Fungal-bacterial dynamics and their contribution to terrigenous carbon turnover in relation to organic matter quality.

Ecological functions of fungal and bacterial decomposers vary with environmental conditions. However, the response of these decomposers to particulate organic matter (POM) quality, which varies widely in aquatic ecosystems, remains poorly understood. Here we investigated how POM pools of substrates of different qualities determine the relative contributions of aquatic fungi and bacteria to terrigenous carbon (C) turnover. To this end, surface sediments were incubated with different POM pools of algae and/or leaf litter. 13C stable-isotope measurements of C mineralization were combined with phospholipid analysis to link the metabolic activities and substrate preferences of fungal and bacterial heterotrophs to dynamics in their abundance. We found that the presence of labile POM greatly affected the dominance of bacteria over fungi within the degrader communities and stimulated the decomposition of beech litter primarily through an increase in metabolic activity. Our data indicated that fungi primarily contribute to terrigenous C turnover by providing litter C for the microbial loop, whereas bacteria determine whether the supplied C substrate is assimilated into biomass or recycled back into the atmosphere in relation to phosphate availability. Thus, this study provides a better understanding of the role of fungi and bacteria in terrestrial-aquatic C cycling in relation to environmental conditions.

Introduction

Freshclass="Chemical">water ecosystems receive aclass="Chemical">nd process large amouclass="Chemical">nts of terrigeclass="Chemical">nous class="Chemical">n class="Chemical">carbon (C) (Cole ), often in the form of leaf litter (Fisher and Likens, 1973; Tank ). Herein, microbial heterotrophs greatly contribute to terrestrial–aquatic C cycling (Battin ; Ward ). The terrigenous subsidy is either assimilated into biomass, providing energy and nutrients for higher trophic levels of the food web (Attermeyer ; Marcarelli ), or mineralized to CO2, contributing substantially to global biogeochemical fluxes (Bardgett ; Rousk and Bengtson, 2014; Battin ).

Notably, the particulate forms of terrigenous C have a highly cclass="Chemical">omplex, arclass="Chemical">n class="Chemical">omatic chemical structure and large C to nutrient stoichiometry (McGroddy ; Lau ); they are therefore considered quite refractory toward microbial decomposition (Kleber ; Attermeyer ). Yet, modern research implies that environmental factors, rather than the chemical structure, control the microbial decomposition of terrigenous particulate organic matter (POM). Microbial communities respond rapidly to changes in their environment in two ways, either through variation in the species composition and/or metabolic activity (Comte and del Giorgio, 2011), both of which shape their performance in terrigenous POM cycling. Therefore, identifying drivers for the microbial turnover of terrigenous POM at the community level is fundamental to understand the role of aquatic decomposers in terrestrial–aquatic C cycling (Rousk and Bengtson, 2014).

Bacterial and fungal heterotrophs dclass="Chemical">omiclass="Chemical">nate microbial decclass="Chemical">n class="Chemical">omposer communities (Hieber and Gessner, 2002). However, these groups are phylogenetically distant and differ in their metabolic requirements and cellular capabilities (Mille-Lindblom and Tranvik, 2003). The typical filamentous structure of most fungi facilitates their penetration into particulate substrates, whereas bacteria are suspended in or attached to substrates (Baldy ). Fungal decomposers also have a lower metabolic nutrient demand (reviewed in Danger ) and wider enzymatic capabilities than bacteria, which allow them to mineralize low-quality substrates like particulate leaf litter (Güsewell and Gessner, 2009). However, bacteria are characterized by shorter turnover and higher metabolic activities (Bardgett ; Attermeyer ). Consequently, fungi and bacteria occupy different functional niches in POM decomposition: fungi act as primary degraders of particulate, predominately terrigenous C and bacteria act as rapid recyclers of simply structured nutrient-rich organic matter (OM) compounds (Gessner ; Krauss ).

The relative abundance of fungi and bacteria varies among different Pclass="Chemical">OM pools (Ficlass="Chemical">ndlay ), which caclass="Chemical">n have profouclass="Chemical">nd effects oclass="Chemical">n ecosystem fuclass="Chemical">nctioclass="Chemical">niclass="Chemical">ng (Riclass="Chemical">nclass="Chemical">naclass="Chemical">n aclass="Chemical">nd Bååth, 2009; Stricklaclass="Chemical">nd ) with regard to their fuclass="Chemical">nctioclass="Chemical">nal class="Chemical">niche iclass="Chemical">n the miclass="Chemical">neralizatioclass="Chemical">n of terrigeclass="Chemical">nous class="Chemical">n class="Chemical">OM (De Graaff ). However, studies simultaneously evaluating the different ecological roles of both fungi and bacteria are scarce, especially in the relation to environmental factors (Schneider ), and such investigation is mainly limited to soil studies (Rousk and Bååth, 2007). Moreover, the contribution of fungi and bacteria to OM degradation is not solely a function of their abundance, but of their metabolic activity that varies with changes in POM quality (Meidute ; Brandstätter ).

The availability of leaf litter for microbial degradation increases in the presence of class="Species">algae-derived class="Chemical">n class="Chemical">OM (Danger ; Ward ) denoting that energy-rich OM aid in the microorganisms' metabolic capabilities to degrade more complex, low-quality C (Kuehn ). Yet, whether this newly available substrate is assimilated or mineralized seems to be determined by the phosphate (P) availability (Rier ). Hence, this study is aimed at deliberately exploring how POM pools of substrates of different qualities, such as algae and leaf litter, determine the relative contributions of fungal and bacterial groups to POM degradation. We hypothesized that (i) the quality of POM pools determines the microbial performance in POM decomposition, (ii) algal POM stimulates the metabolic activity of fungi and bacteria toward leaf litter POM decomposition rather than their abundance and (iii) the contribution of bacteria to terrigenous C turnover varies with the C:P stoichiometry of algal substrates.

Using a stable-isotope approach, we incubated natural streambed sediments with different Pclass="Chemical">OM pools. These pools were geclass="Chemical">nerated through the cclass="Chemical">n class="Chemical">ombination of high-quality (algal POM) and low-quality terrigenous (litter POM) substrates that were 13C-labeled to directly trace their microbial mineralization through continuous measurements of 13CO2: total CO2 production. We further combined microbial respiration measurements with the analysis of aquatic phospholipid-derived fatty acid biomarkers (PLFA) to link microbial activities to microbial community dynamics for fungi and bacteria. The combination of PLFA biomarker analysis with stable-isotope analysis further provides information on the acquisition of different substrates distinct in their isotopic signature. Thus, we were able to directly determine group-specific changes in substrate utilization (Neufeld ) in relation to the substrate composition of the applied POM pools.

Consequently, the results frn class="Chemical">om this study provide importaclass="Chemical">nt implicatioclass="Chemical">ns for our uclass="Chemical">nderstaclass="Chemical">ndiclass="Chemical">ng of terrigeclass="Chemical">nous C traclass="Chemical">nsformatioclass="Chemical">n iclass="Chemical">n aquatic ecosystems aclass="Chemical">nd thus coclass="Chemical">ntribute to a better uclass="Chemical">nderstaclass="Chemical">ndiclass="Chemical">ng of the role of microbial decclass="Chemical">n class="Chemical">omposers in terrestrial–aquatic C cycling.

Materials and methods

Experimental procedures

Microbial sediment cclass="Chemical">ommuclass="Chemical">nities were iclass="Chemical">ncubated for 46 days uclass="Chemical">nder stable climate coclass="Chemical">nditioclass="Chemical">ns (iclass="Chemical">n the dark at 20 °C) together with Pclass="Chemical">n class="Chemical">OM pools that differed in their quality composition (Figure 1a). Therefore, natural sandy sediments were incubated in gastight sealed cylinders (acrylic glass, d=11.5 cm, h=20 cm) with artificial water (modified after Lehman, 1980, containing 20 mg CaCl2 l−1, 15 mg MgSO4 ·7H2O l−1, 20 mg NaHCO3 l−1 and 20 μl l−1 SL10 solution of trace elements mixed and autoclaved in bi-distilled water, without PO4−3 or NO3−2 addition) that recirculated through the sediment in the downward direction at a pore water velocity of ~0.9 × 10−4 m s−1 (after Angermann ). Artificial water was allowed to enter the chamber through a side port in the chamber wall, ~2 cm above the water level, to ensure fast equilibration of CO2 between water and the headspace needed for respiration analysis and to prevent anoxic conditions (Figure 1b). We chose artificial water as the percolation medium to avoid uncontrolled introduction of microbial communities and organic C substrates or other nutrients.

Figure 1

(a) Treatments applied on the sampled sediments with respect to organic C modifications. Beech litter, algae (High C:P) and algae (Low C:P) were added as single-substrate modifications (algae or beech modification) or a mixture of these was used for the mixed modifications (mixed modifications: algae+beech). (b) Experimental set-up of incubation chambers with a focus on the flow path for artificial water.

We sampled the sediments frclass="Chemical">om a ripple regioclass="Chemical">n (upper 2 cm) of a lowlaclass="Chemical">nd stream (Rheiclass="Chemical">nsberger Rhiclass="Chemical">n, 52°34'25"N 14°6'12"E, Braclass="Chemical">ndeclass="Chemical">nburg, Germaclass="Chemical">ny) characterized by regular sedimeclass="Chemical">nt movemeclass="Chemical">nt as well as iclass="Chemical">nteclass="Chemical">nsive vertical class="Chemical">n class="Chemical">water exchange and thereby intrusion of oxygen. The sampling area is surrounded by mixed forests and characterized by low-nutrient loading and low OM sandy sediments. After sampling, we sieved the sediment with stream water through 90- and 1000-μm meshes to remove the silt and clay as well as the coarse POM fraction. We let the resulting sediment slurries acclimate from in situ (2 °C) to experimental conditions in a stepwise (1–1.5 °C d−1), 2-week-long procedure while allowing percolation by artificial water. After acclimation, we mixed the organic substrates into the sediment according to our treatments and subsequently filled the resulting slurries into the incubation cylinders (four replicates each). The slurry approach was essential to ensure equal starting conditions for all treatments with respect to the composition of the microbial community and POM treatments and is commonly used in studies interrelating environmental effects on microbial performance (for example Marotta ; Dyksma ).

The different sediment Pclass="Chemical">OM pools were geclass="Chemical">nerated through the cclass="Chemical">n class="Chemical">ombination of three 13C-labeled POM substrates that differed in quality with respect to chemical structure and nutrient content. These were: pre-leached beech litter (1–2 mg per g sediment dry weight, δ13C=4500‰) as a proxy for a low-quality, stream C substrate of terrestrial origin, characterized by a highly complex aromatic structure and low-nutrient content (C:N:P=187:2:1); two algal substrates (6–12 μg per g sediment dry weight) as proxy for high-quality, stream C of aquatic origin. Both algal substrates differed in their P content and are referred to as algae HighCP (C:N:P=80:10:1, δ13C=503‰) and algae LowCP (C:N:P=36:5:1, δ13C=488‰) throughout the manuscript. Compared with the beech litter, the two algal substrates are characterized by a less-complex molecular structure and higher nutrient content that is assumed to be easily available for microbial decomposition. A detailed description of the applied OM substrates, including their preparation, has been provided in the Supplementary Methods. We amended the sediment with these three substrates in five different ways, resulting in five treatments that differed in their substrate composition of the sediment POM (Figure 1a): sediment with algae HighCP; sediment with algae LowCP; sediment with beech litter; sediment with beech litter and algae HighCP; and sediment with beech litter and algae LowCP. In addition to our treatments, we incubated four chambers with non-amended sediment (control) and applied the same handling to the entire preparation and incubation process.

We monitored microbial C respiration by sampling the headspace class="Chemical">CO2 every 6 h throughout the iclass="Chemical">ncubatioclass="Chemical">n. Iclass="Chemical">n additioclass="Chemical">n, we sampled the class="Chemical">n class="Chemical">water and sediment at the beginning (day 0) and the end (day 46) to obtain information on differences in the composition of the microbial decomposers, their metabolic activity (respiration and substrate specific assimilation), and nutrient consumption. During sampling, we collected the sediment in sterile falcon tubes using a laboratory spoon and the water in an Erlenmeyer flask through a port in the chambers' sidewall. For the start sample (day 0), we collected the sediment and water ~6 h after amendment with OM substrates and filling into incubation chambers to allow settling of the initially suspended sediment particles before the sampling.

Community composition of the microbial decomposers

Information on the cclass="Chemical">ompositioclass="Chemical">n of microbial decclass="Chemical">n class="Chemical">omposers was obtained through PLFA analysis from sampled sediments at days 0 and 46. PLFA are present in the membranes of all living cells and rapidly degrade to neutral lipids upon cell death (reviewed in Willers ), hence offer sensitive and reproducible measurements for characterizing microbial communities (for example, Boschker and Middelburg, 2002; Weise ). We extracted total lipids from 15 g of freeze-dried sediment using a well-established modified method described by Frostegard and Steger . A detailed description on the extraction and analysis of PLFA from the sediments is given in the Supplementary Methods. The concentration of each identified microbial PLFA was corrected to the total concentration of microbial PLFA to compensate for variations resulting from PLFA extraction. We identified 15 PLFAs that occur in the membranes of bacteria (i15:0, a15:0, i17:0), fungi (18:2ω6) and both microbial groups (14:0, 15:0, 16:1ω9c/7c, 16:1ω9t, 17:0, cy17:0, 18:0, 18:1ω9c, 18:1ω9t/7c, 18:3ω6/3, cy19:0). Certain PLFA were assigned to separate fungi (18:2ω6) from heterotrophic bacteria (i15:0) based on previous research (Boschker and Middelburg, 2002; De Carvalho and Caramujo, 2014) to disentangle their metabolic activity with respect to C assimilation from the introduced substrates.

Microbial C mineralization

Microbial C mineralization (C min), and thereby microbial activity, was measured as gaseous class="Chemical">CO2 emitted (class="Chemical">n class="Chemical">emCO2) during a 6 hour incubation period (equation 1). Total C mineralization was calculated from the sum of water total inorganic carbon (ΔTICWater) and emCO2 produced during 46 days (d) of incubation, expressed as a percentage of sediment total organic carbon (TOC), according to equations 2 and 3.

The concentration and isotopy of class="Chemical">emCO2 was measured every 6 h usiclass="Chemical">ng aclass="Chemical">n Off-Axis Iclass="Chemical">ntegrated-Cavity Output Spectroscope (Off-class="Chemical">n class="Disease">Axis ICOS CCIA, Los Gatos Research, CA, USA) connected to the chambers' headspace through tubes. The superscripts ‘t=6 h' and ‘t=0 h' represent the end and start of the incubation period. Every 12–24 h, we calibrated the instrument against an internal standard gas (70% N2, 30% O2 and 0.15% CO2, Airliquide, Germany) and against the international standard Vienna Pee Dee Belemnite values to correct for isotope drifting, yielding a precision of 2‰ for δ13C and 1 ppm for CO2 concentration. The detailed description of headspace CO2 analysis is provided in the Supplementary Methods. TIC was analyzed immediately after sampling according to DIN EN 1484 (DEV, H3) on a multi N/C 2100 Analyzer (Jena Analytics, Jena, Germany).

Food source elucidation of the respired (emCO2) and microbial fixed C (PLFA)

For single class="Chemical">OM modificatioclass="Chemical">ns, we applied a two-source mixiclass="Chemical">ng model approach proposed by Checlass="Chemical">ng (1996) oclass="Chemical">n class="Chemical">n class="Chemical">emCO2 to separate the sediment OM-derived CO2 emission (CSED) and algae-derived CO2 emission (CA) using equations 4 and 5:

For mixed class="Chemical">OM modificatioclass="Chemical">ns (beech aclass="Chemical">nd class="Chemical">n class="Species">algae together), we applied a two-source mixing model approach (equations 6 and 7) to separate for sediment+algae OM-derived CO2 respiration (CSED+A) and beech litter-derived CO2 respiration (CBEECH). In this approach, we treated the sediment and algal C as one C source and beech litter C as the second possible C source.

We obtained δ13emCO frclass="Chemical">om Keeliclass="Chemical">ng plot aclass="Chemical">nalyses of headspace class="Chemical">n class="Chemical">CO2 (Pataki, 2003). δ13C, δ13C and δ13C were assumed to be equal to δ13C of the POM, based on the report by Mary , where fractionation during microbial degradation processes was shown to be negligible.

We used the same mixing model approach on specific PLFA biclass="Chemical">omarkers to separate for sedimeclass="Chemical">nt class="Chemical">n class="Chemical">OM and substrate-derived C assimilated into the different biomarkers. We obtained δ13C for each specific PLFA biomarkers after correcting using a fractionation factor of 5‰, which is an average according to Boschker and Middelburg (2002). Fractionation factors may vary slightly between 1 and 3‰ for different PLFA biomarkers and according to environmental conditions; however, these variations are negligible because of the high enrichment in 13C achieved with our tracer approach.

Nutrient analysis

Prior to analysis, we freeze-dried and grinded the sediments sampled at days 0 and 46. For TP analysis, we applied the class="Chemical">ammonium molybdate spectrclass="Chemical">n class="Chemical">ometric method (DIN EN ISO 6878 2004). Determination of total organic carbon (TOC) and total nitrogen were performed according to DIN EN 1484 1997 using an Elementar Vario EL cube (Elementar Analysensysteme GmbH, Hanau, Germany).

Dissolved nutrients were analyzed frclass="Chemical">om class="Chemical">n class="Chemical">water sampled at day 0 and 46 after filtration through a 0.45 μm cellulose acetate filter (Sartorius Stedim Biotech GmbH, Goettingen, Germany). Soluble reactive phosphorus was determined photometrically on a UV/VIS-Photometer CARY 1E (VARIAN, Darmstadt, Germany) according to DIN EN ISO 11732 modified after Murphy and Riley (1962). Dissolved inorganic nitrogen (sum of NO32− and NH4+) was analyzed according to DIN EN 1189 on a CFA-Photometer Skalar SAN (Skalar Analytical B.V., Breda, The Netherlands) and dissolved organic carbon according to DIN EN 1484 on a multi N/C 3100 Analyzer (Jena Analytics).

Statistical analysis

To evaluate and visualize changes in the cclass="Chemical">ompositioclass="Chemical">n of microbial cclass="Chemical">n class="Chemical">ommunities according to our treatments, we performed a non-metric dimensional scaling (NMDS, three dimensions) ordination and a non-parametric multivariate analysis of variance (PERMANOVA) on computed pairwise Bray–Curtis dissimilarities on proportional lipid data (after Hall ). We also performed a principal component analysis (PCA) to identify important fatty acids that showed different results between our POM treatments at day 46. PCA was performed on a correlation matrix generated from proportional lipid data and the ratio of bacterial to fungal PLFA biomarker (i15:0 to 18:2ω6) from sampling day 46. We further applied non-parametric Kruskal–Wallis rank sum tests with subsequent post hoc Dunn's tests with Bonferroni correction to test for treatment effects (n=3–4) with regard to the consumption of nutrients, assimilation of beech C into microbial PLFA, and changes in fungal (18:2ω6) and bacterial (i15:0) group-specific PLFA biomarkers. We determined the significances of differences between treatments for cumulative respiration of beech-derived C throughout the final incubation period (30⩽t⩽46) by computing a linear mixed-effects model. The fixed structure was set as the interaction for treatment+control (6 levels) and sampling time (11 levels), and for the random structure, we allowed different intercepts for each replicate. Each linear mixed-effects model was followed by a model validation to check the residuals for normal distribution and homogeneity of variances. Statistical significance of the interaction was tested using a likelihood-ratio test by comparing the model with and without the interaction. When the interaction was significant, we analyzed each sampling time individually. The linear mixed-effects models were followed by the conservative Turkey's post hoc test to test significant differences between treatments. Further, we performed single linear regression analysis to relate C turnover to the ratio of group-specific biomarkers. All statistical analyses were performed in the statistic software R using packages dunn.test, MASS, multcomp, nmle, vegan and psych at a significance level of P⩽0.05 (Team R, 2010).

Results

Sediment Pclass="Chemical">OM modificatioclass="Chemical">ns resulted iclass="Chemical">n five Pclass="Chemical">n class="Chemical">OM pools (treatments) that differ in their composition of labile and refractory substrates (Figure 1a) and nutrient content (Table 1).

Table 1

Overview of sediment nutrient content at the start of incubation (day 0) given in total amount and nutrient stoichiometry

Treatment	TN (mg gDW−1)	TOC (mg gDW−1)	TP (μg gDW−1)	C:N	C:P	N:P
Beech	0.1 (0.0)	0.6 (0.1)	109.4 (22.1)	10.5	14.9	1.4
Beech+Algae HighCP	0.1 (0.0)	0.6 (0.1)	110.9 (20.4)	7.5	13.5	1.8
Beech+Algae LowCP	0.1 (0.0)	0.6 (0.0)	153.4 (41.2)	9.3	10.8	1.2
Algae HighCP	0.1 (0.0)	0.5 (0.0)	65.1 (21.5)	7.9	21.4	2.7
Algae LowCP	0.1 (0.0)	0.5 (0.0)	95.5 (17)	7.9	14.6	1.9
Control	0.1 (0.0)	0.6 (0.1)	67.0 (19.0)	8.2	21.6	2.6

Data are presented as mean of replicates (n=3−4) with s.d. given in brackets. DW, sediment dry weight; T, total; N, nitrogen; P, phosphorus; O, organic; C, carbon.

Composition of microbial decomposers in relation to POM pools

NMDS ordination based on Bray–Curtis dissimilarity on proportional PLFA data (Figure 2a) showed a significant shift in the cclass="Chemical">ompositioclass="Chemical">n of all the ideclass="Chemical">ntified PLFA after our treatmeclass="Chemical">nt (PERMANOVA, Pseudo-F=2.901, P=0.003). Thereiclass="Chemical">n, the PLFA cclass="Chemical">n class="Chemical">omposition significantly differed between POM pools at the end of the incubation period (day 46) and was related to both the presence of algae as well as beech litter substrates within POM pools (PERMANOVA, Pseudo-F=4.237, P=0.003). In addition, PCA on proportional PLFA data (Figure 2b) indicated that variation in the ratio of the abundance of fungi (18:2ω6) to heterotrophic bacteria (i15:0) explains 42.4% of the observed interrelation between POM pools and PLFA composition. Thus, the abundance of fungal PLFA (Supplementary Figure S1b) increased in the presence of beech litter (Kruskal–Wallis, χ2(5)=15.663, P=0.01), which was most pronounced for sediments incubated without algae (Beech, Dunn's test, P=0.01) and was independent of algal C:P stoichiometry in sediments amended with beech (Beech+Algae, Dunn's test, P=1). Further, PCA revealed that another 22.5% of the variation in PLFA data relates to the abundance of bacterial-specific PLFA i15:0 but was not related to our POM modifications (Supplementary Figure S1a, Kruskal–Wallis on i15:0, χ2(5)=8.666, P=0.12).

Figure 2

(a) Non-metric multidimensional scaling ordination based on Bray–Curtis dissimilarity on PLFA data, Kruskal's stress=0.02. Open dots (dashed ellipses) and colored squares or triangles (solid ellipses) separate sediment samples taken 6 h and 46 days after organic matter modifications, respectively. Colors separate the samples according to the modification: black=algae (LowCP); gray=algae (HighCP); dark blue=algae (LowCP) with beech litter; blue=algae (HighCP) with beech litter; light blue=beech litter. Ellipses represent 95% intervals around centroids for each treatment. (b) Principal component analysis of fatty-acid profiles of sediments sampled at day 46 reveal a main shift in PLFA biomarkers for heterotrophic bacteria (i15:0) versus fungi (18:2ω6) based on POM pool quality.

Mineralization of C substrates in relation to POM pools

Figure 3 and Supplementary Figure S2 show the resulting changes in C mineralization after our Pclass="Chemical">OM modificatioclass="Chemical">ns. Total C miclass="Chemical">neralizatioclass="Chemical">n was sigclass="Chemical">nificaclass="Chemical">ntly related to the cclass="Chemical">n class="Chemical">omposition of POM pools (Supplementary Figure S2a, Kruskal–Wallis, χ2(3)=11.514, P=0.01). Throughout the final incubation period (day 30–46), C mineralization patterns in sediments amended with algal substrates (Supplementary Figure S2a) were significantly related to algal nutrient stoichiometry (Dunn's test, P<0.01). Further, C mineralization was significantly related to algae C:P stoichiometry in sediments amended with beech litter (Kruskal–Wallis, χ2(2)=7, P=0.03). Significantly higher mineralization rates were observed for POM pools of beech litter and P-enriched algae (LowCP, Dunn's test, P=0.018).

Figure 3

Community dynamics versus microbial respiration. Fungal (18:2ω6)–bacterial (i15:0) dynamics relate to changes in C mineralization, where higher turnover rates were observed in the sediments with OM pools of higher quality.

Stable-isotope analysis of the respired class="Chemical">CO2 provides iclass="Chemical">nformatioclass="Chemical">n oclass="Chemical">n class="Chemical">n class="Chemical">OM substrates metabolized by the microbial community. Unfortunately, we lost the isotope data for the first 29 days of incubation owing to technical problems. However, the available data indicate a general increase in the mineralization of beech litter POM and a decrease in the mineralization of algal POM throughout the incubation period (data not shown).

Figure 4 gives an overview of the effect of class="Disease">algal Pclass="Chemical">n class="Chemical">OM on the mineralization of beech litter. Thus, two-end member mixing model analysis of emCO2 revealed a significant relation between POM pools and the proportion of beech-derived C in mineralized C (Figure 4b, Kruskal–Wallis, χ2(2)=7.86, P=0.01). Even so, the stimulative effect of algae OM on total C mineralization resulted in a higher amount of beech C respired throughout incubation (Figure 4a, linear mixed-effects model, LowCP: P=0.026, High CP: P=0.058), which was unrelated to algal nutrient stoichiometry.

Figure 4

Effect of algal OM on (a) overall and (b) fractional beech C turnover. Addition of algal OM high in P stimulated overall microbial respiration of beech-derived C but decreased the fractional utilization of beech C. Data are shown for the incubation period from day 30 to day 46. Data are presented as the mean of replicates (n=3–4) with s.d.

The observed differences in C mineralization were significantly related to fungal–bacterial ratios (Figure 3; single linear regression, R2=0.61, P<0.001), that is, higher C mineralization, as measured for sediments with a higher ratio of bacterial to fungal PLFA, which was the case when n class="Disease">algal Pclass="Chemical">n class="Chemical">OM, in particular that of high nutrient content, was added.

Effect of algal POM on the assimilation of beech litter by fungi and bacteria

class="Chemical">13C isotope aclass="Chemical">nalysis of class="Chemical">n class="Chemical">phospholipids provides information on the substrates metabolized by fungi and bacteria in relation to our POM treatments. Figure 5 provides an overview of the 13C isotope enrichment (δ13C) in all the identified PLFA at day 46, which was for all treatments up to 10-fold higher in fungal- (18:2ω6) then in bacterial- and unspecific PLFA biomarkers. Incubation of sediments with POM pools of beech and beech+algae HighCP resulted in similar 13C isotope enrichment in PLFA.

Figure 5

Overview of δ13C enrichment in microbial PLFA for all OM modifications including beech OM. Similar stable-isotope enrichment among PLFA is observed for OM modifications with only beech OM and in mixture with highCP algae OM. Data are presented as mean of replicates (n=3–4).

Total amount of beech C assimilated into PLFA specific for heterotrophic bacteria (i15:0) was related to the cclass="Chemical">ompositioclass="Chemical">n of Pclass="Chemical">n class="Chemical">OM pools (Figure 6, Kruskal–Wallis, χ2(2)=5.6, P=0.06), more precisely stimulated in presence of High C:P algae (Dunn's test, P=0.04). In contrast, the amount of beech C assimilated into fungal PLFA was independent of algal presence (Kruskal–Wallis, χ2(2)=3.84, P=0.15).

Figure 6

Amount of beech-derived C fixed into microbial PLFA after 46 days of incubation for (a) fungal PLFA (18:2w6) and (b) bacterial PLFA (sum of all bacterial PLFA) given in % of beech used. The presence of algae HighCP enabled a 1.5–2% assimilation of beech C into bacterial PLFA but not into fungal PLFA. Asterisk boxes indicate significant differences between the marked treatments.

Consumption of dissolved inorganic and organic nutrients

The consumption of soluble class="Chemical">reactive phosphorus (Kruskal–Wallis, χ2(5)=1.274, P=0.94), dissolved class="Chemical">n class="Chemical">inorganic nitrogen (Kruskal–Wallis, χ2(5)=9.016, P=0.11) and dissolved organic carbon (Kruskal–Wallis, χ2(5)=7.179, P=0.21) was similar between all the treatments, except for lower consumption of dissolved inorganic nitrogen in the sediments modified using algae HighCP. Dissolved inorganic nitrogen was limiting at the end of incubation. For data, see Supplementary Table S1.

Discussion

As expected, class="Species">algae class="Chemical">n class="Chemical">OM greatly stimulates microbial decomposition of terrigenous POM in aquatic ecosystems (Danger ; Ward ). Nevertheless, this study showed that microbial fungal and bacterial decomposers respond differently to substrate variations in POM pools and thus contributes to a better understanding of the role of microbial decomposers in terrestrial–aquatic C cycling.

Pronounced contrasts in substrate quality drive fungal–bacterial occurrence within the degrader communities

With regard to their distinct quality, algal and leaf litter are considered to differ in their availability to microbial decclass="Chemical">omposers (for example, Marcarelli ; Attermeyer ). This is reflected by the measured differeclass="Chemical">nces iclass="Chemical">n the C-respiratioclass="Chemical">n rates observed for sedimeclass="Chemical">nts ameclass="Chemical">nded with class="Chemical">n class="Species">algae or beech litter (Figure 3 and S2a).

Thus, the interrelation between Pclass="Chemical">OM quality aclass="Chemical">nd bacterial to fuclass="Chemical">ngal occurreclass="Chemical">nce observed iclass="Chemical">n this study agree with the reported disticlass="Chemical">nct metabolic capabilities of these orgaclass="Chemical">nisms to break dowclass="Chemical">n substrates of differeclass="Chemical">nt qualities (Riclass="Chemical">nclass="Chemical">naclass="Chemical">n aclass="Chemical">nd Bååth, 2009). Beech litter withiclass="Chemical">n Pclass="Chemical">n class="Chemical">OM pools favored the occurrence of fungi (Figure 3 and Supplementary Figure S1b), confirming that they dominate microbial communities when growing on refractory, nutrient-poor OM (Hieber and Gessner, 2002). However, amendments with algal OM favored the dominance of bacteria over fungi, denoting their competitive colonization and turnover of these high-quality substrates (for example, Bardgett ; Fontaine ). In addition, the lower abundance of fungal lipids in sediments amended with algal POM (Supplementary Figure S1b) indicates bacterial antagonism against fungi (Wargo and references therein), which was shown to be stimulated in the presence of high-quality OM (Mille-Lindblom ) such as our applied algal substrates.

Bacterial to fungal ratios did not differ between the class="Disease">algal P treatmeclass="Chemical">nts (Figure 3), but NMDS oclass="Chemical">n proportioclass="Chemical">nal data (Figure 2a) iclass="Chemical">ndicates a differeclass="Chemical">nt cclass="Chemical">n class="Chemical">omposition of bacterial and fungal PLFA between the two algal C:P treatments. This mismatch between the results of the two analyses implies that the quantity of labile and refractory substrates within POM pools determines the ratio between fungi and bacteria, whereas other environmental factors such as the C:P ratio of the labile substrate might primarily affect the species composition within each group.

Contribution of aquatic fungi and bacteria to terrigenous C turnover varies between POM pools

In accordance with previous findings (for example, Danger ), class="Species">algae Pclass="Chemical">n class="Chemical">OM stimulated the decomposition of beech litter (Figure 4a). However, our results support that the contribution fungi and bacteria to beech decomposition is likely not affected by algal POM in the same manner.

The more pronounced class="Chemical">13C eclass="Chemical">nrichmeclass="Chemical">nt iclass="Chemical">n the fuclass="Chemical">ngal cclass="Chemical">n class="Chemical">ompared with that in bacterial specific phospholipids (Figure 5) implies that fungi primary degrade beech litter, yet, fungal abundance was reduced in presence of algal OM. In result, a stimulated decomposition of beech litter by algal OM (Figure 4a) related to a more pronounced dominance of bacteria within the degrader communities (Figure 3) and thus resulted in an increased contribution of bacteria to beech C processing.

Bacterial abundance varied only slightly with our Pclass="Chemical">OM modificatioclass="Chemical">n, implyiclass="Chemical">ng that the additioclass="Chemical">nal C substrate was primarily expeclass="Chemical">nded for eclass="Chemical">nergy required for respiratioclass="Chemical">n, which is also supported by the observed less proclass="Chemical">nouclass="Chemical">nced effect of class="Chemical">n class="Disease">algal POM on the assimilation of beech C into microbial lipids (Figure 6). In addition, algal POM did not increase the use of beech C in preference over other C substrates by microbial decomposers (Figure 4b), thus confirming that high-quality POM stimulated the contribution of bacteria to leaf breakdown through an increase in their metabolic activity rather than through affecting their abundance or substrate preference.

Nutrient availability modifies the metabolic performance of bacteria in terrigenous C turnover

The availability of class="Chemical">nitrogen (N) aclass="Chemical">nd class="Chemical">n class="Chemical">phosphorus (P) frequently controls ecological processes (Elser ) such as litter decomposition (Gulis and Suberkropp, 2003; Rier ). In the present study, N and P were almost exclusively provided in particulate form through algal and/or beech litter substrates, whereas the pool of external inorganic N and P remained to be small to limiting (Supplementary Table S1). In addition, the C to nutrient stoichiometry of our applied leaf litter by far exceeds the nutrient demands of fungi and bacteria in contrast to the applied algal substrates (reviewed in Danger ). Thus, we assumed that the additional N and P provided with the algal substrates determine whether the beech-derived C is assimilated into microbial biomass or mineralized to CO2.

In this respect, mineralization of beech litter was stimulated in presence of nutrient-rich, class="Disease">algal Pclass="Chemical">n class="Chemical">OM (Figure 4a), yet independent of algal C:P stoichiometry, implying that the quantity of algal POM primarily relates to the extent to which these labile C compounds stimulate the decomposition of refractory C substrates. With regard to the assimilation of beech C into biomass (Figure 6), we also did not observe a significant response to our algal treatments for fungal lipids, but a slight response for bacterial lipids was observed, confirming that both groups of organisms may acquire different amounts (Güsewell and Gessner, 2009) and pools of nutrients during litter decomposition. That is, fungi may fulfill their energy and nutrient demand directly from the decomposition of litter (Suberkropp and Chauvet, 1995; Gadd, 2006). In contrast, our results indicate that the bacterial assimilation of low-nutrient, refractory C relies on additional nutrient pools, in particular with respect to P (Figure 6). This is further supported by the stronger variation in 13C in bacterial than fungal specific PLFA with our POM treatment indicating that bacteria respond more sensitive to our algae treatments than fungi (Figure 5). As previously noted, the composition of POM pools with different algal C:P substrates resulted in distinct compositions of microbial decomposers (NMDS, Figure 2a). Thus, the higher amount of beech C assimilated into bacterial PLFA in the presence of high C:P algae may have resulted from a shift in the microbial community composition toward species that have a lower P demand and are more competitive in the assimilation of litter compounds.

Conclusion

This study greatly improves our understanding of the effect of labile high-quality substrates on the fate of terrigenous class="Chemical">OM iclass="Chemical">n aquatic eclass="Chemical">nviroclass="Chemical">nmeclass="Chemical">nts. We showed that the abuclass="Chemical">ndaclass="Chemical">nce of fuclass="Chemical">ngi aclass="Chemical">nd bacteria withiclass="Chemical">n degrader cclass="Chemical">n class="Chemical">ommunities and their performance in metabolizing terrigenous OM are not intimately linked to each other, but largely depend on the composition of OM pools with respect to substrate quality. Clearly, the quantity of labile and refractory substrates determined the dominance of fungi and bacteria within the degrader communities and thereby the turnover of terrigenous C. However, the presence of specific environmental parameters such as P availability may also determine whether terrigenous substrates are assimilated and thereby transported into other trophic levels.

Our results suggest that aquatic fungi contribute to terrigenous C turnover by providing these C substrates for the microbial loop, whereas aquatic bacteria determine the amount of terrigenous C to be potentially recycled into the terrestrial environment through its emission into the atmosphere in the form of class="Chemical">CO2. However, giveclass="Chemical">n the large variety of Pclass="Chemical">n class="Chemical">OM pools in aquatic ecosystems and the few studies that exist on particulate C turnover, our results highlight the need for more sophisticated research to fully understand the contribution of fungi and bacteria to aquatic turnover of terrigenous C in relation to environmental conditions. Further, this study points out that the effects on both metabolic processes, assimilation and mineralization, have to be considered to fully understand these relations. Thus, the combinations of microbial and biogeochemical analyses with stable isotopes facilitated a more detailed observation of the underlying processes, and we highly recommend their application in further research.