Origin of litters
For experiments 1–3, we collected litter of four moss species, Sphagnum magellanicum, Sphagnum majus, Leucobryum glaucum and Polytrichum strictum. The moss litter was defined as recently dead (brown), usually 2 cm long shoot segments, which did not show signs of structural breakdown due to decomposition. S. magellanicum, S. majus and P. strictum were collected in a mountain ombrotrophic bog Rokytecká slať, Šumava National Park and Biosphere Reserve, Czech Republic (49°1′N, 13°25′E). S. magellanicum dominated in drier habitats such as lawns and low hummocks while S. majus occupied exclusively wet carpets. P. strictum dominated the highest hummocks, which were already too dry for any Sphagnum species. Leucobryum glaucum formed typical cushions in the understory of nutrient-poor Scotch pine forest near Třeboň, Třeboňsko Protected Landscape Area and Biosphere Reserve, Czech Republic (48°59′N, 14°50′E). In the laboratory, we removed remaining living and decomposing tissues and other coarse plant remnants from the litter and washed all the litter with distilled water for about 15 min to remove fine debris. The litter was dried at 25°C and relative air humidity of 40%. Before it was fully dry, it was briefly homogenized using a kitchen mixer to obtain particles <5 mm long. This material, referred to as fresh litter, was poor in soluble phenolics (<1.0 mg of tannic acid standard per g of dry mass), as estimated using the Folin-Ciocalteu assay (Bärlocher and Graça 2005).
For experiment 4, recently senesced (brown) leaves of Eriophorum angustifolium were collected from Clara Bog, Ireland (53°51′N, 6°27′W). We choose this litter type for its common occurrence in peatlands and its relatively high decay rate (Limpens and Berendse 2003), likely ensuring measurable effect sizes. The leaves were dried at 70°C and ground to approximately 2-mm fragments. C:N of 48 indicated nutrient-poor litter quality.
Microbial inoculum
We inoculated litters with a suspension of a natural microbial community. We made a leachate from S. magellanicum litter collected above the water table from a bog.
For experiments 1–3, 1 kg of fresh litter was shaken in 3 l of de-ionized water for 1 h. The slurry was filtered through a fine polyamide mesh and coarse filter paper. Based on pre-experimental tests we added mineral nutrients to avoid nutrient limitation of microbial C mineralization rate (concentrations in mg l−1 are parenthesized): NH4Cl (1000), K2HPO4 (1000), MgSO4·7H2O (200), CaCl2·2H2O (20), FeCl3·6H2O (10), MnSO4·H2O (4), CuSO4·5H2O (1). For experiment 2, we used also doubled concentrations of salts of multivalent cations Mg2+, Ca2+, Fe3+, and Mn2+. The latter was done to avoid potential limitation of the decomposer community by essential metallic cations complexed preferably on carboxyl groups of the chlorite-treated then carboxyl-reduced litter (Thomas and Pearce 2004). The initial ionic strength of the inoculum was 31–35 mM, which is still favourable for sphagnan-protein complexation (Ballance et al. 2008).
For experiment 4, the litter of S. magellanicum was inserted into tubes (each 5 g fresh-weight) together with 30 ml of artificial rainwater (Garrels and Christ 1965) and 20 glass beads to facilitate dislodging of micro-organisms from the fibrous Sphagnum peat (Hopkins et al. 1991). The tubes were shaken at 5°C for 2 h and centrifuged at 500×g
n for 2 min and the supernatant collected. This procedure was repeated two times after re-suspending the residue with fresh rain water. To concentrate the micro-organisms collected with the water (c. 1120 ml in total), approximately half of the water was drained overnight using submerged Teflon-coated Rhizon soil moisture samplers (Eijkelkamp Agrisearch Equipment, Giesbeek, the Netherlands) at 5°C.
Experiment 1
To separate the effects of cell-wall polysaccharides and phenolics on microbial respiration, litters of four moss species: S. magellanicum, S. majus, L. glaucum and P. strictum were subjected to three different chemical pre-treatments to selectively remove a number of compounds. This resulted in three litter forms: (i) fresh, i.e., untreated dried litter, (ii) chlorite-treated and (iii) borohydride-treated which was incubated for 64 days, during which respiration was followed. In total there were 4 species × 3 treatments × 3 replicates = 36 samples. The pH was set to 3.5 in all samples using HCl.
Chlorite treatment involved removing the majority of aromatic polymers bound in cell walls. To this end fresh litter was boiled with acetone:methanol (2:1 v/v) for 6 times 5 min and bleached with sodium chlorite (Ballance et al. 2007). We thus obtained papery-white material which is assumed to be free of lignin-like phenolic polymers. This and subsequent treatments required careful manipulation with L. glaucum and particularly P. strictum because the bleached leaves became very fragile; Sphagnum structure was not noticeably affected. The C:N of fresh litter was about 65–90, and increased to about 110–270 after the chlorite treatment (lower values: S. magellanicum and P. strictum, higher values: S. majus and L. glaucum). We assume that adding 2.05 mg of NH4+–N per tube minimized potential effects of the initial litter N content (0.11–0.25 mg of N per tube) on C mineralization.
The chlorite treatment is commonly used for holocellulose quantification in plant tissues. It removes most of the lignin and other phenolics, as analyzed by optical spectrometry and confirmed by pyrolysis mass spectrometry (Morrison and Mulder 1995), and has a low impact on carbohydrate composition (Reeves 1993).
For the borohydride treatment, chlorite-bleached litter was further treated with sodium borohydride in order to reduce any carbonyl groups according to the standard procedure described in Ballance et al. (2007). The final borohydride-treated litter is assumed to be free of carbonyl groups. The litter C:N increased to 366–825 after borohydride treatment.
About 70.0 mg (oven-dry weight) of fresh or treated litter was inserted in 12-ml Exetainer glass tubes with screw septum cap. Samples were inoculated with 6 ml of microbial inocula and tubes were placed into an incubator at 15°C and relative humidity (RH) of >95% to minimize water evaporation from the uncapped tubes. 12 h prior to respiration measurements, the uncapped tubes were thoroughly vortexed, closed with a septum cap, and placed in the incubator horizontally to enhance gas exchange between the suspension and air. After the incubation period of 12 h, the capped tubes were vortexed and a 200-μl air sample was taken with a gas-tight syringe to analyze the CO2 concentration on an Agilent 6850 gas chromatograph (Agilent Technologies, USA). During the incubation period 11–14 samplings were made; the CO2 production was measured every second day in the beginning up to about once per 12 days at the end. Samples were vortexed at least every 3rd day to avoid anoxic conditions.
Measured CO2 concentrations were converted to CO2 production per initial C content per hour and these values were integrated in cumulative C loss for the whole 64-day period of the experiment. After the last measurement (64 days), the samples were lyophilized and weighted to obtain actual C loss. As the starting CO2 concentration in the tubes was not measured directly after capping, this cumulative C loss is an overestimation of the actual C loss and needed to be proportionally adjusted based on the actual C loss (using Goal Seek function in MS Excel). Changes in relative C content of the samples during incubation were neglected in the calculations.
Experiment 2
To estimate the effect of cell-wall carboxyls on microbial respiration, the litters of two moss species, S. magellanicum and L. glaucum, were subjected to two chemical pre-treatments: (i) chlorite-treatment and (ii) carboxyl-reduction, amended with two concentrations of divalent metal ions (standard and doubled) in the inoculum, and incubated for 69 days during which respiration was followed. In total there were 2 species × 4 treatments × 3 replicates = 24 samples. pH was set to 3.5 in all samples using HCl.
Doubling the concentration of divalent cations did not affect C mineralization (Table 2), suggesting that the microbial community was not limited by availability of essential divalent cations and that the addition of cations (ionic strength of 4 mM) did not affect potential electrostatic interactions between litter carboxyls and microbial enzymes.
For details on chlorite treatment see experiment 1. Carboxyl-reduction involved the following procedure. We adapted the method based on the reduction of the carbodiimide-activated carboxyls (Stenutz et al. 2004). 1.2 g of chlorite-treated litter were suspended in 250 ml of water in a 2-l bottle and 50 ml of 0.2 M 2-(4-Morpholino-) ethanesulfonic acid, pH 4.75, were added, followed by 12 g of new N-(3-Dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (carbodiimide). After 4 h of stirring at room temperature, 250 ml of 2 M TRIS pH 9 were added and followed by a few drops of octanol to prevent foaming. After the sample had cooled in an ice bath, 250 ml of 17.5 g of NaBH4 in 0.05 M NaOH was added and the suspension was kept overnight at 4°C and then slowly neutralized to pH 6 by addition of glacial acetic acid. Litter was collected by vacuum filtration, washed in 0.5 M NaCl and then in 20 mM HCl to convert the carboxyls into their H+ form, washed in three changes of a large volume (8 l) of distilled water and once with acetone (500 ml). Finally, the litter was dried at room temperature and ready for incubation which followed the same procedure as for experiment 1.
To test the effectiveness of our carboxyl-reduction treatment, we determined the uronic acid (UA) content before and after carboxyl reduction using the assay of Ahmed and Labavitch (1978). Samples were mill-ground before analysis. The original contents of UA were 27.5 and 21.9% in S. magellanicum and L. glaucum, respectively. It was reduced about 40% in S. magellanicum and 70% in L. glaucum. The C:N of 66 (S. magellanicum) and 98 (L. glaucum) in carboxyl-reduced litter suggests that some carboxyls were successfully activated by carbodiimide but have not been reduced by borohydride, as indicated by the lower C:N and smaller carboxyl reduction in S. magellanicum.
Experiment 3
To evaluate the effect of pH on microbial respiration of cell-wall polysaccharides, chlorite-treated litter of two moss species, S. majus and L.glaucum, was incubated at three levels of pH (3.5, 5.0 and 6.5) for 68 days during which respiration was followed. In total there were 2 species × 3 treatments × 3 replicates = 18 samples. The pH of the suspension was adjusted to 3.5, 5.0 or 6.5 by addition of either HCl or NaOH 6 h after inoculation and readjusted every 3–10 days. After each pH adjustment, a minimum calculated amount of NaCl was added to the samples to ensure equal ionic strength of the solution among all samples (up to 15 mM of Na+). About 50% of carboxyls are dissociated at pH 3.5 and about 97% at pH 5.0.
For details on the chlorite treatment or respiration measurements, see experiment 1.
Experiment 4
To elucidate the effect of free soluble sphagnan on microbial respiration, Eriophorum angustifolium litter was amended with two polysaccharides: sphagnan (prepared from chlorite-treated litter of S. papillosum after Ballance et al. 2008) and polygalacturonic acid (PGA, Sigma P-3850, 79% of galacturonic acid) in five concentrations (0, 5, 10, 20, and 50 mg g−1 l; dry weight equivalent) and incubated for 30 days. Sphagnan and PGA were added in the same concentrations, as both compounds have a similar C-content per monomer. In total there were: 10 treatments × 3 replicates = 30 samples. After the incubation cumulative C mineralization and changes in dissolved inorganic (DIN) and organic nitrogen (DON) fractions were measured.
The ground litter of E. angustifolium was mixed with quartz sand (Merck) and inoculum in a weight-ratio of 1:3:5. This ratio was found to be optimal for microbial respiration in a previously conducted pilot experiment. The substrate was then covered and pre-incubated at 20°C for 1 week, during which the microbial community could build up and stabilize (Kraus et al. 2004). Hereafter the substrate was mixed thoroughly and portions of 15 g were divided over 35 plastic containers (4.5 cm in diameter, height of 5.0 cm). Five containers with substrate were kept apart for chemical analyses. To compensate for the N lost during the pre-incubation period, 0.23 mg of N was added as NH4NO3 to each of 30 containers. The added amount was based on the loss of mineral N during the pre-incubation period of the pilot experiment. After this, freshly prepared solutions of sphagnan or PGA (pH 4.0, adjusted with 0.1 M NaOH) were mixed through the substrates, and the whole containers were placed in 0.5 l airtight mason jars and fitted with a septum-covered outlet. A thin layer of de-ionized water was added to the jars to keep RH constant. The jars were placed in an incubator under 20°C for 30 days.
It was not possible to completely dissolve PGA to prepare a 50 mg g−1 solution so this treatment was omitted from the experiment. The pH (KCl) at the end of the experiment ranged between 3.7 for the lowest concentration and 4.3 or 4.2 for the highest concentrations of sphagnan or PGA.
Chemical analyses
The initial C and N content of all the moss species and their litter forms (experiment 1–3) was determined by an Elemental analyzer (ThermoQuest Italia, Milan, Italy) in two replicates (coefficient of variation <8%). To be aware of possible N-limitation of C mineralization in experiments 2 and 3, we centrifuged the tubes at 5000×g
n for 10 min at the end of incubation and used flow injection analysis to determine the concentration of available NH4
+ and NO3
− in the supernatant. The concentration of NO3
− was negligible in all samples so NH4
+ represented the available DIN. No losses of DIN are expected through denitrification due to aerobic experimental conditions.
In experiment 4, we measured cumulative C mineralization as well as the changes in the DIN and DON. Soil samples were taken after pre-incubation before adding extra N (five containers), as well as at the end of the experiment (all containers). Prior to sampling, the substrate was mixed thoroughly in the containers and sub-samples were taken to determine water and ash contents, C:N, total dissolved nitrogen (TDN), nitrate nitrogen (N–NO3) and ammonium nitrogen (N–NH4). Water content was determined after drying at 105°C and ash content by ignition at 550°C. The substrate was ground in a ball-mill and its C content was determined on an elemental analyzer (Fisons Instruments EA 1108, Milan Italy). To determine the available N fractions, 8 g of fresh substrate was extracted with 20 ml 1 M KCl by shaking for 2 h. N–NO3, and N–NH4 were measured colorimetrically using a continuous flow analyser (SKALAR SAN plus system, the Netherlands). TDN was measured as N–NH4 after Kjeldahl digestion.
We express C loss per initial C content in %. The amount of extra C added to the substrate as sphagnan or PGA was included in these calculations. DIN represents the sum of N–NO3 and N–NH4 contents, DON = TDN − DIN.
Statistical analyses
Factorial general linear models (GLM) ANOVA were used to test effects of factors (species, treatment) on cumulative respiration at the end of incubation periods (experiments 2–3). Repeated measures (RM) GLM ANOVA was used to test factor effects on the course of respiration rates during incubation periods for all experiments. We employed nested design in case of experiment 1 to test the effect of groups (Sphagnum and non-Sphagnum), which both include two different species. Differences between means within factors were determined with Tukey’s honestly significant difference (HSD).
In experiment 4, the highest sphagnan concentration (50 mg g−1 in which PGA did not dissolve) was not included in the RM ANOVA to keep the design balanced. The effect of adding the highest concentration of sphagnan was tested against the lowest concentration (no sphagnan added) using a two-tailed Student’s t-test. The effect of C-mineralization rate on TDN and DIN was tested with an ANCOVA with substance as factor and concentration and C-mineralization as co-variables. C-mineralization rate had no effect on either TDN, DIN or DIN:DON ratio.
We used STATISTICA v. 8.0 software package. Statements given in Result section represent statistically significant differences (α = 0.05) unless otherwise indicated.