Peatlands contain approximately one third of all soil organic carbon (SOC). Warming can alter above- and belowground linkages that regulate soil organic carbon dynamics and C-balance in peatlands. Here we examine the multiyear impact of in situ experimental warming on the microbial food web, vegetation, and their feedbacks with soil chemistry. We provide evidence of both positive and negative impacts of warming on specific microbial functional groups, leading to destabilization of the microbial food web. We observed a strong reduction (70%) in the biomass of top-predators (testate amoebae) in warmed plots. Such a loss caused a shortening of microbial food chains, which in turn stimulated microbial activity, leading to slight increases in levels of nutrients and labile C in water. We further show that warming altered the regulatory role of Sphagnum-polyphenols on microbial community structure with a potential inhibition of top predators. In addition, warming caused a decrease in Sphagnum cover and an increase in vascular plant cover. Using structural equation modelling, we show that changes in the microbial food web affected the relationships between plants, soil water chemistry, and microbial communities. These results suggest that warming will destabilize C and nutrient recycling of peatlands via changes in above- and belowground linkages, and therefore, the microbial food web associated with mosses will feedback positively to global warming by destabilizing the carbon cycle. This study confirms that microbial food webs thus constitute a key element in the functioning of peatland ecosystems. Their study can help understand how mosses, as ecosystem engineers, tightly regulate biogeochemical cycling and climate feedback in peatlands.

International audienceGlycerol dialkyl glycerol tetraethers (GDGTs) are membrane-spanning lipids from Bacteria and Archaea that are ubiquitous in a range of natural archives and especially abundant in peat. Previous work demonstrated that the distribution of bacterial branched GDGTs (brGDGTs) in mineral soils is correlated to environmental factors such as mean annual air temperature (MAAT) and soil pH. However, the influence of these parameters on brGDGT distributions in peat is largely unknown. Here we investigate the distribution of brGDGTs in 470 samples from 96 peatlands around the world with a broad mean annual air temperature (−8 to 27 °C) and pH (3–8) range and present the first peat-specific brGDGT-based temperature and pH calibrations. Our results demonstrate that the degree of cyclisation of brGDGTs in peat is positively correlated with pH, pH = 2.49 x CBTpeat + 8.07 (n = 51, R2 = 0.58, RMSE = 0.8) and the degree of methylation of brGDGTs is positively correlated with MAAT, MAATpeat (°C) = 52.18 x MBT5me’ – 23.05 (n = 96, R2 = 0.76, RMSE = 4.7 °C). These peat-specific calibrations are distinct from the available mineral soil calibrations. In light of the error in the temperature calibration (∼ 4.7 °C), we urge caution in any application to reconstruct late Holocene climate variability, where the climatic signals are relatively small, and the duration of excursions could be brief. Instead, these proxies are well-suited to reconstruct large amplitude, longer-term shifts in climate such as deglacial transitions. Indeed, when applied to a peat deposit spanning the late glacial period (∼15.2 kyr), we demonstrate that MAATpeat yields absolute temperatures and relative temperature changes that are consistent with those from other proxies. In addition, the application of MAATpeat to fossil peat (i.e. lignites) has the potential to reconstruct terrestrial climate during the Cenozoic. We conclude that there is clear potential to use brGDGTs in peats and lignites to reconstruct past terrestrial climate

No abstract