lable at ScienceDirect

Soil Biology & Biochemistry 112 (2017) 153e164

Contents lists avai

Soil Biology & Biochemistry

journal homepage: www.elsevier .com/locate/soi lb io

Using community trait-distributions to assign microbial responses topH changes and Cd in forest soils treated with wood ash

Carla Cruz-Paredes a, Håkan Wallander b, Rasmus Kjøller a, Johannes Rousk b, *

a Department of Biology, Terrestrial Ecology Section, University of Copenhagen, Universitetsparken 15, 2100 Copenhagen, Denmarkb Section of Microbial Ecology, Department of Biology, Lund University, 22362 Lund, Sweden

a r t i c l e i n f o

Article history:Received 19 September 2016Received in revised form8 April 2017Accepted 5 May 2017

Keywords:Plantesoil (below-ground) interactionsWood ashDecomposer ecologySoil organic matterpHHeavy metals

* Corresponding author. Section of Microbial EcolLund University, Ecology Building, 22362 Lund, Swed

E-mail address: johannes.rousk@biol.lu.se (J. Rous

http://dx.doi.org/10.1016/j.soilbio.2017.05.0040038-0717/© 2017 Elsevier Ltd. All rights reserved.

a b s t r a c t

The identification of causal links between microbial community structure and ecosystem functions arerequired for a mechanistic understanding of ecosystem responses to environmental change. One of themost influential factors affecting plants and microbial communities in soil in managed ecosystems is thecurrent land-use. In forestry, wood ash has been proposed as a liming agent and a fertilizer, but has beenquestioned due to the risk associated with its Cd content. The aim of this study was to determine theeffects of wood ash on the structure and function of decomposer microbial communities in forest soilsand to assign them to causal mechanisms. To do this, we assessed the responses to wood ash applicationof (i) the microbial community size and structure, (ii) microbial community trait-distributions, includingbacterial pH relationships and Cd-tolerance, to assign the microbial responses to pH and Cd, and (iii)consequences for proxies of the function soil organic matter (SOM) turnover including respiration andmicrobial growth rates. Two sets of field-experiments in temperate conifer forest plantations werecombined with laboratory microcosm experiments where wood ash additions were compared to addi-tions of lime and Cd. Wood ash induced structural changes in the microbial community in both fieldexperiments, and striking similarities were observed between the application of ash and that of lime inthe microcosm experiments. Wood ash increased pH, and led to a shift toward faster SOM decompositionand a reduced importance of fungi. This coincided with shifts in bacterial community trait distributionsfor pH, with pH optima closely tracking the new soil pH. A Cd solution could induce Cd-tolerance in themicrocosm experiments, but the ash did not affect the microbial tolerance to Cd in field or microcosmexperiments. We demonstrate that the microbial community responded strongly to the application ofwood ash to forest soils with consequences for its functional capabilities in terms of respiration andgrowth rates. The bacterial community's trait distributions revealed that the increased pH directly causedthe microbial responses, while the wood ash associated Cd has no detectable effects on the microbialcommunity. The study demonstrates the power of community trait distributions to (i) causally linkmicrobial structural responses to environmental change and (ii) potential to predict the ecosystemfunctional consequences.

© 2017 Elsevier Ltd. All rights reserved.

1. Introduction

A central interest in ecology is the connection between thecomposition of organisms and the processes and elemental fluxesthrough an ecosystem (Odum and Odum, 1976). Global biogeo-chemical cycles are dominated by microorganisms. In mostterrestrial ecosystems, the turnover of carbon (C) and the

ogy, Department of Biology,en.k).

regeneration of the nutrients that sustain primary productivity arerate-limited by the microbial turnover of soil organic matter (SOM).Thus, it has been argued that a mechanistic understanding ofecosystem functioning is only attainablewith a clear understandingof the factors that structure microbial communities (Raes and Bork,2008; Schmidt et al., 2011). Research of recent decades has high-lighted environmental factors that are important for structuringmicrobial communities, including e.g. pH in terrestrial ecosystems(Lauber et al., 2009; Rousk et al., 2010a) and salinity in aquaticecosystems (Lozupone and Knight, 2007; Mohamed and Martiny,2011; Rath and Rousk, 2015). Although these observations have

C. Cruz-Paredes et al. / Soil Biology & Biochemistry 112 (2017) 153e164154

advanced our understanding of how microbial communities areshaped by their environment, important knowledge gaps remain tobe filled, including the causal links between community structureand function (Jansson, 2013; Prosser, 2013). Within the field ofmicrobial ecology, parallel responses of microbial structure andfunction to experimental treatments reported in most contempo-rary work is suggestive, but notoriously difficult to disentangle(Bier et al., 2016). For instance, it has frequently been observed thata community composition has responded to a change in its envi-ronment such as warming (Bradford, 2013). However, what actuallycaused the change e indirect effects such as nutrient availability, Cquality or direct effects such as microbial tolerance to temperaturee normally remains unresolved. Extending ideas from plant andanimal ecology (Grime, 1977; Southwood, 1977) to the realm ofmicrobial ecology, Green et al. (2008) and Webb et al. (2010) arguethat a trait based approach can bridge from structure to function.

The responses of communities to their environment, and sub-sequent feedback, is determined by the traits of included species orindividuals (Allison, 2012; Norberg et al., 2001;Webb et al., 2010). Alogical way to estimate a community's dependence on, and feed-back to, its environment is thus to identify traits of all individualorganisms, weigh up their relative abundance, and add them up toestablish the contribution of a community to ecosystem function(Raes and Bork, 2008). Problematically, although some reports arebeginning to appear characterizing single traits for selected mi-crobial taxa (Martiny et al., 2015), the determination of traits for allthe microbial taxa that make up natural communities is a dauntingtask. While this strategy may be acceptable for communities wheremost species can be identified, this is a nearly impossible enterprisefor microbial communities given their staggering diversity(Delmont et al., 2011). As an alternative strategy, estimates of thedistribution of traits that define whole communities could be moretractable (Norberg et al., 2001). The determination of an aggregatetrait distribution for a whole community could then be used as away forward (B�arcenas-Moreno et al., 2016; Norberg et al., 2001).The dependence of the intrinsic growth rate of a microbial com-munity on environmental factors, or its tolerance to e.g. toxins,could define such community trait distributions (Bååth andKritzberg, 2015; B�arcenas-Moreno et al., 2016). This approach hasbeen used to determine bacterial community trait distributions fortemperature (B�arcenas-Moreno et al., 2009; Birgander et al., 2013;Li and Dickie, 1987), salt (Rath et al., 2016), and pH (Bååth andKritzberg, 2015; Fern�andez-Calvi~no et al., 2011). For instance, thepH trait distributions of bacterial communities are readily deter-mined in an environmental sample bymeasuring the growth rate ata range of different pHs, forming a unimodal distribution withhighest values at the pH-optimum (e.g. Fern�andez-Calvi~no et al.,2011). Bacterial community trait distributions for inhibitory envi-ronmental factors like salt can be determined by establishing dose-response relationships to estimate tolerance, often forming asigmoidal relationship from highest values without toxins, andlower values at increasing toxin concentrations (e.g. Rath et al.,2016). Based on these approaches, the relative importance ofenvironmental factors have been compared with the initialcomposition of the source community for formed community traitdistributions (B�arcenas-Moreno et al., 2016; Pettersson and Bååth,2013, 2004). Recently, the alignment between community traitdistributions and environmental conditions have also been linkedto functional consequences (B�arcenas-Moreno et al., 2016). Incontrast with most alternative methods to assess microbial com-munity change, monitoring responses in community trait distri-butions also have the power to assign community change to causalfactors. For instance, if microbial communities becomemorewarm-tolerant in response to warming, this provides unambiguous evi-dence that the community was, at least partly, driven by the

temperature increase, rather than indirect effects (B�arcenas-Moreno et al., 2009; Li and Dickie, 1987).

The current land-use characterizes the environment controllingplant and microbial communities in most managed ecosystems. Inforestry, various treatments are made to enhance and maintain soilfertility. One such treatment is the application of wood ash, whichhas been proposed as a liming agent and a slow release fertilizersuitable for forest soils (Jacobson, 2003). In addition to increasingsoil pH (Bååth et al., 1995; Fritze et al., 2000; Knapp and Insam,2011; Perki€om€aki and Fritze, 2002), wood ash contains nutrients,e.g. calcium, potassium and magnesium, along with heavy metals,including e.g. cadmium (Cd) (Knapp and Insam, 2011). Of these, Cdis of major environmental concern, often setting the limits for theuse of wood ash (Perki€om€aki and Fritze, 2004). Given that wood ashis a cocktail of potential drivers of microbial community change, theunderlying causal factors e its alkaline properties, its nutrientcontent, or Cd content e often remain undifferentiated.

The aim of this study was to determine the effects of wood ashon soil microbial communities and their ability to use anddecompose soil organic matter (SOM). Specifically, our objectiveswere to: (1) assess the structural responses of the microbial com-munity in terms of biomass and community structure, (2) deter-mine how causal factors associated with the wood ash (i) pH and(ii) Cd had affected the community trait distribution and had thusstructured the community composition, and (3) assess the conse-quences these microbial responses had for decomposer functioningof the soil, as indicated by the performance of the communitymeasured as microbial growth rates and respiration. To accomplishthis, we combined a range of field-experiments of ash application atfield scale with parallel laboratory microcosm experimentscomparing the effects of ash with factorial additions of lime and Cd.These experiments were thus designed to (a) disentangle the pHand Cd effects of ash amendments, and (b) to characterize the dy-namics of how the anticipated induction of tolerance to pH andmetals developed over time.

Soil pH is known as a strong driver of microbial communitystructure (Lauber et al., 2009; Rousk et al., 2010a) along with fungaland bacterial growth rates (Rousk et al., 2009) both when soil pHvaries due to pedology (Lauber et al., 2009) and when it iscontrolled by liming (Rousk et al., 2010a). Thus we hypothesized(H1) that the dominant factor affecting microbial communities andprocesses would be the soil pH change induced by the ash appli-cation (Bj€ork et al., 2010; Demeyer et al., 2001; Insam et al., 2009;Reid and Watmough, 2014). We predicted that (a) lime additionsshould be able to reproduce the effect of ash in microcosms e thesoil pH changes induced by both treatments likely dominating theirinfluence on microbial communities e and that (b) microbial pHtrait distributions would be induced to shift by ash in field exper-iments, and that microbial pH trait distributions would respondsimilarly in ash and lime additions in the microcosm experiments.It is known that Cd can induce tolerance in soil microbial com-munities when applied in high concentrations (Díaz-Ravi~na andBååth, 1996). However, at the rates and forms of metal associatedwith ash, earlier studies have been unable to detect a microbialresponse linked to Cd (Perki€om€aki and Fritze, 2004). Using ap-proaches to measure microbial community tolerance to metals insoil (Berg et al., 2012; Lekfeldt et al., 2014), we expected that Cdadditions could induce microbial Cd-tolerance in the microcosmexperiments, but we also hypothesized (H2) that the Cd associatedwith ash applications in the field experiments (up to 90 t ash ha�1)would not result in any responses in microbial Cd tolerance. Inaccordancewith earlier findings, we hypothesized (H3) that the ashapplications leading to higher pH in the field experiment wouldfavor bacterial over fungal growth and result in a faster decompo-sition of soil organic matter (Rousk et al., 2009), with a similar

C. Cruz-Paredes et al. / Soil Biology & Biochemistry 112 (2017) 153e164 155

outcome for ash and lime treatments also in the microcosms. Dueto the previously reported higher tolerance of fungal growth thanbacterial growth to metal exposure (Rajapaksha et al., 2004), wehypothesized (H4) that the Cd addition in the microcosms wouldinhibit bacterial more than fungal growth, and lead to a reducedrate of decomposition.

2. Materials and methods

2.1. Field experiments

The study plots were established in April 2014 in GedhusPlantation (56� 16.3922N; 9� 5.672E) located in mid Jutland,Denmark, 51 m above sea level. The climate is temperate with amean annual temperature of 8.4 �C and a mean annual precipita-tion of 850 mm. A Norway spruce (Picea abies (L.) Karst.) mono-culture stand was established 57 years before ash treatments wereinitiated. The forest is a second-generation forest on land previ-ously occupied by heathland, with a well-developed podzolationformed on a well-drained, sandy glacial till. The following treat-ments were performed in randomized split block design: control,and wood ash at three different application rate 0, 3, 4.5 and6 t ha�1, with 20 � 25 m plots including 24e39 trees per plot.Treatments were organized in a randomized block design withthree blocks each including all treatments (n ¼ 3). These arehenceforth referred to as “big plots”. In a proximal area, a reducedscale experiment including higher application rates was alsoestablished. A control treatment (0 t ha�1) and wood ash at fivedifferent ash application rates 3, 9, 15, 30, 90 t ha�1 were estab-lished, also in a randomized block design with 2 � 2 m plots fivereplicate blocks (n ¼ 5). These are henceforth referred to as“extreme plots”. In both sets of field experiments wood ash wasspread manually with a spade evenly over the soil surface, in asingle addition. The wood ash used in this experiment is a woodchip fuel mixed ash from a combustion plant in Brande, Denmark,with a particle size smaller than 1 mm, a pH of 13, and a Cd contentof 4 mg kg�1.

2.2. Soil sampling

Soil samples were collected in January 2015 using a soil corer5 cm in diameter to 3 cm depth, including only O-horizon. Com-posite samples were formed from at least 4 randomly taken cores.All 12 big plots and 30 extreme plots were sampled. The fresh soilwas sieved through 6 mm mesh. Samples were stored at 4 �C forone week before characterization of field experiments were con-ducted and microcosm experiments run.

2.3. Microcosm laboratory incubation experiments of ash, lime, andCd

Laboratory experiments were conducted to isolate the potentialliming (pH) and Cd effects. Microcosms were established in plastic200 mL containers with 30 g of soil from the control treatments ofthe big plots. Soil samples were exposed to three levels of ash, lime

Table 1Amounts added of the different treatments with ash, lime, and Cd to themicrocosmsfor the laboratory incubation experiment.

Treatments Low Medium High

Ash 0.4 t ha�1 1.5 t ha�1 6 t ha�1

Lime 0.1 t ha�1 0.5 t ha�1 2 t ha�1

Cd 0.006 M 0.03M 0.1MLime þ Cd 2 t ha�1þ0.006 M 2 t ha�1þ0.03 M 2 t ha�1þ0.1 M

(CaCO3), Cd (CdSO4), or limeþ Cd (Table 1) with three replicates foreach treatment (n ¼ 3) maintaining the replicate blocks of the bigplot field experiment. Six time-points were sampled; 1 d, 2 d, 5 d,9 d, 19 d and 36 d; and fungal and bacterial growth rates wereestimated as well as respiration rate (method descriptions followbelow). At the final time-point, PLFA, along with bacterial pH re-lationships and Cd tolerance were evaluated as described below.

2.4. Microbial measurements used for field- and microcosmexperiment characterizations

The bacterial growth was estimated using leucine incorporationin bacteria extracted from soil using the homogenization/centri-fugation technique (Bååth et al., 2001) with modifications (Rousket al., 2009; Supplemental methods). Fungal growth and biomassconcentrations were estimated using the technique of acetateincorporation into ergosterol (Newell and Fallon, 1991) with mod-ifications (Rousk et al., 2009; Supplemental methods). The respi-ration rate and SIR-biomass were estimated in headspace vialsmeasured on a gas chromatograph equippedwith amethanizer anda flame ionization detector (Supplemental methods). The soil mi-crobial community structure was analyzed by phospholipid fattyacid (PLFA) (Frostegård et al., 1993) with modifications includingusing pre-packed solid phase silica columns (Bond Elut 1 cc,100 mg, Agilent, USA) for the lipid fractionation (Supplementalmethods).

2.5. Bacterial community trait distributions: pH relationships andCd tolerance

The pH relationships of bacterial communities were assessed bymeasuring bacterial growth (as above) adjusting the pH in thebacterial suspension to a range of different pH levels (Bååth, 1996;Fern�andez-Calvi~no et al., 2011). Aliquots (1.35 mL) of the bacterialsuspensionwere transferred to 2mLmicrocentrifugation tubes thatcontained 0.15 mL pH buffer. The pH of the bacterial suspensionwas adjusted to 11 different verified pHs using buffers, between pH4 (final concentration 1.1 mM K2HPO4 and 0.5 mM citric acid) and 8(final concentration 0.25 mM KH2PO4 and 6.4 mM K2HPO4). Inaddition one sample was treated with only water instead of buffer(in situ pH). The buffer concentrations used did not affect bacterialLeu incorporation rates within the short time-frame studied(Fern�andez-Calvi~no et al., 2011). The bacterial growth was thenestimated using Leu incorporation as described above.

The Cd tolerance of the bacterial community was estimatedsimilarly by exposing 1.35ml of the bacterial suspension to 150 ml ofdifferent concentrations of CdSO4 and measuring bacterial growthas above. The final CdSO4 concentrations in the bacterial suspen-sionwere between 24 mmol and 10mmol per l bacterial suspension.Distilled water was used as a control.

2.6. Statistical analyses

Differences in studied microbial variables in both sets of fieldexperiments were separately analyzed using a randomized blockANOVA with ash treatment as a fixed factor and block as a randomfactor, and Tukey's HSD pair-wise comparisons were used to correctfor multiple comparisons. In the microcosm experiments estimatesof microbial growth rates and respiration over the course of theexperiment were summed up in cumulative values during the 36day study which subsequently were subjected to statistical tests.Here, the effect of treatment concentration was evaluated inseparate one-way ANOVAs for each of the different treatments (ash,lime, Cd and limeþ Cd, yielding 4 ANOVAs per variable). For factorswithmultiple levels, Tukey's HSD pair-wise comparisons were used

C. Cruz-Paredes et al. / Soil Biology & Biochemistry 112 (2017) 153e164156

to correct for multiple comparisons. All values were log-transformed prior to analysis to stabilize variance. Linear re-gressions were used to establish relationships between measuredmicrobial variables (fungal and bacterial growth, respiration,biomass estimates, etc.) and soil pH levels. All the analyses weredone with the R project 3.1 (R Core Team, 2014).

The composition of the microbial PLFAs were analyzed with aprincipal component analysis (PCA) after standardizing to unitvariance using Multi Variate Statistical Package (MVSP) 3.0 (Kovachcomputing services, Anglesey, Wales, UK). To test for the factorsthat affected the PLFA composition in the microcosm experiments,the PLFA data was split into 4 subsets defined by treatment factors,as for the othermicrobial variables (see above). The determined PCswere subjected to ANOVAs to test for treatment differences withTukey HSD corrections for multiple comparisons, as described forthe other microbial variables above.

In order to estimate microbial tolerance to Cd, dose-responserelationships were determined. To do this, values measured atdifferent levels of Cd additions were first normalized by dividingthem by the average of the values measured in the samples thatreceived no Cd and where no inhibition was observed. Normalizedvalues would then fall on a range between 1 (no inhibition ofprocess) and 0 (complete inhibition of process). Values obtained inrepeated runs of the experiment were combined to generate asingle curve. Log (IC50) values (the logarithm of the metal con-centration resulting in 50% inhibition) were estimated using a lo-gistic model (Eq. (1)):

y ¼ c.h

1þ ebðx�aÞi

(1)

where y is the relative normalized process rate, x is the logarithm ofthe Cd concentration, a is the value of log (IC50), b is a fittedparameter (slope) indicating the inhibition rate and c is the processrate measured in the control without added Cd. Kaleidagraph 4.5.2(Synergy Software, Reading, PA, USA) was used to fit inhibitioncurves using the equation above using least-square curve-fits andPearson's R estimates. To statistically compare tolerance betweentreatments, 95% confidence intervals were estimated for the log(IC50) values based on the logistic model. We used the criterion ofnon-overlapping 95% confidence intervals as our significancethreshold, which has been shown to be a conservative threshold forstatistical comparison of single value estimates along with esti-mates of variance derived from the fitted curves (Payton et al.,2000; Rath et al., 2016).

The pH relationships for bacterial communities were estab-lished using a double-logistic function, recently pioneered for thisuse in aquatic samples (Bååth and Kritzberg, 2015), using

Table 2Soil chemical properties in the different ash applications in the big and the extreme plo

Treatments pH Soil organ

Big plots0 t ha�1 4.05 ± 0.1b 79.8 ± 8.23 t ha�1 4.36 ± 0.2ab 74.3 ± 154.5 t ha�1 4.03 ± 0.1b 74.1 ± 9.16 t ha�1 4.77 ± 0.1a 77.0 ± 14Extreme plots0 t ha�1 4.22 ± 0.1d 63.6 ± 8.93 t ha�1 4.68 ± 0.1cd 85.2 ± 6.29 t ha�1 5.59 ± 0.4bc 79.7 ± 4.515 t ha�1 6.12 ± 0.5ab 84.6 ± 4.230 t ha�1 7.39 ± 0.2a 54.6 ± 7.190 t ha�1 7.09 ± 0.5ab 62.0 ± 10

abc Different letters show significant differences (p < 0.05). Soil pH was measured in wateestimate SOM. Water content was estimated gravimetrically (105 �C during 24 h).

Kaleidagraph 4.5.2 (Synergy Software, Reading, PA, USA). This isbased on the assumption of two inhibition curves, where deviationfrom the optimal pH for growth decreases towards both low andhigh pHs.

Bacterial growth

¼ Copt.n

1þ exphBlow�pH

�pH� Alow�pH

�io

þ Copt.n

1þ exphBhigh�pH

�pH� Ahigh�pH

�io� Copt

(2)

where B is a fitted slope parameter indicating the rate of decreasetowards lower or higher pHs. A is a fitted parameter defining the pHresulting in 50% inhibition of bacterial growth (IC50) towards loweror higher pH, and Copt is the bacterial growth at the optimal pH. Theindices “low-pH” and “high-pH” denote which part of the curve isdescribed (towards lower or higher pH compared with the sample'soptimum, respectively). The slope (“B”) and IC50 parameters (“A”)parameters were numerically fitted using the Levenberg-Marquardt algorithm in Kaleidagraph 4.5.2. Bacterial growth datawere normalized prior to establishing the pH-tolerance curves,yielding Copt values of unity. As such, the highest values in all curveswere set to unity, to only resolve the pH-trait distribution differ-ences. Following normalization, the pH at optimal growth wassolved from equation (2) with bacterial growth set to 1. Followingthe determination of the optimal pH for growth for all replicates,one-way ANOVAs were used to compare optimal pH betweentreatments using Tukey's HSD corrections for multiple compari-sons, as described for the other studied microbial variables above.In addition, linear regression analyses were used to link the meanpH optima of each treatment to soil pH.

3. Results

3.1. Soil chemical properties

In the field experiments, soil pH increased significantly with ash(big plots F3,6 ¼ 7.6; p ¼ 0.02; extreme plots F5,20 ¼ 13; p < 0.001;Table 2, Tables S1, S2). In the “big plots” we only found an increasefrom pH 4 to 4.8 in the highest ash application used (6 t ha�1). Inthe reduced scale experiments, “extreme plots”, we found that thepH increased by ca. one unit with 9 t ha�1, by ca. two units with15 t ha�1, and by ca. three units with the two highest applications ofash. The SOM andwater content were unaffected by ash treatmentsin the big plots (Table 2 and Table S1). In the extreme plots, SOMcontent was significantly different (F5,20 ¼ 4.3; p ¼ 0.008; Table 2and Table S2), but water content was unaffected.

ts.

ic matter (% dry matter) Water content (% wet weight)

74.4 ± 1.9.5 72.3 ± 5.9

71.5 ± 3.4.3 73.7 ± 4.1

ab 68.3 ± 3.0a 74.5 ± 1.9ab 72.2 ± 2.1a 74.1 ± 0.8b 65.4 ± 2.1.3ab 67.6 ± 4.6

r extractions in 1:5 (w:w) ratios. Loss on ignition at 600 �C during 24 h was used to

Fig. 1. Functional responses to wood ash application in the two sets of field experi-ments (Big plots, and Extreme plots, respectively). Values (mean ± SEM) for a) bacterialgrowth, b) fungal growth and c) respiration rate in the big plots (circles) and theextreme plots (squares). Different letters show significant differences (p < 0.05).

C. Cruz-Paredes et al. / Soil Biology & Biochemistry 112 (2017) 153e164 157

3.2. Microbial responses to experimental treatments

3.2.1. Microbial growth, respiration and biomassIn the field experiments, bacterial growth measured by [3H]Leu

incorporation increased significantly with higher application ratesof ash in the extreme plots (F5,20 ¼ 4.7; p¼ 0.005; Table S2), but notin the big plots (Fig. 1a; Table S1). Applications from 3 to 6 t ha�1

had minor effects, while applications from 9 to 90 t ha�1 increasedbacterial growth by ca. 3e4-fold. This resulted in positive linearrelationship between bacterial growth and soil pH (R2 ¼ 0.50;p < 0.001). Fungal growth measured by acetate incorporation intoergosterol decreased significantly with higher ash applications inthe extreme plots (F5,20¼ 7.3; p < 0.001; Table S2), but not in the bigplots (Fig. 1b; Table S1). While only marginal effects could be seenup to 15 t ash ha�1, the highest application rates reduced fungalgrowth by half (Fig. 1b). Although the respiration rate tended toincrease with higher ash application, the differences were notsignificant (Fig. 1c; Tables S1 and S2).

The treatments administered in the microcosms resulted inpronounced changes in microbial growth and respiration over time(Figs. S1eS3, Fig. 2). With higher ash additions, cumulative bacte-rial growth increased (F3,6 ¼ 92; p < 0.001; Table S3) while cu-mulative fungal growth was unaffected (Fig. 2a; Table S3). Whenadding lime, we reproduced the effect of ash (F3,6 ¼ 64; p < 0.001;Table S3) and could observe nearly identical results (Fig. 2b). In themicrocosms with the addition of only Cd the opposite effect wasobserved; the cumulative bacterial growth decreased (F3,6 ¼ 40;p < 0.001; Table S3) and the cumulative fungal growth increased(F3,6 ¼ 340; p < 0.001; Table S3) (Fig. 2c). The microcosm treatedwith lime and Cd resulted in no differences in cumulative bacterialgrowth and a significant decrease (F3,6 ¼ 20; p¼ 0.002; Table S3) infungal growth (Fig. 2d). The cumulative respiration rate tended toincrease with the ash and lime treatments (Fig. 2a and b) andsignificantly decreased (F3,6 ¼ 12; p ¼ 0.006; Table S3) in the Cdtreatments (Fig. 2c).

The microbial biomass estimated with the SIR method alongwith the fungal biomassmeasured as ergosterol were unaffected bythe ash application in both big plots and extreme plots (Table 3,Tables S1, S2). Total, bacterial and fungal PLFA concentrations werenot significantly affected by ash addition in the big plots butincreased in the extreme plots (F5,19 ¼ 3.8; p ¼ 0.01, F5,19 ¼ 4.2;p ¼ 0.009, F5,19 ¼ 4.1; p ¼ 0.01, respectively; Table 3, Tables S1, S2).

In the microcosm experiment, total and bacterial PLFA concen-trations were unaffected by all treatments (Table 4 and Table S3).Total PLFA ranged between 610 and 770 nmol g�1 and bacterialPLFA were between 300 and 420 nmol g�1. Fungal PLFA concen-trations decreased in the treatment with Cd andwith lime and highCd additions (F3,6 ¼ 8.5; p ¼ 0.01, F3,6 ¼ 15; p ¼ 0.004, respectively;Table S3) with a value of 28 ± 1 nmol g�1 and 19 ± 2 nmol g�1

compare to the control with 41 ± 2 nmol g�1. In contrast, the fungalbiomarker ergosterol measured in the last sampling day, did notdiffer between treatments (Table 4 and Table S3).

3.2.2. Microbial PLFA compositionIn the field experiments, the microbial PLFA composition was

affected by ash application in the extreme plots (Fig. 3a andFig. S4c) with similar tendencies in the big plots (Fig. S4a). The PC1of the PLFA pattern accounted for 28.4% of the variation in the data,while PC2 accounted for 20.5% (Fig. 3). Changes in the microbialcommunity structure in the extreme plots were significantlydifferent along the PC2 axis (F5,19¼ 2.9; p¼ 0.004; Table S2), but notalong PC1. PC1 was correlated to soil pH (R2 ¼ 0.40; p ¼ 0.003). Theloading plot of the individual PLFAs (Fig. 3b, Fig. S4b, d) showed thatthe effect of ash on the PLFA pattern was driven by higher con-centrations of the PLFAs 16:1u5, 18:1u7 16:1u7c and cy17:0 at

higher ash applications, while at lower applications PLFAs i16:0,cy19:0, 10Me17:0, 10Me16:0 and i15:0 were in higher relativeabundances.

Fig. 2. Functional responses to wood ash application, lime application, and Cd application in microcosm experiments. Cumulative values (mean ± SEM) for bacterial growth,respiration rate and fungal growth in the microcosms with different levels of a) ash, b) lime (CaCO3), c) Cd (CdSO4) and d) lime and Cd.

Table 3Microbial biomass (total, bacterial and fungal) evaluated by SIR, ergosterol concentration, and total, bacterial and fungal PLFAs in the different ash applications in the big andextreme plots.

Treatments SIR (mg C g�1) Ergosterol (mg) Total PLFA (nmol g�1) Bacterial PLFA (nmol g�1) Fungal PLFA (nmol g�1)

Big plots0 t ha�1 2.0 ± 0.2 14.7 ± 0.8 808 ± 79 390 ± 33 56 ± 43 t ha�1 1.7 ± 0.3 11.2 ± 1.0 689 ± 201 352 ± 90 46 ± 134.5 t ha�1 1.5 ± 0.3 11.9 ± 0.9 672 ± 60 352 ± 14 38 ± 86 t ha�1 2.2 ± 0.3 15.2 ± 2.3 765 ± 163 342 ± 55 60 ± 23Extreme plots0 t ha�1 1.4 ± 0.2 11.9 ± 1.7 645 ± 101ab 324 ± 33ab 37±9b

3 t ha�1 2.2 ± 0.2 16.4 ± 1.4 896 ± 63a 427 ± 29a 70±7a

9 t ha�1 2.9 ± 0.6 17.2 ± 0.9 883 ± 101a 431 ± 43a 66 ± 13ab

15 t ha�1 3.1 ± 0.3 16.4 ± 1.1 730 ± 34ab 347 ± 16ab 47±5ab

30 t ha�1 2.5 ± 0.2 11.0 ± 1.7 530 ± 80b 267 ± 37b 34±7b

90 t ha�1 2.7 ± 0.5 15.1 ± 3.0 649 ± 113ab 320 ± 54ab 44±9ab

ab Different letters show significant differences (p < 0.05).

C. Cruz-Paredes et al. / Soil Biology & Biochemistry 112 (2017) 153e164158

Also in the microcosm experiments, the microbial PLFAcomposition responded to the treatments (Fig. 3c and Fig. S5). ThePC1 of the PLFA pattern accounted for 35.4% of the variation in thedata, while PC2 accounted for 17.1% (Fig. 3). As in the field experi-ment the PC1 was correlated to soil pH (R2 ¼ 0.87; p < 0.001). Theloading plot of the individual PLFAs (Fig. 3d) showed that the effectof ash and lime on the PLFA pattern was higher concentrations of

the PLFAs 16:1u5, 16:1u7c, 18:1u7 and cy17:0 as in the fieldexperiment, but also others including cy19:0 and i17:0 at thehighest ash and lime applications. At lower ash and lime applica-tions and Cd applications PLFAs i16:0, 10Me17:0, 10Me16:0,10Me18:0, 16:1u7t and br18:0 occurred at higher relative abun-dances (Fig. 3d). The microcosms with lime and Cd showed verysimilar PCA scores to the high ash and lime applications (Fig. 3c).

Table 4Microbial biomass (total, bacterial and fungal) evaluated by ergosterol concentration, and total, bacterial and fungal PLFAs in the different microcosms at the end of the in-cubation experiment (day 36).

Treatments Ergosterol* (mg g�1) Total PLFA (nmol g�1) Bacterial PLFA (nmol g�1) Fungal PLFA (nmol g�1)

Ash0 t ha�1 10.7 ± 0.8 766 ± 42 396 ± 25 41 ± 20.4 t ha�1 10.2 ± 0.4 739 ± 70 376 ± 40 40 ± 51.5 t ha�1 9.9 ± 0.2 746 ± 64 392 ± 38 37 ± 46 t ha�1 9.4 ± 0.4 731 ± 56 386 ± 26 33 ± 4Lime0 t ha�1 10.7 ± 0.8 766 ± 42 396 ± 25 41 ± 20.1 t ha�1 11.1 ± 1.1 736 ± 73 385 ± 39 37 ± 40.5 t ha�1 9.2 ± 1.0 743 ± 101 390 ± 57 37 ± 42 t ha�1 9.9 ± 0.5 687 ± 9 351 ± 16 33 ± 3Cd0 M 10.7 ± 0.8 766 ± 42 396 ± 25 41±2a

0.006 M 10.8 ± 1.4 682 ± 19 349 ± 10 37±4ab

0.03M 11.6 ± 1.4 682 ± 46 357 ± 26 28±1b

0.1M 10.9 ± 1.1 633 ± 41 298 ± 19 38±3a

Lime þ Cd0 t ha�1 þ 0 M 10.7 ± 0.8 766 ± 42 396 ± 25 41±2a

2 t ha�1þ0.006 M 10.2 ± 0.9 755 ± 13 390 ± 4 34±3a

2 t ha�1þ0.03 M 9.5 ± 1.2 774 ± 38 421 ± 17 27±1ab

2 t ha�1þ0.1 M 11.0 ± 1.7 609 ± 81 329 ± 44 19±2b

ab Different letters show significant differences (p < 0.05).*Last sampling point value.

Fig. 3. Microbial community PLFA composition responses to different treatments. . The effect of wood ash applications on the PLFA pattern of the soil microbial community in thefield experiments (Big and Extreme plots, respectively) showed in a) PCA plot, score values (mean ± SEM) for big plots (circles) and the extreme plots (squares); b) The loadings ofthe individual PLFAs from the PCA data. PLFAs to the right bottom in the plot indicate those that are more common in high ash applications, while those in the left indicate thosethat are more common in the lower applications. The effect of the different treatments on the PLFA composition of the soil microbial community in the microcosms showed in c)PCA plot, score values (mean ± SEM); d) The loadings of the individual PLFAs from the PCA data. PLFAs to the right bottom in the plot indicate those that are more common in thehighest ash and lime applications (high pH), while those in the left indicate those that are more common in the Cd alone applications (low pH).

C. Cruz-Paredes et al. / Soil Biology & Biochemistry 112 (2017) 153e164 159

C. Cruz-Paredes et al. / Soil Biology & Biochemistry 112 (2017) 153e164160

However, as Cd increased the scores separated slightly along PC1and more pronounced, along PC2. The microcosms with only Cdaddition separated from the control in both PC1 and PC2 withhigher rates of Cd application, with a stronger separation along PC1(Fig. 3c). To statistically compare the effects by the different treat-ment factors, separate PCAs were also run on subsets of themicrocosm samples defined by treatment factors (Fig. S5, Table S3).These analyses showed that the PLFA compositionwas substantiallyaffected by ash, lime, Cd, and the combined ash and Cd treatmentwhen compared with the untreated control samples (P < 0.001 forall four ANOVAs; Table S3).

3.2.3. Bacterial community trait distribution for Cd toleranceIn the field experiments, the logistic model described the dose-

response curves well (with a mean R2 ¼ 0.89) and the 95% confi-dence intervals of estimates EC50 values overlapped, showing thatthe bacterial tolerance to Cd did not increase with the applicationrate of ash (Fig. S6). For the microcosm experiments, the logisticmodel described the dose-response curves well (with a meanR2 ¼ 0.99), enabling comparisons of tolerance. Ash and limetreatments did not increase the bacterial tolerance to Cd in themicrocosms (Fig. 4a and b). Rather, we found that the IC50decreased, as shown by non-overlapping 95% confidence intervalsof IC50 values, suggesting a lowered tolerance, when either ash orlime were added (Figs. S7a and b). However, in the microcosms

Fig. 4. Cd-trait distribution responses to wood ash application, lime application, and Cdlationships in the different microcosms a) ash, b) lime (CaCO3), c) Cd (CdSO4) and d) lime an

with Cd additions we were able to induce tolerance to Cd (Fig. 4cand d), leading to 95% confidence intervals clearly distinguishable(Figs. S7c and d). The logistic model described the dose-responsecurves for these treatments well (with a mean R2 ¼ 0.97). Thehighest Cd addition had an IC50 value of 30 mM while for controlsoils the IC50 was 0.1 mM (Fig. S7).

3.2.4. Bacterial community trait distribution for pH relationshipsIn the field experiments, the bacterial community trait distri-

butions for pH varied with the ash application rate in the extremeplots (R2 ¼ 0.82; p ¼ 0.008). The increase in pH induced by the ashapplicationwas matched by a similar change in the pH-relationship(Fig. 5a). As such, with higher ash application rates, the curvesshifted to be centered around higher pH optima. The double logisticmodel described the relationships well (with a mean R2 ¼ 0.82).The shape of the curves appeared to remain similar. This was shownby the difference in pH between 10% inhibition at low pH and 10%inhibition at high pH, which remained similar at ca 1.0e1.3 at allash application rates, and had no correlation with soil pH. Thebacterial community trait distributions in the big plots were shiftedby a 1-pH unit increase in pH optimum in the 6 t ha�1 additiontreatments (Fig. 5a). The responses in bacterial community traitdistributions were similar in the extreme plots. At higher additionrates, over 9 t ha�1, a further increase of 1.5 units higher than thecontrol was observed (Fig. 5a). Finally, the pH optimumwas up to 2

application in microcosm experiments. Microbial tolerance to Cd dose-response re-d Cd at the end of the incubation experiment (day 36), estimated with logistic curves.

Fig. 5. pH-trait distribution responses to wood ash application in the two sets of field experiments (Big plots, and Extreme plots, respectively). (a) pH relationships for bacterialcommunities estimated with double-logistic curves for the big and extreme plots with different ash applications and (b) the relationship between the pH-optima and the soil pH ineach big plot (circles), extreme plot (squares) and microcosm (diamonds) estimated with linear regressions.

Fig. 6. pH-trait distribution responses to wood ash application, lime application, and Cd application in the microcosm experiments. pH relationships for bacterial communitiesestimated with double-logistic curves in the different microcosms a) ash, b) lime (CaCO3), c) Cd (CdSO4) and d) lime and Cd at the end of the incubation experiment (day 36).

C. Cruz-Paredes et al. / Soil Biology & Biochemistry 112 (2017) 153e164 161

units over the control in the 90 t ha�1 treatment. The pH-optimaestimated correlated with the soil pH in each treatment, indi-cating a strong positive relationship (Fig. 5b), but with a gradualincrease in the offset of the pH optima over the soil pH towards acidpHs.

In the microcosm experiments, the double logistic modeldescribed the relationship curves for pH well (with a meanR2 ¼ 0.97). The responses of the community trait distributions forpHwere similar to those found in the field (Fig. 6a and b). However,the curves weremuchwider in the low concentration treatments in

C. Cruz-Paredes et al. / Soil Biology & Biochemistry 112 (2017) 153e164162

the microcosms compared to the field treatments (10% inhibitionlevel pH spans had a mean width of 1.1 pH units in the field ex-periments). In the microcosm experiments, 10% inhibition spansranged from 2.2 to 1.1 pH units with a mean of 1.7. There was nocorrelation between the size of the 10% inhibition span and soil pH.The bacterial trait distribution for pH shifted to have higher pH-optimum following the ash and lime additions. With higher addi-tion rates of ash or lime the soil pH increased and therefore thebacterial pH-trait distribution followed (Fig. 6a and b). Addition of6 t ha�1 ash or 2 t ha�1 lime increased the pH-optimum comparedto the control by almost two units, while the rest of the additionsresulted in smaller changes (Fig. 6a and b). The Cd treatmentsresulted in the opposite shift in the bacterial community trait dis-tributions for pH, shifting the optimal pH to more acid values(Fig. 6c and d). The community pH optima correlated positivelywith the soil pH in each treatment, similarly to results from thefield experiment, and aligning with the same slope (Fig. 5b).

4. Discussion

4.1. Microbial responses assigned to pH changes

We hypothesized (H1) that the dominant factor affecting mi-crobial communities and processes would be the soil pH changeinduced by the ash application. In the field experiments the mi-crobial PLFA composition changed drastically with the addition of30 t ha�1 and 90 t ha�1. We observed an increase in the relativeabundance of specific PLFAs such as 16:1u7c, 16:1u5 and 18:1u7with a decrease in the relative abundance of i16:0 and cy19:0.These patterns have previously been identified as biomarkers forpH-induced microbial PLFA composition changes (Bååth andAnderson, 2003; Nilsson et al., 2007; Rousk et al., 2010b; Rouskand Bååth, 2011). In the microcosms with ash and lime we couldverify these results. The additions of 2 t ha�1 of lime and 6 t ha�1 ofash changed the relative abundance of specific PLFAs in a similarway as in the field experiments. In addition, most of the variation inPLFA composition in both field and microcosm experiments werecorrelated with pH (i.e. PC1 correlated with soil pH; Fig. 3)strengthening the evidence for the causal link between the changesof the community structure and that of pH.

The pronounced changes of community structures coincidedwith shifts in community trait distributions for pH. The pH optimaof community trait distribution curves for pH correlated with theenvironmental soil pH (Fig. 5b), as predicted (H1). These responseswere consistent in both sets of field experiments and in themicrocosm experiments with ash and lime applications. This rela-tionship was similar to those found before in natural and experi-mental ranges of pH (Bååth, 1996; B�arcenas-Moreno et al., 2016;Fern�andez-Calvi~no et al., 2011).

Similar responses in bacterial community pH optima werefound in the microcosm with ash and lime compared to the fieldexperiment. Still, we observed interesting qualitative differencesbetween the shorter-term microcosm experiments and the fielddata. The widths of pH trait distribution curves in the microcosmswere wider compared with the narrower curves of the field ex-periments (c.f. Figs. 5a and 6). The curves in the field experimentshowed that a 10% inhibition will occur with a change of ±1.1 pHunits, while the wider curves of the microcosm experiments yiel-ded more variable indices with an average of ±1.7 pH units. Bååthand Kritzberg (2015) found that pH optima for bacterial commu-nity growth in lakes and streams were closely correlated with insitu pHs, and that the variation in the width of the pH tolerance (at10% inhibition) varied between 1 and 3 pH units. They proposedthat these differences inwidth could be attributed to the stability ofpH in the environment. This would suggest that environments

exposed to larger recent fluctuations in pH, or other disturbance,would have pH tolerance curves with broader optima compared tomore stable environments, without recent perturbation. In the fieldexperiments wood ash was added ca. 1 year before analysis. Themicrocosms were constructed from soil samples following ho-mogenization and treatment with various applications during 36days. It would be reasonable, therefore, that the field experimentalcommunity trait alignment had reached a situation closer to stable-state, while the shorter-term microcosm communities were still ina state of ongoing change undergoing further alignment of its traitvariation to the new environmental conditions.

4.2. Microbial responses assigned to Cd

The microcosm experiments were designed to disentangle thetoxicological effects of ash amendments, and characterize the dy-namics of how the induction of tolerance to Cd developed overtime. In the field experiment Cd additions corresponded to12 kg ha�1 in the 3 t ha�1 ash application, up to 360 kg ha�1 in the90 t ha�1 ash application. In the microcosms, up to 24 kg ha�1

(0.12 mg g�1 soil) Cd was added with 6 t ha�1 ash application.When Cd was added in a solution the concentrations ranged from0.11 up to 1.69 mg g�1 soil. With an assumed bulk density of ca1 g cm�3 for the fresh soil in the top 3 cm of the O-horizon thiswould correspond to ca. 4 and 60 kg Cd ha�1. With soluble Cd ad-ditions, we found that the community structure evaluated withPLFAs changed drastically, contrasting with the results found fromash-associated Cd in the field experiment. In the microcosmstreated with Cd solution we observed an increase in the relativeabundance of i16:0, 10Me16:0, a17:0 and 10Me17:0 previouslyassociated with metal toxicity (Åkerblom et al., 2007; Frey et al.,2006; Pennanen et al., 1996). In addition, in the microcosms withCd solution and lime the relative abundance of cy19:0 increased,which is a marker previously associated with Cd stress (Åkerblomet al., 2007).

One of the major concerns of the use of ash is the content ofheavy metals such as Cd. However, we did not find any response inmicrobial Cd tolerance (up to 90 t ha�1) in accordance with ourhypothesis (H2). In the field experiments and the microcosms withash, the bacterial community trait distribution showed that Cdcontained in wood ash did not induce tolerance to Cd (Fig. S6 andFig. 4a). On the other hand, soluble Cd additions with or withoutlime clearly induced tolerance to Cd (Fig. 4c and d). Many heavymetals like Cd are less bioavailable at high pH values (Brady andWeil, 1996). Previous studies reported that Cd in ash is notbioavailable (Fritze et al., 2001; Perki€om€aki et al., 2003), evenwhenadded at rates of 400 and 1000 mg Cd g�1 ash, because it is probablyin forms with low water solubility, such as CdO or CdSiO3 (Hansenet al., 2001). Studies in contaminated soils have shown that ispossible to induced tolerance to Cd (Almås et al., 2004;Witter et al.,2000), as was also observed inmicrocosms treatedwith Cd solution(Fig. 4c and d), emphasizing that the bioavailable metal concen-tration in soil is decisive. As previously reported, we observed thatwood ash could even counteract the toxic effect of Cd due to itscapacity to increase soil pH and decrease Cd bioavailability. Thesefindings imply that even though wood ash application certainlyaffects the microbial community, the changes are not related to theCd content.

4.3. Ash-induced changes in microbial functions

We hypothesized (H3) that ash applications leading to higherpH in the field experiment would favor bacterial over fungalgrowth. Ash applications increased soil pH substantially, matchingresults of many earlier studies (Bj€ork et al., 2010; Demeyer et al.,

C. Cruz-Paredes et al. / Soil Biology & Biochemistry 112 (2017) 153e164 163

2001; Insam et al., 2009; Perki€om€aki and Fritze, 2002; Reid andWatmough, 2014). In response to these changes, we observed abroad shift toward higher bacterial and reduced fungal dominancein the field experiment assessed by microbial growth rates. As ex-pected, we found a similar outcome in themicrocosms treated withash and lime as in the field experiment. These results echo a seriesof recent reports where acid environments favor fungal growthwhile neutral or slightly alkaline conditions favor bacterial growth(H€ogberg et al., 2007; Rousk et al., 2010a, 2009). The mechanismsby which the pH of the soil environment regulates the relativeimportance of the decomposer groups remain unresolved (but seediscussion about candidate mechanisms in Rousk et al., 2010b).However, the extreme control by which the pH-trait distributionsof bacterial communities were matched to the soil pH of theirenvironment (Fig. 5b) strongly suggest that soil pH exerted a directeffect on the bacterial community. The shifts in fungal-to-bacterialgrowth induced by the application of ash coincided with anincreased respiration rate, matching earlier reports (e.g. Perki€om€akiand Fritze, 2002). However, the relatively limited responses inrespiration compared to shifts in fungal-to-bacterial growth seen inthe more controlled microcosms experiments (Fig. 2), suggest ahigh degree of functional redundancy within the microbial com-munity. As such, the effect of ash induced increases in C availabilitydriven by a higher solubility of SOC with higher pH (Curtin et al.,2016) are more likely to have characterized the higher respirationobserved in the field experiment than the induced changes in themicrobial decomposer groups.

When assessing functional consequences of wood ash additions,our approach to use microbial trait distributions to link microbialresponses to Cd could link none of the observed variation in thefield treatments to Cd. This was consistent with results from mi-crocosms treated with ash where we did not find functional effectsdue to Cd added with ash. However, as we hypothesized (H4) sol-uble Cd addition at relatively high concentrations in the micro-cosms inhibited bacterial more than fungal growth.With soluble Cdadditions, we observed a shift toward higher fungal and reducedbacterial dominance. It is known that bacteria and fungi tend tocompete for resources (Rousk et al., 2008). Therefore, these resultssuggest that Cd reduced bacterial growth which allowed fungi toproliferate; however, overall microbial activity, measured asrespiration, remained unaffected. Presumably this was due to thefungal growth increases compensating for bacterial growth re-ductions terms of overall decomposer function.

5. Conclusions

Using the comparative analysis of two sets of field experimentswith wood ash application to parallel microcosm experimentsincluding wood ash along with its two candidate component fac-tors pH (lime) and Cd we could show that the microbial structuraland ecosystem functional responses induced by wood ash wereassociated with its pH increasing effect. Moreover, by also investi-gating how ash affected the trait distributions of the microbialcommunities we could unambiguously link the detected microbialresponses to being directly induced by soil pH changes rather thancaused by indirect effects of the ash. Finally, using the same ap-proaches, we could determine that the levels of Cd associated withthe wood ash applied (360 kg Cd ha�1 carried with the 90 t ha�1

wood ash application) has no effect on the soil microbial commu-nity or their ability to provide decomposer functioning as proxiedby growth rates and respiration.

Acknowledgments

This work was supported by a STSM grant from the EU Cost

Action “BioLink” (fp1305), and by grants from the Swedish researchcouncil (VR, grant no 2015e04942) and the Swedish researchcouncil Formas (grant no 942-2015-270). This study was part of theDanish ASHBACK project (grant no 0603-00587B).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.soilbio.2017.05.004.

References

Åkerblom, S., Bååth, E., Bringmark, L., Bringmark, E., 2007. Experimentally inducedeffects of heavy metal on microbial activity and community structure of forestmor layers. Biology and Fertility of Soils 44, 79e91. http://dx.doi.org/10.1007/s00374-007-0181-2.

Allison, S.D., 2012. A trait-based approach for modelling microbial litter decom-position. Ecology Letters 15, 1058e1070. http://dx.doi.org/10.1111/j.1461-0248.2012.01807.x.

Almås, Å.R., Bakken, L.R., Mulder, J., 2004. Changes in tolerance of soil microbialcommunities in Zn and Cd contaminated soils. Soil Biology and Biochemistry36, 805e813. http://dx.doi.org/10.1016/j.soilbio.2004.01.010.

Bååth, E., 1996. Adaptation of soil bacterial communities to prevailing pH indifferent soils. FEMS Microbiology Ecology 19, 227e237. http://dx.doi.org/10.1016/0168-6496(96)00008-6.

Bååth, E., Anderson, T.-H., 2003. Comparison of soil fungal/bacterial ratios in a pHgradient using physiological and PLFA-based techniques. Soil Biology andBiochemistry 35, 955e963. http://dx.doi.org/10.1016/S0038-0717(03)00154-8.

Bååth, E., Frostegård, Å., Pennanen, T., Fritze, H., 1995. Microbial communitystructure and pH response in relation to soil organic matter quality in wood-ashfertilized, clear-cut or burned coniferous forest soils. Soil Biology andBiochemistry 21, 229e240.

Bååth, E., Kritzberg, E., 2015. pH tolerance in freshwater bacterioplankton: traitvariation of the community as measured by leucine incorporation. Applied andEnvironmental Microbiology 81, 7411e7419. http://dx.doi.org/10.1128/AEM.02236-15.

Bååth, E., Pettersson, M., S€oderberg, K., 2001. Adaptation of a rapid and economicalmicrocentrifugation method to measure thymidine and leucine incorporationby soil bacteria. Soil Biology and Biochemistry 33, 1571e1574.

B�arcenas-Moreno, G., Bååth, E., Rousk, J., 2016. Functional implications of the pH-trait distribution of the microbial community in a re-inoculation experimentacross a pH gradient. Soil Biology and Biochemistry 93, 69e78. http://dx.doi.org/10.1016/j.soilbio.2015.10.024.

B�arcenas-Moreno, G., Brand�on, M.G., Rousk, J., Bååth, E., 2009. Adaptation of soilmicrobial communities to temperature: comparison of fungi and bacteria in alaboratory experiment. Global Change Biology 15, 2950e2957. http://dx.doi.org/10.1111/j.1365-2486.2009.01882.x.

Berg, J., Brandt, K.K., Al-Soud, W.A., Holm, P.E., Hansen, L.H., Sørensen, S.J.,Nybroe, O., 2012. Selection for Cu-tolerant bacterial communities with alteredcomposition, but unaltered richness, via long-term Cu exposure. Applied andEnvironmental Microbiology 78, 7438e7446. http://dx.doi.org/10.1128/AEM.01071-12.

Bier, R.L., Bernhardt, E.S., Boot, C.M., Graham, E.B., Hall, E.K., Lennon, J.T.,Nemergut, D.R., Osborne, B.B., Ruiz-Gonzales, C., Schimel, J.P., Waldrop, M.P.,Wallenstein, M.D., 2016. Linking microbial community structure and microbialprocesses: an empirical and conceptual overview. FEMS Microbiology Ecology1e38.

Birgander, J., Reischke, S., Jones, D.L., Rousk, J., 2013. Temperature adaptation ofbacterial growth and 14C-glucose mineralisation in a laboratory study. SoilBiology and Biochemistry 65, 294e303. http://dx.doi.org/10.1016/j.soilbio.2013.06.006.

Bj€ork, R.G., Ernfors, M., Sikstr€om, U., Nilsson, M.B., Andersson, M.X., Rütting, T.,Klemedtsson, L., 2010. Contrasting effects of wood ash application on microbialcommunity structure, biomass and processes in drained forested peatlands.FEMS Microbiology Ecology 73, 550e562. http://dx.doi.org/10.1111/j.1574-6941.2010.00911.x.

Bradford, M.A., 2013. Thermal adaptation of decomposer communities in warmingsoils. Frontiers in Microbiology 4, 1e16. http://dx.doi.org/10.3389/fmicb.2013.00333.

Brady, N., Weil, R., 1996. The Nature and Properties of Soils, ninth ed. CollierMacmillan Publishers, London, UK.

Curtin, D., Peterson, M.E., Anderson, C.R., 2016. pH-dependence of organic mattersolubility: base type effects on dissolved organic C, N, P, and S in soils withcontrasting mineralogy. Geoderma 271, 161e172. http://dx.doi.org/10.1016/j.geoderma.2016.02.009.

Delmont, T.O., Robe, P., Cecillon, S., Clark, I.M., Constancias, F., Simonet, P.,Hirsch, P.R., Vogel, T.M., 2011. Accessing the soil metagenome for studies ofmicrobial diversity. Applied and Environmental Microbiology 77, 1315e1324.http://dx.doi.org/10.1128/AEM.01526-10.

Demeyer, A., Voundi Nkana, J.C., Verloo, M.G., 2001. Characteristics of wood ash andinfluence on soil properties and nutrient uptake: an overview. Bioresource

C. Cruz-Paredes et al. / Soil Biology & Biochemistry 112 (2017) 153e164164

Technology 77, 287e295.Díaz-Ravi~na, M., Bååth, E., 1996. Development of metal tolerance in soil bacterial

communities exposed to experimentally increased metal levels. Applied andEnvironmental Microbiology 62, 2970e2977.

Fern�andez-Calvi~no, D., Rousk, J., Brookes, P.C., Bååth, E., 2011. Bacterial pH-optimafor growth track soil pH, but are higher than expected at low pH. Soil Biologyand Biochemistry 43, 1569e1575. http://dx.doi.org/10.1016/j.soilbio.2011.04.007.

Frey, B., Stemmer, M., Widmer, F., Luster, J., Sperisen, C., 2006. Microbial activity andcommunity structure of a soil after heavy metal contamination in a modelforest ecosystem. Soil Biology and Biochemistry 38, 1745e1756. http://dx.doi.org/10.1016/j.soilbio.2005.11.032.

Fritze, H., Perki€om€aki, J., Pet€anen, T., Pennanen, T., Romantschuk, M., Karp, M.,Yrj€al€a, K., 2001. A microcosmos study on the effects of Cd-containing wood ashon the coniferous humus fungal community and the Cd bioavailability. Journalof Soils and Sediments 1, 146e150.

Fritze, H., Perki€om€aki, J., Saarela, U., Katainene, R., Tikka, P., Yrj€al€a, K., Karp, M.,Haimi, J., Romantschuk, M., 2000. Effect of Cd-containing wood ash on themicroflora of coniferous forest humus. FEMS Microbiology Ecology 32, 43e51.

Frostegård, Å., Tunlid, A., Bååth, E., 1993. Phospholipid fatty acid composition,biomass, and activity of microbial communities from two soil types experi-mentally exposed to different heavy metals. Applied and EnvironmentalMicrobiology 59, 3605e3617.

Green, J.L., Bohannan, B.J.M., Whitaker, R.J., 2008. Microbial biogeography: fromtaxonomy to traits. Science 320, 1039e1043. http://dx.doi.org/10.1126/science.1153475.

Grime, J.P., 1977. Evidence for the existence of three primary strategies in plants andits relevance to ecological and evolutionary theory. The American Naturalist 111,1169e1194. http://dx.doi.org/10.2307/2678832.

Hansen, H.K., Pedersen, A.J., Ottosen, L.M., Villumsen, A., 2001. Speciation andmobility of cadmium in straw and wood combustion fly ash. Chemosphere 45,2e7.

H€ogberg, M.N., H€ogberg, P., Myrold, D.D., 2007. Is microbial community composi-tion in boreal forest soils determined by pH, C-to-N ratio, the trees, or all three?Oecologia 150, 590e601. http://dx.doi.org/10.1007/s00442-006-0562-5.

Insam, H., Knapp, B.a., Plank, R., Holzasche, W., Von, 2009. Use of wood ash andanaerobic sludge for grassland fertilization: effects on plants and microbes. DieBodenkultur 60, 39e51.

Jacobson, S., 2003. Addition of stabilized wood ashes to Swedish coniferous standson mineral soils - effects on stem growth and needle nutrient concentrations.Silva Fennica 37, 437e450.

Jansson, J.K., 2013. Exploring Earth's dark matter. Nature 494, 40. http://dx.doi.org/10.1038/494040a.

Knapp, B.A., Insam, H., 2011. Recycling of biomass ashes: current technologies andfuture research needs. In: Recycling of Biomass Ashes, pp. 1e16.

Lauber, C.L., Hamady, M., Knight, R., Fierer, N., 2009. Pyrosequencing-basedassessment of soil pH as a predictor of soil bacterial community structure at thecontinental scale. Applied and Environmental Microbiology 75, 5111e5120.http://dx.doi.org/10.1128/AEM.00335-09.

Lekfeldt, J.D.S., Magid, J., Holm, P.E., Nybroe, O., Brandt, K.K., 2014. Evaluation of theleucine incorporation technique for detection of pollution-induced communitytolerance to copper in a long-term agricultural field trial with urban wastefertilizers. Environmental Pollution 194, 78e85. http://dx.doi.org/10.1016/j.envpol.2014.07.013.

Li, W.K., Dickie, P.M., 1987. Temperature characteristics of photosynthetic and het-erotrophic activities: seasonal variations in temperate microbial plankton.Applied and Environmental Microbiology 53, 2282e2295.

Lozupone, C.a., Knight, R., 2007. Global patterns in bacterial diversity. Proceedingsof the National Academy of Sciences of the United States of America 104,11436e11440. http://dx.doi.org/10.1073/pnas.0611525104.

Martiny, J.B.H., Jones, S.E., Lennon, J.T., Martiny, A.C., 2015. Microbiomes in light oftraits: a phylogenetic perspective. Science 350. http://dx.doi.org/10.1126/sci-ence.aac9323 aac9323-1eaac9323-8.

Mohamed, D.J., Martiny, J.B.H., 2011. Patterns of fungal diversity and compositionalong a salinity gradient. International Society for Microbial Ecology Journal 5,379e388. http://dx.doi.org/10.1038/ismej.2010.137.

Newell, S.Y., Fallon, R.D., 1991. Toward A method for measuring instantaneousfungal growth rates in field samples. Ecological Society of America 72,1547e1559.

Nilsson, L.O., Bååth, E., Falkengren-Grerup, U., Wallander, H., 2007. Growth ofectomycorrhizal mycelia and composition of soil microbial communities in oakforest soils along a nitrogen deposition gradient. Oecologia 153, 375e384.http://dx.doi.org/10.1007/s00442-007-0735-x.

Norberg, J., Swaney, D.P., Dushoff, J., Lin, J., Casagrandi, R., Levin, S.A., 2001.Phenotypic diversity and ecosystem functioning in changing environments: atheoretical framework. Proceedings of the National Academy of Sciences of theUnited States of America 98, 11376e11381. http://dx.doi.org/10.1073/pnas.171315998.

Odum, H.T., Odum, E.P., 1976. Energy Basis for Man and Nature. McGraw-Hill, NewYork, USA.

Payton, M.E., Miller, A.E., Raun, W.R., 2000. Testing statistical hypotheses usingstandard error bars and confidence intervals. Communications in Soil Science

and Plant Analysis 31, 547e551. http://dx.doi.org/10.1080/00103620009370458.Pennanen, T., Frostegard, A., Fritze, H., Baath, E., 1996. Phospholipid fatty acid

composition and heavy metal tolerance of soil microbial communities alongtwo heavy metal-polluted gradients in coniferous forests. Applied and Envi-ronmental Microbiology 62, 420e428.

Perki€om€aki, J., Fritze, H., 2004. Cadmium in upland forests after vitality fertilizationwith wood ash-a summary of soil microbiological studies into the potential riskof cadmium release. Biology and Fertility of Soils 41, 75e84. http://dx.doi.org/10.1007/s00374-004-0816-5.

Perki€om€aki, J., Fritze, H., 2002. Short and long-term effects of wood ash on theboreal forest humus microbial community. Soil Biology and Biochemistry 34,1343e1353. http://dx.doi.org/10.1016/S0038-0717(02)00079-2.

Perki€om€aki, J., Kiikkil€a, O., Moilanen, M., Issakainen, J., Tervahauta, A., Fritze, H.,2003. Cadmium-containing wood ash in a pine forest: effects on humusmicroflora and cadmium concentrations in mushrooms, berries, and needles,2451, 2443e2451. http://dx.doi.org/10.1139/X03-169.

Pettersson, M., Bååth, E., 2013. Importance of inoculum properties on the structureand growth of bacterial communities during recolonisation of humus soil withdifferent pH. Microbial Ecology 66, 416e426. http://dx.doi.org/10.1007/s00248-013-0208-1.

Pettersson, M., Bååth, E., 2004. Effects of the properties of the bacterial communityon pH adaptation during recolonisation of a humus soil. Soil Biology andBiochemistry 36, 1383e1388. http://dx.doi.org/10.1016/j.soilbio.2004.02.028.

Prosser, J.I., 2013. Think before you sequence. Nature 494, 41.R Core Team, 2014. R: a Language and Environment for Statistical Computing.Raes, J., Bork, P., 2008. Molecular eco-systems biology: towards an understanding of

community function. Nature Reviews Microbiology 6, 693e699. http://dx.doi.org/10.1038/nrmicro1935.

Rajapaksha, R.M.C.P., Tobor-Kaplon, M.A., Bååth, E., 2004. Metal Toxicity AffectsFungal and Bacterial Activities in Soil Differently 70, 2966e2973. http://dx.doi.org/10.1128/AEM.70.5.2966.

Rath, K.M., Maheshwari, A., Bengtson, P., Rousk, J., 2016. Comparative toxicity ofsalts to microbial processes in soil. Applied and Environmental Microbiology 82,2012e2020. http://dx.doi.org/10.1128/AEM.04052-15. AEM.04052-15.

Rath, K.M., Rousk, J., 2015. Salt effects on the soil microbial decomposer communityand their role in organic carbon cycling: a review. Soil Biology and Biochemistry81, 108e123. http://dx.doi.org/10.1016/j.soilbio.2014.11.001.

Reid, C., Watmough, S.A., 2014. Evaluating the Effects of Liming and Wood-ashTreatment on Forest Ecosystems Through Systematic Meta-analysis 885,867e885.

Rousk, J., Bååth, E., 2011. Growth of saprotrophic fungi and bacteria in soil. FEMSMicrobiology Ecology 78, 17e30. http://dx.doi.org/10.1111/j.1574-6941.2011.01106.x.

Rousk, J., Bååth, E., Brookes, P.C., Lauber, C.L., Lozupone, C.A., Caporaso, J.G.,Knight, R., Fierer, N., 2010a. Soil bacterial and fungal communities across a pHgradient in an arable soil. Isme Journal 4, 1340e1351. http://dx.doi.org/10.1038/ismej.2010.58.

Rousk, J., Brookes, P.C., Bååth, E., 2010b. Investigating the mechanisms for theopposing pH relationships of fungal and bacterial growth in soil. Soil Biologyand Biochemistry 42, 926e934. http://dx.doi.org/10.1016/j.soilbio.2010.02.009.

Rousk, J., Brookes, P.C., Bååth, E., 2009. Contrasting soil pH effects on fungal andbacterial growth suggest functional redundancy in carbon mineralization.Applied and Environmental Microbiology 75, 1589e1596.

Rousk, J., Demoling, L.A., Bahr, A., Bååth, E., 2008. Examining the fungal and bac-terial niche overlap using selective inhibitors in soil. FEMS Microbiology Ecol-ogy 63, 350e358. http://dx.doi.org/10.1111/j.1574-6941.2008.00440.x.

Schmidt, M.W.I., Torn, M.S., Abiven, S., Dittmar, T., Guggenberger, G., Janssens, I. a.,Kleber, M., K€ogel-Knabner, I., Lehmann, J., Manning, D.a.C., Nannipieri, P.,Rasse, D.P., Weiner, S., Trumbore, S.E., 2011. Persistence of soil organic matter asan ecosystem property. Nature 478, 49e56. http://dx.doi.org/10.1038/nature10386.

Southwood, T.R.E., 1977. Habitat, the templet for ecological strategies? Journal ofAnimal Ecology 46, 337e365.

Webb, C.T., Hoeting, J.A., Ames, G.M., Pyne, M.I., LeRoy Poff, N., 2010. A structuredand dynamic framework to advance traits-based theory and prediction inecology. Ecology Letters 13, 267e283. http://dx.doi.org/10.1111/j.1461-0248.2010.01444.x.

Witter, E., Gong, P., Bååth, E., Marstorp, H., 2000. A study of the structure and metaltolerance of the soil microbial community six years after cessation of sewagesludge applications. Environmental Toxicology and Chemistry 19, 1983e1991.http://dx.doi.org/10.1002/etc.5620190805.

Further reading

Evans, S.E., Wallenstein, M.D., 2012. Soil microbial community response to dryingand rewetting stress: does historical precipitation regime matter? Biogeo-chemistry 109, 101e116. http://dx.doi.org/10.1007/s10533-011-9638-3.

Schimel, J., Balser, T.C., Wallenstein, M., 2007. Microbial stress-response physiologyand its implications for ecosystem function. Ecology 88, 1386e1394. http://dx.doi.org/10.1890/06-0219.