Integrating multi-source geospatial data to monitor peatland red-listed plant species, greenhouse gas dynamics, and the water table

Abstract

Peatlands are important ecosystems that support biodiversity, regulate greenhouse gases (GHGs) and flooding, and help to purify water. In Finland, however, more than 50 % of peatlands have been drained to support forestry, agriculture, and peat extraction. These activities have reduced peatland biodiversity, increased GHG emissions, and altered peatland hydrology, resulting in widespread peatland ecosystem degradation. Consequently, large-scale restoration efforts have been initiated to mitigate these impacts and restore peatland functioning. As restoration progresses, regular monitoring of peatland biodiversity, GHG fluxes, and the water table (WT) is needed to track peatland conditions over time.

One approach to monitoring peatlands is field-based observation. However, this method is constrained by high costs, limited accessibility, and restricted spatial coverage. To address these limitations, this research explores the use of geospatial environmental and remote sensing (RS) data, combined with statistical modelling, as an alternative means to support peatland monitoring. First, environmental data were used to analyse whether different restoration and climate change scenarios affect the national-scale distribution of potential habitats for red-listed peatland plant species in Finland. Second, environmental data and RS data (Sentinel-1 and Sentinel-2 imagery) were compared and combined to detect national-scale patterns of potential peatland GHG sinks and sources and to evaluate which data sources provided better predictive performance. Third, environmental data were used to improve RS-based peatland WT models in three study areas ranging from 18 to 175 ha.

Environmental data were able to predict the potential habitats of peatland red-listed plant species at the national scale. Drainage and climate variables were the most important predictors. Model projections indicated that climate warming will reduce potential suitable habitats for many red-listed species, especially in the northern and middle boreal zones. However, restoration increased the amount of potential suitable habitat, reduced habitat loss, and moderated future northward shifts in species distributions. The benefits of restoration were strongest under mild climate scenarios and gradually weakened towards the end of the twenty-first century under severe warming. Although the models performed well, future studies concerning the interaction between restoration and climate change should include ecological constraints such as dispersal and species interactions to improve the realism of long-term predictions.

In detecting potential GHG sinks and sources, models combining both environmental and RS data performed best, while RS-only models had the lowest accuracy. Maps generated from environmental data alone and those from the combined dataset showed similar patterns, whereas RS-only maps displayed some differences in spatial extent. These results suggest that RS data alone are not sufficient for reliable national-scale GHG predictions and that integrating environmental and RS data is necessary to obtain accurate and realistic estimates of potential peatland GHG sink–source distributions.

At the local scale, combining RS and environmental data improved peatland WT models, both spatially and across seasons. In patterned northern boreal aapa mires, the topographic wetness index and topographic position index were important for explaining spatial WT variation. In a southern boreal drained peatland forest site, canopy height played a larger role in WT dynamics, especially across different management treatments. Uncrewed aerial vehicle data offered very high-resolution information that captured microtopography and vegetation patterns, while Sentinel-2 data allowed repeated seasonal monitoring. These results demonstrate that multi-source RS, combined with environmental variables, can support efficient peatland WT monitoring.