There is growing awareness about the potential benefit of harnessing data collected or generated by citizens. This is particularly attractive in fields that are traditionally data-poor, such as environmental and conservation sciences. However, the use of this data source requires careful attention to data quality. A common example of this issue in conservation is imperfect detection of a cryptic species or imprecise estimation of a quantity of interest. In this presentation, we focus on statistical methods that integrate different sources of data, such as ground-truth observations and citizen-provided data, and account for the differential quality impacts. We describe these approaches in the context of three of our projects: creating a jaguar corridor across the Peruvian Amazon using the knowledge of local people and international experts; monitoring the health of the Great Barrier Reef using observational data, UAVs and divers' photos; and counting koalas in Queensland using drones and VR.