Submitted 2 March 2018Accepted 29 May 2018Published 6 July 2018

Corresponding authorRebecca E Hewittrebeccahewittnauedu

Academic editorXavier Le Roux

Additional Information andDeclarations can be found onpage 17

DOI 107717peerj5008

Copyright2018 Hewitt et al

Distributed underCreative Commons CC-BY 40

OPEN ACCESS

Variable retention harvesting influencesbelowground plant-fungal interactions ofNothofagus pumilio seedlings in forests ofsouthern PatagoniaRebecca E Hewitt12 Donald Lee Taylor23 Teresa N Hollingsworth4Christopher B Anderson56 and Guillermo Martiacutenez Pastur5

1Center for Ecosystem Science and Society Northern Arizona University Flagstaff AZUnited States of America

2 Institute of Arctic Biology University of AlaskamdashFairbanks Fairbanks AK United States of America3Department of Biology University of New Mexico Albuquerque NM United States of America4Pacific Northwest Research Station Boreal Ecology Cooperative Research Unit US Forest Service FairbanksAK United States of America

5Centro Austral de Investigaciones Cientiacuteficas (CADIC) Consejo Nacional de Investigaciones Cientiacuteficas yTeacutecnicas (CONICET) Ushuaia Tierra del Fuego Argentina

6 Instituto de Ciencias Polares Ambiente y Recursos Naturales (ICPA) Universidad Nacional de Tierra del Fuego(UNTDF) Ushuaia Tierra del Fuego Argentina

ABSTRACTBackground The post-harvest recovery and sustained productivity of Nothofaguspumilio forests in Tierra del Fuego may be affected by the abundance and compositionof ectomycorrhizal fungi (EMF) Timber harvesting alters EMF community structurein many managed forests but the impacts of harvesting can vary with the managementstrategy The implementation of variable retention (VR) management can maintainincrease or decrease the diversity of many species but the effects of VR on EMF in theforests of southern Patagonia have not been studied nor has the role of EMF in theregeneration process of these forestsMethods We evaluated the effects of VR management on the EMF communityassociated with N pumilio seedlings We quantified the abundance composition anddiversity of EMF across aggregate (AR) and dispersed (DR) retention sites within VRmanaged areas and compared them to primary forest (PF) unmanaged stands EMFassemblage and taxonomic identities were determined by ITS-rDNA sequencing ofindividual root tips sampled from 280 seedlings across three landscape replicates Tobetter understand seedling performance we tested the relationships between EMFcolonization EMF taxonomic composition seedling biomass and VR treatmentResults The majority of EMF taxa were Basidiomycota belonging to the familiesCortinariaceae (n= 29) Inocybaceae (n= 16) and Thelephoraceae (n= 8) whichwas in agreement with other studies of EMF diversity in Nothofagus forests EMFrichness and colonization was reduced in DR compared to AR and PF FurthermoreEMF community composition was similar between AR and PF but differed fromthe composition in DR EMF community composition was correlated with seedlingbiomass and soil moisture The presence of Peziza depressa was associated withhigher seedling biomass and greater soil moisture while Inocybe fibrillosibrunnea andCortinarius amoenus were associated with reduced seedling biomass and lower soil

How to cite this article Hewitt et al (2018) Variable retention harvesting influences belowground plant-fungal interactions of Nothofa-gus pumilio seedlings in forests of southern Patagonia PeerJ 6e5008 DOI 107717peerj5008

moisture Seedling biomass was more strongly related to retention type than EMFcolonization richness or compositionDiscussion Our results demonstrate reduced EMF attributes and altered compositionin VR treatments relative to PF stands with stronger impacts in DR compared toAR This suggests that VR has the potential to improve the conservation status ofmanaged stands by supporting native EMF in AR Our results also demonstrate thecomplex linkages between retention treatments fungal community composition andtree growth at individual and stand scales

Subjects Biodiversity Ecology Mycology ForestryKeywords Lenga Dispersed retention Ectomycorrhizal fungi Tierra del fuegoForest sustainability Recruitment Silviculture Aggregate retention

INTRODUCTIONSouthern Patagonia is globally recognized for the conservation value of its extensive forests(Olson amp Dinerstein 2002Mittermeier et al 2003) These forests also constitute regionallyimportant economic and cultural resources as working timberlands and rangelands(Peri et al 2016a Martiacutenez Pastur et al 2017) During early European colonizationPatagonian forests were subjected to extensive clearing burning and grazing but in thelast several decades new silviculture techniques have been implemented (Gea-Izquierdo etal 2004) that attempt to achieve greater forest sustainability (Gamondegraves Moyano Morganamp Martiacutenez Pastur 2016) Variable retention (VR) management has been adapted fromtimber practices in the Northern Hemisphere (Franklin et al 1997) and implemented insouthern Patagonia as a way to reconcile the goals of preserving forest biodiversity andproduction values (Soler et al 2015 Soler et al 2016) The VR management strategy inour study system in Tierra del Fuego leaves 30 of a given managed area as intact patchesof primary or secondary forest called lsquolsquoaggregate retentionrsquorsquo (AR) These patches are locatedwithin a harvested area called lsquolsquodispersed retentionrsquorsquo (DR) where 20 of the basal arearemains standing to serve as seed trees regeneration shelter or for conservation purposes

Nothofagus pumilio (Popp et Endl) Krasser (commonly named lenga) is one of thedominant canopy tree species and the most important timber species in the deciduoustemperate forests of the Magellanic subpolar ecoregion (Martiacutenez Pastur et al 2000) Treeseed banks do not persist in the soils of N pumilio forests (Cuevas amp Arroyo 1999) andthus seedbed and environmental conditions at the time of dispersal and germination arekey determinants of forest regeneration after timber harvesting (Martiacutenez Pastur et al2011) Within a VR management system seed dispersal and germination do not limitforest regeneration due to close proximity to seed sources (less than twofold the dominanttree height) and adequate seedbed conditions regardless of the degree of canopy coverremaining after harvest (Martiacutenez Pastur et al 2014 Torres et al 2015) Instead lengaestablishment and performance is influenced by environmental modifications resultingfrom forest harvesting such as changes in soil moisture and light levels (Martiacutenez Pastur et

Hewitt et al (2018) PeerJ DOI 107717peerj5008 225

al 2007 Martiacutenez Pastur et al 2011) and potentially biotic factors such as availability ofcompatible symbiotic fungi

Like other Nothofagus species lenga relies upon an obligate symbiosis withectomycorrhizal fungi (EMF) (Smith amp Read 2008) In general EMF colonize the fineroot systems of host plants improving the plantrsquos ability to access soil resources includingwater and nutrients and in turn receive carbon in the form of photosynthate (Smith ampRead 2008) This symbiosis is considered a mutualism yet at the seedling phase it canhave either beneficial or detrimental effects on seedling performance depending on thetaxon-specific carbon costs and nutrient benefits conferred to the seedling by themycobiont(Jones amp Smith 2004 Koide et al 2008 Van der Heijden amp Horton 2009 Hoeksema et al2010) Few ecological studies of fungal diversity have been carried out in Patagonia (butsee Truong et al 2017) and even less is known about root-associated EMF dynamics inPatagonian forests (but see Palfner Canseco amp Casanova-Katny 2008 Nouhra et al 2013)however other understudied Southern Hemisphere forests have EMF host trees includingNothofagus (Orlovich amp Cairney 2004 Tedersoo et al 2008 Dickie Richardson amp Wiser2009 Tedersoo et al 2009 Horton et al 2013) and collectively these studies inform ourinterpretation of diversity patterns and plant-fungal interactions in Tierra del Fuego

In managed forests around the world successful seedling establishment has been linkedto adequate EMF inoculum fungal species assemblage and diversity (Amaranthus ampPerry 1987 Perry Molina amp Amaranthus 1987 Perry et al 1989) Timber harvesting canalter EMF community structure decrease EMF inoculum potential and reduce fine rootcolonization levels (Parke Linderman amp Trappe 1984 Borchers amp Perry 1990 Hagermanet al 1999 Jones Durall amp Cairney 2003) These findings are particularly evident afterclear-cutting where host plants have been killed and the survival of EMF species on maturetrees is minimal and therefore limited in the timeframe of natural or out-planted seedlingrecruitment (Dahlberg et al 2001 Simard 2009) However with VR management ARpatches may act as islands of ecological integrity where diversity structure and functionare preserved In northern forested ecosystems both AR (Kranabetter De Montigny ampRoss 2013) and individual seed trees can maintain EMF colonization and diversity afterharvest (Outerbridge amp Trofymow 2004 Cline Ammirati amp Edmonds 2005) with positiveeffects on seedling performance (Cline Vinyard amp Edmonds 2007) Yet other studieshave shown that seed trees alone fail to maintain EMF diversity for upwards of severaldecades after harvesting (Varenius Lindahl amp Dahlberg 2017) In comparison to EMFstudies from the Northern Hemisphere relatively little is known about post-harvest EMFdynamics in Southern Hemisphere forests In New Zealand ten years after harvest DickieRichardson amp Wiser (2009) reported reduced occurrence of EMF roots and altered EMFcomposition in Nothofagus forests Together with post-harvest observations from theNorthern Hemisphere these findings suggest that there may be strong effects of harvestingdisturbance on EMF-seedling dynamics In this context EMF-seedling dynamics inmanaged Patagonian forested ecosystems could be crucial for understanding the post-harvest recovery and sustainability of forestry practices

In this study we sought to (i) understand how belowground mycobiont diversity andcomposition was impacted by VR management treatments (AR and DR) in comparison

Hewitt et al (2018) PeerJ DOI 107717peerj5008 325

to primary forest (PF) stands (ii) determine whether variation in lenga biomass wascorrelated with EMF factors ( colonization structure and occurrence of specific taxa)and environmental factors (soil moisture) that are known to be influenced by harvesting(iii) evaluate whether our results support the concept that AR patches act as refugia forEMF thus contributing to biodiversity conservation in southern Patagonian forests and(iv) compare the taxonomic diversity observed in our study to EMF reported in otherSouthern Hemisphere forests We tested four hypotheses (i) seedling root colonizationwill decline from PF to AR to DR (ii) fungal communities will be more similar between PFand AR than PF and DR (iii) fungal composition will be related to seedling biomass withlarger seedlings associated with taxa in PF stands and (iv) dominant EMF taxa observedin association with N Pumillo seedlings will be similar to those found in other SouthernHemisphere forests This study builds on an extensive body of research investigatingpost-harvest dynamics in Northern Hemisphere forests and extends this research to theSouthernHemisphere by providing a first investigation of these dynamics in a VR-managedstands in Patagonia As such these findings are expected to contribute to our understandingof EMF diversity patterns in an underexplored ecosystem and provide further evidence ofthe impacts of VR harvesting on biodiversity in Tierra del Fuegorsquos Nothofagus forests

MATERIALS amp METHODSStudy siteThis study was carried out in the Argentine portion of Tierra del Fuego Island whereworking timberlands constitute ca 220000 ha of monospecific lenga stands (Collado2001) We studied three old-growth N pumilio stands and adjacent areas of VR timbermanagement (with AR and DR treatments) at Los Cerros Ranch (5418primeS 6749primeW)Primary forest stands are 5ndash10 ha with uneven age structure and older trees aged up to 400years old The PF stands presented a closed canopy with a sparse understory of relatively fewwoody and herbaceous plants (Lencinas et al 2008) Managed stands with the VR systeminclude areas of AR intact forest lsquolsquoislandsrsquorsquo (60 m radius one per hectare) surroundedby DR In DR areas evenly dispersed seed trees maintain a basal area of dominant trees at10ndash15 m2 haminus1 (Martiacutenez Pastur et al 2009) These sites are part of the Parcelas de Ecologiacuteay Biodiversidad de Ambientes Naturales en Patagonia Austral (PEBANPA) a long-termmonitoring network in southern Patagonia (Peri et al 2016b) and have been monitoredfor seedling recruitment dynamics since 2004

Seedling sampling and regeneration measurementsTo characterize seedling EMF communities across timber management treatments inrelation to seedling biomass we sampled seedlings in three replicate sites of each PF ARand DR treatment Logging occurred in the growing season of 2004ndash2005 and in 2012 wefocused our sampling on young seedlings most of which were under seven years old andestablished after logging At each site we sampled 10 lenga seedlings along three randomlypositioned 50 m transects in close proximity (10ndash50 m) to permanent PEBANPA networkregeneration plots collecting one seedling approximately every five meters for a total of 30seedlings per site and 90 seedlings per treatment In all sites our sampling did not account

Hewitt et al (2018) PeerJ DOI 107717peerj5008 425

for distance to the rooting zone of lenga trees and therefore represents the full variation ofEMF-seedling dynamics At one of PF sites we sampled an additional transect In total wesampled 280 seedlings across 28 transects and nine sites

Harvested seedlings were transported back to the Agroforestry Lab at the Centro Australde Investigaciones Cientiacuteficas (CADIC-CONICET) in Ushuaia Argentina Shoots wereseparated from roots Aboveground biomass was dried at 70 C for 48 h and weighedThe age of the seedling was verified by counting leaf whirls (Martiacutenez Pastur et al 2011)Thirteen of 280 seedlings established before 2005 eg prior to timber harvest across allthree treatments (three in AR seven in DR and three in PF) We chose not to exclude theseseedlings because they were relatively well distributed across treatments and comparativebi- and multi-variate analyses indicated that they did not impact statistical results

We obtained lenga regeneration metrics including sapling (gt1 year-old plants) andseedling (lt1 year old plants) density in the permanent plots of the PEBANPA networkat our sites (Peri et al 2016b) Survey methods for lenga regeneration are fully describedin Martiacutenez Pastur et al (2011) In brief seedling and sapling density was surveyed in18 regeneration plots of 1 m2 (500 times 20 cm) in each treatment (PF AR DR) We alsomeasured volumetric soil water content (VSW) in the upper 10 cm of soil using a MP406moisture probe (ICT Armidale Australia) at the beginning and end of the summer

EMF root colonizationThe root system of each seedling was rinsed gently to remove soil and then was cut into5 cm segments Root segments were dispersed in water in a gridded Petri dish under adissecting microscope at 10ndash40times magnification We used the gridline intercept method(Brundrett et al 1996) to count the proportion of fine roots colonized by EMF in relationto the total number of root-gridline intersections We randomly selected five healthyEMF-colonized root tips per seedling and preserved them in RNAlater (Ambion IncAustin Texas USA)

Molecular analysis of EMF compositionEMF root tips were rinsed with distilled water to remove excess RNAlater and then werelyophilized Root tips were ground in lysis buffer with a motorized sterile pestle (KontesRockwood TN USA) DNA was extracted from each root tip using the DNEasy Plant MiniKit (QIAGEN Inc Valencia CA USA) according to the manufacturerrsquos instructions withminormodifications suggested in Bent amp Taylor (2010) Genomic DNAwas amplified in 25microl PCR reactions containing 25 mMMgCl2 10 mM dNTPs 50 microM forward primer ITS1-F(CTTGGTCATTTAGAGGAAGTAA (Gardes amp Bruns 1993)) 50 microM reverse primer ITS4(TCCTCCGCTTATTGATATGC (White et al 1990)) 10 mgml bovine serum albuminPromega Go-Taq (Sigma-Aldrich St Louis MO USA) 5X Promega Green Taq bufferand 5 microl of genomic DNA Reaction mixes were prepared in 02 ml tubes and thermocycledin an MJ Research PTC-225 thermal cycler as follows 96 C for 3 min 35 cycles of 94 Cfor 30 s 52 C for 30 s then 72 C for 3 min followed by 72 C for 10 min PCR productswere gel checked and Sanger sequenced by Functional Biosciences (Madison WI)

Hewitt et al (2018) PeerJ DOI 107717peerj5008 525

BioinformaticsSanger sequences were trimmed quality filtered and grouped into operational taxonomicunits (OTUs) synonymous with EMF taxa as follows Initial end-trimming (error ratebelow 01 in a 25 base window for both ends) and assembly of paired end reads (whereavailable) were carried out in Codoncode Aligner 403 Sequences with fewer than 200 highquality base calls (phred gt 20) were deleted and all remaining sequences were exported infastq format Next a more stringent end-trimming was carried out using FASTQ QualityTrim program in Galaxy (Goecks Nekrutenko amp Taylor 2010) with window sizes of 5 20and 40 and a phred threshold of 20 Sequences were converted to fasta format inGalaxy thenaligned using Muscle (Edgar 2004) in AliView (Larsson 2014) to identify and manuallyremove primer sequences We then used QIIME 19 (Caporaso et al 2010) to clustersequences into OTUs via open-reference with the USEARCHmethod (Edgar 2010) a 95sequence identity threshold and theUNITE sh_refs_qiime_ver7_97_s_20112016 database(Kotildeljalg et al 2013) Singletons were kept at this step (min_otu_size 1) Representativesequences for each OTU were submitted to the RDP Naiumlve Bayesian classifier using theWarcup 2 ITS training set to estimate OTU taxonomies (Deshpande et al 2016) We alsosearched GenBank using discontiguousmega-BLAST with uncultured sequences excludedto confirm and in some cases improve identifications produced by RDP OTU trophicguilds were inferred based on investigator knowledge and literature searches All sequencesare deposited in GenBank under accession numbers MH019772ndashMH019959

Following the taxonomic identification of EMF OTUs were eliminated if they weresaprotrophs yeasts wood decay fungi pathogens or if identification was uncertain Forfurther ordination analysis singletons (OTUs occurring once) were also eliminated toreduce the disproportionate influence of extremely rare taxa (McCune amp Grace 2001)resulting in a matrix of 49 OTUs times 228 seedlings

Data analysisWe conducted comparisons for stand metrics including basal area soil moisture andthe regeneration metrics including seedling and sapling density for each VR treatment(Table 1) These comparisons were tested using one-way ANOVA and further pairwisecomparisons using the Tukey test of honest significant difference (plt 005) All statisticalanalyses were performed in R (R Core Team 2016) with the exception of EMF ordinationanalysis and rarefaction curves described below The seedling biomass and colonizationdataset and ectomycorrhizal community composition dataset are archived with BonanzaCreek LTER doi 106073pasta350b701beb4ac2ebd2ef3f29893e6ed5 httpwwwlteruafedudatadata-detailid687 doi 106073pastadf801ec9b7f3493b6eb3bf191124e41fhttpwwwlteruafedudatadata-detailid686 Seedling recruitment and density data areavailable on PEBANPA databases through CADIC-CONICET Ushuaia

Analysis of EMF root colonizationTo compare mean root colonization for each retention type we used the KruskalndashWallistest with a post hoc Dunn Test of multiple comparisons with the Bonferroni correctionwith the R package dunntest (Dinno 2017) We used a non-parametric test because EMF

Hewitt et al (2018) PeerJ DOI 107717peerj5008 625

Table 1 Site and EMF variables for primary forest and variable retention timber treatments (aggregateand dispersed retention)Data included show means followed by SE

Primary forest Aggregate retention Dispersed retention

Mean basal area (m2ha) 9489plusmn 217a 7194plusmn 255b 85plusmn 126c

Mean soil moisture (watersoil) 2130plusmn 179a 1942plusmn 160a 3448plusmn 228b

Mean seedling density (thousandha) 203plusmn 107a 134plusmn 83a 18plusmn 8b

Mean sapling density (thousandha) 157plusmn 48a 251plusmn 58a 86plusmn 10b

Site EMF α diversity 2767plusmn 122a 2700plusmn 153a 1567plusmn 285b

Seedling EMF α diversity 173plusmn 008a 143plusmn 007b 115plusmn 008c

Mean colonization () 972plusmn 001ab 990plusmn 001a 95 5plusmn 001b

Notespost hoc Tukey HSD at plt 005post hoc Dunn Test at plt 005a b c indicate significant differences detected by post hoc tests

colonization levels were strongly skewed due the large number of root systems colonizedat high levels (mean colonization across treatments (972 plusmn 001 SE)) and regardlessof data transformation the data did not meet the assumption of normality

Analysis of EMF community structureFor EMF taxa richness and evenness we counted the EMF taxa occurring across all seedlingswithin each treatment to compute a total richness estimation for PF AR and DR areas Atthe treatment level we also summed the number of species shared between each two-wayand the full three-way comparison of treatments To compute EMF α diversity at theseedling level we counted the number of EMF taxa associated with each seedling fromour sequence dataset We tested for significant treatment differences in seedling-level αdiversity using one-way ANOVA and tested for differences between two-way comparisonsusing the Tukey test of honest significant difference (plt 005) We used the same modelstructure to test for treatment differences in α diversity at the site level Furthermorewe computed sample-based abundance rarefaction curves for each treatment to assessvariation in richness across treatments Diversity statistics for rarefaction curves werecalculated using EstimateS (Colwell 2013) where we used 100 randomization runs andrandomized the samples without replacement

To evaluate patterns in seedling- and transect-level EMF composition (Kruskal ampWish 1978McCune amp Mefford 2011) we performed Nonmetric Multidimensional Scaling(NMDS) ordinations We used the Sorensen (BrayndashCurtis) distance measure to representcompositional dissimilarity We identified the solution with the lowest stress relative tothe number of axes (McCune amp Grace 2002) Finally we calculated a post hoc proportionof variance represented by each axis by calculating the squared Pearson correlation (r2)between distances in the original distance metric and pair-wise distances of objects in theordination space (McCune amp Grace 2002)

Preliminary analysis of composition associated with individual seedlings suggested a1-dimensional solution which would be highly unstable and not significantly different thanrandom We therefore excluded all OTUs that occurred on less than four seedlings Thisresulted in a final matrix of 26 OTUs times 206 seedlings Our final matrix was transformed

Hewitt et al (2018) PeerJ DOI 107717peerj5008 725

using Bealrsquos smoothing to relieve the lsquolsquozero truncation problemrsquorsquo (Beals 1984) whichis common with heterogeneous data with a large number of zeros We exported threeordination axes and used those as representations of EMF composition in Random Forestand liner mixed-effects model analyses described below (Fig S1)

We combined EMF composition on individual seedlings across each transect forordination analysis of EMF composition in relation to environmental factors Pooling EMFcomposition data at the transect level allowed us to evaluate variation in compositionwithinand across sites We correlated this main OTU matrix with a secondary environmentalmatrix and expressed high correlations (r ge 020) with a biplot overlay to indicate thedirection andmagnitude of significantly correlated environmental variables The secondarymatrix was made up of seven environmental variables treatment (PF AR DR) site (n= 9)volumetric soil water content average seedling biomass (g) average root colonizationand average age of seedlings across the transect We tested whether EMF compositionvaried by VR treatments using Multiple-Response Permutation Procedure (MRPP) (BerryKvamme amp Mielke 1983) and tested whether the abundance of any taxon was an indicatorof the VR treatments with Indicator Species Analysis (Dufrene amp Legendre 1997) Allordination analyses of EMF communities were performed in PC-ORD 60 (MJM SoftwareDesign Gleneden Beach OR USA)

Analysis of relationships between seedling biomass and EMF metricsTo evaluate the importance of retention treatments seedling age and EMF variables(colonization richness and composition) in predicting seedling biomass we used RandomForest (RF) regression trees in the randomForest package in R (Liaw ampWiener 2015) RFis an ensemble decision tree method that uses an algorithmic approach to make predictionsbased on the input variables (Breiman 2001b) The RF approach allowed us to includeall our measured variables including untransformed root colonization and biomassdata because RF can accurately predict variable response with data that is not normallydistributed We used RF to calculate variable importance scores for treatment seedlingand fungal variables when predicting seedling biomass We built 500 regression trees usingrandom samples of the 206 observations At each node in the tree two predictor variableswere chosen at random from the seven explanatory variables management treatmentseedling age and EMF composition represented by three NMDS axes EMF richnessand root colonization For each predictor RF computes the prediction error (meansquare error) The difference between the classification of observed bootstrap data andthe classification of permuted data is averaged across all trees and divided by the standarderror (Breiman 2001a Cutler et al 2007)

We also created linear mixed-effects models using the nlme package (Pinheiro et al2018) to explore the fixed-effects of timber management treatment seedling age EMFcomposition represented by three NMDS axes and EMF richness on seedling biomass withseedling nested in transect and site as a random factor We evaluated data distributions andcorrelations between all predictor variables (habitat age root colonization EMF richnessAxis1 Axis2 and Axis3) with a threshold spearman correlation of 04 None were excludedbased on this criterion Seedling biomass was log-transformed

Hewitt et al (2018) PeerJ DOI 107717peerj5008 825

To further evaluate whether the presence of specific EMF taxa affected seedling biomasswe correlated the transect-level species matrix with itself yielding a correlation matrix ofOTUs that strongly correlated (gt020) with each axis (Hollingsworth et al 2013) We thenused a simple ANOVA to evaluate differences in seedling biomass for seedlings that werecolonized (yesno) by each of these strongly correlated taxa

RESULTSVR impacts on seedling recruitmentForest structure significantly varied among treatments Basal area was greatest in PFfollowed by AR and then DR sites (F = 24814 plt 001 Table 1) due to the removalof trees during timber harvesting Harvesting treatment greatly influenced soil moisture(F = 1566 plt 001) Dispersed retention sites had greater soil moisture compared to PFand AR sites (Table 1) We found marginal differences in seedling density by treatment(F = 273 p= 008 Table 1) Primary forest and AR sites had higher seedling densitiescompared to DR sites Seedling densities ranged from 0 to 166 thousand haminus1 in DR sites0 to 786 thousand haminus1 in AR and 0 to 980 thousand haminus1 in PF sites Sapling densityfollowed the same trend but with significant differences among treatments (F = 627plt 001 Table 1) Sapling densities ranged from 35 to 203 thousand haminus1 in DR sites78ndash620 thousand haminus1 in AR and 6ndash477 thousand haminus1 in PF sites

VR impacts on EMF root colonizationOn average colonization rates were high (972 plusmn 001 SE) across all treatments EMFcolonization varied from 38 to 100 among seedlings Colonization levels varied withtimber management treatment (KruskalndashWallis χ2

= 1163 plt 001) On average AR siteshad the highest mean colonization which was significantly higher than colonization levelsin the DR sites but similar to those in PF sites (Table 1)

EMF taxonomic diversityOur sequencing effort yielded 188 OTUs 89 of which were EMF or likely EMF (TablesS1 and S2) The most abundant OTUs in the sequencing dataset were taxa in the EMFand likely EMF trophic niche with 17 EMF taxa occurring in the Phylum Ascomycota(SubphylumPezizomycotina) and 72 in the Basidiomycota (SubphylumAgaricomycotina)All of the fungi in the Basidiomycota were in class Agaricomycetes and belonged to theorders Agaricales Boletales Cantharellales Russulales Sebacinales and ThelephoralesThe families Cortinariaceae (n= 29) Inocybaceae (n= 16) and Thelephoraceae (n= 8)were the most species rich Despite the Basidiomycota showing higher EMF taxonomicdiversity the most abundant OTUs were ascomycetes (Table S1)

VR impacts on EMF taxa richness and occurrenceSummaries of our sequence data indicate that of the EMF taxa 52 occurred in PF comparedto 53 in AR and only 30 in DR Twenty-eight taxa observed in PF stands also occurred inAR while only 13 PF taxa occurred in DR Seedlings from AR and DR areas shared 15 taxaOverall 10 of 89 taxa occurred in all treatments (Table S2) Despite some of the dominant

Hewitt et al (2018) PeerJ DOI 107717peerj5008 925

0

20

40

60

0 25 50 75

Seedlings

Mao

Tau

ric

hnes

s es

timat

e

Figure 1 EMF species richness for both variable retention treatments and primary forest Values areMao Tau abundance based richness estimates with 95 CI (Colwell Mao amp Chang 2004) Primary for-est black aggregate retention blue dispersed retention orange The rarefaction curves indicate that rich-ness was reduced in dispersed retention compared to aggregate retention and primary forest stands

Full-size DOI 107717peerj5008fig-1

taxa occurring across all VR timber treatments the abundances of many of these dominantsvaried with VR treatment For example the most abundant taxon OTU 1 Peziza depressaoccurred primarily in the DR areas Conversely some dominant Cortinarius Tomentellaand Inocybe species had low abundances or were not observed in the DR sites (Table S2)

At the seedling level EMF richness was low ranging from one taxon to three taxa perseedling (Table 1) These richness estimates include EMF that were observed once (egsingletons) Seedlings in PF stands had significantly greater EMF richness than seedlingsin AR and DR and seedlings in AR showed greater richness than seedlings in DR areas(F = 1500 plt 0001 Table 1) These patterns of α diversity shifted slightly when weassessed the richness at the site level Here again richness varied by treatment (F = 1116p= 001) with PF and AR sites showing greater richness than DR sites (Table 1) but nosignificant difference in richness was detected between AR and PF Richness ranged from25 to 29 taxa detected in PF sites 24ndash29 taxa in AR sites and 10ndash19 in DR sites This wassupported by the species rarefaction curves which indicated that EMF species richness waslower in DR sites (Fig 1) The shape of the curves also suggest that we did not saturateEMF diversity with our sampling in any of the treatment categories but we likely camecloser in the DR sites

Hewitt et al (2018) PeerJ DOI 107717peerj5008 1025

VR impacts on EMF compositionTo determine the effects on EMF composition of VR management treatments (AR andDR) in comparison to PF stands we used NMDS ordination to visualize the variation intaxon abundance and occurrence within and across our study sites Three major gradientscaptured sim75 of the variance in EMF composition across transects (Axis 1 = 215Axis 2 = 184 and Axis 3 = 353 of the variation) The NMDS ordination had afinal stress of 1445 with a final instability of 000046 after 278 iterations (Fig 2) Soilmoisture and seedling biomass were correlated with fungal composition along Axis 2(seedling biomass r2 = 047 soil moisture r2 = 045) Additionally there was a relativelystrong visual pattern of differentiation in composition between treatments (Fig 2) Thiswas confirmed by MRPP analysis (T =minus743 A= 008 plt 001) with the strongestdifferentiation between DR and AR (T =minus642 A= 008 plt 001) or PF (T =minus672A= 010 plt 001) There was no significant difference in species composition between PFand AR sites (T =minus157 A= 002 p= 007) Indicator species analysis showed that theabundance of OTU14 Timgrovea ferruginea was an indicator of DR (IV= 4270 p= 003)and OTU34 Cortinarius cycneus was an indicator of AR (IV = 5830 plt 001)

Relationships between seedling biomass and EMF metricsRandom Forest analysis of individual seedlings indicated that treatment (PF AR or DR)was the most important variable predicting seedling biomass followed by seedling ageEMF composition EMF richness and root colonization (Fig 3) Supporting theseresults the mixed-effects model analysis also indicated that seedling biomass was affectedby treatment (χ2

= 14473 plt 001) and seedling age (χ2 = 3307 plt 001) whileEMF composition and richness were not significantly important to explaining variationin seedling biomass (Table 2) The mixed-effects model accounted for 707 of thevariance explaining log-transformed seedling biomass (R2

marginal= 071 R2conditional= 076)

Mixed-effects model parameter estimates indicate that seedlings from the AR and DR siteshad greater biomass than seedlings harvested from the PF sites (PF 006 g plusmn 000 SE AR011 g plusmn 001 SE DR 056 g plusmn 004 SE Table 2) The mean age of seedlings in PF was430 years plusmn 012 SE slightly younger than in DR 458 years plusmn 016 SE and significantlyolder than in AR (376 yearsplusmn 016 SE) (F = 807 plt 001 where D-A diff 082 plt 001PF-A diff 054 p= 002 PF-D diff minus028 p= 036) Although seedling biomass was higherin the DR seedling and sapling density were lower in DR sites than AR or PF (Table 1)

Taxon-specific relationships between EMF and seedling biomassFive OTUs were correlated with Axis 2 (Axis 2 = 187 of 75 total variance explained)the same axis that was correlated with seedling biomass Peziza depressa Clavulina cirrhataCenococcum geophilum Inocybe fibrillosibrunnea and Cortinarius amoenus The presenceof P depressa and I fibrillosibrunneawere significantly related to seedling biomass while Camoenus was marginally related to seedling biomass (Table 3) The presence of P depressawas correlated with higher mean seedling biomass (Table 3) Peziza depressa was mostabundant in DR areas (Table S2) Conversely the presences of I fibrillosibrunnea andC amoenus were correlated to lower mean seedling biomass (Table 3) and these OTUsoccurred in the PF and AR but not the DR sites (Table S2)

Hewitt et al (2018) PeerJ DOI 107717peerj5008 1125

BiomassSoil moisture

Figure 2 NMDS ordination of EMF communities associated with each transect (10 seedlings per tran-sect) Symbols represent the two treatments in the variable retention timber management system and pri-mary forests stands Primary forest black triangles aggregate retention blue circles dispersed retentionorange squares Proportion of variance explained by each axis Axis 1= 213 Axis 2= 187 and Axis3= 352 The vectors indicate the magnitude and direction of the relationship The ordination indicatesthat differences in composition were greatest between dispersed retention and aggregate retention or pri-mary forest stands

Full-size DOI 107717peerj5008fig-2

Table 2 Linear mixed-effects model coefficients and associated SE explaining log- transformedseedling biomass The intercept or reference level is set as the log-transformed seedling biomass forseedlings occurring in primary forest Bolded p-values indicate parameters that were significanlty relatedto seedling biomass

Variable Value SE p-value

Primary forest minus155 008 lt001Aggregate retention 025 007 001Dispersed retention 081 007 lt001Seedling age 007 001 lt001Axis1 004 003 021Axis3 005 004 019Axis2 001 003 082Seedling EMF α diversity 002 003 040

Hewitt et al (2018) PeerJ DOI 107717peerj5008 1225

-001 0 001 002 003 004 005 006 007

Timber treatment

Seedling age

EMF composition Axis 1

EMF composition Axis 3

EMF composition Axis 2

EMF richness

Root colonization

Increase in MSE

Figure 3 Variable importance scores based on Random Forest regression trees explaining seedlingbiomass in relation to variable retention timber management treatment seedling and fungal variablesScores are derived from the increase in the mean square error ( increase MSE) when a variable is per-muted Axis1 2 and 3 were derived from NMDS ordination of EMF composition of each seedling Vari-able retention treatment and seedling age were more important predictors of seedling biomass than EMFvariables

Full-size DOI 107717peerj5008fig-3

Table 3 ANOVA comparison of the relationships between the presence of five EMF taxa onmean seedling biomass These five OTUs werehighly correlated with NMDS Axis 2 the same ordination axis correlated with seedling biomass Bolded OTUs indicate significant effects on seedlingbiomass

OTU Species Sum squares OTU absent OTU present df F pBiomassplusmn SE Biomassplusmn SE

1 Peziza depressa 492 020plusmn 002 041plusmn 005 1 2865 000151 Clavulina cirrhata 041 023plusmn 002 029plusmn 006 1 240 01219 Cenococcum geophilum 010 023plusmn 002 009plusmn 002 1 058 04520 Inocybe fibrillosibrunnea 080 023plusmn 002 006plusmn 001 1 466 00355 Cortinarius amoenus 065 024plusmn 002 007plusmn 001 1 381 005

DISCUSSIONForest management for high mycorrhizal inoculum potential provides an important aidin forest recovery after harvesting (Perry Molina amp Amaranthus 1987 Marx Marrs ampCordell 2002) We investigated how VR management impacted seedling-EMF interactionsafter timber harvest and whether it supported EMF dynamics similar to those found inPF stands In our study we observed a reduction in EMF root colonization and taxonomicrichness along with distinct fungal community composition inDR sites compared to PF andAR sites which shared similar EMF attributes Overall EMF composition was significantly

Hewitt et al (2018) PeerJ DOI 107717peerj5008 1325

correlated with soil moisture and seedling biomass however the effects of EMF metrics onseedling biomass were not as important as the management treatment itself In DR sitesseedlings had greater biomass and there was greater soil moisture Together these findingssuggest a stronger abiotic filter such as soil moisture on seedling performance than thebiotic filter of plant-fungal interactions Yet concurrent with reduced EMF metrics in DRsites we also observed reduced recruitment leading us to hypothesize that EMF dynamicsmay play a role in seedling establishment beyond what we detected by analyzing seedlingbiomass However future experimental studies are necessary to quantify any effects VRmay have on seedling recruitment due to EMF attributes Overall we observed that ARmore than DR treatments maintained EMF diversity in the VR managed forests that westudied

EMF composition in relation to other Southern Hemisphere studiesThe most abundant EMF taxa observed in our study were ascomyceteous fungi of the orderPezizales However the basidiomycetes were more speciose than the ascomycetes Withinthe Basidiomycota the most species-rich order and family were the Agaricales and theCortinariaceae respectively These patterns of taxonomic diversity mirror those in surveysof both root tips (Nouhra et al 2013) and fungal fruiting bodies in the Nothofagus forestsof South America (Garnica Weiszligamp Oberwinkler 2003 Truong et al 2017) BLAST resultsfor the five most abundant OTUs in our EMF dataset matched taxa observed in Chileand Argentina Some of these taxa were previously thought to only occur in Australasia(eg OTU1 top BLAST description Ruhlandiella sp (Truong et al 2017)) Interestinglythe dominant families (Cortinariaceae Inocybaceae and Thelephoraceae) observed inour root tip survey are also abundant mycobionts in Australian forests (Tedersoo et al2009 Horton et al 2017) and the most species-rich families observed at high latitudes inthe Northern Hemisphere (Timling et al 2012) Our sequence dataset supports recentlydescribed patterns of EMF diversity in South America (Truong et al 2017) and drawsattention again (Tedersoo et al 2008) to parallels between EMF taxonomic diversity at highlatitudes in both the Northern and Southern Hemispheres

VR impacts on EMFWe found that N pumilio seedling root systems were highly colonized by EMF (sim97across all the studied treatments) These findings are similar to reports that both youngand mature Nothofagus roots have high colonization levels of gt70 (Diehl Mazzarinoamp Fontenla 2008 Fernaacutendez et al 2015) and reinforce findings from the NorthernHemisphere where inoculum levels in logged areas remain high (Perry et al 1982 JonesDurall amp Cairney 2003) even when strong differences in fungal composition occur afterharvest Despite overall high colonization levels mycorrhizal colonization was lowest inDR sites akin to observations of reduced EMF root presence in harvested New ZealandNothofagus forests (Dickie Richardson amp Wiser 2009) Nonetheless even in DR sitescolonization levels weresim95 indicating that EMF inoculum was not severely limited forseedlings that recruited successfully

Dispersed retention sites supported lowerα diversity than PF andAR sites Several studiesfrom the NorthernHemisphere report reduced EMF richness with increasing distance from

Hewitt et al (2018) PeerJ DOI 107717peerj5008 1425

the forest edge or individual green trees for both seedlings that naturally established beforelogging (advanced regeneration) and experimentally established seedlings (Kranabetter1999 Hagerman Sakakibara amp Durall 2001 Kranabetter amp Friesen 2002 Outerbridge ampTrofymow 2004 Teste Simard amp Durall 2009) We did not directly measure the impactof seedling proximity to seed trees or the forest edge in the DR sites Instead our resultsshow that on average the DR areas have reduced EMF richness compared to the othertreatments This does not exclude the possibility that there aremycorrhizal hot-spots withinDR areas due to proximity to mature tree rooting zones however our measurements ofEMF attributes were not highly variable from seedling to seedling in the DR which wewould expect if there were great contrasts in EMF dynamics within versus beyond therooting zone of seed trees The low density of seed trees in the DR area may thereforeaccount for the lower EMF richness on seedlings

We observed thatmore taxawere shared between the PF andAR treatments than betweenPF and DR supporting the idea that the AR patches act as refugia for species diversity(Soler et al 2015) more effectively than the individual seed trees in the DR AlthoughEMF composition was similar between PF and AR sites and differed in DR sites therewere only two taxa that showed strong gradients in abundance related to VR treatment(Timgrovea ferruginea and Cortinarius cycneus) Both of these fungi have been observed inenvironmental sampling in the Southern Hemisphere but their individual ecologies arenot well known (Danks Lebel amp Vernes 2010 Johnston et al 2010 Nouhra et al 2013)

EMF effects on seedling performanceWe detected species-level correlations between five individual EMF taxa and seedlingbiomass corroborating findings from greenhouse and field studies demonstrating thatthe loss or gain of an EMF taxon can influence seedling performance (Jones Durall ampCairney 2003 and references therein) We also observed significant correlations betweensoil moisture seedling biomass and EMF community composition Yet the causalityof these relationships is unclear Monitoring at the permanent PEBANPA network sitesshows that DR areas have overall lower seedling and sapling density which may releaseseedlings from competition resulting in overall larger seedlings (Martiacutenez Pastur et al2009 Martiacutenez Pastur et al 2011) These larger seedlings may in turn select for particularfungi Alternatively the EMF in DR sites may promote greater seedling growth as hasbeen observed in early-successional northern forests (Kranabetter 2004) Outside of theintraspecific competition dynamics between seedlings and their EMF the differences in soilmoisture related to VR treatment could be a driving factor in explaining the variation inEMF composition (Boucher amp Malajczuk 1990) as well as the positive correlation betweenEMF and host plant biomass in sites with higher moisture (Kennedy amp Peay 2007) Yet afourth explanation of these correlations between seedling size recruitment density andEMF may be the strong abiotic gradients across the VR treatments (Martiacutenez Pastur et al2007Martiacutenez Pastur et al 2011Martiacutenez Pastur et al 2014) The high light andmoisturelevels in DR compared to AR and PF sites may support relatively high biomass accrualwhen seedlings do recruit successfully However the consistently lower recruitment levelsregardless of seed availability and good seedbed conditions (Martiacutenez Pastur et al 2011

Hewitt et al (2018) PeerJ DOI 107717peerj5008 1525

Torres et al 2015 Table 1) in DR sites may be due in part to reduced EMF metrics andwe therefore suggest that EMF may play a larger role in recruitment than we were able tomeasure with our biomass metric Further experimentation is needed to disentangle thehierarchy of VR effects on the outcomes of seedling-EMF interactions (sensu Kranabetter2004 Nicholson amp Jones 2017)

VR impacts and conservation of forest biodiversityOur study of the VR management regime can be used to infer how forests managed formultiple goals impact the diversity of EMF and plant-fungal interactions at the seedlingphase in southern Nothofagus forests Unlike other studies from our study region that havefound that AR and DR sites increase the species richness of various taxa of insects birds andplants (Soler et al 2015 Soler et al 2016) we found that AR sites maintained EMF richnessand DR supported a lower number of taxa The contradiction with other observations inour study system of VR effects on biodiversity likely has to do with the obligate symbiosisof EMF with N pumilio host plants and is supported by findings from EMF studies in theNorthern Hemisphere (Kranabetter De Montigny amp Ross 2013) Within our study regionSoler et al (2015) found that AR maintained the forest structure and microenvironmentconditions found in PF stands With variation in the physical environment more similarbetween PF and AR than AR and DR or PF and DR it follows that treatment-level αdiversity evenness and root colonization would vary more significantly between DR andthe other treatments The DR treatment negatively affected forest structure (Soler et al2015) resulting in a reduction in EMF host plants Therefore the shift in EMF compositionbetween treatments is likely due to the presence and abundance of adult trees and maybe further modulated by EMF dispersal capabilities (Peay et al 2012 Horton 2017)competitive abilities (Kennedy et al 2007 Kennedy et al 2011) and disturbance toleranceof EMF present in DR sites A study of the impacts of dispersed green tree retention and ARshowed that together these treatments maintained sporocarp production of EMF (Luomaet al 2004) while in our study AR seem most important for the maintenance of EMFcommunity dynamics

CONCLUSIONSThe VR management applied to N pumilio forests had a clear impact on recruitmentmetrics like seedling density and biomass and the structure of EMF communities The DRsites had higher soil moisture and fewer seedlings but the seedlings had greater biomassalong with lower EMF diversity colonization and different composition than seedlingsin AR and PF stands In contrast EMF richness taxonomic diversity and compositionwere maintained in AR compared to PF stands and are likely important to seedlingrecruitment However the shifts in EMF attributes across VR treatments were not themost important variables explaining variation in seedling biomass This suggests that theeffects of EMF post-harvest on seedling biomass may be secondary to or muted by otherpresumably abiotic variables such as seedling access to resources like soil moisture whichhas been shown to be important to seedling success in our study area Our results provideboth additional support for the notion that AR sites act as islands of ecological integrity

Hewitt et al (2018) PeerJ DOI 107717peerj5008 1625

maintaining biological legacies from PF stands and highlight the effects of the complexinteractions between retention treatments the post-harvest abiotic environment and EMFon seedling recruitment dynamics

ACKNOWLEDGEMENTSWe thank Gaston Kreps for help aging seedlings and Ian Herriott for assistance withmolecular analyses

ADDITIONAL INFORMATION AND DECLARATIONS

FundingRebecca E Hewitt conducted this study as part of a US National Science Foundation (NSF)International Research Experience for Students grant (OISE 0854350) to Christopher BAnderson Further support to Rebecca E Hewitt came from an NSF Graduate ResearchFellowship (DGE-0639280 and 1242789) Alaska EPSCoR (EPS-0701898) as well as fromthe Bonanza Creek Long-Term Ecological Research (NSF DEB-1636476 and USDA ForestService PNW Research Station JVA-11261952-231) Lab reagents and materials werefunded by an NSF grant (ARC-0632332) to Donald Lee Taylor and by the University ofNorth Texas (Office of the Vice-President for Research) to Christopher B Anderson Therewas no additional external funding received for this study The funders had no role in studydesign data collection and analysis decision to publish or preparation of the manuscript

Grant DisclosuresThe following grant information was disclosed by the authorsUS National Science Foundation (NSF) International Research Experience for StudentsOISE 0854350NSF Graduate Research Fellowship DGE-0639280 1242789Alaska EPSCoR EPS-0701898Bonanza Creek Long-Term Ecological Research NSF DEB-1636476USDA Forest Service PNW Research Station JVA-11261952-231

Competing InterestsThe authors declare there are no competing interests

Author Contributionsbull Rebecca E Hewitt conceived and designed the experiments performed the experimentsanalyzed the data contributed reagentsmaterialsanalysis tools prepared figures andortables authored or reviewed drafts of the paper approved the final draftbull Donald Lee Taylor analyzed the data contributed reagentsmaterialsanalysis toolsprepared figures andor tables authored or reviewed drafts of the paper approved thefinal draftbull Teresa N Hollingsworth analyzed the data prepared figures andor tables authored orreviewed drafts of the paper approved the final draft

Hewitt et al (2018) PeerJ DOI 107717peerj5008 1725

bull Christopher B Anderson conceived and designed the experiments authored or revieweddrafts of the paper approved the final draft logisticsbull Guillermo Martiacutenez Pastur conceived and designed the experiments performed theexperiments analyzed the data contributed reagentsmaterialsanalysis tools authoredor reviewed drafts of the paper approved the final draft

DNA DepositionThe following information was supplied regarding the deposition of DNA sequences

The fungal sequences have been submitted to GenBank under accession numbersMH019772ndashMH019959

Data AvailabilityThe following information was supplied regarding data availability

Hewitt Rebecca Eliza Taylor D Lee Hollingsworth Teresa Nettleton Martinez PasturGuillermo Anderson Christopher B 2018 Ectomycorrhizal community compositionassociated withNothofagus pumilio seedlings harvested fromVariable Retention treatmentsat Los Cerros Ranch Tierra del Fuego Argentina Bonanza Creek LTERmdashUniversityof Alaska Fairbanks BNZ686 httpwwwlteruafedudatadata-detailid686 doi106073pastadf801ec9b7f3493b6eb3bf191124e41f

Hewitt Rebecca Eliza Taylor D Lee Hollingsworth Teresa Nettleton Martinez PasturGuillermo Anderson Christopher B 2018Mycorrhizal colonization age and abovegroundbiomass (g) of Nothofagus pumilio seedlings harvested from Variable Retentiontreatments at Los Cerros Ranch Tierra del Fuego Argentina Bonanza Creek LTERmdashUniversity of Alaska Fairbanks BNZ687 httpwwwlteruafedudatadata-detailid687doi106073pasta350b701beb4ac2ebd2ef3f29893e6ed5

These files are also available as Supplemental Files

Supplemental InformationSupplemental information for this article can be found online at httpdxdoiorg107717peerj5008supplemental-information

REFERENCESAmaranthusMP Perry DA 1987 Effect of soil transfer on ectomycorrhiza formation

and the survival and growth of conifer seedlings on old nonreforested clear-cutsCanadian Journal of Forest Research 17944ndash950 DOI 101139x87-147

Beals EW 1984 BrayndashCurtis ordination an effective strategy for analysis of multivariateecological data Advances in Ecological Research 141ndash55DOI 101016S0065-2504(08)60168-3

Bent E Taylor DL 2010 Direct amplification of DNA from fresh and preservedectomycorrhizal root tips Journal of Microbiological Methods 80206ndash208DOI 101016jmimet200911011

Berry KJ Kvamme KL Mielke PW 1983 Improvements in the permutation test forthe spatial-analysis of the distribution of artifacts into classes American Antiquity48547ndash553 DOI 102307280561

Hewitt et al (2018) PeerJ DOI 107717peerj5008 1825

Borchers SL Perry DA 1990 Growth and ectomycorrhiza formation of Douglas-firseedlings grown in soils collected at different distances from pioneering hardwoodsin southwest Oregon clear-cuts Canadian Journal of Forest Research 20712ndash721DOI 101139x90-094

Boucher NL Malajczuk N 1990 Effects of high soil moisture on formation of ecto-mycorrhizas and growth of karri (Eucalyptus diversicolor) seedlings inoculatedwith Descolea maculata Pisolithus tinctorius and Laccaria laccata New Phytologist11487ndash91 DOI 101111j1469-81371990tb00377x

Breiman L 2001a Random forestsMachine Learning 455ndash32DOI 101023A1010933404324

Breiman L 2001b Statistical modeling the two cultures Statistical Science 16199ndash231Brundrett M Bougher N Dell B Grove T Malajczuk N 1996 Working with mycor-

rhizas in forestry and agriculture In AClAR Monograph 32 Canberra AustralianCentre for International Agricultural Research Canberra i-374

Caporaso JG Kuczynski J Stombaugh J Bittinger K Bushman FD Costello EK FiererN Pena AG Goodrich JK Gordon JI Huttley GA Kelley ST Knights D Koenig JELey RE Lozupone CA McDonald D Muegge BD PirrungM Reeder J SevinskyJR Turnbaugh PJ WaltersWAWidmann J Yatsunenko T Zaneveld J KnightR 2010 QIIME allows analysis of high-throughput community sequencing dataNature Methods 7335ndash336 DOI 101038nmethf303

Cline ET Ammirati JF Edmonds RL 2005 Does proximity to mature trees influenceectomycorrhizal fungus communities of Douglas-fir seedlings New Phytologist166993ndash1009 DOI 101111j1469-8137200501387x

Cline ET Vinyard B Edmonds R 2007 Spatial effects of retention trees on mycor-rhizas and biomass of Douglas-fir seedlings Canadian Journal of Forest Research37430ndash438 DOI 101139x06-229

Collado L 2001 Los bosques de Tierra del Fuego anarsquolisis de su estratificacioacuten medianteimaacutegenes satelitales para el inventario forestal de la provincia de Tierra del FuegoMultequina 101ndash15

Colwell RK 2013 EstimateS statistical estimation of species richness and shared speciesfrom samples Version 90 Available at http purloclcorg estimates

Colwell RK Mao CX Chang J 2004 Interpolating extrapolating and compar-ing incidence-based species accumulation curves Ecology 852717ndash2727DOI 10189003-0557

Cuevas JG ArroyoMT 1999 Ausencia de banco de semillas persistente en NothofagusRevista Chilena de Historia Natural 7273ndash82

Cutler DR Edwards Jr TC Beard KH Cutler A Hess KT Gibson J Lawler JJ2007 Random forests for classification in ecology Ecology 882783ndash2792DOI 10189007-05391

Dahlberg A Schimmel J Taylor AFS Johannesson H 2001 Post-fire legacy of ectomy-corrhizal fungal communities in the Swedish boreal forest in relation to fire severityand logging intensity Biological Conservation 100151ndash161DOI 101016S0006-3207(00)00230-5

Hewitt et al (2018) PeerJ DOI 107717peerj5008 1925

DanksM Lebel T Vernes K 2010 lsquoCort short on a mountaintoprsquomdashEight new species ofsequestrate Cortinarius from sub-alpine Australia and affinities to sections withinthe genus Persoonia Molecular Phylogeny and Evolution of Fungi 24106ndash126DOI 103767003158510X512711

Deshpande VWang Q Greenfield P CharlestonM Porras-Alfaro A Kuske CR ColeJR Midgley DJ Tran-Dinh N 2016 Fungal identification using a Bayesian classifierand the Warcup training set of internal transcribed spacer sequencesMycologia1081ndash5 DOI 10385214-293

Dickie IA Richardson SJ Wiser SK 2009 Ectomycorrhizal fungal communities and soilchemistry in harvested and unharvested temperate Nothofagus rainforests CanadianJournal of Forest Research 391069ndash1079 DOI 101139X09-036

Diehl P MazzarinoMJ Fontenla S 2008 Plant limiting nutrients in Andean-Patagonian woody species effects of interannual rainfall variation soil fertilityand mycorrhizal infection Forest Ecology and Management 2552973ndash2980DOI 101016jforeco200802003

Dinno A 2017 Package dunntest Dunnrsquos test of multiple comparisons using rank sumsversion 134 CRAN Available at https cranr-projectorgwebpackagesdunntestindexhtml

DufreneM Legendre P 1997 Species assemblages and indicator species the needfor a flexible asymmetrical approach Ecological Monographs 67(3)345ndash366DOI 1018900012-9615(1997)067[0345SAAIST]20CO2

Edgar RC 2004MUSCLE multiple sequence alignment with high accuracy and highthroughput Nucleic Acids Research 321792ndash1797 DOI 101093nargkh340

Edgar RC 2010 Search and clustering orders of magnitude faster than BLAST Bioinfor-matics 262460ndash2461 DOI 101093bioinformaticsbtq461

Fernaacutendez NV Marchelli P Gherghel F Kost G Fontenla SB 2015 Ectomycorrhizalfungal communities in Nothofagus nervosa (Rauliacute) a comparison between domesti-cated and naturally established specimens in a native forest of Patagonia ArgentinaFungal Ecology 1836ndash47 DOI 101016jfuneco201505011

Franklin JF Berg D Thornburgh D Tappeiner J 1997 Alternative silviculturalapproaches to timber harvesting variable retention harvest systems In Kohm KFranklin JF eds Creating a forestry for the 21st Century New York Island Press111ndash140

Gamondegraves Moyano IM Morgan RK Martiacutenez Pastur G 2016 Reshaping forestmanagement in southern Patagonia a qualitative assessment Journal of SustainableForestry 3537ndash59 DOI 1010801054981120151043559

Gardes M Bruns TD 1993 ITS primers with enhanced specificity for basidiomycetesmdashapplication to the identification of mycorrhizae and rustsMolecular Ecology2113ndash118 DOI 101111j1365-294X1993tb00005x

Garnica S WeiszligM Oberwinkler F 2003Morphological and molecular phylogeneticstudies in South American Cortinarius speciesMycological Research 1071143ndash1156DOI 101017S0953756203008414

Hewitt et al (2018) PeerJ DOI 107717peerj5008 2025

Gea-Izquierdo G Martiacutenez Pastur G Cellini JM Lencinas MV 2004 Forty years ofsilvicultural management in southern Nothofagus pumilio primary forests ForestEcology and Management 201335ndash347 DOI 101016jforeco200407015

Goecks J Nekrutenko A Taylor J 2010 Galaxy a comprehensive approach for sup-porting accessible reproducible and transparent computational research in the lifesciences Genome Biology 11Article R86 DOI 101186gb-2010-11-8-r86

Hagerman SM Jones MD Bradfield GE Gillespie M Durall DM 1999 Effects of clear-cut logging on the diversity and persistence of ectomycorrhizae at a subalpine forestCanadian Journal of Forest Research 29124ndash134 DOI 101139x98-186

Hagerman SM Sakakibara SM Durall DM 2001 The potential for woody understoryplants to provide refuge for ectomycorrhizal inoculum at an interior Douglas-fir forest after clear-cut logging Canadian Journal of Forest Research 31711ndash721DOI 101139x00-199

Hoeksema JD Chaudhary VB Gehring CA Johnson NC Karst J Koide RTPringle A Zabinski C Bever JD Moore JCWilson GWT Klironomos JNUmbanhowar J 2010 A meta-analysis of context-dependency in plant re-sponse to inoculation with mycorrhizal fungi Ecology Letters 13394ndash407DOI 101111j1461-0248200901430x

Hollingsworth TN Johnstone JF Bernhardt E Chapin III FS 2013 Fire severity filtersregeneration traits to shape community assembly in Alaskarsquos boreal forest PLOSONE 8e56033 DOI 1051371journalpone0056033

Horton BM GlenM Davidson NJ Ratkowsky D Close DCWardlaw TJ MohammedC 2013 Temperate eucalypt forest decline is linked to altered ectomycorrhizal com-munities mediated by soil chemistry Forest Ecology and Management 302329ndash337DOI 101016jforeco201304006

Horton BM GlenM Davidson NJ Ratkowsky DA Close DCWardlaw TJ MohammedC 2017 An assessment of ectomycorrhizal fungal communities in Tasmaniantemperate high-altitude Eucalyptus delegatensis forest reveals a dominance of theCortinariaceaeMycorrhiza 2767ndash74 DOI 101007s00572-016-0725-0

Horton TR 2017 Spore dispersal in ectomycorrhizal fungi at fine and regional scales InTedersoo L ed Biogeography of mycorrhizal symbiosis Cham Springer InternationalPublishing 61ndash78

Johnston P Park D Dickie I Walbert K 2010 Using molecular techniques to combinetaxonomic and ecological data for fungi reviewing the Data Deficient fungi list2009 In Science for conservation Wellington Department of Conservation 31

Jones MD Durall DM Cairney JW 2003 Ectomycorrhizal fungal communities inyoung forest stands regenerating after clearcut logging New Phytologist 157399ndash422DOI 101046j1469-8137200300698x

Jones MD Smith SE 2004 Exploring functional definitions of mycorrhizas aremycorrhizas always mutualisms Canadian Journal of Botany 821089ndash1109DOI 101139b04-110

Kotildeljalg U Nilsson RH Abarenkov K Tedersoo L Taylor AFS BahramM Bates STBruns TD Bengtsson-Palme J Callaghan TM Douglas B Drenkhan T Eberhardt

Hewitt et al (2018) PeerJ DOI 107717peerj5008 2125

U Duentildeas M Grebenc T Griffith GW HartmannM Kirk PM Kohout P LarssonE Lindahl BD Luumlcking R Martiacuten MP Matheny PB Nguyen NH Niskanen TOja J Peay KG Peintner U PetersonM Potildeldmaa K Saag L Saar I Schuumlszligler AScott JA Seneacutes C SmithME Suija A Taylor DL Telleria MTWeiss M LarssonK-H 2013 Towards a unified paradigm for sequence-based identification of fungiMolecular Ecology 225271ndash5277 DOI 101111mec12481

Kennedy PG Bergemann SE Hortal S Bruns TD 2007 Determining the outcomeof field-based competition between two Rhizopogon species using real-time PCRMolecular Ecology 16(4)881ndash890 DOI 101111j1365-294X200603191x

Kennedy PG Higgins LM Rogers RHWeber MG 2011 Colonization-competitiontradeoffs as a mechanism driving successional dynamics in ectomycorrhizal fungalcommunities PLOS ONE 6e25126 DOI 101371journalpone0025126

Kennedy PG Peay KG 2007 Different soil moisture conditions change the outcomeof the ectomycorrhizal symbiosis between Rhizopogon species and Pinus muricataPlant and Soil 291155ndash165 DOI 101007s11104-006-9183-3

Koide RT Sharda JN Herr JR Malcolm GM 2008 Ectomycorrhizal fungi and thebiotrophyndashsaprotrophy continuum New Phytologist 178230ndash233DOI 101111j1469-8137200802401x

Kranabetter JM 1999 The effect of refuge trees on a paper birch ectomycorrhizacommunity Canadian Journal of Botany 771523ndash1528 DOI 101139b99-132

Kranabetter J 2004 Ectomycorrhizal community effects on hybrid spruce seedlinggrowth and nutrition in clearcuts Canadian Journal of Botany 82983ndash991DOI 101139b04-077

Kranabetter JM DeMontigny L Ross G 2013 Effectiveness of green-tree reten-tion in the conservation of ectomycorrhizal fungi Fungal Ecology 6430ndash438DOI 101016jfuneco201305001

Kranabetter JM Friesen J 2002 Ectomycorrhizal community structure on westernhemlock (Tsuga heterophylla) seedlings transplanted from forests into openingsCanadian Journal of Botany 80861ndash868 DOI 101139b02-071

Kruskal J WishM 1978Multidimensional scaling Thousand Oaks SAGE PublicationsLtd DOI 1041359781412985130

Larsson A 2014 AliView a fast and lightweight alignment viewer and editor for largedatasets Bioinformatics 303276ndash3278DOI 101093bioinformaticsbtu531

Lencinas MVMartiacutenez Pastur G Rivero P Busso C 2008 Conservation valueof timber quality versus associated non-timber quality stands for understorydiversity in Nothofagus forests Biodiversity and Conservation 172579ndash2597DOI 101007s10531-008-9323-6

Liaw AWiener M 2015 Package randomForest Breiman and Cutlerrsquos random forestsfor classification and regression version 46-12 CRAN Available at https cranr-projectorgwebpackages randomForest indexhtml

Luoma DL Eberhart JL Molina R AmaranthusMP 2004 Response of ectomycorrhizalfungus sporocarp production to varying levels and patterns of green-tree retentionForest Ecology and Management 202337ndash354 DOI 101016jforeco200407041

Hewitt et al (2018) PeerJ DOI 107717peerj5008 2225

Martiacutenez Pastur G Cellini JM Lencinas MV Barrera M Peri PL 2011 Environmentalvariables influencing regeneration of Nothofagus pumilio in a system with combinedaggregated and dispersed retention Forest Ecology and Management 261178ndash186DOI 101016jforeco201010002

Martiacutenez Pastur G Cellini JM Peri PL Vukasovic RF FernaacutendezMC 2000Timber production of Nothofagus pumilio forests by a shelterwood system inTierra del Fuego (Argentina) Forest Ecology and Management 134153ndash162DOI 101016S0378-1127(99)00253-4

Martiacutenez Pastur G Lencinas MV Cellini JM Peri PL Soler R 2009 Timber manage-ment with variable retention in Nothofagus pumilio forests of Southern PatagoniaForest Ecology and Management 258436ndash443 DOI 101016jforeco200901048

Martiacutenez Pastur G Lencinas MV Peri PL ArenaM 2007 Photosynthetic plasticity ofNothofagus pumilio seedlings to light intensity and soil moisture Forest Ecology andManagement 243274ndash282 DOI 101016jforeco200703034

Martiacutenez Pastur G Peri PL Huertas Herrera A Schindler S Diacuteaz-Delgado R LencinasMV Soler R 2017 Linking potential biodiversity and three ecosystem services insilvopastoral managed forest landscapes of Tierra del Fuego Argentina Interna-tional Journal of Biodiversity Science Ecosystem Services amp Management 131ndash11DOI 1010802151373220161260056

Martiacutenez Pastur G Soler R Cellini JM Lencinas MV Peri PL NeylandMG 2014 Sur-vival and growth of Nothofagus pumilio seedlings under several microenvironmentsafter variable retention harvesting in southern Patagonian forests Annals of ForestScience 71349ndash362 DOI 101007s13595-013-0343-3

Marx DHMarrs LF Cordell CE 2002 Practical use of the mycorrhizal fungal tech-nology in forestry reclamation arboriculture agriculture and horticultureDendrobiology 4727ndash40

McCune B Grace JB 2001 Analysis of ecological communities Gleneden Beach MjMSoftware Design

McCune B Grace JB 2002 Analysis of ecological communities Gleneden Beach MjMSoftware Design

McCune B MeffordMJ 2011 PC-ORD multivariate analysis of ecological data Version6 Gleneden Beach MjM Software Design Available at httpswwwpcordcom

Mittermeier RA Mittermeier CG Brooks TM Pilgrim JD KonstantWR Da FonsecaGA Kormos C 2003Wilderness and biodiversity conservation Proceedings ofthe National Academy of Sciences of the United States of America 10010309ndash10313DOI 101073pnas1732458100

Nicholson BA Jones MD 2017 Early-successional ectomycorrhizal fungi effectivelysupport extracellular enzyme activities and seedling nitrogen accumulation inmature forestsMycorrhiza 27247ndash260 DOI 101007s00572-016-0747-7

Nouhra E Urcelay C Longo S Tedersoo L 2013 Ectomycorrhizal fungal communitiesassociated to Nothofagus species in Northern PatagoniaMycorrhiza 23487ndash496DOI 101007s00572-013-0490-2

Olson DM Dinerstein E 2002 The Global 200 priority ecoregions for global conserva-tion Annals of the Missouri Botanical Garden 89199ndash224 DOI 1023073298564

Hewitt et al (2018) PeerJ DOI 107717peerj5008 2325

Orlovich DA Cairney JG 2004 Ectomycorrhizal fungi in New Zealand currentperspectives and future directions New Zealand Journal of Botany 42721ndash738DOI 1010800028825X20049512926

Outerbridge RA Trofymow JA 2004 Diversity of ectomycorrhizae on experimentallyplanted Douglas-fir seedlings in variable retention forestry sites on southernVancouver Island Canadian Journal of Botany 821671ndash1681 DOI 101139b04-134

Palfner G CansecoMI Casanova-Katny A 2008 Post-fire seedlings of Nothofagusalpina in Southern Chile show strong dominance of a single ectomycorrhizalfungus and a vertical shift in root architecture Plant and Soil 313237ndash250DOI 101007s11104-008-9697-y

Parke JL Linderman RG Trappe JM 1984 Notes inoculum potential of ectomycor-rhizal fungi in forest soils of Southwest Oregon and Northern California ForestScience 30300ndash304

Peay KG Schubert MG Nguyen NH Bruns TD 2012Measuring ectomycorrhizalfungal dispersal macroecological patterns driven by microscopic propagulesMolecular Ecology 214122ndash4136 DOI 101111j1365-294X201205666x

Peri PL Bahamonde HA Lencinas MV Gargaglione V Soler R Ormaechea SMartiacutenez Pastur G 2016a A review of silvopastoral systems in native forestsof Nothofagus antarctica in southern Patagonia Argentina Agroforestry Systems90933ndash960 DOI 101007s10457-016-9890-6

Peri PL Lencinas MV Bousson J Lasagno R Soler R Bahamonde H Martiacutenez PasturG 2016b Biodiversity and ecological long-term plots in Southern Patagonia tosupport sustainable land management the case of PEBANPA network Journal forNature Conservation 3451ndash64 DOI 101016jjnc201609003

Perry DA AmaranthusMP Borchers JG Borchers SL Brainerd RE 1989 Boot-strapping in ecosystems internal interactions largely determine productivity andstability in biological systems with strong positive feedback Bioscience 39230ndash237DOI 1023071311159

Perry D Meyer M Egeland D Rose S Pilz D 1982 Seedling growth and mycorrhizalformation in clearcut and adjacent undisturbed soils in montana a green-housebioassay Forest Ecology and Management 4261ndash273DOI 1010160378-1127(82)90004-4

Perry DA Molina R AmaranthusMP 1987Mycorrhizae mycorrhizospheres andreforestation current knowledge and research needs Canadian Journal of ForestResearch 17929ndash940 DOI 101139x87-145

Pinheiro J Bates D DebRoy S Sarkar D R Core Team 2018 nlme linear and nonlinearmixed effects models R package version 31-137 Available at httpsCRANR-projectorgpackage=nlme

R Core Team 2016 R a language and environment for statistical computing Vienna RFoundation for Statistical Computing Available at httpwwwR-projectorg

Simard SW 2009 The foundational role of mycorrhizal networks in self-organizationof interior Douglas-fir forests Forest Ecology and Management 258S95ndashS107DOI 101016jforeco200905001

Hewitt et al (2018) PeerJ DOI 107717peerj5008 2425

Smith SE Read DJ 2008Mycorrhizal symbiosis New York Academic PressSoler R Schindler S Lencinas MV Peri PL Martiacutenez Pastur G 2015 Retention forestry

in Southern Patagonia multiple environmental impacts and their temporal trendsInternational Forestry Review 17231ndash243 DOI 101505146554815815500589

Soler R Schindler S Lencinas MV Peri PL Martiacutenez Pastur G 2016Why bio-diversity increases after variable retention harvesting a meta-analysis forsouthern Patagonian forests Forest Ecology and Management 369161ndash169DOI 101016jforeco201602036

Tedersoo L Gates G Dunk CW Lebel T May TW Kotildeljalg U Jairus T 2009 Establish-ment of ectomycorrhizal fungal community on isolated Nothofagus cunninghamiiseedlings regenerating on dead wood in Australian wet temperate forests does fruit-body type matterMycorrhiza 19403ndash416 DOI 101007s00572-009-0244-3

Tedersoo L Jairus T Horton BM Abarenkov K Suvi T Saar I Kotildeljalg U 2008Strong host preference of ectomycorrhizal fungi in a Tasmanian wet sclerophyllforest as revealed by DNA barcoding and taxon-specific primers New Phytologist180479ndash490 DOI 101111j1469-8137200802561x

Teste FP Simard SW Durall DM 2009 Role of mycorrhizal networks and tree prox-imity in ectomycorrhizal colonization of planted seedlings Fungal Ecology 221ndash30DOI 101016jfuneco200811003

Timling I Dahlberg AWalker DA Gardes M Charcosset JY Welker JM Taylor DL2012 Distribution and drivers of ectomycorrhizal fungal communities across theNorth American Arctic Ecosphere 31ndash25 DOI 101890ES12-002171

Torres AD Cellini JM Lencinas MV Barrera MD Soler R Diacuteaz-Delgado R MartiacutenezPastur GJ 2015 Seed production and recruitment in primary and harvestedNothofagus pumilio forests influence of regional climate and years after cuttingsForest Systems 24endash016

Truong C Mujic AB Healy R Kuhar F Furci G Torres D Niskanen T Sandoval-LeivaPA Fernaacutendez N Escobar JM Moretto A Palfner G Pfister D Nouhra E SwenieR Saacutenchez-Garciacutea M Matheny PB SmithME 2017How to know the fungicombining field inventories and DNA-barcoding to document fungal diversity NewPhytologist 214913ndash919 DOI 101111nph14509

Van der HeijdenMGA Horton TR 2009 Socialism in soil The importance of myc-orrhizal fungal networks for facilitation in natural ecosystems Journal of Ecology971139ndash1150 DOI 101111j1365-2745200901570x

Varenius K Lindahl BD Dahlberg A 2017 Retention of seed trees fails to lifeboatectomycorrhizal fungal diversity in harvested Scots pine forests FEMS MicrobiologyEcology 93(9) DOI 101093femsecfix105

White TJ Bruns T Lee S Taylor JW 1990 Amplification and direct sequencing offungal ribosomal RNA genes for phylogenetics In Innis MA Gelfand DH SninskyJJ White TJ eds PCR Protocols a Guide to methods and applications New YorkAcademic Press 315ndash322

Hewitt et al (2018) PeerJ DOI 107717peerj5008 2525