High Temporal and Spatial Variability ofAtmospheric-Methane Oxidation inAlpine Glacier Forefield Soils

Eleonora Chiri,* Philipp A. Nauer,* Edda-Marie Rainer,* Josef Zeyer,Martin H. SchrothInstitute of Biogeochemistry and Pollutant Dynamics (IBP), ETH Zurich, Zurich, Switzerland

ABSTRACT Glacier forefield soils can provide a substantial sink for atmosphericCH4, facilitated by aerobic methane-oxidizing bacteria (MOB). However, MOB activity,abundance, and community structure may be affected by soil age, MOB location indifferent forefield landforms, and temporal fluctuations in soil physical parameters.We assessed the spatial and temporal variability of atmospheric-CH4 oxidation in anAlpine glacier forefield during the snow-free season of 2013. We quantified CH4 fluxin soils of increasing age and in different landforms (sandhill, terrace, and floodplainforms) by using soil gas profile and static flux chamber methods. To determine MOBabundance and community structure, we employed pmoA gene-based quantitativePCR and targeted amplicon sequencing. Uptake of CH4 increased in magnitudeand decreased in variability with increasing soil age. Sandhill soils exhibited CH4

uptake rates ranging from �3.7 to �0.03 mg CH4 m�2 day�1. Floodplain andterrace soils exhibited lower uptake rates and even intermittent CH4 emissions.Linear mixed-effects models indicated that soil age and landform were the domi-nating factors shaping CH4 flux, followed by cumulative rainfall (weighted sum�4 days prior to sampling). Of 31 MOB operational taxonomic units retrieved,�30% were potentially novel, and �50% were affiliated with upland soil clustersgamma and alpha. The MOB community structures in floodplain and terrace soilswere nearly identical but differed significantly from the highly variable sandhillsoil communities. We concluded that soil age and landform modulate the soilCH4 sink strength in glacier forefields and that recent rainfall affects its short-term variability. This should be taken into account when including this environ-ment in future CH4 inventories.

IMPORTANCE Oxidation of methane (CH4) in well-drained, “upland” soils is an im-portant mechanism for the removal of this potent greenhouse gas from the atmo-sphere. It is largely mediated by aerobic, methane-oxidizing bacteria (MOB). Whereasthere is abundant information on atmospheric-CH4 oxidation in mature upland soils,little is known about this important function in young, developing soils, such asthose found in glacier forefields, where new sediments are continuously exposed tothe atmosphere as a result of glacial retreat. In this field-based study, we investi-gated the spatial and temporal variability of atmospheric-CH4 oxidation and associ-ated MOB communities in Alpine glacier forefield soils, aiming at better understand-ing the factors that shape the sink for atmospheric CH4 in this young soil ecosystem.This study contributes to the knowledge on the dynamics of atmospheric-CH4 oxida-tion in developing upland soils and represents a further step toward the inclusion ofAlpine glacier forefield soils in global CH4 inventories.

KEYWORDS atmospheric-methane oxidation, glacier forefield soil, high-affinity MOB,methanotroph, methane flux, proglacial landforms, pmoA

Received 22 May 2017 Accepted 30 June2017

Accepted manuscript posted online 7 July2017

Citation Chiri E, Nauer PA, Rainer E-M, Zeyer J,Schroth MH. 2017. High temporal and spatialvariability of atmospheric-methane oxidationin Alpine glacier forefield soils. Appl EnvironMicrobiol 83:e01139-17. https://doi.org/10.1128/AEM.01139-17.

Editor Patrick D. Schloss, University ofMichigan—Ann Arbor

Copyright © 2017 American Society forMicrobiology. All Rights Reserved.

Address correspondence to Martin H. Schroth,martin.schroth@env.ethz.ch.

* Present address: Eleonora Chiri, School ofEcosystem and Forest Sciences, BurnleyCampus, University of Melbourne, Melbourne,VIC, Australia; Philipp A. Nauer, School ofEcosystem and Forest Sciences, BurnleyCampus, University of Melbourne, Melbourne,VIC, Australia; Edda-Marie Rainer, Departmentof Arctic and Marine Biology, University ofTromsø, Tromsø, Norway.

ENVIRONMENTAL MICROBIOLOGY

crossm

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 1Applied and Environmental Microbiology

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

Aerobic oxidation of the potent greenhouse gas methane (CH4) is one of manyimportant ecosystem services provided by soils (1, 2). It is mediated by methane-

oxidizing bacteria (MOB), which can utilize CH4 as a sole source of carbon and energy(3). The activity of MOB in soils can substantially attenuate CH4 emissions to theatmosphere from natural and anthropogenic sources, such as wetlands, rice paddies,and landfills (4–6). Moreover, in well-drained (upland) soils, so-called “high-affinity”MOB are capable of utilizing atmospheric CH4 at ambient concentrations (�1.8 �lliter�1) (7). As a result, upland soils are usually a sink for atmospheric CH4. In fact, theyare CH4’s only known terrestrial sink, contributing �5 to 15% of the total loss of CH4

from the atmosphere (8–10).As cultivation attempts with high-affinity MOB have been mostly unsuccessful, to

date (Methylocystis sp. strain SC2 is the only potential high-affinity MOB isolate [11]),molecular and biochemical methods are employed for their identification in uplandsoils. Identification is typically based on the amplification of the pmoA gene, a widelyused MOB biomarker that closely reflects phylogenies based on 16S rRNA genesequences (e.g., see reference 12). The pmoA gene encodes a subunit of the particulateform of the enzyme methane monooxygenase (pMMO), which catalyzes the first stepin the oxidation of CH4 (3, 13). Based on pmoA, high-affinity MOB have mainly beenassigned to the upland soil cluster (USC) alpha and gamma clades, and it was notedthat they are distantly related to two groups of cultivable MOB, i.e., type II and type I,respectively (14–16). High-affinity MOB have been studied intensively in mature uplandsoils (e.g., tundra, forests, grasslands, savannahs, and deserts [17–21]), but only limitedinformation is available on their performance in young, developing upland soils (e.g.,recent volcanic deposits or proglacial sediments).

Whereas developing upland soils are still excluded from national and global CH4

inventories (e.g., see references 8 and 22), recent studies indicate that they may providea substantial sink for atmospheric CH4. For example, Hawaiian volcanic deposits(andosols [FAO soil classification]) were found to be a sink for atmospheric CH4 already40 years after deposition. They exhibit CH4 uptake rates (expressed by convention asnegative soil-atmosphere CH4 flux values) of �1.8 to �0.7 mg CH4 m�2 day�1, similarto those of mature forest and grassland soils (22). Atmospheric-CH4 oxidation was alsodetected in proglacial sediments from both Arctic (23) and Alpine (24, 25) environ-ments. Specifically, young Alpine proglacial sediments, referred to here as glacierforefield soils (lithosols [FAO soil classification]), can provide a substantial sink foratmospheric CH4 already at an early stage of soil development (�15 years; up to �0.7mg CH4 m�2 day�1) (26). As the extent of glacier forefields is expected to increasefurther with ongoing glacial retreat (26, 27), it is important to better understand thefactors modulating community composition and activity of high-affinity MOB in thisdeveloping soil environment.

Glacier forefields are heterogeneous and dynamic environments which exhibitfeatures that render their soils interesting but challenging systems for the study ofmicrobial structures and functions. First, glacier forefields often feature a well-definedsequence of soil ages (soil chronosequence) (e.g., see references 28 to 30). This is aresult of glacial retreat, when sediments are successively subjected to a radical shiftfrom a subglacial to a proglacial environment, whereby soil formation is initiated. Thus,a soil chronosequence is a good model ecosystem for studies of microbial primarysuccession (e.g., see reference 31). Second, glacier forefield soils can be derived fromdiverse bedrock types. As bedrock type is a key factor determining nutrient availability,it may shape microbial communities in glacier forefield soils (31, 32). Third, the cooccur-rence of different geomorphological processes (e.g., glacial erosion and deposition, debrismovement, and physical and chemical weathering) can lead to the formation of numer-ous proglacial landforms (33, 34). Individual landforms often exhibit specific physico-chemical properties, which may affect microbial community composition and activity.Finally, glacier forefield soils are also subject to dramatic short-term and seasonalvariability in, e.g., physical parameters such as soil temperature and water content,which are known to affect the performance of microbial communities, including

Chiri et al. Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 2

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

high-affinity MOB (e.g., see reference 9). Whereas such variability is usually low duringthe snow-covered season (35, 36), with the melting of snow the glacier forefield soilsundergo sudden changes in soil water content, and they can be subject to largefluctuations in temperature and water regimens throughout the snow-free season (e.g.,see references 37 to 39).

In a recent field study, we investigated atmospheric-CH4 oxidation and MOB com-munity composition along soil chronosequences (soil age, 6 to 120 years) in two Swissglacier forefields situated on contrasting (siliceous versus calcareous) bedrock types(39). Soil age was found to be the main factor affecting atmospheric-CH4 uptake, whichincreased with increasing soil age and ranged from �2.2 to �0.08 mg CH4 m�2 day�1.In contrast, observed differences in MOB community composition were related mainlyto bedrock type rather than to soil age, indicating that distinct, low-diversity MOBcommunities provided similar ecosystem services in the two forefields. However, asfield sampling was restricted to two time points during the snow-free season and asingle proglacial landform, the temporal variability of and effects of different landformson atmospheric-CH4 oxidation and MOB community composition in glacier forefieldsoils remained unknown.

Thus, the objectives for the present study were (i) to investigate temporal variabilityin atmospheric-CH4 oxidation and MOB community composition as a function of soilage along an established glacier forefield soil chronosequence, (ii) to assess the effectsof different landforms on the soil CH4 sink within a single soil age class, and (iii) toprovide an improved, high-resolution analysis of MOB diversity in glacier forefield soils.To this end, we conducted field campaigns to quantify soil-atmosphere CH4 fluxthroughout the snow-free season of 2013 (Fig. 1), and targeted amplicon sequencingwas used to assess MOB diversity and community composition in soil samples collectedduring this period. We employed linear mixed-effects (LME) models to test the depen-dence of the response variables soil-atmosphere CH4 flux and MOB abundance on themodel predictors soil age, landform, sampling time point, soil temperature, soil watercontent, and cumulative rainfall.

RESULTSSoil physical parameters. Soil water content was low at all sampling locations and

time points, indicating that glacier forefield soils remained relatively dry throughout thesnow-free season (Fig. 2a). Data from locations A to C indicate that topsoil was

FIG 1 Sampling locations at the Griessfirn Glacier forefield, mapped using Swiss square-projectioncoordinates (CH1903). Soil chronosequence locations are delimited according to soil age class (A, 0 to 20years; B, 20 to 50 years; and C, 50 to 120 years). (Inset) Landform locations distinguish sandhill (S), terrace(T), and floodplain (F) landforms.

Atmospheric-Methane Oxidation in Alpine Soils Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 3

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

generally drier than bulk soil, with seasonal mean water contents ranging from 0.06 to0.09 m3 m�3 for topsoil and from 0.10 to 0.12 m3 m�3 for bulk soil. Furthermore, topsoilwater content increased with increasing soil age (from locations A to C) (Fig. 2a). Thehighest temporal variability in soil water content at individual locations was measuredin soil age class A (topsoil, 0.04 to 0.09 m3 m�3; bulk soil, 0.06 to 0.19 m3 m�3) (datanot shown). Among landform locations, the highest topsoil water content was ob-served at location F sites (Fig. 2a). Water content data for location S sites agreedreasonably well with topsoil data measured in close proximity, at location B sites, althoughdifferent measurement techniques were employed.

FIG 2 Heat maps showing median values for volumetric soil water content (a), soil temperature (b), andsoil-atmosphere CH4 flux (c) measured at soil chronosequence locations (A, B, and C) and landformlocations (S, T, and F). For soil temperature and water content, “t” refers to topsoil measurements, and“b” refers to averaged (bulk soil) measurements across all depths. The column at right shows seasonalmean values � 1 SD; white areas indicate that no measurements were performed on the correspondingdates.

Chiri et al. Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 4

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

After an initial increase in July following snow melting, the soil temperature de-creased during the remainder of the snow-free season (from August to October) (Fig.2b). During this period, the median topsoil temperature at locations A to C decreased�12°C, on average, whereas the bulk soil temperature decreased �8°C, on average.Moreover, bulk soil temperature moderately increased with increasing soil age (fromlocations A to C).

Rainfall events were quite evenly distributed throughout the sampling season, witha maximum of six dry days between any two consecutive events (see Fig. S2 in thesupplemental material). Calculated cumulative rainfall rates ranged from 0.7 to 13.2 mmday�1, with a mean value of 5.1 mm day�1.

Soil-atmosphere CH4 flux. (i) Chronosequence locations A to C. Nearly all CH4

concentrations measured in soil gas profiles were subatmospheric (example profiles areshown in Fig. S3), with a few measurements in the youngest soils falling at or slightlyabove the atmospheric value. As a result, soils at locations A to C exhibited mostly netCH4 uptake during the sampling season (Fig. 2c). Median fluxes on individual samplingdates ranged from �2.34 to �0.03 mg CH4 m�2 day�1, with no apparent temporaltrend. However, CH4 uptake increased substantially with increasing soil age. Forexample, the seasonal mean uptake rate was �0.30 mg CH4 m�2 day�1 for location Asites and �1.31 mg CH4 m�2 day�1 for location C sites (Fig. 2c). Conversely, thetemporal variability of soil CH4 concentrations and flux at individual locations washighest in the youngest soils (location A sites) and decreased with increasing soil age(Fig. S3). The highest spatial variability in CH4 flux was also measured at location A sites(not shown).

(ii) Landform locations S, T, and F. The three landforms exhibited substantialdifferences in soil-atmosphere CH4 flux (Fig. 2c). Sites at location S exhibited substantialCH4 uptake throughout the sampling season (median CH4 flux of �0.54 to �0.11 mgCH4 m�2 day�1). However, on comparing CH4 uptake rates between adjacent locationsS and B (Fig. 1), estimates for location B sites were �3-fold larger than the CH4 uptakemeasured at location S sites. This may be due in part to the different methodsemployed to estimate/measure CH4 flux. A slight apparent trend of decreasing CH4

uptake with the ongoing sampling season at location S sites (Fig. 2c) was refuted by theLME model (see below). Uptake of CH4 measured at locations T and F was substantiallyless than that at location S. In fact, positive CH4 flux values for locations T and F evenindicated intermittent net CH4 emissions, particularly early in the sampling season (Fig.2c). Similar spatial variabilities in CH4 flux were observed for all landforms (not shown).

Abundance of MOB. High-quality genomic DNA was isolated from all samples butS1-Jul25, which was consequently excluded from all further molecular analyses. pmoAcopy numbers spanned 3 orders of magnitude, ranging from 8 � 102 to 5 � 105 copiesper g of dry soil (Fig. 3). At locations A to C, pmoA copy numbers increased substantiallywith soil age, and the highest copy numbers were detected in location C samples. Thehighest pmoA copy numbers among different landforms were obtained for location Ssamples (similar in magnitude to those for location B samples), whereas landformlocation F samples exhibited the lowest pmoA copy numbers, which were slightly lowerthan the copy numbers for location T samples. Notably, pmoA copy numbers forlocations C and F differed significantly from those for other sampling locations (A, B, S,and T) (based on Mann-Whitney-Wilcoxon tests). Whereas location F samples showedthe lowest variability in pmoA copy number during the sampling season, location Csamples by far exhibited the highest spatial and temporal variability (Fig. 3).

Diversity of MOB communities. We obtained an average of �130,000 pmoA

sequences per sample, distributed among sampling locations as follows: 87,000 �

43,000 (mean � standard deviation [SD]) at location A sites, 115,000 � 43,000 atlocation B sites, 184,000 � 82,000 at location C sites, 131,000 � 26,000 at location Ssites, 154,000 � 78,000 at location T sites, and 147,000 � 66,000 at location F sites.High-throughput sequencing of pmoA genes allowed identification of 31 operationaltaxonomic units (OTUs). Analysis of the phylogenetic distance of the protein-derived

Atmospheric-Methane Oxidation in Alpine Soils Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 5

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

pmoA partial sequences showed that most retrieved OTUs grouped with MOB-likesequences (Fig. 4). Half of the MOB-like OTUs grouped with either type Ic (mostlyUSC-gamma) or type IIb (mostly USC-alpha) sequences. Twenty percent of the MOB-likeOTUs clustered with type Ib, 13% with type IIa, and 7% with type Ia. The remaining 10%of OTUs clustered with the pmoA/amoA-like group, designated for sequences clusteringbetween the pmoA gene and the homologous amoA gene of ammonia-oxidizingbacteria. The applied taxonomic system for pmoA genes follows one reported previ-ously (40). Thirty percent of the MOB-like OTUs were previously undetected pmoAsequences, which may represent novel species. Specifically, OTUs 07, 18, and 28branched with USC-gamma but showed low nucleotide sequence identity with knownpmoA sequences (79 to 86%) (Fig. 4). Similarly, OTUs 09 and 23 branched with type IIband showed identities of 75% and 84%, respectively, with publicly available pmoAsequences. The OTUs grouping with pmoA/amoA-like sequences and OTU 31 (type IIa)also showed low identities with known pmoA sequences.

The presence and relative abundances of OTUs indicated the highest variability andalpha diversity at locations A to C and S (all part of the sandhill landform) (Fig. 5),whereas the lowest variability and alpha diversity were measured at locations T and F.Among the few exceptions were the ubiquitous and most abundant OTUs 01 (type Ic)and 02 (type IIb). Thirty-five percent of the retrieved OTUs were found only in thesandhill landform, with some OTUs being further specific to location C sites.

Analysis of the beta diversity of MOB communities in glacier forefield soils revealedsimilar results, i.e., significant differences in community composition related to land-form (permutational multivariate analysis of variance [PERMANOVA]; P � 0.04) (Fig. 6).Based on pairwise tests among landform locations, MOB communities in location Ssamples differed significantly from communities in both location F (P � 4 � 10�3) andT (P � 6 � 10�3) samples, whereas no significant differences in community composi-tion were detected between locations S and A to C. In fact, the high total variability inMOB community composition in location S samples put these communities closer tothose at locations A to C than to those at locations T and F (Fig. 6). Total variability wascomprised of both spatial and temporal variabilities; the latter was particularly notice-able at locations A and B (Fig. S4). Conversely, MOB community compositions atlocations T and F were almost identical (with T and F data points plotting in the sameregion in space [Fig. 6]) and showed little variability.

Linear mixed-effects model. Several trends perceived in the soil physical data wereconfirmed to be significant by the LME model. For example, topsoil water content at

FIG 3 Box-and-whisker plot showing the total (spatial and temporal) variabilities in MOB abundance(pmoA gene copy number per gram of dry soil) for all samples collected for molecular analyses at soilchronosequence (A, B, and C) and landform (S, T, and F) locations (Table S1). The bottom and top of eachbox indicate the first and third quartiles, and the horizontal line inside the box shows the second quartile(median). Whiskers indicate maximum and minimum values.

Chiri et al. Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 6

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

FIG 4 Maximum likelihood tree with 100-bootstrap support showing the phylogenetic affiliation of the pmoA gene based on thederived amino acid sequence (partial sequence [150 amino acids]). Bootstrap numbers are shown for branches with bootstrapsupport of �50. Operational taxonomic units (OTUs) retrieved in this study are depicted in bold. GenBank accession numbers forrepresentative sequences deposited in the database are given in parentheses. Clusters that do not include OTUs from this study arecollapsed. The scale bar represents 0.2 change per amino acid position.

Atmospheric-Methane Oxidation in Alpine Soils Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 7

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

locations A to C increased significantly with increasing soil age, whereas soil watercontent at locations S, T, and F was significantly influenced by the predictors samplingtime point and landform (Table 1). On the other hand, soil temperature significantlydecreased over the sampling season at all locations, whereas it significantly increasedwith increasing soil age at locations A to C.

Soil CH4 uptake significantly increased with increasing soil age (locations A to C)(Table 1). In addition, CH4 flux was significantly affected by landform (locations S, T, andF). The results of the LME model further revealed that other than soil age and landform,the only predictor significantly influencing CH4 flux was cumulative rainfall, whereassoil temperature and soil water content were not significant (P � 0.05) predictors (notshown). The perceived trend of decreasing CH4 uptake with the ongoing samplingseason for location S sites was refuted as not significant by the LME model.

The results of the LME model fits further indicated that the observed increase inMOB abundance with soil age at locations A to C was statistically significant and thatMOB abundance was also significantly affected by landform (locations S, T, and F)(Table 1). As in the case of CH4 flux, soil temperature and soil water content did notsignificantly influence MOB abundance (not shown).

DISCUSSIONTemporal variability of the soil CH4 sink as a function of soil age. Soil-

atmosphere CH4 flux exhibited substantial temporal variability at locations A to Cduring the snow-free season. This variability was clearly attenuated with increasing soilage (Fig. 2c; see Fig. S3 in the supplemental material). Thus, not only is soil age the mainfactor affecting the strength of the soil CH4 sink (39), but increasing soil age alsocontributes to the stability of atmospheric-CH4 uptake in these developing soils. Inthis context, soil age serves as a proxy for all edaphic factors that change with soildevelopment. The importance of soil age in modulating the soil CH4 sink wascorroborated by our LME model, in which soil age was the main environmentalfactor (P � 5.7 � 10�3) (Table 1) explaining CH4 flux.

FIG 5 Heat map showing the presence and relative abundances of operational taxonomic units (OTUs) retrieved throughtargeted pmoA gene amplicon sequencing. Individual samples are displayed on the x axis for soil chronosequence (A, B, andC) and landform (S, T, and F) locations, separated by a dashed line. OTUs shown on the y axis are grouped according to theirphylogenetic affiliation (40). The average alpha diversity value � 1 SD (Simpson index [D]; one-dimensional notation) is shownfor all sampling locations. Higher values indicate higher diversity.

Chiri et al. Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 8

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

Temporal variability in soil physical parameters (Fig. 2a and b) appeared to havelittle effect on CH4 flux in glacier forefield soils. A weak dependence on soil temperaturewas previously explained by gas diffusion being the main, but only mildly temperature-dependent, factor limiting atmospheric-CH4 uptake (9, 18). Low soil water content andthus a high CH4 availability in these fast-draining glacier forefield soils (39) may partiallyexplain the lack of an expected dependence of CH4 flux on soil water content (e.g., seereference 9). In addition, small but potentially important variations in soil water contentmay have been masked by the measurement uncertainty (0.04 m3 m�3), which wasoften large compared to measured values, in particular for topsoils (0.038 to 0.12 m3

m�3) (Fig. 2a).Conversely, atmospheric-CH4 uptake at locations A to C showed a significant,

positive dependence on cumulative rainfall (P � 0.015) (Table 1). At first glance, thisresult appears to be at odds with our findings for soil water content. However, it mayindicate that atmospheric-CH4 uptake in these soils was at least occasionally limitedby water availability, i.e., when the latter dropped below the lower limit of favorableconditions (�20% water saturation) (41, 42). At a low soil water content, water existsprimarily in small pores and as films coating mineral particles (43, 44). Throughcontinued evaporation, water films may disappear and soil water be present only in theform of small pendular rings between particles (45, 46). Under these conditions,microorganisms inhabiting particle surfaces will be faced with highly unfavorableconditions, which can lead to cessation of their primary metabolism. Cumulative rainfallcan be seen as an integrative parameter accounting for recent rainfall events, whichreplenish water films (47) and thus contribute to maintaining or reestablishing micro-bial activity. Therefore, MOB survival and activity may strongly depend on cell distri-

FIG 6 Principal coordinate analysis of MOB community beta diversity calculated with the weightedUniFrac metric. Average MOB community beta diversity values are shown for chronosequence (A, B, andC) and landform (S, T, and F) locations, with error bars (� 1 SD) representing total, i.e., spatial andtemporal, variability. Together, the PCoA 1 and PCoA 2 axes explain 94.3% of the total variance. AverageMOB community dissimilarities between and within sampling locations (A, B, C, S, T, and F) are displayedin the inset.

Atmospheric-Methane Oxidation in Alpine Soils Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 9

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

TAB

LE1

Pva

lues

ofLM

Em

odel

sfit

ted

tosa

mp

letim

ese

ries

ofch

rono

sequ

ence

(Ato

C)

and

land

form

(S,T

,and

F)lo

catio

nsa

Pred

icto

r

Cor

rela

tion

,Pva

lue

for

resp

onse

vari

able

b

Soil-

atm

osp

her

eC

H4

fluxc

MO

Bab

und

ance

Soil

wat

erco

nte

nt

Soil

tem

p

Loca

tion

sA

toC

Loca

tion

sS,

T,an

dF

Loca

tion

sA

toC

Loca

tion

sS,

T,an

dF

Loca

tion

sA

toC

Loca

tion

sS,

T,an

dF

Loca

tion

sA

toC

Loca

tion

sS,

T,an

dF

bt

tb

tt

Soil

age

�,5

.7�

10�

3N

A�

,7.5

�10

�3

NA

NS

�,5

.3�

10�

3N

A�

,2.5

�10

�5

�,4

.9�

10�

3N

ALa

ndfo

rmd

NA

2.8

�10

�5

NA

1.3

�10

�4

NA

NA

0.01

3N

AN

AN

SSa

mp

ling

time

poi

ntN

SN

SN

SN

SN

SN

S�

,1.8

�10

�3

�,2

.2�

10�

16

�,2

.2�

10�

12

�,1

.86

�10

�5

Cum

ulat

ive

rain

fall

�,0

.015

�,0

.032

�,0

.039

NS

NS

NS

NS

NA

NA

NA

aW

ete

sted

the

dep

ende

nce

ofth

ere

spon

seva

riab

les

soil-

atm

osp

here

CH

4flu

xan

dM

OB

abun

danc

eon

the

pre

dict

ors

soil

age,

land

form

,sam

plin

gtim

ep

oint

,cum

ulat

ive

rain

fall,

soil

tem

per

atur

e,an

dvo

lum

etric

soil

wat

erco

nten

t(t

heda

tafo

rth

ep

redi

ctor

sso

ilte

mp

erat

ure

and

soil

wat

erco

nten

tar

eno

tsh

own,

asin

divi

dual

Pva

lues

wer

eno

tsi

gnifi

cant

).In

addi

tion,

we

also

test

edth

ede

pen

denc

eof

bul

kso

il(b

)an

dto

pso

il(t

)te

mp

erat

ure

and

wat

erco

nten

ton

soil

age,

land

form

,sam

plin

gtim

ep

oint

,and

cum

ulat

ive

rain

fall.

b�

,pos

itive

corr

elat

ion;

�,n

egat

ive

corr

elat

ion;

NS,

not

sign

ifica

nt(P

�0.

05);

NA

,not

app

licab

le.

c Not

eth

atC

H4

upta

keis

exp

ress

edb

yco

nven

tion

asa

nega

tive

soil-

atm

osp

here

flux

valu

e.Th

us,a

nega

tive

corr

elat

ion

ofso

il-at

mos

phe

reflu

xw

itha

pre

dict

or(e

.g.,

soil

age)

indi

cate

sa

pos

itive

corr

elat

ion

ofC

H4

upta

kew

ithth

ese

pre

dict

ors,

and

vice

vers

a.dTh

edi

rect

ion

ofco

rrel

atio

nis

not

app

licab

le.

Chiri et al. Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 10

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

bution among microhabitats and, potentially, small changes in soil water content (e.g.,see reference 48).

Total variability in MOB community composition at locations A to C (Fig. 6) includeda substantial amount of temporal variability (Fig. S4), which is in agreement withprevious findings on the variability of total microbial community composition in otherglacier forefields (e.g., see reference 49). In particular, MOB communities sampled atlocations B1 and B2 during our first campaign (June 18) markedly differed fromcommunities sampled at later time points at these locations. This shift in MOB com-munity composition may be a consequence of changes in environmental conditionsthat occurred during and immediately after snow melting (50). Although high-affinityMOB are thought to be slow-growing microorganisms (51), the observed temporalvariability in community composition indicated that they are capable of promptlyadapting to changes in environmental conditions. This may be facilitated, for instance,by the ubiquitous Methylocystis-like MOB (Fig. 4 and 5), which may switch to a dormantstage (cysts) under unfavorable conditions and rapidly initiate excystation once con-ditions become favorable, thus bypassing the growing phase (e.g., see reference 52).

Effect of landform on the soil CH4 sink. Our measurements indicated substantialdifferences between landforms in their contributions to the overall soil CH4 sink ofglacier forefields; seasonal mean CH4 uptake at locations F and T was significantly lowerthan that at location S (Fig. 2). As our measurements represent net fluxes, i.e., the sumof simultaneous CH4 transport, production, and consumption (53), differences in any ofthese processes may explain differences in CH4 flux between the landforms. Althoughwe were unable to distinguish between individual processes, occasionally observedCH4 emissions at locations F and T support the presence of a CH4 source of as yetunknown origin in these calcareous glacier forefield soils (54). We found no evidence ofmicrobial CH4 production on screening of the top 15 cm of soils at locations S, F, andT for presence of the mcrA gene, a biomarker for methanogenic Archaea (data notshown). However, we cannot fully exclude the possibility of microbial CH4 productionin deeper, water-logged soils at locations F and T. Alternatively, the CH4 source may berelated to slow dissolution of carbonate rock or particle aggregates in water-loggedsoils, thereby releasing entrapped CH4 (54).

Low CH4 uptake at locations F and T may also have resulted from low CH4

availability due to gas diffusion limitation, particularly at location F (9), or from lowerMOB abundances at locations F and T than at location S. The latter may be related todifferences in soil texture between landforms. A particle size analysis of the �2-mmfraction of soils from locations S, F, and T showed that the clay and silt contents weresignificantly higher in the location S samples (E.-M. Rainer, E. Chiri, and M. H. Schroth,unpublished data). Bacterial biomass associated with mineral particles has been shownto be substantially higher in clay-silt minerals than in sand (55, 56), likely becausemicrohabitats in fine mineral aggregates provide more favorable conditions for bacte-rial functioning (e.g., see reference 57). Fine mineral aggregates may have been washedout of soils at locations F and T as a result of flash floods during snow melting orsummer rainstorms.

Diversity of glacier forefield MOB communities. Our study employed a high-throughput sequencing technique (amplicon sequencing on a MiSeq platform [Illu-mina]) to investigate pmoA gene-based MOB diversity in young, developing glacierforefield soils. Using this technique, the number of identified OTUs was �6-fold largerthan that in MOB diversity studies based on pmoA gene sequencing of clone libraries(25, 58).

Functional gene diversity can be low in extreme environments, such as glacierforefields (59). Nonetheless, we retrieved 31 OTUs (of which 30% were previouslyundetected pmoA sequences), which embrace all phylogenetic affiliations identified inprevious studies. The USC-gamma clade was most prominent in all samples in terms ofpresence and relative abundance, as reported in a previous study of Alpine glacierforefield soils (25). These data confirm that high-affinity, USC-gamma MOB are wide-

Atmospheric-Methane Oxidation in Alpine Soils Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 11

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

spread in Alpine soil environments (12, 39, 60). In addition, high-affinity MOB belongingto USC-alpha were prominently detected in the Griessfirn Alpine glacier forefield, likelyas a result of using a high-throughput sequencing technique. Three other OTUsgrouped with the Methylocystis-like cluster, including the ubiquitous and second mostabundant OTU, OTU 02. This agrees with previous findings in which a well-representedMethylocystis-like OTU was exclusively retrieved from glacier forefield soils on calcare-ous bedrock (25). It may indicate that Methylocystis-like MOB possess an ecologicaladvantage in calcareous glacier forefield soils, which may be linked to the existence oftwo pMMO isoenzymes, with different affinities for low (2 to 600 �l liter�1) and high(600 �l liter�1) CH4 concentrations, previously detected in cultures of a Methylocystisstrain (11). Thus, Methylocystis-like MOB may profit from transient release of CH4 (atelevated concentrations) entrapped in calcareous soil aggregates (54).

Conclusions. This study contributes to our knowledge on the dynamics ofatmospheric-CH4 oxidation and MOB communities in developing soils and providessupporting evidence to include mountainous lithosols in global CH4 inventories. Dif-ferent landforms and their relative extents in a glacier forefield clearly modulate theoverall CH4 sink strength. We confirmed that the soil CH4 sink and the MOB communitydriving the latter are subject to temporal and spatial variability during the snow-freeseason. Further work is needed to elucidate the underlying cause of this variability.Hence, to fully characterize atmospheric-CH4 oxidation in this heterogeneous environ-ment, an integrative approach will be required, combining, e.g., areal imagery andlandscape classification maps with eddy covariance measurements of CH4 flux. Finally,field studies of CH4 oxidation during the snow-covered season, as well as investigationsof interannual changes in CH4 oxidation, are still needed to give more comprehensiveknowledge on atmospheric-CH4 oxidation dynamics in Alpine environments.

MATERIALS AND METHODSField site. We performed sampling and measurements in the forefield of Griessfirn Glacier (Canton

Uri, Switzerland), which has been described extensively elsewhere (39, 54). Sampling locations werepositioned along two transects (Fig. 1). For the first transect, eight locations were selected along awell-defined soil chronosequence with increasing distance from the glacier terminus, on a band of lateraldebris deposits (a landform referred to here as a sandhill). They represent a subset of sampling locationspreviously installed and categorized into three soil age classes (location A, 0 to 20 years; location B, 20to 50 years; and location C, 50 to 120 years) (39). In the present study, each soil age class comprised 2or 3 sampling locations.

The second transect was located in soil age class B and comprised five sampling locations each infloodplain (F), terrace (T), and sandhill (S) landforms (Fig. 1, inset), which were identified based ontopographical features. Floodplains may be described as streamlined bed forms parallel to the glacialstream, which feature a shallow groundwater table. In contrast, terraces are elevated, ancient floodplainsthat present dryer conditions with a deeper groundwater table, but otherwise they exhibit a structuresimilar to that of floodplains. Finally, sandhills are depicted as an unoriented hummocky landformexhibiting a disorganized deposition pattern and the deepest groundwater table among the threelandforms (34).

Sampling and measurement procedures. At locations A to C, we employed the soil gas profilemethod to determine soil-atmosphere CH4 flux (milligrams of CH4 per square meter per day) (e.g., seereference 19). To this end, depth-resolved soil gas samples were collected using a polyuse multilevelsampling system (61). Details of the installation, sampling, and measurement procedures were describedpreviously (54). Soil gas sampling was performed under dry weather conditions during nine samplingcampaigns in 2013 (June to October) (see Table S1 in the supplemental material). Sampling started onJune 18, when most of the glacier forefield was still snow covered, followed by more frequent samplingshortly after the snow melted (July 2 to 18). The sampling frequency was reduced toward the end of thesnow-free season. Concurrent with soil gas sampling, we measured the depth-resolved volumetric soilwater content (cubic meters per cubic meter of soil) (PR2/6 capacitance probe; Delta-T Devices Ltd.,Cambridge, United Kingdom) at each location, and we recorded the depth-resolved soil temperature(iButton temperature loggers; Maxim Integrated, San Jose, CA) in 1-h intervals throughout the samplingseason for one location per soil age group (Table S1). We report topsoil water content and temperatureby using the uppermost measurement points (7.5 to 10 cm in depth) and bulk soil values by using themean for all measured depths (7.5 to 97.5 cm).

For locations S, T, and F, we used the static flux chamber method to quantify CH4 flux (e.g., seereference 62). Deployment of polyvinyl chloride chambers (31-cm diameter � 27-cm height) took placeshortly after the snow melted (July 9 and 10). Chambers comprised a base collar inserted �15 cm intothe ground. To allow soil consolidation around the collars, an idle phase of 8 days (which included severalrainfall events) followed chamber installation. Flux chamber measurements were performed during sixsampling campaigns (July 18 to September 19) (Table S1). For measurements, collars were fitted with

Chiri et al. Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 12

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

gaskets and detachable lids, resulting in chamber headspace volumes ranging from 9.2 to 11.4 liters.Each lid featured an aluminum foil cap, to minimize temperature increases in the headspace from solarirradiation, and a port connected to a three-way valve, which allowed extraction of 20-ml gas samplesfrom the headspace by use of a gastight syringe. Measurement periods lasted 90 min, with four gassamples collected at regular time intervals. Gas samples were immediately transferred to serum bottlescapped with butyl rubber stoppers for subsequent analysis of the CH4 concentration. Concurrent with fluxchamber measurements, topsoil water content was quantified by time domain reflectometry (TDR100;Campbell Scientific, Loughborough, United Kingdom) at 2 or 3 spots per landform in the close vicinity ofthe flux chambers, using pairs of 30-cm-long brass rods permanently installed in the ground. Nearby, thetopsoil temperature was measured at a depth of 11 cm at 1-h intervals, using iButton temperatureloggers mounted on wooden rods.

A total of 56 soil samples were collected for molecular analyses on 6 days over the course of thesampling season (Table S1). Soil was collected from all eight sampling sites at locations A to C and froma total of nine sampling sites at locations S, T, and F. Details of the soil-sampling procedure weredescribed previously (39). Individual samples and measurements are referred to by sampling location (A,B, C, S, F, or T) and number (1 to 5) and by sampling time point (date), e.g., A1-Jul02 or F5-Jul25.

Data for daily rainfall (millimeters per day) were obtained from the nearest automated weatherstation of the Federal Office of Meteorology and Climatology (MeteoSwiss). Cumulative rainfall (milli-meters per day) for each sampling date was calculated as the weighted sum of daily rainfall within 4 daysprior to sampling, using an exponential decay function with a decay constant of 1 day�1 for weighting(i.e., the effect of preceding rainfall was halved every 16.6 h).

Determination of soil-atmosphere CH4 flux. Methane concentrations in all soil gas samples werequantified on a gas chromatograph equipped with a flame ionization detector (63). To determine soil-atmosphere CH4 flux at locations A to C, analytical solutions to a steady-state diffusion-reaction model werefitted to individual soil CH4 profiles in the R v3.2.1 software environment (64). Analytical solutions assumedCH4 oxidation to be governed either by a single first-order reaction over the entire profile (65) (referredto here as the one-layer model) or by individual first-order reactions in two distinct (top and bottom) soillayers (39) (referred to here as the two-layer model). Applying Fick’s first law of diffusion, we thencomputed the flux from the CH4 concentration gradient at the soil-atmosphere boundary and best-fitparameters of that model, which yielded better agreement with measured soil CH4 profiles. In caseswhere model convergence failed, CH4 concentration gradients and fluxes were approximated bylinear regression (39). For locations S, T, and F, flux values were obtained from the slope of CH4

concentrations in the flux chambers plotted against measurement time by linear regression analysis(e.g., see reference 66).

pmoA gene marker-based molecular analyses of the MOB community. (i) Total DNA isolationand pmoA gene amplification. Genomic DNA was isolated using a FastDNASpin kit for soil (MPBiomedicals, Solon, OH) according to the manufacturer’s instructions, with minor modifications. Recoveryand purity of the DNA extracts were tested prior to further molecular analyses. Applied protocols for DNAextraction and quality control were described previously (39). Amplification of the pmoA gene was thefirst step for all subsequent molecular techniques applied in this study. Primer sequences and amplifi-cation procedures are reported in Table S2.

(ii) qPCR. To determine MOB abundance, copies of pmoA genes in DNA extracts were quantified byquantitative PCR (qPCR) on an ABI 7500 system (Applied Biosystems [now Thermo Fischer Scientific],Waltham, MA). Quantification of pmoA copy numbers followed a previously described procedure (67) andthe assay specifications shown in Table S2. All samples were analyzed in triplicate. The efficiencies of thethree assays performed were always �100%, with R2 values above 0.99.

(iii) Targeted amplicon sequencing. For each sample, we prepared, indexed, and paired-endsequenced amplicon libraries of the pmoA gene from genomic DNA extracts according to the targetedamplicon sequencing method (68). Primer sequences and amplification procedures are reported in TableS2. Amplicon library preparation, high-throughput amplicon sequencing, and sequence data processingare described in detail in the supplemental material.

Statistical analyses. (i) Linear mixed-effects models. Using lme4 v1.1-10 (69) in R v3.2.1, LMEmodels were fitted to sample time series, with random effects included for individual sampling locations.The drop1 function was used to find the optimal LME model for each response variable, based on theAkaike information criterion. We tested the dependence of the response variables CH4 flux and MOBabundance on the model predictors soil age, landform, sampling time point, cumulative rainfall, soiltemperature, and water content. Initial checks indicated a correlation between soil temperature andsampling time point, and thus an additional interaction term between these variables was included.Where residual diagnostics indicated nonlinearity and/or nonnormality, variables were log transformed(soil age, MOB abundance, cumulative rainfall, and sampling time point) or arcsine transformed (all watercontent data). Missing values were replaced with the mean for the respective soil age group or landform.

(ii) Diversity and structure of MOB communities. Phylogenetic distances of the assigned opera-tional taxonomic units (OTUs) were assessed through nucleotide sequence alignment and phylogenetictree building on the derived-protein level, using the software Seaview v4.5.4 (70). To identify the bestevolution model, we tested 96 amino acid substitution models by using the software ModelGeneratorv0.851 (71) and selected the one showing the lowest value for the Akaike information criterion. Aphylogenetic tree was built according to the maximum likelihood method, with 100-bootstrap support,using the phylogeny software PhyLM v3.1 (72).

To assess the alpha diversity of MOB communities, we computed Simpson indices (D) and visualizedthe results in one-dimensional notation, in which alpha diversity ranges from 0 (low diversity) to 1 (high

Atmospheric-Methane Oxidation in Alpine Soils Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 13

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

diversity). We further employed Mann-Whitney-Wilcoxon tests to assess differences in alpha diversitybetween sample locations (A, B, C, S, T, and F). To identify factors explaining differences in communitystructure, we analyzed the beta diversity of MOB communities (defined as the variation in compositionbetween a set of communities) coupled with standard multivariate statistics (73, 74). Alpha and betadiversity calculations, as well as read count normalization of the pmoA sequences, were performed withthe package phyloseq v1.12.2 (75) from the open source software Bioconductor. We applied a read countthreshold of �35,000 counts per sample. To account for differences in numbers of reads betweensamples, we used rarefied OTU counts to an even sampling depth. The resulting data set was used forall subsequent analyses. Phylogenetic beta diversity was calculated using the tree-based unique fraction(UniFrac) metric weighted by the abundances of individual OTUs (76). Principal coordinate analysis(PCoA) ordination of the distance metric was used to identify community grouping (77). To determinewhether the observed between-group distances were statistically significant, we performed permuta-tional multivariate analysis of variance of the distance metric by using the PERMANOVA� package of thesoftware PRIMER-E v7 (PRIMER-E Ltd., Plymouth, United Kingdom).

Nucleotide sequence accession number(s). Sequence reads of pmoA genes obtained from targetedamplicon sequencing and representative pmoA sequences of identified OTUs were deposited at theEuropean Nucleotide Archive (ENA) under study number PRJEB20489.

SUPPLEMENTAL MATERIAL

Supplemental material for this article may be found at https://doi.org/10.1128/AEM.01139-17.

SUPPLEMENTAL FILE 1, PDF file, 0.8 MB.

ACKNOWLEDGMENTSWe are grateful to I. Erny, M. Folini, and other helpers during field trips. We

acknowledge J. Walser, M. Kaestli, and C. Renaux for bioinformatics and statisticalsupport and R. Henneberger for a critical reading of the manuscript. High-throughputamplicon sequencing and quantitative PCR analyses were performed at the GeneticDiversity Center, ETH Zurich. We thank the three anonymous reviewers for valuablesuggestions.

This study was funded by the Swiss National Science Foundation (grant 200021-13772); additional financial support was provided by ETH Zurich.

REFERENCES1. Dominati E, Patterson M, Mackay A. 2010. A framework for classifying

and quantifying the natural capital and ecosystem services of soils. EcolEcon 69:1858 –1868. https://doi.org/10.1016/j.ecolecon.2010.05.002.

2. Myhre G, Shindell D, Bréon F-M, Collins W, Fuglestvedt J, Huang J, KochD, Lamarque J-F, Lee D, Mendoza B, Nakajima T, Robock A, Stephens G,Takemura T, Zhan H. 2013. Anthropogenic and natural radiative forcing, p659–740. In Stocker TF, Qin D, Plattner G-K, Tignor M, Allen SK, Boschung J,Nauels A, Xia Y, Bex V, Midgley PM (ed), Climate change 2013: the physicalscience basis. Contribution of Working Group I to the fifth assessmentreport of the Intergovernmental Panel on Climate Change. Cam-bridge University Press, Cambridge, United Kingdom.

3. Hanson RS, Hanson TE. 1996. Methanotrophic bacteria. Microbiol MolBiol Rev 60:439 – 471.

4. Qiu Q, Noll M, Abraham W-R, Lu Y, Conrad R. 2008. Applying stable isotopeprobing of phospholipid fatty acids and rRNA in a Chinese rice field to studyactivity and composition of the methanotrophic bacterial communities insitu. ISME J 2:602–614. https://doi.org/10.1038/ismej.2008.34.

5. Graef C, Hestnes AG, Svenning MM, Frenzel P. 2011. The active metha-notrophic community in a wetland from the High Arctic. Environ Micro-biol Rep 3:466 – 472. https://doi.org/10.1111/j.1758-2229.2010.00237.x.

6. Henneberger R, Chiri E, Bodelier PEL, Frenzel P, Lüke C, Schroth MH.2015. Field-scale tracking of active methane-oxidizing communities in alandfill cover soil reveals spatial and seasonal variability. Environ Micro-biol 17:1721–1737. https://doi.org/10.1111/1462-2920.12617.

7. Bender M, Conrad R. 1992. Kinetics of CH4 oxidation in oxic soils exposedto ambient air or high CH4 mixing ratios. FEMS Microbiol Ecol 101:261–270.

8. Dutaur L, Verchot LV. 2007. A global inventory of the soil CH4 sink. GlobalBiogeochem Cycles 21:1–9. https://doi.org/10.1029/2006GB002734.

9. Dunfield PF. 2007. The soil methane sink, p 152–170. In Reay D, HewittK, Smith K, Grace J (ed), Greenhouse gas sinks. CABI, Wallingford, UnitedKingdom.

10. Bridgham SD, Cadillo-Quiroz H, Keller JK, Zhuang Q. 2013. Methane

emissions from wetlands: biogeochemical, microbial, and modeling per-spectives from local to global scales. Glob Chang Biol 19:1325–1346.https://doi.org/10.1111/gcb.12131.

11. Baani M, Liesack W. 2008. Two isozymes of particulate methane mono-oxygenase with different methane oxidation kinetics are found inMethylocystis sp. strain SC2. Proc Natl Acad Sci U S A 105:10203–10208.https://doi.org/10.1073/pnas.0702643105.

12. Knief C. 2015. Diversity and habitat preferences of cultivated and un-cultivated aerobic methanotrophic bacteria evaluated based on pmoAas molecular marker. Front Microbiol 6:1346. https://doi.org/10.3389/fmicb.2015.01346.

13. Semrau JD, Chistoserdov A, Lebron J, Costello A, Davagnino J, Kenna E,Holmes AJ, Finch R, Murrell JC, Lidstrom ME. 1995. Particulate methanemonooxygenase genes in methanotrophs. J Bacteriol 177:3071–3079.https://doi.org/10.1128/jb.177.11.3071-3079.1995.

14. Henckel T, Jäckel U, Schnell S, Conrad R. 2000. Molecular analyses ofnovel methanotrophic communities in forest soil that oxidize atmo-spheric methane. Appl Environ Microbiol 66:1801–1808. https://doi.org/10.1128/AEM.66.5.1801-1808.2000.

15. Holmes AJ, Roslev P, McDonald IR, Iversen N, Henriksen K, Murrell JC.1999. Characterization of methanotrophic bacterial populations in soilsshowing atmospheric methane uptake. Appl Environ Microbiol 65:3312–3318.

16. Knief C, Lipski A, Dunfield PF. 2003. Diversity and activity of metha-notrophic bacteria in different upland soils. Appl Environ Microbiol69:6703– 6714. https://doi.org/10.1128/AEM.69.11.6703-6714.2003.

17. Whalen SC, Reeburgh WS. 1990. Consumption of atmospheric methaneby tundra soils. Nature 346:160 –162. https://doi.org/10.1038/346160a0.

18. Adamsen APS, King GM. 1993. Methane consumption in temperate andsubarctic forest soils: rates, vertical zonation, and responses to water andnitrogen. Appl Environ Microbiol 59:485– 490.

19. Born M, Dörr H, Levin I. 1990. Methane consumption in aerated soils ofthe temperate zone. Tellus B 42:2– 8.

Chiri et al. Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 14

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

20. Livesley SJ, Grover S, Hutley LB, Jamali H, Butterbach-Bahl K, Fest B,Beringer J, Arndt SK. 2011. Seasonal variation and fire effects on CH4,N2O and CO2 exchange in savanna soils of northern Australia. Agric ForMeteorol 151:1440 –1452. https://doi.org/10.1016/j.agrformet.2011.02.001.

21. Hiller RV, Bretscher D, DelSontro T, Diem T, Eugster W, Henneberger R,Hobi S, Hodson E, Imer D, Kreuzer M, Künzle T, Merbold L, Niklaus PA,Rihm B, Schellenberger A, Schroth MH, Schubert CJ, Siegrist H, Stieger J,Buchmann N, Brunner D. 2014. Anthropogenic and natural methanefluxes in Switzerland synthesized within a spatially explicit inventory.Biogeosciences 11:1941–1959. https://doi.org/10.5194/bg-11-1941-2014.

22. King GM, Nanba K. 2008. Distribution of atmospheric methane oxidationand methanotrophic communities on Hawaiian volcanic deposits andsoils. Microbes Environ 23:326 –330. https://doi.org/10.1264/jsme2.ME08529.

23. Bárcena TG, Yde JC, Finster KW. 2010. Methane flux and high-affinitymethanotrophic diversity along the chronosequence of a receding gla-cier in Greenland. Ann Glaciol 51:23–31.

24. Hofmann K, Reitschuler C, Illmer P. 2013. Aerobic and anaerobic micro-bial activities in the foreland of a receding glacier. Soil Biol Biochem57:418 – 426. https://doi.org/10.1016/j.soilbio.2012.08.019.

25. Nauer PA, Dam B, Liesack W, Zeyer J, Schroth MH. 2012. Activity anddiversity of methane-oxidizing bacteria in glacier forefields on siliceousand calcareous bedrock. Biogeosciences 9:2259 –2274. https://doi.org/10.5194/bg-9-2259-2012.

26. Paul F, Frey H, Le Bris R. 2011. A new glacier inventory for the EuropeanAlps from Landsat TM scenes of 2003: challenges and results. AnnGlaciol 52:144 –152. https://www.igsoc.org/annals/52/59/a59A054.html.

27. Haeberli W, Paul F, Zemp M. 2013. Vanishing glaciers in the EuropeanAlps, p 1–9. In Haeberli W, Paul F, Zemp M (ed), Fate of mountain glaciersin the Anthropocene. Pontificia Academia Scientiarum, Vatican City,Vatican.

28. Stevens P, Walker T. 1970. The chronosequence concept and soil forma-tion. Q Rev Biol 45:333–350. https://doi.org/10.1086/406646.

29. Nemergut DR, Anderson SP, Cleveland CC, Martin AP, Miller AE, SeimonA, Schmidt SK. 2007. Microbial community succession in an unvegetated,recently deglaciated soil. Microb Ecol 53:110–122. https://doi.org/10.1007/s00248-006-9144-7.

30. Sigler WV, Crivii S, Zeyer J. 2002. Bacterial succession in glacial forefieldsoils characterized by community structure, activity and opportunisticgrowth dynamics. Microb Ecol 44:306 –316. https://doi.org/10.1007/s00248-002-2025-9.

31. Lazzaro A, Abegg C, Zeyer J. 2009. Bacterial community structure ofglacier forefields on siliceous and calcareous bedrock. Eur J Soil Sci60:860 – 870. https://doi.org/10.1111/j.1365-2389.2009.01182.x.

32. Meola M, Lazzaro A, Zeyer J. 2014. Diversity, resistance and resilience ofthe bacterial communities at two Alpine glacier forefields after a recip-rocal soil transplantation. Environ Microbiol 16:1918 –1934. https://doi.org/10.1111/1462-2920.12435.

33. Mavris C, Egli M, Plötze M, Blum JD, Mirabella A, Giaccai D, Haeberli W.2010. Initial stages of weathering and soil formation in the Morteratschproglacial area (Upper Engadine, Switzerland). Geoderma 155:359 –371.https://doi.org/10.1016/j.geoderma.2009.12.019.

34. Gregory KJ, Goudie AS. 2011. The SAGE handbook of geomorphology.SAGE Publications, London, United Kingdom.

35. Sokratov SA. 2002. Intraseasonal variation in the thermoinsulation effectof snow cover on soil temperatures and energy balance. J Geophys Res107:ACL 13-1–ACL 13-6.

36. Bartlett MG. 2004. Snow and the ground temperature record of climatechange. J Geophys Res 109:F04008. https://doi.org/10.1029/2004JF000224.

37. Lipson DA, Schadt CW, Schmidt SK. 2002. Changes in soil microbialcommunity structure and function in an alpine dry meadow followingspring snow melt. Microb Ecol 43:307–314. https://doi.org/10.1007/s00248-001-1057-x.

38. Lazzaro A, Hilfiker D, Zeyer J. 2015. Structures of microbial communitiesin Alpine soils: seasonal and elevational effects. Front Microbiol 6:1330.https://doi.org/10.3389/fmicb.2015.01330.

39. Chiri E, Nauer PA, Henneberger R, Zeyer J, Schroth MH. 2015. Soil-methane sink increases with soil age in forefields of Alpine glaciers. SoilBiol Biochem 84:83–95. https://doi.org/10.1016/j.soilbio.2015.02.003.

40. Dumont MG, Lüke C, Deng Y, Frenzel P. 2014. Classification of pmoAamplicon pyrosequences using BLAST and the lowest common ancestormethod in MEGAN. Front Microbiol 5:34. https://doi.org/10.3389/fmicb.2014.00034.

41. Von Fischer JC, Butters G, Duchateau PC, Thelwell RJ, Siller R. 2009. Insitu measures of methanotroph activity in upland soils: a reaction-diffusion model and field observation of water stress. J Geophys ResBiogeosci 114:1–12. https://doi.org/10.1029/2008JG000731.

42. Khare E, Arora NK. 2015. Effects of soil environment on field efficacy ofmicrobial inoculants, p 381. In Arora NK (ed), Plant microbes symbiosis:applied facets. Springer India, New Delhi, India.

43. Or D, Tuller M. 2000. Flow in unsaturated fractured porous media: hydraulicconductivity of rough surfaces. Water Resour Res 36:1165–1177. https://doi.org/10.1029/2000WR900020.

44. Bachmann J, van der Ploeg RR. 2002. A review on recent developmentsin soil water retention theory: interfacial tension and temperature ef-fects. J Plant Nutr Soil Sci 165:468. https://doi.org/10.1002/1522-2624(200208)165:4�468::AID-JPLN4683.0.CO;2-G.

45. Tokunaga TK. 12 June 2009. Hydraulic properties of adsorbed water filmsin unsaturated porous media. Water Resour Res 45:W06415. https://doi.org/10.1029/2009WR007734.

46. Lebeau M, Konrad J-M. 2010. A new capillary and thin film flow modelfor predicting the hydraulic conductivity of unsaturated porous media.Water Resour Res 46:12. https://doi.org/10.1029/2010WR009092.

47. Bear J, Cheng AH-D. 2010. Modeling groundwater flow and contaminanttransport. Springer Netherlands, Dordrecht, Netherlands.

48. Ebrahimi A, Or D. 2015. Hydration and diffusion processes shape microbialcommunity organization and function in model soil aggregates. WaterResour Res 51:9804–9827. https://doi.org/10.1002/2015WR017565.

49. Lazzaro A, Brankatschk R, Zeyer J. 2012. Seasonal dynamics of nutrientsand bacterial communities in unvegetated Alpine glacier forefields. ApplSoil Ecol 53:10 –22. https://doi.org/10.1016/j.apsoil.2011.10.013.

50. Lazzaro A, Wismer A, Schneebeli M, Erny I, Zeyer J. 2015. Microbialabundance and community structure in a melting Alpine snowpack.Extremophiles 19:631– 642. https://doi.org/10.1007/s00792-015-0744-3.

51. Priemé A, Sitaula JIB, Klemedtsson ÅK, Bakken LR. 1996. Extraction ofmethane-oxidizing bacteria from soil particles. FEMS Microbiol Ecol21:59 – 68. https://doi.org/10.1111/j.1574-6941.1996.tb00333.x.

52. Dunfield PF, Yimga MT, Dedysh SN, Berger U, Liesack W, Heyer J. 2002.Isolation of a Methylocystis strain containing a novel pmoA-like gene.FEMS Microbiol Ecol 41:17–26. https://doi.org/10.1111/j.1574-6941.2002.tb00962.x.

53. Bouwman AF. 1999. Approaches to scaling of trace gas fluxes in eco-systems. Elsevier, Amsterdam, Netherlands.

54. Nauer PA, Chiri E, Zeyer J, Schroth MH. 2014. Technical note: disturbanceof soil structure can lead to release of entrapped methane in glacierforefield soils. Biogeosciences 11:613– 620. https://doi.org/10.5194/bg-11-613-2014.

55. Kanazawa S, Filip Z. 1986. Distribution of microorganisms, total biomass,and enzyme activities in different particles of brown soil. Microb Ecol12:205–215. https://doi.org/10.1007/BF02011205.

56. Brussaard L, Kooistra MJ. 1993. Soil structure/soil biota interrelationships,1st ed. Elsevier, Amsterdam, Netherlands.

57. Wang G, Or D. 2012. A hydration-based biophysical index for the onsetof soil microbial coexistence. Sci Rep 2:881. https://doi.org/10.1038/srep00881.

58. Bárcena T, Finster K, Yde J. 2011. Spatial patterns of soil development,methane oxidation, and methanotrophic diversity along a recedingglacier forefield, Southeast Greenland. Arct Antarct Alp Res 43:178 –188.https://doi.org/10.1657/1938-4246-43.2.178.

59. Bakermans C. 2015. Microbial evolution under extreme conditions. Wal-ter de Gruyter GmbH, Berlin, Germany.

60. Zheng Y, Yang W, Sun X, Wang S-P, Rui Y-C, Luo C-Y, Guo L-D. 2012.Methanotrophic community structure and activity under warming andgrazing of Alpine meadow on the Tibetan Plateau. Appl Microbiol Biotech-nol 93:2193–2203. https://doi.org/10.1007/s00253-011-3535-5.

61. Nauer PA, Chiri E, Schroth MH. 2013. Poly-use multi-level samplingsystem for soil-gas transport analysis in the vadose zone. Environ SciTechnol 47:11122–11130. https://doi.org/10.1021/es401958u.

62. Livingston GPP, Hutchinson GLL. 1995. Enclosure-based measure-ment of trace gas exchange: applications and sources of error, p 14 –51.In Matson PA, Harriss RC (ed), Biogenic trace gases: measuring emissionsfrom soil and water. Blackwell Science Ltd, Oxford, United Kingdom.

63. Nauer PA, Schroth MH. 2010. In situ quantification of atmosphericmethane oxidation in near-surface soils. Vadose Zone J 9:1052–1062.https://doi.org/10.2136/vzj2009.0192.

Atmospheric-Methane Oxidation in Alpine Soils Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 15

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

64. R Development Core Team. 2011. R: a language and environment forstatistical computing. R Foundation for Statistical Computing, Vienna,Austria.

65. Dörr H, Münnich KO. 1990. 222Rn flux and soil air concentration profilesin West-Germany. Soil 222Rn as tracer for gas transport in the unsatu-rated soil zone. Tellus B 42:20 –28.

66. Schroth MH, Eugster W, Gómez KE, Gonzalez-Gil G, Niklaus PA, Oester P.2012. Above- and below-ground methane fluxes and methanotrophicactivity in a landfill-cover soil. Waste Manag 32:879 – 889. https://doi.org/10.1016/j.wasman.2011.11.003.

67. Henneberger R, Lüke C, Mosberger L, Schroth MH. 2012. Structure andfunction of methanotrophic communities in a landfill-cover soil. FEMSMicrobiol Ecol 81:52–65. https://doi.org/10.1111/j.1574-6941.2011.01278.x.

68. Bybee SM, Bracken-Grissom H, Haynes BD, Hermansen RA, Byers RL,Clement MJ, Udall JA, Wilcox ER, Crandall KA. 2011. Targeted ampliconsequencing (TAS): a scalable next-gen approach to multilocus, multitaxaphylogenetics. Genome Biol Evol 3:1312–1323. https://doi.org/10.1093/gbe/evr106.

69. Bates D, Mächler M, Bolker B, Walker S. 2015. Fitting linear mixed-effectsmodels using lme4. J Stat Softw 67:1– 48. https://doi.org/10.18637/jss.v067.i01.

70. Gouy M, Guindon S, Gascuel O. 2010. SeaView version 4: a multiplatformgraphical user interface for sequence alignment and phylogenetic tree

building. Mol Biol Evol 27:221–224. https://doi.org/10.1093/molbev/msp259.

71. Keane T, Creevey C, Pentony M, Naughton T, Mclnerney J. 2006. Assess-ment of methods for amino acid matrix selection and their use onempirical data shows that ad hoc assumptions for choice of matrix arenot justified. BMC Evol Biol 6:29. https://doi.org/10.1186/1471-2148-6-29.

72. Guindon S, Dufayard J-F, Lefort V, Anisimova M, Hordijk W, Gascuel O.2010. New algorithms and methods to estimate maximum-likelihoodphylogenies: assessing the performance of PhyML 3.0. Syst Biol 59:307–321. https://doi.org/10.1093/sysbio/syq010.

73. Anderson MJ, Ellingsen KE, McArdle BH. 2006. Multivariate dispersion asa measure of beta diversity. Ecol Lett 9:683– 693. https://doi.org/10.1111/j.1461-0248.2006.00926.x.

74. Lozupone C, Lladser ME, Knights D, Stombaugh J, Knight R. 2011. UniFrac:an effective distance metric for microbial community comparison. ISME J5:169–172. https://doi.org/10.1038/ismej.2010.133.

75. McMurdie PJ, Holmes S. 2013. phyloseq: an R package for reproducibleinteractive analysis and graphics of microbiome census data. PLoS One8:e61217. https://doi.org/10.1371/journal.pone.0061217.

76. Lozupone C, Knight R. 2005. UniFrac: a new phylogenetic method forcomparing microbial communities. Appl Environ Microbiol 71:8228–8235.https://doi.org/10.1128/AEM.71.12.8228-8235.2005.

77. Legendre P, Legendre L. 1998. Numerical ecology, 2nd ed. ElsevierScience, Amsterdam, Netherlands.

Chiri et al. Applied and Environmental Microbiology

September 2017 Volume 83 Issue 18 e01139-17 aem.asm.org 16

on March 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from