Received: 13 December 2016

|

Accepted: 30 May 2017

DOI: 10.1111/gcb.13790

PRIMARY RESEARCH ARTICLE

Differential sensitivity of total and active soil microbial communities to drought and forest management Felipe Bastida1

s-Abell | Irene F. Torres1 | Manuela Andre an2 | Petr Baldrian3 |

 n Lo  pez-Monde jar3 | Tom trovsky 3 | Hans H. Richnow4 | Robert Starke5 | Rube as Ve ~o1 | Carlos Garcıa1 | Francisco R. Lo  pez-Serrano2 | Nico Jehmlich5 Sara Ondon 1

Department of Soil and Water Conservation, CEBAS-CSIC, Murcia, Spain 2

Department of Science and Agroforestry Technology and Genetics, Higher Technical School of Agricultural and Forestry Engineering, University of Castilla-La Mancha, Albacete, Spain 3

Laboratory of Environmental Microbiology, Institute of Microbiology of the CAS, Praha 4, Czech Republic 4

Department of Isotope Biogeochemistry, Helmholtz-Centre for Environmental Research - UFZ, Leipzig, Germany 5

Department of Molecular Systems Biology, Helmholtz-Centre for Environmental Research - UFZ, Leipzig, Germany

Abstract Climate change will affect semiarid ecosystems through severe droughts that increase the competition for resources in plant and microbial communities. In these habitats, adaptations to climate change may consist of thinning—that reduces competition for resources through a decrease in tree density and the promotion of plant survival. We deciphered the functional and phylogenetic responses of the microbial community to 60 years of drought induced by rainfall exclusion and how forest management affects its resistance to drought, in a semiarid forest ecosystem dominated by Pinus halepensis Mill. A multiOMIC approach was applied to reveal novel, community-based strategies in the face of climate change. The diversity and the composition of the total and active soil microbiome were evaluated by 16S rRNA gene (bacteria) and ITS (fungal) sequencing, and by metaproteomics. The microbial

Correspondence Felipe Bastida, CEBAS-CSIC. Department of Soil and Water Conservation. Campus Universitario de Espinardo, Murcia, Spain. Email: [email protected]

biomass was analyzed by phospholipid fatty acids (PLFAs), and the microbially mediated ecosystem multifunctionality was studied by the integration of soil enzyme activities related to the cycles of C, N, and P. The microbial biomass and ecosystem multifunctionality decreased in drought-plots, as a consequence of the lower soil

Funding information MINECO, Grant/Award Number: AGL2014n y Cajal MINECO and 54636-R; Ramo FEDER funds, Grant/Award Number: RYC2012-10666; MINECO, Grant/Award n Number: AGL2014-55658-R; Fundacio neca, Grant/Award Number: 19896/ Se GERM/15; Institute of Microbiology of the CAS, Grant/Award Number: RVO61388971

moisture and poorer plant development, but this decrease was more notable in unthinned plots. The structure and diversity of the total bacterial community was unaffected by drought at phylum and order level, but did so at genus level, and was influenced by seasonality. However, the total fungal community and the active microbial community were more sensitive to drought and were related to ecosystem multifunctionality. Thinning in plots without drought increased the active diversity while the total diversity was not affected. Thinning promoted the resistance of ecosystem multifunctionality to drought through changes in the active microbial community. The integration of total and active microbiome analyses avoids misinterpretations of the links between the soil microbial community and climate change. KEYWORDS

climate change, drought, forest management, genomics, metaproteomics, microbial biomass, microbial community, semiarid

Glob Change Biol. 2017;23:4185–4203.

wileyonlinelibrary.com/journal/gcb

© 2017 John Wiley & Sons Ltd

|

4185

4186

|

1 | INTRODUCTION

BASTIDA

ET AL.

biomass after drought (Barnard, Osborne, & Firestone, 2013; Manzoni, Taylor, Richter, Porporato, & Agren, 2012), while others

Climate change is an unavoidable phenomenon that will affect ter-

did not find consistent effects on the microbial community composi-

restrial ecosystems worldwide. In Southern Europe, a reduction in

tion (Rousk, Smith, & Jones, 2013; de Vries & Shade, 2013; Canarini,

precipitation is expected to affect plant development and soil fertility

Carrillo, Mariotte, Ingram, & Dijkstra, 2016). Furthermore, the

(IPCC, 2013). However, drought may have additional adverse effects

response of a microbial community to drought may depend on the

on ecosystems. The advent of reduced rainfall and high tempera-

physiological tolerance and metabolic flexibility of the microbiome

tures, as well as pernicious human actions, is expected to increase

(Allison & Martiny, 2008).

the frequency of fires in forest areas of the Iberian peninsula

The soil microbial community is a complex “blackbox” that

(Lindner et al., 2010; Hedo Santiado, Lucas-Borja, Wic-Baena,

requires a multi-conceptual approach (Hultman et al., 2015; Bastida

s-Abellan, & las Heras, 2015). After such fires, plant developAndre

et al., 2016). Most methods focus on evaluation of the total micro-

ment proceeds slowly, particularly in semiarid areas with limited

bial community and fail to determine its active fraction (Blago-

water availability. Hence, it is expected that plant growth after a fire

datskaya & Kuzyakov, 2013) which can be directly connected to

will be even more constrained as a consequence of the drought in

soil resistance against drought. This has ecological consequences

these areas. This is the reason why particular silvicultural treatments,

since the behavior of the active community is more important (or

such as thinning, are required to foster plant survival in these

even essential) and can differ from that of the total community.

ecosystems. Usually, thinning reduces tree densities and, tradition-

The sensitivity of the active microbial community can be considered

ally, it has been carried out to manage forest stands so that competi-

a biological mechanism that regulates the functional responses of

tion for resources (i.e., water and nutrients) between neighboring

soil to direct (e.g., forest management) and indirect (e.g., climate

trees is avoided, as a way to improve tree growth and survival and

change) human-induced alterations. Indeed, it has been shown that

pez-Serrano, & las ecosystem productivity (Gonzalez-Ochoa, Lo

the diversity of the active community (analyzed by metaproteomics)

pez-Serrano, las Heras, Gonzalez-Ochoa, & GarcıaHeras, 2004; Lo

is related more closely to soil functionality than that of the total

Morote, 2005). Moreover, it has been observed that thinning affects

community (analyzed by 16S rRNA genes and ITS sequencing) (Bas-

the structure of the soil microbial community through changes in soil

tida et al., 2016). Recently, the increasing application of soil

temperature and the abundance of fungi-feeding nematodes in pine

metaproteomics has provided unprecedented, in-depth characteriza-

plantation in eastern Tibetan Plateau (Yang, Pang, Hu, Bao, & Tian,

tion of the composition and functionality of active microbial com-

2017). Furthermore, Rietl and Jackson (2012) concluded that thin-

munities, giving deeper insights into terrestrial microbial ecology

ning reduces the amount of litter in a mesic forest in northern Mis-

(Chourey et al., 2010;

sissippi and promoted the efficiency of soil enzymes involved in

et al., 2016; Keiblinger, Fuchs, Zechmeister-Boltenstern, & Riedel,

litter degradation.

2016).

Bastida, Garcıa et al., 2015; Bastida

Moreover, it has been highlighted that many of the impacts of

We hypothesized that the active soil microbial community will

climate change on the soil microbial community may occur via direct

be more sensitive to drought and forest management than the total

or indirect effects on plant communities (Bardgett, Freeman, & Ostle,

microbial community. Indeed, it is expected that fungal community

2008; von Rein et al., 2016). Hence, thinning may interact with cli-

will be more sensitive to drought than bacterial community

mate change factors and alter the ecological responses of the soil

(Hawkes et al., 2011; Kaisermann, Maron, Beaumelle, & Lata, 2015).

microbial community. However, the impacts of thinning on the

Moreover, given the lower tree density and the consequent open

microbiota are largely unknown in climate change-induced scenarios

spaces after thinning treatments (Yang et al., 2017), that may lead

although it is widely known that soil biodiversity is connected to the

to a higher frequency of microtemporal low-moisture events, we

functionality and productivity of ecosystems (van der Heijden,

hypothesized that thinning will promote a microbial community that

Bardgett, & van Straalen, 2008) and global biogeochemical cycles

is more resistant and adapted to induced drought. Here, we deter-

(Schimel & Schaeffer, 2013; Rousk & Bengston, 2014). In this sense,

mine the responsiveness of the soil microbial community to forest

microbial-mediated processes can result fundamental for the cycling

management in a climate change scenario. Particularly, we aim: (i)

of soil organic matter which is very limited in semiarid areas of

to evaluate the impacts of 6 years of induced drought on the

Southeastern Spain and is a fundamental component of soil sustain-

diversity, biomass, and activity of the microbial community in a

ability (Bastida, Moreno, Hernandez, & Garcıa, 2006). However, a

semiarid forest ecosystem; and (ii) to show whether forest manage-

detailed look into the literature reveals that the observed impacts of

ment (thinning) influences the resistance of the microbial commu-

induced drought on the soil microbial community are inconsistent.

nity to induced drought. To achieve these objectives, several

Studies based on distinct climatic conditions and experimental

properties were evaluated, including soil enzyme activities, microbial

designs, ranging from laboratory incubations to in situ experiments,

biomass, and the diversity and composition of the total and active

have reported a high level of variation in microbial responses. Some

microbial communities, by genomic and metaproteomic approaches.

studies showed profound alterations of community composition and

Overall, this study provides novel insights into the microbially medi-

diversity, while in others these properties were resistant. For

ated processes involved in the adaptations of soil to climate

instance, some studies demonstrated increases in the gram-positive

change.

BASTIDA

|

ET AL.

2 | MATERIAL AND METHODS 2.1 | Study site, experimental design, and soil sampling

4187

2.2 | Chemical and physico-chemical properties, enzyme activities, and phospholipid fatty acids (PLFAs) analysis The total nitrogen (N) and total organic carbon (TOC) in the soil

In August 1994 a severe forest fire affected 14000 ha in Yeste

were determined using an Elemental Analyzer (C/N Flash EA 112

(2°200 W, 38°220 N, Albacete Province, in the Southeast of Spain).

Series-Leco Truspec). Water-soluble C (WSC) was extracted from

Mature forests, mainly of Pinus halepensis, were burnt, but, 5 years

the soil with distilled water (1:5, w:v) by shaking for 2 hr, followed

after the fire, a plentiful natural regeneration had occurred (>7000

by centrifugation at 13000 rpm for 15 min and filtration. The analy-

saplings ha1). The understorey consisted of bushes of Quercus

sis of the carbon (C) content in the extracts was performed in a C

coccifera,

Rhamnus

analyzer for liquid samples (Multi N/C 3100, Analytik Jena). Soil

lycioides, Thymus sp., and Cistus sp. In 2004, a thinning treatment

moisture was analyzed by gravimetry after heating for 24 hr at

(1600 trees ha1) was carried out in some areas. According to the

105°C.

Rosmarinus

officinalis,

Juniperus

oxycedrus,

soil taxonomy (Soil Survey Staff, 2003), the soil was classified as

The urease activity in the soil was determined by the buffered

Entisol (suborder orthents, group xerorthents, subgroup lithic). The

method of Kandeler and Gerber (1988). The phosphomonoesterase

soil textural class was sand, with a pH of 8.6. The N, P, and K

and b-glucosidase activities were determined by following the meth-

concentrations were 0.45% and 3.6 and 200.1 mg L1 respectively.

ods described by Tabatabai and Bremmer (1969) and a modification

Average annual rainfall and temperature for the last 30 years was

method of Tabatabai’s (1982), respectively. Polyphenol oxidase was

503 mm and 13.5°C respectively. In 2015, average temperature in

determined by the method of Allison (2006), while N-acetylglucosa-

summer and autumn was 22.4 and 11.9°C, respectively. In 2015,

minidase (NAG) and cellobiohydrolase (CBH) were determined as

the rainfall in summer and autumn was 14.8 and 58.5 mm

reported by Allison and Jastrow (2006). PLFAs were analysed as

respectively.

described previously (Bastida, Selevsek, Torres, Hernandez, & Garcıa,

A factorial design with two levels of rainfall (natural rainfall vs rainfall reduction or induced drought), two levels of forest manage-

2015). Detailed methods of soil enzyme activities and PLFAs are included in Supporting Information.

ment (thinning present/absent), and two seasons (summer 2015 and autumn 2015) was developed. In July 2009, 12 plots (10 m 9 15 m) were set up, six of them in thinned areas and six

2.3 | DNA extraction and amplification

in unthinned areas. Each group of six plots was split into two

Extraction of DNA from 500 mg of freeze-dried soil from each sam-

groups: three plots received natural rainfall and three plots, par-

ple was performed, using a Fast DNA Spin Kit for soil and the Fas-

tially covered by PVC gutters, received reduced precipitation to

tPrepâ Instrument (MP Biomedicals, Santa Ana, CA, USA). The V4

simulate drought-induced conditions. The PVC gutters were sus-

region of bacterial 16S rRNA was amplified using the barcoded pri-

pended at about 50 cm above the soil surface and occupied about

mers 515F and 806R (Caporaso et al., 2012). The PCR amplification

50% of the plot surface, intercepting the corresponding rainfall. In

of the fungal ITS2 region from DNA was performed using barcoded

total, 12 plots were set up (2 thinned/unthinned 9 2 natural rain-

gITS7 and ITS4 (Ihrmark et al., 2012) in three PCR reactions per

fall/induced drought 9 3 replicate plots). Plots were separated 2 m

sample, as described by Zifcakova, Vetrovsky, Howe, and Baldrian

each other.

(2016).

Soil sampling was performed in summer and autumn 2015—6

The PCR amplification of bacteria and fungi was performed using

years after treatment initiation—to contrast different seasons. Six

short barcoded primers. These were composed of the barcode (4–6

soil samples were taken at a depth of 0–15 cm, after removing the

nucleotides), a spacer (2 nucleotides absent in all GenBank

litter within each plot, and mixed to obtain a composite sample per

sequences at this position to avoid preferential amplification of some

plot. Samples were collected randomly in each plot but never at the

targets Parameswaran et al., 2007), and the primer. Both the forward

plot borders (minimum distance of 50 cm from plot border was

and reverse primers were barcoded to make sure that barcode-

established). The samples were sieved (2 mm) and kept at 4°C for

switching did not affect the results and to avoid the problems with

chemical analysis and at 20°C for molecular analyses.

Caporaso primers. The PCR products were cleaned using a MiniElute

The abbreviations used for the treatments are as follows: CU

Kit (Qiagen) and the concentrations were measured by Qubit. After

(unthinned control soil); DU (unthinned soil submitted to drought);

amplification and purification of amplicons, a TruSeq PCR-Free kit

CT (thinned control soil); DT (thinned soil submitted to drought).

was used for library preparation to avoid further amplification (and

Furthermore, the small letters “s” and “a”, following each abbrevia-

to avoid additional PCR bias).

tion, indicate summer and autumn samples respectively. As a basic

Sequencing of fungal and bacterial amplicons was performed on

plant characterization in 2015, the survival of trees was around

Illumina MiSeq. The amplicon sequencing data were processed using

25000, 15000, 2600, and 1600 trees ha1 in CU, DU, CT, and DT

the pipeline

respectively. The mean trunk diameter at 1.30 m above ground was

reads were merged using

25, 25, 83, and 93 mm in CU, DU, CT, and DT respectively.

cons were processed for bacterial 16S, whereas the ITS2 region was

SEED

1.2.3 (Vetrovsky & Baldrian, 2013). Briefly, pair-end FASTQ-join

(Aronesty, 2013). Whole ampli-

4188

|

BASTIDA

ET AL.

Chimeric

semiarid soils (Bastida, Hernandez, & Garcıa, 2014). The cell lysis and

sequences were detected using USEARCH 7.0.1090 and deleted, and

disruption of soil aggregates were performed by boiling at 100°C for

implemented within USEARCH

10 min in sodium dodecyl sulfate (SDS) buffer. The concentration

(Edgar, 2013) at a 97% similarity level. Consensus sequences were

and purification steps were performed using trichloroacetic acid

constructed for each cluster (Vetrovsky & Baldrian, 2013) and the

(TCA) and three acetone washing steps. Protein pellets were resus-

BLASTn

pended in SDS lysis buffer, containing 4% SDS, 0.1 mM dithiothre-

against the Ribosomal Database Project (Cole et al., 2014) and Gen-

itol, and 100 mM Tris HCl, and incubated for 5 min at 95°C. The

extracted

using

ITS

EXTRACTOR 1.0.8

sequences were clustered using

before

UPARSE

processing.

closest hits at the genus or species level were identified using

(Koljalg et al., 2013) and Gen-

resuspended proteins were loaded on SDS gels (4% stacking gel,

Bank (for fungi). Sequences identified as nonbacterial or nonfungal

12% separating gel) and run into the separating gel for 10 min. After

were discarded. The Shannon–Wiener index was calculated for

electrophoresis, the gels were stained with colloidal Coomassie bril-

15000 and 10000 sequences per sample, respectively, for bacterial

liant blue (Bastida et al., 2016). The gel area containing the protein

and fungal communities. Rarefaction curves and the detailed commu-

mixture of each sample was sliced into one piece. The gel pieces

nity composition of all samples are showed in Supporting Informa-

were destained and subsequently digested using 100 ng of trypsin

1.2.3 was used for data

(Sigma, Munich, Germany), overnight at 37°C. The peptide mixture

preprocessing and diversity calculations (Vetrovsky & Baldrian,

was extracted twice with 100% acetonitrile and concentrated by

2013). The DNA sequences have been deposited by MG-RAST

vacuum centrifugation for 15 min. Tryptic peptides were reconsti-

mgp79662 (http://metagenomics.anl.gov/).

tuted in 0.1% formic acid prior to LC-MS measurement (Bastida

bank databases (for bacteria) or

UNITE

tion (Figs. S1–S6). The pipeline

SEED

et al., 2016).

2.4 | Construction of semiarid soil metagenome

The peptide lysates were separated on a UHPLC system (Ultimate 3000, Dionex/Thermo Fisher Scientific, Idstein, Germany).

The DNA from the soil samples utilized in this study, together

Samples (5 lL) were first loaded for 5 min on the precolumn (l-pre-

with other semiarid soils (Bastida et al., 2016), was utilized for the

column, Acclaim PepMap C18, 2 cm, Thermo Scientific) at 4% mobile

preparation of a semiarid soil metagenome. A TruSeq PCR Free LT

phase B (80% acetonitrile in nanopure water with 0.08% formic acid)

Sample Preparation Kit (Illumina, San Diego, CA, USA) was used for

and 96% mobile phase A (nanopure water with 0.1% formic acid),

library preparation. The library size-distribution was checked on an

then eluted from the analytical column (Acclaim PepMap C18 LC

Agilent 2100 Bionalyser (Agilent Technologies). The libraries were

column, 25 cm, Thermo Scientific) over a 120-min linear gradient of

sequenced on an Illumina HiSeq2000 at Brigham Young University

mobile phase B (4%–55%). Mass spectrometry was performed on an

(Provo, UT, USA).

Orbitrap Fusion mass spectrometer (Thermo Fisher Scientific,

The reads were quality trimmed by removing adapters with Trim-

Waltham, MA, USA) with a TriVersa NanoMate (Advion, Ltd.,

momatic (v 0.27), using ILLUMINA TRUSEQ2-PE adapters with a

Harlow, UK) source in LC chip coupling mode. The MS was set on

seed mismatch threshold, palindrome clip threshold, and simple clip

top speed for 3 s using the Orbitrap analyzer for MS and MS/MS

threshold set at 2, 30, and 10 respectively (Bolger, Lohse, & Usadel,

scans with higher energy collision dissoziation (HCD) fragmentation

2014). Furthermore, the sequencing reads were filtered by base call

at a normalized collision energy of 30%. The MS scans were

quality

(http://hannonlab.cshl.edu/

measured at a resolution of 120,000 in the scan range of 400–

fastx_toolkit/index.html), specifically fastq_quality_filter, with the fol-

1,600 m/z. The MS ion count target was set to 4 9 105 at an injec-

lowing parameters: Q33 q30 p 50. The resulting sequences

tion time of 80 ms. Ions for MS/MS scans were isolated in the quad-

were normalized using methods described previously (Pell et al.,

rupole with an isolation window of 1.6 Da and were measured with

2012; Howe et al., 2014) and Khmer (v 0.7.1) and command normal-

a resolution of 15,000. The dynamic exclusion duration was set to

ize-by-median.py with the following parameters: k20 C20 N4

30 s. The automatic gain control target was set to 5 9 104 with an

x 50e9. Next, errors were trimmed by removing low-abundance

injection time of 100 ms.

using the

FASTX-TOOLKIT

fragments of high coverage reads with Khmer and command filter-

Proteome Discoverer (v1.4, Thermo Scientific) was used for pro-

abund.py V. The paired-end assembly of the remaining reads was

tein identification and the MS/MS spectra acquired were searched

performed with MEGAHIT (v1.0.2, with default parameters; Li, Liu,

with Sequest HT against the specific semiarid soil metagenome data-

Luo, Sadakane, & Lam, 2015). The sequence data of all contig

base (containing 48,094,830 protein-coding sequences). Enzyme

sequences have been deposited in the MG RAST data set

specificity was selected to trypsin with up to two missed cleavages

(number 4697967.3) (http://metagenomics.anl.gov/linkin.cgi?metage

allowed, using 10 ppm peptide ion tolerance and 0.05 Da MS/MS

nome = 4697967.3).

tolerances. Oxidation (methionine) and carbamylation (lysine and arginine)

2.5 | Protein extraction from soil and mass spectrometric analysis

were

selected

as

variable

modifications

and

car-

bamidomethylation (cysteine) as a static modification. Only peptides with a false discovery rate (FDR)

Differential sensitivity of total and active soil microbial communities to drought and forest management.

Climate change will affect semiarid ecosystems through severe droughts that increase the competition for resources in plant and microbial communities...
3MB Sizes 4 Downloads 15 Views