Data and code from: Seed dispersal and seed predation networks highlight important dual roles for plants and vertebrates in a tropical forest
Data files
Jan 15, 2026 version files 106.06 KB
-
analyse_network_metrics.R
21.16 KB
-
Animal_traits.xlsx
11.55 KB
-
calc_network_metrics.R
6 KB
-
Interaction_data.xlsx
18.21 KB
-
multilayer_analysis.R
6.67 KB
-
Plant_traits.xlsx
11.39 KB
-
prep-data.R
1.08 KB
-
ReadMe.docx
23.54 KB
-
README.md
6.47 KB
Abstract
The dispersal and predation of seeds by animals are important, and inextricably linked, drivers of plant species diversity and ecosystem stability. Within tropical forests, these interactions involve diverse vertebrate communities with the same species often contributing to both processes. Yet, multi-level interactions are rarely examined at community-scales, limiting our ability to identify the most important species driving plant recruitment – a crucial aspect of demographic dynamics. Here, we build a multi-layered network of vertebrate seed dispersal and seed predation interactions from a tropical forest in Thailand, to identify important interactors and assess the plant and animal traits that influence species’ roles. The dataset encompassed interactions between 36 plant and 25 ground-foraging vertebrate species observed in camera-traps (>1000 camera-trap days). Over half the plant and animal species were involved in both dispersal and predation, with the dispersal network involving more species than the predation network. Four of six animal species that were major interactors had dual roles and two were seed predators, indicating that analyses of networks in isolation would incompletely reflect their importance in plant recruitment. Abundance was positively associated with species’ roles (degree, species strength) for plants and animals in the seed dispersal network but not the predation network; none of the size traits (fruit/seed size; body mass) nor seed hardness were related to species’ roles. Plant multi-layer versatility, a measure of relative importance within and between interaction types, was not correlated with any measured traits, suggesting that more complex trait associations define versatility. In conclusion, the most important species for influencing patterns of plant recruitment are more accurately identified by measuring roles across multiple levels, because of the dual involvement of plant and animal species in both seed dispersal and predation. However, a broader range of traits that might determine these roles needs to be studied.
This Dryad repository contains all the data and code necessary to reproduce the analyses and figures in the study. A Read Me file containing a description of all the contents of this repository, usage notes and outputs is provided.
Dataset DOI: 10.5061/dryad.2v6wwq01k
Description of the data and file structure
This data repository consists of the following files:
- Information on data files, code, and how to replicate analysis ("ReadMe.docx" and "README.md")
- Data files (“Interaction data.xlsx”, “Animal traits.xlsx”, “Plant traits.xlsx”)
- R scripts to perform all data processing and analyses (“prep-data.R”, “calc_network_metrics.R”, “multilayer_analysis.R”, “analyse_species_metrics.R”)
To run the R script on your local machine, you will first need to specify your local working directory by setting the working directory to the folder containing all the files, using the setwd() function. Output will automatically be written into the defined working directory.
Files and variables
File: analyse_network_metrics.R
Description: R script to perform analysis of interaction importance and network metrics
File: Animal_traits.xlsx
Description: Trait and abundance data for animal species
Variables
- Animal species: Species names of animals recorded using camera traps
- Common name: Common name of animal species
- Family: Taxonomic family of animal species
- Order: Taxonomic order of animal species
- Diet: Broad categorisation of animal species' dietary niche
- Body mass: Body mass in grams
- % camera-trap success (Luskin unpublished): Relative frequency of animal occurrence in camera traps. Empty cells = species was not recorded in the camera trap study.
File: calc_network_metrics.R
Description: R script to calculate species-level and community-level network metrics
File: Plant_traits.xlsx
Description: Trait and abundance data for plant species
Variables
- Plant species: Species name of focal plants observed using camera traps
- Family: Taxonomic family of plant species
- BA m2: Total basal area of species in the Mo Singto ForestGeo plot. Basal area for liana species was not recorded in the plot and hence recorded as "NA (liana)".
- Fruit colour: Dominant colour of fruit or seed (for arillate seeds)
- Fruit length (mm): Length of fruit in millimetres
- Seed length (mm): Length of seed in millimetres
- Seed protection: Hardness of seed
File: multilayer_analysis.R
Description: R script to perform multi-layer network analysis
File: Interaction_data.xlsx
Description: Interaction frequency data for seed predation and seed dispersal interactions
The dataset contains two sheets, one for seed predation data ("Seed predation" sheet) and seed dispersal data ("Seed dispersal" sheet). They both contain the following columns:
Variables
- Plant species: Species name of focal plants observed using camera traps
- Family: Taxonomic family of plant species
- #camera trap days: The number of days that fruits/seeds for a plant species were observed using camera traps (the same for seed predation and seed dispersal sheets as they are based on the same camera trap data)
- Animal columns: Interaction frequency (seconds per camera trap day) of each animal species with each plant.
File: prep-data.R
Description: R script to prepare interaction data for further analysis
File: ReadMe.docx
Description: Word Doc version of README file.
Code/software
To run this code you will need an installation of R and some R packages. A list of these R packages can be found below, but this list may not include all of their dependencies. Feel free to get in contact if you have trouble running the script.
Code was tested on R version 4.4.0 with the following packages attached (version number in parentheses): BiMultiNetPlot (v.0.1.0), bipartite (v.2.2.0), cowplot (v.1.1.3), dplyr (v.1.1.4), ggplot2 (v.3.5.1), muxViz (v.3.1), plyr (v.1.8.9), readxl (v.1.4.3) , reshape2 (v.1.4.4), stringr (v.1.5.1), tidyverse (v.2.0.0).
To fully replicate the analysis, the following R scripts should be run in order. Expected output files are described and named below:
1. Prepare data (prep-data.R)
- Reads in raw interaction data and generates a long-form versions: “dispersal_clean.csv” and “predation_clean.csv”
2. Calculate network metrics (calc_network_metrics.R)
- Generates a long-form data frame containing both dispersal and predation information for all animal and plant species pairs (combined_df.csv).
- Calculates species-level network metrics for both seed dispersal and seed predation networks, and animal and plant species, separately (“dispersal_animal_metrics.csv”, “dispersal_plant_metrics.csv”, “predation_animal_metrics.csv”, “predation_plant_metrics.csv”)
3. Performs multi-layer network analysis (multilayer_analysis.R)
- Calculates versatility (node centrality) for each plant species (“plant_node_centrality.csv”)
- Plots the multi-layer network as a tripartite network (“tripartite_plot_lab.pdf”)
4. Performs analysis of interaction importance and network metrics (analyse_network_metrics.R)
- Calculate absolute and relative importance of plant species in both seed dispersal and predation networks (“plant_importance.csv”), and plots the respective figures (“plant_abs_importance.pdf”,“plant_rel_importance.pdf”, “plant_combined_importance.pdf”)
- Calculate absolute and relative importance of plant species in both seed dispersal and predation networks (“animal_importance.csv”), and plots the respective figures (“animal_abs_importance.pdf”,“animal_rel_importance.pdf” , “animal_combined_importance.pdf”)
- Compiles data of all species-level network metrics and traits for plant and animal species (“animal_res.csv” and “plant_res.csv”)
- Performs correlation tests between species-level network metrics (including plant multi-layer versatility scores) and trait/abundance variables, exports the results (“plant_network_analysis_res.csv”, “animal_network_analysis_res.csv”), and plots them (“plant_network_cor.pdf”, “animal_network_cor.pdf”, “versatility_combined.pdf”)