Ectomycorrhizal Colonization and Diversity in Relation to Tree Biomass and Nutrition in a Plantation of Transgenic Poplars with Modified Lignin Biosynthesis

Wood from biomass plantations with fast growing tree species such as poplars can be used as an alternative feedstock for production of biofuels. To facilitate utilization of lignocellulose for saccharification, transgenic poplars with modified or reduced lignin contents may be useful. However, the potential impact of poplars modified in the lignification pathway on ectomycorrhizal (EM) fungi, which play important roles for plant nutrition, is not known. The goal of this study was to investigate EM colonization and community composition in relation to biomass and nutrient status in wildtype (WT, Populus tremula × Populus alba) and transgenic poplar lines with suppressed activities of cinnamyl alcohol dehydrogenase, caffeate/5-hydroxyferulate O-methyltransferase, and cinnamoyl-CoA reductase in a biomass plantation. In different one-year-old poplar lines EM colonization varied from 58% to 86%, but the EM community composition of WT and transgenic poplars were indistinguishable. After two years, the colonization rate of all lines was increased to about 100%, but separation of EM communities between distinct transgenic poplar genotypes was observed. The differentiation of the EM assemblages was similar to that found between different genotypes of commercial clones of Populus × euramericana. The transgenic poplars exhibited significant growth and nutrient element differences in wood, with generally higher nutrient accumulation in stems of genotypes with lower than in those with higher biomass. A general linear mixed model simulated biomass of one-year-old poplar stems with high accuracy (adjusted R2 = 97%) by two factors: EM colonization and inverse wood N concentration. These results imply a link between N allocation and EM colonization, which may be crucial for wood production in the establishment phase of poplar biomass plantations. Our data further support that multiple poplar genotypes regardless whether generated by transgenic approaches or conventional breeding increase the variation in EM community composition in biomass plantations.


Introduction
The growing world population inevitably entails an increasing energy demand along with diminishing fossil fuel resources [1]. Renewable energies from biomass can be used as an alternative to partially replace conventional energy supplies. Trees, especially fast-growing species such as poplars, are an appealing feedstock for this purpose because they can be grown in dense short rotation plantations allowing several harvests without the need to re-plant [2]. Furthermore, poplars have a low nitrogen demand compared with other potential bioenergy crops [3]. Thus, their cultivation may contribute to the mitigation of nitrogen emissions from intensely used agricultural areas [4].
The conversion process of biomass to biofuels requires the breakdown of plant cell walls, which mainly consist of cellulose, hemicelluloses, and lignin [5]. Lignin is a recalcitrant polymer composed of phenylpropanoid units that hinder chemical and enzymatic cellulose degradation necessary for bioethanol production [6]. To amend wood utilization cell wall properties have been changed by targeted genetic approaches [7]. Genes of the biosynthetic pathway of lignin and cellulose have been isolated and characterized [8][9][10]. Suppression of cinnamyl alcohol dehydrogenase (CAD), an enzyme which converts cinnamyl aldehydes to the respective alcohols [5] and caffeate/5-hydroxyferulate O-methyltransferase (COMT), an enzyme involved in biosynthesis of syringyl lignin [5] result in altered lignin composition compared to wildtype (WT) poplars [11][12][13]. Overexpression of ferulate 5-hydroxylase (F5H), an enzyme that catalyzes an intermediate step in lignin biosynthesis, also results in compositional changes and less polymerization of monolignol units compared to the WT [14]. Suppression of cinnamoyl-CoA reductase (CCR) causes reduced lignin contents [15]. Transgenic poplars with alterations in lignin content and composition have been tested for industrial usage and display improved Kraft pulping [16]. The saccharification efficiency is also increased by genetic engineering of the lignin biosynthetic pathway [17].
If the use of genetically modified (GM) poplars with improved wood properties for bioenergy production was expanded, it will be necessary to know whether nutrient status and ecological interactions of GM poplars are changed compared with the WT. In a preceding study we compared whole fungal communities in soil and roots of poplars with suppressed CAD activities and of the WT by pyrosequencing and found a strong dominance of ectomycorrhizal (EM) in roots, whereas saprophytes were prevalent in soil [18]; significant differences of these traits between the CAD lines and WT were not found [18]. The interaction of poplar roots with EM fungi is of particular importance for nutrient acquisition [19]. But other benefits have also been reported such as higher survival rates of EM-inoculated young poplar saplings [20][21][22][23] and increased resistance to drought stress [24][25][26], issues gaining importance with increasing poplar cultivation in a warming climate. Currently it is still unclear if changes in the lignification pathway have significant ecological implication for interacting organisms. Lignin is the end product of the phenylpropanoid pathway, whose modification generally has consequences for the biosynthesis of other phenol-bearing compounds. For example, the suppression of CCR results in decreased lignin, but increased concentrations of phenolic compounds [15]. Phenolic compounds have been implicated in a wide range of ecological interactions. Greenhouse studies have shown that enzymatic activities of microbial communities are altered in soil of poplars with reduced lignin concentrations [27]. Field studies on the EM communities in relation to the performance of poplars with changes in the lignin composition and reduction of the lignin concentrations are lacking.
The aim of this study was to characterize the EM community composition and dynamics in the first cycle of a short rotation plantation with poplars modified in the lignification pathway. To assess the relationship between EM diversity, plant nutrient status and dendromass we analyzed height growth, biomass, and nutrient element composition in leaves, stem and roots of transgenic Populus 6 canescens with suppressed activities of COMT (L9 and L11), CCR (L5 and L7), or CAD (L18, L21 and L22) and the wildtype (WT). We further compared the EM assemblages in the GM plantation with those of commercial poplar clones (P. x euramericana, syn, Populus deltoides 6 Populus nigra c.v. Ghoy, I-214, and Soligo). Our study shows that in the first year after plantation establishment, EM fungal colonization and diversity were linked with tree productivity and low stem nitrogen concentrations. The variation of the EM fungal community composition found on roots of different transgenic poplar genotypes was similar to that found on different commercial poplar genotypes.

Plant material and field site
One hybrid clone of Populus tremula 6 Populus alba (INRA #717-1B4, syn. P. 6 canescens) referred to as wild type (WT) and seven transgenic lines from this WT clone modified in key enzymes of the lignin biosynthetic pathway were used to establish a field trial. The transgenic lines were down regulated in one of the following enzymes of the lignin biosynthesis pathway: CCR (cinnamoyl coenzyme A reductase) with line FS3 = L5 and FAS13 = L7 [15], COMT (caffeic acid O-methyl transferase) with line ASOMTB2B = L9 and ASOMTB10B = L11 [11], and CAD (cinnamyl alcohol dehydrogenase) with line ASCAD21 = L21, ASCAD52 = L18, and SCAD1 = L22 [28]. After multiplying the clones by micropropagation [29] 120 plants of each of the 8 different poplar lines were planted in a plowed area of 1365 m 2 on sandy soil with flint in June 2008, next to INRA in Orléans, Sologne, France (47u839N, 1u919E). The field trial with GM poplars with modified lignin (application B/FR/07/06/01) has been approved by the ''Bureau de la réglementation alimentaire et des Biotechnologies'' from the ''Direction Générale de l''Alimentation'' from the French ''Ministère de l''Agriculture et de la Pêche'' (ministerial decision #07/015 on September 21, 2007 for a 5 year period). The land, where the field trial was conducted, is owned by INRA. Protected species were not sampled.
In this area the mean annual temperature is 10.4uC and precipitation 600 mm. The plant density was chosen according to short rotation coppice practice as follows: the space between trees of one double row was 0.55 m while the interspace between the two double rows was 1.5 m, and the planting distance within a line was 1 m (Fig. S1). The poplar lines were planted in a randomized block design with 5 blocks. Each block consisted of eight plots, one for each line. Each plot consisted of 24 trees (466) planted in two double rows. To prevent edge effects the experimental plantation was bordered with one row of WT clones (Fig. S1). During the growing season the poplars were drip irrigated.
A second plantation with 11 commercial clones of Populus deltoides 6 P. nigra including the cultivars Blanc de Poitou, Carpaccio, Dorskamp, Flevo, Ghoy, I-214, Koster, Lambro, Robusta, Soligo, and Triplo was established in May 2009 in the same area. The random block design consisted of three blocks. Each block consisted of 11 plots. Each plot consisted of 16 trees (464) of one commercial clone. The space between trees of one double row was 0.6 m while the interspace between the two double rows was 1.5 m, and planting distance within a line was 0.6 m (Fig. S2) Sampling of soil cores for analyses of roots and soil Soil cores were harvested immediately after planting (July 2008) to assess the heterogeneity of soil fungi and nitrogen at the beginning. After plantation establishment soil were collected for ECM fungal community analysis in October 2009 and October 2010. In July 2008, 25 soil cores (diameter: 8 cm, depth: 20 cm) were taken randomly in the experimental field, the border area, and the area between the experimental field and a nearby poplar plantation.
In October 2009 and 2010 three plots per clone (i.e. 1 WT +7 GM lines) were randomly chosen and soil cores (diameter: 5 cm, depth: 20 cm) were collected within these plots. Three trees per plot were chosen and three soil cores per tree were taken at a distance of 0.25 m from the trunk. In total 27 soil cores per line were collected. Soil cores were transported on ice and stored at 4uC until further processing.
Sampling in the P. deltoides 6 P. nigra plantation took place in October 2010, one year after planting. The same sampling strategy was used for the plantation with the commercial poplar clones as described above for the transgenic poplars. Three clones were selected for the analysis based on growth differences, which were mainly caused by differences in Melampsora larici-populina leaf rust infection: Soligo (high growth and high rust resistance), Ghoy (low growth and low rust resistance) and I-214  soil samples were sieved and 250 mg sieved soil was used for DNA isolation with the PowerSoil TM DNA Isolation Kit (MO BIO Laboratories, Inc., Canada). The primer pair ITS1 and ITS4 [30] was used to amplify the rDNA ITS-region of fungi. A GC-clamp was added to the 59 end of the ITS4 primer to stabilize the melting behavior of the Polymerase Chain Reaction (PCR) products in the gel according to Muyzer et al. [31].
The separation of the rDNA sequences was achieved in a 7.5% polyacrylamide (37.5: 1 = acrylamide: bis-acrylamide) gel with a linear denaturing gradient from 32-65% of denaturant (100% denaturant containing 40% (v/v) formamide and 7 M urea). After 2 h of polymerization 7.5 ml of 7.5% polyacrylamide gel without denaturant was added (stacking gel). After 20 min of polymerization the gel was loaded with 4 ml of PCR product per lane of each of the 25 samples. Running buffer contained 0.56 TAE (20 mM tris(hydroxymethyl)-aminomethane, pH 7.4, 10 mM sodium acetate, 0.5 mM disodium ethylenedinitrilo-tetraacetic acid). An INGENYphorU-2 system (Ingeny International, Goes, The Netherlands) was used for the DGGE at a constant temperature of 58uC, 120 V and a running time of 16 h. DNA bands were visualized by silver staining following the ''SILVER SEQUEN-CE TM '' protocol (Promega Corporation, Madison, USA). The stained gels were scanned on a flat-bed scanner. The band patterns were manually converted into a present/absent matrix, which was subjected to similarity analyses (Table S1).

Free amino acids, nitrate and ammonium in soil samples
At the time point of plantation establishment (June 2008), the concentrations of nitrogen compounds (nitrate, ammonium, amino acids) in the soil solution were determined. Soil samples were sieved (mesh width 5 mm) and 40 g of fresh soil were mixed with 40 ml 1 mM CaCl 2 , incubated for 10 min and filtered through a WhatmanH folded filter (Ø185 mm, Ref.No. 10314747, Whatmann, Dassel, Germany). After 1 h the resulting filtrate was passed through a glass fiber filter (pore size 1 mm, Pall Life Science, Port Washington, NY, USA) and subsequently through a sterilization filter (0.2 mm Sarstedt Filtropur S, Nümbrecht, Germany). After volume determination, the filtrate was freezedried and dissolved in 0.5 ml double deionized H 2 O. Amino acids were analyzed by high-performance liquid chromatography (Pharmacia/LBK, Freiburg, Germany) according to Tilsner et al. [32]. Nitrate and ammonium were determined by photometric measurements (Shimadzu UV 1602, Hannover, Germany) using enzymatic ammonium and nitrate test kits (Merck 100683, Merck 109713, Merck, Darmstadt, Germany). The concentrations of inorganic nitrogen and amino acids are reported in Table S2.

Ectomycorrhizal colonization and morphotyping
For the investigation of the EM fungal community of roots, soil cores were divided longitudinally, and the three samples, which had been collected around the stem of one tree, were pooled resulting in nine samples per poplar line. Roots were carefully separated from the soil by washing in a sieve under running tap water. The washed roots were inspected under a stereomicroscope (M205 FA, Leica, Wetzlar, Germany) and non-poplar roots were removed from the sample. The root samples were weighed and aliquots were removed, dried and used for nutrient element analyses.
Subsequently, living and dead root tips were counted until a total number of 300 living roots tips per sample was reached. The numbers of the different morphotypes and of the dead root tips were recorded applying a simplified method after Agerer [33]. Dead root tips exhibited a shrunken and dry appearance. EM morphotypes were distinguished by color, shape, texture of the mantle, and absence or presence of rhizomorphes and/or hyphae. Samples of each morphotype were collected and stored at 220uC for molecular analysis.
The vitality index of root tips was determined as: number of living root tips 6100/total number of counted root tips.

Sanger sequencing of the fungal ITS region
For the extraction of genomic DNA of frozen EM root tips the ''innuPREP Plant DNA kit'' (Analytik jena, Jena, Germany) was used following the instructions of the manufacturer. The primer pair ITS4 and ITS5 [30] was used to amplify the rDNA ITSregion by PCR with the PCR protocol described above for the DGGE. Cloning and sequencing or direct sequencing were conducted according to Druebert et al. [34]. The following databases were used for nucleotide BLAST searches: UNITE (http://unite.ut.ee/), Fungal RSyst (http://mycor.nancy.inra.fr/ RSyst/), and NCBI BLASTn (http://www.ncbi.nih.gov/). Fungal sequences have been deposited at NCBI with the accession numbers JQ409279 to JQ409296 and JQ824878 to JQ824884, respectively.

Stem heights and biomass
Heights of trees chosen for EM fungal analysis were measured in October 2009 and 2010, respectively, when seasonal growth had stopped. In 2010 in addition to the height (h) of the leader shoot the number and lengths of side shoots, and stem diameters (d) of all shoots (15 cm above ground) were measured. Fully expanded top leaves were collected (Oct 2009) and dried for nutrient analyses.
Trees were coppiced in March 2010 and above ground stem biomass was determined after drying at 40u for two weeks to constant weight. Since there is no growth between October and March (fall/winter season), the stem biomass data measured in March 2010 represent that of the preceding year (2009).
Biomass in October 2010 was calculated as: V N r with V = 1/3 N r 2 N p N h where r = d/2 and r = 0.50 g N cm 23 [35,36].

Nutrient element and d 13 C analyses
Dry stem wood (March 2010), roots (October 2010) and leaves (October 2010) were cut into small pieces, mixed and aliquots were removed and milled to a fine powder (MM2, Retsch, Hannover, Germany). Nutrient elements were pressure-extracted in HNO 3 and measured by inductively coupled plasma optical emission spectrometry (ICP-OES) after Heinrichs et al. [37]. For N and C analyses powdered dry tissues were weighed into tin cartridges (Hekatech, Wegberg, Germany) and measured with an element analyzer (Element Analyzer EA-1108, Carlo, Erba Instruments, Rodano, Italy). Leaf and wood samples for d 13 C analysis were weighed into tin cartridges (Hekatech, Wegberg, Germany) and analyzed with an isotope mass spectrometer (Delta plus XP, Finnigan MAT, Bremen, Germany) coupled with an element analyzer (EuroVektor, HEKAtech GmbH, Wegberg, Germany).

Statistical analyses
Statistical analyses were conducted using R statistics version 2.9.2 [38]. To identify potential clusters in the distribution of soil fungi (detected by DGGE) and soil nutrients (soluble amino acids, nitrate, and ammonium) across the plots non metric multidimensional scaling (NMDS) was conducted with package: ''vegan'' [39]. Input parameters were Jaccard distance for soil fungi and Euclidean distance for soil nutrients, respectively. To find out if the soil fungal assemblages were related to the composition of the soluble nitrogen compounds in soil, data were subjected to a Mantel test with the package ''vegan'' [39].
Similarities of EM fungal community structures in 2009 and 2010 were analyzed by NMDS using Bray-Curtis distance as input parameter. In all cases a maximum of 100 starts were used to find a stable solution. The procedure was repeated with the best previous solution to prevent local optima. Function envfit() was used to fit grouping factors (different lines) onto the ordination. 95% confidence ellipses were drawn with function ordiellipse(), package:''vegan'' [39].
Data for height, biomass, mycorrhizal colonization, vitality index, nutrient element concentrations and 13 C signature are shown as means (6SE). Significant differences at p # 0.05 were detected by one-way ANOVA followed by multiple comparisons with TukeyHSD (package: ''stats''). Residuals of the models were analyzed by Kolmogorov-Smirnov and Levene''s test to check for normal distribution and homogeneity of variances, respectively. If one of the assumptions of the ANOVA had to be rejected, Kruskal-Wallis rank sum test followed by Mann Whitney U test was conducted.
Rarefied diversity indices (Shannon-Wiener Index (H9), species richness, and Pielou''s Evenness) based on 850 root tips per sample were calculated using the EcoSim software version 7.72 [40]. Since cumulative rarefied diversity indices for the EM fungi community were calculated, only one value per line and year was obtained. Regression analysis and general mixed models (GLM) were calculated with Statgraphics Centurion (StatPoint Technologies, Inc.,Warrenton, VA). Residuals of the regression models were tested by Shapiro Wilks normality test to check the assumption of normal distribution. If the assumption of normal distribution had to be rejected the Null Hypothesis that the slope is equal to zero was tested by Spearman''s rank correlation. Before starting the analysis the data were checked graphically for outliers followed by Dixon test for outliers, package: ''outliers'' [41].

Absence of fungal clusters and nutrient patches in the soil of a poplar plantation
When the poplar plantation was established in June 2008, nitrogen in the soil solution and fungal distribution were determined to detect potential patchy distribution of soil nutrients and fungi. NMDS did neither reveal any clustering for the patterns of soil fungi (Fig. 1a, permutation test, R 2 = 0.30, p = 0.144) nor for soluble nitrogen in the soil solution at different sampling spots in the plantation (Fig. 1b, R 2 = 0.34, p = 0.101). Other soil nutrient elements and soil pH neither showed positional effects [18].
The mean concentration of the sum amino acids was 415638 nmol kg 21 soil. Glycine, alanine, serine, phenylalanine and isoleucine were the most abundant amino acids in the soil (Fig. 1c). The mean soil concentrations of inorganic nitrogen were 82.667.0 mmol kg 21 for nitrate and 16.660.9 mmol kg 21 for ammonium. To test if the concentrations of the soluble nitrogen compounds in the soil were correlated with the fungal distribution a Mantel test was conducted. No correlation of those parameters was found (r = 20.065, p = 0.634). Since we did not detect clustering of soil fungi or nutrient patches when the plantation was established it is unlikely that further results were influenced by local variations of these environmental factors.
Ectomycorrhizal colonization show temporal dynamics and genotype-but not gene-specific effects in GM poplars One year after planting (2009) the EM colonization varied between the different transgenic poplar lines and WT from 58% to 86% (Table 1). CAD line L22 showed the lowest and CAD line L18 the highest colonization (Table 1). At the end of the following growing season (2010) almost all vital root tips were colonized with EM (Table 1). There was only very little variation between the lines ( Table 1).
The higher EM colonization of roots after two years than after one was also accompanied by higher EM species richness: only eight different EM species were detected after one, however, 30 after two years (Fig. 2, Table S3). Of the 30 EM species, six (Paxillus involutus, Laccaria tortilis, Hebeloma sacchariolens, Hebeloma sp., Cenococcum geophilum and Peziza ostracoderma) had already been present in the preceding year (Table S3). The increases in total ECM species numbers were also reflected in the Shannon-Wiener Index, which increased from a mean across all poplar lines of 1.  Table 1).
To investigate potential genotype-related effects on EM associations, we analyzed the EM community composition in greater detail. One year after plantation establishment, four of the total number of eight detected EM species were dominant colonizing .90% of the mycorrhizal root tips of all poplar lines; no significant differences between CAD, CCR, COMT and WT lines were found (Fig. 2). NMDS of the ECM fungal community on 1-year-old poplars neither revealed significant separation of different poplar lines (permutation test R 2 = 0.1649, p = 0.073, Fig. 3a).
At the end of the second year (2010), eight EM species were relatively frequent on the root tips (.10% colonization per EM species) with some significant differences between the poplars lines ( Fig. 2): an uncultured Peziza was more abundant on WT than on CCR line L7 and COMT line L11 roots (Mann-Whitney U-Test, p = 0.022 and p = 0.031, respectively). Laccaria tortilis was more abundant on COMT line L11 than on CAD lines L21 and CAD line L22 (p = 0.0077 and p = 0.0087, respectively) (Fig. 2). The changes in fungal abundance and composition resulted in genotype-related shifts in the EM communities as documented by NMDS (Fig. 3b, permutation test R 2 = 0.43, p = 0.001). The transgenic lines CCR L7 and CAD L22 showed a complete separation of their EM community structures (Fig. 3b) (Fig. 3b).
To elucidate the ecological importance of these observations we also analyzed the EM species composition of three genotypes of high-yielding, commercial clones of P. x euramericana (Ghoy, I-214, and Soligo) in comparison with the WT of P. 6 canescens. The ordination shows a clear separation of the EM communities of one-and two-year-old P. 6 canescens (permutation test: R 2 = 0.76, p = 0.001, Fig. 3c). Among the three P. x euramericana clones studied Ghoy and I-214 showed overlapping EM communities with P. 6 canescens, whereas Soligo was almost completely separated from P. 6 canescens and had less overlap with Ghoy and I-214 than those two genotypes among each other (Fig. 3c). These results support that the EM communities underlie temporal and genotype-specific differentiation. However, a separation of EM communities according to the modification of lignification genes was not found.

Early genotype-specific variation of growth is related to stem N concentrations and ectomycorrhizal root colonization
Since EM fungi can affect nutrient uptake and plant performance, we investigated growth and nutrient status of the poplars in the GM plantation. Significant differences were found for height growth and biomass among the poplar genotypes (Table 2). CAD line L22 generally exhibited the lowest performance and CAD line L18 the best (Table 2). CAD line L18 also produced more side shoots than the other poplar genotypes (Table 2). In comparison    with the WT the lines CAD L22 and CCR L7 showed reduced biomass production, whereas biomass of the other genotypes was unaffected by the genetic modification (Table 2). To find out whether the growth differences of the different poplar genotypes were the results of compromised nutrient supply, the nutrient element status was characterized for leaves, wood and stem, and carbon allocation was assessed by analyses of the d 13 C signatures in leaves and stem biomass (Table S5). The mean d 13 C value of leaves was 227.3460.11% and that of stems 224.9260.03% (p,0.001). This indicates differences in carbon discrimination between leaves and stem; but no genotype-related effects within leaves or stems were found. We have, therefore, no evidence that the growth differences were caused by genotyperelated differences in photosynthetic carbon allocation to wood.
The nutrient element concentrations did not reveal nutritional deficits in comparison with other poplars [42], but significant differences between the analyzed poplar genotypes were detected ( Table 3, Table S5). The highest number of differences in nutrient element concentrations among the genotypes was found in stems (P, N, K, Mg, Ca, Mn), an intermediate number in leaves (P, N, K, C, S) and the lowest number of differences were found in roots (P, K, Mn). These results indicate genotype-specific differences in internal nutrient element allocation. The macronutrients P and K showed genotype-related effects in all tissues and N in leaves and stems. The latter three nutrient elements were analyzed in greater detail since their uptake is known to be regulated by EM fungal associations [19].
Multiple variable analyses revealed no significant correlations of the P concentrations in any of the analyzed tissues with EMrelated parameters such as root colonization, EM species richness, the Shannon Wiener index or root tip vitality (Table S6). To find out if the P concentrations were related to the abundance of specific EM fungi, i.e., related to fungal identity, multiple variable analyses were carried out for the dominant fungi with the tissue nutrient concentrations. None of the nutrient elements (stem concentrations of P, K, or N) showed significant correlations with the abundance of any of the major EM fungi in 2009. In 2010, the leaf P and K concentrations were negatively correlated with the relative abundance of Peziza ostracoderma (for P: R = 20.808, p = 0.015; for K: R = 20.713, p = 0.047) and the leaf P concentrations were positively correlated with the abundance of an unknown ascomycete JQ409294 (R = 0.747, p = 0.033). Although leaf P concentrations were correlated with height (Table  S6), a link between height and the abundance of the ascomycete JQ409294 could not be established (p = 0.19). Therefore, we have no evidence for interactions between distinct EM fungal species, P concentrations and growth.
To further evaluate the relationship between growth, tissue nutrient element concentrations and EM assemblages, we searched the correlation matrix for significant p values (Table  S6). Stem biomass (2009) was significantly correlated with EM fungal species richness (2009), root tip colonization (2009), stem K and stem N concentrations. GLM analyses with these parameters and stepwise removal of the factor with the least significant P-value revealed that stem biomass (2009) was modeled with high accuracy by only two factors: stem N concentrations and mycorrhizal root colonization (adjusted R 2 = 97%, F (model) = 108.4, P (model) = 0.0003, F (N) = 101.1, P (N) = 0.0006, F (EM) = 10.8, P (EM) = 0.03, Fig. 4). Stem biomass was negatively related to N concentrations and positively with the degree of EM root tip colonization (Fig. 4).

Influence of gene modification on mycorrhizal colonization and community structure
Poplars can form mutualistic associations with both arbuscular mycorrhizal and EM fungi [19]. However, in poplar plantations associations with EM fungi are the dominant symbiotic form [18,21]. Age-related increases in root tip colonization and EM species diversity as observed here for GM and WT poplars are well known for non-transgenic as well as transgenic poplars (e.g., suppression of the rolC gene in P. x canescens [43], wildtype P. tremuloides [44]). Besides the dynamic fungal succession, we observed initially differences in root tip colonization, which vanished in the second year and a differentiation of distinct EM communities on different poplar genotypes.
A main question of the current study, therefore, was if the changes in EM colonization and fungal species composition were  caused by the suppression of genes of the lignification pathway. Decreases in lignin as caused by CCR suppression or changes in the lignin composition as caused by CAD and COMT suppression interfere with secondary metabolism and entail changes in the profiles of phenolic compounds [45]. Since phenolic compounds belong to the defense arsenal of poplars [46][47][48][49], negative effects on biotic interactions with EM fungi may be anticipated in transgenic trees with changed lignin biosynthesis. Although we found differences in the EM community composition in the second year after planting, these differences could not be related to the suppression of CCR, CAD or COMT. The composition of EM communities can be influenced by abiotic and biotic environmental factors such as fungal competition [50], soil nutrient and water availability [51][52][53] and the physiology and genetic constitution of the host [34,54,55]. Variations of abiotic factors and patchiness of soil fungi were not detected in our study plantation. Therefore, EM species composition and abundance might have been influenced by host factors. During transformation the positioning of the introduced DNA in the genome cannot be controlled. Thus, the insertion may have side-effects when the introduced DNA fragment unintentionally hits a functional plant gene locus. Therefore, each transformation event may cause intra-specific variation of traits, in addition to the target gene. Controlled experiments testing the colonization efficiency of the EM fungus Laccaria bicolor with the F1 progeny of an inter-specific poplar hybrid revealed that the ability to form mycorrhizas underlies natural intra-specific variation [55][56][57]. Different EM assemblages were also observed in the present study for different varieties of P. x euramericana, a poplar hybrid bred for biomass plantations [58,59]. The intra-specific and interspecific variation in EM assemblages on the WT hybrids of P. x euramericana and P. x canescens was similar to that between CCR line L7 and CAD line 22, which exhibited the largest difference of EM species composition. Our study, therefore, supports that the host genotype can affect the colonization ability of distinct mycorrhizal fugal species. However, the intra-specific variation introduced by Table 3. P, N and K concentrations in stems of wildtype and transgenic poplar (P. 6 canescens). the transformation of poplars with the antisense constructs to suppress CCR, COMT or CAD activities did not result in larger differences in the EM community composition than those observed for different varieties of conventionally bred highyielding poplar clones.

The link between EM colonization and diversity and poplar dendromass and nutrient status
The GM poplars with suppressed activities of enzymes of lignin biosynthesis showed strong (ca. 5-fold) differences in growth and biomass in the plantation. This was not surprising since similar results had been obtained by others studying the performance of lignin-modified plants. For example, Leplé et al. [15] found reduced growth in two of five investigated CCR-suppressed poplar lines under field conditions. Voelker et al. [60] observed extensive variations in aboveground biomass of 14 different lines of P. 6 canescens down-regulated in 4-coumarate:coenzyme A ligase (4CL). Furthermore, greenhouse-grown transgenic poplars with suppressed coumaroyl 39-hydrolase (C39H) activity showed drastic growth reductions [61]. The suppression of C39H activity also reduced the water use efficiency resulting in lower d 13 C signatures in the transgenic compared to WT poplars [61]. If the growth reductions found here were due to impairment of photosynthesis such as reduced stomatal conductance, we would have expected a shift in the d 13 C signature to higher values because of decreased carbon discrimination. However, this was not observed and, therefore, effects on water use and carbon allocation to wood are unlikely reasons for growth reductions in the GM poplars of our study.
Another possibility is that changes in EM colonization and changes in the EM communities had negative impact on tree nutrition leading to reduced growth. This option is not unlikely since the interactions of mycorrhizas with their hosts cover the whole range from beneficial to parasitic effects [62,63]. For example, colonization of P. x euramericana (cv Ghoy) with different arbuscular mycorrhizal fungal species caused reductions in plant biomass [64]. Although the P concentrations of the aboveground tissues increased, P content of the shoot was diminished because of overall biomass loss [64]. In our study, the abundance of the EM fungi Peziza ostracoderma and the ascomycete JQ JQ409294 on root tips of the transgenic poplar genotypes showed negative and positive correlations with foliar P concentrations, respectively. Paxillus involutus, which was present in our plantation, has been shown to increase K and P nutrition of poplars [20][21][22][23]65]. These observations might imply that distinct EM-poplar genotype associations contributed to facilitating or suppressing P or K transfer to their host trees. However, this suggestion is currently speculative since a full nutrient budget of the trees was not possible and the regulation of tree-fungal-environmental interaction is barely understood. Further functional analyses of EM fungi are, therefore, required.
N is one of the most important nutrient elements for plant growth [66]. In young strongly growing poplars N is mainly present in leaves, but a significant fraction is resorbed in fall, present in woody tissues during the dormant season and re-utilized for sprouting in spring [67,68]. Here, we observed a negative relationship between stem N concentrations and stem biomass indicating higher storage in the wood of smaller poplars than in those of taller plants. The biomass differences of stems were maintained in the following season, and could obviously not be compensated by increased internal N utilization of smaller trees for stem growth. Thus, poplars with low growth have the additional disadvantage of wasting N when utilizing woody biomass. There is evidence that N allocation differs between fast and slow growing poplar species since trees with inherently higher biomass production exhibit lower N concentrations in the wood and higher nitrogen productivity [69][70][71]. Poplars grown on a previous agricultural field also showed increased biomass production, decreased N concentrations, and increased nitrogen use efficiency in response to long-term free air CO 2 enrichment [72,73]. Our present data support that, at least in the initial phase, EM colonization is linked with these traits. Positive relationships for growth, nitrogen utilization and EM colonization rates have also been found in Douglas fir [74]. Based on the current data it is not possible to distinguish if poplar growth was stimulated because of higher rates of EM colonization or if trees with higher growth were more amenable to EM colonization. However, the latter possibility is more likely since other studies have already shown that EM colonization and diversity were driven by carbon availability and productivity of the host tree and not vice versa [34,54,74]. Since the root tips of the GM poplars were almost completely colonized with EM at the end of the second growing season, it is clear that the GLM model developed for biomass, nitrogen and root colonization will not be applicable in older plantations. The establishment phase is, however, very important and biomass increments realized during this crucial period will result in further gains because of the exponential nature of growth.

Conclusion
Genetically modified poplars are a potential alternative for the production of renewable energy since their properties can be optimized to facilitate saccharification. The release of transgenic organisms into the field needs to be carefully controlled to avoid negative effects on environmental interactions, especially with potentially beneficial soil microbes. In this study we demonstrated that transgenic poplar lines modified in the lignin biosynthesis pathway show normal abilities to form ectomycorrhizas. Genespecific effects of the transformed poplars on mycorrhizal community structure were not found. Variations in EM community structures found between different GM poplar genotypes were in a range similar to the intra-specific variation of commercial poplar clones. The transgenic lines displayed strong differences in stem biomass production. Wood production in the initial phase of plantation establishment was positively correlated with EM colonization rates and negatively with stem N concentrations. Growth advantages realized in the establishment phase were pertained in the following year. Our results suggest that initial differences in EM colonization may have consequences for long term biomass production.