Spot-mapping:
As in the original study, the primary census method was spot-mapping. We spot-mapped nearly every morning from August 10th to November 8th, 2018 and replicated the original technique of estimating distance and direction to all birds detected perpendicular to the nearest trail markers found at 25 m intervals along all trails. Whenever possible, we recorded the position of counter-singing territorial neighbors, which were then used as boundary indicators for delimiting territorial boundaries. On each day, routes were randomized in their assignment to observers who rotated their starting and ending positions on each route to provide even coverage of the plot. Most routes were covered in the early morning hours, starting pre-dawn and ending two to three hours later. Routes were covered after dusk in order to census nocturnal species. Opportunistic detections at other parts of the day were also noted and observed. Over the 2018 field season, we logged 18,430 spot-map registrations of 224 species on the plot. These observations were digitized and corrected for observer sampling error (see Supplementary Information) and then used for several methods of comparison between census and the re-census.
After correcting for distance, we wrote a user-friendly, graphically interactive R program to detect and plot all counter-calling events and then produce automatically the map of each species recorded in 2018. These maps were inspected independently by AEM, SKR and JT to determine the number of territories within the plot. Knowledge of the natural history of each species, and, in particular, the number of individuals per territory/group, then resulted in a density estimate within the 97-ha plot, following the methodology used in Terborgh et al(1990). To carry out the Kernel Density Estimation (KDE) and analytical comparison of the geo-referenced observations in 1982 and 2018, we manually digitalized and geo-referenced the maps for 247 species produced in the original study. Producing these 247 maps involved writing and designing software-specific tools which are presented as supplementary material.
Mist-netting :
We replicated the sampling effort for each of the original mist-net lanes designated in the original study. We captured and banded birds as part of a standardized mist-netting regime, and also while target-netting understory mixed-species flocks throughout the entirety of the plot. We conducted standardized netting in seven locations on six trails, with lines of 18 mist-nets (12 x 3.2 m, 36 mm mesh) along 250-350 m of trail in each location, and operated the nets from dawn to dusk on three consecutive days when weather allowed, as in Terborgh et al (1990). During standardized netting we captured 360 different individuals of 79 species using the same methodology as the original study. When target-netting understory mixed-species flocks, we operated the nets in conjunction with playback from dawn until either we had captured a significant portion of individuals in the flock or until bird activity had slowed between the hours of 09:30 and 11:00. Target mist-netting was not conducted near any standardized mist-netting lanes. Core flock species and some additional species strongly associated with flocks (e.g., Isleria hauxwelli , Myrmoborus myotherinus ) were color-banded in order to facilitate following flocks and mapping out flock territories.
We were unable to band captured hummingbirds due to not having appropriate bands for the family, and, despite past use of the technique, we did not clip rectrices due to the high number of species that incorporate the tail in courtship displays. However, we photographed the right wing, tail, and body of all captured individuals, and with recent advances in the understanding of identifying individual birds through the use of digital photography (Pyle & Sullivan 2010; Nelson & Pyle 2013), we feel that a combination of age, sex, measurements, and molt pattern allowed us to identify all captured hummingbirds and to account for recaptures.
Colonial birds :
One observer surveyed the entire study plot for active nesting colonies of icterids. Once all colonies were found, the observer counted the number of active nests on the plot. The number of individuals of each species was pro-rated on the plot after taking into consideration foraging ecology and natural history observations compiled by one of the authors (Robinson 1985a, b; Robinson & Terborgh 1995).