Sun-induced fluorescence spectrum as a tool for assessing peatland vegetation productivity in the framework of warming and reduced precipitation experiment

Abstract

Northern peatlands store a large amount of carbon in the form of partially decomposed organic matter. Because the majority of northern peatlands are located in remote areas, remote sensing serves as a suitable alternative to traditional surveys, enabling to enhance our understanding of peatland vegetation. Among various optical remote sensing signals, sun-induced fluorescence (SIF) is the most directly connected to carbon assimilation by plants, making it a promising early-response indicator for assessing the impact of climate change on natural ecosystems. However, the behavior of SIF throughout the season and the strength of the relationship between SIF and carbon assimilation for peatland vegetation, which consists of peat mosses and vascular plants with diverse anatomy, morphology, physiology, and phenology, have not yet been studied. Therefore, we conducted the first comprehensive assessment of the full spectrum SIF, reflectance, and gross primary production (GPP) for two distinct peatland vegetation communities under control (C), warming (W), and warming with reduced precipitation (WP) conditions throughout an entire season. While we could detect clear differences in SIF and reflectance between the two vegetation communities for original, C, vegetation during the main growing season, these differences diminished when W and WP were applied. The W and WP conditions caused a more pronounced change in plant biomass for vegetation characterized by a higher proportion of peat mosses and low creeping shrubs, which resulted in significant changes in the SIF and reflectance spectrum. Our findings demonstrate that the domination of peatland by vascular plants that is expected due to future warmer conditions causes stronger seasonal variation of SIF, reflectance, and GPP. We observed that far-red SIF and the spectrally integrated full SIF spectrum strongly correlate (r2 > 0.85) with GPP regardless of vegetation community, temperature, and precipitation regime. However, the use of a novel multiple wavelength regression model using ten bands from the full SIF spectrum allowed for higher accuracy in the estimation of GPP compared to the use of single bands or integrated total SIF value. Moreover, such a model has a more stable performance when transferring from one vegetation community to another. Conversely, the correlation strength of traditional vegetation indices like the Normalized Difference Vegetation Index or MERIS Terrestrial Chlorophyll Index depends on the peatland vegetation community. While the SIF:GPP relationship exhibits similar r2 values for vegetation communities with different ratios of planophile and electrophile leaves, the slope of the linear model depends on this ratio. This study performed at the ground level shows for the first time the importance of full spectrum SIF for the monitoring of a heterogeneous ecosystem like peatland, which will help to better utilize the SIF products obtained through current and future satellite missions.