Page 68 - A bird’s-eye view of recreation - Rogier Pouwels

A bird's-eye view of recreation
available from two sources: monitoring in sampling plots (‘BMP’) and large scale breeding bird surveys. Both the monitoring project and the surveys are based on the method of territory mapping (Hustings et al., 1985, Bibby et al., 2000). It involves standardised fieldwork and interpretation of the data to infer the presence and location of a breeding pair (Van Turnhout et al., 2008). For the monitoring scheme 6–12 visits are made to the sampling plots by volunteers coordinated by Sovon, Dutch Centre for Field Ornithology. The plots vary in size between 10 and 500 ha. Breeding bird surveys by professionals usually have less visits (3-5) and cover larger areas (>500 ha). Both data sources deliver number and location of breeding territories in specific areas that reflect breeding bird densities. From plots with multiple censuses in different years we used only the most recent one. Plots were excluded when they were smaller than 25 ha or larger than 1000 ha, contained less than 10% heathland or had extreme values for recreation densities. This selection resulted in 61 available plots for further analysis.
4.3.2.2 Habitat and other environmental variables
Bird occurrence can be explained by a number of habitat related factors such as vegetation type and soil type. Based on the ecological requirements of the bird species under study a number of potentially explanatory habitat variables were included in the statistical models describing land use and specific habitat features and soil condition (Appendix 5). These data were retrieved from a set of GIS maps (CBS, 1985, CBS, 2008, Clement, 2001, De Vries et al., 2003). Information about the disturbance by roads was also included in the analysis because traffic disturbance has been shown to affect the presence of these species (Reijnen and Foppen, 1995, Reijnen et al., 1996). For the traffic variable we used the maximum noise level (in dB) within the survey site.
4.3.2.3 Statistical procedure
For each species we constructed statistical models to link the abundance of the species in a particular sampling plot to the chosen explanatory variables. We developed generalized linear models with a Poisson distribution and log-link function using the statistical program ‘R’ version 3.2.2 (R Development Core Team, 2015). In the models the number of breeding birds in a plot was used as the dependent variable. In order to adjust for temporal trends in population levels the year of the census was included in the analysis. The mapping protocol was also included in order to adjust for differences in densities between the two bird monitoring datasets that were used. Area of the plot was included to account for the differences in plot size, but also added as a variable, Area2, to account for possible non-linear relationships with area (Nee and Cotgreave,
66