Dataset DOI: 10.5061/dryad.2ngf1vj33

Description of the data and file structure

We compiled data on understorey bird captures using mist-nets at 67 Amazonian forest sites, including 716 plots (86,278 individuals captured representing 646 species) incorporating information from published articles, MSc and PhD theses, unpublished data from collaborating researchers, environmental licensing data generated by major infrastructure projects, and the avian ecology sampling conducted by the Amazon Biodiversity and Carbon (ABC) Expeditions Project. Filtering our data excluded 23 sites, 484 plots, 415 species, and 51,556 individuals, which amounted to a total of 232 plots retained in our analyses of eco-morphological traits (including non-terra firme plots). We excluded 24 sites, 505 plots, 456 species, and 56,984 individuals, which amounted to a total of 211 plots retained in our analyses of environmental and spatial predictors of species richness (only terra firme plots).

Table_S1.csv. Bird sampling plots compiled for this study (n = 211). The number of individuals refers to the total used (only understorey birds) after the filtering process described in the main text.

Table_S2.csv. Species list and eco-morphological traits used in our analyses, including those extracted from range maps and observed data (n = 284). Taxonomy follows HBW & BirdLife International (January 2024).

Body mass = species traits extracted from Wilman et al. (2014)

Trophic level = From the proportional use of food items available in Wilman et al (2014), we converted categorical dietary data into a continuous measure of trophic level (see Bueno et al. 2018), where the sum of the percentages of each food item consumed by each species was weighted by an energy level: (1) foliage and other vegetative plant materials, (2) fruits and nectar, (3) seeds, (4) invertebrate prey, and (5) vertebrate prey. For example, an insectivorous species relying entirely on invertebrates was assigned a value 4 (1.0 × 4), whereas a species relying equally on fruits (50%) and invertebrates (50%) was assigned a value 3 [(0.5 × 2) + (0.5 × 4)].

Number of habitats = number of distinct habitat types used by each species, extracted from Stotz et al. (1986).

Geographic range size: information extracted for each species from Tobias et al. (2022).

Table_S3.csv. Table used in Species richness analyses.

We distinguish two spatial scales of sampling: (i) plots refer to individual sampling units (typically, a net-line or the smaller sample unit in each study, which may be a location containing several net-lines), and (ii) site pertains to an Amazonian landscape, protected area, municipality or other spatial designation adopted by individual authors to identify their study area. For instance, one of our sites consists of the 10,000-ha Adolpho Ducke Forest Reserve, in which 87 plots of 250 m each were sampled.

Relative_abundance = calculated for each plot.

Observed richness = numbers of species observed in each plot

RM richness = number of species extracted from range maps

GSI richness = estimated species richness using Geometric Series Index

All environmental variables are described bellow:

Aboveground biomass (AGB) can be used as a measure of forest structure and complexity (Zhang et al. 2013) and is related to the spectrum of resources available for aboveground consumers (MacArthur and MacArthur, 1961; Zelaya et al. 2022). We extracted AGB estimates derived from a combination of Earth observation data, the Copernicus Sentinel-1 mission, and the Envisat ASAR instrument (see Santoro and Cartus, 2023) for the years 2010, 2017, 2018, 2019, and 2020, depending on the time of bird sampling. 

Canopy height (CH) data was extracted from the Global Land Analysis and Discovery (GLAD) dataset. The maximum forest canopy height map has a 30-m spatial resolution and was developed by integrating NASA's spaceborne LiDAR measurements available from the Global Ecosystem Dynamics Investigation (GEDI; dates April – October 2019; Dubayah et al. 2020) and the Landsat temporal data series (see Potapov et al. 2021). 

Global aridity index (GAI) provides a measure of water availability for potential vegetation growth (Zomer and Trabucco, 2022) and is an excellent large-scale predictor of plant diversity (O’Brien, 1993; 1998). The GAI is a high-resolution (30-sec) dataset spanning the 1970–2000 period and describes the precipitation deficit for potential vegetative growth. This index represents the ratio of precipitation to evapotranspiration, with values decreasing with more arid conditions and increasing with more humid conditions (see Trabucco and Zomer, 2019). 

Temperature (TEMP) is a measure of energy availability in an ecosystem, and is related to metabolic and speciation rates, with most species worldwide adapted to warm places (Brown et al. 2004). According to the species-energy theory (Wright, 1983), energy availability, such as temperature, drives species diversity. Thus, for each plot, we extracted the mean annual TEMP data (bio1) from WorldClim (Fick and Hijmans, 2017), with a 10-m resolution.

Soil fertility (SOIL). Soil base cation concentrationdata were extracted from a Pan-Amazonian map of the sum of exchangeable base cation concentration (Zuquim et al. 2023). High soil heterogeneity sites are expected to support more plant species (Tuomisto et al. 2014), and spatial heterogeneity in plant species composition is expected to influence species diversity of other taxonomic groups (Pomara et al. 2012). More nutrient-rich sites in Amazonian forests also support higher arboreal vertebrate diversity and biomass (Peres, 2008). 

Table_S4.csv. List of species extracted from range maps that were not recorded in the compiled studies. List of species per plot.

Table_S5_Complete_dataset.csv - Complete species list per plot.

Code/software

All of our analyses were run in R software, and the scripts are available for replication (____R_code-_Ecography_ECOG-07625..txt).