The potential of microbial Volatile Organic Compounds as a microbial proxy for metabolic functioning in grassland soils

Microbial functioning is central to soil health, yet robust and integrative proxies for monitoring it under field conditions remain limited. Microbial volatile organic compounds (mVOCs) have recently emerged as a promising non-destructive fingerprint of soil microbial activity, but their capacity to evaluate consequences of land use for soil functioning has not been evaluated yet. Here, we compare the performance of mVOC profiles with that of extracellular enzyme activities (EEA), which are used as a proxy for soil microbial functioning. We compared both approaches across 18 grassland fields spanning a gradient of land-use intensity from conventional, extensive, to semi-natural). We assessed (i) whether the unsupervised multivariate structure of each proxy aligns with post-priori land-use types, (ii) if the two proxies covary across sites, and (iii) how their supervised classification performance compares. Results showed that in unsupervised ordinations neither proxy produced clear separation among land-use types, and that the multivariate structures of the two proxies did not correspond. Therefore the results suggested that mVOCs and enzymes capture distinct biological signals and that soil functioning does not map neatly onto discrete management labels. Despite this, canonical correlation analysis revealed a strong shared gradient between mVOC and enzyme profiles (r = 0.82), demonstrating meaningful multivariate coupling between the two proxies. Unexpectedly, mVOC profiles outperformed enzyme activities in supervised classification of land-use types (balanced accuracy 0.95 vs. 0.81), with dimethyl sulfide and methyl acetate emerging as the most discriminating compounds for conventional land use. These results suggest that mVOCs capture emergent metabolic properties of microbial communities not fully represented by enzyme assays, and that the two proxies may be considered complementary rather than competing indicators. We discuss implications for multi-proxy soil monitoring and the broader challenge of establishing reference values that provide a benchmark for soil functioning.