Fungi participate in driving home-field advantage of litter decomposition in a subtropical forest

Transcription

1 REGULAR ARTICLE Fungi prticipte in driving home-field dvntge of litter decomposition in subtropicl forest Dunmei Lin & Mei Png & Nicols Fnin & Hongjun Wng & Shenhu Qin & Ling Zho & Yongchun Yng & Xingcheng Mi & Keping M Received: 25 My 218 /Accepted: 24 October 218 # Springer Nture Switzerlnd AG 218 Abstrct Bckground nd ims Home-field dvntge (HFA) hypothesis predicts tht plnt litter decomposes fster beneth the plnt species from which it ws derived thn beneth other plnt species. However, it remins uncler, which groups of soil orgnisms drive HFA effects cross wide rnge of litter qulity nd forest types. Methods We set up reciprocl trnsplnt decomposition experiment to quntify the HFA effects of brodlef, coniferous nd bmboo litters. Litterbgs of different mesh sizes nd high-throughput pyrosequencing of microbil rrna gene were used to test the contribution of different decomposer groups to HFA effect. Results The reclcitrnt brodlef litter nd the lbile bmboo litter exhibited HFA. Presence of meso-nd mcrofun did not substntilly chnge the HFA effects. Bcteril nd fungl community composition on litters were significntly influenced by litter type. Bcteril community composition remined unchnged when the sme litter ws decomposed in different forest types, wheres fungl community composition on brodlef nd bmboo litters were significntly influenced by incubtion site. Conclusions Our dt demonstrte specific ssocition between fungl community composition nd fster litter decomposition in the home site, suggesting tht fungi probbly prticipte in driving the HFA effect of brodlef nd bmboo litters. Responsible Editor: Frnçois Teste. Electronic supplementry mteril The online version of this rticle ( contins supplementry mteril, which is vilble to uthorized users. D. Lin (*): M. Png : S. Qin : L. Zho : Y. Yng Key Lbortory of the Three Gorges Reservoir Region s Eco-Environment, Ministry of Eduction, Chongqing University, 174th, Shpingb Zhengjie Street, Shpingb District, Chongqing 445, Chin e-mil: lindunmei@cqu.edu.cn N. Fnin Interction Soil Plnt Atmosphere (ISPA), UMR 1391, INRA - Bordeux Sciences Agro, 71 venue Edourd Bourlux, Villenve-d Ornon cedex, Frnce H. Wng Biotechnology Reserch Center, Chongqing Acdemy of Agriculturl Sciences, Chongqing 41329, Chin S. Qin : L. Zho : Y. Yng Ntionl Centre for Interntionl Reserch of Low-crbon nd Green Buildings, Ministry of Science & Technology, Chongqing University, Chongqing 445, Chin X. Mi : K. M Stte Key Lbortory of Vegettion nd Environmentl Chnge, Institute of Botny, The Chinese Acdemy of Sciences, Beijing 193, Chin

2 Keywords Home-field dvntge. Litter-decomposer interctions. Litter trits. Locl dpttion. Functionl redundncy Introduction Primry production nd decomposition of detritus re two fundmentl ecologicl processes on erth (Berg nd McClugherty 214; Swift et l. 1979). Most primry production in terrestril ecosystems enters the soil s litter or ded orgnic mtter (Cebrin 1999). Decy of this mteril provides substntil nutrients nd energy tht supports detritl food webs, influencing soil fertility nd the globl crbon cycle (Berg nd McClugherty 214; Gessner et l. 21; Swift et l. 1979). It is commonly ccepted tht climte nd litter qulity re the min fctors controlling litter decomposition t brod sptil scles (Meentemeyer 1978; Moore et l. 1999; Zhng et l. 28). However, t the locl scle vriety of biotic fctors such s composition of decomposer communities or interctions mong litter species cn lso influence decomposition rtes (Austin et l. 214; Ayres et l. 29; Gessner et l. 21). Among the potentil mechnisms explining the vribility in decomposition rtes t the locl scle, severl recent studies hve reported tht litter tends to decompose fster in the site from which it ws derived (i.e. home site) thn when plced in wy sites, phenomenon clled home-field dvntge (HFA) (Ayres et l. 29; Gholz et l. 2). Although the HFA effect for litter decomposition hs gined much ttention over the lst decde (e.g. Ayres et l. 29b; Milcu nd Mnning 211; Chomel et l. 215), mny studies did not find ny evidence of such n effect (e.g. Giesselmnn et l. 211; StJohnetl. 211; Veenetl. 215), thereby chllenging the generlizbility of this phenomenon in terrestril ecosystems. Furthermore, most previous studies did not disentngle the true HFA effect, i.e. dpttion of soil biot to decomposition of litter t home, from the functionl bredth hypothesis, i.e. the bility of soil biot to decompose more efficiently mny different litter types t the sme time (Fnin et l. 216; Keiseretl. 211, 214). It is cler tht mgnitude of HFA effect is difficult to predict unless we uncover the underlying mechnisms of HFA effect of litter decomposition. HFA could result from long-term ecologicl interctions between decomposer communities nd litter tht they encounter most often. In prticulr, microbil decomposers my differ in their bility nd preference to utilize different crbon sources (Durll et l. 1994; Hnson et l. 28; McGuire et l. 21). Therefore, fster decomposition rtes t home my rise becuse of niche differences mong microbil communities t different sites. For instnce, the high proportion of complex nd/or toxic compounds contined in reclcitrnt litter requires specilist decomposers to brek them down (Ayres et l. 29; Milcu nd Mnning 211). By contrst, the high proportion of rich nd lbile crbon compounds contined in some litters my stimulte competition mong copiotrophic microbil decomposers. Therefore, fster decomposition rtes my rise becuse of higher decomposer growth rte t home site. Bsed on this theoreticl frmework, it s likely tht both low qulity nd high qulity litter cn decompose fster in their home sites, but for different resons. Despite the incresing recognition tht HFA depends on the functionl ttributes of soil communities, little is known bout the min groups of soil orgnisms (fun, bcteri nd fungi) contributing to HFA effects. It hs been shown recently tht the direction nd strength of interction between soil fun nd litter trits my vry mong sites nd biomes (Grci-Plcios et l. 213; Milcu nd Mnning 211; Perez et l. 213), suggesting n importnt context-dependency of HFA effects (Chomel et l. 215; Milcu nd Mnning 211). Such vribility in the effects of soil fun on decomposition rtes my depend on their impct on microbil ctivity (Brdgett 25; Berg nd McClugherty 214). In ddition, empiricl evidence on the reltive contribution of soil microorgnisms to HFA re scrce in the literture (e.g. Fnin et l. 216). In prticulr, the role of fungi nd bcteri my differ cross contrsting ecosystems nd/or litter qulity becuse of their differences in functionl ttributes nd crbon requirements. Fungi re generlly thought to possess stronger cpcities thn bcteri for decomposing recciltrnt plnt mteril, wheres bcteri re thought to be more efficient t exploiting lbile crbon compounds (de Grff et l. 21; Huntetl.1987; Pterson et l. 28). However, whether, how nd why the min contributors of HFA effect vry between lbile nd reclcitrnt litters hs received only wek ttention in the literture. In this study, we im to investigte HFA during lef litter decomposition cross contrsting ecosystems by performing reciprocl litter trnsplnttion experiment using three tree species (Cstnopsis eyrei, Cunninghmi lnceolt nd Phyllostchys heterocycl

3 cv. Pubescens) originting from three different forest types (brodlef, coniferous nd bmboo forests) in subtropicl Chin. We used litterbgs of two different mesh sizes nd pyrosequencing technique to ddress the role of soil biot, i.e. soil fun, bcteri nd fungi, in explining the HFA effects. More specificlly, becuse of the supposedly specilized role of bcteri nd fungi during litter decomposition (de Grff et l. 21; Hunt et l. 1987; Pterson et l. 28), we hypothesized tht different groups of decomposer microorgnisms should dominte depend on the litter qulity: fungi should be the min group of orgnism prticipting HFA effect in reclcitrnt litter becuse of their greter functionl cpcities in using low qulity crbon resource (H 1 ), nd bcteri should be the min group of orgnism prticipting to HFA effects in lbile litter becuse of their fster rte in resource cquisition nd exploittion (H 2 ). Finlly, given tht some studies reported soil fun ply n importnt role in driving the HFA effects (e.g. Milcu nd Mnning 211), we hypothesized tht meso- nd mcrofun presence should substntilly enhnce the strength of HFA effects (H 3 ). Mteril nd methods Study site We conducted this study in subtropicl forest within Gutinshn Ntionl Nture Reserve (GNNR, pproximtely 81 km 2 in re), locted in the western prt of Zhejing Province, Est Chin ( N, E). The region hs subtropicl monsoon climte, with men nnul temperture of 15.3 C, nd men nnul precipittion of 1964 mm (most of which occurs between Mrch nd September). The prent rock of the mountin rnge is grnite, nd the predominnt soil types re red, red-yellow, yellow-red, nd mrsh soil. The GNNR comprises lrge re of nturl evergreen brodlef forest (57% of the reserve re) dominted by C. eyrei, Schim superb, Cycloblnopsis gluc, swellsconiferousplnttions (C. lnceolt or Pinus mssonin), te-seed oil plnttions (Cmelli oleifer) nd bmboo plnttions (P. heterocycl cv. Pubescens) (Linetl.216). Experimentl design We selected three forest types which differ significntly in species compositon for the study: brodlef forest, coniferous forest, nd bmboo forest. C. eyrei, C. lnceolt nd P. heterocycl cv. Pubescens re the dominnt tree species of the brodlef, coniferous nd bmboo forest, respectively (Tble 1). The shortest nd longest distnce between ny two forest sites is 1.6 nd 4.6 km, respectively. We selected one plot t ech forest site for the estblishment of our litter decomposition experiment. All selected plots hd similr ltitude, slope nd hence present similr microclimtic conditions. In Jnury 216, three soil cores (5 cm depth) were rndomly collected within ech plot for soil physicochemicl nlyses. Soil smples were ir-dried, sieved with 2-mm mesh nd grounded. Four edphic vribles were mesured: ph, soil orgnic crbon, totl nitrogen nd phosphorus (Tble 1).Soilwsshkenin1molL 1 KCl solution, nd ph ws mesured using ph meter. Soil orgnic crbon ws mesured by dry combustion in solid module (Shimdzu SSM- 5, Jpn) coupled with TOC/TN nlyzer (Shimdzu TOC-L CPH, TNM-1, Jpn). After digestion of soil powder with concentrted H 2 SO 4 nd 3% H 2 O 2 (Jones 21), totl nitrogen ws mesured by using the TOC/TN nlyser, nd totl phosphorus content ws determined spectrophotometriclly by using molybdenum blue method. We instlled full reciprocl trnsplnt decomposition experiment using the litterbg method in the three forest types. Lef litter of the three dominnt species, i.e. C. eyrei, C. lnceolt nd P. heterocycl cv. Pubescens were used nd we referred to these litter types herefter s brodlef litter, coniferous litter, nd bmboo litter, respectively. Litter ws collected fter nturl bscission using litter trps (1 m 1 m) plced underneth tree cnopies. Litter smples for ech species were bulked, homogenised nd oven dried t 4 C for 48 h in the lbortory. Five subsmples of the dried mteril from ech litter type were then oven dried t 6 C to determine the weight conversion fctor between 4 C nd 6 C. To determine the contribution of soil fun to decomposition we constructed two different types of litterbgs: corse mesh size (4-mm upper side nd.5- mm lower side fcing the soil surfce to void loss of litter frgments during field exposure) nd fine mesh size (25-μm onbothside).the25μm meshsize llowed for the ccess of only microorgnisms nd microfun, while the 4 mm mesh size lso llowed meso-nd mcrofun to enter the litterbg. Both types of litterbgs were 1 cm 15 cm in size, filled with pproximtely 3. g of 4 C-dried litter, nd lbeled

4 Tble 1 Tree community chrcteristics, soil chemicl properties (-5 cm depth) nd initil litter qulity of the three tree species Cstnopsis eyrei Cunninghmi lnceolt Phyllostchys heterocycl cv. Pubescens Site description Forest type Brodlef Coniferous Bmboo Reltive bsl re (%) 55.1% 81.9% 84.5% DBH rnge (cm) Soil ph 3.72 ± ± ±.8 Soil orgnic C (mg g 1 ) 41.7 ± ± ± 12.7 Soil N (mg g 1 ) 2.9 ± ± ± 1.97 Soil P (mg g 1 ).158 ±.6.92 e±.5.21 ±.5 Soil C/N 15.3 ± ± ±.7 Initil litter trits C (mg g 1 ) ± ± b ± c N(mgg 1 ) 9.8 ± ±.1 b 7.6 ±.88 c P (mg g 1 ).27± ± ±.21 K(mgg 1 ) 1.861± ± ±.88 b Mn (mg g 1 ).795±.9.74 ± ±.63 b C/N ± ±.52 b ±.422 c Proximte lignin (mg g 1 ) ± ± 4.45 b ± 6.3 c Totl phenols (mg g 1 ) 24.9 ± ± 1.68 b 6.9 ±.6 c Tnnin (mg g 1 ) 15. ± ± 1.52 b.2 ±.8 c Plnt Soil Reltive bsl re nd rnge of tree dimeter t brest height (DBH) were clculted bsed on 2 m 2 m census plot for ech forest type. Dt re men ± SE (n = 3). Different letters indicte significnt differences (P <.5) on the bsis of one-wy ANOVA followed by pirwise multiple comprisons (Tukey s HSD test) with plstic tgs. In totl, 18 litter bgs (3 incubtion sites 2 mesh size 3 litter types 6 replictes = 18) were used to determined decomposition rtes. Three dditionl fine-mesh bgs for ech litter types nd ech incubtion site (27 in totl) were used to monitor the microbil communities of the decomposing litter. Six blocks beneth tree crowns of C. eyrei, C. lnceolt nd P. heterocycl cv. Pubescens in brodlef, coniferous nd bmboo forest plots, respectively, were selected for litterbg incubtion. The distnce between ny two blocks is t lest 1 m. All litterbgs were fixed on the surfce of the forest floor using plstic nils fter recently fllen litter ws removed. After 222 dys of exposure in the field (25th of Jnury to 3rd of September, 216) ll litter bgs were hrvested nd put in seled polyethylene bgs. In the lbortory, the remining litter ws removed from litterbgs nd gently brushed to remove dhering soil prticles nd other foreign mterils. Smples were then dried t 6 C nd weighed to determine the remining litter mss. Litter mss loss (M l,%)ws clculted s M l =(M i -M f )/M i 1%, where M i nd M f re the initil nd finl litter mss (dry t 6 C), respectively. Litter mss loss dt re vilble in the Electronic supplementry mteril. Litter qulity mesurements We used three replictes per species to ssess litter qulity. Totl crbon (C) nd nitrogen (N) concentrtions were mesured by dry combustion using n elementl nlyzer (MACRO Cube Elementl Anlyzer, Elment, Itly). C/N rtio ws clculted bsed on the C nd N concentrtions. Phosphorus (P), potssium (K) nd mngnese (Mn) concentrtions were mesured using inductively coupled plsm emission spectroscopy (ICP-OES; Therm Jrrel-Ash, IRIS Advntge, MA, USA) fter digestion of ground lef litter mteril with concentrted HNO 3 nd 3% H 2 O 2 (Jones 21). Proximte lignin ws determined by the cid-detergent fiber method s described in Grç et l. (25). Totl phenols were extrcted with 75% cetone solution, nd concentrtions of totl phenols, tnnins were mesured colorimetriclly with the Folin-Cioclteu regent

5 following the method of Mkkr (23), using tnnic cid s stndrd. Microbil community on decomposing litter Litterbgs for ssessing microbil community structure (three replicte smples per tretment) were lso hrvested on September 3rd, 216. They were plced in het insultion box cooled by ice immeditely fter collection nd shipped to the lbortory in Chongqing University (< 24 h). Smples were stored in 8 C freezer until further processing. We extrcted DNA from.3 g of frozen lef litter using FstDNA Spin Kit for Soil ( , MPBIO, Cliforni, USA), ccording to the mnufcturer s protocol nd quntified the extrcted DNA using NnoDrop 2 (Thermo Fisher Scientific Inc., Wilmington, DE, USA). The fungl nd bcteril bundnces were estimted by using quntittive rel-time PCR (qpcr). A frgment of the fungl 18S rrna gene ws mplified using the forwrd primer 5 -GGCAAGTCTGGTGCCAG-3 combined with reverse primer 5 - ACGGTATC T(AG)ATC(AG)TCTTCG-3, whilefrgmentofthe bcteril 16S rrna gene ws mplified using the forwrd primer 5 -ACTCCTACGGGAGGCAGCAG-3 combined with reverse primer 5 -TACNVGGGTATCT AATCC-3. Stndrd curves were generted using 1- fold seril dilutions of plsmid (pgem-t) contining the trgeted gene inserts for the 18S nd 16S rrna gene, respectively. The qpcr rections were performed in duplicte 1 μl mixtures, ech contining 5 μl GoTq qpcr Mster Mix Technicl Mnul (A61, Promeg, USA), 1 μl of ech forwrd nd reverse primers, nd 2 μl sterile DNA-free wter, nd 1 μl stndrd or soil DNA smples. The rection ws crried out on n ABI ViiA 7 Rel Time PCR System (ABI, USA), using rection conditions of 95 Cfor 3 min followed by 4 cycles of 95 Cfor 15 s, 6 Cfor 3s nd 72 Cfor 3s. Fungl nd bcteril gene copy numbers were clculted ccording to the stndrd curves tht relte the cycle threshold vlue to the known number of copies in the stndrds (Rousk et l. 21). Fungl nd bcteril communities tht colonized on litter were ssessed using high-throughput sequencing methods. Fungl nd bcteril mplicon librries were obtined for pyrosequencing using custom fusion primers. The primer pir 338-F (5 - ACTCCTAC GGGAGGCAGCAG-3 ) nd 86-R (5 - GGACTACHVGGGTWTCTAAT-3 ) wereusedto mplify the V3-V4 region of the bcteril 16S rrna gene (Zhou et l. 216), nd the primer pir ITS1-F (5 - CTTGGTCATTTAGAGGAAGTAA-3 ) ndits2-243r (5 -GCTGCGTTCTTCATCGATGC-3 ) were used to mplify the fungl internl trnscribed spcer (ITS) rrna region (McGuire et l. 213). Amplifiction ws conducted on ABI GeneAmp 97 with the following settings: initil denturtion t 95 C for 3 min, followed by 27 or 35 cycles seprtely for bcteril nd fungl genes, consisting of denturtion t 95 C for 3 s, nneling t 55 C for 3 s, extension t 72 C for 45 s, nd finlly extension t 72 C for 1 min. Ech smple ws mplified in triplicte nd the three PCR products of ech smple were pooled nd purified. Smples were then evluted for quntity nd qulity vi electrophoresis with 2% grose gel. The resulting gene mplicon smples were sequenced on the Illumin MiSeq pltform (Sn Diego, CA, USA) t Mjorbio BioPhrm Technology Co., Ltd. (Shnghi, Chin), using pired-end 3-bp red lengths. For both 16S nd ITS genes, sequences were preprocessed nd qulity filtered before downstrem nlyses using Trimmomtic progrm (Bolger et l. 214) nd FLASH progrm (Mgoc nd Slzberg 211). Reds contining mbiguous bse BN^ were removed. Reds were truncted t ny site receiving n verge qulity score < 2 over 5 bp sliding window, nd ny trunctedredsshorterthn5bpwerediscrded. Reds without exct brcode mtching or > 2 nucleotide mismtches in primer mtching were removed. Forwrd nd reverse reds of sme sequence with t lest 1 bp overlp nd < 2% mismtches were ssembled into contigs. Reds tht could not be ssembled were lso discrded. Singleton contigs were discrded nd the remining contigs were hndled s OTUs (Opertionl Txonomic Units). OTUs were clustered t 97% sequence similrity level for both fungl nd bcteril sequences nd n OTU tble ws creted by identifying the number of sequences of ech OTU in ech smple using UPARSE progrm (Edgr 213). Dt nlysis Dt ws tested for norml distribution using the Shpiro Wilk test (P >.5) nd homogeneity of vrinces using Breusch-Pgn test (P >.5). We used onewy ANOVA to test for interspecific differences in litter chemistry nd post hoc comprisons between litter species were evluted using Tukey s HSD.We performed

6 rndom forest nlysis to ssess which litter trit ws the best fctor explining litter mss loss, with litter trits s explntory vribles nd litter mss loss of ll litterbgs s response vrible. We performed three-wy ANOVA to exmine the effects of incubtion site (brodlef, coniferous, nd bmboo forest), mesh size of litterbgs (corse nd fine), litter type (brodlef, coniferous nd bmboo litter), s well s ll possible interctions mong these fctors on litter mss loss. We then nlysed the litter mss loss for ech litter type individully using two-wy ANOVA in which the incubtion site nd mesh size of litterbg were treted s fixed effects nd were llowed to interct. We then performed the regression model developed by Keiser et l. (214), which seprtes the overll bility of soil biot to decompose different litter types nd the rel HFA. The model sttes tht litter mss loss is equl to litter bility (β l, i.e. litter qulity index) plus soil bility (γ s, i.e. inherent functionl cpcity of soils) plus home interction (η h, i.e. HFA), s follow (Keiser et l. 214): Y i ¼ α þ N l¼1 β l Litter li þ M s¼1 γ s Soil si þ K h¼1 η h Home hi þ ε i where Y i is the litter mss loss for observtion i. α is the intercept term, represents the verge litter mss loss for ll observtions in the dtset fter controlling for litter, soil, nd HFA effects. β l is the bility of litter species l (from species 1 to N), γ s is the bility of the soil community s (from community 1 to M), nd η h is the HFA of h (from home combintions 1 to K). The estimted prmeters re β l, γ s nd η h.bothof N l¼1 β l nd M s¼1 γ s were restricted to zero to prevent perfect collinerity (Keiser et l. 214). Litter l,soil s nd Home h re dummy vribles tht is set s 1 or depending on the presence or bsence of the litter species, soil community or home combintion, respectively (Keiser et l. 214). ε is the error term. Using this model, we clculted the litter qulity index (β l ), the functionl bility index of soil decomposer communities (γ s ), nd the HFA index (η h ) on the litter mss loss. We nlysed the dt from finend corse-mesh litterbgs seprtely. We rn the models in SAS version 9.3 (SAS Institute 21) using the SAS code provided in Keiser et l. (214). We used one-wy ANOVA to test the effects of incubtion site on the bundnces of fungi nd bcteri on decomposing litters nd post hoc comprisons between incubtion sites were evluted using Tukey s HSD.We used nonmetric multidimensionl scling (NMDS) nlysis to visulize the differences in the structure of bcteril nd fungl community on decomposing litters. We then used permuttionl multivrite ANOVA (PERMANOVA) to test the effect of incubtion site, litter type, nd their interctions on bcteril nd fungl communities with 9999 rndom permuttions. We lso rn PERMANOVA for ech litter type seprtely. Bry- Curtis dissimilrity mtrices were used in ll the nlyses. We performed sttisticl nlyses in the softwre R (R Development Core Tem 216), using the R pckge rndomforest (Liw nd Wiener 22) for rndom forest nlysis, vegn (Oksnen et l. 216) for NMDS nd PERMANOVA nlysis. P <.5 ws considered s sttisticl significnce. Results Litter qulity Litter chemicl trits differed significntly mong the three litter types, except for the P content (P <.5; Tble 1). The interspecific vribility of litter chemistry ws prticulrly pprent for N content, C/N rtio, proximte lignin, totl phenols, nd tnnins contents which vried by 2.1-, 2.2-, 3.-, 3.6-nd 75-fold, respectively (Tble 1). Litter qulity strongly differed mong the three litter types ccording to the two first xes of the PCA plot (Online Resource 1). Bmboo litter ws segregted from brodlef nd coniferous long the first xis wheres brodlef ws opposite to coniferous long the second xis. Bmboo litter presented significntly lower content of tnnins tht ws bout 54-nd 74-fold lower thn coniferous nd brodlef litters, respectively (Tble 1). Coniferous litter ws chrcterized by significntly higher lignin content thn other two litter types, nd brodlef litter ws significntly higher in totl phenols nd tnnins contents (Tble 1). Overll, litter qulity of bmboo litter ws reltively higher thn tht of brodlef nd coniferous litter. Litter mss loss nd home-field dvntge After 222 dys of exposure in the field litter mss loss differed significntly mong the different litter types (ANOVA, F 2,15 = 15.3, P <.1; Fig. 1), with the highest mss loss observed for bmboo litter (24.6 ± 1.2%), followed by coniferous litter (24.3 ±.6%) nd brodlef litter (17.8 ± 1.%). Rndom forest nlysis

7 Fig. 1 Litter mss loss (%) of the threetreespeciesthtwere enclosed in (-b-c) fine-mesh (25 μm, left column) nd (d-e-f) corse-mesh (4 mm, right column) litterbgs fter 222 dys of filed exporsure. Grey br denotes the litter ws decomposed in its home site. Error brs respresent ± SE (n = 6). Different letters bove the brs indicte significnt differences (P <.5; Tukey s hsd test) Mss loss (%) Mss loss (%) Mss loss (%) 35 () Brodlef litter b (b) Coniferous litter (c) Bmboo litter Brodlef Coniferous Bmboo Incubtion site (d) Brodlef litter + (e) Coniferous litter + (f) Bmboo litter b + Brodlef Coniferous Bmboo Incubtion site b showed tht tnnins content ws the most importnt litter qulity trit tht influences litter mss loss (Online Resource 2). The Keiser s model showed tht the litter qulity index (β l ) estimtes were higher for coniferous nd bmboo litter compred to brodlef litter (Fig. 2). Litter type explined the lrgest proportion (22.5%; P <.1) of the vrition in litter mss loss (Tble 2). Although there ws no significnt effect of incubtion site (P =.17), the interction between incubtion site nd litter type ws the second most importnt driver explining 2% of the vrition in litter mss loss (Tble 2). There were greter mss losses for brodlef nd bmboo litter when they were incubted in their home site thn in wy sites, but not for coniferous litter (Fig. 1). Consistently, the HFA index estimted from Keiser s model displyed significnt positive HFA index (η h ) for brodlef nd bmboo litter (Fig. 2b), confirming tht they decomposed fster in the re dominted by the plnt species from which they were derived. The functionl bility index (γ s ) estimtes showed tht the coniferous forest hd the highest functionl bility to decompose ll litter (Fig. 2c). Mesh size of litterbg hd significnt negtive impct on mss loss for coniferous nd bmboo litter (P =.34 nd P <.1, respectively), but not for brodlef litter (P =.87). This led to significnt interction between mesh size nd litter type on litter mss loss (P <.1; Tble 2). In ddition, HFA index (η h ) estimted from Keiser s model between fine-nd corse-mesh litterbgs for ech litter type hd pprent overlpped stndrd errors (Fig. 2b). Microbil community on decomposing litter There ws no significnt effect of incubtion site on the bundnce of fungi (determined using qpcr) for brodlef nd coniferous litters (Fig. 3, b). However, we found tht the bundnce of fungi ws significntly higher when bmboo litter ws incubted in its home site (P <.5; Fig. 3c). Incubtion site hd no significnt effect on the bcteril bundnce for ll litter types (Fig. 3d f). Litter type ws the most importnt vrible explining 2.8% of the the vrition in the fungl community structure (P <.1; Tble 3). Ordintion plot from NMDS clerly distinguished the fungl communities of the three litter types (Fig. 4). Although there ws no significnt effect of incubtion site (P =

8 Prmeter estimtes (%) Prmeter estimtes (%) Prmeter estimtes (%) () (b) (c) *** *** *** * *** ** ** Fine-mesh litterbg Corse-mesh litterbg *** ** *** *** ** *** Brodlef Coniferous Bmboo Fig. 2 Prmeter estimtes (men ± SE, n = 6) of the litter mss loss clculted using the pproch developed by Keiser et l. (214) for() litter qulity index, (b) home-field dvntge (HFA) index nd (c) functionl bility index. Litter qulity index reltes to the reltive bility of ech different litter (brodlef, coniferous nd bmboo) to be decomposed by ll the decomposer communities used in our study, HFA estimtes the interction between the litter decomposition nd the decomposer communities in ech forest (brodlef, coniferous nd bmboo), nd functionl bility quntifes the overll bility of decomposer community. Dt from fine-nd corse-mesh litterbgs were nlysis seprtely. Positive vlues men positive effect, while negtive vlues mens negtive effect. Estimtes tht differ significntly from zero re indicted by sterisk (* P <.5, ** P <.1 nd ***P <.1).59; Tble 3), the interction between incubtion site nd litter type ws the second most importnt driver, ** ** significntly explining 16.3% of the vrition in the fungl community structure (P =.15; Tble 3). This ws minly driven by the significnt effect of incubtion site on fungi colonizing brodlef nd bmboo litter (P =.3 nd P =.35, respectively), but not for coniferous litter (P =.775; Tble 3). Similr to wht ws observed for the fungl community, we found tht the identity of litter species ws the most importnt driver, significntly explining 33.8% of the vrition in the community structure of bcteri tht colonised in the litter (P <.1;Tble3). Ordintion plot from NMDS distinguished the bcteril communities of brodlef litter from coniferous nd bmboo litters (Fig. 4b). However, incubtion site nd interction between incubtion site nd litter type hd no significnt effects on bcteril communities (P =.584 nd P =.329, respectively; Tble 3), in contrst to our findings regrding fungi. Discussion Due to the hyper-diverse nture of soil biot, soil orgnisms re thought to be functionlly redundnt in terrestril ecosystems nd soil community composition hs long been ignored in globl prediction models (Allison nd Mrtiny 28). Recently, ecologists hve strted to consider the functionl differences mong soil decomposer communities in driving ecosystem processes, nd their results chllenged the ssumptions of functionl redundncy (Delgdo-Bquerizo et l. 216; Fnin et l. 216; Stricklnd et l. 29). Consistent with previous reports cross contrsting forest types (Ayres et l. 29b; Chomel et l. 215; Milcu nd Mnning 211), we found HFA effects for two of three litter types, i.e. bmboo nd brodlef litter decomposed fster t home site (Tble 2; Figs. 1, 2b). Furthermore, our study suggests tht soil fungi prticipte ctively in driving HFA effects for both lbile nd reclcitrnt litters (Tble 3;Fig.4), highlighting tht better understnding of plnt-soil interctions is necessry for predicting how litter decomposition vries cross contrsting ecosystems. Litter qulity s mjor control of litter mss loss It is generlly ccepted tht initil litter qulity strongly influences litter decomposition rtes, with high-qulity litter (e.g. high N content, low lignin, phenols nd tnnins contents) decomposing more rpidly thn poor-qulity

9 Tble 2 ANOVA results for the effects of incubtion site, mesh size, litter type nd ll possible interctions on lef litter decompositon rte Source of vrition df SS %SS F P All litter Incubtion site Mesh size <.1 Litter type <.1 Incubtion site mesh size Incubtion site litter type <.1 Mesh size litter type <.1 Incubtion site mesh size litter type Error Brodlef litter Incubtion site <.1 Mesh size Incubtion site mesh size Error Coniferous litter Incubtion site Mesh size Incubtion site mesh size Error Bmboo litter Incubtion site Mesh size <.1 Incubtion site mesh size Error Significint effects (P <.5) re shown in bold. SS% represents the percentge sum of squres explined litter (e.g. Coq et l. 21; Melillo et l. 1982; Vivnco nd Austin 28). In greement with this theory, we found tht litter type explined the lrgest proportion of the vrition in litter mss loss (Tble 2). Rndom forest nlysis showed tht tnnins content is the most importnt litter qulity trit tht influences litter mss loss (Online Resource 2). Higher tnnins content cn inhibit litter decomposition vi inhibiting microbil growth nd soil enzyme ctivity, forming insoluble complexes with biologicl polymers such s proteins (Chomel et l. 216; Hättenschwiler nd Vitousek 2), nd reducing litter pltbility for soil fun (Coq et l. 21). Interction between decomposers nd their substrtes Besides the effect of litter qulity, we found tht the interction between decomposers nd their substrtes explined significnt, but low proportion of the vribility in decomposition rtes (Tble 2). Contrry to the ide tht HFA should be more pronounced for reclcitrnt thn lbile litter (Ayres et l. 29; Chomel et l. 215; Milcu nd Mnning 211), we found tht both the reclcitrnt litter (brodlef) nd the reltive lbile litter (bmboo) exhibited HFA (Fig. 2b). Although high concentrtions of complex nd/or toxic compounds require specific decomposers with the enzymtic cpcities to brek them down, thereby generting HFA in reclcitrnt substrtes (Durll et l. 1994; Hnson et l. 28; McGuire et l. 21), our results suggested tht HFA my lso occur when litter resources re bundnt nd esily decomposble. This my rise becuse of selecting copiotrophic decomposer communities tht re well dpted to exploit energy-rich resources t home site. This hypothesis ws supported by the higher fungl biomss when the bmboo litter ws decomposed in its home site (Fig. 3c).

10 Fungl copy number () Brodlef litter Bcteril copy number (d) Brodlef litter Fungl copy number Fungl copy number (b) Coniferous litter (c) Bmboo litter Brodlef Coniferous Bmboo Incubtion site Fig. 3 The bundnce of fungi (-c) nd bcteri (d-f) on decomposing litters, s indicted by the number of 18S nd 16S ribosoml DNA (rdna) copies mesured using quntittive PCR b Bcteril copy number Bcteril copy number (e) Coniferous litter (f) Bmboo litter Brodlef Coniferous Bmboo Incubtion site (qpcr). Grey br denotes the litter ws decomposed in its home site. Error brs respresent ± SE (n = 3). Different letters bove the brs indicte significnt differences (P <.5;Tukey s HSD test) Another wy to ddress functionl dissimilrity between contrsting microbil communities is the functionl bredth hypothesis, i.e. the bility of soil biot to decompose more efficiently ll litter types t thesmetime(fninetl.216; Keiseretl.211, 214). Here, we found tht ll litter decomposed fster in the coniferous forest (Fig. 1), suggesting tht the decomposer community hs broder functionl bility to decompose vrious litter types. This ws confirmed by the higher functionl bility index (γs) for the coniferous forest (Fig. 2c). This my rise becuse the long-term coniferous litter inputs, rich in lignin nd presenting high C/N rtios, shpe reltively poor nd reclcitrnt environment for the decomposer communities (Tble 1), thereby stimulting their cpcities to decy wide rnge of substrtes (Keiser et l. 211). Fungi prticipte in driving HFA effect Becuse of the supposedly contrsting role of bcteri nd fungi during litter decomposition, we tested the hypotheses tht fungi nd bcteri should be the min drivers of HFA in reclcitrnt nd lbile litters, respectively. Our results provide support for our first hypothesis (H 1 ) tht fungi re probble the min group of orgnisms prticipting to HFA effects in reclcitrnt litter (i.e. brodlef litter). More specificlly, we found tht incubtion site explined 32.3% of the vrition in the fungl community on brodlef litter (Tble 3), nd NMDS plot showed tht fungl communities on brodlef litter were most dissimilr between litter incubted t home sites (i.e. brodlef forest) nd in the bmboo forest (Fig. 4). Similrly, Chomel et l. (215) reported greter fungl biomss in

11 Tble 3 PERMANOVA evluting the effects of incubtion site, litter type nd their interctions on the fungl nd bcteril community structure Fungl community Bcteril community Source of vrition df SS %SS F P SS %SS F P All litter Incubtion site Litter type < <.1 Incubtion site litter type Error Brodlef litter Incubtion site Error Coniferous litter Incubtion site Error Bmboo litter Incubtion site Error Significint effects (P <.5) re shown in bold. SS% represents the percentge sum of squres explined reclcitrnt spruce plnttion, supporting the ide tht fungi re probbly the most importnt contributors to HFA effects in reclcitrnt environments. By contrst, our results do not support our hypothesis (H 2 ) predicting tht bcteri should be the min group of orgnisms prticipting to HFA effects in lbile litter (i.e. bmboo litter). In prticulr, we did not find significnt effect of the incubtion site on bctericl community structure (P =.65, Tble 3). On the contrry, incubtion site explined 39.5% of the vrition in fungl NMDS () Fungl community stress =.172 Incubted in home site Incubted in home site Incubted in home site Brodlef litter Coniferous litter Bmboo litter NMDS1 Fig. 4 Nonmetric multidimensionl scling (NMDS) ordintion bsed on Bry-Curtis dissimilrities depicting fungl () nd bcteril (b) community composition for brodlef, coniferous nd NMDS2 communities on the bmboo litter (Tble 3), nd NMDS plot showed tht the dissimilrity mong fungl communities on bmboo litter ws highest between litter incubted t home (i.e. bmboo forest) nd in the brodlef forest (Fig. 4). Contrry to wht we expected, this result suggests tht HFA effect in lbile litter is probbly not medited by bcteri, but by fungi. This does not support the widely held views tht bcteril-dominted decomposition pthwys control the decomposition of more lbile orgnic substrtes (de Grff et l. 21; (b) Bcteril community stress =.137 Incubted in home site Incubted in home site Incubted in home site NMDS1 bmboo litter which were incubted in brodlef (blue colour), coniferous (red colour) nd bmboo (green colour) forests. Smples tht re closer together hve similr microbil communities

12 Hunt et l. 1987; Pterson et l. 28). In greement with this result, Rousk nd Frey (215) hve recently found n ssocition between fungi nd highqulity soil crbon fter two decdes of litter mnipultion, chllenging the trditionl view of contrsting life strtegies between bcteri nd fungi during the decomposition process. It is lso noteworthy tht s response to litter qulity chnges during decomposition, fungl communities on litters my chnge (Voriskov nd Bldrin 213). Meso-nd mcrofun did not lter HFA effects Our results do not support our third hypothesis (H 3 ) predicting tht soil meso-nd mcrofun would substntilly enhnce the strength of HFA effects. Insted, we found tht HFA effects were reltively consistent between fine-nd corse-mesh litterbgs (Fig. 2b). Depending on litter type nd climte, there is considerble vrition in the mgnitude of soil fun effect on litter decomposition (Brdford et l. 22; Grci-Plcios et l. 213; Wlletl.28). For instnce, Gonzlez nd Sestedt (21) showed tht funl contribution to litter decomposition vriedfrom1.6%to66%ndcoqetl.(21) reported tht the fun effect vried bout 2-folds mong 16 litter types originting from the sme tropicl forest. Furthermore, Fujii et l. (216) even found significntly negtive fun effect in temperte forest in Jpn. In line with this result, we found lower mss loss in corse-mesh litterbgs (Fig. 1), suggesting tht the inclusion of meso-nd mcrofun reduced litter decomposition probbly becuse of grzing nd/or disruption of microbil decomposer communities by soil fun (Brdford et l. 22; Crowther et l. 212; Newell 1984). However, significntly lower mss loss in corse-mesh litterbgs were only found for coniferous nd bmboo litters but not for the reltively reclcitrnt brodlef litter (Fig. 1; Tble 2). This might be due to selective feeding on microbil decomposers by soil fun (e.g. Brdgett et l. 1993). Conclusion We found tht two out of three litter types, i.e. brodlef (reclcitrnt) nd bmboo (lbile) litter, exhibited HFA effects. This indictes tht HFA is not restricted to either lbile or reclcitrnt litter types nd suggests locl dpttion of decomposer communities to the substrtes tht they encounter the most often. Further, we found tht HFA effects of brodlef nd bmboo litter were probbly driven by fungi, highlighting tht specific plnt-fungi interctions re importnt for litter decomposition. Acknowledgements We thnk Pei Wng nd Yn Liu for their help in perpring litterbgs, nd Zhenxi Li, Pengpeng Dou nd Fng Wng for their help in the field nd lbortory. We would lso like to thnk Alison Bemish t the University of British Columbi for her ssistnce with English lnguge nd grmmticl editing of the mnuscript, nd nonymous reviewers for constructive comments on the mnuscript. This work ws supported by the Ntionl Nturl Science Foundtion of Chin [No ], Chongqing Reserch Progrm of Bsic Reserch nd Frontier Technology [No. cstc216jcyja4], Fundmentl Reserch Funds for the Centrl Universities [No. 218CDXYCH14] nd the 111 Project [No. B1341]. References Allison SD, Mrtiny JBH (28) Resistnce, resilience, nd redundncy in microbil communities. Proc Ntl Acd Sci U S A 15: Austin AT, Vivnco L, Gonzlez-Arzc A et l (214) There's no plce like home? An explortion of the mechnisms behind plnt litter- decomposer ffinity in terrestril ecosystems. New Phytol 24: Ayres E, Steltzer H, Simmons BL, Simpson RT et l (29) Home-field dvntge ccelertes lef litter decomposition in forests. Soil Biol Biochem 41:66 61 Ayres E, Steltzer H, Berg S et l (29b) Soil biot ccelerte decomposition in high-elevtion forests by specilizing in the brekdown of litter produced by the plnt species bove them. J Ecol 97: Brdgett RD (25) The biology of soil: community nd ecosystem pproch. Oxford University Press, New York Brdgett RD, Whittker JB, Frnklnd JC (1993) The diet nd food preferences of Onychiurus procmptus (Collembol) from uplnd grsslnd soils. Biol Fertil Soils 16: Berg BR, McClugherty C (214) Plnt litter: decomposition, humus formtion, crbon sequestrtion, 3rd edn. Springer, Berlin Bolger AM, Lohse M, Usdel B (214) Trimmomtic: flexible trimmer for Illumin sequence dt. Bioinformtics 3: Brdford MA, Tordoff GM, Eggers T et l (22) Microbiot, fun, nd mesh size interctions in litter decomposition. Oikos 99: Cebrin J (1999) Ptterns in the fte of production in plnt communities. Am Nt 154: Chomel M, Guittonny-Lrcheveque M, DesRochers A et l (215) Home field dvntge of litter decomposition in pure nd mixed plnttions under borel climte. Ecosystems 18:

13 Chomel M, Guittonny-Lrcheveque M, Fernndez C et l (216) Plnt secondry metbolites: key driver of litter decomposition nd soil nutrient cycling. J Ecol 14: Coq S, Souquet JM, Meudec E et l (21) Interspecific vrition in lef litter tnnins drives decomposition in tropicl rin forest of French Guin. Ecology 91: Crowther TW, Boddy L, Jones TH (212) Functionl nd ecologicl consequences of sprotrophic fungus-grzer interctions. ISME J 6: de Grff MA, Clssen AT, Cstro HF et l (21) Lbile soil crbon inputs medite the soil microbil community composition nd plnt residue decomposition rtes. New Phytol 188: Delgdo-Bquerizo M, Girmid L, Reich PB et l (216) Lck of functionl redundncy in the reltionship between microbil diversity nd ecosystem functioning. J Ecol 14: Durll DM, Todd AW, Trppe JM (1994) Decomposition of C-14- lbeled substrtes by ectomycorrhizl fungi in ssocition with Dougls-fir. New Phytol 127: Edgr RC (213) UPARSE: highly ccurte OTU sequences from microbil mplicon reds. Nt Methods 1: Fnin N, Fromin N, Bertrnd I (216) Functionl bredth nd home-field dvntge generte functionl differences mong soil microbil decomposers. Ecology 97: Fujii S, Mkit N, Mori AS et l (216) Plnt species control nd soil funl involvement in the processes of bove- nd below-ground litter decomposition. Oikos 125: Grci-Plcios P, Mestre FT, Kttge J et l (213) Climte nd litter qulity differently modulte the effects of soil fun on litter decomposition cross biomes. Ecol Lett 16: Gessner MO, Swn CM, Dng CK et l (21) Diversity meets decomposition. Trends Ecol Evol 25: Gholz HL, Wedin DA, Smithermn SM et l (2) Long-term dynmics of pine nd hrdwood litter in contrsting environments: towrd globl model of decomposition. Glob Chng Biol 6: Giesselmnn UC, Mrtins KG, Brndle M et l (211) Lck of home-field dvntge in the decomposition of lef litter in the Atlntic rinforest of Brzil. Appl Soil Ecol 49:5 1 Gonzlez G, Sestedt TR (21) Soil fun nd plnt litter decomposition in tropicl nd sublpine forests. Ecology 82: Grç MAS, Bärlocher F, Gessner MO (25) Methods to study litter decomposition: prcticl guide. In: Springer. Dordrecht, New York Hnson CA, Allison SD, Brdford MA et l (28) Fungl tx trget different crbon sources in forest soil. Ecosystems 11: Hättenschwiler S, Vitousek PM (2) The role of polyphenols in terrestril ecosystem nutrient cycling. Trends Ecol Evol 15: Hunt HW, Colemn DC, Inghm ER et l (1987) The detritl food web in shortgrss pririe. Biol Fertil Soils 3:57 68 Jones JB (21) Lbortory guide for conducting soil tests nd plnt nlysis. CRC Press, Boc Rton Keiser AD, Keiser DA, Stricklnd MS et l (214) Disentngling the mechnisms underlying functionl differences mong decomposer communities. J Ecol 12:63 69 Keiser AD, Stricklnd MS, Fierer N et l (211) The effect of resource history on the functioning of soil microbil communities is mintined cross time. Biogeosciences 8: Liw A, Wiener M (22) Clssifiction nd regression by rndomforest. R News 2:18 22 Lin DM, Anderson-Teixeir KJ, Li JS et l (216) Trits of dominnt tree species predict locl scle vrition in forest boveground nd topsoil crbon stocks. Plnt Soil 49: Mgoc T, Slzberg SL (211) FLASH: fst length djustment of short reds to improve genome ssemblies. Bioinformtics 27: Mkkr HPS (23) Quntifiction of tnnins in tree nd shrub folige: lbortory mnul. In: Kluwer cdemic publishers. Dordrecht, Boston McGuire KL, Bent E, Bornemn J et l (21) Functionl diversity in resource use by fungi. Ecology 91: McGuire KL, Pyne SG, Plmer MI et l (213) Digging the New York city skyline: soil fungl communities in green roofs nd city prks. PLoS One 8:e582 Meentemeyer V (1978) Mcroclimte nd lignin control of litter decomposition rtes. Ecology 59: Melillo JM, Aber JD, Murtore JF (1982) Nitrogen nd lignin control of hrdwood lef litter decomposition dynmics. Ecology 63: Milcu A, Mnning P (211) All size clsses of soil fun nd litter qulity control the ccelertion of litter decy in its home environment. Oikos 12: Moore TR, Trofymow JA, Tylor B et l (1999) Litter decomposition rtes in Cndin forests. Glob Chng Biol 5:75 82 Newell K (1984) Interction between two decomposer bsidiomycetes nd collemboln under Sitk spruce: grzing nd its potentil effects on fungl distribution nd litter decomposition. Soil Biol Biochem 16: Oksnen J, Blnchet FG, Friendly M et l (216) Vegn: community ecology pckge. R Pckge Version 2:4 1 Pterson E, Osler G, Dwson LA et l (28) Lbile nd reclcitrnt plnt frctions re utilised by distinct microbil communities in soil: independent of the presence of roots nd mycorrhizl fungi. Soil Biol Biochem 4: Perez G, Aubert M, Decens T et l (213) Home-field dvntge: mtter of interction between litter biochemistry nd decomposer biot. Soil Biol Biochem 67: R Development Core Tem (216) R: A lnguge nd environment for sttisticl computing. R Foundtion for Sttisticl Computing, Vienn, Austri. URL Rousk J, Bth E, Brookes PC et l (21) Soil bcteril nd fungl communities cross ph grdient in n rble soil. ISME J 4: Rousk J, Frey SD (215) Revisiting the hypothesis tht fungl-tobcteril dominnce chrcterizes turnover of soil orgnic mtter nd nutrients. Ecol Monogr 85: SAS Institute (21) SAS for Windows, version 9.3. SAS Institute, Cry, North Crolin. USA Stricklnd MS, Luber C, Fierer N et l (29) Testing the functionl significnce of microbil community composition. Ecology 9: St John MG, Orwin KH, Dickie IA (211) No 'home' versus 'wy' effects of decomposition found in grsslnd-forest reciprocl litter trnsplnt study. Soil Biol Biochem 43:

14 Swift MJ, Hel OW, Anderson JM (1979) Decomposition in terrestril ecosystems. University of Cliforni Press, Berkeley Veen GF, Sundqvist MK, Wrdle DA (215) Environmentl fctors nd trits tht drive plnt litter decomposition do not determine home-field dvntge effects. Funct Ecol 29: Vivnco L, Austin AT (28) Tree species identity lters forest litter decomposition through long-term plnt nd soil interctions in Ptgoni, Argentin. J Ecol 96: Voriskov J, Bldrin P (213) Fungl community on decomposing lef litter undergoes rpid successionl chnges. ISME J 7: Wll DH, Brdford MA, St John MG et l (28) Globl decomposition experiment shows soil niml impcts on decomposition re climte-dependent. Glob Chng Biol 14: Zhng DQ, Hui DF, Luo YQ et l (28) Rtes of litter decomposition in terrestril ecosystems: globl ptterns nd controlling fctors. J Plnt Ecol 1:85 93 Zhou JZ, Deng Y, Shen LN et l (216) Temperture medites continentl-scle diversity of microbes in forest soils. Nt Commun 7:1283