Habitatnutzung brütender Rotmilane (Milvus milvus): Evaluation der Habitatpotenzialanalyse anhand von GPS-Tracking-Daten
Description
Here, we publish the analysis scripts for our publication "Habitatnutzung brütender Rotmilane (Milvus milvus): Evaluation der Habitatpotenzialanalyse anhand von GPS-Tracking-Daten", along with the dataset with GPS data of red kites from 2000 - 2023 in Hesse, Germany. Since spatial usa data for the red kite is sensitive, we have removed the spatial reference from the data; therefore, it does not contain any coordinates. Our analysis is divided into two R scripts.
Script "01_analysis_absolute_gps_ha.R": Analysis of GPS-based habitat use intensity (GPS points per hectare)
This script analyses the habitat use intensity of Red Kites within the central (zPb) and extended (ePb) inspection areas, measured as the number of GPS points per hectare (GPS/ha). After loading and preparing the tracking dataset, habitat-specific GPS counts and area proportions are calculated for each individual, year, and inspection area. For the central inspection area, a Generalized Linear Mixed Model (GLMM) with a Gamma distribution is fitted to test whether GPS/ha differs significantly between habitat types (forest interior, forest edge buffer, and other habitats), with year as a fixed effect and individual as a random effect. The same modelling approach is applied to the extended inspection area, where habitat types additionally distinguish between grassland patches above and below the 10% area threshold and the 10 ha minimum size criterion. Post-hoc comparisons are conducted using estimated marginal means. For the extended inspection area, an additional linear mixed model tests whether GPS/ha in grassland habitats decreases with increasing grassland area proportion, providing evidence for or against a grassland area threshold effect. Model fit is assessed using residual diagnostics via DHARMa. Results are visualised as combined raincloud/boxplot figures and as a regression plot with confidence intervals.
Script "02_analysis_proportions.R": Analysis of area-based versus GPS-based habitat proportions
Script: Analysis of area-based versus GPS-based habitat proportions
This script tests whether the observed proportion of GPS points per habitat type differs significantly from the corresponding area-based proportion (as assumed by the Habitat Potential Analysis, HPA) within the central (zPb) and extended (ePb) inspection areas. After loading the tracking dataset, area-based (HPA) and GPS-based proportions are calculated for each individual, year, inspection area, and habitat type, and reshaped into long format for joint modelling. For grassland habitats in the extended inspection area, an additional threshold classification distinguishes whether the grassland area proportion exceeds or falls below the 10% threshold. For both inspection areas, a Beta Generalized Linear Mixed Model (Beta-GLMM) with a logit link is fitted, including year, habitat type, reference basis (HPA vs. GPS), and their interaction as fixed effects, individual as a random effect, and a method-specific variance structure to account for differences in dispersion between the two reference bases. Significant differences between HPA- and GPS-based proportions within each habitat type are assessed using pairwise contrasts based on estimated marginal means. Model fit is evaluated using residual diagnostics via DHARMa. Results are visualised as combined raincloud/boxplot figures for both inspection areas.
License
Except where otherwise noted, this item's license is described as Attribution-NonCommercial-ShareAlike 4.0 International