Building and using detection models for ecological surveys – Cindy Hauser

Ecologists routinely conduct surveys to identify the presence, absence or abundance of a species they’re interested in. However there’s always a risk that they fail to detect what’s there: surveys can rarely cover the entire territory of interest; a bird might be silent or have moved to another part of its home range; a plant might be misidentified or simply overlooked. Furthermore, detection success often depends on variable conditions like the weather, the expertise of the surveyor or the surrounding terrain. To avoid biases and misleading conclusions, it’s important that such imperfect detection is accounted for when surveys are designed and survey data are analysed. This presentation will outline how these challenges have been addressed in a Victorian state government program to eradicate invasive hawkweeds from the Alpine National Park. We conducted field experiments to test the detectability of hawkweeds under a variety of conditions, fit statistical models to the experimental data, and employed optimisation methods to identify where Park surveyors can most efficiently search for and destroy hawkweeds.


Comments are closed.