Soil organic carbon under lockdown: Fresh plant litter as the nucleus for persistent carbon


 The largest terrestrial organic carbon pool, carbon in soils, is regulated by the intricate connection between plant carbon inputs, microbial activity, and soil matrix. This is manifested by how microorganisms, the key players in transforming plant-derived carbon into soil organic carbon, are controlled by the physical arrangement of organic and inorganic soil particles. We studied the role of soil structure on the fate of litter-derived organic matter and we propose that the persistence of soil carbon pools is directly determined at plant–soil interfaces. We show that while microbial activity and fungal growth is controlled by soil structure, occlusion of organic matter into aggregates and formation of organo-mineral associations occur in concert on litter surfaces regardless of soil structure. These two mechanisms—the two most prominent processes contributing to the persistence of organic matter—occur directly at fresh litter that constitutes a key nucleus in the build-up of soil carbon persistence.


Introduction
physical soil structure defines the porous network, affecting the movement and bioavailability of gases 48 (e.g., CO 2 and O 2 ) and water 5 . Determined by pore size, the differences in soil water contents can shape 49 ecological niches suitable for certain microbial taxonomic groups. The size of pores also controls the 50 contact between microorganisms and their essential source of energy and nutrients-the litter 6 . The 51 effect of soil structure on the functionality of the microbial community can be predicted, e.g., via oxygen 52 availability, which regulates C turnover 7 . 53 Over the last decades, it has become more evident that inherent recalcitrance, i.e., the reduced 54 decomposability due to the chemical composition of OM, is of less importance to SOM persistence 55 compared to soil structure-driven mechanisms that rely on soil aggregation and accessibility of reactive 56 mineral surfaces 8,9 . Soil C is mostly stored in the following two major pools: as particulate OM (POM; 57 particulate organic residues mostly of plant origin) and mineral-associated OM (MAOM; OM adhering 58 to mineral surfaces) 10, 11 . Physical mechanisms, such as the potential of OM compounds to adhere to 59 mineral surfaces 12 , or the accessibility of substrates for microorganisms 3 , are now paving the way for a 60 better understanding of OM cycling and persistence in soils. This persistence of soil C is regulated in 61 microscale hot spots at which microorganisms transform plant-derived OM into SOM. The functioning 62 of biogeochemical interfaces between plant litter substrates, microorganisms, and soil mineral surfaces 63 requires chemical, physical and biological factors to be considered in consortium. 64 We applied a systemic approach by investigating how physical soil texture governs the pathway of litter-65 derived C compounds from initial plant litter into more persistent SOM pools via microbial 66 transformation in a relevant process scale (µm-mm). To disentangle mineral-microorganism 67 interactions that regulate these processes, we incubated two differently textured soils together with 13 C-68 labeled litter in a 95-day microcosm experiment. Aside from monitoring CO 2 production and litter-69 derived 13 CO 2 release, we followed the alterations in the chemical composition of SOM in POM and 70 MAOM, and the microbial communities and their uptake of litter-derived 13 C into phospholipid fatty 71 acids (PLFA). The intact biogeochemical interfaces between plant residues, microorganisms, and soil 72 minerals were, for the first time, directly studied using nano-scale secondary ion mass spectrometry 73 (NanoSIMS). Our objective was to quantify the interactions between microbial litter decay and the 74 parallel formation of more persistent soil C pools in regard to aggregate formation, and the association 75 of microbial C with mineral surfaces controlled by soil texture. 76

77
Litter decomposition and native soil carbon priming. We measured how different soil textures 78 and litter addition (enriched in 13 C, δ 13 C = 2129 ± 82 ‰ V-PDB) affected the soil heterotrophic 79 respiration by monitoring the soil-CO 2 emissions. By analyzing 13 CO 2 , we were able to differentiate CO 2 80 derived from native soil organic C from CO 2 derived from the added litter. We report the CO 2 -derived 81 C per amount C in incubated samples to directly showcase the mechanistic process level. In the coarse-82 textured soil, the total native respiration (105.6 mg CO 2 -C g −1 C bulk ) and the net litter-derived CO 2 (47.7 83 mg CO 2 -C g −1 C bulk ) were significantly higher than in the fine-textured soil (65.3 mg CO 2 -C g −1 C bulk 84 and 29.0 mg CO 2 -C g −1 C bulk , p < 0.001, t = −7.512 and t = −6.593 respectively; df = 8 for both, Fig. 1). 85 While the litter-derived CO 2 accounted for around 30% of the total respiration in both soil textures, the 86 litter addition induced a higher total priming effect in the coarse-textured soil ( Fig. 1 b and d), accounting 87 for a net release of 27.5 mg CO 2 -C g −1 C bulk from the native soil organic C in the coarse-textured soil 88 compared to 12.8 mg CO 2 -C g −1 C bulk in the fine-textured soil (p < 0.001, t = −7.686, df = 8). 89 respired CO2-C in soil with b coarser and d finer texture is displayed on the right (means, SDs displayed 93 with errors bars, n = 5), together with the total priming effect. Asterisks represent significant differences 94 between the textures (***p < 0.001).

95
Fate of litter-derived carbon in particulate and mineral-associated OM fractions. We 96 assessed the contribution of OC derived from the decaying litter to the formation of differently stabilized 97 OM pools in two soils with contrasting textures divided into three depths (top, center, and bottom) by 98 soil fractionation according to density and size. The soil-derived C in mg g −1 C bulk was similarly 99 distributed across OM fractions for both differently structured soils. The MAOM fraction dominated the 100 C storage in both soils (Fig. 2 a). In the coarse-textured soil, we found a significantly higher litter-101 derived C content occluded within aggregates (oPOM) (71.1 mg C g −1 C bulk compared to 36.8 mg C g −1 102 C bulk in the fine-textured soil, p = 0.007, t = −5.03, df = 4) and a slightly higher content in the MAOM 103 fraction (101.3 mg C g −1 C bulk in the soil with coarser texture compared to 48.8 mg C g −1 C bulk in the soil 104 with finer texture, p = 0.08, W = 0) (Fig. 2 b). Although not statistically significant, a tendency of a 105 higher contribution of litter-derived C recovered as oPOM and MAOM in the coarse-textured soil further 106 extended down with soil depth to the center layer of the microcosms (p = 0.06 in both cases, t = −2.66 107 and −2.60, respectively; df = 4 for both).   were similar to those of the controls (Fig. 4 a and b). While the differences in soil texture had no effect 140 on the overall community structure, a strong response to litter addition was detected in fungal textures (means, SDs displayed with errors bars, n = 3). Lowercase letters represent the significant 174 differences (p < 0.05) between microbial groups and asterisks (*p < 0.05, **p < 0.01, ***p < 0.001) 175 represent the differences between textures. Differences between depths were significant in all groups.

177
Formation of MAOM fostered by microbial activity on decaying POM surface. We gained 178 a direct insight on the biogeochemical interface between decaying plant residues (POM), mineral 179 particles, and microorganisms at the microscale using scanning electron microscopy (SEM) and 180 NanoSIMS. Large areas of litter-derived POM particles were covered in 13 C-enriched microbial-derived 181 extracellular polymeric substances (EPS), forming a biofilm-like structure that was intertwined with 182 fungal hyphae and unicellular microorganisms (presumably bacteria). Clay-sized minerals were directly 183 enclosed into the biofilm on the POM surface (Fig. 6 a and b). The microorganisms and EPS were 184 significantly enriched in N compared to the underlying POM, with a higher 12 C 14 N − : 12 C − ratio obtained 185 for EPS, followed by hyphae and bacteria (Fig. 6 d). The 13 C − :( 12 C − + 13 C − ) ratios for fungal hyphae, 186 bacteria, and EPS (3.0, 2.3 and 3.0 atom % 13 C, respectively) were well over the natural abundance level 187 (1.1 atom % 13 C), and the hyphae showed a significantly higher enrichment compared to bacteria and 188 POM (p < 0.05, df = 3, Fig. 6 c).  Soil texture, and thus the 3D structure of soils, controls overall microbial activity; the coarser soil texture 212 entailed both higher decomposition of litter-derived OM, and an increased priming effect, fostering the 213 mineralization of native soil C (Fig. 1 b and d). Plant litter fragments located in larger soil pores of 214 coarse-textured soils are more easily accessible; therefore, litter decomposition is enhanced [16][17][18] . In the 215 coarse-textured soil, the increased accessibility and, hence, increased bioavailability of litter-derived C 216 was further demonstrated by consistently higher proportions of labeled PLFAs in gram-negative bacteria 217 across all soil depths (Fig. 5 b); bacteria that are specialized in the processing of labile plant C 218 sources 19,20 . 219 We show that the coarse-textured soil offered a more favorable habitat for fungi in a micro-environment 220 rich in bioavailable substrates formed by fresh unprotected litter. Fungal abundance increased by more 221 than five-fold following the litter addition in the coarse-textured soil. Furthermore, a substantial part (92 222 % in the coarse-textured soil) of the fungal biomass was directly derived from the added plant litter, as 223 demonstrated by the PLFA profiles (labeled PLFA profiles; Fig. 5 b). This highlights the key role of the 224 fungal community for rapid litter decomposition, particularly in coarse-textured soils. Soil structure with 225 a distinct soil pore network determines the abundance and community structure of microbiota 21 . Fungi 226 are mainly found in macropore spaces (> 10 µm) that are noticeably larger than the hyphae itself 16,22,23 . 227 The filamentous growth of the mycelium enables fungi to bridge air-filled pore spaces, supporting them 228 to overcome capillary boundaries between wet and dry soil, and to adapt to heterogeneous pore 229 networks 24-26 . Consequently, under the physical conditions of coarse-textured soils, fungi have a clear 230 advantage over other microorganisms to reach OM in hard-to-access soil compartments that are not 231 connected via water nor biofilms 21 . We stress that in sandy soils, fungi are key to sustain crucial soil 232 functions such as C and nutrient cycling by the transformation of litter-derived OM into SOM. 233 In the coarse-textured soil, fungal activity extended away from the litter source, thereby promoting a 234 downward transfer of litter-derived C into deeper soil layers (PLFA depth profiles; Fig. 4). This pattern 235 can partly be attributed to the apical properties of the fungal mycelium, enabling the translocation of C 236 sources throughout the fungal colony 27-29 . The expansion of hyphal networks facilitates the incorporation 237 of litter into aggregates 30,31 . The stabilization of aggregated soil structures can be ascribed to the 238 exudation of EPS (e.g., polysaccharides, Fig. 6 a and b) from the hypha 32 . We propose that the expansion 239 of fungal hyphae, together with its interactions with mineral particles, results in the build-up of litter-240 derived oPOM in the deeper soil layers away from the litter source (Fig. 2 b). This intricate interaction 241 between fungal hyphae, plant residues, and mineral particles adhering to microbial-derived EPS was 242 underlined by spectromicroscopic imaging (Fig. 6). With the direct measurement of intact plant-fungi 243 interfaces, we emphasize the key role of fungi in the translocation of litter-derived C within soils, as 244 well as in the formation of aggregates and mineral-associated OM-a process which ultimately drives 245 the stabilization of litter-derived C compounds in soils. 246 We were able to demonstrate the incorporation of plant C into microbial biomass directly in the interface 247 of plant residues and soil minerals. This was quantified with high levels of 13 C enrichment in fungal 248 hyphae and microbial EPS on the POM surface (Fig. 6). The direct contact between minerals ( 16 O − 249 distribution; Fig. 6) and microbial biomass ( 12 C 14 N − ; Fig. 6), together with the enmeshment of fresh litter 250 (free POM) with fungal hyphae and microbial-derived EPS (Fig. 6 a and b), promotes the gluing of fine-251 sized soil minerals. This agglomeration of fine mineral particles, driven by microbial activity and 252 regulated by the bioavailability of litter-derived C, drives aggregate formation and soil structure 253 development directly at the plant-soil interface. In addition, the chemical composition of the litter-254 derived OM that got entrapped in soil aggregates (oPOM) by this soil structure formation resembled the 255 undecomposed litter (Fig. 3). Thus, particulate OM acts as an important precursor for aggregate 256 formation and parallel occlusion of litter-derived POM into aggregated soil structures (Fig. 7). 257 simultaneous universal processes across soils of different structure (Fig. 7). These two mechanisms 273 strongly rely on the spatial proximity of particulate litter and its surfaces, microbial residues, and fine- The experimental design involved four treatments; soils of two textures, either with or without 13 C-288 labeled maize stalks. In order to obtain a coarse-textured soil (sandy clay loam; 24% clay, 15% silt, and 289 60% sand), half of the initial soil was mixed with quartz sand (Quarzwerke, Frechen, Germany). 290 Approximately 120 g (for coarser texture) and 90 g (for finer texture) of soil was filled homogeneously 291 and gently packed (bulk density 0.9-1. in-between the two samplings was adapted to the current respiration rates. During the incubation period 305 of 95 days, the CO 2 concentration, as well as the 13 C abundance in the respired CO 2 , was measured via 306 gas chromatography isotope ratio mass spectrometry (GC/IRMS) (Delta Plus, Thermo Fisher, Dreieich, 307 Germany). The CO 2 levels were calibrated against three calibration gases (890, 1500 and 3000 ppm 308 CO 2 ; Linde AG, Pullach, Germany). Then, source carbonic acid with known isotopic composition 309 diluted in helium was used as a lab standard. This standard was in turn calibrated against three 310 international standards (RM 8562, RM 8563, and RM 8564; International Atomic Energy Agency, 311 Vienna, Austria) with a dual inlet system. The temperature and water holding capacity were kept 312 constant at 21 °C and 60%, respectively, along with the incubation period. 313

314
After 95 days of incubation, each microcosm was cut into three horizontal sections with a razor blade, 315 separating the top, center, and bottom layer (each 1.67-cm high). Subsamples for subsequent microbial 316 analyses were freeze-dried and stored at 4 °C, and dried aliquots for fractionation were stored in sealed 317 plastic containers at 20 °C. Furthermore, a few POM particles were selected manually for NanoSIMS 318 measurements. 319

320
The soil was separated into five distinct OM fractions using a combined density and particle size 321 fractionation scheme 33 . Air-dried soil (18-20 g) was gently capillary-saturated with sodium 322 polytungstate solution (Na 6 [H 2 W 12 O 40 ]; 1.8 g cm −3 ) and after 12 h, the free-floating particulate organic 323 matter (fPOM) was collected using a vacuum pump. oPOM was released from aggregated soil structures 324 via ultrasonic dispersion (Bandelin, Sonoplus HD 2200; energy input of 440 J ml −1 ) allowing its 325 separation from heavier minerals. The excess salt was removed from the oPOM by washing it with 326 deionized water over a sieve (20-µm mesh size), which yielded an oPOM fraction of < 20 µm 327 (oPOM small ). Both fPOM and oPOM fractions were washed for several times using deionized water and 328 pressure filtration (20-µm mesh) until the solution dropped below an electric conductivity of < 5 µS/cm 329 via pressure filtration. The oPOM small fraction was cleaned via saturation with deionized water for 24 h. 330 While sand and coarse silt fractions were separated by wet sieving, mineral fractions < 20 µm were 331 separated via sedimentation, and later combined as one MAOM fraction. The C, N, and 13 C contents 332 were determined for freeze-dried and milled OM fractions, as well as milled bulk soil, via dry combustion with an isotope ratio mass spectrometer (Delta V Advantage, Thermo Fisher, Dreieich, 334 Germany) coupled with an elemental analyzer (Euro EA, Eurovector, Milano, Italy). Acetanilide was 335 used as a lab standard for calibration and to determine the isotope linearity of the system, and was in 336 turn calibrated against several suitable isotope standards (International Atomic Energy Agency, Vienna, 337 Austria). International and lab isotope standards were included in every sequence to create a final 13 C 338 correction. Since the samples did not contain carbonates, the C contents were assumed to be equal to 339 organic C contents. Along with the incubation period, the amount of C respired per hour was computed as 353 where ΔCO 2 /Δt is CO 2 increase over time, V HSP is the volume of the headspace of Mason Jars. The 355 volume of an ideal gas is set at 22.4, and 12 represents the atomic mass of C. 356 Subsequently, the precentage of respired CO 2 originating from the litter was calculated as 357 where δ 13 C resp emission gives the δ 13 C for the current CO 2 emission between the two samplings (‰ V-359 PDB), δ 13 C control is the average δ 13 C of the control soils at the time of measurement, and δ 13 C litter is the 360 δ 13 C signature of the labeled litter. Finally, the respired C originating from the soil was computed as 361 where δ 13 C labeled is the 13 C enrichment in labeled samples, δ 13 C control is the 13 C enrichment in controls 365 (natural abundance level, i.e., 28 ‰ V-PDB), and δ 13 C litter is the 13 C enrichment in the added litter (i.e., 366 2129 ‰ V-PDB) from which the amount of litter-derived C within each OM fraction could then be 367 determined as 368 where C fraction is the amount of C in mg g −1 , and m is the recovered mass (g) of each fraction after the 370 fractionation. 371

372
The PLFA patterns were analyzed 37  Chemstation, Santa Clara, USA) connected to a flame ionization detector equipped with a capillary 381 column (SGE, BPX5, 60 m × 0,25 mm × 0,25 mm). The FAME concentrations were quantified relative 382 to methyl nonadecanoate (19:0), enabling methylated lipids to be identified. A standard soil was used 383 and extracted in parallel to detect potential deviations between the extraction rounds, expressed in nmol 384 C-FA per g of soil. Mono-unsaturated and cyclopropylated PLFA (C16:1w7c, C18:1w9c, and 385 C18:1w9t) were assigned to gram-negative bacteria, iso-and anteiso-branched PLFA (iC15:0, aC15:0, 386 iC16:0, i-C17:0, C:17, and C18:0) were assigned to gram-positive bacteria and C18:2w6c, C18:3w3c 387 respectively C20:5w3c were assigned to fungi 39 . The total content of bacteria was expressed by adding 388 gram-positive, gram-negative together with the markers C14:0, C16:0, C20:0, and C15:1. Lastly, the 389 13 C-labeling of FAME was concluded by correcting for the added methyl moieties during methanolysis 390 and relating it to the chain length of fatty acids 391 where δ 13 CF A represents the δ 13 C of the fatty acid, C n the number of C atoms in the fatty acid, δ 13 C FAME 393 is the δ 13 C of the fatty acid methyl ester, and δ 13 C MeOH is the δ 13 C of the methanol used for the 394 methylation (−63 ‰) to calculate the isotope ratios of the fatty acids. The relative incorporation of 13 C 395 into four microbial groups was calculated by relating the proportions of each fatty acid to the total 13 C 396 incorporation, and the absolute incorporation of 13 C in each microbial group was calculated by dividing 397 the amount of 13 C enriched fatty acid with the total amount of extracted fatty acid for that particular 398 group. 399

400
In order to gain insights on the microscale distribution of the assemblages of litter with microbes and 401 minerals, we used SEM and NanoSIMS. Free POM from non-fractionated soil was hand-picked and 402 fixed onto graphene sample substrates on metal stubs (10 mm in diameter). To avoid the charging 403 phenomena, samples were gold-coated prior to SEM analyses by physical vapor deposition under argon 404 atmosphere (Emitech Sputtercoater SC7620, Gala Instrumente, Bad Schwalbach, Germany). To analyze 405 the microscale structures of the assemblages of POM, microorganisms and soil minerals of the samples 406 were first analyzed using SEM (Jeol JSM 5900LV, Freising, Germany), and subsequently the spots that 407 best exemplified the microbial transformation on the decaying litter (POM) surface were analyzed using 408 a Cameca NanoSIMS 50 L (Cameca, Gennevilliers, France) 40 . For the NanoSIMS measurements, a 270-409 pA high primary beam was used to locally sputter away impurities and gold coating, and to implant 410 primary ions (Cs + ) into the samples surface (impact energy of 16 keV) to enhance the yields of secondary 411 ions. Subsequently, secondary ions were measured using electron multipliers; 12 C − , 13 C − , 12 C 14 N − to 412 display OM fragments and 16  The instrument was tuned to a high mass resolution in order to accurately separate mass isobars at mass 414 13 ( 13 C − , 12 C 1 H − ). The ion images were acquired with a 25 × 25 µm field of view, 40 planes and 1 ms 415 pixel −1 dwell time for all measurements. Charging effects were compensated for with an electron flood 416 gun if necessary. The acquired measurements were dead time (44 ns) and drift corrected using the 417 OpenMIMS plugin of the ImageJ software. The 13 C − :( 12 C − + 13 C − ) and 12 C 14 N − : 12 C − ratios were computed 418 for distinct regions of interests which were chosen manually with respect to the major compartments: 419 continuous fragments of fungal hyphae, individual bacteria, EPS patches as well as exposed POM 420 surfaces. To account for instrumental mass fractionation, the electron multipliers were carefully 421 checked, and the control measurements of non-labeled POM samples were conducted regularly along 422 the sessions. Here, the mean 13 C − :( 12 C − + 13 C − ) ratios were in line with the level of natural abundance, 423 which meant that a correction of ratios for labeled POM samples was not necessary. 424

Statistical analyses
425 All parameters were separately tested for normality with Shapiro-Wilk test and for homoscedasticity 426 with Bartlett's test. In addition, the distribution of the datasets was checked with Q-Q plots. In cases 427 where the assumptions of normality or homoscedasticity were not met, a log-transformation was applied 428 on the raw data and analyses were carried out on the log-transformed data. The differences caused by 429 texture and litter addition were tested using unpaired t tests, and depth differences were tested using 430 one-way analysis of variance with Tukey's honestly significant difference as the post-hoc test. In cases 431 where the log-transformed data did not meet the requirements for parametrical testing, the unpaired two-432 samples Wilcoxon test or Kruskal-Wallis test was applied. The statistical findings were considered 433 significant if the confidence limits were in excess of 95% (p < 0.05). All statistical testing was carried 434 out in the R statistical environment 41 using agricolae 42