A multiscale analysis of factors influencing Blackpoll Warbler occupancy and abundance during the non-breeding season in eastern Colombia

The dataset includes two survey methods to monitor Blackpoll Warbler populations and assess habitat structure in tropical ecosystems across Colombia. The first method involved surveys conducted along 128 transects (regional-scale data), each 100 meters in length and 50 meters in width, across multiple sites during two sampling periods (2017 and 2018). The transects, situated in both uniform habitats (foothill tropical rainforests and terra firma tropical humid forests) and transitional/agricultural systems, were surveyed by trained observers using visual and auditory species identification techniques. Each transect was surveyed at least twice per season, with multiple repetitions to reduce biases.

The second method, point counts (vegetation and landscape scale), was conducted across different habitat types in the Meta and Guaviare regions, with points strategically placed to ensure habitat homogeneity, at 26 sites using 160 point-count surveys, each within a 30-meter radius. Both transect and point count data were analyzed based on observer and repetition to determine species presence and absence. This combined methodology provides a comprehensive understanding of species distribution and habitat preferences across different landscape scales.

Methods
Regional scale data collection     

Surveys were carried out along 128 transects (6 to 11 transects per site), each 100 m long and 50 m wide, and separated by at least 200 m to avoid pseudoreplication (Bibby et al. 2000) and spaced equally along available access routes e.g., tertiary roads (unpaved), trails and footpaths, between January and March during 2017 (Sample A) and 2018 (Sample B). Surveys were conducted by four different observers trained in species identification by sight and sound. At least half of the transects (72) were located in blocks of uniform habitat, mainly foothill tropical rainforest (Boyacá-Casanare-Meta) and terra firma tropical humid forest (Guaviare), and the remaining transects were placed in transitional habitats or agricultural systems (e.g. forest edge, shade-grown cacao, cattle pastures; Table S1).

Transects were surveyed during 10 minutes by one observer slowly walking the entire transect and registering the number of Blackpoll Warblers encountered. All observations were carried out between 06:00 and 10:00 h (local time UTC-5). Each site was visited at least twice during the same season, with two repetitions of each transect per day, per visit, for a total of 8 repetitions per transect between January-February 2017 and 4 repetitions for transects in Boyacá between February and March 2018 (Table S1). Observers, as well as the order in which transects were visited, were randomized to minimize temporal or observer-based biases. Transects were not conducted in weather conditions that might affect detection probabilities, like rain or strong winds.

Landscape scale and vegetation structure data collection

We selected 26 sites within the departments of Meta and Guaviare that included one or more of the focal habitat types and established 160, 30-m fixed radius, point counts (Ortega-Álvarez et al. 2018), divided equally in 40 points per habitat (Fig. 1b). Site selection was carried out to minimize the distance between sites (and therefore the effect of broader elevation/precipitation patterns) but was driven primarily by the location of properties that contained the relevant habitats and granted access. Within properties, points were distributed such that the desired habitat was the dominant cover in a 60 m diameter around points and that point centers were at least 30 m from the edge of the habitat, as well as being 200 m away from nearby points (Bibby et al., 2000). Points were surveyed during 10 minutes on 6 occasions between January and March 2020 by three different observers. Repetitions were carried out over a two-day period, to ensure population closure, surveying each point three times a day between 6:00 h and 11:00 h (local time). Observers, as well as the order in which points were visited, were randomized to minimize temporal or observer-based biases.

The processed data are presented by transect and point for each scale, with the presence and absence for each repetition and for each observer.