| 30 Sep 2026
Measurement report: Moisture and light driven Volatile Organic Compound emissions of biocrusts from the Succulent Karoo, South Africa
Abstract. Biological soil crusts (biocrusts) cover around 12 % of Earth’s land surface and play critical roles in dryland ecosystems, yet their contribution to volatile organic compound (VOC) emissions remains poorly understood. Despite their extensive distribution, biocrusts are not currently included in global VOC emission inventories, possibly creating a significant gap in the understanding of atmospheric chemistry in arid regions. Here, we quantify VOC emissions and CO2 fluxes from soil samples dominated by four different biocrust types, namely light cyanobacteria-, dark cyanobacteria-, chlorolichen-, and moss-dominated crusts and bare soil samples, collected from the Succulent Karoo, South Africa. Using a climate-controlled chamber, we measured biocrusts at high air humidity, during and after wetting with artificial rainwater, and during subsequent drying. Interestingly, VOC emissions were strongest in the absence of photosynthesis before wetting, when samples were exposed to humid air in the dark and in the light with strong emissions of ethanol, methanol, and acetone. After wetting most VOC emissions decreased, but some of them, like geosmin and 2-methylisoborneol, compounds associated with petrichor, increased. In general, moss-dominated biocrusts exhibited the highest isoprene and sulfur compound emissions. Dimethyl sulfide emission peaked during darkness, suggesting a light-dependent source or sink in moss dominated biocrusts. The emission of up to 75±48 nmol m–2 s–1 of total VOCs across the studied biocrust types highlights them as a significant, overlooked source of reactive VOCs to the atmosphere, potentially influencing atmospheric oxidation capacity and secondary organic aerosol formation, particularly in regions with low vascular vegetation cover.
In their manuscript “Measurement report: Moisture and light driven Volatile Organic Compound emissions of biocrusts from the Succulent Karoo, South Africa”, the authors investigated VOC emissions and CO2 exchange from four different biological soil crust types and bare soil collected in South Africa. Using a controlled laboratory chamber coupled to PTR, GC, and gas analyzers, the authors characterized the responses of VOC emissions to humidity, light, and wetting. The topic is of general interest to the readers of Atmospheric Chemistry and Physics and novel because biocrusts are largely absent from existing BVOC emission inventories.
However, several aspects of the experimental design and data analysis are currently insufficient to fully support the conclusions. In particular, 1) the sequential experimental treatments could introduce confounding effects, 2) the high emissions of low-molecular-weight VOCs require more validation, 3) the statistical analysis may not adequately account for the limited number of biological samples, and 4) the ecosystem-scale extrapolation is highly uncertain given the limited representativeness of the laboratory measurements. Although the study has potential value as a Measurement Report, substantial revisions are required before I could recommend publication.
Major issues:
1. In Section 2.2, the experimental design does not adequately separate the effects of humidity, light, and wetting. The authors exposed the samples sequentially to dry air in darkness (1 hr), humid air in darkness (10 hr), humid air under light (12 hr), and finally liquid water under light (12 hr). Consequently, changes in VOC emissions between phases may also reflect elapsed time, adaptation, or depletion of stored VOCs rather than the individual treatment variables. The additional light–dark cycling experiment with moss samples does not resolve these uncertainties for all biocrust types. In fact, Table A1 shows considerable variability in initial sample wetness (and even for repeated measurements #4 and #7? Also regarding Table A1, I did not see a lichen_02, and it appears to me there are 14 distinct physical samples, not 15 claimed in Section 2.4? And I'm mostly concerned for the replicates #4 and #7. From the table, the dry weight of dark_cyano_03 differs by more than 75% (14.7 g vs. 25.3 g). How is that explainable if the two are genuinely the same sample?). The authors should better distinguish the treatment effects from time-dependent changes and previous sample history, ideally supported by appropriate control experiments (e.g. re-order/mix-match your four treatment sequences within 24 hrs to the same sample to isolate/verify each individual effect) with additional analyses. Otherwise, the attribution of VOC emission changes to specific environmental drivers remains inconclusive.
2. In Section 2.4 and Appendix D (Tables D1–D2), the statistical analysis needs justification. Only three independent physical samples were collected for each biocrust type, while some samples were measured repeatedly during different experimental runs. Although the authors applied linear mixed-effects models and aggregated the data for PERMANOVA, treating repeated measurements of the same sample as independent observations may introduce pseudoreplication in the permutation-based analysis. The reported statistical significance should be reassessed using an approach that preserves the dependence between repeated measurements of the same sample by definition. The authors should also clarify the interpretation of coefficients from the log1p-transformed models and their conversion to percentage changes or differences in the original flux units.
3. In Sections 2.5 and 3.5, the ecosystem-scale upscaling is too far stretched. The authors extrapolated measurements from only three replicates of each biocrust type collected from one geographical location to regional and global VOC emissions, estimating up to 0.61 Tg/yr globally. However, the laboratory measurements were conducted at a fixed temperature and under artificial humidity and wetting conditions, while actual emission rates are influenced by far more missing constraints such as regional climate, soil properties, microbiological community, and precipitation frequency, just to name a few. To pick one of the most simple VOC, methane, as an example, the global budget/emission inventory could only be modeled with long-term, in situ CH4 flux measurements across the globe, not from a few laboratory chamber experiments. Furthermore, the upscaling uses post-wetting emissions, whereas the highest VOC emissions were observed before wetting under humid air conditions. Even if the authors insisted to perform a "global estimation", they still need to justify the representativeness of the selected fluxes and provide sensitivity ranges that account for uncertainties in wetness duration, emission rates, and biocrust coverage. For the coherency and scientific robustness of the manuscript, I personally would suggest removing the ecosystem upscaling section from the Measurement Report. If not removed, with the current available lab data, the global estimate should be specified as a highly simplified illustrative scenario rather than a scientific assessment of global biocrust VOC emissions.
4. In Sections 3.2 and 4.3, the high emissions of ethanol and other low-molecular-weight VOCs require validation. Ethanol was reported as the dominant emitted compound, with a mean flux of 190 ± 121 nmol m-2 s-1 before wetting, even though little CO2 exchange was observed during that period. The authors proposed microbial fermentation mechanisms, but the experimental evidence does not distinguish biological production from the release of previously accumulated compounds or other sample-handling effects. The samples were collected, dried, frozen, and reconditioned before measurement while ethanol was used during chamber cleaning with background measurements and heating procedures were performed. Conventional alcoholic fermentation should co-emit CO2. Methane fermentation should co-emit CH4 but under anaerobic condition which is unlikely. The authors should provide stronger evidence that these large ethanol signals were not influenced by the experimental/human artifacts. Since low MW compounds dominate the total VOC flux, uncertainties and explanations in their origin affect the main findings of this study.
If I were the lead investigator, I might also scrutinize the Vocus housekeeping files on potential cal gas valve leak (unlikely, but the small VOCs are all in your cal tank, which is Tee-ed at the Vocus inlet per Figure A1) and humidity-dependent reagent-ion chemistry or imperfect response-factor corrections which introduce apparent changes in VOC signals. If available to provide, your Vocus raw ethanol and other calibrant VOC time series from the empty chamber, before background subtraction, alongside the sample signals could be a very useful diagnostic tool. You could also consider normalizing your VOCs to less biologically plausible markers to isolate the tubing/instrument/contamination artifacts. But these are all personal suggestions that the authors could take or ignore.
Minor issues:
1. In Section 3.2 (lines 268–271), the authors stated that measurement phase explained a substantially larger proportion of VOC compositional variance than sample type but the reported PERMANOVA R2 values are only 0.503 and 0.518?
2. In Section 4.2 (lines 485–490), the reported mass-based VOC flux for lichen-dominated biocrusts is 452 ± 400 µg m-2 h-1 both before and after wetting, despite substantially different molar fluxes. This appears to be a typo and should be verified?
3. In Section 2.2 (line 134), the tubing diameter is given as “1/4 inch (0.625 nm outer diameter)”. The nm unit appears to be a typo
4. Figure 2, caption. The HR m/z of MEK should be 73.065 (as mentioned in Line 598) not 73.0242.