Non-mycorrhizal root associated fungi of a tropical montane forest are relatively robust to the long-term addition of moderate rates of nitrogen and phosphorus

Andean forests are biodiversity hotspots and globally important carbon (C) repositories. This status might be at risk due to increasing rates of atmospheric nutrient deposition. As fungal communities are key in the recirculation of soil nutrients, assessing their responses to soil eutrophication can help establish a link between microbial biodiversity and the sustainability of the C sink status of this region. Beyond mycorrhizal fungi, which have been studied more frequently, a wide range of other fungi associate with the fine root fraction of trees. Monitoring these communities can offer insights into how communities composed of both facultative and obligate root associated fungi are responding to soil eutrophication. Here we document the response of non-mycorrhizal root associated fungal (RAF) communities to a long-term nutrient manipulation experiment. The stand level fine root fraction of an old growth tropical montane forest was sampled after seven years of nitrogen (N) and phosphorus (P) additions. RAF communities were characterized by a deep sequencing approach. As per the resource imbalance model, we expected that asymmetries in the availability of C, N and P elicited by fertilization will lead to mean richness reductions and alterations of the community structure. We recovered moderately diverse fungal assemblages composed by sequence variants classified within a wide set of trophic guilds. While mean richness remained stable, community composition shifted, particularly among Ascomycota and after the addition of P. Fertilization factors, however, only accounted for a minor proportion of the variance in community composition. These findings suggest that, unlike mycorrhizal fungi, RAF communities are less sensitive to shifts in soil nutrient availability. A plausible explanation is that non-mycorrhizal RAF have fundamentally different nutrient acquisition and life history traits, thus allowing them greater stoichiometric plasticity and an array of functional acclimation responses that collectively express as subtle shifts in community level attributes.

Together, these responses might affect the composition of C pools belowground.

125
That is, an increased production of leaves and fine roots with shorter lifespans could 126 signal an increased proportion of recalcitrant C inputs, yet a reduction in the 127 exudation of more labile C compounds from living roots. Alterations of the C pool 128 composition, and the N:P ratio of the substrate are key, given fungal stoichiometry 129 appears to closely track the stoichiometry of soils in montane tropical forests 130 (Nottingham et al., 2015). Indeed, non-mycorrhizal soil fungal communities have 131 increased in biomass after the simultaneous addition of a C source together with a 132 mineral nutrient in these forests (Krashevska et al., 2010;Nottingham, Hicks, et al., 133 2018). Hence, it is reasonable to expect that asymmetric rates of nutrient addition 134 will elicit the response of RAF communities. The direction and magnitude of such 135 change, however, is still unclear.

136
To test these ideas, we surveyed the fine root fraction of an old growth tropical 137 Andean forest subjected to long-term nutrient manipulation (Homeier et al., 2012). 138 We characterized the taxonomic and trophic guild structure of RAF communities by 139 means of a meta-barcoding approach, and estimated community level attributes 140 across fertilization treatments. To further our assessments of the effects of 141 fertilization on RAF communities, we compared the relative read abundance of 142 higher level taxonomic and guild categories across treatments with a differential 143 abundance method that corrects for distortions in these parameters. We expected 144 that an increase in the availability of organic and inorganic N and P in the soil will 145 lead to 1) a reduction in RAF community diversity because fungal taxa for which the 146 altered resource availability is detrimental will be suppressed; 2) shifts in community 147 composition, not only due to reductions in diversity, but also due to the colonization 148 of empty niches or the competitive dominance of taxa with better dispersal abilities or 149 more flexible coping mechanisms; and 3) major fungal lineages or trophic guilds will 150 respond differently to fertilization in terms of adjusted relative read abundances, due 151 to evolutionary fixed differences in their abilities to obtain limiting resources.    Briefly, primer oligos were removed from every sequence using Cutadap (v.2.1, 200 Martin, 2011). Read ends were truncated if the quality score of base calls fell below 201 two; and filtered out when reads were smaller than 50 bp or presented more than 202 two erroneous or undefined base calls per sequence. Remaining reads were de-203 replicated and then subjected to the Dada2 sample inference algorithm. Denoised 204 forward and reverse reads were then merged if they had an overlap of at least 12 bp 205 between them. At the end of this process, an ASV abundance table was generated 206 followed by a chimera removal step. Hereafter, the acronym ASV and the term 207 'variant' will be used interchangeably.

208
To allow comparisons to prior studies, ASVs were clustered into 97% OTUs using , where θ is an overdispersion parameter allowing the variance to increase randomly excluded to achieve this balance (n=80).

299
Beta diversity was defined here as turnover of phylotypes across samples.   ASVs were defined as those present in at least 5 samples across the dataset. 357 Consequently, rare ASVs were those observed in fewer than 5 samples. appreciable proportion of reads among orders (Fig. 1c). Archaeorhizomycetales was 406 also present, albeit in low proportion. Fertilization effects on alpha diversity 433 We observed that seven years of moderate fertilization rates did not elicit changes 434 on RAF communities' alpha diversity (Fig 2a, Table 1 have been observed by random variation of the data is larger than 5% (Fig. 2a).

441
Dunnett's t tests supported the CI assessment, confirming that no fertilization 442 treatment elicited a statistically significant reduction in mean ASV richness at the 443 kingdom level (Table 1).

444
Trends observed at the kingdom level remained consistent across focal phyla (Fig.   445 2b). That is, no fertilization regime changed mean ASV richness among these phyla.

446
Dunnett's t tests supported the CI assessments, showing none of the differences 447 elicited by fertilization treatments were statistically significant (Table 1). Assumption nutrients. In stark contrast to the overall trend, mean richness of rare Basidiomycota 457 variants was higher than the mean richness of frequent ones (Fig. S5b).

458
Fertilization effects on beta diversity 459 Beta diversity of RAF communities, as characterized by the Jaccard index, was large within Control, and +P plots, it was not significant when considering assemblages 468 within the N addition treatments (Table S3).   Fig. 3). Test at the phylum level were rather inconsistent, with  (Table S5). Yet two notable differences arose when analyzing 488 only rare ASVs. The addition of N did not shift the structure of any of the clades 489 considered and, the addition of P did affect the structure of the two most frequent 490 clades recovered in this study (i.e. Ascomycota and Basidiomycota).

491
Variance partitioning showed that none of the covariate groups analyzed contributed 492 to explain a considerable portion of RAF community composition variability (Fig. 4). Relative abundance of reads amongst the great majority of phyla, orders or guilds 514 did not change in response to fertilization (Fig. 5). Univariate tests indicated technical 515 variation was as great as variation between treatments in almost all cases (Fig. S7).

675
It is also possible that RAF communities are simply not structured by the same 676 drivers affecting soil fungal or mycorrhizal communities. We observed high beta 677 diversity, which was weakly correlated with distance between samples. This is 678 remarkable, given the small spatial extent covered in this study (0.64 ha). Indeed, 679 high turnover rates at local scales appear to be a generic feature of soil fungal    Sample metadata is available with accessions ERS6425851-ERS6425946.