Surveys of Forest Birds on Puerto Rico, 2015

Christopher C. Rimmer, John D. Lloyd, and Jose Salguero Faria

The study area consisted of 211 locations on Puerto Rico where we conducted point-count surveys for birds during January–March 2015. Survey points were located by first drawing a generalized random tessellation stratified (GRTS) sample of 1-km² cells that were identified from the Bicknell’s Thrush winter-habitat model (McFarland et al. 2013). We then visited each cell and identified 3-5 locations suitable for point-count surveys. Suitability was based on extent of forest cover – at least 50% of the area within 50 m of the point was forested – and accessibility; all points were along public roads or trails. To maintain independence of counts conducted at different points, we placed each point at least 250 m from its nearest neighbor. We were able to establish survey points and conduct surveys in 43 different cells. Most points were in submontane evergreen forest or lower montane elfin, sierra palm, transitional, and tall cloud forest (Kennaway and Helmer 2007). The elevation of survey points across all cells ranged from 0–1,297 meters above sea level (MASL), with a median of 705 MASL.

National Fish and Wildlife Foundation grant no. 48324 and private donations from friends of the Vermont Center for Ecostudies

Surveys were conducted in forested areas at a variety of elevations on Puerto Rico.

We conducted standardized point-count surveys for birds at 211 locations on Puerto Rico during January – March 2015. Survey points were located by first drawing a GRTS sample of 60 1-km² cells that were identified from a predictive model of Bicknell’s Thrush winter habitat (McFarland et al. 2013). The sample included an unequal mix of cells with a high (>0.5; n = 30), medium (0.25–0.50; n = 20), and low (<0.25; n = 10) predicted probability of occurrence of Bicknell’s Thrush. Once we had drawn a sample of cells to survey, we visited each cell and identified 3-5 locations suitable for point-count surveys. Suitability was based on the extent of forest cover – at least 50% of the area in a 50-m radius around each point was forested – and accessibility; all points were along public roads or trails. To maintain independence of counts conducted at different points, we placed each point at least 250 m from its nearest neighbor. Survey points were georeferenced via GPS. Of the 60 selected cells, we were able to visit and survey 43 during 2015 Fig. 1. Most (n = 29) of the cells that we sampled had a high predicted probability of occurrence of Bicknell’s Thrush; relatively fewer of the sampled cells had a medium (n = 5) or low (n = 9) predicted probability of occurrence. Survey points within the cells covered a wide range of elevations: from 0–1,297 meters above sea level (MASL), with a median of 705 MASL. Exact geographic coordinates of each survey point are contained in the data package available at KNB (see link below in section 'Data resources').

Figure 1.  

Location of 1-km2 cells on Puerto Rico surveyed for forest birds during 2015. Geographic distribution of sampling locations was based on the predicted probability of occurence of Bicknell's Thrush: 29 cells had a high predicted probability of occurrence of Bicknell’s Thrush (black dots); 5 had a medium (black triangles) and 9 had a low (black squares) predicted probability of occurrence (note that GPS coordinates were not recorded for 1 cell in each category, so only the 40 cells with coordinates are shown).

At each survey point, an observer conducted 4, 2.5-minute counts during which all birds heard or seen were recorded (i.e., a repeated-counts design). Counts were conducted in sequence, one after the other. A 1-minute playback of Bicknell's Thrush vocalizations preceded the second and fourth counts. Individuals of all species detected during each count were recorded into four distance bands: 0–10 m from the observer, 10–25 m, 25–50 m, and >50 m from the observer. All surveys were conducted between sunrise and 10:00 Atlantic Standard Time. Surveys were not conducted during rain or high winds.

Step 1. Identification of sampling locations.

We downloaded the raster output of the winter habitat model of Bicknell's Thrush (McFarland et al. 2013) from Data Basin (http://databasin.org). Each 1-km² cell in the raster included an associated predicted probability of occurrence of suitable habitat. We selected 60 cells in which to conduct point-count surveys using a GRTS sampling design. The sample included 30 cells with a high (>0.50) probability of containing suitable habitat, 20 with a medium (0.25–0.50) probability, and 10 with a low (<0.25) probability.

Step 2. Establishing sampling points.

We visited 43 of the selected cells during 2015 (due to the paucity of detections of the focal species, Bicknell's Thrush, we did not return as planned in 2016 to sample the remaining cells). In each cell, we identified 3–5 locations suitable for conducting point-count surveys. Point locations within each cell were chosen systematically based on ease of access - all points were located in forests adjacent to public roads or trails - and so as to cover as much of the cell as possible. Each point was also placed so that forest cover within a 50-m radius was at least 50%. We located survey points >250 m from one another so as to maintain the independence of counts at each point. The location of each point was established in the field using a handheld GPS unit. Coordinates were not properly recorded for 28 points and so the exact the location of the survey point is unknown.

Step 3. Conduct bird surveys.

At each survey point, an observer conducted 4, 2.5-minute counts during which all birds heard or seen were recorded (i.e., a repeated-counts design). Counts were conducted in sequence, one after the other. A 1-minute playback of Bicknell's Thrush vocalizations preceded the second and fourth counts. Individuals of all species detected during each count were recorded into four distance bands: 0–10 m from the observer, 10–25 m, 25–50 m, and >50 m from the observer.

Puerto Rico

17.375 and 19.125 Latitude; -67.875 and -65.0 Longitude.

This dataset is released to the public and may be freely downloaded. Please keep the designated Contact person informed of any plans to use the dataset. Consultation or collaboration with the original investigators is strongly encouraged. Publications and data products that make use of the dataset must include proper acknowledgment.