever, we used one-time surveys to compare forest sites across gradients of fragmentation in studies of ticks and infectious diseases in the Netherlands and in Panama, and to compare the mammal community of different-sized islands and different types of forest plantation in Panama.
We have several ongoing monitoring projects in which camera traps are deployed at the same points repeatedly or even continuously. In National Park De Hoge Veluwe, the Netherlands, we run a permanent network of 50+ camera traps to monitor habitat use by wildlife. In the tropical forest sites of the Tropical Ecology Assessment and Monitoring (TEAM) program, we annually survey wildlife at 60 permanent camera-trap points. In Barro Colorado Island, Panama, we have continuously monitoring of wildlife at 30 stations to study the consequences of the El Niño Southern Oscillation.
Frugivory & Seed Dispersal
A special use of camera traps is in our studies of frugivory and seed removal, in particular by scatter-hoarding rodents, such as agoutis and acouchies in the Neotropics and mice in Europe. We study fruit and seed removal with cameras aimed at feeding stations, food preferences using cafeteria-style experiments, and the fate of cached seeds by monitoring individual caches with cameras.
Part of our work concerns methodological development, on which we closely collaborate with colleagues from the Zoological Society of London, the Smithsonian Institution and the North Carolina Museum of Natural Sciences. Examples are the extraction of movement speed and activity patterns from camera trap data, image recognition and automated processing.
Digital camera trapping comes with the challenge of processing and storing vast quantities of photographs or videos. We manage Agouti, a web application with tools that allow scientists to quickly and easily process photographs, and safely store them on our servers. Agouti follows metadata standards and is fully compatible with the repositories of Smithsonian's eMammal and Wildlife Insights. The system is available for scientists, NGO's and commercial parties. We are working on making the system suitable for use by citizens, based on our experience with the volunteers that use Agouti to process the millions of photos from a project at National Park De Hoge Veluwe.