As in classical microplate assays (Marx et al., 2001) and in soil zymography (Bilyera and Kuzyakov, 2024), the determination of EEA in SEAR is based on the reaction of soil enzymes with artificial fluorogenic substrates and measuring the increase in product fluorescence over time. We currently use substrates for enzymes that release either 7-Amino-4-methylcoumarin (AMC) or 4-Methylumbelliferone (MUF).

The SEAR setup consists of a reaction plate and a reader, as well as a soil tray and a 4 mm sieve (Figure 1). The reaction plate is composed of a passive carrier containing wells for substrates (reaction wells) and calibration substances (reaction products in different concentrations, calibration wells). Both substrates and calibration substances are dissolved in a transparent gel that consists of 99.0% water (Figure 1b). The current standard reaction plate contains reaction wells with substrates (3 replicates per substrate-type) as well as calibration wells with the reaction products MUF and AMC, each in 5 concentrations as basis for the respective calibration curves. The substrate wells target five different groups of hydrolytic enzymes involved in the carbon (C), nitrogen (N), and P cycles (Supplementary Table 1). Sealed reaction plates have a limited storage time as biotic or abiotic hydrolysis may decompose substrate over longer time periods. Therefore, storage at 4 °C is recommended.

For the measurement, sieved soil is filled into the soil tray, the foil on one side of the reaction plate is removed, and the open side of the reaction plate is pressed onto the soil (see Supplementary Material 1 for a detailed user protocol). Coarse homogenization using the 4 mm sieve included with the setup, should work with most field-fresh soils. However, if soil consistency allows, finer sieving, e.g., to 2 mm, may likely decrease variability among replicate measurements. The tray with soil and reaction plate is inserted into the reader, and over a period of 40 min the reader will register the change in fluorescence in reaction and calibration wells. Every 2 min, a picture is taken of the reaction plate from the opposite side of the soil following illumination with LED diodes matching the reaction product excitation and emission spectra. Temperature and time are recorded during the measurement period. This results in a set of images (= raw data) that are then analyzed in the data analysis pipeline to derive EEA rates.

The software included with the SEAR device derivates EEA rates from the generated images using the following steps (see also Supplementary Figure 1):

In the first step, in each image, all pixels for a given well are extracted. A single luminosity value for each well is computed from the 8-bit grayscale image by using the 90th percentile of all pixel values. This accounts for heterogeneity within the well that can be caused by non-perfect contact, border effects, or single spots that are outliers (dirt on the foil, single overexcited pixels). However, in most cases, the distribution of luminosity values is quite homogeneous (see example of reaction plate shown in Supplementary Figure 1, top left).

Next, the observed signal in the reaction wells is corrected based on the signal drop during the measurement period in the calibration wells, i.e., the respective AMC and MUF values, based on a fitted drop function. Depending on soil properties and moisture, this drop may be up to 25% of the initial signal (data not shown). This correction serves to account for the net effect of losses, such as transport and quenching processes on the fluorescence produced by the enzymatic reactions, whichis further discussed in Section 4.1. Third, using the calibration function, which is linear up to 30 μmol L^-1 for both MUF and AMC (see Supplementary Figure 1, top right), the luminosity values for each well in each image are converted to product concentrations (in mol L^-1). Luminosity values above the linear range of the calibration function are not considered by the data analysis pipeline. Soil background fluorescence as measured in the zero calibration well, has been found to be maximum 5 luminosity values.

Last, for each reaction well, the increase of corrected product concentration with time is used to compute the respective EEA rate. The concentration data used for this are taken only from the range of linear increase with time which is constrained as follows. Based on 20 recorded data points, an automated, proprietary code determines a range of at least 10 consecutive data points with a linearity requirement based on R². If the quality requirements are not met, the measurement is labeled invalid. An example of product concentration increase with time is shown in Supplementary Figure 1 (bottom middle). The obtained rate is multiplied by a conversion factor of 5100, to convert from μ M s^-1 to pmol min^-1, resulting in the number of product molecules released in each well. The final EEA is the average of the analytical replicates. The rate can also be converted to per unit time and unit area (as used often in zymography) by using the conversion factor of 115 to get from μ M s^-1 to pmol mm^-2 min^-1, i.e., dividing 5′100 by the well surface of 44.2 mm². In addition to correcting product concentrations in the reaction wells, as described above, the data from the calibration wells are used in the following way: The range of the product concentrations in the calibration wells defines the valid detection and therefore measurement range, in particular assuring to be in the linear range and avoiding oversaturation. Comparison with default values, e.g., for a blank in form of pure agarose, serves to check the overall functioning of the hardware (e.g., whether LEDs work properly).

Statistical analysis was conducted using R version 4.3.1 (2023–06-16). To determine if there were significant differences in EEA among the soil moisture levels, a one-way analysis of variance (ANOVA) followed by Tukey’s Honest Significant Differences (HSD) test was performed. ANOVA assessed the overall variability in EEA attributed to different moisture levels, while Tukey’s HSD test enabled pairwise comparisons between treatment levels to identify statistically significant differences. A significance level of ɑ = 0.05 was chosen to determine statistical significance.

The substrate saturation curves resulted for all four (GLS) and two (PHO) test samples in a similar pattern: with increasing substrate concentration, EEA showed an increase and then reached a plateau at 200–250 μmol L^-1 (Figure 3). For all GLS amended samples and PHO amended soil 1 this was followed by a decrease in EEA for the highest substrate concentrations. Based on these results, a substrate concentration of 250 μmol L^-1 was selected for all further measurements.

Enzyme specific LODs were determined as average soil specific LODs that were calculated as described in Section 2.3.3., exemplarily shown for Leucine-Aminopeptidase (LEU) in Figure 4 (5 technical replicates per sand/soil). Values for four enzymes ranged between 0.6 and 2.1 pmol min^-1 (Table 2). Because of the particularly high variability of the response for phosphatase (PHO) activity at values close to 0 no value is given. More research is needed to improve the instrument performance for measuring low phosphatase activities.The repeatability in terms of coefficient of variations (CV) of soil measurements with SEAR varied between 5% and 50%, except for 94% for a β-Xylosidase (XYL) value close to the detection limit (Table 2; Supplementary Table 4). Generally XYL activities were low compared to other enzymes, which likely was the reason for the often relatively high CVs.

For all four tested model enzymes, EEA increased linearly with increasing enzyme concentrations in the solution the sand was spiked with (Figure 5). The R² was above 0.9, except for LEU.

The observed patterns of EEA as response to five levels of soil moisture appear to depend more on the soil than on the type of enzyme (Figure 6; Supplementary Figure 2; Supplementary Table 5). Overall, the influence of moisture was small, and in the two sands and most soils no clear increasing or decreasing trend in EEA with increasing soil moisture could be established. Exceptions were soils 5 and 7 with increasing and decreasing moisture effects, respectively. Of the 400 tested pairs, 20 resulted in significant (p < 0.05) differences according to Tukey’s HSD test, while all other pairs and all other soils were not significantly different from each other (see Supplementary Table 6).

Effects of temperature during the measurement were assessed using the topsoil and the subsoil of a temperate beech forest site (Soils 5 and 6, Table 1). Within the tested range from 6 °C to 55 °C, the rates of all tested enzymes showed an exponential increase, except for EEA of GLS which reached a plateau at around 50 °C (Figure 7). The Q₁₀ values, calculated for the range from 15 °C to 50 °C ranged between 1.4 and 1.8.

While values per unit time and unit area obtained with the SEAR device cannot be directly compared with values per unit time and unit mass obtained with classical slurry-based enzymatic assays (see discussion Section 4.2), we include a relative comparison of EEAs obtained with both methods from experimental studies of research partners that were kindly provided to us (Supplemantary Data 9). Overall, SEAR data for 5 different enzymes from a model ecosystem experiment investigating drought and warming effects on various tree species, and for GLS from a field experiment in pastures comparing different management practices correlated significantly with respective data obtained by microplate assays (Supplementary Figure 3). However, the data from the model ecosystem experiment revealed that there were differences among the enzymes, e.g., a particularly strong correlation for GLS data and a much weaker one for LEU data.

The method is based on the reaction of enzymes in a reactive layer of homogenized soil with fluorogenic substrates upon contact with an agarose gel in which the substrates are dissolved. The reaction products are detected with a camera on the opposite side of the transparent gel by their fluorescence upon excitation with a set of diodes, employing a time series of images taken by a camera sensor. As such, the method incorporates the latest developments in soil zymography, specifically camera detection (Spohn and Kuzyakov, 2014) and time-lapse operation (Guber et al., 2021). However, using agarose as substrate medium corresponds to earlier versions of soil zymography (e.g., Baldrian and Větrovský, 2012), while substrate-impregnated membranes are preferred in current soil zymography methodology because of their better lateral resolution needed for mapping the small-scale heterogeneity of the EEA, e.g., around root systems (Bilyera and Kuzyakov, 2024). For our purposes, the gel is a better choice because it allows for a fast contact of soil with substrate in a well-defined concentration. Based on observations, we assume a fast uptake of water from the gel by the soil, the extent of which depends on the inherent moisture of the soil and its properties, most importantly texture. As a consequence, substrate is rapidly transported from the gel into the reactive soil layer, also in very dry soils.

In most cases, the pixel-wise data analysis indicated a quite homogeneous spatial distribution of the detected fluorescence, and thus, despite a limited degree of soil homogenization achieved by coarse sieving, a large part of the soil surface can be assumed to participate in the enzymatic reaction. However, in a given soil, homogeneity of EEA is a function of soil aggregation as well as the spatial distribution of enzyme hotspots, and thus the degree of homogeneity varies depending on the soil.

We define the reactive soil layer as the layer where reaction of soil enzymes with substrate leads to product fluorescence that reaches the detector within the measuring period. Based on theoretical considerations, the thickness of this layer is affected by various factors including convective and diffusive transport of substrate from the gel into the soil, the diffusive transport of reaction product from the location of formation towards the gel and to deeper soil layers, the convective transport of reaction product to deeper soil layers, the penetration depth of exciting light into the soil, and quenching or absorbance of product fluorescence in the soil or the gel. Soil background fluorescence as another factor affecting the measured luminescence, was found to be negligible in all our case studies to date, indicating low to no impact of autofluorescence within the emission wavelength interval.

In contrast to the importance of transport processes affecting the distribution of substrate and the reaction product, Guber et al. (2018), Guber et al. (2021) showed that movement of enzymes within the soil is negligible. In these studies, the maximum thickness of the reactive layer for direct soil zymography with membranes as substrate carriers was estimated to be about 200 µm. Since in the latter method the convective substrate transport is negligible and fluorescence detection is limited to the membrane itself, this result cannot be transferred unrestrictedly to our method. In our data-analysis pipeline, we account for the net effect of all afore-mentioned processes on product detection by correcting for the observed disappearance of product fluorescence from the reactive layer in the calibration wells, assuming that the freshly generated products in the reaction wells disappear with the same rate from the reactive layer as the products in the calibration wells. Although diffusive flux of product from the soil to the gel is not considered, applying this automated correction procedure improved the linearity of the response in our positive control experiment, and generally decreased the variability of the technical replicates.

In contrast to the need to correct for decreasing product fluorescence during the measuring period, no such correction should be necessary for transport of substrate to deeper soil layers. Since we work under conditions of substrate saturation, as is further discussed below, we can assume that disappearance of substrate from the reactive layer should have a negligible effect on the measured enzymatic rates.

Table 3 presents an overview of several factors that were defined crucial for any new method aiming at assessing EEA in soils (Dick W. A., 2011), and how this is implemented in the SEAR system.

An overall goal that a soil enzymological method needs to achieve is to be able to measure a reaction rate that is proportional to the concentration of enzymes in the soil (Dick W. A., 2011). The linear increase of EEA in our experiments with sand, spiked with solutions of increasing concentrations of four different enzymes, confirms that our method in the presented set-up achieves this basic goal. However, the thickness of the reactive layer for a given soil, which can be considered the equivalent of the extraction efficiency in slurry-based methods, is unknown (see Section 4.1), and thus activities per unit mass or volume of soil cannot be derived from the measured rates. Therefore, the data obtained with our method cannot be directly compared to data obtained with other soil enzymatic methods, in particular classical slurry-based assays. However, a first relative comparison of EEA obtained with SEAR and slurry-based microplate assays, based on experimental data provided by research partners, indicates that treatment effects can indeed be captured by our method in a similar way as by classical assays. However, these data also show that the strength of correlations between EEA obtained by the two methods can vary depending on the type of enzyme and the particular study. This may be attributed to inherent operational differences between the two methods. In particular, this concerns the slurry formation in an excess of buffer or water, often followed by ultrasonic dispersion, in the classical assays as compared to the gentle sample preparation by sieving employed in our method. Thus, our method on the one hand may not assess enzymes that are well protected in soil aggregates, on the other hand may avoid operational bias imposed by the slurry formation such as Birch-type re-wetting effects of dry soils (Birch, 1958; Fierer and Schimel, 2003), e.g., from drought experiments (see also introduction). A more in-depth discussion of results obtained by our method with classical slurry-based enzyme assays is beyond the scope of this manuscript but may be possible as soon as more such comparisons become available.

When related to unit surface area, the rates obtained with our method are up to one order of magnitude higher than those measured with membrane-based soil zymography (e.g., Meller et al., 2020; Spohn and Kuzyakov, 2014), which may be explained by the faster contact between substrate and soil enzymes when using a gel rather than a membrane due to the additional convective transport of substrate to the soil or generally, due to a thicker gel layer.We found our method to operate well for a wide range of different soils, comprising textures from sand to silty clay loam (USDA, 2017), acid forest soils with pH < 4 to carbonate containing agricultural soils, and soil organic carbon contents up to 18%. However, so far, we have limited experience with soils that are low in soil organic carbon (SOC <1%), and thus likely also exhibit low EEA. Furthermore, by successfully testing soil moistures from 2% to 173% of the respective water holding capacity and temperatures from 6 °C to 55 °C, we could show that measurements under most conditions encountered in the field are possible. Thus, there is no need for storage and related sample pretreatments such as drying that may affect the measured rates (Burns et al., 2013; Lorenz and Dick, 2011; and references therein). In all tested cases, we found sieving at 4 mm as a minimum soil pretreatment to be suitable for homogenizing the soil samples. This mesh size is within the range recommended by Lorenz and Dick (2011), although they argue for a 2 mm mesh size to be optimum for most soils and advise larger mesh sizes only for highly organic and clayey soils.

Overall, the effects of short-term changes in soil water content on the measured activity of five different enzymes were small, which can be explained by the equalizing wetting effect upon contact with the gel. The drier the soil the more water is lost from the gel to the soil, leading to a homogenization of moisture conditions among different samples. The weak patterns of EEA vs. soil moisture, observed in some cases, appear to be both soil and enzyme dependent, which is similar to observations by Gomez et al. (2020), Gomez et al. (2021). The small extent of the effects suggests that our method is able to capture climatic or longer-term experimental effects on EEA irrespective of the current soil moisture during the measurement.

We could show that our device is operational almost up to the temperature limit of 60 °C above which most enzymes in soils are expected to denature (Tabatabai, 1994). Within the range of 15 °C–50 °C, the EEA of GLS, β-glucosaminidase (GLA), and XYL in both the topsoil, rich in organic matter, and the mineral subsoil of an acidic beech forest site exhibited an exponential increase, with Q₁₀ values of 2 or smaller, as is expected for catalyzed reactions in soil (Tabatabai, 1994). For GLS, a plateau at around 50 °C indicated that the optimum temperature for this enzyme was reached (Dick W. A., 2011). We conclude that with our device, EEA can be measured either at the current temperature, or if a thermostatic device is available, at any chosen standard temperature. In classical soil enzymatic assays, often standard temperatures of 25 °C or 37 °C have been chosen in order to allow for better comparison among different studies (Dick W. A., 2011; ISO/TS 22939, 2019). In order to extrapolate measured EEA with SEAR in the field to other temperatures, average Q10 values may be used.

The need to buffer soil enzyme assays at an enzyme’s optimum pH has been discussed controversially and depends on the research question or purpose of the analyses (German et al., 2011). Since SEAR is mainly intended for field assessments under inherent soil conditions, including pH, we chose a buffer-free gel for our standard reaction plate. However, should optimum comparability among different studies be required, the method can be easily adapted by producing reaction plates with buffered gel.

Based on GLS and PHO rates in spiked sand and soils with increasing substrate concentrations, we chose 250 µM as the concentration of substrates for all tested hydrolytic enzymes in our standard reaction frame. In soils, GLS and PHO rates have been often found to be larger than those of most other hydrolytic enzymes (Sinsabaugh et al., 2008; Sinsabaugh and Shah, 2012), and thus we assume that the chosen concentration ensures substrate-saturated conditions for the enzymes included in our standard reaction plate under a wide range of soil properties and conditions. The concentration of 250 µM is also in the same range as is found for substrate saturated conditions in slurry-based assays (e.g., German et al., 2011). Nevertheless, should particular soil properties or conditions require higher substrate concentrations, the method can be easily adapted, considering the good solubility of substrates in the gel, tested up to 500 µM (Figure 3).

The total incubation time of 40 min was sufficiently long to achieve a robust linear increase of product for all enzymes, soils and conditions tested. It is also short enough to minimize the potential effect of microbial growth or reactions on the number of detectable enzymes (Dick W. A., 2011), and thus, there is no need for adding an antiseptic reagent, as has been often done in classical slurry-based assays (Tabatabai, 1994). Considering the measurements of sterilized soils as negative controls, the potential contribution of abiotic hydrolysis of substrates to the measured signals cannot be larger than the LODs derived from these measurements.

Overall, the variability of replicate measurements of EEA was within the range found for fluorometric microplate assays (Creamer et al., 2009). The effect of inherent soil moisture on reproducibility was small, which might also be related to the equalizing wetting effect upon contact with the gel. The variability of EEA measurements for particularly inhomogeneous soils with SEAR can potentially be decreased by further increasing the number of replications and/or establishing a protocol for identifying and removing outliers.