Protected Areas (PAs) are spatially representative management tools that impose various levels of protection for conservation purposes. As spatially regulated places, ensuring compliance with the rules represents a key element of effective management and positive conservation outcomes. Wildlife crime, and in particular poaching, is a serious global problem that undermines the success of PAs. This study applies a socio-ecological approach to understanding the opportunity structure of illegal recreational fishing (poaching) in no-take zones in Australia's Great Barrier Reef Marine Park. We use Boosted Regression Trees to predict the spatio-temporal distribution of poaching risk within notake Marine National Park zones. The results show that five risk factors account for nearly three quarters (73.6%) of the relative importance for poaching in no-take zones and that temporally varying conditions influence risk across space. We discuss these findings through the theoretical lens of Environmental Criminology and suggest that law enforcement strategies focus on reducing the negative outcomes associated with poaching by limiting the opportunity of would-be offenders to undertake illegal activity.