Soil microbial communities associated with giant sequoia: How does the world’s largest tree affect some of the world’s smallest organisms?

Giant sequoia (Sequoiadendron giganteum) is an iconic conifer that lives in relic populations on the western slopes of the California Sierra Nevada. In these settings it is unusual among the dominant trees in that it associates with arbuscular mycorrhizal fungi rather than ectomycorrhizal fungi. However, it is unclear whether differences in microbial associations extends more broadly to non-mycorrhizal components of the soil microbial community. To address this question we characterized microbiomes associated with giant sequoia and co-occurring sugar pine (Pinus lambertiana) by sequencing 16S and ITS1 of the bulk soil community at two groves with distinct parent material. We found tree-associated differences were apparent despite a strong grove effect. Bacterial/archaeal richness was greater beneath giant sequoia than sugar pine, with a unique core community that was double the size. The tree species also harbored compositionally distinct fungal communities. This pattern depended on grove but was associated with a consistently elevated relative abundance of Hygrocybe species beneath giant sequoia. Compositional differences between host trees correlated with soil pH, calcium availability, and soil moisture. We conclude that the effects of giant sequoia extend beyond mycorrhizal mutualists to include the broader community, and that some but not all host tree differences are grove-dependent.


Introduction 25
There is increasing evidence that tree species influence a combination of soil 26 chemical, physical, and biological properties (Hobbie et al., 2007;Mitchell et al., 2010;27 Langenbruch et al., 2012). For example, variation in litter chemistry, patterns of nutrient 28 uptake, root exudation, and microclimate among tree species can alter rates of 29 decomposition, soil nitrogen (N) and carbon (C) availability, and pH (Binkley & 30 Giardina, 1998). Tree-induced differences in resource availability and microclimate can, 31 in turn, modify soil bacterial, archaeal, and fungal community composition (Ushio et  Although soil microorganisms can directly and indirectly influence plant 42 dynamics (Abbott et al., 2015) and may mediate how plant communities respond to 43 anthropogenic threats (Malcolm et al., 2006;Stephens et al., 2014), information on soil 44 microbial communities associated with many rare or endemic tree species is limited. One 45 such tree is the giant sequoia (Sequoiadendron giganteum)-a species that epitomizes 46 charismatic mega-flora (Hall et al., 2011). Endemic to the western slope of the Sierra 47 symbiotic fungi that form relationships with the vast majority of herbaceous plants 71 (Fahey et al., 2012). This contrasts with most other trees in the Sierra Nevada, such as 72 those of the Pinaceae and the Fagaceae families, whose woody roots instead associate 73 with symbiotic ectomycorrhizal fungi (EMF; Brundrett & Tedersoo, 2018). Few studies 74 have probed the mycorrhizal dynamics of giant sequoia (Kough et al., 1985;Molina, 75 1994;Fahey et al., 2012), and none have intensively assessed soil bacterial/archaeal and 76 fungal community structure (composition and diversity) using molecular techniques. 77 Given that giant sequoia occur on soils derived from various rock types-including 78 granite, diorite, and andesite (Weatherspoon, 1990)-it is important to evaluate whether 79 microbial dynamics beneath giant sequoia remain consistent or if they vary across groves 80 with different parent material. Parent material exerts a strong influence on soil properties, 81 and differences in underlying geology often interact with trees to shape soil microbial 82 community structure (Ulrich & Becker, 2006;Carletti et al., 2009;Wagai et al., 2011). 83 For example, parent material and vegetation type interacted to affect soil macroaggregate 84 size, and both factors also shaped microbial community structure following 30 years of 85 surface exposure of reclaimed surface mining sites (Yarwood et al., 2015). The degree to 86 which this occurs with the iconic giant sequoia, however, remains unknown. 87 In this paper, we sequenced bacterial, archaeal, and fungal communities from beneath 88 giant sequoia and a co-dominant ectomycorrhizal tree, sugar pine (Pinus lambertiana 89 Dougl.). Sugar pine are prevalent on the western slope of the Sierra Nevada and are 90 second only to giant sequoia in total volume, with individuals reaching 76 m in height 91 and living up to 600 years (Hardin et al., 2001). We sampled soil from 32 individuals of 92 each tree species across two groves with contrasting geologic substrates in Yosemite 93 National Park, USA. By comparing these two tree species within and between groves, our 94 experimental design allows us to evaluate for the first time the relative impact of these 95 tree hosts and parent material on soil microbial structure. were generally the limiting experimental unit (especially in the Merced Grove), we first 128 selected mature giant sequoia individuals whose crowns did not overlap with adjacent 129 trees. We then selected the closest mature sugar pine trees to each giant sequoia 130 individual that shared similar aspect, slope, landscape position, and understory species 131 (when present). Relatively little understory vegetation occurred beneath each focal tree, 132 and areas containing nitrogen-fixing species such as Ceanothus spp. were avoided. As 133 with giant sequoia, we ensured selection of sugar pine individuals whose crowns did not 134 overlap with adjacent trees. This selection procedure was designed to minimize any 135 confounding influences that may affect soil properties besides tree species. Our 136 experimental design resulted in 32 total giant sequoia-sugar pine pairs that were typically 137 within 30 m of each other. 138

Soil Sampling 139
In August 2013, we sampled bulk surface soil from beneath each tree individual 140 (i.e., we did not directly target rhizosphere soil surrounding plant roots). We selected 141 sampling locations mid-crown and downslope of the tree bole with the assumption that 142 aboveground litter would accumulate most at these locations and therefore the influence 143 of trees would be maximal (Zinke, 1962). Five replicate soil cores (0-5 cm depth of 144 mineral soil) per tree were taken within an approximately 20 cm x 20 cm area using an 145 Oakfield corer (1.9 cm diameter; Oakfield Apparatus Co, Fond du Lac, WI, USA), and 146 composited into a single sample within a sterile plastic bag (Whirl-Pak®, Nasco, Fort 147 Atkinson, WI, USA). The soil corer was sanitized after each composite sample using a 148 rinse of 10% bleach followed by a rinse of 95% ethanol. Soil samples were stored at 4 149 °C, transported to University of California (UC) Merced, sieved (< 2 mm; sanitized 150 between samples as described above), and subsampled for microbial analysis. 151

DNA extraction 152
Each subsample was extracted immediately upon returning to the laboratory 153

16S Sequence analysis 202
We obtained the 16S rRNA gene sequences already demultiplexed from the UC 203 Riverside Genomics Core Facility and processed them using Quantitative Insights into 204 Microbial Ecology (QIIME; Caporaso et al., 2010). After we joined the forward and 205 reverse reads (allowing for 20% maximum difference within the region of overlap), we 206 used default parameters to conduct quality control: reads were excluded if the length was 207 less than 75 bases, if there were more than three consecutive low-quality base calls, if 208 less than 75% of the read length was consecutive high-quality base calls, if a Phred score 209 was below three, or if one or more ambiguous calls were present (Bokulich et al., 2013). 210 After quality filtering, 3.9 M sequence reads remained. to 36,345 reads per sample, and as a result, one sugar pine sample was dropped. 218

ITS Sequence analysis 219
We obtained the ITS gene sequences already demultiplexed from the UC 220 Berkeley Vincent J Coates Genomic Sequencing Laboratory and processed them as in 221 (Glassman et al., 2016) using UPARSE (Edgar, 2013). We removed distal 222 priming/adapter sites, trimmed the remaining untrimmed, low-quality regions from the 223 ends, and then joined the forward and reverse reads. Paired reads were then quality 224 filtered using the fastq_filter command in usearch and employing a maximum expected 225 number of errors of 0.25. After quality filtering, 2.9 M sequence read pairs remained. We 226 picked 97% OTUs, then reference based chimera detection was employed using usearch 227 and referencing against the UNITE database accessed on 10.09.2014 (Kõljalg et al., 228 2005

Statistical Analysis 235
We used a multifaceted approach to assess microbial community structure of 236 giant sequoia soils and to compare the structure of these communities with those beneath 237 sugar pine. We first tested the main and interactive effects of plant species and grove on 238 bacterial/archaeal and fungal OTU richness (alpha diversity) by performing a two-way 239 analysis of variance (ANOVA, ɑ = 0.05), transforming the data for normality and 240 homogeneity of variance when necessary. We then visualized similarities in microbial 241 community composition between tree species and grove using non-metric 242 multidimensional scaling (NMDS) of the Jaccard (presence-absence) and Bray-Curtis 243 (relative abundance) dissimilarity metrics. To determine if beta diversity differed 244 significantly between tree species and grove, we performed the multivariate permutation 245 test perMANOVA using the 'adonis' function in the R VEGAN package (permutations = 246 999; Oksanen et al., 2012). Because an assumption of perMANOVA is equal variance 247 between groups, we also performed an analysis of multivariate homogeneity of group 248 dispersions (permDISP; Supplemental Information Table S1). 249 In addition to running beta diversity analyses that included all taxa, we ran 250 separate analyses for two subsets of fungal taxa that included only EMF or AMF. This 251 allowed us to determine whether these root symbionts differed significantly between 252 samples (by tree and grove). Specifically, EMF taxa were bioinformatically parsed as 253 previously established (Glassman et al., 2015) and AMF were bioinformatically parsed to 254 include only individuals of Glomeromycotina. The resultant OTU tables were rarefied to 255 even sampling depths (EMF = 3022; AMF = 40), visualized using NMDS, and analyzed 256 using perMANOVA in the same way as described above. 257 As a complement to these multivariate tests, we compared the relative abundance 258 of bacterial/archaeal and fungal phyla within each grove using non-parametric Mann-259 Whitney U test on ranks. Given that abundant taxa tend to contribute significantly to 260 ecosystem functioning (Dai et al., 2016), we also analyzed the frequency and relative 261 abundance of dominant OTUs across species and grove. Specifically, we filtered both 262 bacterial/archaeal and fungal OTU tables to include only those OTUs that comprised 263 >1% of the total sequences. While 28 fungal OTUs met this criterion, only one 264 bacterial/archaeal OTU met it (a Bradyrhizobium species); therefore, we summarized the 265 results of the fungal OTUs only. Specifically, we visualized OTU frequency (presented as 266 a percentage of the total number of samples) and relative abundance, and assessed 267 significant differences in OTU relative abundance within each grove using non-268 parametric Mann-Whitney U tests. 269 Finally, we identified core OTUs that were unique to each tree species across both 270 groves, where a core OTU was defined as an OTU that occurred in 100% of the samples 271 recovered beneath a tree species. This enabled us to capture and identify the microbial 272 members that were shared among all giant sequoia or sugar pine soils, but not both (i.e., 273 the unique or host-specific core microbiome of each tree species). Core OTUs were 274 identified using the compute_core_microbiome.py script in QIIME and graphically 275 represented using Venn diagrams. Core bacterial/archaeal OTUs that were unique to each 276 tree species were summarized at the phylum level. In addition, we determined the 277 taxonomy of core bacterial/archaeal OTUs that were significantly more frequent (> 20% 278 difference) beneath giant sequoia than sugar pine using EzTaxon (Kim et al., 2012), and 279 illustrated the presence/absence of these OTUs from all samples with the heatmap3 280 package (Zhao et al., 2014). We provide no such summary for fungi, as neither giant 281 sequoia nor sugar pine had a core microbiome based on our 100% occurrence definition. 282 In addition to determining differences in microbial community structure, we 283 aimed to identify whether any structural differences may be attributed to tree-induced 284 changes in soil parameters. To that end, univariate Spearman rank correlations were 285 conducted to determine if any of the measured physiochemical parameters correlated 286 with microbial richness. In addition, we used simple and partial Mantel tests to examine 287 correlations between each physiochemical parameter and microbial community 288 composition. A simple Mantel test is a non-parametric method that compares two 289 distance matrices, and which calculates a correlation coefficient and p-value using 290 permutations. It determines whether samples that are similar in one measure (e.g., soil 291 pH) are similar in another (e.g., microbial composition Averaged across both the Mariposa and Merced groves, Proteobacteria comprised the 308 majority of the 16S sequences recovered beneath giant sequoia (31.5% ± SE 0.5%), 309 followed by Acidobacteria (15.9% ± SE 0.5%), Actinobacteria (15.6% ± SE 0.4%), 310 Planctomycetes (11.2% ± SE 0.3%), and Verrucomicrobia (9.0% ± SE 0.2%; 311 Supplemental Information Figure S1A). Together, these five phyla accounted for greater 312 than 80% of the sequences. Forty-one less abundant phyla were also recovered from giant 313 sequoia soils, three of which were archaeal. 314 315

Characterizing Giant Sequoia Fungal Communities (Q1) 316
Averaged across both the Mariposa and Merced groves, Basidiomycota comprised the 317 majority of ITS sequences recovered beneath giant sequoia (82.1% ± SE 2.4%), followed 318 by Ascomycota (12.5% ± SE 1.5%), and Zygomycota (2.1% ± SE 1.4%). Less than 2% 319 were Glomeromycota or unidentified fungi (Supplemental Information Figure S1B). Bacterial/archaeal communities were most strongly structured by grove effects, 328 followed by host tree differences ( Figure 1A, Table 1A). Bacterial/archaeal richness was 329 greater under giant sequoia (mean richness ± SE, Mariposa Grove: 5,924.6 ± 129.2, 330 Merced Grove: 6,164.6 ± 104.6) compared to sugar pine (mean richness ± SE, Mariposa 331 Grove: 5,327.7 ± 178.5, Merced Grove: 5,875.4 ± 221.5; Figure 2A). These differences 332 remained constant across grove (no significant tree x grove interaction, P > 0.1). At the 333 phylum level, Proteobacteria, Actinobacteria, and Gemmatimonadetes were relatively 334 more abundant-and Acidobacteria, Armatimonadetes, and TM7 were relatively less 335 abundant-in giant sequoia compared to sugar pine soils (Supplemental Information 336 Figure S1A). However, these differences were not consistent across groves and some 337 were only marginally significant (P = 0.05-0.10). 338 Fungal communities were also structured most strongly by grove effects, followed 339 by host tree differences ( Figure 1B; Table 1B). However, in contrast to bacteria/archaea, 340 there was a significant tree by grove interaction (P < 0.05), and fungal richness did not 341 significantly differ between tree species (giant sequoia mean richness ± SE, Mariposa 342 Grove: 160.7 ± 12.7, Merced Grove: 200.1 ± 13.1; sugar pine mean richness ± SE, 343 Mariposa Grove: 187.9 ± 17.7, Merced Grove: 187.3 ± 14.8; Figure 2B). At the phylum 344 level, Basidiomycota and Glomeromycota were relatively more abundant-and 345 Ascomycota and Zygomycota were relatively less abundant-in giant sequoia compared 346 to sugar pine soils (Supplemental Information Figure S1B). These phylum-level 347 differences were not consistent across groves and some were marginally significant (P = 348 0.05-0.1). In addition, the composition of EMF and AMF communities showed small but 349 significant effects of tree host and grove with no interaction (Supplemental Information 350 Only one bacterial OTU, a Bradyrhizobium (phylum Proteobacteria), comprised more 356 than 1% of the total bacterial sequences. In contrast, 28 fungal taxa each comprised more 357 than 1% of the total fungal taxa (Figure 3). Of these 28 fungal taxa, 64% were EMF and 358 18% belong to the genus Hygrocybe. We observed a number of these dominant OTUs 359 whose frequency, relative abundance, or both differed consistently between tree species. 360 For example, in both Mariposa and Merced groves, an unidentified Cryptococcus species 361 was recovered from 100% of samples, but was relatively more abundant beneath sugar 362 pine. An unidentified Byssocorticium species, an EMF taxon, also showed a consistent 363 trend across groves, where it was more frequent and relatively abundant beneath sugar 364 pine compared to giant sequoia. In contrast, species of Hygrocybe, which is generally 365 considered saprophytic (although see discussion below), were almost always more 366 frequent and relatively more abundant in giant sequoia soils, although these differences 367 were not always statistically significant (Figure 3). Finally, some OTUs differed 368 significantly between tree species in one grove but not the other. This included Russula 369 acrifolia, another EMF taxon (Tedersoo et al., 2010), which was more frequent and 370 relatively abundant beneath sugar pine than giant sequoia in the Mariposa grove, and an 371 unidentified Geminibasidium species, which is a xerotolerant basidiomycete yeast 372 (Nguyen et al., 2013), was relatively more abundant beneath sugar pine than giant 373 sequoia in the Merced grove. 374 There were very few fungal OTUs that were recovered from 100% of giant 375 sequoia or sugar pine samples in either grove. Accordingly, neither giant sequoia nor 376 sugar pine had a core fungal microbiome comprised of OTUs that were recovered from 377 all samples ( Figure 4A). Even when we relaxed the definition of a core to require only 378 80% frequency, giant sequoia soils contained only two unique core fungal OTUs and 379 sugar pine soils contained one-both of which were Zygomycota. In contrast, 380 bacterial/archaeal communities beneath giant sequoia and sugar pine harbored a number 381 of OTUs that made up a unique core (i.e., were recovered from 100% of giant sequoia 382 samples or 100% of sugar pine samples, but not 100% of both; Figure 4B). Both tree 383 species contained the phyla Proteobacteria, Acidobacteria, Planctomycetes, 384 Actinobacteria, Bacteroidetes, and Verrucomicrobia in their core community ( Figure 4C  Gemmatimonadetes, and TM7, while sugar pine did not ( Figure 4C). Similarly, only 388 sugar pine contained core members from Armatimonadetes, Chlorobi, and OD1 ( Figure  389   4D). In addition to these phylum-level differences, the size of the core for giant sequoia 390 and sugar pine differed, with giant sequoia containing substantially more core OTUs (101 391 OTUs) than sugar pine (50 OTUs; Figure 4B). However, of the 101 core OTUs beneath 392 giant sequoia, only 13% were notably less frequent (frequency < 80%) in sugar pine soils 393 (Figure 4E; Supplemental Information Table S2). Similarly, of the 50 core OTUs beneath 394 sugar pine, only 10% were notably less frequent in giant sequoia soils (Supplemental 395 Information Table S2). In all other cases, OTUs that comprised the core of one tree 396 community were often missing from only a few samples in the other tree community (80-397 95% frequency). partial Mantel test. The composition of bacterial/archaeal and fungal communities also 407 tended to be strongly related to differences in soil moisture (gravimetric water content) in 408 both the simple and partial Mantel tests. In addition, bacterial/archaeal and fungal 409 community composition correlated with extractable aluminum and the sum of base 410 cations; however, these relationships were inconsistent between groves for fungi, and 411 often disappeared for both microbial groups when the effects of other soil parameters 412 were accounted for ( Figure 5). 413 In contrast to microbial composition, bacterial/archaeal and fungal richness were 414 largely unrelated to the measured physiochemical parameters (Supplemental Information 415 Table S3) Differences in bacterial/archaeal community composition between groves tended 439 to be greater than those differences associated with tree species. Still, there was evidence 440 that giant sequoia influenced underlying bacteria and archaea in unique ways compared 441 to sugar pine. Specifically, we found that bacterial and archaeal richness was greater 442 beneath giant sequoia than sugar pine (Figure 2A), possibly because giant sequoia are 443 larger and create more niche space within the soil . In addition, 444 communities of bacteria/archaea were compositionally distinct from those beneath sugar 445 pine ( Figure 1A). These differences remained constant across the two groves, despite the 446 fact that soils from each grove were derived from geochemically distinct substrates. 447 Fungal community composition also differed between giant sequoia and sugar pine; 448 however, the specific ways that fungi differed between tree species depended on the 449 grove ( Figure 1B) despite relatively large grove differences, a "host signal" can still be observed. 453 Giant sequoia and sugar pine are known to associate with two contrasting groups 454 of mycorrhizal fungi. The former with AMF (Fahey et al., 2012) and the latter with EMF 455 (Walker, 2001). While AMF and EMF were recovered from beneath both tree hosts-456 possibly because of understory influences or root overlap-communities associated with 457 giant sequoia differed from those associated with sugar pine (Supplemental Information 458 Figure S2). AMF communities were considerably more diverse beneath giant sequoia, 459 and EMF taxa such as an unidentified Byssocorticium species and Russula acrifolia were 460 more frequent and relatively abundant beneath sugar pine. While this is to be expected, 461 our findings show that differences in mycorrhizal communities in mixed stands of AMF 462 and EMF trees can be seen at the bulk soil scale, differences that can have cascading 463 effects on global scale biogeochemical processes including the cycling of carbon (Averill 464 et al., 2018) and nitrogen (Mushinski et al., 2019). 465 Microbial taxa that are abundant within the community can contribute 466 significantly to ecosystem function (Dai et al., 2016). To discern patterns in frequent and 467 abundant taxa between trees, we filtered both the bacterial/archaeal and fungal OTU 468 tables to include only those OTUs that comprised greater than 1% of the total sequences. 469 In accordance with the idea that prokaryotic microbial communities often include a very 470 small number of dominant members, and are instead comprised of many rare members 471  In contrast to bacteria/archaea, we recovered 28 abundant OTUs that each 480 comprised greater than 1% of the total ITS sequences (Figure 3). Notably, 18% of these 481 OTUs were from the genus Hygrocybe (waxcaps). Hygrocybe are widespread and can be 482 In addition to identifying common and abundant OTUs, it can be useful to 499 distinguish core members of a microbial community that remain constant across space or 500 time (Shade & Handelsman, 2012). Doing so helps define a healthy (or alternatively a 501 degraded) community, and can improve our understanding how that community will 502 respond to future perturbations. In contrast to fungi, which had no discernable core 503 community, we identified considerable core bacterial/archaeal communities associated 504 with both tree species (Figure 4A & B). Interestingly, giant sequoia's unique core was 505 double the size of sugar pine's, indicating that this giant, long-lived tree maintains a 506 relatively large and consistent set of prokaryotic OTUs in its surrounding soil. The larger 507 community associated with giant sequoia could be due to its age and size, as larger trees 508 are known to host more microbial taxa . Of these OTUs, thirteen 509 were also considerably (at least 20%) less frequent in sugar pine soils. However, there 510 were no clear trends in the taxonomy or ecology of these thirteen OTUs, with family 511 associations ranging from Norcardioidaceae (contains endophytes and species capable of 512 degrading organic matter; Tóth & Borsodi, 2014) to Mycobacteriaceae (contains animal 513 pathogens and species capable of degrading hydrocarbons; Lory, 2014). Metagenomic 514 data could provide a more complete picture of core community dynamics, as some 515 evidence suggests that communities assemble at the functional rather than the 516 phylogenetic level (Burke et al., 2011). Regardless, our data contribute to a small set of 517 previously published work that explicitly identify core communities in bulk soil (Andrew 518 et al., 2012;Orgiazzi et al., 2013), and indicate that giant sequoia and sugar pine each 519 harbor unique core communities of bacteria/archaea, but lack consistent core fungal 520 OTUs at the spatial scale studied here. 521 It is possible that having a large and diverse core community of bacteria/archaea 522 aids in the long-term success of giant sequoia individuals. In addition to providing critical 523 biogeochemical functions (e.g., decomposition of organic matter and nutrient cycling), 524 core and abundant microorganisms within soil may act as a source that "seeds" 525 rhizospheric and endospheric communities, ultimately contributing directly to plant 526 health. Indeed, it has been proposed that some foliar endophytes persist through the 527 winter as saprobes of litter only to re-invade host leaves in the spring (Unterseher et al., 528 2013;Baldrian, 2017). A recent study assessing foliar bacterial endophytic communities 529 of giant sequoia and coastal redwoods found that giant sequoia contained a diverse 530 endophytic community, with major phyla including Acidobacteria, Actinobacteria, 531 Bacteroidetes, Firmicutes, Fusobacteria, Proteobacteria, and TM7 (Carrell & Frank, 532 2015). Notably, five of the twenty dominant orders recovered from giant sequoia foliage 533 samples were represented in our core bacterial/archaeal community dataset 534 (Actinomycetales, Burkholderiales, Rhizobiales, Rhodospirillales, and 535 Sphingobacteriales). It is conceivable that at least some of these endophytes are derived 536 from the soil community; however, (at minimum) comparative sampling of giant sequoia 537 microbial communities within the same site, and (at maximum) use of more advanced 538 tracing techniques (e.g., isotopes or quantum dots), will be required to assess this. Future 539 studies that focus on connecting the giant sequoia holobiont with soil communities should 540 provide promising insights to the stability of this tree species over millennia. Supplemental Information Methods S1)-and while bacterial/archaeal and fungal 550 richness were generally insensitive to these parameters, community composition 551 correlated with a number of them. Most strongly and consistently, bacterial/archaeal and 552 fungal community composition related to soil pH and soil moisture ( Figure 5). Soil 553 moisture dynamics are coupled to water stress, oxygen diffusion, and substrate supply; 554 differences in soil moisture can therefore alter metabolic activity of microbial functional 555 groups, the occurrence of aerobic/anaerobic processes (e.g., aerobic 556 decomposition/denitrification), and ultimately microbial community composition 557 cations, a composite measure that was composed primarily of calcium. However, this 574 relationship almost always disappeared when the effects of the other variables were 575 accounted for (Figure 5), suggesting that the influence of calcium was mediated by other 576 soil parameters-namely soil pH. Base cations compete with H + and Al 3+ for exchange 577 sites on soil particle surfaces; therefore, the concentration of calcium in soil mediates soil 578 pH, such that higher amounts of exchangeable calcium (and other base cations) create 579 less acidic soils (Reich et al., 2005). The idea that soil calcium indirectly mediates Using next-generation sequencing techniques, we show for the first time that 590 microbial communities of bulk soil differ between giant sequoia and a co-occurring 591 conifer, sugar pine. Namely, giant sequoia supported unique bacterial/archaeal and fungal 592 community composition, greater bacterial/archaeal richness, and a large unique core 593 community of bacteria/archaea. These host tree differences, which were at least partially 594 driven by soil pH, calcium availability, and soil moisture, were discernible despite 595 concurrently large grove effects. In some cases, the influence of host tree differed 596 between the two groves under study, which were close in proximity but had contrasting 597  Figure 1. Influence of tree species (giant sequoia and sugar pine) and grove (Mariposa and Merced Grove) on (A) bacterial/archaeal and (B) fungal community composition. Left panel = non-metric multidimensional scaling (NMDS) of Jaccard (presence-absence) dissimilarity metric. Right panel = NMDS of Bray-Curtis (relative abundance) dissimilarity metric. Each symbol corresponds to a sample collected from one of two groves, and each color corresponds to a tree species. Points that are close together represent samples with similar community composition, and the dashed ovals represent 95% confidence intervals of sample ordination grouped by unique tree x grove combinations. The stress values for the bacterial/archaeal ordinations were 0.06 (Jaccard) and 0.07 (Bray-Curtis); the stress values for the fungal ordinations were 0.14 (Jaccard) and 0.18 (Bray-Curtis). In (A), an asterisk denotes significant differences (P < 0.05) between tree species. P-values were derived from a two-way ANOVA. Figure 3. The relative abundance of the most abundant fungal OTUs in giant sequoia and sugar pine soils across both groves. Error bars = 1 standard error of the mean (n = 16). The size of each point is scaled by the frequency of an OTU (how many samples it was recovered from), with larger circles corresponding to greater frequency. Significant differences in OTU relative abundance between tree species were assessed using Mann-Whitney U test on ranks (* p < 0.05, ** p < 0.01, *** p < 0.001). The Venn diagrams show absolute number of OTUs shared between core microbiomes of each tree species across two groves. Phylum level taxonomic information is also provided for the OTUs comprising the (C) giant sequoia and (D) sugar pine core bacterial/archaeal communities. (E) Heatmap illustrating presence/absence and taxonomy of giant sequoia's core OTUs that were considerably less frequent (20% difference) in sugar pine soils. Taxonomic information was derived from EzTaxon, and % similarity is the sequence similarity between the OTU and its nearest cultured match. Colors of the bar beneath the heatmap correspond to tree type (pink = sugar pine, blue = giant sequoia). Figure 5. Results of (top panel) simple (r) and (bottom panel) partial (ρ) Mantel tests relating soil parameters that varied between tree types to bacterial/archaeal and fungal community composition. Only those parameters that correlated significantly with composition at least once are presented. pH = soil pH; Al = soil xtractable aluminum; Sum BC = sum of the base cations (Ca 2+ , Mg 2+ , K + , and Na + ); soil moisture = soil gravimetric water content. N.S. = non-significant (p > 0.05) correlation.