Arbuscular mycorrhizal fungi composition in soils of switchgrass harvested for bioenergy under 1 alternate nitrogen management

Learning more about the biodiversity and composition of arbuscular mycorrhizal fungi (AMF) 18 under alternative agricultural management scenarios may be important to the sustainable 19 intensification of switchgrass grown as a bioenergy crop. Using PacBio single-molecule 20 sequencing and taxonomic resolution to the level of amplicon sequence variant (ASV), we 21 assessed the effects of nitrogen amendment on AMF associating with switchgrass and explored 22 relationships between AMF and switchgrass yield across three sites of varying productivity in biodiversity our GS24 , studies treatment paired

2 Wisconsin. Nitrogen amendment had little effect on AMF diversity metrics or community 24 composition. While AMF ASV diversity was not correlated with switchgrass yield, AMF family 25 richness and switchgrass yield had a strong, positive relationship at one of our three sites. Each 26 of our sites was dominated by unique ASVs of the species Paraglomus brasilianum, indicating 27 regional segregation of AMF at the intraspecific level. Our molecular biodiversity survey 28 identified putative core members of the switchgrass microbiome as well as novel clades of AMF, 29 especially in the order Paraglomerales and the genus Nanoglomus. Furthermore, our phylogenies 30 unite the unknown, cosmopolitan, soil-inhabiting clade GS24 with Pervetustaceae, an enigmatic 31 family prevalent in stressful environments. Future studies should isolate and characterize the 32 novel genetic diversity found in switchgrass agroecosystems and explore potential yield benefits 33 of AMF richness. 34 to characterize AMF, which fail to detect certain clades while masking the extensive functional 92 diversity present among AMF after over 500 million years of evolution (28). For example, AMF 93 staining procedures to measure root colonization and nutrient-transfer structures do not detect 94 AMF in the order Paraglomerales, which have weakly staining or completely non-staining 95 tissues (29-30). Furthermore, morphological metrics have no inherent meaning for plant 96 performance given that AMF species exist along a symbiotic spectrum ranging from mutualism 97 to parasitism (31-32). In fact, alternative strains within a given AMF species can result in 98 completely different outcomes for plant performance (33-35). In addition, the 16:1ω5cis fatty 99 acid biomarker commonly used to measure AMF abundance in soils (36-37) is only present at 100 miniscule quantities or is completely absent from species in Paraglomeraceae and 101 Gigasporaceae (38). On the other hand, fatty acid indicators for these two families (e.g., 102 16:1ω7cis, 18:1ω9cis, 20:1ω9cis) are typically treated as biomarkers for saprotrophic fungi and 103 gram-negative bacteria, conflating the abundance of multiple guilds (38-40). Finally, AMF 104 metabarcoding of the small subunit ribosomal DNA (rDNA) gene region with short-read 105 sequencing platforms does not discriminate AMF to the species level (27, 41-42). This problem 106 is compounded when operational taxonomic units are clustered at 97% similaritya 107 conservative threshold and arbitrary proxy for species-level delimitation of filamentous fungi 108 (43)(44). 109 In the context of a long-term, replicated bioenergy cropping systems experiment, we 110 conducted a metabarcoding study of AMF using PacBio Sequel Single Molecule, Real-Time 111 6 highly variable regions of rDNA for an unprecedented opportunity to investigate AMF at a fine 115 taxonomic resolution (42). ASVs are exact, error-free sequences and thus serve as biologically 116 meaningful representatives of AMF genotypes, revealing intraspecific variation and strains of 117 interest for bioprospecting (9, 43). To our knowledge, this study is the first to employ PacBio 118 Sequel SMRT Sequencing with ASVs in the analysis of AMF communities and one of only a 119 handful to use PacBio for AMF metabarcoding (18,41,(45)(46)(47). We asked the following 120 questions: Does long-term N amendment affect AMF ASV diversity or community structure in 121 Wisconsin switchgrass agroecosystems? Does AMF diversity explain variation in switchgrass 122 yield? We hypothesized that N amendment would decrease AMF diversity (27,48) and that 123 AMF diversity would be correlated with switchgrass yield because of its positive linkage to 124 ecosystem functioning (3). 125 126 Methods 127

Sites and experimental design 128
The US Department of Energy's Great Lakes Bioenergy Research Center designed a 129 Marginal Lands Experiment to test the efficacy of growing perennial biofuel feedstock crops on 130 abandoned agricultural fields or land deemed unsuitable for high-productivity agriculture. This 131 study focused on three Wisconsin sites: Rhinelander in northern Wisconsin, Hancock in the 132 central part of the state, and Oregon in the south (Fig. S1). The coordinates, soil texture, and 133 mean soil physicochemical properties of the three sites are presented in Table S1; for a 134 comprehensive assessment of the soils at each site, see Kasmerchak and Schaetzl (49)

Taxonomy and phylogeny of ASVs recovered from switchgrass agroecosystems 261
A total of 190,700 high-quality, non-chimeric amplicons clustered into 436 ASVs (for a 262 complete breakdown of the number of sequences remaining after each bioinformatics processing 263 step, see Table S2). Rarefaction curves of ASV accumulation versus sample size reached a clear 264 asymptote for every experimental unit, indicating relatively complete sampling of AMF diversity 265 in taxonomic assignment showcases both the biological (e.g., myriad undescribed taxa in the 285 Nanoglomus plukenetiae sensu lato clade) and artificial (e.g., incomplete data curation of 286 Paraglomerales) issues that hinder accurate identification of AMF environmental sequences 287 using even curated reference databases. 288 A few clades stood out for their high sequence abundance or consistent presence across 289 the experimental units, which suggests a role in the core microbiome of switchgrass ( Fig. 2; 76). 290 The previously mentioned clade encompassing Nanoglomus plukenetiae included ASVs with 291 some of the greatest absolute abundances (ASVs 1, 4, and 9). While no ASV belonging to this 292 clade occurred at more than one site, representatives of the Nanoglomus plukenetiae sensu lato 293 clade were present in both N-amended and control plots at all three sites (Figs. S3-S4). In total, 294 64 of the 436 ASVs (~15%) belonged to this diverse but poorly understood clade, which may 295 represent various undescribed species in the genus Nanoglomus or new genera within 296 on November 4, 2020 by guest http://aem.asm.org/

Downloaded from
Glomeraceae. Within Paraglomerales, 73 ASVs (~17%) had high LSU sequence similarity to 297 Paraglomus brasilianum and 51 more (~12%) to Paraglomus laccatum. Since ASVs belonging 298 to these species also occurred at all three sites (Figs. S3-S4), Paraglomus spp. appear to be 299 important members of the core switchgrass microbiome as well. Of the 436 ASVs recovered in 300 this study, only three ASVs occurred at all three sites, two of which belonged to Paraglomus 301 brasilianum (ASVs 8 and 27); the last -ASV 78belonged to Microdominikia litorea, another 302 potential member of the core switchgrass microbiome (Fig. 2). 303 In addition to the Nanoglomus plukenetiae sensu lato clade, some of the less-abundant 304 clades identified in this study might be undescribed lineages of AMF. Our phylogenetic analysis 305 showed that the unknown, worldwide, soil-inhabiting clade deemed GS24 (77) corresponds to 306 the family Pervetustaceae in Paraglomerales, which contains one species, Pervetustus simplex, 307 described in 2017 from the deserts of Oman (29). We recovered seven ASVs in three clades 308 belonging to this enigmatic family, two of which were well-supported (Fig. 3). These ASVs were 309 most prevalent in the sandy soils of our Hancock site, indicating they may confer some degree of 310 stress tolerance to switchgrass under water-limiting or nutrient-poor conditions (76). Finally, two 311 ASVs -ASVs 187 and 427were phylogenetically resolved as basal to the rest of 312 Paraglomeraceae, suggesting a new genus or a new family in Paraglomerales might be 313 necessary to accommodate this newfound genetic diversity (Fig. 3). 314

Relationship between soil variables and switchgrass yield, and indicator taxa 326
No metric of AMF ASV diversity was a significant predictor of switchgrass yield at any 327 site in linear mixed effects regression models including N treatment. At the family level, AMF 328 richness was a marginally significant predictor of yield at Oregon (F 1,4.92 = 21.3, p = 0.006, R c 2 = 329 0.88); every additional family represented in the AMF community at Oregon was associated with 330 an increase in 0.61 Mg ha -1 of switchgrass yield (Fig. 7). There was also a significant 331 relationship between switchgrass yield and cation exchange capacity at Hancock ( information exists on other taxa prevalent in our study (e.g., Paraglomerales), limiting our ability 391 to determine whether community structure in relation to soil N was a function of morphology 392

Positive correlation trend between yield and AMF family richness at most productive site 393
We found no universal relationships between AMF metrics and switchgrass yield at the 394 ASV level. Indeed, improved taxonomic resolution may not allow one to detect relationships 395 between AMF and plant productivity if fine-scale taxonomic groupings of AMF are not good 396 proxies for functional trait differences. However, at the family level in our most productive site 397 (Oregon), we detected a relationship between switchgrass yield and AMF family richness. Along 398 with N treatment and plot, these variables could account for 88% of the variation in switchgrass 399 yield. In another bioenergy study on productive soils, Emery et al.

Switchgrass soil microbiome dominated by poorly studied taxa 425
Our understanding of AMF is biased towards those that are most prevalent in disturbed 426 habitats (90). Concerted efforts to understand uncultured fungi in less disturbed environments 427 could benefit perennial agriculture as a whole given that these agroecosystems might harbor 428 AMF communities that are more similar to non-agricultural ecosystems. In addition to the 429 numerous undescribed clades identified in this study, little is known about the ecology of the 430 putative core AMF of the switchgrass microbiome.