lable at ScienceDirect

Soil Biology & Biochemistry 80 (2015) 251e259

Contents lists avai

Soil Biology & Biochemistry

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

Long-term litter decay in Canadian forests and the influence of soilmicrobial community and soil chemistry

C.E. Smyth*, D. Macey, J.A. TrofymowNatural Resources Canada, Canadian Forest Service, 506 West Burnside Road, Victoria, BC V8Z 1M5, Canada

a r t i c l e i n f o

Article history:Received 13 June 2014Received in revised form4 September 2014Accepted 28 September 2014Available online 27 October 2014

Keywords:Forest litter decayPLFASoil chemistryCarbon dynamicsCanadian Intersite DecompositionExperiment

* Corresponding author. Tel.: þ1 250 298 2313.E-mail address: carolyn.smyth@nrcan.gc.ca (C.E. S

http://dx.doi.org/10.1016/j.soilbio.2014.09.0270038-0717/Crown Copyright © 2014 Published by Els

a b s t r a c t

Long-term rates of litter decay have been shown to be primarily influenced by temperature, moisture andlitter quality. However, while decomposition is a biological process, the relative importance of microbialcommunities and other soil chemistry factors is not well understood. Our analysis examined long-termlitter decay parameters, microbial community composition via phospholipid fatty acid (PLFA) analysis,and soil organic horizon chemistry at 14 upland forested sites. Data were collected as part of the Ca-nadian Intersite Decomposition Experiment (CIDET), a 12-year national litter decomposition experiment.Residual errors from a two-pool exponential decay model with decay rates modified by mean annual airtemperature and moisture stress were compared to PLFA marker groups and chemistry variables. Re-sidual errors were not well explained by soil PLFA marker group abundance or concentration, soil pH, norsoil C:N ratios. The best predictor of residual error was soil carbon percent (%C), with higher %C asso-ciated with slower than predicted decomposition.

Crown Copyright © 2014 Published by Elsevier Ltd. All rights reserved.

1. Introduction

The terrestrial biosphere holds large carbon (C) pools, and thedecomposition of plant detritus and soil organic matter releasesmore carbon dioxide (CO2) to the atmosphere annually, than fossilfuel and industrial sources (IPCC, 2007). Despite the critical role ofdecomposition in the global C balance, our understanding ofdecomposition is rudimentary in comparison to our current un-derstanding of C inputs through primary production (Adair et al.,2008). Decomposition is a key ecosystem process that influencesthe recycling of nutrients and thus ecosystem fertility and it is alsothe main return pathway to the atmosphere of CO2 fixed duringphotosynthesis. Decomposition is mediated by microbes that useplant primary production from above- and belowground litter andsoil as their sources of C (Brant et al., 2006). During decomposition,the microbial community controls the partitioning of litter- anddead root-C between CO2 via respiration and storage in semi-permanent soil-C pools (Moore-Kucera and Dick, 2008; Prescott,2010). The size and composition of the soil microbial communityhas been related to complex interactions with tree species (Hobbieet al., 2012), climate (Fierer et al., 2009), disturbances (Brant et al.,2006; Moore-Kucera and Dick, 2008), soil nutrients (Leckie, 2005;

myth).

evier Ltd. All rights reserved.

Lauber et al., 2008), soil chemistry (Nilsson et al., 2005; H€ogberget al., 2007), and net primary production (Brant et al., 2006).

Studies comparing the influences of several edaphic variables onmicrobial community structure and composition have identifieddifferent dominant drivers of the microbial community. H€ogberget al. (2007) found that soil chemistry had a larger influence thantree species on the soil microbial community, but Mitchell et al.(2010) found that plant community composition better predictedchanges in microbial community composition than soil properties.You et al. (2014) found that soil water, soil organic C, soil temper-ature, soil clay content, fine root mass, and soil C to N ratio were allsignificant drivers of variations in soil microbial communitystructure.

Although past research has shown that temperature, precipita-tion, and litter chemistry strongly control rates of litter decompo-sition (e.g. Aber et al., 1990; Aerts, 1997), how these factorsindirectly or interactively influence the soil microbial community,and hence litter decay, across large spatial and temporal scalesremains unclear. Decomposition studies are most often local orregional in scale, use a low diversity of litter types, and do notconsider the microbial community. Extrapolating to continentalscales, or non-represented litters, or different ecosystems istherefore problematic. Similarly, because most studies are con-ducted for less than 5 years, there are few data available to definewhat factors control long-term or late-phase decomposition(Trofymow et al., 2002; Hobbie et al., 2012).

Table 1CIDET sites, locations, climate variables and soil organic horizon %C. Climate vari-ables are mean annual air temperature (T), water stress (W) total annual precipi-tation (P).

Site Location Latitude(N)

Longitude(W)

T (�C) W P(mm)

Soil %C

INU Inuvik, NT 68� 190 133� 320 �7.64 0.60 237 37.5SCH Schefferville, QC 54� 520 66� 390 �4.16 0.90 830 16.5GI1 Gillam, MB 56� 190 94� 510 �3.77 0.69 482 32.0NH1 Nelson House, MB 55� 550 98� 370 �2.88 0.70 471 22.2WHI Whitehorse, YT 60� 510 135� 120 0.01 0.47 241 24.6TOP Topley, BC 54� 360 126� 180 1.50 0.66 557 35.0KAN Kananaskis, AB 51� 000 115� 000 3.62 0.65 625 34.0TER Termundee, SK 51� 500 104� 550 3.68 0.43 370 17.0GAN Gander, NL 48� 550 54� 340 4.18 0.87 1265 52.5CBR CB Rocky Harbour,

NL49� 320 57� 500 4.49 0.85 1258 43.0

HID Hidden Lake, BC 50� 330 118� 500 6.55 0.67 717 49.6MAR Morgan Arboretum,

QC45� 250 73� 570 6.70 0.76 978 34.2

PMC Port McNeill, BC 50� 360 127� 200 8.72 0.82 1912 53.2SHL Shawnigan Lake, BC 48� 380 123� 420 9.33 0.64 1266 19.2

C.E. Smyth et al. / Soil Biology & Biochemistry 80 (2015) 251e259252

The Canadian Intersite Decomposition Experiment (CIDET) wasa national litterbag experiment that examined decomposition over12 years for a range of litter types and ecosystems. Data from thisexperiment has been used for quantitative assessment of modelparameters in three dynamic C models (Palosuo et al., 2005; Zhanget al., 2007; Smyth et al., 2010), for testing of several combined Cand nitrogen models (Manzoni et al., 2008; Zhang et al., 2008;Manzoni and Porporato, 2009) and for studying the effects ofclimate and litter quality on decomposition processes (Trofymowet al., 2002; Moore et al., 2006).

In this paper we present results from the 12-year C-remainingtime series and examine the relationships between decomposition,climate, soil organic horizon chemistry and microbial community.We used phospholipid fatty acid (PLFA) analysis to measure soilmicrobial biomass and community composition. The PLFAsextracted from soils represent living microorganisms and indicatorPLFA are used as markers for taxonomic groupings (Zelles, 1999;Bååth and Anderson, 2003). This method is quantitative, andmultivariate statistical procedures can be used to determine sig-nificant differences in abundance and composition of soil microbialcommunities (Frostegård et al., 1991, 1993). PLFA analyses areeconomical and allow relatively high sample throughput comparedto nucleic acid-based methods, which is an advantage for largescale field based studies (Moore-Kucera and Dick, 2008). Thismethod has been shown to be proportional to other microbialbiomass measures for forest soils (Fritze et al., 2000; Fierer et al.,2003; Leckie et al., 2004).

We predicted C-remaining time series from a two-pool modelwith decay rates modified by temperature and water stress andcompared it to measured C-remaining time series (Smyth et al.,2011). Then we compared the residuals to estimates of PLFAmarker groups to understand the influence of bacterial and fungalgroups on residual errors. Our hypothesis was that residual errorsin the litter decomposition, which reflect decomposition that isfaster or slower than predicted, are significantly related to soil PLFAmarker group abundance or concentration. We also assessed therelationships between soil PLFA marker groups and soil chemistryvariables to determine which variables were significant drivers ofvariations in community composition.

2. Materials and methods

2.1. The CIDET study and sample selection

The CIDET study is a 12-year litter decomposition experiment inwhich roughly 11 000 litterbags were surface-placed at 21 sites (18upland, three wetland) that represented the major forested eco-climatic provinces of Canada (Ecoregions Working Group, 1989).Litterbags were 20� 20 cm constructed from polyproplylene meshwith 0.5 mm openings, each containing 10 g (dry weight) of one of11 different standard litter types. All litterbags were placed incontact with litter layers just before or during litterfall in autumn1992 on four replicate plots on each site. Litterbags were collectedannually each autumn until 2000, and biennially for the last twocollections (2002, 2004). Further details of site descriptions, plotlayouts, details of litter collection, sample processing, and initiallitter and soil chemistry were published previously (Trofymow andCIDET Working Group, 1998; Trofymow et al., 2002). For thisanalysis we included initial and exposed samples of eight treefoliar litters (trembling aspen: Populus tremuloides, Americanbeech: Fagus grandifolia, Douglas-fir: Pseudotsuga menziesii, whitebirch: Betula papyrifera, jack pine: Pinus banksiana, black spruce:Picea mariana, tamarack: Larix laricina, western redcedar: Thujaplicata) from 14 upland forested sites for which we had PLFA data(Table 1). Two of the sites (MAR, TER) had predominantly

deciduous stands, while the remaining sites had predominantlyconiferous stands.

2.2. Climate indicators

Climate data were from nearby Meteorological Services Canada(MSC) climate stations (http://www.msc-smc.ec.gc.ca) and ANU-CLIM interpolated climate data (McKenney et al., 2001). Climateindicators were mean annual temperature (T), total annual pre-cipitation (P) and water stress (W), which is defined in thedecomposition model section.

2.3. Litter measurements

As described previously (Trofymow et al., 1995; Trofymow andCIDET Working Group, 1998), litters were removed from litter-bags after each fall collection, oven-dried at 55 �C, weighed, andground to 0.2 mm mesh in a Wiley mill. Weighted compositesamples were prepared from four replicate litter types for each siteand analyzed for total C by dry combustion on a LECO CR-12analyzer or a LECO CNS2000 Combustion Analyzer (LECO Corpo-ration, St. Joseph, MI).

Carbon remaining at time t was estimated for all litters as:

Cr�t� ¼ 100

CcðtÞMðtÞCcð0ÞMð0Þ (1)

where Cc is the carbon concentration (%), M is the mass of thesample, Cc(0) is the initial carbon concentration, and M(0) is theinitial mass.

2.4. Soil measurements

2.4.1. Soil organic horizon chemistryIn 2004, samples of the surface soil organic horizon were

collected immediately adjacent to the string of litterbags in three tofour replicate plots at each site. After removing the litter layer andgreen material, a 10 � 10 cm sample was excavated to the mineralhorizon and composited into a single sample bag. The field moistsoils were kept at 2 �C and shipped to the laboratory within 5 days,where they were sieved to 8 mm, gently homogenized, andportioned into subsamples for analysis.

Soil pH was measured with Ag/AgCl pH electrode using 1:2 ratioof soil to calcium chloride solution (0.01 M) and settling time of

C.E. Smyth et al. / Soil Biology & Biochemistry 80 (2015) 251e259 253

30 min (Kalra and Maynard, 1991). Moisture content was measuredby weighing soil samples before and after oven-drying at 105 �Cand the loss in weight expressed as a percentage of the oven-driedweight. Extractable inorganic N was determined from the fieldfresh soil with 2.0 M KCl (soil-solution ratio 1:10 w/v) and gravityfiltered extracts were analyzed for ammonium (NH4eN) and nitrate(NO3eN) colorimetrically by segmented flow analysis (Technicon®

AutoAnalyzer II, Technicon Corp). Total P was determined throughwet oxidation of air-dried samples in a block digester (Parkinsonand Allen, 1975) and P the digest measured using the Auto-Analyzer II. Results for NH4eN, NO3eN and total P were expressedon a mg per gram dry-soil basis.

Subsamples of 70 �C oven-dried soil were ground to 0.2 mmusing a Wiley mill for measurement of total C, N, and S. Total C, Nand Swere determined by dry combustion on CNS2000 (LECO Corp,St Joseph, Michigan). All data were expressed on percent dry soilbasis.

2.4.2. Phospholipid fatty acid analysisLipids were extracted from subsamples of lyophilized soil using

a single phase mixture of methanol, chloroform, and 0.15 M (pH4.0) citrate buffer (1:1:0.9 v/v/v) following the method of Bligh andDyer (1959) adapted for forest soils (White et al., 1979; Frostegårdet al., 1991). Lipids were fractionated on silicic acid SPE columns(6 ml 500 mg�1) SampliQ silica, Agilent Technologies, Santa ClaraCA, USA) by sequential elution with chloroform, acetone andmethanol. Phospholipids collected in the methanol fraction werederivatized to corresponding methyl esters by mild alkalinemethanonlysis (White et al., 1979; White and Ringelberg, 1998)and re-suspended in hexane containing known amount of 19:0Me(methyl nonadecanoate, SigmaeAldrich) as internal standard.PLFAs were identified and quantified on an HP 5890A Series II gaschromatograph fitted with an HP-Ultra 2 column (25 m, 0.33 mm5%PHeMeSilicone, J&W Scientific) and FID detector (Agilent Tech-nologies Hewlett Packard Co, Palo Alto CA, USA) using MIDISherlock Version 4.5 software (MIDI, Inc., Newark, DE, USA). Peakswere identified by matching retention times against a knownstandard solution and abundance of individual PLFA reported as aproportion (mol %) of all identified PLFAs. The PLFA nomenclaturefollows standard formatting as described by Frostegård et al.(1993).

PLFAs between 14 and 20 carbon units in length and greaterthan 1% abundance (mol %) were used in the analysis and abun-dance was re-scaled to 100% based on these PLFAs. Molar concen-tration of individual PLFAs were calculated based on the internalstandard and expressed as nmol g�1 dw soil. Total microbialbiomass for each sample was calculated by summing molar con-centrations of all reported PLFAs. Indictor PLFAs were used asmarkers to classify soil microbial community into broad taxonomicgroups. Total bacteria (B) were represented by summing i14:0, 15:0,i15:0, a15:0, i16:0, 16:1 u7c, 17:0, i17:0, a17:0, cy17:0, 18:1 u7c, andcy19:0 (Frostegård and Bååth, 1996; Bååth and Anderson, 2003;Steger et al., 2007); Gram-positive bacteria (GMþ) were repre-sented by summing i14:0, i15:0, a15:0, i16:0, a16:0, i17:0, and a17:0(Frostegård and Bååth, 1996; Bååth and Anderson, 2003); Gram-negative bacteria (GM�) were represented by summing 16:1 u 7c,cy17:0, 18:1 u7c, and cy19:0 (Frostegård et al., 1993; Frostegård andBååth, 1996; Bååth and Anderson, 2003); actinobaceteria (A) wererepresented by summing 10Me16:0, 10Me17:0 and 10Me18:0(Steger et al., 2007); and fungi (F) were represented by 18:2 u6,9.

2.5. Decomposition model

It is assumed that as decomposition proceeded, the litterbagscontained C from the original un-decayed litter as well as semi-

stable or humified organic C. Decay was modeled using power-series decay in a two-pool model based on the Carbon BudgetModel of the Canadian Forest Sector (CBM-CFS3) where decay rateswere modified by temperature (Kurz et al., 2009) and water stress.The relationships between decay rates, temperature and waterstress have been investigated earlier (Smyth et al., 2010, 2011), andthe derived relationships are used here.

C-remaining in the litterbags was averaged over the eight littertypes, and then modeled as:

C�t� ¼ 100

"ð1� kÞt þ kt

Xti¼1

Cði� 1Þð1� ksÞt�iþ1

#(2)

where C is the carbon stock [%], t is the annual time step, k is thelitter decay rate (% yr�1), t is the proportion transferred to the semi-stable (slow) C pool, and ks is the decay rate of that pool (% yr�1)(Kurz et al., 2009). In Eq. (2), the first term represents the un-decayed litter C percent remaining, and the second term repre-sents the semi-stable C.

The litter decay rate is estimated as base decay rate modified bytemperature and water stress:

k ¼ kbQ10

�T�10T

�W (3)

where kb is the base decay rate, T is the mean annual air temper-ature in �C, Q10 is the Quotient 10 function with a reference tem-perature of 10 �C, and W is the water stress modifier. Parametervalues for Eq. (3) were taken from Smyth et al. (2011): kb was 55%yr�1, the Q10 value was 2.95, and t (the transfer of decaying litter Cto semi-stable C) was 18.5% with the remainder released to theatmosphere. For the slowly decaying semi-stable C pool, the basedecay rate was 1.5% yr�1, and the Q10 value was 2.95. The waterstress modifier was estimated as:

Wi ¼ min�1;

0:52PiPETi

�(4)

where Pi is the monthly precipitation (mm), PETi is the monthlypotential evapotranspiration (mm), and Wi was set to 1 if PETi ¼ 0.An annual value of the water stress was estimated from the averageof monthly values (Smyth et al., 2011). PET, the potential loss ofwater to the atmosphere through evaporation and transpiration,was defined using the method of Hamon (1963):

PETi ¼ 0:1651Ldrsnd (5)

rs ¼ 216:7Es

Ti þ 273:3(6)

Es ¼ 6:108 exp�17:29693TiTi þ 273:3

�(7)

where PETi is the monthly potential evapotranspiration (mm); Ti isthe monthly mean air temperature (�C); Ld is the daylight length inmultiples of 12 h; nd is the number of days in the month; rs is thesaturated vapor density (g m�3) (Federer and Lash, 1983); and Es isthe saturated vapor pressure (mb) (Murray, 1967).

The residual error, E, was defined as the predicted C-remainingminus the measured values. Errors were averaged over all littertypes at six years, E(6), and 12 years, E(12).

Fig. 1. PLFA markers in the soil organic horizon. PLFA amounts were scaled by taking the natural logarithm of the nmol g�1 dry weight of soil. Bacterial (B) and Fungal (F) markersare indicated on the right-hand side. Site mean annual air temperature increases to the right. Three letter site codes are described in Table 1.

C.E. Smyth et al. / Soil Biology & Biochemistry 80 (2015) 251e259254

2.6. Statistical analysis

2.6.1. Regression analysisFour dependant decomposition variables were selected for

regression analysis (i) measured C-remaining after 12 years, Cr(12);(ii) the site-specific decay rate, k; (iii) the residual error betweenpredicted andmeasured C remaining after 6 years, E(6); and (iv) theresidual error after 12 years, E(12). Dependent variables wereregressed against 20 explanatory variables which included totalPLFA, 6 PLFA marker groups (B, GMþ, GM�, A, F and B:F ratio), threeclimate variables (T, P andW) and 10 soil chemistry variables (pH, %moisture, NH4eN, NO3eN, total P, %C, %N, %S, C:N ratio, and C:Sratio). One- and two-variable regressions (proc reg, SAS, 2008)were performed for all possible combinations, with results onlyreported for the highest squared correlation coefficient (r2) withreliable predictors (p-values less than 0.1), and uncorrelated

explanatory variables. Collinearity was assumed if the varianceinflation factor was greater than 10. The hypothesis tested if totalPLFA or PLFA marker groups were related to the four dependentdecomposition variables. The null hypothesis, that total PLFA andPLFA marker groups were unrelated to decomposition variables,was rejected if p-values were less than 0.05. Regression residualswere tested for normality using the ShapiroWilk test for normality(proc univariate, SAS, 2008). The hypothesis that the regressionresiduals were normally distributed was rejected for p < 0.05.

2.6.2. Factor analysisA factor analysis was performed on the residual error after 12

years using E(12), PLFA variables (Total, Bacterial and Fungal) andtwo soil chemistry variables (pH and C:N ratio). Variables werestandardized to unit variance (SAS proc factor, principal componentmethod), and only eigenvalues greater than one were retained.

C.E. Smyth et al. / Soil Biology & Biochemistry 80 (2015) 251e259 255

Solutions were rotated using a varimax rotation and variables withcomponent loadings greater than 0.4 were examined.

2.6.3. Significance testingIndependent group t-tests (proc ttest, SAS, 2008) were used to

compare means of PLFA variables for high and low categories of pHand C:N ratios (H€ogberg et al., 2007). p-values from the two-tailedprobability computed using the t-distributionwere compared to analpha level of 0.05. Cases where the variances were equal used thePooled method, and cases where the variances were unequal usedthe Satterthwaite method (Satterthwaite, 1946).

3. Results

Named PLFAs accounted for more than 90% of the total numberof peaks, and more than 99 mol % of reported PLFAs. Fig. 1 showsthat the PLFA individual marker concentration was remarkablyconsistent across sites, even though the mean annual air temper-ature varied from �7.6 �C to 9.3 �C. PLFA abundance (not shown)had a similar distribution across the sites.

The sites had an average of 34 PLFAs, with a range of 27e39markers. Overall, the richness, defined as the number of identifiedPLFAs present, did not relate to the abundance or concentration ofPLFA (Fig. 2). One of the sites, SHL, had particularly low bacterialconcentration (Fig. 2a) and total PLFA (Fig. 2c), although its richness

Fig. 2. PLFA marker group for (a) bacteria, (b) fungi, (c) ratio of bacteria to fungi, (d) total PLsites with the largest residual errors. Average concentrations in nmol g�1 dw soil were 113average richness was 34. Average abundance was 37% for bacteria, and 8% for fungi.

was higher than the average richness (Fig. 2e). The site with thelowest richness, GAN, did not have the lowest amount total PLFA,but did have relatively little bacterial PLFA. Several PLFAs that werenoticeably absent at either GAN or SHL included i17:0, i15:1, 18:1w5c and 11Me18:1w7c.

PLFAs that contributed the most to the bacterial PLFA concen-tration were 18:1 w7c, i15:0, 10Me16:0 and cy 19:0. These markerswere highest for TOP, and lowest at SHL. Similarly, the total PLFAand bacterial concentration were highest at TOP and lowest at SHL(Fig. 2), although bacterial abundances for these sites were notextreme.

Total PLFA was on average 312 nmol g�1 dw soil, and rangedfrom 171 to 422 nmol g�1 dw soil. Bacterial PLFA concentration andabundance were larger than for the fungal marker (Fig. 2). Thebacterial group had an average abundance of 37% and an averageconcentration of 113 nmol g�1 dw soil. In contrast, the fungal grouphad an average of 8% abundance and an average concentration of26 nmol g�1 dw soil. The ratio of bacterial to fungal PLFA variedfrom 1.7 to 10.5, with an average of 5.6.

Regression analysis of the total PLFA and PLFA marker groupsagainst the 13 explanatory variables (3 climate indicators and 10soil chemistry indicators) found that none of the marker groupswere correlated with climate variables. However, bacterial PLFAconcentrationwas significantly correlatedwith soil organic horizon%S, with r2 values of 0.39, 0.41 and 0.64 for B, GMþ and A,

FA abundance, (e) richness, defined as the number of PLFA peaks. Shaded bars indicatefor bacteria, 26 for fungi and 312 for the total. The average B:F ratio was 5.6, and the

Table 2CIDET site names, litter decay rates, C-remaining after 12 years, and residual errorsin C-remaining after 6 and 12 years.

Site k (% yr�1) Cr(12) (%) E(6) (%) E(12) (%)

INU 4.9 67.1 �7.2 �4.0SCH 10.7 32.4 5.9 6.8GI1 8.5 52.8 0.4 �6.6NH1 9.5 29.5 1.9 13.2WHI 8.7 47.4 �12.3 �1.9TOP 14.9 25.3 8.5 4.4KAN 17.9 25.4 �1.8 0.1TER 11.8 36.9 7.3 �0.8GAN 25.5 25.9 �5.1 �6.2CBR 25.8 22.6 �1.9 �3.0HID 25.4 20.6 0.1 �0.8MAR 29.1 20.1 3.8 �1.6PMC 39.3 21.9 �10.5 �5.2SHL 33.0 17.8 �3.3 �0.2

C.E. Smyth et al. / Soil Biology & Biochemistry 80 (2015) 251e259256

respectively (Table S2). Weaker relationships were also found for %N and %C soil chemistry variables. In addition, total PLFA wassignificantly correlated with %C and had an r2of 0.4. Litter decayrates for the sites ranged from 4.9% yr�1 at the coldest site to 39.3%yr�1 at one of the warmest sites (Table 2). The average litter C-remaining after 12 years was strongly correlated with climatevariables. The coldest site was in the intermediate stage of decaywith 67.1% C-remaining, and the warmer sites were in the maturestage of decay with roughly 20% C-remaining.

The top three sites where litter decayed faster than predicted bythe two-pool model (E(12)>0) were SCH, NH1 and TOP (Table 2).The top three sites where litter decayed slower than predicted bythe two-pool model (E(12)<0) were GI1, GAN and PMC.

Regression of the litter decomposition variables against the 20explanatory variables found that both C-remaining after 12 yearsand site-specific decay rates were strongly linked (r2 > 0.7) toclimate variables (Table 3, Table S1). It was anticipated that thelitter decay rate would be correlated with climate variables becauseit was calculated from a base decay rate modified by a temperaturequotient (Eq. (3)). The best two-variable model for the litter decayrate, when climate variables were excluded, was a weak correlation(r2 ¼ 0.46) with the soil organic horizon %C (significant) and fungalPLFA concentration (not significant) (Table 3).

The residual error between C-remaining measured and pre-dicted time series was correlated with soil organic horizon %C forthe best single-variable model, and %C and total PLFA for the besttwo-variable model (Fig. 3), although the correlation with totalPLFA was not significant. Normality testing of the regression

Table 3Results of regressing litter decomposition dependant variables against independent clim

Decay indicator Variable 1 Slope p-level V

Cr(12) T �2.17 <0.001 Wk T 0.011 <0.001 Pk* %C 0.0059 <0.05 Fk* %C 0.0043 <0.1E(6) %S �95.7 <0.1 AE(12) %C �0.394 <0.01 TPE(12) %C �0.236 <0.05 e

T: Mean annual air temperature.W: Water stress.P: Total annual precipitation.*: Regression without climate variables.%C: Soil organic horizon C percentage.%S: Soil organic horizon Sulfur percentage.TPLFA: Total PLFA concentration (nmol g�1 dry soil).A: Actinobacteria PLFA concentration (nmol g�1 dry soil).F: Fungal PLFA concentration (nmol g�1 dry soil).

residuals did not reject normality assumptions. In general, soilchemistry variables were better predictors than PLFA group abun-dances or concentration for residual errors.

Despite the non-significant relationships between the residualerror in the litter decay and the microbial community revealed inthe regressions, there were some connections between sites atwhich litter decayed faster than predicted by the two-pool decaymodel, and high-ranking PLFA markers. Litter decayed faster thanpredicted at Topley (TOP), which had highest bacterial concentra-tion, and at NelsonHouse (NH1), which had the third highest fungalconcentration. Decay was slower than predicted at Gander (GAN),which had the lowest richness. Litter decay at PMC was also slowerthan predicted, but this site did not have low-richness or low PLFAconcentration. Both of these sites with lower than expected decayhad high soil organic horizon %C, which suggests inhibiteddecomposition.

A factor analysis of the 12-year residual error (E(12)) with PLFAconcentrations (total, bacterial, fungal) and two soil chemistryvariables (pH and C:N ratio) found three eigenvalues that explained48% of the variance. The first eigenvalue had high componentloadings on the three PLFA markers (Fig. 4). The second had highcomponent loadings on the pH and C:N ratio, and the thirdeigenvalue was related to the 12-year residual error. The factoranalysis did not connect marker groups with either soil chemistryor residual errors. The first component separated sites with highPLFA concentration (TOP) from regions with low PLFA concentra-tion (SHL, SCH). The second component separated sites with highC:N ratios and lowpH (e.g. GAN) from thosewith lowC:N ratios andhigh pH (e.g. TER and GI1). The third component separated regionswith high residual error.

4. Discussion

The total PLFA concentration for these sites averaged312 nmol g�1 dw soil, and was lower than in previous studies.Hannam et al. (2006) and H€ogberg et al. (2007) found total PLFAconcentration in excess of 1 mmol g�1 dw soil within F and H soillayers. However the present estimates of PLFA concentration weresampled near the end of the season, and total microbial biomasshas been found to higher at the beginning of the season and lowerat the end of the season (Wallander et al., 2001; Brant et al., 2006).Present estimates of total microbial biomass were higher thanthose reported by in previous studies by Drenovsky et al. (2010) andBrant et al. (2006) who found total PLFA concentration of20e40 nmol g�1 dw soil. However, both of these studies sampleddeeper soil depths (up to 15 cm), and total microbial biomass has

ate, PLFA marker group concentration, and soil chemistry variables.

ariable 2 Slope Intercept p-level r2

�35.1 60.9 <0.05 0.780.0001 0.0698 <0.001 0.97

�0.0032 0.0757 <0.1 0.460.0433 0.27

1.11 �10.9 <0.1 0.32LFA 0.0452 �1.25 <0.1 0.52

e 7.52 0.31

Fig. 3. Error between predicted litter decay and measurements at 12 years versus (a) %C in the soil organic horizon and (b) total PLFA concentration. Shaded areas indicatesites with the largest residual errors. p-values were 0.006 for %C and 0.053 for totalPLFA concentration, with an overall r2 of 0.52.

Fig. 4. Factor analysis of the PLFA marker groups, soil chemistry pH and C:N ratioindicators and residual error in litter C-remaining after 12 years. (a) scores for the firsttwo principal components, and (b) scores for principal components 2 and 3. Shadedareas indicate sites with the largest residual errors.

C.E. Smyth et al. / Soil Biology & Biochemistry 80 (2015) 251e259 257

been found to decrease with depth (Leckie et al., 2004). As noted byLeckie (2005), PLFA concentrations are difficult to compare becausethe variability in methods and the nature of the soil matrix.

The richness, as expressed by the average number of namedPLFAmarkers, averaged 34 in the present study, and ranged from27to 39. An earlier study by Staddon et al. (1998) found the mineralsoil microbial functional diversity decreased with increasing lati-tude and suggested diversity may be lower due to nutrient limi-tation and higher acidity. In this study, the number of markers wasnot correlated with any soil chemistry or climate variables(Table S2); however it was weakly correlated with the actino-bacterial PLFA concentration.

The average proportion of bacterial markers (37%) and fungalmarkers (8%) for the present study was similar to those found in aprevious study by Leckie et al. (2004). In addition, the bacterial tofungal ratios for the present study (ranging from 1.7 to 10.5, with anaverage of 5.6) were similar to those observed by Leckie et al.(2004) who found ratios that ranged from 3.3 to 14.3 in theirstudy and from 1.7 to 8.3 in published studies. Low B:F ratios havebeen linked to systems with a higher proportion of fungal de-composers due to recalcitrant litter (Lauber et al., 2008), nutrient-poor sites (de Vries and Shade, 2013), low pH (Bååth andAnderson, 2003) and water-logged soils (Drenovsky et al., 2010).The present study found no significant relationships between theB:F ratios and soil organic horizon chemistry variables.

Previous studies in a diverse range of ecosystems (Staddon et al.,1998; Fierer et al., 2009; Drenovsky et al., 2010) have found

relationships between the microbial community structure andclimate. We did not observe such a relationship in our study,perhaps because we examined only boreal and cool temperateforest systems, a less diverse range of ecosystems than the previousstudies.

Many studies have reported that pH is a good predictor of themicrobial community structure (Nilsson et al., 2005; Lauber et al.,2008; Fierer et al., 2009; Frostegård et al., 2011) at landscape andcontinental scales. For boreal forest soils, H€ogberg et al. (2007)found that soil pH and C:N ratio was a good predictor of micro-bial community structure. In their study of three forests in Swe-den, fungal biomass was found to be high in acid soil (pH < 4.1)with a high C:N ratio (>38). In this study, the fungal concentrationwas not significantly different (p ¼ 0.37) for high acid soils(pH < 4.1) with a high C to N ratio (>38). Nor were there anysignificant differences in PLFA group concentrations (or abun-dances) tested separately for soil pH levels (threshold 4.1) or soil Cto N ratio levels (threshold 38). H€ogberg et al. (2007) also foundthat the soil C:N ratio and pH were good predictors of microbialcommunity structure in a PCA, but our study found that neithersoil C:N ratio nor pH were good predictors of microbial commu-nity structure. Factor analysis of PLFA groups and pH as well asC:N ratio did not find any clustering of PLFA and soil chemistryvariables.

In the present study, there were no significant relationshipsfrom the regression analysis to support the hypothesis thatdecomposition parameters were related to PLFA markers, althoughthere were some weak correlations. Other studies have also iden-tified disconnections between litter and soil processes even thoughthey are considered part of the same continuum. For example, Ballet al. (2014) found a correlation between community size and mass

C.E. Smyth et al. / Soil Biology & Biochemistry 80 (2015) 251e259258

and nitrogen loss in the litter but not in the soil, suggesting theresponses for the litter and soil are not the same.

The lack of connection between the soil microbial communityand the litter decomposition suggests that the litter decay microbeswere different from the soil community, perhaps because soil mi-crobes are preferentially selected for different litter types (Norriset al., 2013), and/or because the microbial community is affectedby soil fauna. Additional data on PLFA marker groups or DNAmarker groups within the litterbags over time may also provideinsights into litter decomposition dynamics. Such data are notavailable for this study, however represents potential objectives forfuture long-term decomposition studies.

Criticisms of litterbags include that they may impede coloni-zation of litter by soil microbes and fauna (Wise and Schaefer, 1994;Kampichler and Bruckner, 2009) and modify litter microenviron-ment (Lousier and Parkinson, 1976; Bradford et al., 2002) whencompared to unconstrained litter. While soil biota colonization oflitter in litterbags may be slower than on free litter over the firstfew months, over the long time period of this study there wouldhave been ample opportunity for colonization as the litterbagsbecame covered with native litterfall and fungal hyphal growth,water flow, and phoretic transport by soil fauna would havetransferred microbes to litter inside the bag. The mesh size of thelitterbags used does not prevent colonization by collembola andmites (Set€al€a et al., 1996); the dominate soil faunal groups in borealand cool temperate forests.

Another possible reason for the lack of connection between thePLFA markers and the decomposition variables may be related tothe temporal mismatch between the litter variables, which arebased on a long-term litter decomposition time series, and asnapshot of the soil PLFA and soil chemistry variables. We assumedthat the soil PLFA markers and chemistry variables were repre-sentative of long-term site conditions, but we don't have anyadditional data to test this assumption.

5. Concluding remarks

We found little connection between 12-year litter decay pa-rameters and soil microbial community structure as estimated byPLFA markers. Although some of the soil chemistry variables didprovide some predictive power in explaining residuals in decaytime series, beyond that explained by climate variables, in a two-pool model of litter decay.

We did not include any initial litter quality variables in thisstudy, although previous analysis has found the ratio of acidunhydrolyzable residue to nitrogen to have some predictive capa-bilities, particularly for more resistant litter types (Trofymow et al.,2002; Smyth et al., 2010).

Variation in soil microbial communities amongst study sites wasrelated to soil chemistry variables, in particular between bacterialconcentration and soil C, N, and S percent. There was not a strongrelationship betweenmicrobial concentration and pH or C:N ratios,which has been found in other studies.

Acknowledgments

Climate station data were provided by Environment Canada,Atmospheric and Environment Services. We acknowledge all of theCIDET Working Group, without whom this analysis would not havebeen possible: D. Anderson (University of Saskatchewan, SK), C.Camir�e (Universit�e de Laval, PQ), L. Duchesne (Canadian ForestService, ON), J. Fyles (McGill University, PQ), L. Kozak (University ofSaskatchewan, SK), M. Kranabetter (Ministry of Forests, BC), T.Moore (McGill University, PQ), I. Morrison (Canadian Forest Service,ON), C. Prescott (University of British Columbia, BC), M. Siltanen

(Canadian Forest Service, AB), S. Smith (Agriculture Canada, BC), B.Titus (Canadian Forest Service, BC), S. Visser (University of Calgary,AB), R. Wein (University of Alberta, AB), D. White (Dep. Indian andNorthern Affairs, YT). Technical support from R. Ferris and A. Harris,PFC Chemical Services Lab are also gratefully acknowledged. Wethank two anonymous reviewers and the editor for their thoughtfuland insightful comments which improved this manuscript. Moreinformation on CIDET including a complete list of publications isavailable at: http://cfs.nrcan.gc.ca/projects/76.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.soilbio.2014.09.027

References

Aber, J.D., Melillo, J.M., McClaugherty, C.A., 1990. Predicting long-term patterns ofmass loss, nitrogen dynamics, and soil organic matter formation from initialfine litter chemistry in temperate forest ecosystems. Canadian Journal of Botany68, 2201e2208.

Adair, E.C., Parton, W.J., Del Grosso, S.J., Silver, W.L., Harmon, M.E., Hall, S.A.,Burke, I.C., Hart, S.C., 2008. Simple three-pool model accurately describes pat-terns of long-term litter decomposition in diverse climates. Global ChangeBiology 14, 2636e2660.

Aerts, R., 1997. Climate, Leaf Litter Chemistry and Leaf Litter Decomposition inTerrestrial Ecosystems: a Triangular Relationship. Blackwell, Oxford, UK.

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.

Ball, B.A., Carrillo, Y., Molina, M., 2014. The influence of litter composition across thelitteresoil interface on mass loss, nitrogen dynamics and the decomposercommunity. Soil Biology and Biochemistry 69, 71e82.

Bligh, E.G., Dyer, W., 1959. A rapid method of total lipid extraction and purification.Canadian Journal of Biochemistry and Physiology 37, 911e917.

Bradford, M.A., Tordoff, G.M., Eggers, T., Jones, T.H., Newington, J.E., 2002. Micro-biota, fauna, and mesh size interactions in litter decomposition. Oikos 99,317e323.

Brant, J., Myrold, D., Sulzman, E., 2006. Root controls on soil microbial communitystructure in forest soils. Oecologia 148, 650e659.

de Vries, F.T., Shade, A., 2013. Controls on soil microbial community stability underclimate change. Frontiers in Microbiology 4, 265.

Drenovsky, R.E., Steenwerth, K.L., Jackson, L.E., Scow, K.M., 2010. Land use and cli-matic factors structure regional patterns in soil microbial communities. GlobalEcology and Biogeography 19, 27e39.

Ecoregions Working Group, 1989. Ecoclimatic Regions of Canada, First Approxi-mation. Environment Canada, Ecoregions Working Group of Canada,Ottawa,Ontario.

Federer, C.A., Lash, D., 1983. BROOK: a Hydrologic Simulation Model for EasternForests. University of New Hampshire, New Hampshire.

Fierer, N., Carney, K.M., Horner-Devine, M.C., Megonigal, J.P., 2009. The biogeog-raphy of ammonia-oxidizing bacterial communities in soil. Microbial Ecology58, 435e445.

Fierer, N., Schimel, J.P., Holden, P.A., 2003. Variations in microbial communitycomposition through two soil depth profiles. Soil Biology and Biochemistry 35,167e176.

Fritze, H., Pietik€ainen, J., Pennanen, T., 2000. Distribution of microbial biomass andphospholipid fatty acids in Podzol profiles under coniferous forest. EuropeanJournal of Soil Science 51, 565e573.

Frostegård, A., Bååth, E., 1996. The use of phospholipid fatty acid analysis to esti-mate bacterial and fungal biomass in soil. Biology and Fertility of Soils 22,59e65.

Frostegård, Å., Bååth, E., Tunlio, A., 1993. Shifts in the structure of soil microbialcommunities in limed forests as revealed by phospholipid fatty acid analysis.Soil Biology and Biochemistry 25, 723e730.

Frostegård, Å., Tunlid, A., Bååth, E., 1991. Microbial biomass measured as total lipidphosphate in soils of different organic content. Journal of MicrobiologicalMethods 14, 151e163.

Frostegård, Å., Tunlid, A., Bååth, E., 2011. Use and misuse of PLFA measurements insoils. Soil Biology and Biochemistry 43, 1621e1625.

Hamon, W.R., 1963. Computation of direct runoff amounts from storm rainfall. In-ternational Association of Hydrological Sciences 63, 52e62.

Hannam, K., Quideau, S., Kishchuk, B., 2006. Forest floor microbial communities inrelation to stand composition and timber harvesting in northern Alberta. SoilBiology and Biochemistry 38, 2565e2575.

Hobbie, S.E., Eddy, W.C., Buyarski, C.R., Adair, E.C., Ogdahl, M.L., Weisenhorn, P.,2012. Response of decomposing litter and its microbial community to multipleforms of nitrogen enrichment. Ecological Monographs 82, 389e405.

C.E. Smyth et al. / Soil Biology & Biochemistry 80 (2015) 251e259 259

H€ogberg, M., H€ogberg, P., Myrold, D., 2007. Is microbial community composition inboreal forest soils determined by pH, C-to-N ratio, the trees, or all three?Oecologia 150, 590e601.

IPCC, 2007. In: Metz, B., Davidson, O.R., Bosch, P.R., Dave, R., Meyer, L.A. (Eds.),Contribution of Working Group III to the Fourth Assessment Report of theIntergovernmental Panel on Climate Change. Cambridge University Press,Cambridge, UK.

Kalra, Y.P., Maynard, D.G., 1991. Methods manual for forest soil and plant analysis.Information Report NOR-X-319E. Natural Resources Canada, Canadian ForestService, Northern Forestry Centre, Edmonton, Alberta, p. 116.

Kampichler, C., Bruckner, A., 2009. The role of microarthropods in terrestrialdecomposition: a meta-analysis of 40 years of litterbag studies. Biological Re-views 84, 375e389.

Kurz, W.A., Dymond, C.C., White, T.M., Stinson, G., Shaw, C.H., Rampley, G.J.,Smyth, C., Simpson, B.N., Neilson, E.T., Trofymow, J.A., Metsaranta, J., Apps, M.J.,2009. CBM-CFS3: a model of carbon-dynamics in forestry and land-use changeimplementing IPCC standards. Ecological Modelling 220, 480e504.

Lauber, C.L., Strickland, M.S., Bradford, M.A., Fierer, N., 2008. The influence of soilproperties on the structure of bacterial and fungal communities across land-usetypes. Soil Biology and Biochemistry 40, 2407e2415.

Leckie, S.E., 2005. Methods of microbial community profiling and their applicationto forest soils. Forest Ecology and Management 220, 88e106.

Leckie, S.E., Prescott, C.E., Grayston, S.J., Neufeld, J.D., Mohn, W.W., 2004. Charac-terization of humus microbial communities in adjacent forest types that differin nitrogen availability. Microbial Ecology 48, 29e40.

Lousier, J.D., Parkinson, D., 1976. Litter decomposition in a cool temperate deciduousforest. Canadian Journal of Botany 54, 419e436.

Manzoni, S., Jackson, R.B., Trofymow, J.A., Porporato, A., 2008. The global stoichi-ometry of litter nitrogen mineralization. Science 321, 684e686.

Manzoni, S., Porporato, A., 2009. Soil carbon and nitrogen mineralization: theoryand models across scales. Soil Biology and Biochemistry 41, 1355e1379.

McKenney, D.W., Hutchinson, M.F., Kesteven, J.L., Venier, L.A., 2001. Canada's planthardiness zones revisited using modern climate interpolation techniques. Ca-nadian Journal of Plant Science 81, 129e143.

Mitchell, R.J., Campbell, C.D., Chapman, S.J., Cameron, C.M., 2010. The ecologicalengineering impact of a single tree species on the soil microbial community.Journal of Ecology 98, 50e61.

Moore-Kucera, J., Dick, R., 2008. PLFA profiling of microbial community structureand seasonal shifts in soils of a Douglas-fir chronosequence. Microbial Ecology55, 500e511.

Moore, T.R., Trofymow, J.A., Prescott, C.E., Fyles, J., Titus, B.D., 2006. Patterns ofcarbon, nitrogen and phosphorus dynamics in decomposing foliar litter in Ca-nadian forests. Ecosystems 9, 46e62.

Murray, F.W., 1967. On the computation of saturation vapor pressure. Journal ofApplied Meterology 203e204.

Nilsson, L.O., Giesler, R., Baath, E., Wallander, H., 2005. Growth and biomass ofmycorrhizal mycelia in coniferous forests along short natural nutrient gradi-ents. New Phytologist 165, 613e622.

Norris, C.E., Quideau, S.A., Macey, D.E., 2013. Processing of 13C glucose in mineralsoil from aspen, spruce and novel ecosystems in the Athabasca oil sands region.Applied Soil Ecology 71, 24e32.

Palosuo, T., Liski, J., Trofymow, J.A., Titus, B.D., 2005. Litter decomposition affectedby climate and litter qualitydTesting the Yasso model with litterbag data fromthe Canadian intersite decomposition experiment. Ecological Modelling 189,183e198.

Parkinson, J.A., Allen, S.E., 1975. A wet oxidation procedure suitable for the deter-mination of nitrogen and mineral nutrients in biological material. Communi-cations in Soil Science and Plant Analysis 6, 1e11.

Prescott, C.E., 2010. Litter decomposition: what controls it and how can we alter itto sequester more carbon in forest soils? Biogeochemistry 101, 133e149.

SAS, 2008. SAS/STAT 9.2 User's Guide. SAS Institute Inc., Cary, NC.Satterthwaite, F.E., 1946. An approximate distribution of estimates of variance

components. Biometrics Bulletin 2, 110e114.Set€al€a, H., Marshall, V.G., Trofymow, J.A., 1996. Influence of body size of soil fauna on

litter decomposition and 15N uptake by poplar in a pot trial. Soil Biology andBiochemistry 28, 1661e1675.

Smyth, C.E., Kurz, W.A., Trofymow, J.A., 2011. Including the effects of water stress ondecomposition in the Carbon Budget Model of the Canadian Forest Sector CBM-CFS3. Ecological Modelling 222, 1080e1091.

Smyth, C.E., Trofymow, J.A., Kurz, W.A., Group, C.W., 2010. Decreasing uncertainty inCBM-CFS3 estimates of forest soil C sources and sinks through use of long-termdata from the Canadian Intersite Decomposition Experiment. BC InformationReport BC-X-422. Natural Resources Canada, Canadian Forest Service, PacificForestry Centre, Victoria. http://cfs.nrcan.gc.ca/pubwarehouse/pdfs/31205.pdf.

Staddon, W., Trevors, J., Duchesne, L., Colombo, C., 1998. Soil microbial diversity andcommunity structure across a climatic gradient in western Canada. Biodiversityand Conservation 7, 1081e1092.

Steger, K., Sj€ogren, Å.M., Jarvis, Å., Jansson, J.K., Sundh, I., 2007. Development ofcompost maturity and Actinobacteria populations during full-scale compost-ing of organic household waste. Journal of Applied Microbiology 103,487e498.

Trofymow, J.A., CIDET Working Group, 1998. CIDETdthe Canadian IntersiteDecomposition Experiment: Project and Site Establishment Report., BC Infor-mation Report BC-X-378. Natural Resources Canada, Canadian Forest Service,Pacific Forestry Centre, Victoria, BC. http://cfs.nrcan.gc.ca/pubwarehouse/pdfs/5030.pdf.

Trofymow, J.A., Moore, T.R., Titus, B., Prescott, C., Morrison, I., Siltanen, M.,Smith, S., Fyles, J., Wein, R., Camirt, C., Duschene, L., Kozak, L., Kranabetter, M.,Visser, S., 2002. Rates of litter decomposition over 6 years in Canadian forests:influence of litter quality and climate. Canadian Journal of Forest Research 32,789e804.

Trofymow, J.A., Preston, C.M., Prescott, C.E., 1995. Litter quality and its potentialeffect on decay rates of materials from Canadian forests. Water, Air and SoilPollution 82, 215e226.

Wallander, H., Nilsson, L.O., Hagerberg, D., Bååth, E., 2001. Estimation of the biomassand seasonal growth of external mycelium of ectomycorrhizal fungi in the field.New Phytologist 151, 753e760.

White, D.C., Davis, W.M., Nickels, J.S., King, J.D., Bobbie, R., 1979. Determination ofthe sedimentary microbial biomass by extractable lipid phosphate. Oecologia40, 51e62.

White, D.C., Ringelberg, D.B., 1998. Signature lipid biomarker analysis. In:Burlage, R.S., Atlas, R., Stahl, D., Geesey, G., Sayler, G. (Eds.), Techniques in Mi-crobial Ecology. Oxford University Press, New York, pp. 255e272.

Wise, D.H., Schaefer, M., 1994. Decomposition of leaf litter in a mull beech forest:comparison between canopy and herbaceous species. Pedobiologia 38,269e288.

You, Y., Wang, J., Huang, X., Tang, Z., Liu, S., Sun, O.J., 2014. Relating microbialcommunity structure to functioning in forest soil organic carbon trans-formation and turnover. Ecology and Evolution 4, 633e647.

Zelles, L., 1999. Fatty acid patterns of phospholipids and lipopolysaccharides in thecharacterisation of microbial communities in soil: a review. Biology and Fertilityof Soils 29, 111e129.

Zhang, C.F., Meng, F.-R., Trofymow, J.A., Arp, P.A., 2007. Modeling Mass and NitrogenRemaining in Litterbags for Canadian Forest and Climate Conditions. Agricul-tural Institute of Canada, Ottawa, Ontario, 20 pp.

Zhang, C.F., Meng, F.R., Bhatti, J.S., Trofymow, J.A., Arp, P.A., 2008. Modeling forestleaf-litter decomposition and N mineralization in litterbags, placed acrossCanada: a 5-model comparison. Ecological Modelling 219, 342e360.