REGIONAL-SCALE ANALYSIS OF ARBUSCULAR MYCORRHIZAL FUNGI : THE CASE OF BURGUNDY VINEYARDS

1: Université de Bourgogne, UMR Agroécologie INRA 1347/Agrosup/Université de Bourgogne, Pôle Interaction Plantes-Micro-organismes, ERL CNRS 6300, 17 rue Sully, BP 86510, 21065 Dijon Cedex, France 2: Institut National de la Recherche Agronomique, UMR Agroécologie INRA 1347/Agrosup/Université de Bourgogne, Pôle Interaction Plantes-Micro-organismes, ERL CNRS 6300, 17 rue Sully, BP 86510, 21065 Dijon Cedex, France 3: Present addresses: Department of Soil Ecology, UFZ-Helmholtz Centre for Environmental Research, D-06120 Halle (Saale), Germany; and German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, D-04103 Leipzig, Germany


INTRODUCTION
Vitis vinifera (L) is an economically important crop, in particular in some regions where vine production is a major economic resource.Grape production and quality is highly dependent on soil condition and composition (Renouf et al. 2010).However, soil quality and structure is affected by erosion caused by environmental factors like water-induced erosion (Quiquerez et al. 2008), but also by anthropogenic factors (Chevigny et al. 2014).The costs of erosion can counterbalance, to a non-negligible part, the income generated by the sale of wine, in particular because of the loss of N and P provided by chemical fertilization.In Burgundy, which is one of the major regions of vine production and exportation worldwide, soil erosion is a threat for this highquality vine production.It is therefore essential to better understand the relationship between vines and their soil, in terms of physical, chemical and biological properties.
Arbuscular mycorrhiza is a widespread symbiosis involving 80 % of land plant species and arbuscular mycorrhizal fungi (AMF), which all belong to the monophyletic phylum Glomeromycota.AMF are well known for their potential in soil stabilization, which can be useful to limit soil erosion (Rillig and Mummey 2006).These fungi have beneficial effects on plants, enhancing nutrient uptake, in particular phosphate, or resistance against pathogens, including pathogenic fungi and nematodes well known to destroy a large amount of crops (Gianinazzi et al. 2010).They can potentially reduce the input of chemical fertilizers, but often are not taken into account in agriculture because of the still relatively poor understanding of their ecology in agricultural ecosystems.
In vineyards, AMF were shown to enhance plant performance, including nutrient uptake (Schreiner 2007), plant growth (Linderman and Davis 2001) or water use efficiency (Valentine et al. 2006).Previous studies have shown that plant age or genotype may influence AMF community composition, as well as soil properties or cultural practices (Gianinazzi et al. 2010;Peyret-Guzzon et al. 2016).Conversely, different AMF species may have a different impact on plant growth or health (Linderman and Davis 2001).However, a better understanding of the impact of biotic (plant diversity) and abiotic (soil management) factors on AMF composition in vineyards may contribute to improving wine production or quality through the ecosystem services they can provide (for review see Trouvelot et al. 2015).
In this project, in order to better understand AMF diversity in vineyards, the objectives were to (i) identify parameters influencing AMF communities in Burgundy vineyards and (ii) determine a possible core set of species commonly present in these vineyards.In this study, 16 vineyard plots were selected from those monitored by the Bureau Interprofessionnel des Vins de Bourgogne (BIVB, Beaune, France), which are representative of soil properties and agricultural practices in Burgundy vineyards.AMF colonization levels were determined and diversity was analyzed in these plots using 454 pyrosequencing based on the ITS marker gene, designated as the universal barcode marker for Fungi (Schoch et al. 2012).

MATERIALS AND METHODS 1. Sites and sampling
Across Burgundy (France), 16 vineyard plots (planted between 1976 and 1987) were selected in 2013 among those monitored by the BIVB, in Yonne, Côted'Or and Saone-et-Loire (Figure 1, Tables S1 and  S2).They were selected based on their agricultural practices (conventional, conversion or organic) and rootstock.At four sampling points in each plot, young thin roots of nine consecutive vines were collected.Root samples were washed with water to remove the soil.A part of each sample was set aside for mycorrhizal colonization estimation using the method described in Trouvelot et al. (1986).For each sample, mycorrhizal frequency (F%), intensity of mycorrhizal root colonization (M%), and relative arbuscular richness (A%) were calculated.The remaining root samples were stored at -80° until DNA extraction.Soil was sampled in triplicate for the 16 plots, and soil properties were analyzed in the Laboratoire d'Analyse des Sols d'Arras (France).

DNA extraction, PCR amplification and pyrosequencing
Root samples were ground in liquid nitrogen using mortar and pestle.Genomic DNA was extracted from 70mg of roots using a DNeasy plant mini kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol.DNA was eluted in 70 µl and 50 µl AE buffer and kept at -20 °C.Samples were prepared for 454 pyrosequencing in a two-step PCR procedure in order to amplify the ITS 2 region used as gene marker.A first PCR was performed using 0.4 U of Phusion High Fidelity DNA polymerase (Thermo Fisher Scientific, Courtaboeuf, France), 1x Phusion HF buffer, 0.5 µM of the primers SSUmCf and LSUmBr (Krüger et al. 2009), 0.2 mM of each dNTPs and 1µl of genomic DNA, in a final volume of 20 µl.The PCR conditions used were 5 min at 99 °C, 35 cycles of 10 s at 99 °C, 30s at 63 °C and 1min at 72 °C, followed by 10min at 72 °C, using an Eppendorf Mastercycler epgradient S (Vaudaux-Eppendorf, Schönenbuch, Switzerland).Each PCR product was checked on an agarose gel and diluted at 1/100 to be used as template in the nested PCR.The nested PCR was performed using 1U of Phusion High Fidelity polymerase, 1x HF buffer, 0.5µM of the primers ITS3m (Zhong et al. 2010) and ITS4 (White et al. 1990) with 5-bp barcodes, 0.2 µM of each dNTPs and 2 µl of diluted PCR product, in a total volume of 50 µl.PCR conditions were 30 s at 98 °C, 30 cycles of 10 s at 98 °C, 30 s at 64 °C and 20s at 72 °C, followed by 10min at 72 °C, in an Eppendorf Mastercycler ep gradient S. PCR success was checked on an agarose gel.Each DNA extract was amplified three times.The three PCR products, with a size around 400 bp, were pooled and purified using the High Pure PCR Product Purification Kit (Roche Applied Science, Meylan, France) following the manufacturer's instructions.After quantification using the Quant-iT TM Picogreen ® dsDNA Assay kit (Invitrogen, Life Technologies, Saint-Aubin, France), the purified PCR products were mixed equimolarly to prepare sequencing libraries.The libraries were sent to Genoscreen (Lille, France) for sequencing, using a 454 GS FLX Roche instrument.Raw data of 454 pyrosequencing were submitted to Sequence Read Archive within the Bioproject SRP070752.

Sequence and data analysis
Raw data were split in different fasta files after removal of primer and barcode sequences.The filtered sequences were trimmed and denoised using Mothur v.1.32.0 (Schloss et al. 2009).The resulting sequences were clustered using UCLUST algorithm of USEARCH (Edgar 2010) at 97 % identity threshold to create Operational Taxonomic Units (OTUs), and singletons were excluded from further analysis.Hence OTUs were assigned to taxonomy: firstly, a Blast search against UNITE database was performed (e-value=1.10 -5 ) and non-Glomeromycota sequences were excluded; Secondly, the EPA algorithm of RAxML (Berger et al. 2011) was used to correct and improve the taxonomic assignment of the OTUs to Molecular Taxa at the specieslevel wherever possible.In each sample, 1200 sequences were subsampled from the total number of sequences obtained to avoid an overestimation of the diversity in some of the samples.
Rarefaction curves, diversity indices (Shannon) and richness indices (Chao1) of species-level MTs were obtained using EstimateS software v.9.0.0 (Colwell 2013).Statistic Shannon diversity differences among plots were tested using Kruskal-Wallis test in the stats package implemented in R software (version 2.15.1, R Core Team, 2013).Bray-Curtis distances were calculated from AMF abundance tables at the OTUs level and two MT-levels (genus and species).
Variations of species-level MT abundances among plots were tested using a pairwise Wilcoxon test with Bonferroni correction from the stats package implemented in R software.Soil physical properties were log-transformed to normalize data, and differences among plots were statistically tested by ANOVA using stats package implemented in R software.
RESULTS AND DISCUSSION

Arbuscular mycorrhiza colonization rate
AMF colonization rates were calculated to analyze whether agricultural practices influence the proportion of mycorrhiza in vine roots.Intensity of mycorrhizal colonization, as well as the abundance of the arbuscules, which are the fungal organs active in plant-fungi exchanges, were calculated using the method described by Trouvelot et al. (1986), based on root staining and microscopy.Results showed that in all vineyard plots, arbuscular mycorrhiza was present in vine roots (Suppl.Figure 1), which is the first systematic evidence for this fact in Burgundy vineyards, whatever soil properties or cultural practices.The intensity of mycorrhizal colonization and arbuscule abundance was varying among vineyard plots (Suppl.Figure 1).However, due to the large variability of arbuscule abundance and intensity of mycorrhization in roots in the plots, no correlation with cultural practices could be determined.A significantly lower mycorrhizal colonization was found in the plots in Yonne, with soils with the highest clay content.Lower colonization rate in clay soils has been linked to a lower amount of spores (Aliasgharzadeh et al. 2001, Mathimaran et al. 2005), probably due to lower oxygen concentration or mechanical barriers affecting hyphal growth.At the same time, these samples were the only ones from rootstock 41B.It is well established that different rootstocks may react differently to AMF (Linderman and Davis 2001), thus in this case it remains unclear whether soil or rootstock are responsible for the low colonization.
Root colonization levels by AMF were previously shown to be influenced by plant species and fertilization level (Zubek et al. 2012), or soil biochemical properties (Zaller et al. 2011, Chmura and Gucwa-Przepióra 2012, Krishnamoorthy et al. 2014).In the present study, soil micro-and macroelements, soil structure and chemical properties were analyzed for all the plots.Each parameter was set into relation to arbuscular mycorrhiza colonization rates.Significant correlations (P<0.05) were observed between mycorrhizal frequency (F%) and C/N ratio (r=0.33),CEC (r=-0.38),magnesium (r=-0.46)and calcium (r=-0.34)soil content.The only significant positive correlation of the intensity of mycorrhizal colonization (M%) with soil parameters was found for copper levels (Cu; r=0.38,P<0.05), whereas arbuscule abundance (A%) was not influenced significantly at all by soil chemical properties.Nutrient uptake is generally known to be enhanced by AMF due to better soil exploration, but the majority of studies has concentrated on P uptake.
Copper is essential for plant photosynthesis and Cu nutrition in plants has been shown to be improved by the AMF symbiosis (Lehmann and Rillig 2015).Conversely, AMF can reduce Cu toxicity for plants in contaminated soils through several copper transporters (Tamayo et al. 2014).Copper is frequently used for disease prevention in vineyards and is the only molecule allowed in organic viticulture, resulting in elevated Cu levels in many vineyard soils.AMF are highly tolerant to this metal and may play a major role in regulating plant copper content in vineyards, especially in organic culture.

AMF diversity across Burgundy
To analyze AMF diversity, a high-throughput sequencing method was used.After denoising, 252704 sequences classified in 2300 clusters were obtained.Only samples with at least 2100 sequences were kept to analyze AMF species diversity.Results showed, in agreement with root staining results, that samples from Yonne had a low number of AMF sequences, and therefore they were not included in the following analyses.Concerning the samples analyzed, a relatively low diversity at the phylogenetic level of the order was observed in the vine roots, and 99 % of the sequences belonged to the Glomerales, an order known to be highly represented in vine roots (Schreiner and Mihara 2009, Balestrini et al. 2010, Lumini et al. 2010).Among them, six molecular taxa (Claroideoglomus sp., Glomeraceae sp. 1, 4 and 5, Glomerales sp. and Rhizophagus irregularis) correspond to 77% of the total number of sequences.They were found in all monitored plots in a large proportion, so it can be hypothesized that they are generally present in Burgundy vineyards as core set of AMF species.Interestingly, the majority of clusters could not be assigned to known species but only to the genus or family level, as no close matches were available in the sequence databases, suggesting that these may be new to science.No clear trend was observed for a correlation between AMF species composition or diversity index found in the roots on one hand and cultural practices (Figure 2) or rootstock type on the other hand.Therefore, the combination of the variations between plots (soil, local climatic conditions, etc) has a stronger effect on AMF community composition than the cultural practices or rootstock type.The location was also shown to influence AMF diversity but not the cropping system in Croatian vineyards (Radic et al. 2014), which suggests the importance of local environment on AMF diversity in vineyards.The AMF species Rhizophagus irregularis was found in all the plots, even if the proportion of this species was affected by organic carbon, organic matter and total nitrogen soil content (P<0.05).This species was reported to be ubiquitous with its ecotypes adapted to various environments (Börstler et al. 2008).In grapevine, this species was previously found in AMF communities from soil or roots (Balestrini et al. 2010, Lumini et al. 2010, Holland et al. 2014).Therefore, it would be interesting in the future to use the molecular markers developed by Börstler et al. (2008) to study intraspecific diversity of this species in relation to cultural practices, in particular in the comparison between grass-covered vineyards and tilled vineyards.In this project we had the opportunity to study a plot inoculated 17 years ago with Funneliformis mosseae.However, this AMF species, which was also never found in the other plots, was not detected in the inoculated plot.In fact, no species from this widespread genus were detected.Funneliformis mosseae is known as a disturbance colonizer with an r-type strategy, a quick sporulation and colonization of soil and roots (Sýkorová et al. 2007), therefore it is frequently found in arable fields.Our results suggest that even if it may have a role in the first colonization stages, it may not persist on the long term, unlike Rhizophagus irregularis.The majority of taxa found in vine roots may indeed be Kstrategists, well adapted to competition in a system stable for a long time.Overall, a large variability in AMF diversity was found among the different plots, but these variations are multifactorial, including soil properties, cultural practices and host genotype.A simplified sampling design taking into account the results presented here, using a limited number of variable factors in a same plot, could increase our understanding of the impact of these factors on AMF communities.

CONCLUSION
Here we report the first large-scale analysis of the diversity of the arbuscular mycorrhiza in French vineyards using high-throughput sequencing.In spite of the relatively low diversity observed and a significant proportion of molecular taxa shared between the sites, some differences in the AMF community composition were observed between the plots.By identifying the spectrum of fungal taxa involved, this study contributes to a better understanding of ecology of these fungi in vine roots, thus providing essential knowledge for future applications in sustainable agriculture in vineyards.
Céline Berthier (Institut Français de la Vigne et du Vin) for her helpful participation in this experimentation.

Figure
Figure 1 -Map with the location of the 16 vineyards in Burgundy.

Figure 2 -
Figure 2 -Number of sequences found for the Glomeromycota taxa identified in each monitored plot.Species with statistically different number of sequences between plots are indicated with *, after Kruskal-Wallis test and P<0.05.Taxa were defined at species level wherever possible, and the box named "other Glomeromycota" corresponds to taxa containing less than 1% of the total number of sequences.