Redistribution of soil water by a saprotrophic fungus enhances carbon mineralization

Significance

This work shows a mechanism behind the observed higher drought resistance of soil fungi compared with bacteria. It also demonstrates the relevance of hydraulic redistribution by saprotrophic fungi for ecosystem ecology by influencing the carbon and water cycle in soils and terrestrial ecosystems under drought. Furthermore, we documented a so far underrated pathway of water in desiccated soils.

Abstract

The desiccation of upper soil horizons is a common phenomenon, leading to a decrease in soil microbial activity and mineralization. Recent studies have shown that fungal communities and fungal-based food webs are less sensitive and better adapted to soil desiccation than bacterial-based food webs. One reason for a better fungal adaptation to soil desiccation may be hydraulic redistribution of water by mycelia networks. Here we show that a saprotrophic fungus (Agaricus bisporus) redistributes water from moist (–0.03 MPa) into dry (–9.5 MPa) soil at about 0.3 cm⋅min⁻¹ in single hyphae, resulting in an increase in soil water potential after 72 h. The increase in soil moisture by hydraulic redistribution significantly enhanced carbon mineralization by 2,800% and enzymatic activity by 250–350% in the previously dry soil compartment within 168 h. Our results demonstrate that hydraulic redistribution can partly compensate water deficiency if water is available in other zones of the mycelia network. Hydraulic redistribution is likely one of the mechanisms behind higher drought resistance of soil fungi compared with bacteria. Moreover, hydraulic redistribution by saprotrophic fungi is an underrated pathway of water transport in soils and may lead to a transfer of water to zones of high fungal activity.

Drought is one of the most important and frequent abiotic stresses in terrestrial ecosystems (1). With respect to soil processes, soil desiccation limits microbial activity and decreases soil enzyme activity (2), carbon mineralization (3, 4) and nitrogen mineralization (5). In addition, drought can also alter soil microbial community composition (2, 6).

During desiccation and dry periods, soil fungal communities and fungal-based food webs are better adapted to drought than bacterial communities and bacteria-based food webs (7, 8). Bacteria are more strongly restricted than fungi (9), as bacterial activity needs a constant supply of water (10). One reason for the better adaptation of fungi compared with bacteria to low soil water potentials is seen in their strong cell walls, preventing water losses (1). The strength of fungal cell walls can even be enhanced by cross-linking of polymers and thickening under stress. Another reason for the better adaptation of fungi to soil desiccation might be hydraulic redistribution of water by mycelia networks. Hydraulic redistribution is defined as the passive transport of water in soils through organisms along a gradient in soil water potential and was first observed for plant roots (11). Hydraulic redistribution through plant roots improves plant survival and nutrient uptake by extending the life span and activity of roots (12) and by favoring decomposition of soil organic matter (13). Mycorrhiza fungal hyphae can also relocate water along gradients in soil water potential (12, 14–16). In addition, some studies reported the transport of nutrients and water over larger distances (>1 m) by saprotrophic fungal hyphae in nonsoil systems. Further, water leakage from hyphae into dry growth medium was observed (17). The water transport in hyphae was attributed to gradients in osmotic potentials (18–22).

Saprotrophic fungi are main regulators of soil nutrient cycling, litter decomposition, and soil respiration due to their specific enzymatic activities (23, 24) and due to the high density of hyphae in soil (up to 800 m⋅g⁻¹ soil) (25), and especially in litter layers. The ability of saprotrophic fungi to distribute water would provide a direct and fast connection between water and nutrient sources in soils that would be hardly accessible to bacteria. This could have an enormous impact on decomposition processes under drought conditions.

Here, we show the potential of the saprotrophic fungus Agaricus bisporus for hydraulic redistribution and impact of water redistribution on carbon mineralization in a desiccated soil.

Results and Discussion

We analyzed hydraulic redistribution using 4–5 replicates of two-chamber units filled with homogenized mineral soil. The single chambers of each unit were separated by a 2-mm air gap to prevent bulk flow of water (Fig. S1). After inoculation and the establishment of hyphal bridges through the air gap between the two chambers, the soil in both chambers was desiccated to a soil water potential of about –9.5 MPa. Thereafter, only chamber I was rewetted to field capacity (–0.03 MPa), whereas chamber II of each unit remained dry. Hydraulic redistribution was prevented in the controls by cutting the hyphal bridges between the two chambers. Volumetric water contents, soil water potential, and deuterium-labeling were used for quantification of hydraulic redistribution.

Fig. S1.

Schematic design of mesocosms.

At 72 h after rewetting chamber I with deuterium-labeled water, the volumetric soil water content of chamber II increased on average by about 0.02 cm³ H₂O cm⁻³ soil. The small increase of 0.006 cm³ H₂O cm⁻³ soil in chamber II of the controls can be attributed to diffusion of gaseous water through the air gap. Deuterium signatures of redistributed soil water supported the determined rates of hydraulic redistribution by A. bisporus. After 72 h, the amount of water redistributed from chamber I to chamber II was three times higher with hydraulic redistribution compared with the controls (Fig. 1). This was accompanied by an increase in water potential of the bulk soil in chamber II from –9.5 to –6.9 MPa with hydraulic redistribution, whereas in the controls soil water potential only increased from –9.2 to –8.2 MPa.

Fig. 1.

Hydraulic redistribution (HR) by hyphae. Amount of water redistributed from chamber I to chamber II 72 h after the irrigation of chamber I. Calculation based on hydrogen stable isotope ratios. Black, active hydraulic redistribution; gray, control with no fungal connection (mean + SEM, n = 4; Mann–Whitney U test, U_1,7 = 0, **P < 0.01).

After 72 h, hydraulic redistribution through hyphae had resulted in an average water flux of 67 µL·cm⁻²·d⁻¹ between both chambers. The number of hyphae bridging the two chambers in the air gap was about 2,300 cm⁻². The total water flux between the chambers corresponds to a specific water flux of about 0.03 µL·d⁻¹ through a single hyphae. A flow velocity of about 0.3 cm·min⁻¹ in single hyphae was calculated using the Bernoulli’s equation for the central cell lumen of the hyphae (outer diameter, 4.4 µm; central cell lumen, ∼2/3 of the total hyphal diameter) (26). This corresponds to flow velocities observed for the central cell lumen in arbuscular mycorrhizal hyphae (0.3 cm·min⁻¹) (27).

In general, passive mass flow is considered as the main mechanism of water transport in fungal hyphae (28), but active transport mechanisms for water might also be involved, such as cytoplasmic streaming (with velocities of 0.03–0.05 cm·min⁻¹ in hyphae) (29), vesicles moved by motor proteins (up to 0.02 cm·min⁻¹) (30), and vacuolar pathways (about 0.005 cm·min⁻¹) (31). Given these flow velocities, active transport cannot be the reason for the high velocities observed here, and thus, water redistribution is driven by passive mass flow along the soil water potential gradient. The mass flow of water can be apoplastic as well as symplastic. Symplastic water transport at such flow velocities requires the presence of aquaporins. Those are likely part of the cell membrane in A. bisporus as the encoding genes were identified (32).

Amount and velocity of hydraulic redistribution by saprotrophic fungi might even be higher under natural soil conditions than in our experiment. First, hyphal density in the chamber experiment was small compared with natural soils. From the number of hyphae in the air gap and an estimated tortuosity factor of 2 for hyphal length in soil pores, a total hyphal length of about 30 m·g⁻¹ soil dry weight close to the air gap was calculated. In natural soils, hyphal lengths of up to 800 m·g⁻¹ soil are reported (25). Second, the potential for hydraulic redistribution is likely larger for fungal species that form more complex mycelial structures, such as rhizomorphs or cords—that is, aggregations of longitudinal aligned hyphae. Although A. bisporus is capable of forming such cords, only single hyphae was found in our experiment. Cord-forming fungi were found to be very effective in translocating nutrients (21, 22, 33, 34). The transport of nutrients in cords is much faster than in nondifferentiated mycelium (28), and the distances for nutrient translocation in fungal cords can be >1 m (22). After localization and identification of substrates in the soil, cord connections are strengthened to exploit the substrate while other parts of the mycelium regress (35, 36), preferentially directing water to the substrate. Hence, as with nutrient translocation, hydraulic redistribution through fungal cords and especially rhizomorphs is probably more effective than through single hyphae.

In an additional experiment (Fig. S2), we determined the effect of hydraulic redistribution on carbon mineralization in desiccated soils by measuring soil enzyme activities and mineralization of ¹³C-labeled plant material. Soil zymography allowed us to measure enzyme activity in situ under different soil water contents, thus showing the effect of soil moisture on enzyme activity. Hydraulic redistribution by hyphae increased enzyme activity on average by 350% for N-acetylglucosaminidase and by 250% for cellobiohydrolase compared with the controls (Fig. 2). Enzyme activities decreased with increasing distance to the air gap in chamber II when hydraulic redistribution was active (Fig. 3), whereas no such pattern was observed in the controls. Similar relations of soil enzyme activities were observed between soils with and without irrigation (2, 37), emphasizing the significance of hydraulic redistribution by fungi.

Fig. S2.

Experimental design for measurement of CO₂ effluxes. Solid arrows, path flow for measurement; dashed arrows, alternative path flow for purging.

Fig. 2.

Enzyme activity on the soil surface. Measured in chamber II, 7 d after irrigation of chamber I. Black, active hydraulic redistribution (HR); gray, control with no fungal connection (mean + SEM, n = 5; NAG, lme, F_1;39 = 20.30, **P < 0.01; cellobiohydrolase, lme, F_1,39 = 15.91, **P < 0.01; NAG, N-acetylglucosaminidase).

Fig. 3.

Enzyme activity on the soil surface. (A) Cellobiohydrolase and (B) N-acetylglucosaminidase in chamber II of one mesocosm with active hydraulic redistribution, 7 d after irrigation of chamber I. The calibration line for the enzyme activity is presented at the bottom.

Hydraulic redistribution led to an enrichment in ¹³C of the respired CO₂ that was apparent already after 24 h and increased throughout the rest of the experiment. After 168 h, the cumulative C mineralization amounted to 59.7 g CO₂·kg⁻¹ C with active hydraulic redistribution and 2.1 g CO₂·kg⁻¹ C in controls (Mann–Whitney U test, U_1,9 = 0, P < 0.01; Fig. 4). This relation is similar to variations in C mineralization rates between dry and rewetted litter in forest soils (0.11 and 137 mg CO₂ kg⁻¹ C h⁻¹, respectively) (38), which again illustrates the significance of hydraulic redistribution for mineralization.

Fig. 4.

Cumulative carbon mineralization. Calculation based on ¹³C-CO₂ efflux from labeled plant material in chamber II, following irrigation of chamber I. Black, active hydraulic redistribution (HR); gray, control with no fungal connection (solid lines, mean values; dashed lines, SEM; n = 5).

Therefore, saprotrophic fungi not only have the capability to redistribute water but can also partly compensate unfavorable soil moisture conditions in desiccated soil as long as water is available in other zones of the mycelial network, like in deeper soil horizons. Under dry conditions, desiccation descends from the substrate-rich upper soil layers to the subsoil. The leakage of water from hyphae into the surrounding soil is concentrated to hyphal tips (39), and hyphal tips are concentrated in the growing part of the mycelium close to the substrate (35). Water redistribution through the mycelium, bypassing capillary transport through soil pores, has probably been underrated as a pathway of water movement in desiccated soils. Hydraulic redistribution is of particular relevance for bridging capillary barriers. Overall, hydraulic redistribution likely leads to a transfer of water to hotspots of fungal activity in dry soils with preferential wetting of the surrounding substrate. It enables high fungal enzymatic activity in the growth zone even under low soil water potentials. Hence, hydraulic redistribution likely is one of the mechanisms behind the higher resistance of soil fungi compared with bacteria to soil desiccation.

Materials and Methods

General Setup.

Experiments were carried out in mesocosms (adapted from 39) (Fig. S1) represented by two chambers (a 6 × 20 × 15 cm), filled with a steam sterilized mixture (2:1:1 vol/vol/vol) of loamy soil (17% clay; 76% silt; 7% sand) and medium coarse quartz sand (Dorsilit 8; particle size range, 0.3–0.8 mm) as well as coarse quartz sand (Dorsilit 7, 0.6–1.2 mm; Dorfner GmbH & Co.). The two chambers made of Makrolon (Bayer AG) had openings on the sides facing each other, which were covered by two 160-µm pore size stainless steel mesh screens. A 2-mm-thick air gap between both chambers prevented capillary flow of water and was stabilized by two additional perforated stainless steel mesh screens with 2 mm pore sizes. Chamber tops were removable and were air-tight if closed. The soil surface was compressed slightly to obtain a uniformly flat surface. Fungal cultures [DSM No. 3056, A. bisporus (Lange) Imbach] were received from the Leibniz Institute DSMZ and grown on malt extract peptone agar at 14 °C. The soil of chamber I was inoculated by placing a 1 cm² agar plate with fungal hyphae into the substrate close to the air gap at a depth of approx. 2 cm. A. bisporus was chosen because it is one of the best studied filamentous fungal species, with a complete available genome, that shows a fast growth rate. Both chambers were maintained at 23 °C and irrigated regularly to field capacity with a liquid fungal growth medium (2% glucose, 0.2% peptone, 0.2% yeast extract, 0.1% K₂HPO₄, 0.46% KH₂PO₄, and 0.05% MgSO₄) (40) for 6 wk. The chamber tops were kept open but were covered with glass microfiber filter paper (Grade 934-AH, Whatman Ltd.) during the growth phase to facilitate air exchange and avoid contaminations. Volumetric water content was controlled continuously using soil moisture sensors monitoring the dielectric constant of the media (ECH₂O-10 moisture sensor; Decagon Devices Inc.).

Quantification of Hydraulic Redistribution.

After desiccation for 6 wk at 23 °C to a soil water potential of approximately –9.5 MPa in both chambers, chamber I was rewetted to field capacity (–0.03 MPa) with deuterium-labeled water (3% at deuterium enrichment; ROTH GmbH + Co. KG). Mescosms were then closed air-tight and only opened for sampling of soil cores. Soil cores of chamber II were destructively sampled 72 h after irrigation of chamber I. Water for isotope analyses was extracted from soil samples by cryogenic vacuum extraction (41). Hydrogen-stable isotope analyses were conducted at the Laboratory for Isotopic-Biogeochemistry (University of Bayreuth) using thermal conversion/isotope-ratio mass-spectrometry (isotope mass spectrometer, delta V advantage; Thermo Fisher Scientific).

In addition, soil water potential was measured on collected soil samples (4 cm diameter, 0.5 cm thickness) using the chilled mirror dewpoint method (WP4-T; Decagon Devices Inc.) (42).

Controls were established by mesocosms treated in the same way as described above, but the hyphal connections between the chambers in the air gap were severed by cutting with a thin stainless-steel wire before irrigating chamber I. In total, four mescosms with intact fungal connections as well as four control mescosoms were treated with deuterium-labeled water.

Mineralization of Organic Matter.

CO₂ efflux from the soil is an indicator of the general activity of soil microorganisms and was therefore used to estimate the impact of hydraulic redistribution on mineralization of organic matter under drought conditions. The use of ¹³C-labeled plant material (Triticum aestivum L. green shoots; >97 atom% ¹³C; C/N ratio, 15; IsoLife) enabled us to trace the origin of collected CO₂. Labeled plant material (five ground samples of 20 mg each) were placed at regular intervals of 4 cm from the mesh screen on the soil surface of the nonirrigated chamber II, shortly before rewetting chamber I. Mescosms were then closed air-tight and were not opened for 7 d. CO₂ effluxes were regularly measured for 7 d at 20 °C using the dynamic closed-chamber technique (43) (Fig. S2). CO₂ concentrations in the mesocosms headspace (1.2 L) were measured every 6 min for 30-min periods for the first 24 h with an infrared gas analyzer (LiCOR 820; Licor). Beginning with the second day of experiments, CO₂ concentrations were measured for periods of 48 min. Soil CO₂ effluxes were calculated from the slope of the linear regression between CO₂ concentration and incubation time. An alternative air path flow was opened at the end of each measurement cycle for 30 min on the first day of measurements and subsequently for 12 min to flush the system with CO₂-free synthetic air and reduce the CO₂ concentration. In addition, the extracted air was collected every 12 h for further analysis of ¹³C isotope contents to determine the percentage of decomposed plant material in chamber II. Extracted air was stored in 10 mL butyl rubber septum-capped vials by flushing for 90 s. Septa were heated at 105 °C for 12 h before vial closing to prolong stability of CO₂ isotope composition (44). Vials were flushed with N₂ for 90 s prior to use to remove environmental CO₂. ¹³C-CO₂ efflux for the whole measurement cycle was interpolated from measured ¹³C-CO₂ efflux values using a Gaussian function extended with a linear term to adjust for divergences from a normality distribution. Function fitness was optimized using Solver (Microsoft Cooperation).

δ¹³C analyses were conducted at the Laboratory for Isotopic-Biogeochemistry (University of Bayreuth) using an Elemental analyzer (NA 1108; CE Instruments)–isotope ratio mass spectrometer (delta S; Finnigan MAT) linkage.

In total, five mescosms with intact fungal connections as well as five control mescosoms were treated with labeled plant material.

Analysis of Soil Enzyme Activity.

The impact of hydraulic redistribution on soil enzyme activity was analyzed using soil zymography (ref. 45 modified by ref. 46). This in situ method allows for analysis of the 2D distribution of enzyme activities in soil with high spatial resolution and under different water contents in contrast to more traditional methods that are based on the determination of enzyme activity in solution. Hence, this provided a comprehensive picture of the allocation of redistributed water in chamber II and the resulting influence on enzyme activities. In addition, the study of enzymatic activities provides functional information on specific aspects of organic matter decomposition and can therefore support the results of CO₂ efflux measurements. N-acetylglucosaminidase and cellobiohydrolase activity were analyzed using the artificial substrates 4-Methylumbelliferyl N-acetyl-β-d-glucosaminide (4-MNG) and 4-methylumbelliferyl β-d-cellobioside (4-MC; both Sigma-Aldrich Chemie GmbH), respectively. The fluorogenic 4-methylumbelliferone (MUF) is released from 4-MNG and 4-MC due to hydrolytic cleavage in the presence of compatible enzymes. In soils, the activity of chitinase is considered as a good indicator of fungal biomass and activity (47, 48).

A 1% agarose gel (size of 0.1 × 12.0 × 11.0 cm) was cast in systems usually used for vertical gel electrophoresis (Biometra GmbH). The gel was sliced in four parts at 2 × 11 cm, and all four parts were attached to the soil surface of the nonirrigated chamber II in the space between the labeled plant material samples. Polyamide membrane filters (0.45 µm pore size; Sartolon, Sartorius AG) were sliced in 10 parts at 2 × 11 cm. Half of the slices were saturated with a 4-MNG solution or 4-MC solution (25% wt/vol; Sigma-Aldrich Chemie GmbH), respectively. Four slices of each group were placed in turn on top of the gel slices, starting with the 4-MNG group. The membrane filter was extracted after an incubation time of 25 min at 20 °C for 4-MNG and 20 min for 4-MC and illuminated on a fluorescent transilluminator in the dark (wavelength, 355 nm; Desaga GmbH). Pictures were taken with a digital camera (Nikon D3100) and analyzed in comparison with controls without hydraulic redistribution. To adjust for differences in exposure time, which are necessary to avoid overexposure at high activities and loss of details at low activities, one filter slice was not incubated on the soil but photographed together with the others and served as a standard of zero activity. A calibration line was prepared from membranes soaked in different solutions of MUF concentrations (0, 35, 70, 130, and 200 µM). These calibration membranes were cut into strips of 2 cm and photographed under the UV light in the same way as the zymogram membranes. The amount of MUF on an area basis was calculated from the volume of solution taken up by the membrane and by the size of the membrane.

Image processing and analysis were done using the open source software imageJ 1.46r (Wayne Rasband, National Institutes of Health, Bethesda, MD). The digital images were transformed to 8-bit—that is, grayscale images. To illustrate the results, the values of the grayscale image were depicted in false color. The linear correlation between the MUF concentration and the mean of grayscale in an area of 4 cm² of each calibration gel were calculated using the software R. A segment at 1.5 × 7.5 cm with no visible disturbance was selected from the soil zymograms, and mean values of the grayscale were measured. Values were standardized based on the difference between the standards of zero activity and the calibration membrane with 0 µM MUF concentration. Values were expressed as pmol MUF/h and mm².

Data Analysis.

All statistical analyses and graphics were done using R 3.1.0 (R Developmental Core Team). Normality and homogeneity of the data were tested using Shapiro–Wilk test and Levene's test, respectively. Enzyme activities were analyzed using linear mixed-effect models as implemented in the R package nlme (49). The sample origin from the different chambers was added as a random factor into the model to adjust for random variances among chambers. For pair-wise post hoc comparisons, general linear hypotheses based on Tukey all-pair comparisons were conducted, using the R package multcomp (50).

Kruskal–Wallis test with pair-wise Wilcox tests for post hoc comparisons were used if data were not normal and/or variances were not homogeneously distributed.