Gene-informed decomposition model predicts lower soil carbon loss due to persistent microbial adaptation to warming
Site description and sampling
This experimental site was established in July 2009 at the Kessler Atmospheric and Ecological Field Station (KAEFS) in the US Great Plains in McClain County, Oklahoma (34̊ 59ʹN, 97̊ 31ʹW)14,48. Experimental design and site description were described in detail previously25. Briefly, Ambrosia trifida, Solanum carolinense and Euphorbia dentate belonging to C3 forbs, and Tridens flavus, Sporobolus compositus and Sorghum halapense belonging to C4 grasses are dominant in the site25,48. Annual mean temperature is 16.3 °C and annual precipitation is 914 mm, based on Oklahoma Climatological Survey data from 1948 to 1999. The soil type of this site is Port–Pulaski–Keokuk complex with 51% of sand, 35% of silt and 13% of clay, which is a well-drained soil that is formed in loamy sediment on flood plains. The soil has a high available water holding capacity (37%), neutral pH and 1.2 g cm−3 bulk density with 1.9% total organic matter and 0.1% total nitrogen (N)25,48. Four blocks were used in the field site experiment, in which warming is a primary factor. Two levels of warming (ambient and +3 °C) were set for four pairs of 2.5 m × 1.75 m plots by utilizing a real or dummy infrared radiator (Kalglo Electronics, Bethlehem, PA, USA). In the warmed plots, a real infrared radiator was suspended 1.5 m above the ground, and the dummy infrared radiator was suspended to simulate a shading effect of the device in the control plots.
In this study, eight surface (0–15 cm) soil samples, four from the warmed and four from the control plots, were collected annually at approximately the date of peak plant biomass (September or October) from 2010 to 2016. Three soil cores (2.5 cm diameter × 15 cm depth) were taken by using a soil sampler tube in each plot and composited to have enough samples for soil chemistry, microbiology and molecular biology analyses. A total of 56 soil samples were analyzed in this study.
Environmental and soil chemical measurements
Precipitation data were obtained from the Oklahoma Mesonet Station (Washington Station)48 located 200 m away from our experiment site, and 12-month version of the standardized precipitation-evapotranspiration index (SPEI-12) was used as annual drought index49. Air temperature, soil temperature and volumetric soil water content were described in detail previously25. Specifically, air temperature and soil temperature at the depth of 7.5 cm in the center of each field plot were measured by using Constantan-copper thermocouples wired to a Campbell Scientific CR10x data logger (Campbell Scientific, UT, USA). A portable time domain reflectometer (Soil Moisture Equipment Corp.) was used to measure soil moisture from the soil surface to a 15-cm depth once or twice a month. Three measurements of soil moisture were performed in each plot and the average of three technical replicates were used in further analyses.
All soil samples were analyzed to determine soil total organic carbon (TOC), total nitrogen (TN), soil nitrate (NO3−) and ammonia (NH4+) by the Soil, Water, and Forage Analytical Laboratory at Oklahoma State University (Stillwater, OK, USA). Soil pH was measured using a pH meter with a calibrated combined glass electrode50.
Aboveground plant communities
Aboveground plant community investigations were annually conducted at peak biomass (usually September)48,51. Aboveground plant biomass, separated into C3 and C4 species, was indirectly estimated by a modified pin-touch method48,51. Detailed description of biomass estimation is provided by Sherry et al.52. A pin frame used in this study is 1 m long and have 10 pins 10 cm apart at 30° from vertical. Pins with a 0.75 m length were raised within the frame to count hits up to 1 m high (hits over 1 m are negligible at this site). The pin frame was placed in the center of each plot to record the contact numbers of the pins separately with C3 and C4 plants (e.g., leaves and stems). The contact numbers of C3 and C4 plants were then used to estimate plant biomass using calibration equations derived from calibration plots, which were located near the experimental plots. Biomass in the calibration plots was clipped at a height of 10 cm above the ground at approximately the date of peak plant biomass (September or October). All of the species in plant community within each plot were identified to estimate species richness. Clipped plant materials were oven-dried and then correlated with the total contact number. C3 and C4 plant biomasses were estimated by using the calibration equation of contact number and plant biomass. All of the species within each plot were identified to estimate species richness of plants.
Ecosystem C fluxes and soil respiration
Ecosystem C fluxes and soil respirations were measured once or twice a month between 10:00 and 15:00 (local time) from January 2010 to December 2016 by following previous methods14,48. One square aluminum frame (0.5 m × 0.5 m) was inserted in the soil at 2 cm depth in each plot to provide a flat base between the soil surface and the CO2 sampling chamber. NEE and ecosystem respiration (ER) were measured using LI-6400 portable photosynthesis system (LI-COR). Gross primary productivity (GPP) was estimated as the difference between NEE and ER. Meanwhile, soil surface respiration was monthly measured using a LI-8100A soil flux system attached to a soil CO2 flux chamber (LI-COR). Measurements were taken above a PVC collar (80 cm2 in area and 5 cm in depth) and a PVC tube (80 cm2 in area and 70 cm in depth) in each plot. The PVC tube was permanently fixed on the ground to cut off old plant roots and prevent new roots from growing inside the tube. Any aboveground parts of living plants were removed from the PVC tubes and collars before each measurement. The CO2 efflux measured above the PVC tubes represented heterotrophic respiration (Rh) from soil microbes, while that measured above the PVC collars represented soil total respiration (Rt) including heterotrophic and autotrophic respiration (Rh and Ra) from soil microbes and plant root, respectively.
Soil decomposition rate
Weighted cellulose filter paper (Whatman CAT No. 1442-090) was placed into fiberglass mesh bags and placed vertically at 0–10 cm soil depth in each plot in March 2016. All of decomposition bags were collected back in September 2016, rinsed and dried at 60 °C for weighing. The percentage of mass loss was calculated to represent soil decomposition rate.
Molecular analyses of soil samples
The C substrate utilization patterns of soil microbial communities in 2016 were analyzed by BIOLOG EcoPlateTM (BIOLOG). The BIOLOG EcoPlateTM contains 31 of the most useful labile carbon sources for soil community analysis, which are repeated three times in each plate. In this study, the plates with diluted soil supernatant (0.5 g soil with 45 mL 0.85% NaCl) were incubated in a BIOLOG OmniLog PM System at 25 °C for 4.5 days. The color change of each well was shown as absorbance curve. The net area under the absorbance versus time curve was calculated to represent physiological activity of various C sources53. The average value from three replicates was used for analyses in this study.
Soil total DNA was extracted from 1.5 g soil by freeze-grinding and SDS-based lysis54, and purified with a MoBio PowerSoil DNA isolation kit (MoBio Laboratories)25. Then, 10 ng DNA per sample were used for library construction and amplicon sequencing. Amplicons sequencing was performed with cautions in terms of experimental preparations and data analyses to ensure sequence representativeness and semi-quantitative nature55. The V4 region of bacterial and archaeal 16S rRNA genes were amplified with the primer set 515F (5ʹ-GTGCCAGCMGCCGCGGTAA-3′) and 806R (5ʹ-GGACTACHVGGGTWTCTAAT-3ʹ), and fungal ITSs between 5.8S and 28S rRNA genes were amplified with the primer set ITS7F (5ʹ-GTGARTCATCGARTCTTTG-3ʹ) and ITS4R (5ʹ-TCCTCCGCTTATTGATATGC-3ʹ). PCR products from different samples were sequenced on a MiSeq platform (Illumina, Inc.) using 2 × 250 pair-end sequencing kit. Raw sequences were submitted to our Galaxy sequence analysis pipeline (http://zhoulab5.rccc.ou.edu:8080) to further analyze according to the protocol in the pipeline25. Finally, OTUs were clustered by UPARSE56 at 97% identity for both 16S rRNA gene and ITS. All sequences were randomly resampled to 30,000 sequences for 16S rRNA gene and 10,000 sequences for ITS per sample. Representative sequences of OTUs were annotated taxonomically by the Ribosomal Database Project (RDP) Classifier with 50% confidence estimates.
GeoChip 5.0 M, a functional gene array57, was used for all 56 samples from 2010 to 2016. GeoChip hybridization, scanning and data processing were performed in the Institute for Environmental Genomics, University of Oklahoma57,58. Specifically, 800 ng of purified soil DNA of each sample was mixed with 5.5 µl random primers (Life Technologies, random hexamers, 3 µg/µl), diluted with nuclease-free water to 35 µl, heated to 99 °C for 5 min, and placed on ice immediately. The labeling master mix (15 µl), including 0.5 µl of Cy-3 dUTP (25 nM; GE Healthcare), 2.5 µl of dNTP (2.5 mM dTTP, 5 mM dAGC-TP), 1 µl of Klenow (imer; San Diego, CA; 40 U ml−1), 5 µl Klenow buffer, and 2.5 µl of water, was added in the sample mixed solution. The samples were incubated at 37 °C for 6 h in a thermocycler, and then incubated at 95 °C for 3 min to inactivate the enzyme. Subsequently, samples were protected from the light as much as possible. Labeled DNA was cleaned using a QIAquick purification kit (Qiagen) according the manufacturer’s instructions and then dried thoroughly in a SpeedVac (45 °C, 45 min; ThermoSavant).
Labeled DNA was resuspended into 27.5 µl of DNase-free water, and then mixed completely with 99.4 µl of hybridization solution, containing 63.5 µl of formamide (10% final concentration), 2 × HI-RPM hybridization buffer, 12.7 µl of 10 × aCGH blocking agent, 0.05 μg/µl Cot-1 DNA, and 10 pM CORS58. The mixed solution was denatured at 95 °C for 3 min, and then incubated at 37 °C for 30 min. The DNA solution was centrifuged at 6000 × g for 1 min to collect liquid at the bottom of the tube. 110 µl of the solution was pipetted into the center of the well of the gasket slide. The array slide was placed on the gasket slide, sealed using a SureHyb chamber, hybridized at 67 °C for 24 h at 20 rpm in a hybridization oven. After hybridization, slides were washed in room temperature with Wash Buffer 1 (Agilent) and Wash Buffer 2 (Agilent).
The slides were imaged as a Multi-TIFF with a NimbleGen MS200 Microarray Scanner (Roche NimbleGen, Inc., Madison, WI, United States). The raw signals from NimbleGen were submitted to the Microarray Data Manager on our website (http://ieg.ou.edu/microarray), cleaned, normalized and analyzed using the data-analysis pipeline. Briefly, probe quality was assessed, and poor or low signal probes were removed. Probe spots with coefficient of variance (CV; probe signal SD/signal) >0.8 were removed. Then, the signal-to-noise ratio (SNR) was calculated. As suggested by Agilent, the average signal of Agilent’s negative control probes within each subarray was used as the background signal for the probes in that subarray instead of the local background typically used. The signal intensity for each spot was corrected by subtracting the background signal intensity. If the net difference was30%, aligned length >20 a.a., and e-value More
