Cellular and non-cellular mineralization of organic carbon in soils with contrasted physicochemical properties

Abstract

It has been recently demonstrated that soil organic carbon (SOC) mineralization is supported by intracellular respiration of heterotrophic microorganisms and by non-cellular oxidative processes. However, little is known about the prevalence and drivers of non-cellular SOC mineralization among soils. In this study, untreated and gamma-irradiated soils sampled along a latitudinal gradient and exhibiting contrasted physicochemical properties were incubated in order to quantify potential non-cellular SOC mineralization and to identify its sensibility to soil properties. In sterilized and unsterilized soils, CO₂ emission mirrored O₂ consumption signifying the presence of several coupled redox reactions transferring electrons from organic C to intermediate acceptors and to O₂. This supports the idea that non-cellular mineralization results from extracellular oxidative metabolisms catalyzed by soil enzymes and/or abiotic catalysts. Our findings also show that non-cellular SOC mineralization is ubiquitous and contributes to 24 % of soil respiration on average. Cellular and non-cellular SOC mineralization are positively linked but the contribution of non-cellular processes to soil CO₂ emissions increases with dissolved organic carbon concentration.

Representing one of the largest terrestrial source of CO₂ flux, soil organic carbon (SOC) mineralization significantly impacts atmospheric CO₂ concentration and climate (Lal & Kimble, 2000; Paterson & Sim, 2013). SOC mineralization is traditionally viewed as a two-step process. First, microorganisms secrete extracellular enzymes to convert insoluble SOC polymers into compounds assimilable by microbial cells. In the second step, during which carbon (C) is released as CO₂, assimilated compounds are carried out by an oxidative metabolism including many coupled reactions, enzymes and cofactors as well as required redox and pH conditions. Given this complexity, it is taught that respiration (second step of the C mineralization process) is strictly an intracellular process. However, recurrent observations of persistent substantial CO₂ emissions in soil microcosms where sterilization treatments reduced microbial activities to undetectable levels challenged this paradigm (e.g., Peterson, 1962; Lensi et al, 1991; Maire et al, 2013; Blankinship et al, 2014; Kéraval et al, 2016). Different non-cellular processes (Fig. S1) have been proposed as potential contributors to these CO₂ emissions such as (i) the partial degradation of aromatic compounds induced by reactive oxygenated species and/or metals (Fe^3+, Mn⁴⁺)(Majcher et al, 2000; Wang et al, 2017), (ii) the extracellular decarboxylation of the metabolites of the Krebs cycle supported by some decarboxylases released during cell lysis and retaining their activities in soil (Maire et al, 2013; Blankinship et al, 2014) and (iii) the complete mineralization of organic compounds supported by an extracellular oxidative metabolism (EXOMET)(Maire et al, 2013; Kéraval et al, 2016). EXOMET differs from the two other processes in its complexity. Whereas (i) and (ii) involve single catalytic reactions, EXOMET implies numerous coupled redox reactions capable of complete conversion of organic C into CO₂ with an electron transfer to O₂.

Irrespective of the mechanism, nothing is known about the prevalence and drivers of non-cellular SOC mineralization (R_NON-CELLULAR) in soils with contrasting physicochemical properties. The objectives of this study were to (i) generalize non-cellular SOC mineralization to many soils, (ii) quantify R_NON-CELLULAR and its contribution to total SOC mineralization and (iii) identify soil properties that most influence R_NON-CELLULAR.

Twenty two soils (Table S1) with various physicochemical properties (Table 1, Fig. S2, Table S2) were collected in permanent grasslands along a latitudinal gradient from Europe to Southern Africa. Untreated and sterilized soils (gamma-irradiations) were incubated to measure potential R_NON-CELLULAR and total SOC mineralization (R_SOIL), respectively (see Supplementary information). Soils were incubated at 20°C and at a water potential of -100 kPa. We used γ-irradiations at 45 kGy to sterilize soil samples because previous investigations combining various molecular, microscopic and biochemical techniques showed that such procedure reduce microbial biomass to undetectable levels, prevent any microbial re-colonization (MacNamara et al, 2003; Berns et al, 2008, Maire et al, 2013) and limit the impact of sterilization treatment on the physicochemical soil properties compared to the use of fumigants or thermal sterilization treatments (MacNamara et al, 2003). All manipulations were done under sterile conditions and sterility of soils was checked, at the end of incubation, using culture approach and propidium iodide staining coupled to flow cytometry (see Supplementary information).

Table 1.

Main physicochemical properties of the twenty two soils sampled. (S1 to B6, see Table S1 for correspondence). OC: organic carbon, IC: inorganic carbon, DOC: dissolved organic carbon.

Soil references	OC	IC	DOC	PH	Sand	Clay	Silt
	mg.g-1		μg g-1		%
B1	1.0	0.5	9.6	6.5	96.6	2.2	1.1
B2	1.4	0.9	13.8	6.2	95.6	3.3	1.1
B3	2.2	0.5	17.8	6.2	95.7	1.1	3.2
B4	2.1	0.7	11.3	6.2	96.6	0.0	3.4
B5	2.9	0,0	13.0	6.3	96.7	1.1	2.2
B6	1.6	0.7	9.7	6.0	92.3	2.2	5.5
S1	31.7	52.4	58.7	8.1	59.2	6.0	34.8
S2	20.4	79.2	62.1	8.2	44.4	6.2	49.4
S3	31.1	64.4	81.3	8.0	49.0	7.2	43.8
S4	43.2	67.8	63.4	7.7	48.9	6.2	44.9
S5	31.8	76.8	68.9	8.0	54.4	6.8	38.8
S6	20.3	83.1	49.4	8.1	49.4	3.9	46.8
S7	25.9	49.5	54.4	7.6	52.4	4.1	43.5
S8	17.8	94.7	48.1	8.1	49.8	9.2	41.0
S9	44.9	25.5	115.3	8.0	60.4	2.2	37.4
S10	28.3	66.5	63.8	8.1	59.4	5.8	34.8
S11	29.6	27.4	114.7	7.6	74.3	6.6	19.1
F1	21.9	0.0	84.3	6.4	70,0	6.8	23.3
F2	15.2	6.7	23.4	8.1	42.2	10.0	47.8
F3	21.5	0.6	26.9	6.0	32.2	3.3	64.4
F4	17.4	6.2	21.2	7.6	42.2	16.7	41.1
F5	34.8	0.0	41.9	6.0	55.6	13.3	31.1

We recorded O₂ consumption and CO₂ production over a 16-day incubation period in γ-irradiated and untreated soil microcosms (Figs 1a and 1b, Table S3). As none viable cells were detected in irradiated soil microcosms (Supplementary information, Table S4), we assumed that γ-irradiations allowed quantifying potential R_NON-CELLULAR in sterilized soil microcosms. We acknowledge a possible contribution of carbonates to CO₂ emissions in alkaline soils (Serrano-Ortiz et al, 2010). However, like Maire et al (2013), we observed that the consumption of O₂ mirrored the production of CO₂ (Fig. S3) with no significant difference (p<0.05) in the average value of the respiratory coefficient (CO₂/O₂) between R_SOIL and R_NON-CELLULAR (1.2 and 1.05, respectively). Accordingly, we used O₂ consumption in untreated and γ-irradiated soils as a proxy of R_SOIL and R_NON-CELLULAR, respectively and we estimated the contribution of the cellular respiration (R_CELLULAR) to SOC mineralization.

Figure 1.

(a) O₂ consumption and (b) CO₂ production by γ-irradiated and untreated soil microcosms. Twenty two soils sampled along a latitudinal gradient were incubated (From S1 to B6 in x axis, see Table S1 for correspondence). (c) Contribution (in %) of R_NONCELLULAR and R_CELLULAR to R_SOIL see SI). ns indicates that measured fluxes were not significantly different from 0 (p<.05).

In the wide range of soil types investigated, the contribution of R_NON-CELLULAR to R_SOIL ranged from 13 to 50 %, with an average of 24 % (Fig. 1c). The close positive correlation between R_CELLULAR and R_NON-CELLULAR (Fig. 2a) could result from (i) the stimulation of R_NON-CELLULAR by cell activities such as the secretion of extracellular enzymes solubilizing SOC releasing substrates for R_CELLULAR and R_NON-CELLULAR, (ii) the liberation of intracellular enzymes and substrates after microbial cell lysis responsible for decarboxylation and/or performing EXOMET (Maire et al, 2013, Blankinship et al, 2014, Kéraval et al, 2016) and/or (iii) the dependency of R_CELLULAR and R_NON-CELLULAR on the same rate-limiting factors (e.g., amount and assimilability of C, Figs.2b and 2c). Pearson correlations partly accredited this last hypothesis as these two respiration pathways globally responded to the same environmental factors (Table S6). SOC, especially, was identified as a soil variable exhibiting a highly significant correlation with R_CELLULAR and R_NON-CELLULAR (Fig. 2b, 2c). However, R_CELLULAR and R_NON-CELLULAR exhibited a different level of sensibility to dissolved organic carbon (DOC) content (Fig. 2b, 2c), with R_NON-CELLULAR having a higher sensitivity to DOC content than R_CELLULAR. Consistently, the contribution of R_NON-CELLULAR to soil respiration (R_SOIL) increased with DOC content (Fig. 2d). These findings suggest that these two processes use, at least partly, different sources of carbon. Soil microorganisms would use the large stock of insoluble SOC for their respiration by using extracellular enzymes present on their wall or released in the vicinity of the cell (Burns et al, 2013). In contrast, R_NON-CELLULAR would depend on soluble substrates that must diffuse between catalysts that are likely dispersed in soil pores; hence this respiration pathway may be limited by diffusive process and the concentration of substrates.

Figure 2.

Linear regressions between (a) cellular (R_CELLULAR) and non-cellular (R_NON-CELLULAR) soil organic carbon (SOC) mineralization, (b) R_NON-CELLULAR, SOC and dissolved organic carbon (DOC) contents, (c) R_CELLULAR, SOC and DOC contents, (d) contribution (in %) of R_NONCELLULAR to R_SOIL and DOC. R_NON-CELLULAR and R_SOIL were quantified using O₂ consumption in irradiated and non-irradiated soils, respectively. R_CELLULAR was estimated by the difference between R_SOIL and R_NON-CELLULAR.

The steady production of CO₂ over 16 days combined to a consumption of O₂ (Table S3) strongly suggests the presence of several coupled redox reactions transferring electrons from organic C to O₂. This observation supports the idea that non-cellular mineralization mainly results from an EXOMET catalyzed by soil enzymes and/or abiotic catalysts. However, other processes such as single decarboxylation (e.g. by decarboxylases, reactive oxygenated specie and/or metals) may contribute to R_NON-CELLULAR in sterilized soil. We therefore encourage further investigations using metabolomic approaches to identify potential metabolic intermediates and the chemical reactions in play.

Globally, our results show that R_NON-CELLULAR is ubiquitous substantially contributing to CO₂ emissions (24±3%) in many soils with contrasting physicochemical properties and can be estimated by measuring R_SOIL and DOC contents. Our estimation of R_NON-CELLULAR contribution to R_SOIL (13-50 %, Fig.1c) is in a similar range of that determined by Maire et al (2013) through a modeling approach. However, we acknowledge that methods used here and in Maire et al. (incubation of sterilized soils) can lead to an overestimation of R_NON-CELLULAR because sterilization (i) increases DOC content (McNamara et al, 2003; Kéraval et al, 2016) and (ii) removes the possible competition for DOC between R_CELLULAR and R_NONCELLULAR. It is therefore crucial to develop a method of quantification without these biases. The recent discovery that R_NON-CELLULAR has a specific ¹³C isotope signature open avenues toward a quantification based on isotopes without the use of sterilization (Kéraval et al, 2016).

The greater sensitivity of R_NON-CELLULAR to DOC content compared to R_CELLULAR suggests that its contribution to soil CO₂ emissions may be larger in DOC-rich soils (Fig. 2d) such as permafrost, soils submitted to drying/rewetting cycles and rhizosphere soils (Nguyen, 2003; Frey & Smith, 2005). Soil water availability might be another important factor affecting this contribution due to its control on DOC diffusion and distribution in the different soil spaces (R_CELLULAR and R_NONCELLULAR may occupy different soil pore size, see Maire et al, 2013). Further studies are required to directly study the effect of DOC content and soil water availability on the balance between R_CELLULAR and R_NONCELLULAR.