CoChat

The direct drivers of recent global anthropogenic biodiversity loss

Open access

The DOI matches its Crossref record; 100 of the 100 references it cites were found in a registry or an index; 4 of the 11 sources its claims cite were checked for support.

Abstract

Effective policies to halt biodiversity loss require knowing which anthropogenic drivers are the most important direct causes. Whereas previous knowledge has been limited in scope and rigor, here we statistically synthesize empirical comparisons of recent driver impacts found through a wide-ranging review. We show that land/sea use change has been the dominant direct driver of recent biodiversity loss worldwide. Direct exploitation of natural resources ranks second and pollution third; climate change and invasive alien species have been significantly less important than the top two drivers. The oceans, where direct exploitation and climate change dominate, have a different driver hierarchy from land and fresh water. It also varies among types of biodiversity indicators. For example, climate change is a more important driver of community composition change than of changes in species populations. Stopping global biodiversity loss requires policies and actions to tackle all the major drivers and their interactions, not some of them in isolation.

Overview

CoChat Summary

TL;DR

This paper statistically synthesizes multidriver comparisons to rank the direct anthropogenic drivers of recent global biodiversity loss. It finds land/sea use change dominant overall, direct exploitation second, with marine and biodiversity-indicator-specific hierarchies differing significantly.

Key findings

  • Land/sea use change was the dominant direct driver overall, though not significantly ahead of direct exploitation (bootstrap P = 0.92); both were significantly dominant over climate change (P = 0.034 and P = 0.040, respectively) and invasive alien species (P < 0.0001 for both). Direct exploitation ranked second and pollution third; climate change and invasive alien species were significantly less important than the top two drivers.
  • Driver hierarchies differed significantly among terrestrial, freshwater, and marine realms (randomization P < 0.0001). Direct exploitation was dominant in marine ecosystems with climate change second, whereas land/sea use change ranked first in terrestrial systems and in freshwater ecosystems.
  • The hierarchy varied significantly among the six Essential Biodiversity Variable classes (randomization P = 0.0004). Climate change ranked first for community composition but last for species population changes; land/sea use change ranked first for species populations, ecosystem structure, and ecosystem function; invasive alien species were dominant for species traits; climate change was top-ranked for genetic composition but with an extremely small dataset.
  • Nearly 30% of pairwise comparisons went against the overall ranking, and each driver was the most important in some studies.

Methods

  • Rapid evidence assessment within the IPBES Global Assessment: systematic Web of Science search on 5 September 2018 plus 138 manually included studies/gray literature; 45,162 studies extracted, 3,822 screened, 575 read in full, 189 eligible, and 163 studies used after redundancy checks.
  • Eligible studies were empirical or synthetic studies published in/after 2005 that compared impacts of at least two of five direct drivers—climate change, land/sea use change, direct exploitation, pollution, and invasive alien species—on temporal biodiversity indicators; projected future effects were excluded.
  • Driver rankings were converted into nonredundant pairwise comparisons, weighted by spatial coverage and indicator representation, and synthesized using normalized David’s scores with bootstrapping (10,000 replicates) and randomizations to test differences among realms, IPBES regions, and EBV classes. The overall hierarchy used N = 154 studies.

Limitations

  • Stated: evidence had large-scale geographic and taxonomic biases, with fewer studies from Africa and far more information from vertebrates than plants or invertebrates; most studies did not consider impacts before 1970 (median start year 1983).
  • Stated: comparisons were predominantly qualitative rather than quantitative, and too few studies compared multiple drivers quantitatively to support analysis of those estimates; there were data gaps for species traits and genetic composition, with genetic composition based on an extremely small dataset.
  • Stated: rankings are context dependent—nearly 30% of pairwise comparisons went against the overall ranking, each driver was most important in some studies, and temporal dynamics and interactions among drivers remain poorly understood. Evident from the text: the assessment was a rapid evidence assessment, with some systematic review components simplified or omitted, and the regional/realm/EBV groupings are broad and heterogeneous.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1126/sciadv.abm9982

Cites

100 of 100 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Fig. 1. Dominance hierarchies of the five studied direct drivers of biodiversity loss. (A) Overall hierarchy (N = 154 studies) and (B) hierarchies for terrestrial, marine, and freshwater realms. Area of the circle for each driver is proportional to its dominance score (20) (indicated inside with 95% confidence interval; possible range = 0 to 4). Arrows linking pairs of drivers show the significance of the dominance difference between them based on bootstrapping: Arrow thickness reflects unadjusted P values (thin: P < 0.1, intermediate: P < 0.05, thick: P < 0.01, no arrow: P ≥ 0.1), and arrow shading reflects P values adjusted for multiple testing (black: P < 0.05, gray: P ≥ 0.05). The central triangle in (B) reports the significance of differences in the among-driver dominance hierarchy between pairs of realms (***: randomization P < 0.001; *: P < 0.05; ns: P > 0.5). N gives numbers of studies available for the analysis within each realm. The steepness of the driver hierarchy also rejects the null hypothesis that all drivers have the same impact overall (steepness = 0.405, P = 0.0001), in the terrestrial realm (steepness = 0.465, P < 0.0001), and in the marine realm (steepness = 0.479, P < 0.001), but not in fresh water (steepness = 0.292, P = 0.13).
Fig. 1. Dominance hierarchies of the five studied direct drivers of biodiversity loss. (A) Overall hierarchy (N = 154 studies) and (B) hierarchies for terrestrial, marine, and freshwater realms. Area of the circle for each driver is proportional to its dominance score (20) (indicated inside with 95% confidence interval; possible range = 0 to 4). Arrows linking pairs of drivers show the significance of the dominance difference between them based on bootstrapping: Arrow thickness reflects unadjusted P values (thin: P < 0.1, intermediate: P < 0.05, thick: P < 0.01, no arrow: P ≥ 0.1), and arrow shading reflects P values adjusted for multiple testing (black: P < 0.05, gray: P ≥ 0.05). The central triangle in (B) reports the significance of differences in the among-driver dominance hierarchy between pairs of realms (***: randomization P < 0.001; *: P < 0.05; ns: P > 0.5). N gives numbers of studies available for the analysis within each realm. The steepness of the driver hierarchy also rejects the null hypothesis that all drivers have the same impact overall (steepness = 0.405, P = 0.0001), in the terrestrial realm (steepness = 0.465, P < 0.0001), and in the marine realm (steepness = 0.479, P < 0.001), but not in fresh water (steepness = 0.292, P = 0.13).CC BY-NC · PubMed Central
Fig. 2. Land/sea use change and direct exploitation are the main drivers in all regions. Area of the circle for each driver is proportional to its dominance score (20) (indicated inside with 95% confidence interval; possible range = 0 to 4) within each IPBES region. Arrows linking pairs of drivers show the significance of the dominance difference between them based on bootstrapping: Arrow thickness reflects unadjusted P values (thin: P < 0.1, intermediate: P < 0.05, thick: P < 0.01, no arrow: P ≥ 0.1), and arrow shading reflects P values adjusted for multiple testing (black: P < 0.05, gray: P ≥ 0.05). N gives numbers of studies available for the analysis within each region.
Fig. 2. Land/sea use change and direct exploitation are the main drivers in all regions. Area of the circle for each driver is proportional to its dominance score (20) (indicated inside with 95% confidence interval; possible range = 0 to 4) within each IPBES region. Arrows linking pairs of drivers show the significance of the dominance difference between them based on bootstrapping: Arrow thickness reflects unadjusted P values (thin: P < 0.1, intermediate: P < 0.05, thick: P < 0.01, no arrow: P ≥ 0.1), and arrow shading reflects P values adjusted for multiple testing (black: P < 0.05, gray: P ≥ 0.05). N gives numbers of studies available for the analysis within each region.CC BY-NC · PubMed Central
Fig. 3. The main drivers of biodiversity loss differ among the six EBV classes. Area of the circle for each driver is proportional to its dominance score (20) indicated inside the circle (possible range = 0 to 4). One-way arrows linking pairs of drivers show the significance of the dominance difference between them based on bootstrapping. Two-way arrows between EBV classes indicate significant pairwise differences in their driver dominance hierarchies based on randomization tests. For all arrows, thickness reflects unadjusted P values (thick: P < 0.01, intermediate: P < 0.05, thin: P < 0.1, no arrow: P ≥ 0.1) and shading reflects P values adjusted for multiple testing (black: P < 0.05, gray: P ≥ 0.05). N gives numbers of studies available for the analysis within each EBV class. The question mark indicates the very uncertain rankings for genetic composition because of small sample size. EBV class icons created by C. Gutiérrez of the Humboldt Institute (Bogotá, Colombia) for GEO BON.
Fig. 3. The main drivers of biodiversity loss differ among the six EBV classes. Area of the circle for each driver is proportional to its dominance score (20) indicated inside the circle (possible range = 0 to 4). One-way arrows linking pairs of drivers show the significance of the dominance difference between them based on bootstrapping. Two-way arrows between EBV classes indicate significant pairwise differences in their driver dominance hierarchies based on randomization tests. For all arrows, thickness reflects unadjusted P values (thick: P < 0.01, intermediate: P < 0.05, thin: P < 0.1, no arrow: P ≥ 0.1) and shading reflects P values adjusted for multiple testing (black: P < 0.05, gray: P ≥ 0.05). N gives numbers of studies available for the analysis within each EBV class. The question mark indicates the very uncertain rankings for genetic composition because of small sample size. EBV class icons created by C. Gutiérrez of the Humboldt Institute (Bogotá, Colombia) for GEO BON.CC BY-NC · PubMed Central

Verification

Claims the paper makes

4 Valid · 0 Not valid · 5 Not checked · 67 Own results and interpretations

32 more claims

Create a free account to see all 76 claims and every source verdict.

See all claims — free account
  • It may state two separate claims as one. 16
  • Probably not a substantial claim. 27

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

1 of 1

Citations

Cites

Show all 100 references

Research works printed but not linked: 6

  1. P. Dasgupta The Economics of Biodiversity: The Dasgupta Review (HM Treasury 2021).
  2. P. Balvanera A. Pfaff A. Viña E. Garcia Frapolli S. A. Hussain L. Merino P. A. Minang N. Nagabhatla A. Sidorovich in Global Assessment Report on Biodiversity and Ecosystem Services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services E. S. Brondizio J. Settele S. Díaz and H. T. Ngo Eds. (Secretariat of the Intergovernmental Science-Policy Platform for Biodiversity and Ecosystem Services 2019) pp. 54–200.
  3. A. Purvis Z. Molnár D. Obura K. Ichii K. Willis N. Chettri M. Dulloo A. Hendry B. Gabrielyan J. Gutt U. Jacob E. Keskin A. Niamir B. Öztürk R. Salimov P. Jaureguiberry in Global Assessment Report on Biodiversity and Ecosystem Services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services E. S. Brondizio J. Settele S. Díaz and H. T. Ngo Eds. (Secretariat of the Intergovernmental Science-Policy Platform for Biodiversity and Ecosystem Services 2019) pp. 206–308.
  4. Millennium Ecosystem Assessment Ecosystems and Human Well Being: Synthesis (Island Press 2005).
  5. B. A. Polidoro S. R. Livingstone K. E. Carpenter B. Hutchinson R. Mast N. J. Pilcher Y. Sadovy S. Valenti in Wildlife in a Changing World—An Analysis of the 2008 IUCN Red List of Threatened Species J.-C. Vié C. Hilton-Taylor S. N. Stuart Eds. (IUCN 2008) pp. 55–65.
  6. D. K. Sheehan R. D. Gregory M. A. Eaton P. J. Bubb A. M. Chenery “The Wild Bird Index—Guidance for National and Regional Use” (Cambridge 2010); https://www.rspb.org.uk/globalassets/downloads/documents/conservation-projects/wild-bird-index.-guidance-for-national-and-regional-use.pdf.

Cited by

Items 501–600 of 1,097

Metadata

Authors
Jaureguiberry, P., Titeux, N., Wiemers, M., Bowler, D. E., Coscieme, L., Golden, A. S., Guerra, C. A., Jacob, U., Takahashi, Y., Settele, J., Díaz, S., Molnár, Z., & Purvis, A.
Affiliations
  • Consejo Nacional de Investigaciones Científicas y Técnicas (AR)
  • Helmholtz Centre for Environmental Research (DE)
  • Hot or Cool Institute, Quartiersweg 4, 10829 Berlin, Germany.
  • Rutgers, The State University of New Jersey (US)
  • German Centre for Integrative Biodiversity Research (DE)
  • Alfred-Wegener-Institut Helmholtz-Zentrum für Polar- und Meeresforschung (DE)
  • Institute for Global Environmental Strategies (JP)
  • University of the Philippines Los Baños (PH)
  • Institute of Ecology and Botany (HU)
  • Natural History Museum (GB)
Corresponding author
Purvis, Andy
Venue
Science Advances
Published
2022-11-11
Volume
8
Issue
45
Pages
eabm9982-eabm9982
ISSN
2375-2548 (ISSN-L)
OpenAlex
W4308625497
Cited by
1,121
References
90
Citations by year
2026: 367 · 2025: 396 · 2024: 252 · 2023: 100 · 2022: 6
FWCI
127.38
Type
journal article
Language
en
Open access
Locations
  • Science Advances (publishedVersion)
  • PubMed Central (publishedVersion, cc-by-nc)
  • PubMed (publishedVersion)
  • Helmholtz-Zentrum für Polar-und Meeresforschung (Alfred-Wegener-Institut) (submittedVersion)
Funders
Natural Environment Research Council (NE/V006800/1, NE/M014533/1)
Topics
  • Physical Sciences › Environmental Science › Nature and Landscape Conservation › Ecology and Vegetation Dynamics Studies
  • Physical Sciences › Environmental Science › Global and Planetary Change › Land Use and Ecosystem Services
  • Physical Sciences › Environmental Science › Global and Planetary Change › Conservation, Biodiversity, and Resource Management
Keywords
biodiversity loss, anthropogenic factors, land use change, pollution, climate change, invasive species, marine biodiversity, freshwater biodiversity, community composition change, biodiversity policy

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-02, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-02 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.