Data from: Understory frog communities recover quickly during tropical forest regeneration
Data files
Feb 24, 2026 version files 26.97 KB
-
Plot_data.csv
4.36 KB
-
R_.R
14.72 KB
-
README.md
5.64 KB
-
Species_plot.csv
2.26 KB
Abstract
This dataset contains plot-level data on understory frog communities sampled along a forest regeneration chronosequence in the lowland Chocó forest near Hoja Blanca, Esmeraldas province, northwestern Ecuador. Sampling was conducted within and around the privately managed Canandé (14,000 ha) and Tesoro Escondido (1,800 ha) reserves. A total of 38 plots (50 × 50 m; 2,500 m²) were established across a mosaic landscape, including active pastures (n = 6), active cacao plantations (n = 6), regenerating forests following pasture or cacao abandonment (0–37 years; n = 18), and old-growth forest (n = 8). Plots ranged from 159 to 615 m a.s.l. Regeneration time was determined from local land-use history records and reserve acquisition data. Understory frogs were surveyed using standardized visual encounter surveys with two person-hours per plot per visit. Each plot was sampled six times between March 2022 and June 2023, including both day and night sessions. Species abundances were pooled across visits to obtain one value per species per plot. The dataset includes (a) species abundance data per plot, and (b) associated environmental predictors, including regeneration time, land-use legacy (cacao or pasture), elevation, basal area, distance to the nearest forest older than 40 years, and proportion of surrounding forest older than 40 years within a 500 m radius. The repository also includes the R script used for statistical analyses.
https://doi.org/10.5061/dryad.k0p2ngfj9
Description of the data and file structure
The dataset was collected in the lowland Chocó forest, northwestern Ecuador, across a forest regeneration gradient using a chronosequence approach. A total of 38 plots (2,500 m² each) were established in a mosaic landscape that included active pastures, cacao plantations, regenerating forests (1 - 37 years), and old-growth forests. Each plot was visited six times during three sampling rounds (March–June 2022, July–October 2022, and March–June 2023). Each round consisted of both day and night sessions.
Systematic visual searches were conducted for two person-hours per plot, focusing on understory frogs from the leaf litter to vegetation up to 1.2 meters in height. Frogs encountered were captured, identified to species using field guides and online resources, and released after the survey. Sampling alternated between plots of different habitat types to minimize temporal effects on results.
The study was conducted in Canandé and Tesoro Escondido reserves, Esmeraldas, Western Ecuador.
Files and variables
File: Plot_data.csv
Description: The dataset contains information from 38 study plots located in different habitats within the Chocó rainforest in Ecuador. These plots, measuring 50 x 50 meters, span a range of regeneration stages, including active agricultural areas (cacao plantations and pastures, with age assigned as 0), early and late forest regeneration (ranging from 1 to 37 years), and old-growth forest. Data collected from each plot includes geographic coordinates (latitude and longitude), elevation, regeneration time (RegTime), historical land-use (Legacy), proximity to the nearest forest (Distance_forest), and the percentage of forest within a 500-meter radius around each plot (Forest_500m). The plots represent a chronosequence to study the recovery of understory frog communities in relation to forest regeneration.
Variables
- Plot_ID: Unique identifier assigned to each sampling plot.
- Forest_100m: Proportion (0–1) of forest older than 40 years within a 100 m radius around the plot centroid.
- Forest_500m: Proportion (0–1) of forest older than 40 years within a 500 m radius around the plot centroid.
- Treatment: Current land-use type of the plot (active cacao, active pasture, regenerating forest, old-growth forest).
- Regeneration_year: Calendar year in which agricultural use ceased for regenerating plots. NA indicates that regeneration time does not apply (active plots or old-growth forest).
- latitude: Geographic latitude in decimal degrees (WGS84).
- longitude: Geographic longitude in decimal degrees (WGS84).
- elevation: Elevation in meters above sea level (m a.s.l.).
- RegTime: Number of years since abandonment at the time of sampling. Value = 0 indicates active land use or abandonment in the year sampling began. RegTime was calculated as sampling year minus Regeneration_year.
- Legacy: Previous land-use type prior to regeneration (cacao or pasture).
- Age_cat: Categorical classification of regeneration stage (Active, Early, Late, Old-growth).
- Distance_forest: Distance (in meters) from the plot centroid to the nearest forest older than 40 years.
File: Species_plot.csv
Description: The dataset contains the total abundance of understory frog species recorded in each sampling plot. Each row represents one plot, identified by its unique Plot ID. Each column (after the first column) corresponds to one frog species, identified by its full scientific name.
Abundance values represent the total number of individuals recorded per species per plot, pooled across six standardized visual encounter surveys conducted between March 2022 and June 2023.
The dataset is provided in wide format (species as columns).
Zero values indicate that a species was not detected in a given plot.
File: R_.R
Description: The project contains scripts for analyzing the taxonomic diversity and composition of understory frogs across forest regeneration stages. The codes include:
- Taxonomic Diversity Metrics
Calculation of species richness, exponential Shannon diversity, and Inverse Simpson diversity using the iNEXT and vegan R packages.
Rarefaction/extrapolation analysis and confidence interval estimation.
- Statistical Analysis
Linear models (lm) to explore the effects of land-use legacy, regeneration time, and spatial factors on diversity metrics.
Model validation, including residual diagnostics, variance homogeneity (Bartlett test), and spatial autocorrelation (Moran's I).
- Recovery Time Estimation
A simplified linear model, incorporating regeneration time and land-use legacy as fixed effects and median values from old-growth forests as the reference.
- Community Composition Analysis
PERMANOVA and PERMDISP to evaluate species composition and dispersion among forest successional stages.
Pairwise comparisons with adjusted p-values to identify specific differences.
- NMDS Ordination
Non-metric multidimensional scaling (NMDS) to visualize differences in species composition across successional stages and analyze community recovery trajectories.
- Visualization
Plotting trends and ordinations using the ggplot2 and vegan packages.
Code/software
Software:
R (version 4.3.0, R Core Team, 2023)
Packages used:
iNEXT (Hsieh et al., 2016)
vegan (Oksanen et al., 2020)
ggplot2 (Wickham, 2016)
car (Fox & Weisberg, 2019)
ape (Paradis & Schliep, 2019)
Study area and sampling design
Fieldwork was conducted in the lowland Chocó forest near Hoja Blanca, Esmeraldas province, northwestern Ecuador, within and around the privately managed Canandé (14,000 ha) and Tesoro Escondido (1,800 ha) reserves (0.52°N, 79.2°W; 130–615 m a.s.l.). The landscape consists of a mosaic of old-growth forest, secondary forests of different ages, pastures, and cacao plantations.
A total of 38 plots (50 × 50 m; 2,500 m²) were established following a space-for-time substitution approach (chronosequence). Plots were distributed across active pastures (n = 6), active cacao plantations (n = 6), regenerating forests after pasture or cacao abandonment (0–37 years; n = 18), and old-growth forest (n = 8). Regeneration time for secondary forests was determined from local land-use history records and reserve acquisition information. Old-growth forest plots were not assigned a regeneration age.
Plots were spatially distributed as widely as possible within habitat categories (minimum distance = 184 m; maximum distance = 14 km).
Frog surveys
Understory frogs were sampled using standardized visual encounter surveys. Each plot was surveyed for two person-hours per visit, scanning from leaf litter up to approximately 1.2 m above ground. Each plot was visited six times between March 2022 and June 2023, across three sampling rounds that included both day and night sessions.
All individuals encountered were captured, identified to species level, and released. Species abundances were pooled across the six visits to obtain one abundance value per species per plot.
Environmental predictors
Plot-level environmental variables included regeneration time (years since abandonment), land-use legacy (cacao or pasture), elevation (m a.s.l.), distance to the nearest forest older than 40 years (m), and proportion of forest older than 40 years within a 500 m radius.
Data processing
Species abundances were pooled across visits to reduce seasonal and stochastic variation in detection. The accompanying R script contains the code used to calculate diversity metrics and perform statistical analyses based on the data provided.