Dark microbial CO2 fixation in temperate forest soils increases with CO2 concentration

Dark, that is, nonphototrophic, microbial CO2 fixation occurs in a large range of soils. However, it is still not known whether dark microbial CO2 fixation substantially contributes to the C balance of soils and what factors control this process. Therefore, the objective of this study was to quantitate dark microbial CO2 fixation in temperate forest soils, to determine the relationship between the soil CO2 concentration and dark microbial CO2 fixation, and to estimate the relative contribution of different microbial groups to dark CO2 fixation. For this purpose, we conducted a 13C‐CO2 labeling experiment. We found that the rates of dark microbial CO2 fixation were positively correlated with the CO2 concentration in all soils. Dark microbial CO2 fixation amounted to up to 320 µg C kg−1 soil day−1 in the Ah horizon. The fixation rates were 2.8–8.9 times higher in the Ah horizon than in the Bw1 horizon. Although the rates of dark microbial fixation were small compared to the respiration rate (1.2%–3.9% of the respiration rate), our findings suggest that organic matter formed by microorganisms from CO2 contributes to the soil organic matter pool, especially given that microbial detritus is more stable in soil than plant detritus. Phospholipid fatty acid analyses indicated that CO2 was mostly fixed by gram‐positive bacteria, and not by fungi. In conclusion, our study shows that the dark microbial CO2 fixation rate in temperate forest soils increases in periods of high CO2 concentrations, that dark microbial CO2 fixation is mostly accomplished by gram‐positive bacteria, and that dark microbial CO2 fixation contributes to the formation of soil organic matter.


|
). Yet the contribution of dark, that is, nonphototrophic, microbial CO 2 fixation in soils to the C cycle, and the factors that control dark microbial CO 2 fixation in soils are still not well understood.
It has been reported recently that dark CO 2 fixation rates in arctic tundra soils amounted to 0.4%-16.0% of the respiration rates (Šantrůčková et al., 2018). If dark microbial CO 2 fixation in soil reaches up to 16% of respiration, the process contributes substantially to C cycling in terrestrial ecosystems and needs to be taken into account for calculating soil C balances. However, currently it is difficult to estimate the importance of dark soil microbial CO 2 fixation for the C balance of soils, because there are only a few studies on microbial dark CO 2 fixation, and many of them analyzed indications of CO 2 fixation, such as the abundance of certain gene markers Liu et al., 2018;Yuan, Ge, Chen, O'Donnell, & Wu, 2012) or isotopic labeling of specific compounds (Ge et al., 2016), but most studies did not quantitate the proper fixation rates.
There are indications that dark soil microbial CO 2 fixation occurs in a large range of soils, including Mediterranean soils (Nel & Cramer, 2019), semiarid desert soils (Liu et al., 2018), arctic tundra soils (Šantrůčková et al., 2018), temperate peatland soils Nowak et al., 2015), paddy soils (Ge et al., 2016;Yuan et al., 2012), temperate meadow soils (Šantrůčková et al., 2005) as well as cropland soils (Miltner, Richnow, Kopinke, & Kästner, 2004). Thus, the process might occur ubiquitously in soils, and if quantitatively important, might affect the global C cycle. Yet, little is known about CO 2 fixation in temperate forest soils, and more importantly, proper CO 2 fixation rates have only been quantified in just a few soils so far Šantrůčková et al., 2018).
Microbial fixation of CO 2 in the absence of light can be accomplished by a wide variety of microorganisms using different metabolic pathways. First, chemoautotrophic bacteria and archaea use CO 2 as the only C source (Berg et al., 2010;Saini, Kapoor, Kumar, Siddiqi, & Kumar, 2011). Second, a wide spectrum of heterotrophic bacteria and fungi employ carboxylases to replenish the citric acid cycle through anaplerotic reactions (Erb, 2011;Kornberg & Krebs, 1957). Still, little is known about the contribution of different microbial groups to dark CO 2 fixation in soil. Šantrůčková et al. (2018) indicated that CO 2 in arctic tundra soils was mostly fixed by gram-negative bacteria and to a lesser extent by gram-positive bacteria and fungi. However, aside from the evidence provided by this study, hardly anything is known about the relative contribution of bacteria and fungi to dark CO 2 fixation in soils.
The factors that control dark microbial CO 2 fixation in soil are not completely known, but there is scattered evidence that the CO 2 concentration might affect dark microbial CO 2 fixation. The CO 2 concentration in soils is much higher than in the atmosphere, and ranges typically between 1% and 2%, but can get as high as 13% (Amundson & Davidson, 1990). Thus, the high CO 2 concentration in soil might facilitate dark microbial CO 2 fixation. Krebs (1941) reported that heterotrophic CO 2 -fixing propionic bacteria grew submerged in agar, where the CO 2 concentration was high, and only in experiments, in which the atmospheric CO 2 concentration was increased, they grew on the surface of the agar. More recently, Šantrůčková et al. (2005) reported that dark microbial CO 2 fixation was higher in alkaline than in acid soils, and the authors explained this by the higher HCO − 3 concentration in these soils. Aside from these two studies, nothing is known about the relationship between the CO 2 concentration and dark microbial CO 2 fixation, to our knowledge.
The objective of this study was (a) to quantitate dark microbial CO 2 fixation in three temperate forest soils that are typical for Central European forests; (b) to determine the relationship between the CO 2 concentration and dark microbial CO 2 fixation rate; and (c) to estimate the relative contributions of different microbial groups to dark CO 2 fixation. We hypothesized that (a) dark microbial CO 2 fixation in temperate forest soils contributes substantially to C cycling and the C balance of temperate forest soils; (b) the microbial CO 2 fixation rate increases with the CO 2 concentration in soil; and (c) CO 2 is mostly fixed by prokaryotes and not by fungi in these temperate forest soils. In order to test these hypotheses, we conducted an incubation experiment, in which we determined the dark microbial CO 2 fixation rate under four different CO 2 concentrations using 13 C-CO 2 . In addition, we estimated and visualized the contribution of different microbial groups to dark microbial CO 2 fixation based on phospholipid fatty acids (PLFAs) and nanoscale secondary ion mass spectrometry (NanoSIMS).

| Study sites, sampling, and sample preparation
We selected three forest sites for this study that are typical for Central Europe (Baritz, Seufert, Montanarella, & Ranst, 2010 at 1,023 m a.s.l. and hosts a mature F. sylvatica L. forest. The mean annual rainfall amounts to 1,299 mm and the mean annual temperature is 4.8°C. The parent material is paragneiss, and the soil type is a hyperdystric chromic Cambisol (FAO). The site Rothenkirch (R) is situated in the Franconian Forest (50°27′N, 11°21′E) at 670 m a.s.l. and hosts a mature Picea abies forest. The mean annual rainfall amounts to 1,070 mm and the mean annual temperature is 6.3°C. The parent material is greywacke, and the soil type is a podzolic Cambisol (FAO).
All three soils are acidic and carbonate free (Table 1). At each site, we sampled three depth increments of a soil profile, namely 0-5 cm (Ah horizon), 30-40 cm (Bw1 horizon) and 60-70 cm (Bw2 horizon) in September 2018. The soil samples were immediately transferred to the laboratory of the University of Bayreuth, where all soil samples were sieved (<2 mm) and roots were removed.

| Experimental design
The rates of CO 2 fixation and respiration were determined in an incubation experiment at the University of Bayreuth. For this purpose, 18.0 g of each sample was weighed into incubation jars in eight replications for the Ah and Bw2 horizons, and in 20 replications for the Bw1 horizons. The incubation jars had a volume of 1,057 ml and were equipped with a septum that allowed for removal of gas samples with a syringe. Incubation jars with a large headspace volume with respect to the soil mass were chosen because we wanted the CO 2 concentration and its isotopic signature to change as little as possible during incubation. The water content of all samples was adjusted to 60% of their field capacity. Subsequently, we preincubated the jars at 15°C in a dark incubation chamber for 10 days. After preincubation, all jars were re-opened for several minutes to allow the CO 2 concentration to equilibrate with the ambient CO 2 concentration. Four replicates of each soil horizon were kept with air as 13 C natural abundance (n.a.) control treatment. In the other jars, the CO 2 concentration was adjusted using 99.99 at% 13 C-labeled CO 2 (Sigma-Aldrich). For this purpose, air from the closed jar was removed using a gas-tight syringe, and the same volume was replaced by pure 13 C-labeled CO 2 .
Four replicates of all horizons were incubated under an atmosphere of 2.50 (v/v) % CO 2 . In addition, the Bw1 horizons of the three forest soils were incubated at 1.00, 4.70, and 6.10 (v/v) % CO 2 , each in four replications. The reason for choosing the Bw1 horizon for the experiment with the CO 2 gradient is twofold; first, the CO 2 concentration tends to be higher in the subsoil than in the Ah horizon in many soils, second, the Bw1 horizon has a higher microbial biomass and microbial activity than the Bw2 horizon. After the CO 2 concentration was adjusted, all incubation jars were incubated at 15°C in the dark for 165 hr (7 days). At the very end of the incubation experiment, the jars were opened. The soil was divided, and one part of each soil sample was immediately used for the determination of microbial biomass carbon (MBC), one part was freeze dried and ground using a ball mill (Retsch GmbH), one part was frozen at −14°C for subsequent DNA and PLFA extraction, and one part was chemically fixed for NanoSIMS measurements.

| Respiration, microbial biomass, and DNA
The (net) respiration rate was determined based on measurements of the CO 2 concentrations in the glass jars performed at the beginning and at the end of the incubation experiment using a gas chromatograph (SRI 8610C; SRI Instruments Europe GmbH) equipped with a flame ionization detector. Gas samples were taken from the headspace of the jars using a syringe and then injected directly into the gas chromatograph.
In order to determine the MBC and its delta 13 C signature, we used the chloroform-fumigation-extraction method (Vance, Brookes, & Jenkinson, 1987). For this purpose, each soil sample was divided into two parts. One part was fumigated for 24 hr with chloroform in a desiccator before being extracted with 0.5 M K 2 SO 4 , while the other part was directly extracted in 0.5 M K 2 SO 4 . The dissolved C concentration of the extracts was determined using a total organic carbon (TOC)/TN analyzer (Multi N/C 2100S; Analytik Jena AG). The total MBC was calculated using the conversion factor of 2.22 (Joergensen, 1996). All extracts were freeze dried for isotope analysis.
DNA was extracted from 400 mg of moist soil using a DNA extraction kit (FastDNA™ SPIN Kit for Soil; MP Biomedicals) with small modifications as in Spohn, Pötsch, et al. (2016). The DNA extract was dried in silver capsules at 60°C overnight before isotope analysis.

| Isotope analyses
The δ 13 C signature of the TOC, of the MBC, and of the DNA of each soil sample was determined at the Center for Stable Isotopes (KOSI) at the University of Göttingen. Samples were analyzed on a Delta V isotope ratio mass spectrometer (Thermo Fisher) interfaced to a high temperature conversion elemental analyzer Euro EA 3000 (EuroVector S.p.A).

| Nanoscale secondary ion mass spectrometry
Bulk soil samples were chemically fixed using Karnovsky fixative (Karnovsky, 1965)  used open multi-isotope imaging mass spectrometry plugin in ImageJ and computed the 12 C 13 C − / 12 C − 2 ratio, which is doubled compared to 13 C − / 12 C − . The n.a. at the instrument settings used translates into a ratio of 220 on the color scale used in the figures (Figure 4c,f).

| Calculations and statistical analyses
The isotope ratio of the MBC was calculated based on the extracts of the fumigated and the nonfumigated soil sample as follows: The amount of C that was fixed in the soil and in the MBC pool was calculated from the isotope ratio of the labeled sample and the n.a.
sample, as follows: where C is the C concentration of the respective pool. The 12 C/ 13 C ratio was obtained from the δ 13 C based on the 13 C/ 12 C ratio of the V-PDB standard that amounts to 0.0111802 as follows: The C fixation rates for the total soil pool and the MBC pool were cal- (2) (3) 13 C 12 C = δ 13 C 1,000 + 1 × 0.0111802.
Kruskal test followed by pairwise Wilcoxon rank sum test. In all tests, differences with a p < .05 were considered as statistically significant. All analyses were conducted using R version 3.4.0 (R Core Team, 2013).

| RE SULTS
The rates of dark microbial CO 2 fixation were highest in the Ah horizons, and reached up to 320 µg CO 2 -C kg −1 soil day −1 in the Ah horizon of soil M ( Figure 1A). The fixation rates were 5.8, 8.9, and 2.8 times higher in the Ah horizon than in the Bw1 horizon of soils BB, M, and R, respectively ( Figure 1A). Normalized on the soil TOC concentration, the dark microbial CO 2 fixation rates ranged between 1.1 and 5.3 µg CO 2 -C g −1 TOC day −1 . The fixation rates in the MBC pool per unit MBC were significantly higher in the Bw1 horizon of soil R than in the Bw1 horizon of the other two soils ( Figure 1B). In addition, the fixation rates in the MBC pool per unit MBC were also significantly higher in the Bw2 horizon of soil R than in the Bw2 horizon of the other two soils.
The rates of dark microbial CO 2 fixation were positively correlated with the CO 2 concentration in the Bw1 horizon in all three soils (R 2 = .79, .87, and .87 in soil BB, M, and R, respectively, all p < .001; Figure 2a). The CO 2 fixation rates per unit soil increased by a factor of 1.74, 1.73, and 1.73 in the soil BB, M, and R, respectively, when the CO 2 concentration was increased from 1.0% to 6.1% CO 2 (Figure 2a). Similarly, the fixation rates in the MBC pool were also positively correlated with the CO 2 concentration in all soils ( Figure 2b).
The MBC concentration was 5.8-8.3 times higher, and respiration rates were 6.7-11.5 times higher in the Ah horizon than in the Bw1 horizon in the three soils ( Table 2). The respiration rates of the Bw1 horizons of all three soils were very similar ( Table 2).
Rates of CO 2 fixation amounted to 1.2%-3.9% of the respiration rates ( Table 2). The percentage of the fixation rate on the F I G U R E 1 Dark microbial CO 2 fixation in three horizons of the forest soils Bad Brueckenau (BB), Mitterfels (M), and Rothenkirch (R; a) in the soil and (b) in the soil microbial biomass carbon (MBC) pool determined in an atmosphere with 2.5% CO 2 . Columns depict means ± SDs (n = 4). Different lowercase letters indicate significant (p < .05) differences tested separately for the Ah, Bw1, and Bw2 horizons across all three soils by ANOVA followed by Tukey's test  respiration rate tended to increase with soil depth in soil BB and R ( Table 2).
The DNA pool was only very weakly enriched with 13 C in all three forest soils independently of the CO 2 concentration (Table 3). The δ 13 C enrichment of the DNA pool was higher in the Ah horizon than in the two Bw horizons in all three soils (Table 3). The δ 13 C enrichment of the PLFAs showed a consistent pattern across all three soils.
The δ 13 C enrichment was significantly highest in PLFA 18:1ω9c in all three soils, reaching up to 133‰ δ 13 C (Figure 3). The PLFAs i15:0, a15:0, i16:0, and i17:0 that are produced by gram-positive bacteria were also enriched in 13 C in all three soils. The PLFA 18:2ω6,9 which is produced by fungi showed only a very low 13 C enrichment in all three soils, between 0.4‰ and 3.0‰ δ 13 C. The PLFA cy19:0 that is produced by gram-negative bacteria was more weakly enriched than the PLFAs i15:0 and i17:0 that are produced by gram-positive bacteria. The 13 C enrichment in 18:1ω9c as well as in i15:0, a15:0, i16:0, and i17:0 increased from soil BB, to soil M, and was highest in soil R ( Figure 3). The PLFA 18:1ω9c, which was most strongly enriched in 13 C, was significantly more abundant in soil R than in the other two soils (Table S1).
The NanoSIMS measurements revealed 13 C-enriched micropatches, presumably bacterial cells, in the soils that had a size of about 585 nm × 585 nm (Figure 4). The 13 C-enriched cells were associated with fungal hyphae (Figure 4b), although the hyphae themselves were not enriched in 13 C ( Figure 4c). Furthermore, we observed that 13 C-enriched micropatches were associated with particulate organic matter (Figure 4e). F I G U R E 3 δ 13 C enrichment of nine phospholipid fatty acids in the Bw1 horizon of the soils Bad Brueckenau (BB), Mitterfels (M), and Rothenkirch (R) after labeling with 13 C-CO 2 in an atmosphere with 6.1% CO 2 in relation to the natural abundance samples.
Columns depict means ± SDs (n = 4). Different lowercase letters indicate significant (p < .05) differences tested separately for the three soils by Kruskal test followed by pairwise Wilcoxon rank sum test

| D ISCUSS I ON
We found that the rate of dark CO 2 fixation was positively correlated with the CO 2 concentration in soils (Figure 2), which has never been described before, to our knowledge. The CO 2 concentration in soils is highly dynamic and can increase to up to 13% of the soil air (Amundson & Davidson, 1990). In this light, our results indicate that more CO 2 is fixed in soils in periods of high soil CO 2 concentration and in soil microsites that have a high CO 2 concentration. The linear relationship between the CO 2 concentration and the dark microbial CO 2 fixation rate indicates that the process is controlled by the partial pressure of CO 2 . Reactions that lead to the fixation of CO 2 are carboxylation reactions that are catalyzed by carboxylases in microorganisms.
Carboxylases catalyze carboxylation as well as decarboxylation of organic compounds and the equilibrium of the reaction depends on the concentrations of all compounds involved in the reaction, according to Le Chatelier's principle. An increase in the CO 2 concentration moves the equilibrium of the reaction toward the product of the carboxylation reaction, and thus to an increase in CO 2 fixation. Our findings show that the carboxylation reaction in soil microorganisms is very sensitive to changes in the CO 2 concentration. We found that the CO 2 fixation rate and the CO 2 concentration were linearly correlated in the concentration range between 1.0% and 6.1% CO 2 . It could be that at a higher CO 2 concentration, a saturation is reached and the fixation rate increases less with the CO 2 concentration than below a CO 2 concentration of 6.1%. However, CO 2 concentrations beyond 6.1% do likely not occur very frequently in many soils (Amundson & Davidson, 1990).
Our results suggest that in periods, in which large amounts of CO 2 are respired, a larger percentage of CO 2 is fixed by soil microorganisms, leading to a negative feedback of the CO 2 concentration on soil net CO 2 emission. However, it has to be taken into account that the F I G U R E 4 Image analyses of locations of dark soil microbial CO 2 fixation, including micrographs gained by scanning electron microscope with red squares indicating area measured using NanoSIMS (a and d), composite images derived from NanoSIMS measurements showing organic matter ( 12 C 14 N − ) and mineral particles ( 16 O − ; 56 Fe 16 O − ; b and e), and hue-saturation-intensity (HSI) maps of the ratio of 12 C 13 C − / 12 C − 2 , where the black line in the color bar indicates the ratio of 12 C 13 C/ 12 C 2 at natural abundance (c and f). Upper panel: (a) Microaggregates and a fungal hyphae in soil R, horizon Bw1, after labeling with 13 C-CO 2 in an atmosphere with 6.1% CO 2 , (b) organic matter of the fungal hyphae surrounded by mineral particles, and (c) spots of isotopic enrichment attached to the fungal hyphae (red arrows). Lower panel: (d) Microaggregate with occluded particulate organic matter (oPOM) in soil M, horizon Bw1, after labeling with 13 C-CO 2 in an atmosphere with 6.1% CO 2 , (e) organic carbon of the oPOM with adhering mineral particles, and (f) spots of isotopic enrichment attached to oPOM (red arrows) rates of dark microbial CO 2 fixation were small and only amounted to less than 4% of the net respiration rates. Hence, the negative feedback that the CO 2 concentration exerts on the overall soil CO 2 emission due to its positive effect on dark microbial CO 2 fixation is small.
Although the rates of dark microbial fixation in the temperate forest soils studied here were small compared to the respiration rates, our findings indicate that organic matter formed by microorganisms from CO 2 contributes to the soil organic matter pool. Dead microbial biomass (microbial necromass) is a very important source of soil organic matter since it is relatively stable in soil compared to plant detritus (Liang, Amelung, Lehmann, & Kästner, 2019;Miltner, Bombach, Schmidt-Brücken, & Kästner, 2012;Schimel & Schaeffer, 2012). Thus, C that enters the soil through the microbial biomass pool likely remains much longer in soil than C that enters the soil as plant detritus and mostly leaves the soil again in the form of CO 2 after only a few years (Sierra, Hoyt, He, & Trumbore, 2018). Hence, dark microbial CO 2 fixation might substantially contribute to the formation of soil organic matter in temperate forests. Furthermore, the NanoSIMS images indicate that labeled bacteria were associated with organic matter in microaggregates, which are assumed to be relatively stable in soil (Figure 4 lower panel). This finding provides additional evidence that CO 2 that is fixed in soils by bacteria directly enters relatively stable pools of organic matter in soil. Our results are in accordance with the emerging understanding that the microbial biomass is an important contributor of organic matter in soils (Kallenbach, Frey, & Grandy, 2016;Schimel & Schaeffer, 2012;Spohn, Klaus, Wanek, & Richter, 2016). To what extent dark microbial C fixation contributes to the formation of soil organic matter that remains in soil for a long period of time depends ultimately on the persistence of microbial necromass in soil. Thus, a full evaluation of the contribution of dark microbial CO 2 fixation to the formation of stable organic C in soil would require detailed information on the turnover of microbial necromass in soil.
We found that the dark microbial CO 2 fixation rates in relationship with the respiration rate were in a similar range as in  who reported that dark microbial CO 2 fixation rates in temperate cropland soils amounted to 0.3%-1.1% of the respiration rate. Also the absolute rates in  were similar to our study and amounted to 420 µg C kg −1 soil day −1 . The rates found here are also in accordance with Nowak et al. (2015) who reported that dark microbial fixation rates in organic grassland soils on natural CO 2 vents amounted to 20-400 µg C kg −1 soil day −1 . However, the rates of dark microbial CO 2 fixation found here are much smaller than the rates reported by Šantrůčková et al. (2018) for arctic permafrost soils. They found that the CO 2 fixation rate across all studied soils amounted to up to 14.6 µg C g −1 TOC day −1 (i.e. up to 16% of the respiration rate).
The reason for the much larger rates of dark microbial CO 2 fixation measured by Šantrůčková et al. (2018) might be the very special conditions in arctic permafrost soils. Our data and the data of similar previous studies indicate that C flux calculations based on natural C isotope signatures may be misleading since microbial fixation of CO 2 that has the isotopic signature of atmospheric CO 2 strongly increases the δ 13 C value of soil OC.
Microbial dark CO 2 fixation was higher in the Ah horizon than in the Bw1 horizon of all three soils ( Figure 1A). This finding is in accordance with Nowak et al. (2015) and Ge et al. (2016), showing that dark microbial CO 2 fixation was highest in the topsoils of grasslands on natural CO 2 vents and in the top of paddy soils. The reason for this might be the high concentrations of microbial biomass (Table 2) and the high concentrations of TOC (Table 1) in the Ah horizons. The latter is supported by the NanoSIMS results, revealing that microbial cells that fixed CO 2 were associated with organic matter (Figure 4).
The interpretation that microbial CO 2 fixation depends on organic C is in accordance with Šantrůčková et al. (2018) and , showing that the dark CO 2 fixation rates increased due to organic matter addition to soils. The observation that dark CO 2 fixation is associated with the presence of organic C might suggest that dark microbial CO 2 fixation is mostly catalyzed by heterotrophic microorganisms. However, it could also be that dark microbial fixation is high in soils in the presence of large amounts of labile organic, because large amounts of labile organic matter are usually associated with high CO 2 production in soil. In the present experiment, the availability of organic C for microorganisms was low since root exudation and leaching of dissolved organic C from the organic layer into the mineral soil were excluded. If dark microbial C fixation is carried out mostly by heterotrophic bacteria, as discussed above, it can be expected that the fixation rates are higher if organic C availability is increased. In addition, it could also be that some microbial groups, including gram-negative ones, that are little active under starvation conditions, become more active if more organic C is available.
The PLFA analysis revealed that CO 2 was fixed by the same group of microorganisms in all three soils although the soils and the study sites differed in many properties including parent material, dominant tree species, soil texture, soil TOC content (Table 1), and PLFA abundance (Table S1). Our results indicate that CO 2 was mainly fixed by gram-positive bacteria and only to a much smaller extent by fungi and gram-negative bacteria (Figure 3). The dominant role of bacteria over fungi in dark CO 2 fixation is in accordance with Šantrůčková et al. (2018). However, in contrast to Šantrůčková et al. (2018), we found only a small enrichment of 13 C in the PLFA cy19:0 that is produced by gram-negative bacteria but relatively high enrichments in PLFAs that are produced by gram-positive bacteria. This difference between Šantrůčková et al. (2018) and our study can likely be attributed to the difference in soils, since they investigated arctic permafrost soils that host a different microbial community than the temperate forest soils studied here. However, both studies agree in the finding of a dominant role of bacteria over fungi in dark CO 2 fixation. The NanoSIMS measurements confirmed that dark microbial CO 2 fixation was accomplished by bacteria and not by fungi. In addition, we observed that CO 2 -fixing bacteria were closely associated with fungal hyphae (Figure 4 upper panel).
One might speculate that CO 2 -fixing microorganisms grow preferentially on fungal hyphae because they benefit from the CO 2 respired by the fungus. This speculation is supported by the finding of Krebs (1941) that CO 2 -fixing propionic bacteria grow preferentially at microsites with high CO 2 concentration. However, there might be other reasons why CO 2 -fixing prokaryotes grew preferentially on fungal hyphae; they might, for example, also benefit from water being transported along the hyphae (Guhr, Borken, Spohn, & Matzner, 2015).
We found that the DNA was much less enriched with 13 C than most of the PLFAs. The reason for this might be, first, that the four nucleotides that form DNA are not completely decomposed and newly synthesized in the cell but are (at least partly) re-used, in contrast to PLFAs that are newly synthesized (Nelson, Cox, & Häcker, 2009). Second, CO 2 fixed through anaplerotic reactions enters the microbial cell through the citric acid cycle and is incorporated into citrate, which is an important precursor of fatty acids, while the pentose that contains a large part of the C in DNA is not built up from compounds of the citric acid cycle in heterotrophic organisms (Nelson et al., 2009). Our results are in accordance with Miltner et al. (2004), observing that amino sugars were much less enriched with 13 C than amino acids after incubating soil from a cropland with 13 C-CO 2 . The findings that sugar-containing compounds are less enriched in CO 2 -derived C indicate that dark microbial CO 2 fixation is mostly accomplished by heterotrophic microorganisms.
Significantly more CO 2 was fixed in the microbial biomass in the Bw1 horizon of soil R than in the Bw1 horizon of the other two soils ( Figures 1B and 2b) and also some PLFAs were more strongly enriched with 13 C in soil R than in the other two soils (Figure 3). This is especially noteworthy given that the respiration rates of the three Bw1 horizons were very similar (Table 2), indicating that the overall microbial activity did not differ between the three soil horizons.
The reason for the significantly higher CO 2 fixation rate in the upper subsoil of the spruce forest soil (horizon Bw1 of soil R) might be a different microbial community that fixes more CO 2 compared to the corresponding horizon in the two beech forest soils. This is supported by the finding that the PLFA 18:1ω9c, which was most strongly enriched in 13 C (Figure 3) was significantly more abundant in soil R than in the other two soils (Table S1). The difference in the microbial community might be caused by the difference in forest type (beech forest at sites BB and M, and spruce forest at site R) since tree species affect the microbial community composition in temperate forest soils (Hackl, Pfeffer, Donat, Bachmann, & Zechmeister-Boltenstern, 2005).
In conclusion, our study shows that dark microbial CO 2 fixation rates increase linearly with the soil CO 2 concentration, indicating that more CO 2 is fixed in periods or microsites with high soil CO 2 concentrations. In addition, our findings show that dark CO 2 fixation in temperate forest soils is mostly accomplished by gram-positive bacteria and not by fungi. Although the rates of dark microbial fixation were small compared to the respiration rates, our findings suggest that organic matter formed by microorganisms from CO 2 contributes to the soil organic matter pool, especially given that microbial detritus is more stable in soil than plant detritus.

ACK N OWLED G EM ENTS
M.S. thanks Renate Krauß for technical assistance in the lab, Gerhard Müller for assistance in the field, and Per-Marten Schleuss for help with the gas valve. We thank the Center for Stable Isotopes