Population trend assessment on a large spatial scale: integrating data collected with heterogeneous sampling schemes by means of habitat modelling

The assessment of population trends is fundamental in conservation science, but requires a monitoring programme with a specific sampling scheme that is often unrealizable due to high operating costs. For this reason, and considering that historical data would be very valuable for the assessment of overall trends, we developed a method for analysing population trends by using census data collected with the same survey technique, but acquired with different sampling strategies. Data on three bird species (skylark Alauda arvensis, swallow Hirundo rustica and nightingale Luscinia megarhynchos) coming from different point count surveys performed in Lombardy (Italy) between 1992 and 2005 were used. We corrected the bias due to the different sampling schemes by defining a population index as the ratio between the numbers of observed and expected (from a habitat suitability model) bird pairs in each point count. For each species, the overall trend for the period was assessed by fitting the mean population index calculated for each year. Population indices showed a significant negative trend for the skylark and nightingale and a marginally significant negative trend for the swallow. Trends were more negative in marginally suitable than in highly suitable habitats for all species. Validation of the population indices using an unbiased independent data set showed the effectiveness of our method in estimating population variations. However, its application must be considered carefully when the explained deviance of the suitability model is low or when land use may have changed substantially.