henhaobio
Sign in

ScienceEcologyPeer-reviewed journal article

Wildlife trade drives animal-to-human pathogen transmission over 40 years

Jérôme M. W. GippetiD, Colin J. CarlsoniD, Tristan KlaftenbergeriD, Mattéo Schweizer, Evan A. EskewiD, Meredith L. GoreiD, Cléo BertelsmeieriD

PDFDOIPreprintCodeData

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (10 Feb 2025) · CC BY 4.0

Legally traded wild mammals have higher odds (OR 2.72) of sharing pathogens with humans, and each 9 years in global trade adds roughly one human-shared pathogen.

Wildlife trade has been tied to outbreaks such as COVID-19 and the 2003 monkeypox cases, but earlier empirical studies only tallied zoonotic pathogens in traded animals without comparing traded and non-traded species. The work tests that comparison and whether years spent in trade scale with pathogens shared with humans, controlling for research effort, phylogeny and synanthropy.

Key results

  1. ShownTraded wild mammals are more likely than non-traded mammals to share at least one pathogen with humans, after adjusting for research effort, synanthropy, phylogenetic distance to humans, taxonomy and biogeographic realm.

    Half of 1,505 traded species vs eight percent of non-traded species are zoonotic hosts; binomial mixed model OR 2.72 [2.06–3.6], p < 0.0001 (risk ratio 2.4); synanthropy OR 1.94 [1.43–2.62], p < 0.0001 (risk ratio 1.7). · Abstract; Effect of wildlife trade and synanthropy; Effect of trade on the probability of sharing…

  2. ShownAmong CITES-listed mammals, more years present in international trade predicts both the probability of being a zoonotic host and the number of zoonotic pathogens a species hosts.

    Hurdle negative binomial mixed model on 171,749 records for 546 species (1980-2019): zero-inflation sub-model OR 4.4 [2.8–6.7], p < 0.0001; conditional sub-model incidence rate ratio 1.5 [1.3–1.8], p < 0.0001; pathogen richness ranged 0 to 84. · Time in trade predicts zoonotic pathogen…; Effect of time in trade on the number of…

  3. ShownWhen the fitted time-in-trade curve is converted into a rate, a traded mammal species gains on average one pathogen shared with humans per about 9 years of presence in trade.

    Mean years per additional shared pathogen 8.9 [CI95% 7.4–11.2], derived as the reciprocal of the derivative of the hurdle model response, averaged over 1 to 40 years in trade. · Abstract; Time in trade predicts zoonotic pathogen…; Figure legends

  4. SuggestedA structural equation model supports a direct effect of trade on zoonotic host status that synanthropy does not explain; trade and synanthropy also raise research effort, which partly mediates pathogen detection.

    Trade to zoonotic host status std. est. 0.14 (R^2 marginal 0.49); synanthropy to zoonotic status 0.08 and to trade 0.15, with R^2 marginal 0.02 for the synanthropy-to-trade link. · Effect of wildlife trade and synanthropy; Effect of trade on the probability of sharing…

  5. SuggestedConclusions held across four CLOVER datasets: the full set, one without EID2 records (which include experimental infections), one restricted to viruses (the group judged highest pandemic risk), and one applying both filters.

    Zoonotic host species: 871 (full), 799 (no EID2), 699 (viruses), 691 (viruses, no EID2); in the SEM, the main change was a non-significant phylogenetic distance effect in the virus-only sets (p=0.0548 and 0.0598); full outputs in Supplementary Text 6. · Sensitivity analysis; Sensitivity analysis

Methods, data and reproducibility

Mammal species harmonized to the Mammal Diversity Database were linked to CITES and LEMIS trade records and CLOVER host-pathogen data; a binomial mixed model and piecewise structural equation model (SEM) tested trade status, and a hurdle negative binomial mixed model related years in CITES trade (1980-2019) to zoonotic pathogen richness.

  • CITES trade database (aggregated comparative tabulation version 2023.1); 405,066 mammal records filtered to 290,241, with a 1980-2019 time-in-trade subset · wild mammals (CITES Appendix I/II species) · n = 171,749 records in 546 species (time-in-trade analysis)
  • LEMIS database of wildlife and wildlife products imported into the United States, 2000-2014, covering species not listed by CITES · wild mammals
  • Traded species set from legal CITES and LEMIS records (commercial, personal, medical, scientific and hunting trophy purposes) · wild mammals · n = 1,505 traded species
  • CLOVER host-pathogen association database aggregating GMPD 2.0, EID2, HP3 and Shaw; 8,188 zoonotic associations and 2,424 zoonotic pathogens after filtering · mammals incl. humans; viruses, bacteria, fungi, helminths, protozoa · n = 14,383 associations; 1,312 mammals; 4,446 pathogens
  • Mammal Diversity Database taxonomy (order, family, genus) and biogeographic realm, excluding 99 extinct and 38 domesticated species · mammals · n = 6,456 valid mammal species
  • VertLife mammal phylogeny used to compute mean phylogenetic distance to humans in million years · mammals · n = 20 randomly selected trees
  • IUCN Red List habitat codes for human-modified habitats (urban, rural gardens, water storage, aquaculture, wastewater, canals) defining synanthropy · mammals · n = 875 of 6,456 species synanthropic (14%)
  • PubMed title and abstract counts via easyPubMed, combining total publications with disease-focused share into a research effort index · mammals
Methods & tools
R and RStudio, glmmTMB (binomial and hurdle negative binomial mixed models), piecewiseSEM, DHARMa, performance (R package), sjPlot, taxize, easyPubMed, CLOVER host-pathogen association database, Mammal Diversity Database (MDD), VertLife mammal phylogeny, IUCN Red List
Software & environment
R and RStudio with glmmTMB, performance, DHARMa, sjPlot, piecewiseSEM, taxize and easyPubMed; CITES comparative tabulation version 2023.1; package versions not stated.
Reproducibility gaps
At preprint time the code and data repository was marked to be made public only upon publication; full model summaries and sensitivity outputs are in Supplementary Texts not included in the extracted text.
Limitations (6)
  • Host-pathogen links are coded as presence or absence, ignoring variation in prevalence between and within species, so the analysis cannot separate high-risk reservoirs from dead-end or incidental hosts. · stated by the authors
  • Only legal international trade is analyzed; local wild meat and exotic pet markets and illegal trade are not captured, and illegal records were excluded because they are not systematically detected. · stated by the authors
  • CITES records cover only CITES-listed species (about 13% of mammals) and LEMIS only US imports from 2000 to 2014, which limits trade coverage for most species. · stated by the authors
  • The temporal order and direction of transmission remain unresolved. Spillover is argued to dominate over reverse zoonosis, but confirming this needs detailed temporal and genomic data on pathogen transmission. · stated by the authors
  • The design is observational. Years in trade may also track how long a species has been handled, sampled and reported, so pathogen detection could follow trade; a PubMed-based research effort covariate may not fully remove this bias, which weakens the causal reading of 'drives'. · noted in this summary
  • The time-in-trade estimate rests on 546 species in CITES Appendices I or II that were listed by 1980, a regulated and conservation-focused subset, so the 8.9-year rate may not transfer to unlisted or domestically traded mammals. · noted in this summary

Read this if you model zoonotic spillover risk or inform wildlife trade regulation and want effect sizes linking a species' presence and years in trade to the pathogens it shares with humans.

Citations
14
FWCI
76.84
Field percentile
top 1%

Abstract

Wildlife trade affects a quarter of terrestrial vertebrates(1) and creates novel opportunities for cross-species pathogen transmission(2-4). Yet, its precise role in shaping human-wildlife pathogen exchanges remains unclear. Analysing 40 years of global wildlife trade data, we show that traded mammals are twice as likely to share pathogens with humans than non-traded species. Moreover, time spent in trade strongly predicts the number of zoonotic pathogens that wildlife species host: on average, a species shares an additional pathogen with humans for every 9 years of presence in trade. This work highlights the importance of the wildlife trade in driving animal-to-human pathogen transmission, offering key insights into pathogen biogeography in the Anthropocene and the prevention of future pandemics.

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
THISBempah '21Shadloo '26Eskew '19Lee '20Sauer '25Balasubrama '21
66% semantic matchZoology
G. Bempah, M. A. Nartey, K. B. Dakwa+1 more·bioRxiv·Sep 18, 2021