Temporal and Cultivar-Speciﬁc E ﬀ ects on Potato Root and Soil Fungal Diversity

: The soil fungal community plays an important role in determining plant growth and health. In this study, we investigated the fungal diversity and community composition in the roots and soil of 21 potato ( Solanum tuberosum L.) cultivars using high-throughput sequencing at three di ﬀ erent time points across the growing season. In soil and roots, the fungal richness and relative abundance of pathogens and saprotrophs were mainly a ﬀ ected by sampling time. While sampling time a ﬀ ected fungal composition in soil, root fungal communities were also signiﬁcantly a ﬀ ected by cultivar. The cultivar had the strongest e ﬀ ect on diversity of pathogens and abundance of particular pathogen species. Our results demonstrate changes in soil and root fungal communities of potato over the growing season, as well as highlighting the importance of potato cultivar on root fungal communities and abundance of pathogens.


Introduction
Modern agriculture has increased crop yields significantly [1]; however, it often relies on chemical fertilisers and pesticides [2], and thus can negatively impact the environment [3], including soil ecosystems [4][5][6]. Soils harbour millions of microorganism species, many of which form intimate associations with plant roots [7,8]. These associations may have direct, i.e., intimate mutualistic or pathogenic interactions with plants, or indirect effects, through the action of free-living microbes that affect nutrient availability in the surrounding environment [9]. Both biotrophic and saprotrophic fungi can affect plant productivity.
In agricultural fields, plant-associated and saprotrophic microbial communities are affected by different management practices such as crop rotation, fertilisation and tillage [10][11][12][13][14]. Different plant species and different growth stages may harbour distinct microbial communities [15][16][17]. Furthermore, plants modify their associated and surrounding microbial communities by exuding organic compounds into the surrounding environment, which act as nutrient sources for microbes [18,19]. Higher fertilisation rates may increase root exudation as well as microbial abundance.

Sampling and DNA Extraction
Root samples were collected at the early flowering stage (BBCH [60][61][62] and during senescence (BBCH 93-95) [31]. Cultivars were classified as early-, medium-or late-maturing, based on their maturity period. Maturity classes comprised eight early-, seven intermediate-and six late-maturing cultivars (See Table S1 for list of cultivars). Based on the different flowering times of potato cultivars, root samples were collected on 9 July, 16 July and 22 July. During the plant senescence stage (SSC), root samples of all cultivars were collected on 26 August. The root systems of three individual plants from each plot were randomly chosen and collected using a clean shovel. Roots were washed three times with deionised water to remove residual soil, dried at 50 • C for 24 h, as described in García de León et al. [32], and stored air-tight at room temperature until molecular analysis. Soil samples were collected three times over the study period: during planting (BEM), and again during flowering (FL) and senescence (SSC), simultaneously with root sampling. Each sample consisted of three subsamples of soil taken randomly from the potato root zone at 0-25 cm depth. At BEM, roots were not developed and therefore, these samples represent bulk soil. The subsamples were pooled and air-dried at <35 • C following Tedersoo et al. [33], mixed thoroughly and subjected to molecular analysis.
Total DNA was extracted from 0.075 g dry weight of roots and 0.2 g dry weight of soil, using the PowerSoil DNA Isolation Kit (MO BIO Laboratories Inc., Carlsbad, CA, USA), following manufacturer's instructions with the following modifications: (1) samples were homogenised via bead beating with three 3 mm autoclaved steel beads for 5 min at 30 Hz, using a MixerMill MM400 (Retsch, Haan, Germany); and (2) final elution was performed twice with 50 µL solution C6.

PCR Amplification and High-Throughput Sequencing
The ITS2 region was amplified using ITS3mix1-5 (mixture of six forward primers in equimolar concentration analogous to ITS3) and a degenerate reverse primer ITS4ngs, which was tagged with one of the 108 multiplex identifiers [33]. PCR amplification was performed in a 25 µL reaction volume, and consisted of 18 µL nuclease-free water, 5 µL 5× HOT FIREPol Blend Mastermix (10mM MgCl2) (Solis Biodyne, Tartu, Estonia), 0.5 µL of each primer (20 pmol) and 1 µL DNA extract. PCR was performed using Eppendorf 5341 and Eppendorf 6321 thermal cyclers (Eppendorf AG, Hamburg, Germany) in four replicates, under the following thermocycling conditions: 15 min at 95 • C, 35 cycles of 30 s at 95 • C, 30 s at 55 • C, 1 min at 72 • C and 10 min at 72 • C. PCR products were pooled, and their relative quantities were estimated by running 5 µL of amplicon DNA on 1% agarose gel (SeaKem LE Agarose, Lonza Group Ltd., Basel, Switzerland) for 15 min. PCR products were pooled for library preparation on the basis of relative band strength, as visualized on an agarose gel, which is indicative of amplicon concentration. To obtain sufficient PCR product, DNA samples yielding no visible band, and samples with a very strong band, were re-amplified by altering the number of cycles. The quantities of PCR products were normalised with SequalPrep Normalisation Plate Kit (Invitrogen, Carlsbad, CA, USA). The sequencing libraries were prepared using a Nextera XT kit (Illumina Inc., San Diego, CA, USA). All samples were sequenced on an Illumina MiSeq instrument at the Estonian Genome Center (University of Tartu, Tartu, Estonia).

Bioinformatics
Paired-end sequencing (2 × 300 bp) resulted in 772,326 paired reads. Sequencing reads were quality-filtered and assigned to samples using mothur 1.34.4 [34] (average quality over 15 bp ≥ 26, and no ambiguities allowed). The quality-trimmed data were assembled using PANDAseq Assembler [35], with a minimum overlap of 15 bp, and demultiplexed in mothur. Potential chimeric sequences were removed using USEARCH 7.0.1090 [36]. The remaining chimeric sequences, where full primer strings were detected inside the reads, were removed using PipeCraft in-built module (remove multiprimer artefacts) [37]. The ITS2 subregion was extracted using ITSx 1.0.9 [38], and clustered using a 97% similarity threshold in CD-HIT [39]. Singleton OTUs were removed from further analyses. Representative sequences for BLASTn search were picked in mothur using the abundance method. In addition, BLASTn searches were performed for the representative sequence of each OTU against the UNITE reference dataset v7.0 [40]. OTUs were further checked and filtered based on BLASTn search values as well as positive and negative controls to remove contaminants, non-fungal OTUs, potential artefacts and index-switching errors. OTUs with 75%, 80%, 85%, 90%, 95%, and 97% sequence similarity thresholds were considered to represent the phylum, class, order, family, genus, and species level, respectively [33]. Based on taxonomic assignments, OTUs were parsed to one of the following functional groups of fungal guilds: plant pathogenic fungi, saprotrophic fungi and arbuscular mycorrhizal fungi, based on FUNGuild [41]. OTUs not assigned to plant pathogens by FUNGuild, but reported as potato pathogens according to the United States Department of Agriculture (USDA) Agricultural Research Service (U.S. National Fungus Collections Fungus-Host Database. Available online: https://nt.ars-grin.gov/fungaldatabases/), were additionally assigned as pathogens. The raw data of this study are available through the Sequence Read Archive, BioProject PRJNA638263.

Statistical Analysis
We calculated linear regression between a number of obtained OTUs per sample and square root of the obtained sequences per sample, in order to control for variation in sequencing depth [33]. The standardised residuals of OTU richness were used as a proxy for fungal richness. In roots, linear regression explained 51%, 40.9%, and 36% of variation in OTU richness of all fungi, pathogens, and saprotrophs, respectively. In soil, square-root of sequencing depth explained 72.1%, 20.6%, and 67.2% of variation in OTU richness of all fungi, pathogens and saprotrophs, respectively. The relative abundance of fungal guilds was calculated as a number of sequences corresponding to the particular guild divided by the total number of fungal sequences for each sample.
Differences in richness and relative abundance of total fungal communities, as well as those of fungal guilds, were tested using GLM (Type III SS) followed by Tukey HSD post hoc tests comparing the means of standardised residuals of OTU richness and relative abundance (α = 0.05; Statistica 12.0, Palo Alto, CA, USA). Explanatory variables included potato cultivar, plant growth stage (fixed factor with levels BEM, FL, SSC), time × cultivar interaction, and replicate block (random factor with three levels). PERMANOVA+ [42], implemented in Primer 7 software (PRIMER-E, Auckland, New Zealand), was used to study the community composition of both the overall fungal community, as well as that of separate fungal guilds. Three samples (LK129, LK130 and LK131) were removed from soil total fungal and saprotroph datasets, as they contained only a few sequences and OTUs. PERMANOVA+ tests were carried out with 9999 permutations under the reduced model. The read abundance was standardised by samples and fourth-root transformed before calculating the Bray-Curtis dissimilarity index. Adjusted R 2 values were calculated using the function RsquareAdj in the package "vegan" in R 3.6.0 (R Development Team, http://www.R-project.org). Trends in fungal composition were visualised via non-metric multi-dimensional scaling (NMDS) ordinations, as implemented in the metaMDS function in both "vegan" and "ggplot" packages.

Results
The quality-filtered sequence dataset comprised 224,195 high quality, full length ITS2 reads in 315 samples, with an average number of 712 sequences per sample. These sequencing reads were assigned to 1655 fungal OTUs, with 80 OTUs in root samples and 110 OTUs in soil samples on average (Table S4). In roots, Mortierellaceae was the most abundant family observed in the FL stage, while Plectosphaerellaceae was the most dominant in the SSC stage ( Figure 1A). In soil, Mortierellaceae prevailed in all growth stages ( Figure 1B). In roots of each cultivar, Plectosphaerellaceae, Ceratobasidiaceae, Microdochiaceae and Nectriaceae Lasiosphaeriaceae, Filobasidiaceae and Nectriaceae were among the most predominant fungal families ( Figure 2A). Mortierellaceae had the highest relative abundance in soil of each cultivar ( Figure 2B). Of all sequences, 24.1% were assigned to plant pathogens, whereas saprotrophs and arbuscular mycorrhizal fungi accounted for 44.0% and 0.6% of sequences, respectively. Similar patterns occurred in root and soil samples. Due to their lack of detection from several samples, diversity patterns of arbuscular mycorrhizal fungi are not reported. Furthermore, no correlation studies between beneficial and pathogenic fungi were carried out.

Richness of Fungal Guilds
The highest proportion of variance in root fungal richness was explained by sampling time (F1,82 = 43.75, R 2 adj = 0.184, p < 0.001; Table 1), followed by the sampling time × cultivar interaction (F20,82 = 2.85, R 2 adj = 0.106, p < 0.001) and cultivar (F20,82 = 2.20, R 2 adj = 0.038, p = 0.007). Post hoc analyses showed that the cultivar Merlot had significantly higher total fungal richness than cultivars Concordia, Solist and Laudine (Figure 3a), and revealed that in the FL stage, Merlot had significantly  Of all sequences, 24.1% were assigned to plant pathogens, whereas saprotrophs and arbuscular mycorrhizal fungi accounted for 44.0% and 0.6% of sequences, respectively. Similar patterns occurred in root and soil samples. Due to their lack of detection from several samples, diversity patterns of arbuscular mycorrhizal fungi are not reported. Furthermore, no correlation studies between beneficial and pathogenic fungi were carried out.

Richness of Fungal Guilds
The highest proportion of variance in root fungal richness was explained by sampling time (F1,82 = 43.75, R 2 adj = 0.184, p < 0.001; Table 1), followed by the sampling time × cultivar interaction (F20,82 = 2.85, R 2 adj = 0.106, p < 0.001) and cultivar (F20,82 = 2.20, R 2 adj = 0.038, p = 0.007). Post hoc analyses showed that the cultivar Merlot had significantly higher total fungal richness than cultivars Concordia, Solist and Laudine (Figure 3a), and revealed that in the FL stage, Merlot had significantly Of all sequences, 24.1% were assigned to plant pathogens, whereas saprotrophs and arbuscular mycorrhizal fungi accounted for 44.0% and 0.6% of sequences, respectively. Similar patterns occurred in root and soil samples. Due to their lack of detection from several samples, diversity patterns of arbuscular mycorrhizal fungi are not reported. Furthermore, no correlation studies between beneficial and pathogenic fungi were carried out.

Plant Pathogen, and Saprotroph Abundance
The highest proportion of variance in relative pathogen abundance in roots was explained by sampling time (F1,82 = 66.79, R 2 adj = 0.296, p < 0.001; Table 2), with higher relative pathogen abundance in the SSC stage compared to FL stage. Cultivar also had a significant effect (F20,82 = 1.81, R 2 adj = 0.004, p = 0.033; Table 2), with Rosagold having a significantly higher fungal pathogen abundance than

Plant Pathogen, and Saprotroph Abundance
The highest proportion of variance in relative pathogen abundance in roots was explained by sampling time (F 1,82 = 66.79, R 2 adj = 0.296, p < 0.001; Table 2), with higher relative pathogen abundance in the SSC stage compared to FL stage. Cultivar also had a significant effect (F 20,82 = 1.81, R 2 adj = 0.004, p = 0.033; Table 2), with Rosagold having a significantly higher fungal pathogen abundance than Catania (Table S8). Sampling time had the strongest effect on soil pathogen abundance (F 2,124 = 37.70, R 2 adj = 0.250, p < 0.001; Table 2), with highest values at the SSC stage, and lowest values at the BEM stage (Table S8). Cultivar also had a significant effect on soil pathogen abundance (F 20,124 = 1.69, R 2 adj = 0.011, p = 0.004; Table 2), which was non-significant in post hoc analyses (Table S9). The highest proportion of root saprotroph abundance was explained by sampling time (F 1,82 = 46.60, R 2 adj = 0.242, p < 0.001; Table 2), with the plant FL stage exhibiting significantly higher saprotroph abundance. Furthermore, sampling time had a significant influence on soil saprotroph abundance (F 2,124 = 41.02, R 2 adj = 0.273, p < 0.001) ( Table 2). The highest saprotroph abundance was observed at the BEM stage, compared to both the FL (post hoc; p < 0.001; Table S9) and SSC stages (post hoc, p < 0.001; Table S9). Cultivar also had a weak but significant influence on soil saprotroph abundance (F 20,124 = 1.69, R 2 adj = 0.012, p = 0.040; Table 2). Post hoc analysis revealed that the cultivar Viviana had significantly higher saprotroph abundance than Manitou (Table S9).
In the soil samples, Gibellulopsis nigrescens (5.2%), P. cucumerina (2.4%), Didymellaceae (1.4%) and Fusarium spp. (0.9%) were the most abundant pathogen taxa. Similarily to overall pathogen abundance in root samples, sampling time had the strongest influence on the relative abundance of G. nigrescens (F 2,124 =9.34, R 2 adj = 0.071, p < 0.001; Table S10), with lowest abundance at the BEM stage, compared to FL and SSC stages (Table S12). Sampling time affected the relative abundance of P. cucumerina (F 2,124 = 32.70, R 2 adj = 0.251, p < 0.001; Table S10) and Didymellaceae (F 2,124 = 10.5, R 2 adj = 0.103, p < 0.001; Table S10). The lowest P. cucumerina abundance was observed at the BEM compared to FL and SSC stages (Table S12). The lowest Didymellaceae abundance was observed at both BEM and FL stages compared to the SSC stage (Table S12). In contrast, cultivar was the only significant variable affecting Fusarium spp. abundance (F 20,124 = 1.70, R 2 adj = 0.060, p = 0.040; Table S10), but post hoc analyses revealed that Fusarium spp. abundance in soil near the roots of Merlot is only marginally higher than that of Karlena (Table S12).

Dominant Taxa
Plant soil offers habitat niches and root exudates as nutrients for microorganisms [43,44]. In the present study, root samples harboured only one fifth of the total OTUs detected, suggesting that a few microorganisms overcome plant defense mechanisms and inhabit plant roots. Ascomycota was the most abundant phylum detected in roots for each studied time point. Furthermore, ascomycetes prevailed in the roots of all potato cultivars. Ascomycota is the most diverse fungal phylum, comprising the majority of plant pathogens [45]. Furthermore, ascomycetes dominate as decomposers of organic matter in agro-ecosystems [46]. Mortierellomycota dominated in the soil-this phylum includes the genus Mortierella, mold-like decomposers that contribute to soil phosphorus cycling [47,48].
The observed dominant plant pathogens are common pathogens of potato, with a capability for saprophytic growth and infective spread to other hosts. Plectosphaerella cucumerina prevailed both in soil and root samples, which is in agreement with previous studies in Italy [49,50]. Although this pathogen causes wilting in potato [51], some studies have shown that P. cucumerina acts as a biocontrol agent against potato cyst nematodes [52]. Other abundant pathogens, belonging to genera such as Fusarium and Rhizoctonia, are common potato pathogens causing dry rot and black scurf, respectively. These soil-borne genera are among the most economically-important plant-pathogenic fungi, [26,53,54] that can survive saprophytically on crop residues in the absence of their hosts, growing rapidly when fresh organic matter is available [54]. Here we considered both genera to be tentatively pathogenic, although these groups contain non-pathogenic endophytes and saprotrophs and pathogens on hosts other than potato [55][56][57]. Limited species-level and forma speciales-level resolution of the ITS marker in Fusarium [58,59] and the paucity of SH-level functional reference data in both groups hamper our ability to distinguish effectively pathogenic organisms from closely related non-pathogenic taxa.

Seasonal Variation
Sampling time was the strongest variable shaping total fungal richness and composition, both in soil and roots. The role of temporal change in both bacterial and fungal communities has been observed in other studies [60][61][62]. In the present study, overall root-fungal richness was highest in the plant FL stage, whereas in soil samples, overall fungal richness peaked during the SSC stage. These results imply that in intensively managed agricultural soil during the early stages of plant growth, fungi mainly colonize healthy plant roots, which may provide high amounts of energy to the fungi. In later stages, when pathogens accumulate and plants senesce, nutrient flow ceases. In addition, changes in fungal community composition during the growing season may also contribute to the observed temporal dynamics, possibly due to climatic conditions, time since disturbance (ploughing) and interactions among fungal taxa. During the vegetative period, organic matter content decreases, and is replaced by root exudates. A previous study by Chaparro et al. [63] showed that the surrounding rhizosphere microbiome is affected by plant developmental stage, and is related to root exudation. Plants exude organic compounds into the surrounding environment [64], and microbes use these organic compounds as nutrients; and different microbial groups have distinct nutritional preferences [65]. Therefore, these qualitative changes in root exudation may cause differences in soil fungal diversity and community composition.
Saprotroph richness in roots and soil peaked during the plant FL and SSC stages, respectively, whereas saprotroph relative abundance in both soil and roots decreased in the plant SSC stage. Saprotroph richness-peaks in the FL stage could be attributed to increases in rhizodeposits [66]. In roots, plant FL stage exhibited significantly higher saprotroph relative abundance, whereas in soil, the highest saprotroph abundance was observed at the BEM stage. Thus, it can be expected that continuous tillage operations at the beginning of the growth period, and resource-rich spring wheat residues, being relatively fibrous with more long-term decomposition, promoted saprotroph abundance in the BEM stage.
The abundance of most dominant pathogens increased over the growing season. This is in agreement with the generally observed exponential increase in disease incidence of most crop pathogens over time during the growing season. It can be speculated that, in addition to changes in exudate patterns and root architecture, plants' resistance against pathogens decreases during ageing, allowing virulent pathogens to become prevalent [67].

Effect of Cultivars
In contrast to other fungal guild and habitat combinations, root pathogen richness was mainly affected by plant cultivar. Lowest pathogen richness was observed in the cultivars Viviana, Solist and Concordia, which are considered relatively resistant cultivars with medium nutrient demand (Europlant.biz, Danespo.com). Roots of the cultivar Glorietta comprised the highest pathogen richness, but this variety is not known to be particularly susceptible to pathogens. However, higher pathogen richness may not always be related to higher disease incidence and severity, but may instead lead to microbial competition in the rhizosphere and suppress the dominant pathogens [68].
The present study confirmed previous findings [69] regarding different cultivar susceptibility to Rhizoctonia solani. Our results reveal that some cultivars show almost no occurrence of R. solani when grown in the same soils where other cultivars became infected by this fungus. According to information provided by the breeding company Norika (Norika GmbH), the cultivar Merlot has high resistance to Rhizoctonia spp. Nevertheless, we found that Merlot showed marginally stronger abundance of R. solani than other cultivars. Furthermore, in this field trial, at the end of the growing period, plants with nests of small-sized tubers caused by R. solani were only observed on Merlot. Plant cultivar was the main factor affecting overall fungal, pathogen and saprotroph community composition in roots. It can be speculated that host genotype determined the fungal mycobiome recruited from the soil into the potato roots. Furthermore, a similar, consistent and weak, yet significant, effect of genotype on bacterial community composition in plant roots was observed in previous studies [70][71][72].

Conclusions
In this study, we assessed differences in fungal community structure in the roots and soil near the root zone of 21 potato cultivars, and among different plant growth stages. Potato cultivars (genotypes) developed distinctive fungal communities in their roots, with the background of similar fungal assemblages in soil showing temporal changes over the growing season. The abundances of fungal guild representatives in roots of potato cultivars, as well as fungal guild composition and diversity, showed temporal changes. The occurrence of major pathogens strongly varied among potato cultivars. Overall, our results demonstrate that in roots, cultivar was a primary factor determining overall fungal, pathogen and saprotroph community composition.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4395/10/10/1535/s1, Figure S1: Results of a general linear model estimating the species richness and relative abundance of pathogen and saprotroph communities in root (A,C) and soil (B,D). Table S1: The layout of potato field, and the list of varieties, Table S2: characteristics of the study site. Main field operations and their timings, Table S3: Soil properties on field site, Table S4: OTU table, Table S5: Tukey post hoc test of significant variables for the overall fungal richness in roots and soil, Table S6: Tukey post hoc test of significant variables for the pathogen fungal richness in roots and soil, Table S7: Tukey post hoc test of significant variables for the saprotroph fungal richness in roots and soil, Table S8: Tukey post hoc test of significant variables for the pathogen relative abundance in roots and soil, Table S9: Tukey post hoc test of significant variables for the saprotroph relative abundance in roots and soil, Table S10: Relative abundance (%) of the four most dominant OTUs from potato roots, and rhizosphere sampled in different time points, Table S11: Tukey post hoc test of significant variables for the most abundant pathogens in roots, Table S12: Tukey post hoc test of significant variables for the most abundant pathogens in soil.