Ex situ soil respiration assessment using minimally disturbed microcosms and dried–sieved soils; comparison of methods to assess soil health

Abstract Soil respiration measurements are commonly used as soil health indicators. Several ex situ soil respiration methods exist, but comparative performances between them have rarely been analyzed. Specifically, there is a lack of comparisons between intact microcosms and destructive methods. The objective of this study was to analyze and compare three different ex situ soil respiration methodologies: minimally disturbed microcosms using fresh soil, dried–sieved 24 h burst test, and dried–sieved 10-day incubation. We hypothesized that (i) the respiration rates for the three methods are correlated to each other; (ii) the respiration rates are strongly correlated with soil physico-chemical parameters; (iii) disturbance caused by drying and sieving reduces regression coefficients compared with microcosms; and (iv) drying and sieving soil produces larger respiration rates. Soil was collected in the Province of New Brunswick, Canada. Total carbon and nitrogen (C:N), pH, aggregate stability, total dissolved C and N, NO3 and NH4, texture, and labile C were determined prior to incubations. Our results showed that the three methods had CO2 efflux in similar ranges. However, all the methods had low to no significant correlations between soil physico-chemical parameters and respiration. Total dissolved N had the strongest correlation with CO2 efflux. The results of the microcosm method significantly correlated with the results for 24 h burst test but not with the 10-day incubation method. We conclude that drying and sieving soil prior to performing ex situ soil heterotrophic respiration measurements using the 24 h burst tests can produce cautiously reliable results. Despite the disturbance, results from the 24 h burst tests are comparable with the results of the microcosm method.


Résumé
Il est courant de mesurer la respiration du sol pour en déterminer la vitalité. Plusieurs méthodes ex situ existent pour cela, mais on s'est rarement attardé à en comparer l'efficacité. Plus précisément, on n'a pas comparé les méthodes qui utilisent le microcosme intact à celles qui le détruisent. Les auteurs voulaient analyser et comparer trois méthodes employées pour mesurer la respiration du sol ex situ : l'usage de sol frais au microcosme le plus intact possible, l'essai d'éclatement de 24 heures après tamisage à sec et l'incubation de dix jours, également après tamisage à sec. Ils formulent les hypothèses suivantes : (1) le taux de respiration est corrélé entre les trois méthodes; (2) il présente une corrélation étroite avec les propriétés physicochimiques du sol; (3) comparativement à la méthode du microcosme, les perturbations que suscitent le séchage et le tamisage réduisent les coefficients de régression; (4) le taux de respiration est plus grand quand il y a séchage et tamisage du sol. Pour vérifier leurs hypothèses, les auteurs ont prélevé du sol dans la province du Nouveau-Brunswick, au Canada. Ils en ont mesuré la concentration totale de carbone et d'azote (C:N), le pH, la stabilité des agrégats, le C et le N total dissous, la teneur en NO 3 et NH 4 , la texture, et la concentration de C labile avant incubation. Les résultats indiquent que les dégagements de CO 2 sont du même ordre pour les trois méthodes. Néanmoins, les propriétés physicochimiques du sol ne présentent aucune corrélation importante ou significative avec le taux de respiration, peu importe la méthode. La concentration totale de N dissous est le paramètre qui présente la plus forte corrélation avec les émissions de CO 2 . Les résultats de la méthode du microcosme sont significativement corrélés avec ceux de l'essai d'éclatement de 24 heures, mais pas avec ceux de l'incubation de dix jours. Les auteurs en concluent que sécher et tamiser le sol avant d'en mesurer la respiration hétérotrophe ex situ peut donner des résultats fiables, sous certaines réserves. Malgré les perturbations qu'il entraîne, l'essai d'éclatement de 24 heures donne des résultats comparables à ceux de la méthode du microcosme. [Traduit par la Rédaction] Mots-clés : dégagements de CO 2 , respiration, microcosme, perturbation du sol, éclatement de CO 2 , analyse comparative Introduction Soil heterotrophic respiration (Rh; i.e., CO 2 efflux from soil not including root respiration) releases a significant amount of CO 2 into the atmosphere (Marland 2008). Accordingly, small changes in Rh can have a significant impact on the climate (Schurgers et al. 2018) . Therefore, accurate Rh assessments are important for quantifying flux of CO 2 from the soil to the atmosphere and for estimating the activity of soil microbial communities (Kuzyakov 2006). For these reasons, Rh is often used as an indicator of soil health when evaluating different land-uses or management practices (McGowen et al. 2018). For example, the Cornell Soil Health Laboratory soil respiration operating procedures uses standard scores from 0 to 100 to indicate the presence of active soil microbial communities (Moebius-Clune et al. 2017). However, Rh measurements performed in the field or laboratory often produce distinct results (Davidson et al. 1998;Comeau et al. 2018a). Due to soil heterogeneity across the landscape and difficulties in taking representative Rh measurements in different land forms, ex situ (in-lab incubations) Rh assessments are often preferred. In these ex situ incubations, to compare management practices and land-uses or to test specific hypotheses, excavated soil is usually pooled to generate representative samples (Gutinas et al. 2013;Zhou et al. 2014) and Rh measurements are performed under controlled moisture and temperature conditions (Bao et al. 2016;Yan et al. 2017).
Ex situ assessments of Rh are traditionally performed on dried, sieved, and homogenized (e.g., mixed or pooled) soils (Brinton and Vallotton 2019) to create replicates adequate for statistical analysis. However, this physical disturbance affects the original soil structure, which in turn has consequences on microbial activity and composition (Baveye et al. 2018), and as a result, it may not always give an accurate picture of in situ dynamics. Specifically, concerns of microbial life cycle changes brought on by drying, sieving, and rewetting are noteworthy. To address disturbance while at the same time maintaining the advantages of soil mixing and pooling, Comeau et al. (2018b) developed a new method using minimally disturbed soil microcosms. This method involves extracting intact soil cores, gently breaking the cores, pooling the soil (e.g., per treatment or land-use or management), removing the visible live roots, and repacking the moist nonsieved soil into microcosm cores at their original field bulk densities. This microcosm method was demonstrated to be effective in dividing the influence of soil temperature and moisture on Rh and in simulating seasonal climatic variations (Comeau et al. 2018a). Previous studies have analysed the effects of drying and sieving on soil respiration in comparison with intact cores (e.g., Stenger et al. 2002;Herbst et al. 2016) but comparative analysis between the microcosm method and traditional methods using dry-sieved soil has not been performed.
The objective of this study was to assess the effect on Rh of drying and sieving soils and the correlation between soil physico-chemical properties and CO 2 efflux. Four hypotheses were tested: (i) the three methods will correlate to each other; (ii) the three methods will strongly correlate with soil physicochemical parameters; (iii) disturbance caused by drying and sieving will reduce the regression coefficients compared to the microcosms; and (iv) dried and sieved soil would have a larger Rh compared to the microcosms because disturbing soil by dry-sieving exposes occluded organic matter to microbial degradation and breaks fungal hyphae (Datta et al. 2014).

Soil sampling
Soil was collected from 34 agricultural sites/farms in New Brunswick, Canada (Supplementary Table 1), encompassing a wide range of arable lands as part of a larger Agriculture and Agri-Food Canada cluster project (PSS2224). Composite soil samples were collected from each site with a Dutch auger (0-15 cm depth) using zigzag random sampling (Zebarth et al. 2021b). In the field, the soil samples were transferred into plastic bags kept in a cooler with icepacks, transported immediately to the laboratory and stored at 4 • C to keep them fresh prior to analyses. Soil samples were either dried (35 • C for 24 h) and passed through a 2 mm sieve or kept fresh for the microcosm method.

Soil chemical and physical parameters
The soil chemical analyses included pH (1:1 H 2 O), soil texture, aggregate stability, total C, total N, C:N ratio, dissolved C and N, NH 4 -N and NO 3 -N, and labile C, and were analysed at Agriculture and Agri-Food Canada laboratory in Fredericton, New Brunswick. The labile C was assessed with the permanganate oxidizable carbon (POX-C) method in duplicate samples (Culman et al. 2012). Briefly, air-dried soil (2.5 g) was mixed with 0.02 mol L -1 KMnO 4 . The mixture was shaken for 2 min at 240 rpm on a lateral shaker and allowed to settle for 10 min. A 0.5 mL aliquot of supernatant was diluted in 49.5 mL of deionized water and the absorbance was determined at 550 nm on a Biochrom Libra S60 Spectrophotometer (Biochrom Ltd., UK). The absorbance of four standard solutions was also determined (0.00005, 0.0001, 0.00015, and 0.0002 mol L -1 KMnO 4 ). POX-C was calculated as described by Culman et al. (2012). Wet aggregate stability was determined by slaking using an Eijkelkamp Wet Sieving Apparatus, according to a method modified from Angers and Mehuys (1993). Briefly, 4.0 g of air-dried soil aggregates of 1-2 mm size were placed into a 250 mm sieve, gently moistened, and repeatedly immersed for 3 min in water. The particles and aggregate fragments that passed through the sieve were filtered, dried, and weighed. The particles remaining on the sieve were repeatedly immersed in a 2 g L -1 NaOH dispersing solution for intervals of 5 min until there were only sand particles remaining in the sieve. Soil aggregate stability was calculated according to "Manual for Wet Sieving Apparatus" (Eijkelkamp 2008).
Total dissolved organic carbon and nitrogen were determined on fresh soil extracted with 50 mL 0.5 mol L -1 K 2 SO 4 solution and shaken at 200 rpm for 2 h. Extracts were filtered (40# Whatman), and concentrations of organic C and N in the extracts were determined using a Shimadzu TOC analyzer (TOC-L, Shimadzu Scientific Instruments, Inc.). The concentrations of NH 4 -N and NO 3 -N were analyzed colorimetrically on a QuickChem 8500 Flow Injection Analyzer (Lachat Instruments, Loveland, CO). Total soil organic carbon and total nitrogen concentrations were determined by dry combustion (Elementar varioMACRO; Skjemstad and Baldock 2007). Soil textural class (sand, silt, and clay) was assessed through the pipette sedimentation method following organic matter removal (Kroetsch and Wang 2007).

Soil heterotrophic respiration assays
Three assays were employed to carry out the ex situ Rh measurements. First was the 24 h burst test, equivalent to the 24 h burst test of Haney (Haney et al. 2008), which measures CO 2 -C emissions over a 24 h period following rewetting of dry soil. The second assay was the soil respiration rate measured 10 days after soil rewetting, to avoid any burst of CO 2 -C emissions associated with soil rewetting. The third assay used minimally disturbed microcosms that measure CO 2 -C from fresh soil that has been carefully repacked to field bulk density (Comeau et al. 2018b).
For the 24 h burst test and the 10-day incubation, 40 g of soil was placed in a 50 mL perforated beaker developed for the Solvita test (Solvita and Woods End Laboratories). The perforated beaker was then placed into a 500 mL Mason jar containing 20 mL of water, which allowed the water to wet the soil. The Mason jar was immediately covered with a lid containing a septum and 20 mL of CO 2 -free compressed air was added. Samples were incubated at 25 • C for 24 h and a 20 mL gas sample was collected and injected into a 12 mL preevacuated exetainer. The burst test value was calculated as the mass of C in the headspace at the end of the 24 h incubation per unit weight of oven-dried soil. Following the collection of 24 h burst test gas sample, the air in the chamber was flushed, the lid was replaced with parafilm, and samples were incubated another 9 days at 25 • C. After this, the parafilm was removed and the headspace was flushed with CO 2 -free compressed air for 20 s. A lid containing a septum was placed on the jars and 80 mL of compressed air was added. A 20 mL gas sample was collected at 0, 30, 60, and 90 min using a syringe and placed into a pre-evacuated exetainer. Taking into account the volume of air in the headspace at each time point and the removal of CO 2 by gas sampling, the mass of C was determined at each time point. The respiration rate was then calculated as the slope of the regression of the mass of C per kg of oven-dried soil against time (Haney et al. 2008).
For the microcosms, the fresh soil was repacked to bulk density of 1.1 g cm -3 in the 56.7 cm 3 microcosm cores (inner diameter 3.8 cm and height 5 cm) following Comeau et al. (2018b). Each soil core was placed individually into a hermetically sealed 2.9 L plastic container and left to stabilize in the dark for 2 weeks at 25 • C. After the preincubation was completed, for all microcosms, 20 mL gas samples were collected with an airtight syringe at 0, 24, 48, and 72 h after container closure. All incubations were performed at 40% gravimetric water content. In the microcosms, the water content was adjusted following Comeau et al. (2018b). For all methods, gas samples were analyzed with a gas chromatograph (Varian, Mississauga, ON) at the Dalhousie University in Truro Nova Scotia as described by Burton et al. (2008). Briefly, an electron capture detector (ECD) was connected with a CombiPAL autosampler. The autosampler removes a 2.5 mL volume from the sample tube and injects into a sample valve that delivers 0.5 mL to the ECD. The ECD was operated at 300 • C, with an argon/methane carrier gas (90% Ar and 10% CH 4 ) delivered at 10 mL/min, through a Haysep N 80/100 precolumn (0.32 cm diameter × 50 cm length) and Haysep D 80/100 mesh analytical columns (0.32 cm diameter × 200 cm length) both in a column oven operated at 70 • C. The precolumn was used in combination with a valve to remove water from the sample. Operational conditions and data handling were performed with the Varian Star software. In each analytical run, three concentrations of standard gas mixtures were included for quality assurance/quality control purposes. The Ideal Gas Law was used to determine the amount of CO 2 -C and fluxes (Lang et al. 2011).

Statistical analyses
Calculations and descriptive statistics were performed with Microsoft Excel XP . For the 3 assays, an analysis of variance test to compare the Rh averages between the 3 methods and verification of the homogeneity of variance were performed with the statistical program R version 2.8.1 (R Development Core Team 2008) with the function linear model (anova.lm). Spearman correlations analyses were performed for the soil physico-chemical parameters using SigmaPlot 14.5 (Systat Software, Inc.). Statistical differences were deemed significant at α = 0.05.

Results and discussion
Ex situ assessment of Rh is regularly used to infer soil health and microbial activity. Different methods to determine Rh are commonly performed; however, until now, there have been few studies that perform comparative analyses between procedures. This study verified the correlation of Rh with 11 physico-chemical parameters of 34 different soils collected in New Brunswick, Canada to judge the strength of the different methodological approaches. The premise was that superior methods will show more and stronger correlations between Rh and soil physico-chemical properties at a fixed standard temperature and moisture level.
The 24 h burst tests, 10-day incubations, and microcosms had ex situ Rh of 1.42 ± 0.47 (average ± standard deviation), 1.22 ± 0.58, and 1.60 ± 0.45 mg CO 2 -C/kg dry soil/h, respectively (Table 1). Equivalent values of 0.4-3 mg CO 2 -C/kg dry soil/h have been previously reported for the Atlantic Region of Canada (Miller et al. 2009;Cooper et al. 2011;Zebarth et al. 2022). Drying and sieving of the soil did not increase Rh compared with the microcosms as originally expected. Conversely, the Rh of the 10-day incubation method was significantly lower than the minimally disturbed microcosms. For the 24 h burst test with drying and sieving soil, Rh was slightly and nonsignificantly lower than the minimally disturbed microcosms. Other studies analyzing the effect of soil disturbance on CO 2 efflux from lab incubations have found that disturbance might occasionally enhance microbial activity and respiration. For example, Herbst et al. (2016) and Datta et al. (2014) found that air-drying and sieving can impact the relationship between soil moisture and soil respiration due to alteration of macroaggregates. This was explained by aggregates that protect a fraction of soil organic C from mineralization due to the occlusion of C within the aggregates (Pulleman and Marinissen 2004). However, other authors found no significant difference in soil respiration between intact and sieved soils (Stenger et al. 2002;Thomson et al. 2010), while Adekanmbi et al. (2019) found a decrease in soil respiration due to of sieving disturbance, same as in our study. The smaller ex situ Rh in sieved soil that we observed might be a consequence of changes in water-holding capacity after sieving. Altering soil structure and aggregate sizes through sieving affects both total porosity and pore size distribution (Wu et al. 1990). That is, with the same amount of water, the types of pores filled were likely not the same, which in turn could have affected aerobic microbial activity. Specifically, grinding of soil may reduce the water capacity of a soil and therefore increase the likelihood of anaerobic conditions (Brinton 2020).
As studies show that to avoid destroying the microaggregates, the threshold appears to be around a 2 mm sieve mesh size, this study likely did not cause artifacts that boost microbial activity for the sieved soil. Drying temperature during soil preparation has been linked to changes in total respiration, and it is generally thought that temperatures between 40 and 60 • C are ideal; lower drying temperatures may pre- vent the burst effect (Franzluebbers and Veum 2019; Laffely 2019), while higher temperatures (e.g., 100 • C) can cause destruction of microbial communities and greatly reduce 24 h respiration (Haney et al. 2004). Accordingly, the drying at 35 • C for 24 h used in this study is unlikely to have annihilated major microbial taxa. The physical and chemical soil properties presented in Table 1 are in the same ranges as for other published studies in the region (Cambouris et al. 2006;Nyiraneza et al. 2012;Zebarth et al. 2019;Abedin and Unc 2020;Chen et al. 2022;Nyiraneza et al. 2021;Zebarth et al. 2021a). Accordingly, this experiment can be considered representative of the soils of the Atlantic Region that are dominated by shallow podzols and luvisols (Fhamy et al. 1986;Krzic et al. 2021).
The 24 h burst test significantly correlated with both the 10-day incubation and the microcosm test (r 2 = 0.18 and 0.13, respectively); however, the 10-day incubation method (Table 2) did not correlate with the microcosm test. Table 3 shows that the 24 h burst test, pH, and total dissolved N positively and significantly correlated with ex situ Rh. As soil extracellular enzymes are pH dependent, previous incubation studies have shown the soil pH to influence Rh (Yiqi et al. 2006;Wang et al. 2010). Other incubation studies have shown that the amount of total dissolved N significantly enhances Rh due to the influence of fast-growing microbial communities (Cookson et al. 2007;Soong et al. 2020). Specifically, changes in inorganic N (i.e., NH 4 and NO 3 ) often strongly enhance Rh fluxes from incubations due to N limited status of most soil (Micks et al. 2004;Burton et al. 2012;Zhang et al. 2021). Table 4 shows that in our study, for the microcosm test, only total dissolved N positively and significantly correlated with ex situ Rh. The relationship between total dissolved N and ex situ Rh was similar for the microcosms and 24 h burst tests (Fig. 1). However, the slope of the regression line was almost three times greater for the 24 h burst test than for the microcosms (i.e., 0.014× vs. 0.005×). For the 10-day incubation, no soil properties significantly correlated with Rh (Table 5). Accordingly, this method might be suboptimal as an indicator of soil health and soil functional status since reliable ex situ respiration methods ought to have some stable correlations between soil properties and CO 2 efflux.
The soil physical and chemical parameters determined in this study have been shown in previous studies to influence or correlate to Rh from incubations. For example, Rh rates were shown to be proportional to soil organic carbon content without changing microbial communities (Gan et al. 2020;Nyberg et al. 2020). Due to the fact that microbial respiration per unit microbial biomass depends on organic matter C:N ratio, it has been demonstrated that, in incubations, reduc-  ing the C:N ratio can increase cumulative respiration (Spohn 2015;Nguyen et al. 2017). In addition, because soil microbes use carbon in labile form, it has been found that Rh is notably higher in fresh soil in comparison to incubation-DOC depleted soil (Birge et al. 2015). Similarly, POX-C, a measurement of labile C availability, has been shown to positively correlate with Rh in incubation where POX-C levels were regulated (Cleveland 2007). Likely, in our studies, none of these soil properties were at levels that would allow them to act as limiting factors for microbial activity, explaining why they did not have the measurable influence on Rh we hypothesized. Soil physical properties and Rh have been studied in similar incubation experiments. Zezhou et al. (2022) demonstrated that the disruption of the aggregates liberates soil nutrients, which in turn promotes Rh. Similarly, Peng et al. (2017) found that stable microggregates emit less Rh than dis-turbed macroaggregates. Regarding soil texture, Cable et al. (2008) found that Rh is more sensitive to disturbance and fluctuation on coarse versus fine-textured soils. Moreover, Chodak et al. (2010) found that soil texture had a greater influence on Rh than microbial diversity. In our study, soil texture and aggregate stability did not have a direct influence on Rh. It is possible that the range of texture and aggregate stability between the samples was too small to be able to detect significant correlations.
Overall, we recommend careful consideration when choosing a method of soil sampling and preparation prior to incubation for ex situ Rh assessment. Both homogenizing soil and the use of minimally disturbed microcosms have advantages and disadvantages. For example, when soils are sieved, it is easier to account for the heterogeneity of field conditions since a larger number of samples from multiple locations can be pooled, homogenized, and assigned to treatments. Thus,  in this case, the 24 h burst test is a suitable method. On the other hand, in other types of studies, when Rh needs to be analysed together with microbial diversity, the microcosm method should be considered given its minimal disturbance effect.

Conclusions
To assess the effects of soil management practices on soil health, ex situ assessment of Rh is commonly performed. Different methods to determine ex situ Rh exist, but there have been only a few studies to compare their accuracies. In this study, drying and sieving the soil did not increase the Rh as hypothesized. On the contrary, the 10-day incubation method with dried and sieved soils yielded an average Rh significantly lower than that obtained using the minimally disturbed microcosms. The 24 h burst test using dried and sieved soil was only slightly lower than that obtained using minimally dis-turbed microcosms. The microcosm method was weakly correlated with the 24h burst test, but did not correlate with the 10-day incubation method. Moreover, all three methods had low coefficients of regression with the 11 soil physicochemical properties analyzed. Total dissolved N was the factor with the strongest correlation with Rh. We conclude that drying and sieving soil prior to performing ex situ Rh measurements using 24 h burst tests or the minimally disturbed microcosms can produce cautiously reliable and comparable results for soil from Canada's Atlantic region. Further studies should simultaneously compare ex situ and in situ measurements of soil respiration in this region and use a broader diversity of soil types and land managements.