A combined microbial and biogeochemical dataset from high-latitude ecosystems with respect to methane cycle

Sites overview and characteristics

This study focused on three regions located in subantarctic, arctic, and subarctic latitudes. The respective latitudinal and longitudinal ranges covered in this study were: 54.95 to 52.08 °S, and 72.03 to 67.34 °W in Patagonia; 67.44 to 67.54 °N, and 86.59 to 86.71 °E in Siberia; 63.21 to 68.63 °N, and −150.79 to −145.98 °W in Alaska (Figs. 1 and 2). The exact coordinates for each sample were included in the submitted dataset. The field campaigns were conducted in 2016, during the summer for each respective region: January-February in Chilean Patagonia, June-July in Alaska and July-August in Siberia.

For every site included in the present study, a set of nine qualitative environmental and/or ecological site-scale descriptors was selected and adapted from ENVO Environment Ontology⁴⁰, which included for example permafrost state, biome, environmental feature and vegetation type (Table 1, Fig. 3). Permafrost state was obtained from the NSIDC permafrost map⁴¹. The biome, large-scale descriptor based on climate and vegetation criteria, was derived from Olson et al.⁴². Temperate forest, boreal forest, and tundra biomes were included. The environmental features that were representative for the three regions were considered: lakes, wetlands, broadleaf/coniferous/mixed forest soils, grassland, tundra, and palsa. All the metadata was included in the submitted dataset. Table 2 summarizes the main types of sampled ecosystems and their main characteristics in the three regions, while Supplementary Table S1 provides the details of each sampling site.

In Alaska, the studied area ranged from the Alaska Range and Fairbanks area (interior, continental climate, 63–65°N, discontinuous permafrost) up to Toolik Field Station (North Slope, arctic climate, 66–69°N, continuous permafrost; Fig. 2). The physiochemistry and CH₄ emissions of lakes ALL1 (Killarney lake), ALL2 (Otto lake), ALL3 (Nutella lake), and ALL4 (Goldstream lake) were previously characterized³⁵. A number of heterogeneous soil and wetland samples were collected around the studied Alaskan lakes and/or from monitored sites, as detailed in Supplementary Table S1. In the Alaska Range and Fairbanks area, soils were mostly covered by mixed or taiga forests, alpine tundra, and bogs or fens wetlands. In the norther Brooks Ranges mountain system, the landscape was piedmont hills with a predominant soil of porous organic peat underlain by silt and glacial till, all in a permafrost state, characterized mainly by Sphagnum and Eriophorum vegetation, as well as dwarf shrubs.

In Siberia, the studied area was located in the discontinuous permafrost region surrounding Igarka, on the eastern bank of the Yenisei River (Fig. 2). This region was mainly covered by forest, dominated by larch (Larix Siberica), birch (Betula Pendula), and Siberian pine (Pinus Siberica), and palsa landscapes (frozen peat mounts), the latter being dominated by moss, lichens, Labrador tea and dwarf birch. In degraded areas, thermokarst bogs were dominated by Sphagnum spp. and Eriophorum spp. Land cover was an indicator of permafrost status, since forested areas reflected a deep permafrost table (>2 m) associated with Pleistocene permafrost, while palsa-dominated landscapes were indicative of the presence of near-surface (<1 m) Holocene permafrost. In this area, most of the lakes were of glacial origin and influenced by permafrost degradation⁴³ that has been observed for the last 30 years, while some were thermokarst lakes (Supplementary Table S1). Two studies that focused on methane cycling in SIL1 to SIL4 were recently published^18,20. We sampled organic soils on a degradation gradient from dry palsa to thermokarst bogs⁴⁴, as detailed in Supplementary Table S1.

Subantarctic sites were located in three areas in the Southern part of Chilean Patagonia: the Magellanic region around Punta Arenas, Tierra del Fuego, and Navarino Island (Fig. 2). Most of the sampled lakes from Magellanic and Tierra del Fuego regions were of glacial origin, while Navarino Island lakes were peatland lakes, surrounded by peatland and broadleaf forests. Peatlands were characterized by a very low diversity of Sphagnum species dominated by S. magellanicum from hollows up to hummocks. The typical broadleaf forests of the area were dominated by Nothofagus. Some grassland soil came from an experimental monitored field site (Supplementary Table S1). Samples collected from Patagonian soils and wetlands have been included in a recent survey of soil geochemical characterization (organic content)⁴⁵. Sediment samples collected in lakes PPL1, PPL2, PCL1, PCL2, PCP2 were also included in a recent study by Lavergne et al.⁴⁶ which showed that increasing air temperature led to enhanced CH₄ production and to an associated metabolic shift in the CH₄ production pathway, increasing the relative contribution of hydrogenotrophic methanogenesis compared to acetoclastic methanogenesis, together with consistent microbial community changes.

Surface area for lakes and elevation for all sites were determined using Google Earth Pro. Climate variables (Table 1) for each site were retrieved from WorldClim – Global⁴⁷.

Sampling design

A specific sampling strategy was defined for each kind of ecosystem, i.e. lakes, soils, and wetlands (Fig. 4), as follows.

In lakes, surface (0–10 cm) sediments and water samples were collected from three replicate points A, B, and C (Fig. 4) corresponding to the deepest zone of the lake, at ~ 2–5 meters of distance from each other. Two sampling depths were considered for the water samples: (i) at the oxycline, and (ii) just above the interface with sediment. Water was sampled using a 2.2 L Van Dorn bottle (Wildco, Mexico). Sediments were sampled using an Ekman dredge.

Mineral soil samples were collected from three replicate points A, B, and C (at ~ 2–5 meters of distance from each other), considering two sampling depths for each point (Fig. 4).

In wetlands, microtopography is known to influence organic matter decomposition, CH₄ emissions, microbial community structure, and metabolic pathways^48,49,50. The sampling strategy covered the three main microtopographic features of wetlands: hollows (i.e. small depressions, ponds, that can be filled with water or not at the time of sampling) (points A and D, Fig. 4); flat edges (or lawns) at the water table level or below, usually water-saturated and characterized by Sphagnum moss vegetation (points B and E, Fig. 4); and hummocks (i.e. dryer elevated mounts/raised domes, above the water table level, usually characterized by lichens and shrubs) (points C and F, Fig. 4). Two duplicate transects were considered, i.e. A-B-C and D-E-F transects, collected at ~ 10 meters of distance from each other. At each point, two sampling depths were considered, according to the same strategy as explained below for soils.

For both mineral and organic soils, soil blocks (20 × 20 × 20 cm blocks) were collected with a bread knife or a shovel. If soil layers could be clearly identified, top and bottom samples were defined accordingly and reported in the database. Otherwise, default depths were 0–10 cm for the surface layer and 10–20 cm for the bottom layer.

In addition to ecosystem-scale descriptors, every sample was characterized by point-scale descriptors (latitude, longitude, microtopography and vegetation type) and sample-scale descriptors such as environmental material (water, sediment, organic or mineral soil; Table 1 and Fig. 3). Soil samples were classified between organic and mineral soils using organic matter content (40% threshold) as the discriminating criterion between the two environmental materials⁶.

The material and methods used for characterizing these samples in situ and in the laboratory are described in the following sections. In some sites (ALP3, ALS3, ALS4, ALS6, ALS8, ALS9, PCL3, PCP3, PCS1, PPL3, PPP3, PTL1, PTL2, PTP1, PTP2, PTS1, SIL5, SIP6, SIP7, SIS3, SIS4), a basic characterization was carried out, due to harsh conditions and limited access. This basic characterisation included restrained set of measured parameters as listed in Table 1, yet enabling to fully fill the objective of this project. All the other sites were fully characterized, including the whole set of measured parameters as listed in Table 1, according to the environmental package (water, sediment, soil).

In situ analyses

Physicochemical analyses

At each sampling point and depth in lakes and hollows, dissolved oxygen, temperature, pH, conductivity, and redox potential were measured in water with a multiparametric probe (HI 9828, Hanna Instrument, Mexico). The detection limits for dissolved O₂ was 10 µg L⁻¹. In soil ecosystems, temperature was measured with an insertion thermometer (Isolab, Laborgerate GmbH).

Dissolved CH₄ and CO₂ concentrations

In lakes, the dissolved CH₄ and CO₂ concentrations were measured at each replicated sampling point and depth with the membrane-integrated cavity output spectrometry method using an ultraportable greenhouse gas analyzer (UGGA, Los Gatos Research, USA)⁵¹. The detection limits for dissolved CH₄ and CO₂ concentrations were 5 nmol L⁻¹ and 4 μmol L⁻¹ respectively.

Atmospheric CH₄ and CO₂ emission rates

CH₄ and CO₂ emission rates were estimated with a static opaque chamber coupled in a loop to the UGGA (Los Gatos Research, USA), following the procedure described previously⁹. Briefly, a 0.102 m² floating chamber (7.8 L) was placed at the surface of lakes and ponds and a 0.035 m² chamber (12.3 L) was installed on soil sites. Accumulation of CH₄ and CO₂ was recorded during 5 min, and flux determined from the slope of CH₄ and CO₂. Then the chamber was ventilated and closed to perform another flux measurement. At least three replicate measurements were performed at each location (sampling points defined in Fig. 4). The static chamber method used measures the total flux at the surface, i.e. including both diffusive and ebullitive fluxes. As illustrated in Fig. 5a, the highest CH₄ emission rates were found in hollows, especially in Siberian peatlands of discontinuous permafrost and lakes.

Sample processing in the field

For further analysis, water subsamples were collected into 10 mL glass vials, directly in the field. For δ¹³CCH₄, δ²HCH₄, and total organic carbon (TOC) analysis, samples were acidified (HCl 6 N). For dissolved inorganic carbon (DIC) concentration and δ¹³C-DIC analysis, HgCl₂ was added to the samples to stop any biological activity. After fixation by HCl and/orHgCl₂, water subsamples were stored at 4 °C in dark conditions. Soil samples were also kept at 4 °C for 24 h maximum before further processing.

Laboratory methods

Moisture and organic matter content

Soil and sediment samples were dried at 110 °C overnight to determine the dry weight. Organic matter content was assessed via loss on ignition at 550 °C.

Suspended solids

Lake and hollow water samples (20 mL to 3 L, until clogging) were filtered on pre-weighted combusted GF/F grade glass microfiber filters (0.7 µm pore size, Whatman). The filters were dried overnight at 105 °C to calculate the total suspended solids (TSS). The filters were then incinerated at 550 °C for 2 hrs to determine the concentration of particulate organic matter (POM).

Filtration

After pre-filtration at 80 µm (nylon net filters, Merck Millipore, Cork Ireland), water samples were filtered at 0.22 µm (nitrocellulose GSWP membrane filters, Merck Millipore, Cork Ireland) up to filter clogging (corresponding to 636 ± 521 mL on average, ranging from 70 to 2930 mL depending on the highly variable suspended matter content of the samples). The filter was frozen at −20 °C prior to DNA extraction. The filtrate was recovered and used to prepare four vials for further analysis of dissolved organic carbon (DOC), the isotopic composition (δ¹³C) of DOC, optical properties of dissolved organic matter, cations, anions, and trace elements.

Pore water extraction

The water extraction was carried out on soil and sediment samples to assess the mobile fraction of DOC, major anions and cations, trace elements, and the optical properties of dissolved organic matter (DOM). Following the procedure recommended in Jones & Willet⁵², 40 g of sample were placed in 200 mL of deionized water, and gently agitated with a magnetic stirrer at room temperature for 1 hr. The liquid phase was then recovered using a microRhizon sampler (Rhizosphere, Netherlands). The same procedure as for water samples was used to prepare and analyse these extracts.

Total and dissolved organic carbon

In water samples collected in lakes and hollows, TOC and DOC concentrations were analysed in using a TOC-V CSH analyser (Shimadzu, Japan). For DOC concentrations, samples were acidified to pH 2 using HCl 6 N and stored in 10 mL baked clear glass vials. The limit of quantification (LoQ) was 1 mg L⁻¹.

Anions and cations

Major ions were quantified in water samples collected in lakes and hollows and in pore water using a HPLC (Dionex, USA), a Dionex DX-120 analyser for cations (Thermo Fisher Scientific, France) and a Dionex ICS-5000 + analyser for anions (Thermo Fisher Scientific, France), according to recommandations⁵³. The LoQ was 0.5 mg L⁻¹ for calcium, chloride, sulphate, and magnesium; 0.25 mg L⁻¹ for bromide, sodium, and potassium; 0.025 mg L⁻¹ for ammonium and phosphate; 0.01 mg L⁻¹ for fluoride, nitrate, and nitrite.

Trace elements

For trace element analysis, samples were acidified with ultrapure HNO₃ prior to ICP-MS (7500ce, Agilent Technologies) analysis, and kept in 15 mL polypropylene vials. LoQ were <0.5 µg g⁻¹ for aluminium, iron, manganese, <0.05 µg g⁻¹ for vanadium, chromium, cobalt, nickel, copper, zinc, and <0.005 µg g⁻¹ for arsenic, strontium, cadmium, antimony, lead, uranium.

Optical properties

Subsamples were collected in 30-mL polypropylene vial for optical properties of DOM. The UV absorption spectra of pore water were measured with a spectrophotometer (Secoman UVi-lightXT5) from 190 to 700 nm in a 1 cm quartz cell. The specific UV absorbance at 254 nm (SUVA, L mg C⁻¹ m⁻¹) was calculated as follows: SUVA = A₂₅₄/b*DOC⁵⁴, where A₂₅₄ is the sample absorbance at 254 nm (non-dimensional), b is the optical path length (m), and DOC is in mg L⁻¹. Fluorescence measurements were performed using a spectrofluorometer (Synergy MX, Biotek). The emission spectrum was recorded for a 370 nm excitation wavelength. The fluorescence Index (FI) was determined for a 370 nm excitation wavelength, as the ratio of the 470 nm emission to 520 nm emission^55,56.

Isotopes

The stable isotopic signature of methane (δ¹³C-CH₄, shown in Fig. 5b, and δ²H-CH₄) was analyzed at the Stable Isotope Facility of UC-Davis (https://stableisotopefacility.ucdavis.edu/methane-ch4-gas), using a ThermoScientific Precon concentration unit interfaced to a ThermoScientific Delta V Plus isotope ratio mass spectrometer (ThermoScientific, Germany). Methane was extracted for IRMS analysis following the method of Yarnes et al.⁵⁷. The LoQ was 5 ppm of CH₄ for δ²H and 1.7 ppm of CH₄ for δ¹³C, and standard deviation was typically 2‰ for δ²H and 0.2‰ for δ¹³C. The δ¹³C-CO₂ was analyzed using a mass spectrometer (Isoprime 100, Elementar, UK) coupled with an equilibration system (MultiFlow-Geo, Elementar, UK). Samples were acidified using phosphoric acid and flushed with helium. The δ¹³C-DOC was analysed at the UC Davis Stable Isotope Facility, following the described procedure (http://stableisotopefacility.ucdavis.edu/doc.html). A TOC Analyzer (OI Analytical, College Station, TX) was interfaced to a PDZ Europa 20–20 isotope ratio mass spectrometer (Sercon Ltd., UK) utilizing a GD-100 Gas Trap Interface (Graden Instruments).

DNA extraction

Soil and sediments were subsampled and frozen at −20 °C. DNA was extracted from 0.5 g of the soil or sediment subsamples and from the previously frozen 0.22-µm filters using the PowerSoil and PowerWater DNA isolation kits, respectively (Qiagen, Hilden, Germany), following manufacturer instructions. The DNA extracts were stored at −20 °C.

qPCR assay

The abundances of four genes were measured by quantitative PCR (qPCR): bacterial 16S rRNA gene, archaeal 16S rRNA gene, pmoA gene (marker gene for aerobic methane oxidizing bacteria through the particulate methane monooxygenase), and mcrA gene (marker gene for methanogens and ANMEs through the methyl coenzyme M reductase). Duplicate measurements were run in 20 µL, using the Takyon SYBR master mix (Eurogentec, Belgium) with a CFX96 thermocycler (Bio-Rad Laboratories, Hercules, CA, US) and AriaMX thermocycler (Agilent, CA, US). Primer sequences and concentrations, thermocycling conditions, and standard curve preparation were detailed in Thalasso et al.¹⁸. As an illustration, the abundance of mcrA gene according to habitat (i.e. category combining the environmental material and the microtopography) is displayed in Fig. 6.

High-throughput amplicon sequencing

Archaeal and bacterial diversity was assessed using metabarcoding and targeting the V4-V5 region of 16S rRNA gene. Amplicons were obtained from DNA extracts using 515 F (GTGYCAGCMGCCGCGGTA) and 928 R (CCCCGYCAATTCMTTTRAGT) primers⁵⁸. MTP Taq DNA polymerase was acquired from Sigma (France). The thermocycling procedure was the following: 2 min at 94 °C; 30 cycles of 60 s at 94 °C, 40 s at 65 °C, and 30 s at 72 °C; and finally, 10 min at 72 °C. PCR products were used for pair-end sequencing using Illumina Miseq (2 × 250-bp). After pre-processing of raw reads through the FROGS pipeline⁵⁹, a total of 18 369 310 sequences were obtained from the 387 samples, and clustered into 121 971 OTUs using Swarm⁶⁰. The OTUs were further filtered at 0.005% of relative abundance, as previously recommended⁶¹, and taxonomically annotated against SILVA 132 rRNA database. Community analysis was carried out in R software, version 4.1.1, with ‘phyloseq’ package⁶². The taxonomic composition of bacteria according to habitat was represented by a barplot at the phylum level (Fig. 7a). As an illustration of the microbial diversity outcomes from this dataset and the community variability according to the different habitats, the dissimilarity among the 387 community structures was visualized by a principal coordinate analysis PCoA, a.k.a. Multidimensional scaling (MDS) with the ordinate function using Bray Curtis distance (‘phyloseq’ package⁶²) computed on the filtered and standardized (percentage) OTUs relative abundances (Fig. 7b).

Source: Ecology - nature.com