Earth is not simply a planet that happens to host life. According to a new opinion article published in PLOS Ecosystems, it is a planet whose living and nonliving components have coevolved for roughly 4 billion years, the vast majority of Earth’s history, producing a world where we generally find life wherever we look. Woody Turner, recently retired from the Earth Science Division at NASA Headquarters, argues in the article that this deep coevolution makes biodiversity the essential lens through which the Earth system can be understood, and that the state of biodiversity can serve as a proxy for the state of the entire system.
The argument rests on a striking comparison across the natural sciences. Physics currently counts 17 fundamental particles. Chemistry works with roughly an order of magnitude more elements, 118 in total. Geology catalogs another order of magnitude more minerals, with over 6,100 known. Biology, by contrast, spans estimates from more than 8 million to over 100 million species, a jump of three to five orders of magnitude beyond the mineral count. Turner attributes this astonishing diversity to billions of years of coevolution between life and Earth’s highly dynamic physical, chemical, and geological realms, characterized by fluid air, water, and land masses.
From this comparison the article builds a logical chain with far-reaching consequences. If there is no Earth system science understanding without a concomitant understanding of life on the planet, and if life cannot be understood without accounting for its biodiversity, then by the transitive property one cannot truly know the Earth system without knowing its biodiversity. The corollary, Turner writes, is that biodiversity’s status can function as a proxy for the condition of the whole planetary system. This is not framed as an esoteric philosophical point but as the practical foundation for international environmental policy.
The stakes are concrete. The question of how to capture Earth’s biological diversity, in terms of both its status and its trends, lies at the heart of efforts to address the four goals and 23 targets of the Convention on Biological Diversity’s 2022 Global Biodiversity Framework, as well as at least a half dozen other international environmental conventions and dozens of international agreements. The CBD defines biodiversity as the variability among living organisms from all sources, including terrestrial, marine and other aquatic ecosystems and the ecological complexes of which they are part, encompassing diversity within species, between species and of ecosystems. Knowing biodiversity’s status and trends therefore means observing it from genes through organisms to ecosystems across terrestrial, freshwater, marine, and atmospheric biomes.
A central difficulty is that biodiversity and its change are simultaneously global and local, even microscopic, phenomena, depending on whether one observes at the biome, ecosystem, organismal, or genetic level. Just within the organismal level, viruses and the clonal seagrass Posidonia australis differ in size by roughly 12 orders of magnitude. Determining the state of biodiversity on Earth thus requires scaling observations across a wide array of observing systems and then integrating them into a global whole. Getting to trends requires repeated monitoring, and resource managers, conservationists, and investors additionally want forecasts of future trends, which demands cross-scale observational monitoring networks integrated with modeling frameworks that ingest monitoring outputs and generate projections.
The spatial scale challenge typically manifests as a trade-off between observation extent, the size of the overall area covered by a particular observation, and grain size, the smallest resolvable unit within an observation, such as a pixel. The aspiration is global observational coverage with a grain size capturing the size of the organisms of interest, in some cases descending to the size of their molecular material. Most known organisms, the article notes, are a few millimeters in length or smaller, which makes that aspiration technically demanding.
For observations at global extent, including individual biomes, most ecosystems, and a few large organisms such as large canopies, grasslands, clonal stands, and the largest vertebrates, satellite remote sensing offers repeated, increasingly well-calibrated observations in the ultraviolet through the visible, infrared, and microwave regions of the electromagnetic spectrum. This domain is enjoying a technological revolution across instrument types, with higher spatial and spectral resolution passive optical and thermal infrared sensing, along with active laser and radar sensors. Government civil space agencies lead this revolution, but the private sector is playing an increasingly important role.
A parallel revolution is occurring across in-situ observations of smaller ecosystems, organisms, and organismal genetic material. Smaller-scale optical instruments, including camera traps, terrestrial and aquatic video systems such as imaging flowcytobots, citizen science cell phone cameras, drones, and crewed aircraft instruments, are joined by active and passive acoustics, telemetry tags such as magnetometers for tracking animal movement and behavior, and environmental genetic material such as environmental DNA. Together these exemplify the innovative tools now monitoring organisms, communities, and diversity among and within species.
The consequence of these parallel remote sensing and in-situ revolutions is that biodiversity science is awash in data, and the field may be on the brink of escaping the tyranny of the extent versus grain size trade-off. Nonetheless, while multiscale, multimodal observations are abundant, the ability to integrate these data into modeling frameworks for correlative pattern recognition, much less mechanistic and ultimately predictive understanding, remains nascent. This is where artificial intelligence enters the argument. AI is described as a powerful tool for integrating multimodal data and recognizing patterns, and the article identifies an urgent need for more investment promoting the hybridization of a growing number of biodiversity foundation models and geospatial foundation models into multimodal biodiversity-geospatial foundation models.
These hybrid models could incorporate remote sensing products, in situ environmental and weather or climate data, and organismal datasets derived from images, acoustics, telemetry, and genomic observations, as illustrated in the framework figure accompanying the article. With their emergent properties, such foundation models could, at last, make commonplace ecological forecasting capabilities analogous to those already extant in the weather and climate communities. The article emphasizes that this vital work is already happening, and that funding agencies must expedite and broaden these trends. The concluding message is one of unprecedented opportunity: the observations, the computational resources, and the early-stage AI models now exist to monitor the status and trends of biodiversity across space and time, an opportunity to understand and preserve a living planet whose 4 billion years of coevolution have made it, in the author’s view, knowable only through its diversity.
Subject of Research: The role of biodiversity observation and AI-based integration in understanding the Earth system
Article Title: Understanding the Earth system through its biodiversity: Why and how
Article References: Turner, W. (2026). Understanding the Earth system through its biodiversity: Why and how. PLOS Ecosystems, 1(2), e0000044. https://doi.org/10.1371/journal.pesy.0000044
Image Credits: AI Generated
DOI: 10.1371/journal.pesy.0000044
Keywords: biodiversity, Earth system science, remote sensing, ecological forecasting, artificial intelligence, foundation models, Convention on Biological Diversity, environmental DNA, satellite observation, in-situ monitoring, coevolution, PLOS Ecosystems
News Source: Margaret Porter. (November 5, 2026). Biodiversity holds the key to understanding Earth as a living system. Scienmag.
