Dataset DOI: 10.5061/dryad.m905qfvg5

Description of the data and file structure

This dataset was collected to investigate the dissimilarity of frugivorous bat assemblages, plant species, and their interactions across different forest strata in a conserved Atlantic Forest environment in southern Brazil. The data were generated to evaluate whether sampling focused on a single vertical stratum, such as the understory, provides an accurate or biased characterization of bat–fruit interaction patterns.

Sampling was conducted monthly between September 2013 and August 2014 in the Salto Morato Natural Reserve, Paraná state, comprising 12 sampling campaigns and 61 sampling nights. Bats were captured using mist nets simultaneously deployed in three forest strata (understory, subcanopy, and canopy), allowing a stratified assessment of frugivorous bat assemblages, plant species, and their interactions along the forest vertical gradient.

Fecal samples were collected from captured bats and analyzed to identify consumed plant species, generating records of bat–fruit interactions for each forest stratum. The dataset includes raw and processed data used to construct interaction matrices and quantify patterns of network structure, species specialization, and interaction dissimilarity among forest strata.

Files and variables

File name: Main_script.R

Description: R script used to perform the majority of the analyses presented in the study. The script includes the construction and visualization of bat–fruit interaction networks; calculation of network-level metrics (connectance, specialization index (H2′), nestedness WNODF , and modularity); null model analyses; estimation of sampling completeness and interaction richness (Chao1); species-level metrics for bats and plants (normalized degree, specialization index d′ , and species strength); statistical comparisons among forest strata; and beta diversity of interactions among strata. The script reads the interaction matrices and species-level metric files provided in this dataset and reproduces all figures and numerical results reported in the manuscript.

File name: Three_strata_matrix.txt

Description: Interaction matrix representing bat–fruit interactions pooled across the three forest strata (understory, subcanopy, and canopy). Columns correspond to frugivorous bat species and rows correspond to plant species. Cell values represent the observed frequency of interactions, defined as the number of fecal samples containing at least one intact seed of a given plant species associated with a given bat species. This matrix was used to run interaction networks, calculate species-level and network-level metrics.

Thus, if seeds of the same plant species were detected in four fecal samples from a given bat species, the interaction frequency was recorded as four. Zeros indicate the absence of recorded interactions between a given bat and plant species.

Variables:

• Rows: Plant species (identified to species level when possible; unidentified taxa labeled as sp., sp1, sp2, etc.).
• Columns: Bat species.
• Cell values: Integer counts of observed bat–plant interactions based on fecal sample analyses.

Missing data:
No missing values. Absence of interactions is indicated by zeros.

File name: Two_strata_matrix.txt

Description: This file contains a quantitative interaction matrix of bat–fruit interactions recorded in the subcanopy and canopy forest strata. The matrix was assembled from fecal sample analyses of frugivorous bats, including only interactions detected within these two strata. This matrix was used to run interaction networks, calculate species-level and network-level metrics.

Variables:

• Rows: Plant species (identified to species level when possible; unidentified taxa labeled as sp., sp1, sp2, etc.).
• Columns: Bat species.
• Cell values: Number of fecal samples in which seeds of a particular plant species were found for each bat species.

Missing data:
No missing values. Zero indicates that no interaction was detected between the corresponding bat and plant species in these strata.

File name: Canopy_matrix.txt

Description: This file contains a quantitative interaction matrix of bat–fruit interactions recorded exclusively in the canopy forest stratum. The matrix was compiled from fecal sample analyses of frugivorous bats and includes only interactions observed in this stratum. This matrix was used to run interaction networks, calculate species-level and network-level metrics, and to analyze beta diversity of interactions.

Variables:

• Rows: Plant species (identified to species level when possible; unidentified taxa labeled as sp., sp1, sp2, etc.).
• Columns: Bat species.
• Cell values: Number of fecal samples in which seeds of a particular plant species were found for each bat species.

Missing data:
No missing values. Zero indicates that no interaction was detected between the corresponding bat and plant species in the canopy stratum.

File name: Subcanopy_matrix.txt

Description: This file contains a quantitative interaction matrix of bat–fruit interactions recorded exclusively in the subcanopy forest stratum. The matrix was compiled from fecal sample analyses of frugivorous bats and includes only interactions observed in this stratum. This matrix was used to run interaction networks, calculate species-level and network-level metrics, and to analyze beta diversity of interactions.

Variables:

• Rows: Plant species (identified to species level when possible; unidentified taxa labeled as sp., sp1, sp2, etc.).
• Columns: Bat species.
• Cell values: Number of fecal samples in which seeds of a particular plant species were found for each bat species.

Missing data:
No missing values. Zero indicates that no interaction was detected between the corresponding bat and plant species in the subcanopy stratum.

File name: Understory_matrix.txt

Description: This file contains a quantitative interaction matrix of bat–fruit interactions recorded exclusively in the understory forest stratum. The matrix was compiled from fecal sample analyses of frugivorous bats and includes only interactions observed in this stratum. This matrix was used to run interaction networks, calculate species-level and network-level metrics, and to analyze beta diversity of interactions.

Variables:

• Rows: Plant species (identified to species level when possible; unidentified taxa labeled as sp., sp1, sp2, etc.).
• Columns: Bat species.
• Cell values: Number of fecal samples in which seeds of a particular plant species were found for each bat species.

Missing data:
No missing values. Zero indicates that no interaction was detected between the corresponding bat and plant species in the understory stratum.

File name: GN_bats.txt

Description: This file contains the normalized degree for each frugivorous bat species across different forest strata. The normalized degree is a species-level metric that quantifies the level of generalization, calculated as the proportion of available plant partners with which a bat species interacts. Maximum values of 1 indicate species that interact with all available partners in the network. The dataset includes results for individual strata as well as pooled strata combinations. This file was used to generate graphs and compare species-level metrics across forest strata.

Variables:

• Bats: Species names of frugivorous bats.

• Normalized_degree: Normalized degree of the bat species, representing its level of generalization.

• Strata: Forest stratum or combination of strata where interactions were recorded. Codes are as follows:

  1 = Understory 

  2 = Sub-canopy

  3 = Canopy

  4 = Sub-canopy + Canopy (two strata)

  5 = Understory + Sub-canopy + Canopy (all three strata)

Missing data:
No missing values.

File name: Bats_specialization.txt

Description: This file contains the species-level specialization index (d’) for frugivorous bats across different forest strata. The d’ index quantifies the deviation of interaction frequencies between partners relative to the abundance of interactions. Species with d’ values close to 1 tend to interact with their partners more frequently than expected based on interaction abundances, indicating high specialization. The dataset includes results for individual strata as well as pooled strata combinations. This file was used to generate graphs and compare species-level metrics across forest strata.

Variables:

• Bats: Species names of frugivorous bats.

• Specialization: d’ index for each bat species, representing its level of specialization in the network.

• Strata: Forest stratum or combination of strata where interactions were recorded. Codes are as follows:

  1 = Understory

  2 = Sub-canopy

  3 = Canopy

  4 = Sub-canopy + Canopy (two strata)

  5 = Understory + Sub-canopy + Canopy (all three strata)

Missing data:
No missing values.

File name: Strength_bats.txt

Description: This file contains species-level strength values for frugivorous bats across different forest strata. Species strength quantifies the importance of a species in the network based on the sum of dependencies of all other species with which it interacts. Dependency is calculated as the proportion of interactions that each species represents, considering the entire set of interactions of a focal species. This file was used to generate graphs and compare species-level metrics across forest strata.

Variables:

• Bats: Species names of frugivorous bats.

• Strength: Species strength value for each bat species.

• Strata: Forest stratum or combination of strata where interactions were recorded. Codes are as follows:

  1 = Understory

  2 = Sub-canopy

  3 = Canopy

  4 = Sub-canopy + Canopy (two strata)

  5 = Understory + Sub-canopy + Canopy (all three strata)

Missing data:
No missing values.

File name: GN_plants.txt

Description: This file contains species-level normalized degree values for plants across different forest strata. The normalized degree quantifies the level of generalization of a species, being calculated as the proportion of available partners with which the species interacts. Maximum values of 1 identify species that interact with all available partners in the network. This file was used to generate graphs and compare species-level metrics across forest strata.

Variables:

• Plants: Species names of plants.
• Normalized_degree: Normalized degree value for each plant species.
• Strata: Forest stratum or combination of strata where interactions were recorded. Codes are as follows:
  1 = Understory
  2 = Sub-canopy
  3 = Canopy
  4 = Sub-canopy + Canopy (two strata)
  5 = Understory + Sub-canopy + Canopy (all three strata)

Missing data:
No missing values.

File name: Specia_plants.txt

Description: This file contains species-level specialization (d’) values for plants across different forest strata. The specialization index (d’) quantifies the deviation of interaction frequencies between partners in relation to the abundances of interactions. Species with d’ values close to 1 tend to interact with their partners more frequently than expected based on abundance, showing high specialization. This file was used to generate graphs and compare species-level metrics across forest strata.

Variables:

• Morphotype: Species names of plants.
• Specialization: Specialization (d’) value for each plant species.
• Strata: Forest stratum or combination of strata where interactions were recorded. Codes are as follows:
  1 = Understory
  2 = Sub-canopy
  3 = Canopy
  4 = Sub-canopy + Canopy (two strata)
  5 = Understory + Sub-canopy + Canopy (all three strata)

Missing data:
No missing values.

File name: Stren_plants.txt

Description: This file contains species-level strength values for plants across different forest strata. Species strength quantifies the importance of a species in the network based on the sum of dependencies of all other species with which it interacts. Dependency is calculated as the proportion of interactions that each species represents, considering the entire set of interactions of a focal species. This file was used to generate graphs and compare species-level metrics across forest strata.

Variables:

• Morphotype: Species names of plants.
• Strength: Species strength value for each plant species.
• Strata: Forest stratum or combination of strata where interactions were recorded. Codes are as follows:
  1 = Understory
  2 = Sub-canopy
  3 = Canopy
  4 = Sub-canopy + Canopy (two strata)
  5 = Understory + Sub-canopy + Canopy (all three strata)

Missing data:
No missing values.

File name: Venn_D.R

Description: This R script was used to generate Venn diagrams to visualize the presence, exclusivity, and overlap of species across different forest strata, as well as their interactions. The script uses interaction matrices, which are described below.

File name: Bats_by_stratum.txt

Description: This file lists bat species recorded in each forest stratum. It was used to generate Venn diagrams showing species presence, exclusivity, and overlap across strata.

Variables:

• Species: Bat species name.
• Strata: Forest stratum where the species was recorded (canopy, subcanopy and understory).

Missing data:
No missing values.

File name: Binary_canopy.txt

Description: This file contains a binary interaction matrix of bat–fruit interactions recorded exclusively in the canopy forest stratum. This matrix was used to generate Venn diagrams showing species presence, exclusivity, and overlap, as well as to visualize differences in interactions between strata.

Variables:

• Rows: Plant species.
• Columns: Bat species.
• Values: 1 = interaction observed; 0 = no interaction observed.

Missing data:
No missing values.

File name: Binary_subcanopy.txt

Description: This file contains a binary interaction matrix of bat–fruit interactions recorded exclusively in the sub-canopy forest stratum. This matrix was used to generate Venn diagrams showing species presence, exclusivity, and overlap, as well as to visualize differences in interactions between strata.

Variables:

• Rows: Plant species.
• Columns: Bat species.
• Values: 1 = interaction observed; 0 = no interaction observed.

Missing data:
No missing values.

File name: Binary_understory.txt

Description: This file contains a binary interaction matrix of bat–fruit interactions recorded exclusively in the understory forest stratum. This matrix was used to generate Venn diagrams showing species presence, exclusivity, and overlap, as well as to visualize differences in interactions between strata.

Variables:

• Rows: Plant species.
• Columns: Bat species.
• Values: 1 = interaction observed; 0 = no interaction observed.

Missing data:
No missing values.

File name: Canopy_interactions.txt

Description: This file lists all bat–fruit interactions recorded exclusively in the canopy forest stratum. It was used to generate Venn diagrams showing species and interaction presence, exclusivity, and overlap across forest strata.

Variables:

• Interactions: Unique identifier for each bat–plant interaction, combining the bat and plant species names (e.g., *Sturnira_lilium_Solanum_*sp4).

Missing data:
No missing values.

File name: Subcanopy_interactions.txt

Description: This file lists all bat–fruit interactions recorded exclusively in the sub-canopy forest stratum. It was used to generate Venn diagrams showing species and interaction presence, exclusivity, and overlap across forest strata.

Variables:

• Interactions: Unique identifier for each bat–plant interaction, combining the bat and plant species names (e.g., Sturnira_lilium_Piper_hispidum).

Missing data:
No missing values.

File name: Understory_interactions.txt

Description: This file lists all bat–fruit interactions recorded exclusively in the understory forest stratum. It was used to generate Venn diagrams showing species and interaction presence, exclusivity, and overlap across forest strata.

Variables:

• Interactions: Unique identifier for each bat–plant interaction, combining the bat and plant species names (e.g., Sturnira_lilium_Piper_hispidum).

Missing data:
No missing values.

Code/software

Software/Code and Workflow

All analyses and visualizations were performed using R (www.r-project.org), a free and open-source environment for statistical computing and graphics. The following R packages were used for data processing, network analyses, and plotting:

bipartite (v2.18) – to calculate network- and species-level metrics, including network size, number of interactions, asymmetry, connectance (C), specialization (H2’), weighted nestedness (WNODF), weighted modularity (Qw), normalized degree, species-level specialization (d’), and species strength. Null models for network metrics were generated using the r2dtable.

iNEXT (v3.0.1) – to estimate sampling completeness of interaction networks using the Chao1 estimator.

stats (v3.6.2) – for statistical analyses, including ANOVA, Kruskal–Wallis, and post hoc tests (Tukey or Dunn) to evaluate differences in species-level metrics among strata.

**ggplot2  **– to generate visualizations of species-level metrics and comparisons across strata.

betalink (v2.2.1) – to calculate beta diversity of interactions, including dissimilarity due to species turnover (βST), interactions among shared species (βOS), and total dissimilarity (βWN).

VennDiagram (v1.7.3) – to plot Venn diagrams illustrating the sharing and exclusivity of species and interactions across forest strata.

Workflow:

Interaction data were compiled into weighted matrices for each forest stratum (understory, subcanopy, and canopy), with plant species in rows and bat species in columns. Each cell represents the observed frequency of interactions, defined as the number of fecal samples containing seeds of a given plant species associated with a particular bat species.

Analyses were first conducted separately for each stratum and subsequently for combinations of two and three strata to evaluate gradual effects across the vertical gradient. Network-level metrics, including null models and sampling completeness, were calculated first, followed by species-level metrics, which were then used to generate graphs and perform comparisons across strata.

Interaction dissimilarity between strata was assessed using beta diversity metrics. Species and interaction exclusivity and overlap across strata were visualized using Venn diagrams.

Included scripts: R scripts for data import, calculation of network- and species-level metrics, null model testing, beta diversity calculation, and Venn diagram plotting.

Tab-delimited input files containing species occurrences and interactions for each stratum.

Access information

Other publicly accessible locations of the data:

• The dataset and R scripts supporting this study are publicly available at: https://github.com/karolportosupi/Bat-fruit-interaction.git

Data was derived from the following sources:

• No external datasets were used; all data were generated by the authors.