Calibration of hybrid species distribution models: the value of general-purpose vs. targeted monitoring data

Aim Temporally replicated observations are essential for the calibration and validation of species distribution models (SDMs) aiming at making temporal extrapolations. We study here the usefulness of a general-purpose monitoring programme for the calibration of hybrid SDMs. As a benchmark case, we take the calibration with data from a monitoring programme that specifically surveys those areas where environmental changes expected to be relevant occur. Location Catalonia, north-east of Spain. Methods We modelled the distribution changes of twelve open-habitat bird species in landscapes whose dynamics are driven by fire and forest regeneration. We developed hybrid SDMs combining correlative habitat suitability with mechanistic occupancy models. We used observations from two monitoring programmes to provide maximum-likelihood estimates for spread parameters: a common breeding bird survey (CBS) and a programme specifically designed to monitor bird communities within areas affected by wildfires (DINDIS). Results Both calibration with CBS and DINDIS data yielded sound spread parameter estimates and range dynamics that suggested dispersal limitations. However, compared to calibration with DINDIS data, calibration with CBS data leads to biased estimates of spread distance for seven species and to a higher degree of uncertainty in predicted range dynamics for six species. Main conclusions We have shown that available monitoring data can be used in the calibration of the mechanistic component of hybrid SDMs. However, if the dynamics of the target species occur within areas not well covered, general-purpose monitoring data can lead to biased and inaccurate parameter estimates. To determine the potential usefulness of a given monitoring data set for the calibration of the mechanistic component of a hybrid SDM, we recommend quantifying the number of surveyed sites that are predicted to undergo habitat suitability changes.