Figure 3.

(a) A map of predicted expected marginal value for six different days in 2018, throughout the Greater Sydney Region, showing the highest valued sites that would optimize the collective knowledge on biodiversity trends throughout the Greater Sydney Region. This prediction step is dynamic: predictions are updated as new observations are submitted to the citizen science database. Expected marginal value maps will need to be updated fast in parts of the world where sampling rates are high, but this can be done at a slower rate where sampling is less frequent, and this will also vary among projects. (b) The changes in expected marginal value for 2018 for 10 randomly chosen grid cells at the 5 km² grid size. Some sites' expected marginal values remain relatively constant through time (e.g. grid 371) while others are highly variable (e.g. grid 274), and others undergo distinct step-changes (e.g. grid 405) corresponding to whether that grid was recently sampled or not. This is a dynamic approach, showing how the expected marginal value of a grid changes based on our parametrized model results (https://github.com/coreytcallaghan/optimizing-citizen-science-sampling/blob/master/Figures/dynamic_map.gif). (Online version in colour.)