Integrate Remote Sensing Into Ecological Research to Realise Its Potential for Monitoring in the Antarctic

Significant, rapid technical advances are being made that expand and enhance the potential use of remote sensing-derived data to estimate multiple facets of biodiversity (Reddy et al. 2021). The use in biodiversity models of remotely sensed physical environmental variables and measures of ecosystem extent and connectivity is well established. However, progress is now also being made towards estimating a suite of compositional and functional biodiversity variables, for example, tree species richness, canopy height, leaf nitrogen content, and primary productivity (Cavender-Bares et al. 2022). The integration of multiple forms of RS-derived information into biodiversity models holds enormous promise. However, research to support this advance is still in its infancy (Gillespie et al. 2024). For example, the accuracy of biodiversity indices derived from remote sensing can be low, and RS techniques struggle to capture fine-scale ecological variation. One particular strand of work needed, as pointed out in the Invited Opinion in Global Change Biology, Colesie et al. (2025), is the calibration of remotely sensed signatures for specific and unique habitats that differ in the diversity, identity, and structure of their vegetation. The effectiveness of remote sensing techniques can vary significantly across different ecosystems. Unlike tropical and temperate forests, climatically extreme regions such as polar, alpine, and desert ecosystems, including the Antarctic, are typified by lower vegetation cover and stature. The vertical structural complexity of these ecosystems is also limited to very fine spatial scales. These properties make them no less complex than tropical forests. They similarly demand mechanistic understanding and life history knowledge in order to benefit from the upscaling (spatial and temporal) in biodiversity modelling made possible by RS technology. High-quality evidence for multi-decadal shifts in plant community composition and expansion of vegetation cover is available for few places in the region (one example is the use of photogrammetry in the sub-Antarctic, van der Merwe et al. 2024). The need to gain broad, regional estimates of the extent, rate, and variation in this change is therefore urgent. Terrestrial vegetation on the Antarctic continent and maritime islands is notorious for its variability, driven by microtopography, microclimate, and micro-scale variation in sediment properties, including nutrient availability (Tóth et al. 2025). Colesie et al. (2025) provide a helpful list of biological and technological considerations. If incorporated into research on vegetation change that uses RS-derived response variables as proxies for biodiversity, these considerations can enhance the information value and robustness of inferences made about vegetation change. The biological factors that can affect the spectral properties of vegetation, and hence the values of indices derived from RS imagery, include for example the water status of vegetation and seasonal pigmentation changes in plant tissue. Phenological and irregular fluctuations in biomass can also confound the interpretation of trends in vegetation dynamics if the timing of sensor data capture does not coincide meaningfully with the natural dynamics of plant species and communities. This includes, for example, knowledge of minimum plant growth rates to inform the time periods over which meaningful directional change in vegetation should be interpreted. In addition to species and ecosystem properties, a number of technological factors significantly influence the spectral properties of vegetation and subsequently affect the accuracy of indices derived from remote sensing imagery. As Colesie et al. (2025) point out, the NDVI detection threshold presents a particular challenge. Sensitivity analysis requires tailored thresholds specific to each study area, especially considering that slow-growing vegetation may take decades to accumulate sufficient coverage to exceed established detection limits. Mixed pixel effects further complicate vegetation analysis, as individual pixels often contain multiple vegetation types alongside non-vegetated surfaces, which reduce classification accuracy and can mask small-scale vegetation changes. Spectral overlap creates additional difficulties in distinguishing vegetation of different types, particularly among the major taxonomic groups that dominate Antarctic vegetation (mosses, lichens, and algae) and that exhibit similar spectral properties. This limitation restricts the ability to distinguish between individual species and even different vegetation types. Either higher spectral resolution or more sophisticated spectral analysis techniques are needed when working with multispectral or hyperspectral imagery. Current satellite resolution constraints also play a key role, as different platforms offer varying detection capabilities, from MODIS at 250 m resolution to Landsat at 30 m, Sentinel-2 at 10 m, and WorldView-3 at 1.24 m. These resolution differences mean that small vegetation patches remain undetectable at coarser resolutions and can result in misclassifications (Darabi et al. 2025). Environmental factors, such as cloud and snow cover, introduce temporal variability in observable land area, which can create trends unrelated to actual vegetation growth patterns. Similarly, inconsistent temporal data that arises from historical gaps in observations and differing observation baselines in satellite imagery can lead to misinterpretation of vegetation trends. Historical on-ground vegetation data, alongside imagery with the same spatial coverage and from the same time period, are needed to hindcast vegetation change. This complement of data is rare for the Antarctic. While high-resolution data from platforms like WorldView, QuickBird, and RapidEye provide finer detail and greater accuracy, their high cost and limited coverage in Antarctica present challenges for long-term and broad-scale vegetation monitoring applications (Mohammadimanesh et al. 2024). Therefore, the effective use of spectral signatures to estimate vegetation cover, condition, and community membership in the Antarctic presents a combination of biological and technical challenges, most notably (i) the potentially slow growth rates combined with the fine scale of the response for some plants, such as lichens, and (ii) the patchiness and lack of suitable imagery alongside evidence on the ground, from which near-recent and historical baselines can be established. A non-uniform and highly variable response to the changing climate by plant species and communities, at multiple spatial scales and between locations, is the most likely expectation given the already high variation in the distribution of Antarctic vegetation. These factors mean that realising the potential for monitoring vegetation change at scale in the Antarctic is some way off. Nonetheless, the sooner investment in research in this area ramps up, the sooner the necessary Antarctic-specific methodological advances will be made and current baselines can be established. Despite these constraints, as the research field matures alongside ongoing technological solutions, the integration of research approaches across ecology and remote sensing is emerging as a way forward. There is a growing trend toward multi-scale, multi-platform integrated approaches for studies in the Antarctic and elsewhere that leverage the strengths of each data source: ground surveys for detailed validation and calibration, UAVs (drones) surveys for high-resolution localised monitoring, and satellite imagery for broad-scale, repeated coverage (e.g., Sandino et al. 2023; Tovar-Sánchez et al. 2021). The integration of field-air-space observations into ecological research and monitoring designs from the outset, to address well-defined questions about Antarctic vegetation change, is crucial for accurate vegetation community assessment and monitoring. Such integration will allow researchers to bridge spatial and temporal scales effectively while maintaining accuracy and operational efficiency and addressing the suite of confounding factors discussed by Colesie et al. (2025). In summary, the most rapid and robust advances in understanding how vegetation is changing in the Antarctic will come from integrated research that uses remote sensing as a tool to upscale local ecological observations and mechanistic understanding of the life history, physiology, and morphology of the region's vascular and non-vascular flora. This requires interdisciplinary collaboration and the co-creation of research programs to address this critical question for the Antarctic region. Given the pace of change and the importance of scientists and decision-makers working together to manage the natural environment effectively, interdisciplinary scientific collaboration is critical for delivering consensus-based information for management and to inform policy. Melodie A. McGeoch: conceptualisation, writing – review and editing. Felipe Gonzalez: conceptualisation, writing – review and editing. This work was supported by the Australian Research Council SRIEAS Grant SR200100005, Securing Antarctica's Environmental Future. The authors are members of a comparatively small, extensively networked, and in some cases co-funded collaborative community of Antarctic scientists, including several of the authors of the Invited Perspective and related publications. We declare no material conflicts of interest. This article is a Invited Commentary on Colesie et al., https://doi.org/10.1111/gcb.70294. The authors have nothing to report.