See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/378236791

Harnessing online digital data in biodiversity monitoring

Article in PLoS Biology · February 2024

DOI: 10.1371/journal.pbio.3002497

CITATIONS READS
0 213

10 authors, including:

Andrea Soriano-Redondo Ricardo Correia

University of Lisbon University of Turku
44 PUBLICATIONS 779 CITATIONS 144 PUBLICATIONS 2,822 CITATIONS

SEE PROFILE SEE PROFILE

Thomas Brooks Ivan Jarić

International Union for Conservation of Nature Université Paris-Saclay
325 PUBLICATIONS 47,079 CITATIONS 163 PUBLICATIONS 4,810 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Andrea Soriano-Redondo on 16 February 2024.

The user has requested enhancement of the downloaded file.

 PERSPECTIVE

Harnessing online digital data in biodiversity

monitoring
Andrea Soriano-Redondo ID1*, Ricardo A. Correia1,2,3, Vijay Barve4, Thomas
M. Brooks5,6,7, Stuart H. M. Butchart8,9, Ivan Jarić10,11, Ritwik Kulkarni1, Richard
J. Ladle12,13, Ana Sofia Vaz13,14,15, Enrico Di Minin1,2,16*
1 Helsinki Lab of Interdisciplinary Conservation Science (HELICS), Department of Geosciences and
Geography, University of Helsinki, Helsinki, Finland, 2 Helsinki Institute of Sustainability Science (HELSUS),
University of Helsinki, Helsinki, Finland, 3 Biodiversity Unit, University of Turku, Turku, Finland, 4 Marine
a1111111111 Biodiversity Center, Natural History Museum of Los Angeles County, Los Angeles, California, United States of
a1111111111 America, 5 International Union for Conservation of Nature (IUCN), Gland, Switzerland, 6 World Agroforestry
a1111111111 Center (ICRAF), University of the Philippines Los Baños, Laguna, Philippines, 7 Institute for Marine and
a1111111111 Antarctic Studies, University of Tasmania, Hobart, Tasmania, Australia, 8 BirdLife International, Cambridge,
United Kingdom, 9 Department of Zoology, University of Cambridge, Cambridge, United Kingdom,
a1111111111 10 Université Paris-Saclay, CNRS, AgroParisTech, Ecologie Systématique Evolution–IDEEV, Gif-sur-Yvette,
France, 11 Biology Centre of the Czech Academy of Sciences, Institute of Hydrobiology, České Budějovice,
Czech Republic, 12 Instituto de Ciências Biológicas e da Saúde, Universidade Federal de Alagoas, Maceió,
Brazil, 13 CIBIO, Centro de Investigação em Biodiversidade e Recursos Genéticos, InBIO Laboratório
Associado, Campus de Vairão, Universidade do Porto, Vairão, Portugal, 14 Departamento de Biologia,
OPEN ACCESS Faculdade de Ciências, Universidade do Porto, Porto, Portugal, 15 BIOPOLIS Program in Genomics,
Biodiversity and Land Planning, CIBIO, Campus de Vairão, Vairão, Portugal, 16 School of Life Sciences,
Citation: Soriano-Redondo A, Correia RA, Barve V, University of KwaZulu-Natal, Durban, South Africa
Brooks TM, Butchart SHM, Jarić I, et al. (2024)
Harnessing online digital data in biodiversity * andrea.sorianoredondo@helsinki.fi (AS-R); enrico.di.minin@helsinki.fi (EDM)
monitoring. PLoS Biol 22(2): e3002497. https://doi.
org/10.1371/journal.pbio.3002497

Published: February 15, 2024 Online digital data from media platforms have the potential to com-
Copyright: © 2024 Soriano-Redondo et al. This is plement biodiversity monitoring efforts. We propose a strategy for
an open access article distributed under the terms integrating these data into current biodiversity datasets in light of the
of the Creative Commons Attribution License,
which permits unrestricted use, distribution, and Kunming-Montreal Global Biodiversity Framework.
reproduction in any medium, provided the original
author and source are credited.

Funding: A.S.R. was supported by a Marie

Skłodowska-Curie Actions Postdoctoral Fellowship Biodiversity is declining at unprecedented rates, which is limiting our capacity to respond to
(101022521). R.A.C. acknowledges personal other sustainability challenges, including climate change. Addressing the biodiversity crisis
funding from the Academy of Finland (348352) and requires collecting accurate, large-scale data for monitoring the status of species, habitats, and
from the KONE Foundation (202101976). R.K. and ecosystems over time and space. Without large-scale biodiversity monitoring, our capacity to
E.D.M. thank the European Research Council (ERC)
investigate the human activities that drive biodiversity loss or that support nature conservation
for funding under the European Union’s Horizon
2020 research and innovation programme and restoration would be severely limited, with catastrophic consequences for nature conser-
(802933). R.K. was supported by the KONE vation and people [1]. These requirements are reflected in the Kunming-Montreal Global
Foundation research grant (202103830). I.J. was Biodiversity Framework, an international agreement that sets out a pathway to achieve harmo-
supported by the Czech Science Foundation (23- nious coexistence with nature by 2050. However, biodiversity monitoring suffers from signifi-
07278S). A.S.V. acknowledges support from the
cant taxonomic and geographic gaps owing to inequalities in resources and capacity [2]. To
FCT - Portuguese Foundation for Science and
Technology through the program Stimulus for
address this shortcoming, we propose the strategic use of the enormous volume of available
Scientific Employment - Individual Support online biodiversity data (originally generated for purposes other than biodiversity conserva-
(2020.01175.CEECIND/CP1601/CT0009), and tion) [3], as a cost-efficient method for monitoring biodiversity and human activities.
project ClimateMedia - Understanding climate Such online digital data can be used to strengthen existing assessments of the status and
change phenomena and impacts from digital trends of biodiversity, the pressures upon it, and the conservation solutions being
technology and social media (2022.06965.PTDC).

PLOS Biology | https://doi.org/10.1371/journal.pbio.3002497 February 15, 2024 1/5

PLOS BIOLOGY

The funders had no role in study design, data implemented, as well as to generate novel insights about these aspects, nature’s contributions
collection and analysis, decision to publish, or to people, and human–nature interactions [3,4]. The most common sources of online biodi-
preparation of the manuscript.
versity data include web pages, news media, social media, image- and video-sharing platforms,
Competing interests: The authors have declared and digital books and encyclopedias [5]. These data can be filtered and processed by research-
that no competing interests exist. ers to target specific research questions and are increasingly being used to explore ecological
processes and to investigate the distribution, spatiotemporal trends, phenology, ecological
interactions, or behavior of species or assemblages and their drivers of change [3].
Digital data are also being used to explore human–nature interactions from multiple per-
spectives. For example, photographs from Flickr have been used to explore cultural ecosystem
services, Twitter/X or YouTube have been used to identify instances of illegal wildlife trade,
and Wikipedia page-views have been used to assess observable patterns in seasonal phenom-
ena of plants and animals [5]. Such diversity of uses showcases the great potential of these data
to provide novel insights into nature and human–nature interactions, and to track global bio-
diversity trends through continued monitoring. However, it also shows the unstructured
nature of this area of research, as researchers mostly select digital media platforms ad hoc.
Moreover, digital data are not currently integrated into standardized biodiversity monitoring
programs, thereby preventing a streamlined utilization of these digital media resources for bio-
diversity conservation.
In support of the Kunming-Montreal Global Biodiversity Framework targets and the UN
Sustainable Development Goals, we are calling for urgent action to identify, collect, filter,
extract, store, integrate, share, and disseminate digital online data to better capture biodiversity
status and trends in near real-time (Fig 1). Such data include complementary text, image,

Fig 1. Proposed framework for integrating online digital data into biodiversity monitoring. Workflow of the proposed framework, from data collection to
data harmonization.
https://doi.org/10.1371/journal.pbio.3002497.g001

PLOS Biology | https://doi.org/10.1371/journal.pbio.3002497 February 15, 2024 2/5

PLOS BIOLOGY

video, and/or sound content pertaining to the natural world that is uploaded into the digital
realm, and that can be integrated into current datasets and monitoring schemes on biodiver-
sity, from individual organisms to whole ecosystems, and the way these dimensions intersect
with human society [5].
To collect meaningful digital data for biodiversity monitoring, we propose an approach that
follows standard assessments of species (e.g., the International Union for Conservation of
Nature (IUCN) Red List of Threatened Species), ecosystems (e.g., the IUCN Red List of
Ecosystems [6]), and nature’s contributions to people [7], as well as classification schemes for
threats and conservation actions (e.g., Unified Classifications of Threats and Actions [8]). The
approach would consist of continually searching and retrieving data for a selected list of key-
words or topics that are particularly relevant to complement current data streams, through
direct scraping or via dedicated open Application Programming Interfaces (APIs) of social
media platforms or search engines. For example, a set of keywords could include the list of
names of all described species in multiple languages.
Moving forward, the collection of online digital data will require filtering and the extraction
of relevant information for biodiversity from large corpora, and hence the construction of
automated pipelines that can leverage developments in machine learning methods (Fig 1). To
filter data from digital text, these pipelines would need to deduplicate information and remove
irrelevant entries using, for example, text vectorization algorithms and artificial neural net-
works [9]. Named entity recognition could then be used to extract specific data such as the
name of a species, timestamp, geographic coordinates, and/or other quantifiable information
(e.g., quantities and prices) to be included in structured biodiversity datasets. Classifying infor-
mation from digital images will follow a similar process, as it requires the implementation of
machine vision models to tease apart relevant and irrelevant images [10].
One of the major limitations of these methods is that they require the creation of high-qual-
ity labeled training datasets covering all the necessary content. An additional challenge in the
case of text is the global diversity of languages, which would necessitate multilingual models.
Moreover, after extraction, data would require automated validation and flagging of uncertain
records to minimize erroneous information entering the data stream. Machine learning tech-
niques could be used to partially overcome some of the limitations, such as the uneven cover-
age of digital data across regions, time, taxa, and ecosystems (for example, areas with low
human population densities are likely to be underrepresented in the datasets and charismatic
species to be overrepresented) by extrapolating some of the missing data.
Following the proposed framework, only the specific data required to monitor biodiversity
would be collected and stored, the rest would be discarded, and any information that could be
used to identify individuals would be pseudonymized [11]. Data minimization and pseudony-
mization principles would be included and enforced in the pipelines to minimize the risk of
leaking sensitive information. Publication of data derived from online sources would also need
to take account of recognized best practices in minimizing risk of harm to biodiversity posed
by revealing the precise locations of highly threatened or valuable species [12].
Data generated through the framework in near real-time could be continuously integrated
with other independently collected datasets and used for real-time applications; for example,
as part of integrated spatial conservation planning assessments and adaptive management (Fig
1). Species occurrence data could be collected following Biodiversity Information Standards,
such as the Darwin Core Standard, and uploaded into the Global Biodiversity Information
Facility (GBIF). Data relevant to assessment of species extinction or ecosystem collapse risk,
perhaps most usefully relating to threats, could be mobilized into the workflows for generating
the IUCN Red List of Threatened Species and Red List of Ecosystems, respectively, while data
on sites of global significance for the persistence of biodiversity (and the occurrence of species

PLOS Biology | https://doi.org/10.1371/journal.pbio.3002497 February 15, 2024 3/5

PLOS BIOLOGY

within them) could be served to the appropriate national coordination groups to strengthen
their efforts in identifying Key Biodiversity Areas. Moreover, data on the illegal wildlife trade
could be integrated with the Convention on International Trade in Endangered Species of
Wild Fauna and Flora (CITES) Trade Database or the Trade Records Analysis of Flora and
Fauna in Commerce (TRAFFIC) open-source wildlife seizure and incident data.
The necessary technology to implement this framework is within reach; it only requires
willingness and resources to put it into action. Reaching the full potential of the framework
will require harnessing expertise from multiple sectors and academic disciplines, as well as the
collaboration of digital media companies. The proprietary companies that own social media
platforms, search engines, and other digital platforms and their respective APIs, often limit
access to their content, either by capping it to a certain volume and/or by requiring paid sub-
scription. Data availability varies greatly depending on the platform, and platform policies can
change at any given time, further restricting (or granting) access. Therefore, the full potential
of a digital observatory of biodiversity will only be achieved if digital media companies provide
researchers with unfiltered access to all relevant content. Furthermore, the long-term sustain-
ability of such a system will require the commitment of an established international and inter-
governmental organization to maintain and host the system and update its pipelines, as well as
addressing data quality concerns through engagement of expert communities and response to
feedback.
The current lack of integration of online digital data into biodiversity monitoring under-
mines efforts to collect all critical information relevant to identify conservation priorities and
implement actions. While such a system should not divert resources from other monitoring
efforts, leveraging the ongoing digital data revolution would generate complementary and
timely evidence to inform decision-making on the conservation, restoration, and sustainable
use of nature and its contributions to people.

Acknowledgments
The authors thank Núria Altimir for assistance in developing the conceptual figure and Tim
Hirsch for suggestions during the drafting of the manuscript.

References
1. Dı́az S, Settele J, Brondı́zio ES, Ngo HT, Agard J, Arneth A, et al. Pervasive human-driven decline of
life on Earth points to the need for transformative change. Science. 2019;366:eaax3100. https://doi.org/
10.1126/science.aax3100 PMID: 31831642
2. Moussy C, Burfield IJ, Stephenson PJ, Newton AFE, Butchart SHM, Sutherland WJ, et al. A quantitative
global review of species population monitoring. Conserv Biol. 2022; 36:e13721. https://doi.org/10.1111/
cobi.13721 PMID: 33595149
3. Jarić I, Correia RA, Brook BW, Buettel JC, Courchamp F, Di Minin E, et al. iEcology: Harnessing Large
Online Resources to Generate Ecological Insights. Trends Ecol Evol. 2020; 35:630–639. https://doi.
org/10.1016/j.tree.2020.03.003 PMID: 32521246
4. Ladle RJ, Correia RA, Do Y, Joo G-J, Malhado AC, Proulx R, et al. Conservation culturomics. Front
Ecol Environ. 2016; 14:269–275. https://doi.org/10.1002/fee.1260
5. Correia RA, Ladle R, Jarić I, Malhado ACM, Mittermeier JC, Roll U, et al. Digital data sources and meth-
ods for conservation culturomics. Conserv Biol. 2021; 35:398–411. https://doi.org/10.1111/cobi.13706
PMID: 33749027
6. Keith DA, Ferrer-Paris JR, Nicholson E, Bishop MJ, Polidoro BA, Ramirez-Llodra E, et al. A function-
based typology for Earth’s ecosystems. Nature. 2022; 610:513–518. https://doi.org/10.1038/s41586-
022-05318-4 PMID: 36224387
7. Dı́az S, Pascual U, Stenseke M, Martı́n-López B, Watson RT, Molnár Z, et al. Assessing nature’s contri-
butions to people. Science. 2018; 359:270–272. https://doi.org/10.1126/science.aap8826 PMID:
29348221

PLOS Biology | https://doi.org/10.1371/journal.pbio.3002497 February 15, 2024 4/5

PLOS BIOLOGY

8. Salafsky N, Salzer D, Stattersfield AJ, Hilton-Taylor C, Neugarten R, Butchart SHM, et al. A Standard
Lexicon for Biodiversity Conservation: Unified Classifications of Threats and Actions. Conserv Biol.
2008; 22:897–911. https://doi.org/10.1111/j.1523-1739.2008.00937.x PMID: 18544093
9. Kulkarni R, Di Minin E. Automated retrieval of information on threatened species from online sources
using machine learning. Methods Ecol Evol. 2021; 12:1226–1239. https://doi.org/10.1111/2041-210X.
13608
10. Kulkarni R, Di Minin E. Towards automatic detection of wildlife trade using machine vision models. Biol
Conserv. 2023; 279:109924. https://doi.org/10.1016/j.biocon.2023.109924
11. Di Minin E, Fink C, Hausmann A, Kremer J, Kulkarni R. How to address data privacy concerns when
using social media data in conservation science. Conserv Biol. 2021; 35:437–446. https://doi.org/10.
1111/cobi.13708 PMID: 33749044
12. Chapman AD. Current Best Practices for Generalizing Sensitive Species Occurrence Data. Copenha-
gen: GBIF Secretariat; 2020. Available from: https://doi.org/10.15468/doc-5jp4-5g10.

PLOS Biology | https://doi.org/10.1371/journal.pbio.3002497 February 15, 2024 5/5

View publication stats