Skip to main content
Dryad logo

Species abundances and functional traits data of frog assemblages in Amazonian forests (Tapajós FLONA and Alter do Chão village, Brazil)

Citation

Torralvo, Kelly; de Fraga, Rafael; Lima, Albertina; Magnusson, William (2021), Species abundances and functional traits data of frog assemblages in Amazonian forests (Tapajós FLONA and Alter do Chão village, Brazil), Dryad, Dataset, https://doi.org/10.5061/dryad.gb5mkkwqv

Abstract

We sampled frogs along an edaphic and vegetation-structure gradient in the Brazilian Amazon. We sampled 33 plots organized in four modules (with 10, 6, 8 and 9 plots). Environmental data are composed of are litter depht(cm), distance to water bodies (m), temperature (ºC), vegetation structure (PCA axis), soil sand and silt content (g-kg), and proportion of the area deforested. Additionally, we measured snout-vent length (SVL) and leg length relative to SVL and evaluated the effects of environmental variables on thirteen binary traits used with presence or absence.

Methods

We sampled 33 plots organized in four modules (with 10, 6, 8 and 9 plots) of Tapajós National Forest and Alter do Chão Village, Pará state, Brazil. We used visual and acoustic search from day to twilight (16h30min–18h00min), and night (19h00min–23h00min pm) for frogs detection during the rainy season (January–March) in 2018 and 2019. We identified species based on the literature (e.g. Lima et al. 2006, Oliveira et al. 2017, Carvalho et al. 2020) and comparisons with specimens deposited in the herpetological section of the zoological collection of Instituto Nacional de Pesquisas da Amazônia (INPA, Manaus, Brazil).

We measured litter depth with a Marimon-Hay litter collector (Marimon-Junior & Hay 2008) at six points separated by 50-m intervals in each plot and used mean values in the analyses. We using a set of variables quantifying forest structure (data in https://search.dataone.org/view/PPBioAmOc.577.2 ) summarized with the first axis of a principal component analysis (PCA) that represent a 59% of the variance in the original variables - the metrics are maximum forest height, average forest height, canopy opening ratio, canopy gaps computed at 5 m, canopy gaps computed at 10 m, canopy gaps computed at 15 m, leaf area index in the vertical profile, leaf area index in the vertical profile 0–15 m, leaf area index in the vertical profile above 15 m, collected by a portable Light Detection and Ranging (LIDAR) device. We measured temperature at the beginning of day and night sampling, using AK172-AKSO® data loggers, and used mean values for each plot in the inferential models. We measured the distances from the coordinates of the beginning of each plot to the nearest water body using the linear-distance matrix and Qchainage tools of the QGis 3.16.2 software, applied over combined hydrography shapefiles from public repositories (MMA 2006), following Venticinque et al. (2016). We used SoilGrids data to obtain sand and silt content 0–5 m deep, extracted sand- and silt-content values from centroid coordinates of each plot using the raster R-package. Finally, we used a 30 x 30 m² satellite-based raster layer summarizing deforestation until 2019 from the MapBiomas project (Souza et al. 2020, see full description in http://mapbiomas.org) to extract proportions of deforested areas in 500 m-buffers using the raster R-package. Analyzes were performed using the statistical program R (R Development Core Team 2019).