Topography structures arthropod communities across an invaded tropical island revealed by leaf-derived environmental DNA
Data files
Jul 09, 2026 version files 1.13 GB
-
raw_files.zip
1.13 GB
-
README.md
7.27 KB
Abstract
This dataset contains metabarcoding results and metadata from environmental DNA (eDNA) sampled from tree canopies along elevational gradients on Oʻahu. The primary host tree is ʻōhiʻa lehua, with paired comparisons to koa (native) and strawberry guava (introduced) on shared gradients where available. The data include zOTU/OTU matrices, site and sample metadata, host tree identity, elevation, and native/introduced classifications predicted by NIClassify and cross-checked manually. Analyses focus on beta diversity (turnover vs. nestedness), NMDS ordinations, PERMANOVA tests of elevation/site/tree effects, multivariate dispersion, indicator/threshold analyses (TITAN2), and overlap/uniqueness summaries (Venn/UpSet).
All paths are relative to the Dryad deposit root.
Archive package
Download files from: https://github.com/sven9r/ohia_gradient
Raw sequence folder
raw_files.zip contains raw FASTA files used in sequence processing and classification workflows.
Data files included
data/ohia_gradient_2026-02-09.xlsx- Core sample/site metadata and summary metrics.
- Expected sheets:
samples— one row per eDNA sample.sites— site-level attributes.aux— helper tables (e.g., elevation band breaks, plotting keys).
data/zotu_table.tsv- zOTU (UNOISE/ESV) table; rows = zOTUs, columns = samples, values = read counts.
data/taxonomy_assignments.tsv(optional)- BLAST-based taxonomy labels with identity/coverage and rank.
data/niclassify_labels.tsv(optional)- NIClassify output with native/introduced predictions and confidence.
If tables are generated at render-time, they are written to output/.
Analysis scripts and what they do
scripts/00_source.R— dependency loading, shared data ingestion, helper functions (save_fig()), canonical object setup.scripts/01_map.qmd— site map and sampling context.scripts/02_initial_analyses.qmd— NMDS, PERMANOVA, beta dispersion, distance-decay, decomposition, and associated tables/figures.scripts/03_overlap_invasion.qmd— overlap/invasion framing, UpSet panels, comparative summaries.scripts/04_species_richness.qmd— richness/elevation/rainfall analyses and regression panels.scripts/05_titan_breakpoints.qmd— TITAN2 breakpoint and bootstrap summaries.scripts/06_molecular_framing.qmd— taxonomy resolution framing, NIClassify/BLAST comparison, Sankey synthesis.scripts/07_coadaptation.qmd— coadaptation and additional exploratory/sensitivity analyses.
Rendered outputs
figures/main/— manuscript main figures (F#_*) and table graphics.figures/supp/— supplementary figures (S#_*).figures/exploratory/— exploratory/non-final plots.output/— derived tables/statistics/intermediate files.
Data dictionary (for data/ohia_gradient_2025-03-25_91.xlsx:samples)
| Column | Type | Description | Allowed values / example |
|---|---|---|---|
sample_id |
string | Unique sample identifier (matches zOTU/OTU table sample names). | OHIA_POA_01A |
island |
string | Island name. | Oahu, Hawaii |
site |
string | Local site name/code (ridge/valley). | Poamoho, Pahole, Mānana, Hakalau |
ridge |
string | Ridge identity within mountain block (if applicable). | Poamoho Ridge |
elevation_m |
numeric | Elevation in meters. | 1120 |
elev_band |
factor | Elevation band used in analyses. | low, mid, high |
host_tree |
factor | Canopy host species sampled. | ohia, koa, strawberry_guava |
lat |
numeric | Latitude (decimal degrees, WGS84). | 21.5173 |
lon |
numeric | Longitude (decimal degrees, WGS84). | -158.1065 |
date |
date | Sampling date (YYYY-MM-DD). |
2024-07-02 |
replicate |
string | Technical/biological replicate code. | A, B, C |
primer |
factor | Marker/primer set used. | ANML, MCO |
reads_total |
integer | Total reads post-QC per sample. | 145632 |
zotu_richness |
integer | Number of zOTUs detected in sample. | 312 |
introduced_frac |
numeric | Fraction classified as introduced (specify read- or zOTU-based). | 0.46 |
notes |
string | Free-text notes/flags. | low_template |
Quality control and caveats
- Non-biological controls (field/blank/PCR negatives) were processed with samples; contaminants/tag-jumps were screened.
- Species-level labels are reported only where identity support is sufficient; otherwise records are collapsed to reliable higher rank.
- NIClassify predictions are model-based; manual checks were applied for focal taxa and unexpected assignments.
- Coordinate sharing should follow permit/sensitivity constraints (use rounded coordinates or centroids where required).
Code/software
- R: 4.4.3
- Quarto: >= 1.5
- Key R packages:
tidyverse,vegan,ggplot2,patchwork,sf,geosphere,readxl,here,janitor,TITAN2,UpSetR(exact set documented inscripts/00_source.R/renv.lock). - CLI tools:
vsearch>= 2.21,cutadapt,BLAST+(NCBI), NIClassify.
Reproducibility / How to run
- Install R (4.4.3) and Quarto.
- Unzip
Rproject.zipat the Dryad root. - Confirm directory structure:
data/
raw_fasta_files/
scripts/
figures/
output/
dryad_README.md
- Render the full figure workflow:
make figures
Or run scripts individually, e.g.:
quarto render scripts/02_initial_analyses.qmd
quarto render scripts/05_titan_breakpoints.qmd
Licenses and reuse
- Data license: CC0
- Code license: CC0
Please cite both the Dryad dataset and the associated article when reusing.
Funding and acknowledgments
Supported by SERDP (DoD/EPA/DOE) under PIs Rosemary Gillespie and George Roderick. Fieldwork and permits were coordinated with Hawaiʻi Volcanoes National Park (EDIAs project), Hakalau Forest NWR, and state partners on Oʻahu. We thank collaborators and field teams [add names], and the Essig Museum for specimen curation.
Ethical, legal, and permit compliance
Include permit numbers, issuing agencies, and any sharing restrictions. Confirm no sensitive taxa are disclosed at vulnerable spatial resolution.
Versioning and change log
- v1.0.0 (2025-10-09): Initial Dryad release with metadata workbook, analysis scripts, and derived tables/figures.
- v2.0.0 (2026-04-01): Review process recommendation
- v2.0.1 (2026-07-07): Resubmission and flagged fixes
Related materials
- Associated manuscript (journal to be added upon acceptance).
- GitHub repository for analysis code / NIClassify settings [optional].
How to cite
Dataset: Weber S., Gillespie R.G., Roderick G.K., et al. (2026). Topography structures arthropod communities across an invaded tropical island revealed by leaf-derived environmental DNA. Dryad Digital Repository. https://doi.org/10.5061/dryad.2z34tmq09
Article: Weber S., et al. (2026). Topography structures arthropod communities across an invaded tropical island revealed by leaf-derived environmental DNA. eDNA , [add DOI].
Sampling design: eDNA was sampled from tree canopies along replicate elevational gradients on Oʻahu. Primary host tree was ʻōhiʻa; paired gradients included koa and strawberry guava on the same slopes where those hosts co-occurred. Analyses in this dataset are Oʻahu-only.
Molecular marker & library prep: Metabarcoding using established arthropod primers (e.g., ANML/MCO; ~180 bp amplicon). PCR replication and negative controls followed standard eDNA QA/QC; final pools were cleaned with 1.5× SPRI beads; QC with Qubit and Bioanalyzer prior to sequencing.
Bioinformatics (summary): Reads were merged, trimmed, and dereplicated; zOTUs were inferred (UNOISE3 via vsearch --cluster_unoise). Read mapping generated a zOTU table (--usearch_global --id 1.00).
Native/introduced classification: NIClassify predictions were parameterized by species-delimitation method (bPTP default), Hawaiian geography setting, and taxonomic split level at order. Predictions were cross-checked manually for focal taxa.
Community analyses: Bray–Curtis and Sørensen dissimilarities were calculated; NMDS ordinations summarized compositional differences. PERMANOVA (adonis2, 999 permutations) tested effects of elevation, site/ridge, and host tree. Multivariate dispersion (betadisper) assessed within-group heterogeneity. Beta diversity was decomposed into turnover and nestedness components. TITAN2 identified taxon change-points along elevation. Overlap among hosts/elevations was summarized via Venn/UpSet.