henhaobio
Sign in

Ecology and EvolutionEcologyPeer-reviewed journal article

A breath of fresh air: comparative evaluation of passive versus active airborne eDNA sampling strategies

Hugo Jager, Krijn Baptist TrimbosiD, Jan‐Maarten LuursemaiD, Adrianus G. C. L. Speksnijder, Kathryn A. StewartiD

PDFDOIPreprint
In-depth full-text digest available · Jump to digest ↓

Abstract

Terrestrial biodiversitys rapid decline demands expansion of high-resolution biomonitoring to support science-based policy. Environmental DNA (eDNA) analysis has proven effective in aquatic systems but remains underexplored for terrestrial habitats, partly due to point-source sampling bias and difficulty in upscaling. Air has emerged as a promising substrate yet most studies use active samplers which tend to be expensive, bulky, and require power. We used a newly developed, inexpensive, reusable, and easy-to-use passive airborne eDNA sampler (Nutshell eDNA sampler) to capture eDNA suspended in air across time (6 to 96 hours) within Rotterdam Zoo, the Netherlands. Its performance was compared to two commonly-used active airborne eDNA samplers for vertebrate diversity detection. In total, 88 species were detected, including 24 zoo residents. The Nutshell eDNA sampler was the most effec-tive at detecting zoo residents, surpassing active samplers in species richness within 48 hours and continuing to accumulate new species beyond 96 hours, including detections of both patchy and singleton signals. It also detected the furthest species signal (515 m). Zoo airborne eDNA further demonstrated a positive correlation to species total biomass, suggest-ing larger vertebrates release proportionately more DNA into the air. Our findings indicate that for long, unsupervised biomonitoring, passive airborne eDNA sampling presents a prom-ising approach for assessing vertebrate communities and putatively reduces detection noise in stochastic air eDNA signals. While deeper investigations into airborne eDNA sampling strategies are needed, passive methods can offer a much needed logistically flexible, low-maintenance approach compared to short-burst collection strategies employed by many ac-tive samplers.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (18 Nov 2025) · CC BY 4.0

Passive 3D-printed Nutshell air eDNA samplers hung up to 96 hours at Rotterdam Zoo detected 76 vertebrate species, versus 50 and 25 for two active samplers.

Terrestrial environmental DNA (eDNA) biomonitoring is hard to scale, and airborne eDNA studies mostly rely on costly, bulky, power-hungry active pumps run in short bursts. The authors report that passive and active air collection had not been compared before, so this work tests a cheap, reusable, open-source passive sampler against two active devices for vertebrate detection, accumulation over time, and detection distance.

Key results

  1. ShownPooled over all deployments, the passive Nutshell strategy recovered more vertebrate species than either active sampler. However, the Pollensniffer was more efficient per sample and at matched deployment times.

    Species detected: 76 Nutshell (62.17% of reads), 50 Pollensniffer, 25 Coriolis Micro; 30 unique to Nutshell; per-location richness F(2,12) = 43.21, p < 0.001. Per-sample rate 0.84 species/sample (Nutshell, n = 90) vs 2.78 (Pollensniffer) and 1.39 (Coriolis Micro), n = 18 each; Nutshell below Pollensniffer at every matched time point (n = 3). · Species detection; Accumulating species richness detection over…; Species detection within Rotterdam Zoo

  2. ShownPassive richness per outdoor site kept rising through 96 hours without a plateau. It overtook the Coriolis Micro between 6 and 24 hours and the Pollensniffer between 24 and 48 hours (for zoo species, between 48 and 72 hours).

    Per outdoor site at 96 hours, Nutshell averaged 15.07 (SD ± 2.62) species vs 6.6 (SD ± 1.38) for Pollensniffer and 3.47 (SD ± 1.15) for Coriolis Micro after 10 minutes; zoo species 2.87 (SD ± 1.57) vs 1.67 (SD ± 0.82) and 0.4 (SD ± 0.43); pooled outdoor richness rose from 19 species (6 hours) to 38 (96 hours). · Accumulating species richness detection over…; Accumulation patterns of airborne eDNA

  3. ShownMost taxa caught by the passive samplers appeared only at some deployment durations rather than persisting. The authors attribute this mainly to pulsed eDNA release or uneven mixing of air, rather than to signal lost from the filters.

    Of 74 outdoor Nutshell taxa, 51.4% (n = 38) were singletons, 25.7% (n = 19) patchy, 12.2% (n = 9) found at every time point and 23% (n = 17) retained from first detection to 96 hours; of 19 zoo residents, 47.4% (n=9) singletons, 15.8% (n=3) patchy, 36.8% (n=7) retained. · Accumulating species richness detection over…; Accumulation patterns of airborne eDNA

  4. SuggestedPassive samplers detected zoo residents at the largest mean and maximum distances, including an American bison 515 m from its enclosure. Differences in mean distance between strategies were not significant.

    Mean detection distance 188.67 ± 61.33 m (Nutshell) vs 172.58 ± 91.29 m (Pollensniffer) and 68 ± 50.21 m (Coriolis Micro), F(2,10) = 2.8, p = 0.11; outdoor maximum 373 ± 124.37 m for Nutshell, above Coriolis Micro (p = 0.036); within a 68 m radius, detection accuracy 33.2% ± 20.4% vs 21.6% ± 21.7% (Pollensniffer). · Accuracy, dispersion, and biomass; Spatial distribution of target species

  5. SuggestedTotal airborne eDNA reads for detected zoo species rose with the total biomass housed. This fits the idea that larger vertebrates, or larger populations, shed more DNA into air.

    Pearson correlation between total reads and log-transformed total biomass (published mean mass times individuals housed): r(24) = 0.44, p = 0.031; after removing four large outliers, r(20) = 0.38, p = 0.097. · Quantification and statistical analyses; Accuracy, dispersion, and biomass; Spatial distribution of target species

Methods, data and reproducibility

Passive Nutshell samplers (glass-fibre filters retrieved at 6 to 96 hours) and 10-minute active Pollensniffer and Coriolis Micro runs were deployed at six Rotterdam Zoo sites over three weekly campaigns. Extracts were metabarcoded with 12S and 16S rRNA primers on Illumina MiSeq, denoised with DADA2, assigned by BLASTn, and compared by ANOVA.

  • Passive Nutshell eDNA sampler glass fibre filters (47 mm, 0.7 µm) hung at ∼2.5 m and retrieved at 6, 24, 48, 72 and 96 hours at six zoo sites · vertebrates (mammals, birds, fish, amphibians) · n = 90 samples (75 outdoor)
  • Active Pollensniffer samples on Vaseline-coated slides (∼8 L/min, 10-minute runs at 1 m height) · vertebrates (mammals, birds, fish, amphibians) · n = 18 samples
  • Active Coriolis Micro samples collected into 4 mL PBS (∼100 L/min, 10-minute runs at 1 m height) · vertebrates (mammals, birds, fish, amphibians) · n = 18 samples
  • Field negative controls per device, PCR negatives, and a harbor porpoise (Phocoena phocoena) DNA positive PCR control · n = three field negatives per device
  • 12S and 16S rRNA metabarcoding reads from all DNA extracts on Illumina MiSeq (3,862,876 raw 12S and 5,871,156 raw 16S reads) · vertebrates · n = 131 DNA extracts; 88 species detected · FigShare DOI 10.6084/m9.figshare.28842980
  • Rotterdam Zoo animal inventory, including off-exhibit residents, used to classify detections as zoo species · vertebrates (zoo residents) · n = N = 266
  • Published mean body mass values multiplied by numbers of housed individuals to give total biomass per zoo species · zoo vertebrates
Methods & tools
Nutshell eDNA sampler (passive, mostly 3D printed), Pollensniffer (active, impingement onto slides), Coriolis Micro (active, vortex into liquid), 12S05 and 16Smam1/16Smam2 primers with human blocker, QIAshredder (QIAGEN) DNA extraction, Illumina MiSeq v3, 150 bp paired-end, Nextera XT indexing, Cutadapt v4.9, DADA2, BLASTn vs GenBank with weighted LCA on Naturalis OpenStack GALAXY, vegan v.2.6.6.1 and ggplot2, R 4.4.0 (ANOVA, Tukey's HSD, Pearson correlation), Google Earth (detection distance estimation)
Reagents & environment
R 4.4.0; Cutadapt v4.9; vegan v.2.6.6.1; DADA2 and ggplot2 without stated versions; taxonomy assigned via a custom BLASTn and weighted LCA pipeline on Naturalis OpenStack GALAXY.
Reproducibility gaps
No code or pipeline repository is linked (the Galaxy taxonomy pipeline is custom), and the open-source Nutshell design is described only in Supplementary Information. All data are archived under FigShare DOI 10.6084/m9.figshare.28842980.
Limitations (6)
  • Sampling effort was unbalanced, with 90 passive versus 18 samples per active device; per sample, the Nutshell recovered 0.84 species versus 1.39 for the Coriolis Micro and 2.78 for the Pollensniffer. · stated by the authors
  • The passive sampler hung higher (∼2.5 m vs ≤1 m) and far longer than the active devices, and its cumulative sampled air volume could not be quantified, so what drives its higher richness remains unresolved. · stated by the authors
  • The best Nutshell configuration is still unknown: filter material, deployment duration, season, sampler height and air movement were not explored, and wind direction, thought to matter for passive capture, was not recorded. · stated by the authors
  • At the single indoor site the Pollensniffer outperformed the passive sampler, plausibly because of weak indoor airflow, and indoor active detections up to 491 m away show how hard it is to trace airborne eDNA to its source. · stated by the authors
  • Many nearby residents went undetected (only 24 species, 9% of the inventory, were found), which the authors link to particle heterogeneity, wind, UV degradation, low shedding rates or missing reference sequences. · stated by the authors
  • The biomass-read correlation pools reads across three samplers and two primer sets with different amplification biases, and loses significance (p = 0.097) after four outliers are excluded, so it does not yet support abundance estimation. · noted in this summary

Read this if you plan terrestrial eDNA biomonitoring and want field evidence on whether cheap passive air samplers left out for days can match powered active samplers for vertebrate detection.

Citations
14
FWCI
38.58
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISMartino '25Lynggaard '22Gygax '26Garrett '26Lynggaard '22Clare '21
84% semantic matchEcology
Francesca Martino, Christine Elizabeth Cooper, Josh Newton+8 more·bioRxiv·Sep 2, 2025· 1 citations