Maybe clarify that the GAM trends are not to be understood as an estimate of observations / occupancy

[bramdhondt]

On request, I wish to point out that the trend graphs might easily be misunderstood by readers, in that they might consider the GAM output to be estimates of either the true number of observations, or the true occupancy. This is not the case, since the true number must be above the observed number.

E.g. on the figure below, the true number of occupied cells in 2020 must be above 100 (the observed value). The value from the GAM at this point is indicative for the trend, but should not be interpreted in number. (Read further below graph.)

I support the idea of having the GAM line plotted on the raw data, but I do advice that, at least somewhere, it is mentioned that the Y axis does not apply to the GAM line. This can be included as a notice on the graph, or simply somewhere on the explanation page (maybe with the GAM values). I assume the GAM line is dimensionless; if not, it could be an option to include a second Y axis (right of graph).

Note that techniques exist to estimate the true occupancy. A recent contribution can be watched here (O. Gimenez, International Statistical Ecology Conference 2022, minutes 27:00 to 34:30). It uses Hidden Markov Models, and I think the search effort can be modeled in the same way as performed under the GAM. Maybe an idea for further IAS work...

[damianooldoni]

Thanks @bramdhondt! I think we can tackle this issue together with #72