- A global study integrates field surveys, satellite data, and citizen-science observations to map plant traits, revealing how plants function across ecosystems.
- Citizen science fills data gaps and improves models, but biodiversity data still favour temperate regions, leaving tropical areas like India underrepresented.
- For India, better use of existing ecological data, stronger validation networks, and greater participation in global databases could improve conservation planning and climate research, positioning the country as a key contributor to global biodiversity science.
What does a global map of plant life look like, and what happens when the data behind it is incomplete?
A recent study published in Nature Communications in January 2026, describes such a map, built from field surveys, earth observation systems, and millions of observations recorded by citizen scientists around the world.
This map now offers one of the most in-depth views of how plants function across ecosystems. However, the map also exposes something else. These are large, persistent gaps in the data that scientists rely on to understand the Earth’s vegetation, which means that quite a bit of the world’s plant life is still poorly documented.
The study used 31 plant traits such as size, growth strategy, leaf characteristics, wood density, reproductive traits, and resource use to outline a global ‘plant economics’ spectrum. These characteristics, also known as functional traits, can help us understand how plant strategies change in response to climate and ecosystem stress.
Currently, most global biodiversity data only tell us what species are found where; they don’t tell us what roles they play in carbon storage and ecosystem dynamics. Mapping these traits on a global scale gives us a spectrum of characteristics spanning fast-growing, nutrient-hungry plants to slow-growing, stress-tolerant ones and how these traits support plant growth, survival, adaptation, and persistence in an ever-changing world. This is especially important for informing models on energy, nutrient, and water cycles which are increasingly being used to plan infrastructure, agricultural, and energy strategies in a world faced with climate change.
Sourcing the data from scientists, citizens, and satellites
The researchers used a combination of data from detailed field surveys collected by scientists, millions of observations from citizen scientists, and environmental information derived from satellites and climate records to create this global plant trait map.
They then used machine-learning models to link the plant traits with environmental conditions like temperature, rainfall, and soil properties to predict plant traits in places where direct measurements were unavailable. The models were generated using three approaches, namely, scientific surveys only, citizen science only, and both combined.
But bringing these different datasets together was no small task. As study author Daniel Lusk explains, “scale and integration” were the biggest challenges in this work.
“We were working with ~340 million citizen science observations from GBIF, 2.5 million vegetation survey plots from sPlot, trait data for over 74,000 species from TRY, and roughly 150 Earth observation predictor layers, all at multiple spatial resolutions. Just getting these datasets to talk to each other was an enormous data engineering challenge before we could even begin the science. Spatial cross-validation at global scale is also computationally demanding and required careful design to ensure we were honestly testing the models’ ability to generalise to new regions rather than just memorising local patterns,” says Lusk, a researcher in the Chair of Sensor-based Geoinformatics (geosense) at the University of Freiburg.
How citizen science helped to fill gaps in the maps
This exercise generated maps that estimate plant characteristics across the Earth at different spatial scales, including those at very fine resolutions (~1 square kilometre) and others at broad resolutions.
Adding citizen-science data made the models more reliable, especially when applied to new or poorly sampled regions. This means that the predictions worked better outside the areas where data were originally collected, especially at finer spatial scales. Next, the researchers checked how accurate each data approach was at the finest resolution. Models built only from citizen-science data were generally less accurate than those using structured scientific surveys. However, combining both datasets performed almost as well as the scientific surveys alone, while offering broader coverage.
“This study demonstrates that citizen science observations, when integrated with expert vegetation plots (sPlots) and functional trait databases (TRY-Database), can generate high-resolution global maps of plant functional traits with high accuracy for 15 out of 31 traits studied,” says Shyam S. Phartyal, who is a professor at Mizoram University, and a co-author of this study. “The resulting maps outperform previous ones, filling critical gaps in under-sampled ecosystems and reducing uncertainty in ecological modelling,” he adds.
“Honestly, the most exciting moment was when we saw that models trained on citizen science data alone, without any structured vegetation surveys, could produce trait maps that rivalled or even outperformed previously published maps that relied on direct trait measurements,” says Lusk.
This shift, from using small, carefully curated datasets to massive, crowd-generated ones, is an important step in ecology. Vijay Barve, who is part of the Asia Regional Support Team at the Global Biodiversity Information Facility says, “A key methodological novelty in this study lies in the synthesis of heterogeneous data sources — namely citizen science observations, curated professional trait databases, and remote sensing imagery — to enhance global monitoring capabilities. Citizen science significantly expands our data reach; even with its inherent biases, it provides essential coverage that traditional research datasets compiled by mere scientists simply can’t match.”
What citizen science could not cover
While these results were promising, there was one glaring issue that this study could not address. Both, the scientific surveys and the citizen science data, were clustered around temperate regions such as Europe, North America, Japan, and Australia, which are well-studied. Tropical and subtropical regions including large parts of South Asia, Southeast Asia, and Africa, were comparatively underrepresented, even with the addition of citizen-science observations.
“This is likely driven in large part by data scarcity itself — these regions simply have far fewer training observations — but the higher species richness and structural complexity of tropical systems probably compounds the problem,” says Lusk.
“Approximately 800 vegetation plots associated with sPlot from India have been contributed by three of us and used in this study. These plots span a wide ecological gradient, from the western Himalayan alpine region to the Doon Valley and the savanna ecosystems of Maharashtra. In addition, around 240,000 citizen science observations from India were incorporated into the study, which is still not substantially enough to strengthen the spatial coverage and representation of Indian plant diversity, keeping heterogeneous landscapes of India,” adds Phartyal.
In addition, although citizen science offers a powerful way to expand data coverage, it has certain limitations. For example, observations are often clustered in accessible areas and may be biased toward certain species.
“The landscape of biodiversity monitoring is evolving rapidly as platforms like iNaturalist, eBird, and Pl@ntNet become increasingly sophisticated. While these tools democratise data collection and provide unprecedented global coverage, researchers must remain vigilant regarding their inherent biases — such as spatial clustering and taxonomic skew. By acknowledging these limitations and implementing robust analytical workarounds, scientists can effectively harness the full potential of crowdsourced data for global conservation,” says Barve.
India’s contributions to the study
“India is significantly underrepresented. It contributes about 244,000 georeferenced vascular plant records to GBIF, just 0.07% of the global total of ~339 million, despite being home to roughly 18,000 known vascular plant species. To put that in perspective, India has about 171 GBIF records per million people, compared to over 1.1 million for France and ~66,000 for the USA. From the sPlot vegetation survey database, India has about 5,500 plots, just 0.25% of the global total,” says Lusk.
Phartyal points out that in major global databases such as TRY, India’s representation remains disproportionately low. Although India is recognised as the home of four biodiversity hotspots, our contribution to global ecological datasets is still limited, except for a few well-supported subjects of ecological science where charismatic megafauna (such as tigers and elephants) is a research subject.
Even within India, the data that does exist is fairly concentrated, with iNaturalist accounting for 52% of India’s GBIF plant records, points out Lusk. “Observations are clustered around urban centres and a few well-surveyed states like Tamil Nadu, Maharashtra, and Karnataka. Vast areas of central India, the northeast, and the Himalayas remain extremely data-sparse,” he adds.
Opportunities for India in a changing field
For the researchers involved in the study, these gaps represent opportunities.
“India is in a fascinating position. It has a huge and growing base of citizen scientists (GBIF contributions from India have been growing at roughly 31% per year) and a rich institutional tradition in botanical and ecological research. Yet despite this institutional capacity, India’s representation in global open-data platforms like sPlot and GBIF remains relatively low. The knowledge obviously exists but isn’t yet flowing into the kinds of standardised, open datasets that power global analyses like ours. Bridging that gap is a huge opportunity,” says Lusk.
Phartyal points out that by harnessing scattered vegetation survey data, currently lying underutilised in hundreds of plant ecology and botany theses across Indian universities and research organisations, and integrating them with citizen science data and trait-based ecological approaches, India can substantially strengthen its conservation and climate-change strategies.
Phartyal also says that India needs more researchers in biodiversity, ecology, data science, taxonomy, and trait-based research if it truly aims to become a global leader in ecological science. “India is generating unprecedented volumes of citizen science data to document and map biodiversity. However, significant bottlenecks remain in expert validation, as not all observations reach ‘research-grade’ status and therefore cannot always be used in scientific analyses. We really need to expand expert networks to strengthen data validation,” he adds.
“I’d say this is a genuinely exciting time to be entering ecology, particularly from India. India’s megadiversity is a real scientific treasure trove: the Western Ghats, the Eastern Himalayas, and the Indo-Burma region are biodiversity hotspots where additional data and analysis would have significant global impact,” says Lusk.
As Lusk reflects, even individual contributions matter. “It means that every observation someone uploads to iNaturalist or Pl@ntNet isn’t just a data point in isolation; it’s contributing to a global picture of how ecosystems function,” he says.
For now, that picture is still incomplete. But as more data flows in from scientists, satellites and citizen observers, the global plant trait map is likely to become richer, more detailed and more representative of the world it seeks to describe.
Read more: From diversity to monotony, ecological communities are homogenising
Banner image: A solitary elephant moves through the forests of Nelliyampathy in the Western Ghats. Image by Thangaraj Kumaravel via Flickr (CC BY-NC-ND 4.0).
Credits
