Materials and methods

The materials for the study were soils sampled from the Carbon Action experiment (Mattila et al., 2022) intensive observation set (Mattila, 2020), where 20 farms test carbon sequestration (cover crops, compost, grazing practices, intercrop leys and subsoiling). The experiment started in 2019 and the samples were collected in July 2021. Each farm had a split field, where one side had carbon farming practices implemented and the other was held as a continued-normal-practice control. The samples were collected from 3 GPS located points on each field from a 10 m radius from the center of the point with a 16 mm soil corer from a depth of 0-17 cm. Each field had 30 cores collected, which were pooled, dried at room temperature (fan assisted) and gently sieved through a 5 mm sieve. As the soil sieving and milling can influence the results, all soil sample processing was done following established guidelines (Franzluebbers and Haney, 2018). (Due to an unfortunate laboratory accident, 5 samples were lost during processing, resulting in an overall sample amount of n=35.) The sampled soils covered a large range of soil texture and organic matter: the median clay content was 35% (4-63%) and the OM 6.8 % (2.6-15.5%). The farming systems covered annual cropping, grass in rotation and perennial pastures.
For CO2 burst analysis, the dry samples were rewetted to approximately 50 % pore space (i.e. 30 ml of soil and 9 ml of water; Woodsend lab manual). The sample was placed in a 475 ml container and sealed with a CO2 measurement cap fitted with a datalogger (Woodsend IRTH). The CO2 concentration was measured for 24 h and the increase in CO2 level over time was converted to mg CO2-C/kg3soil by multiplying with container air space, dividing by sample mass and converting to mass units using the ideal gas law.
For the redox analysis, soils were also rewetted to 50% pore space (Husson et al., 2016). The Redox was measured with an Extech RE300 Exstik Platinum oxidation-reduction-potential sensor (platinum electrode, silver/chloride reference electrode). The flat end ORP sensor was pressed to the moist soil sample and allowed to stabilize 1-3 minutes, until the ORP reading changed only slowly. Three repeated measurements were made of the same sample and the average value was used. The sample pH was measured with a Horiba LAqua Twin pH meter, using 1:1 ratio of distilled water:soil. The ORP reading was converted to Eh (mV) by adding the reference electrode voltage (200 mV) and pH corrected to a relative hydrogen score by the equation rH2 = Eh/30 + 2 pH. For interpretation, the readings were compared to suggested norms for ”healthy” soils (Husson, 2013).
For additional interpretation, the soil samples were classified according to organic matter content, soil clay content, soil structure (VESS (Ball and Munkholm, 2015)) and type of crop (perennial, annual). These were collected and published as ongoing monitoring in the Carbon Action experiment (Mattila and Girz, 2021).
The statistical analysis was done in R programming language (R Core team, 2022). Correlations between measured variables were calculated with Pearson’s correlation. Significance between the differences in two groups was tested with Mann-Whitney U-test (Wilcoxon). The effect size of carbon farming practices was tested by fitting a linear model to the data, using each farm as a blocking factor.