No abstract

Abstract1. It is known that tropical grasslands such as Serengeti host large populations of arbuscular mycorrhizal (AM) fungi and that they respond to abiotic and biotic factors.It is also known that AM symbioses are important for the uptake of essential plant nutrients, which, in turn, influences the biomass and nutritional quality of herbivores and their predators. The purpose of this study was to investigate the influence of AM symbioses on the biomass of different trophic levels of an ecosystem.2. To do this, we first measured the neutral lipid fatty acid biomarker 16:1ω5 to estimate the biomass of AM fungi in a long-term grazing exclusion experiment. Then, we used model selection of Bayesian linear regressions to infer the primary factors that influence AM fungal biomass. Using model selection of different combinations of soil characteristics, we selected the best model using the leave-one-out crossvalidation information criterion. Finally, we used the Madingley model to simulate the influence of AM fungi on higher trophic levels. We combined spatially explicit information about soil phosphorus and AM fungal biomass to explore the emergent patterns of the Serengeti resulting from AM symbioses.3. Our Bayesian analysis indicated that total soil phosphorus was the strongest predictor of AM fungal biomass, and there were significant interactions with grazing.Arbuscular mycorrhizal fungal biomass is lowest in soil where phosphorus is limited and increases with increasing phosphorus concentration. Biomass was also significantly higher in plots that were not grazed. The Madingley model indicated that nutritional benefits of AM symbioses maintain a substantial proportion of the biomass across all trophic levels. 4. Synthesis. Our analysis shows that inputs of phosphorus through arbuscular mycorrhizal symbioses substantially increase the ability of plants to grow and maintain nutritional quality, cascading through the biomass of consumers and predators in the ecosystem. Although they account for less than 1% of the total modelled biomass, the predicted nutritional benefit provided by arbuscular mycorrhizal fungi increased the biomass of macro-organisms in the Serengeti by 48%. When considering the management of biodiversity, future ecosystem models should account for the influence of arbuscular mycorrhizal fungi on all trophic levels.

33The diversity of complex microbial communities can be rapidly assessed by high- defining OTUs depended on the method used for OTU picking. Our "default" analysis in QIIME 45 overestimated mock community OTU diversity by at least a factor of ten. Our optimized analysis

33The diversity of complex microbial communities can be rapidly assessed by high- defining OTUs depended on the method used for OTU picking. Our "default" analysis in QIIME 45 overestimated mock community OTU diversity by at least a factor of ten. Our optimized analysis

Illumina MiSeq is the current standard for characterizing microbial communities in soil. The Oxford Nanopore Technologies MinION sequencer is quickly gaining popularity because of the low initial cost and longer sequence reads. However, the accuracy of MinION, per base, is much lower than MiSeq (95% versus 99.9%). The effects of this difference in base-calling accuracy on taxonomic and diversity estimates remains unclear. We compared the output of short MiSeq to short and full-length MinION 16S rRNA amplicons from a mock community and agricultural soil samples. For all three methods, we found that taxonomic assignments of the mock community at both the genus and species level matched expectations with minimal deviation (genus: 80.9–90.5%; species: 70.9–85.2% Bray-Curtis similarity); however, the short MiSeq with error correction (DADA2) resulted in the correct estimate of species richness and much lower alpha diversity for soils. Several filtering strategies were tested to improve these estimates with varying results. The sequencing platform also had a significant influence on the relative abundances of taxa with MiSeq resulting in significantly higher abundances Actinobacteria, Chloroflexi, and Gemmatimonadetes and lower abundances of Acidobacteria, Bacteroides, Firmicutes, Proteobacteria, and Verrucomicrobia compared to the MinION platform. When comparing agricultural soils from two different sites (Fort Collins, CO and Pendleton, OR) methods varied in the taxa identified as significantly different between sites. At all taxonomic levels, the full-length MinION method had the highest similarity to the short MiSeq method with DADA2 correction with 73.2%, 69.3%, 74.1%, 79.3%, 79.4%, and 82.28% of the taxa at the phyla, class, order, family, genus, and species levels, respectively, showing similar patterns in differences between the sites. In summary, although both platforms appear suitable for 16S rRNA microbial community composition, biases for different taxa may make the comparison between studies problematic; and even with a single study (i.e., comparing sites or treatments), the sequencing platform can influence the differentially abundant taxa identified.