We sampled soils from the Mendenhall Glacier forefield in July 2014. The glacier forefield at the west side of the glacier terminus, where sampling took place (Figure 1), is the result of ongoing deglaciation since the Little Ice Age. Exposed soils are Entisols of granitic tills (Burt and Alexander, 1996). The glacier terminus is roughly 20 m above sea level with a mean annual precipitation > 2500 mm (Burt and Alexander, 1996). Soils were collected from three transects of 100 m length, at increasing distance from the glacier, representing ecologically important stages of succession: before vascular plant colonization (pre-colonization: N 58° 26.667′; W 134° 26.456′), initial alder colonized soils (alder colonizer: N 58° 26.673′; W 134° 33.397′), and alder stand soils (alder dominated: N 58° 26.665′; W 134° 33.457′) (Figure 1). Pre-colonization, alder-colonizer, and alder-dominated soils corresponded with approximately 7, 9, and 11 years post-deglaciation, respectively (Figure 1). A total of 10 samples were collected along the pre-colonization transect in unvegetated soils. A total of 10 samples were collected from the alder colonizer transect with spatially patchy alder colonizers from 45 to 60 cm tall. Here, soils were collected at the base of the alder seedling trunks, with no other vegetation within a 40 cm radius of the plant. This area was co-dominated by patchy fireweed. A total of 10 replicate soils were collected in a dense alder stand (1–1.4 m tall). Fresh clippings were collected from the alder stand plants and from young spruce in an adjacent area of > 11-year-old soils. All soils were then transported back to the laboratory in Juneau, AK, United States, and sieved to 2 mm. Subsamples for molecular analysis were placed in a freezer with dry ice and the remaining samples were placed in the fridge held at 4°C within 6 h of sampling. Vegetation was immediately dried at 60°C for 2 days. Soils were transported to Boulder on ice 2 days after collection and stored in a -70°C freezer for molecular analysis and at 4°C for soil chemistry analysis.

Dried leaves from spruce and alder were ground up and 50 g of each plant was added to 800 mL of megapure water, respectively, and shaken at 350 rpm for 1 h and then allowed to sit for 14.5 h. Extracts were then filtered with burned glass microfiber GF/F 47 mm filters with 0.45 micrometer pore size. Filtered extracts were then sterile filtered with 0.22 micrometer 150 mL bottle top filters (nanopyrogenic, sterile, Corning Incorporated, Corning, NY, United States). Samples of each leachate were then run on a Shimadzu TOC-V CSN Total Organic Carbon Analyzer to evaluate non-purgeable organic carbon (NPOC) (e.g., total organic carbon) and total dissolved nitrogen (TDN). Leachates were adjusted through dilution with sterile water to create spruce and alder leachate at equivalent ending C concentrations. C:N ratios of the two leachates based on NPOC and TDN measurements were approximately equivalent. Leachate pH was tested and found to be 4.8 and 4.2 for alder and spruce leachate, respectively. For the microcosm experiment, 40 g of homogenized alder-dominated soils were added to 3 sets of 10 replicate sterilized Mason jars, for a total of 30 microcosms. In this way, all microcosms started with alder-conditioned soil of equivalent starting microbial and edaphic properties. Treatments of alder leachate, spruce leachate, and sterile H₂O (control) were applied to the 3 sets of 10 replicate microcosms. Alder and spruce leachates were adjusted using sterilized, megapure water to ∼680 μg C/mL and ∼8 μg N/mL of leachate. A total of 5 mL of solution was added to each incubation, resulting in the addition of approximately 97 μg C/g soil and 1.1 μg N/g soil for those treated with litter leachate. The addition of alder vs. spruce leachate allowed us to not only assess the different effects of spruce and alder, but also distinguish whether effects were a general C-addition effect or specific to the heterospecific plant. Microcosms were arranged in a randomized block design for the incubation period at 18°C. Incubations were stopped after 5 days. Soils were immediately subsampled and placed in the -70°C freezer for DNA analysis as well as refrigerated at 4°C for enzyme analyses.

For all soils from the end of the microcosm experiment, soil moisture, pH, TDN, extractable organic carbon (NPOC), and ammonium (NH4⁺) were evaluated. A subsample of each soil was dried at 100°C for 48 h to determine gravimetric soil moisture and adjust for real soil weight in all calculations. Immediately following collection of the final microcosm soil sample, 10 g of soil were extracted for 24 h in 40 mL of 2M KCl including 1 h of shaking. Extracts were filtered with Whatman no. 1 paper (Whatman Incorporated, Florham Park, NJ, United States) and frozen until analysis. NH4⁺ was measured on a BioTek Synergy 2 Multidetection Microplate Reader (BioTek, Winooski, VT, USA) and TDN/NPOC were measured on a Shimadzu TOC-V CSN Total Organic Carbon Analyzer (Shimadzu TOCvcpn, Kyoto, Japan). pH was measured with a ratio of 1.5 g dry soil: 3 mL water. Soils were shaken at 250 rpm for 1 h, allowed to equilibrate, and then tested on an Accumet Research AR10 pH meter.

Enzyme activities for β-1,4-N-acetylglucosaminidase (NAG), and acid phosphatase (aP) were evaluated for the microcosm experiment in order to assess microbial investment in N and P acquisition under the effect of alder or spruce litter leachate. The activities of these enzymes are indicative of microbial (both bacterial and fungal) investment in the acquisition of N and P, and the limiting nature of these nutrients. Thus, changes in these decomposition enzymes both indicate shifts in soil nutrient cycling dynamics as well as microbial nutrient limitation (Sinsabaugh et al., 2008; Holden et al., 2012; Knelman et al., 2017). Enzyme activity was measured via fluorometric microplate methods (Sinsabaugh et al., 2002; Weintraub et al., 2012). The methods of Weintraub et al. (2012) were used based on a 96-well assay plate method with 1M sodium acetate buffer titrated to a pH of 7.0 and 4-methylumbelliferone standards. Approximately 1 g of refrigerated soil was used from each sample. Each sample was run with 16 analytical replicates, quench corrections, standards, and negative controls. Fluorescence was measured using a microplate reader (Thermo Labsystems, Franklin, MA, United States) at 365 nm excitation and 460 nm emission to calculate nmol activity h^-1 g soil^-1.

MoBio’s PowerSoil DNA Isolation Kit (MO BIO Laboratories, Carlsbad, CA, United States) was used according the manufacturer’s protocol to extract genomic DNA from each sample within the environmental and microcosm study. Samples were checked for DNA purity using a NanoDrop800 and quantified using the PicoGreen method. All samples were then normalized to 2.85 ng DNA/μL. Samples were then sent to The Genomic Sequencing and Analysis Facility at University of Texas at Austin for barcoded PCR using 515F/806R primers (Hyb515F_rRNA: 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGGTGYCAGCMGCCGCGGTA-3′ and Hyb806R_rRNA: 3′-TAATCTWTGGGVHCATCAGGGACAGAGAATATGTGTAGAGGCTCGGGTGCTCTG-5′), and sequencing of the 16S rRNA gene V4 regions on a Illumina MiSeq platform, according to the sequencing center protocols¹.

Using the R statistical environment (R Development Core Team, 2013) and the pgirmess package in R, Kruskal–Wallis ANOVA contrasts were run to assess changes in dominant bacterial taxa. First, we evaluated changes at the phyla level – as well as genera level for well known plant symbionts within orders Rhizobiales and Actinomycetales (Nemergut et al., 2011; Yeoh et al., 2017) – across the natural chronosequence with increasing alder dominance. Within those phyla that showed significant responses with increasing alder dominance, we examined relative abundances from the microcosm experiment for all genera that constituted, on average, at least 0.1% relative abundance of communities, to better understand how spruce plant-effects impacted these alder-influenced bacterial communities. We used the ggplot2 package in R to plot these results (Figure 3). Using Tukey’s Honestly Significant Differences test, we examined which genera changed significantly due to alder/spruce leachate addition in comparison with control (alder-dominated) soils collected from the natural chronosequence. Enzyme activity data was checked for normality and then assessed for differences across microcosm treatments using One-Way ANOVAs. TDN, NH4⁺, and NPOC were natural log-transformed to fulfill assumptions of normality and along with pH and soil moisture were tested for differences among incubation treatments using One-Way ANOVAs and Tukey’s Honestly Significant Differences tests.

We examined bacterial community composition and diversity using the UPARSE and QIIME software packages (Caporaso et al., 2010b; Edgar, 2013). Cutadapt (Martin, 2011) was used to trim sequencing primers from the forward and reverse reads. When both the forward and reverse primers were not detected (allowed for 10% mismatches for primer search), read pairs were removed. Paired-ends were merged, demultiplexed and stored as fastq output according to the protocol of Andrei et al. (2015) in QIIME (v1.9). Operational taxonomic units (OTUs) were picked and an OTU table was constructed using UPARSE. Using the “parallel_assign_taxonomy_rdp.py” script in QIIME, taxonomy was assigned based on the Ribosomal Database Project (RDP) classifier (Wang et al., 2007) against the “13_8” Greengenes database (DeSantis et al., 2006). A phylogenetic tree was built using FastTree (Price et al., 2009) on masked pyNAST aligned sequences (Caporaso et al., 2010a). This phylogeny was then manually rooted to Archaea in Dendroscope (v3; Huson and Scornavacca, 2012). Based on OTU tables rarified to 8,490 sequences per sample, the lowest minimum sequencing depth for a sample, community dissimilarity matrices using the weighted UniFrac method (Lozupone et al., 2006, 2007) were calculated in QIIME. Permutational MANOVAs and PERMDISP, a permutational test of homogeneity of dispersion, were performed in PRIMER E on the UniFrac dissimilarity matrices (Clarke and Gorley, 2006). Mantel-like Spearman RELATE tests were performed in PRIMER E to assess the correlation between microbial community phylogenetic dissimilarity and NH4⁺. Ammonium data was ranked and converted into a dissimilarity matrix based on Euclidean distance and then compared to the UniFrac dissimilarity matrix via this test.

Sequences, mapping file, and metadata have been made available via FigShare with the DOI 10.6084/m9.figshare.5278297.v1.

There were significant differences in microbial community structure between the three successional time points of alder soil conditioning (Table 1). Sequencing data showed a clear correspondence of particular microbial taxa with the increasing dominance of alder in a directional manner. Increases in the prominence of Actinobacteria, Acidobacteria, Bacteroidetes, Planctomycetes, and Alphaproteobacteria occurred with increasing soil age and alder dominance (Table 2). A higher resolution view of genera within a bacterial order involved with symbioses, Rhizobiales, showed that Agrobacterium and Rhizobium increase significantly in young, alder-dominated soils. Outside of increases in Agrobacterium and Rhizobium, no other genera within Rhizobiales showed significant changes with the rising influence of alder across our sampling of the natural chronosequence. Although there were increases in Actinobacteria, no statistically significant trends were observed in the relative abundance of genera within order Actinomycetales including that of the alder symbiont, Frankia. Betaproteobacteria show decreases with increasing prominence of alder (Table 2).

Sequence analysis of the microcosm experiment showed distinct communities among treatments (Table 1). In particular, PERMDISP analysis showed that spruce had a greater structuring effect on microbial communities of alder-conditioned soils than other treatments, resulting in significantly less dispersion in community phylogenetic dissimilarity than both the alder leachate and the control soil communities (environmental alder-influenced soil bacterial communities) (PERMDISP, S vs. A: P = 0.003; S vs. control: P = 0.019). Controls and alder leachate soil communities did not show any statistically significant differences in the dispersion of phylogenetic community dissimilarity (PERMDISP, P = 0.568). Principle coordinates analysis (PCoA) based on UniFrac phylogenetic dissimilarity among communities shows grouping by treatment as well as this strong selective effect of spruce leachate on alder-conditioned microbial communities in the significant reduction of dispersion (Figure 2). At a higher resolution, leachate addition (both alder and spruce), in general, drove increases in the relative abundance of Acidobacteria (Figure 3). A differential response between alder and spruce leachate addition in relation to the control demonstrated unique spruce-driven shifts in microbial taxa as well. Out of all major taxa at the phyla level, statistically significant relative increases in Planctomycetes and Chloroflexi as well as declines in Deltaproteobacteria were associated with spruce leachate additions in comparison with microcosm controls. Further, we examined all genera within the these phyla and those that responded to increasing Alder dominance in the natural chronosequence (Table 2) to examine how interspecific interactions via spruce leachate addition influenced these bacteria at a higher resolution (Figure 3). This work showed a variety of genera in alder-influenced soils responded to spruce leachate in a significant, unique way. Some genera showed a general response to leachate addition indicating a leachate or C effect. For example, in the case of Arthrobacter, Hydrogenophaga, Polaromonas, and Sphingobium, these genera all responded with significant decreases in their relative abundance vs. control for both alder and spruce leachate addition. In other cases, Spruce uniquely drove declines in relative abundance vs. controls, where alder leachate did not (e.g., Agrobacterium, Bdellovibrio, Lutibacterium, Microbacterium, Novosphingobium, Peredibacter, Rhizobium, Rhodoferax, Sphingomonas, and Zymomonas). Finally, in the cases of Janthinobacterium, Planctomyces, and Thiobacillus, spruce leachate addition drove significant increases in the genera relative abundances vs. controls, whereas alder leachate addition did not.

The microcosm experiment, designed to assess the impact of spruce encroachment colonization on alder-conditioned soil communities, found no differences among treatments in soil moisture, pH, or TDN (Table 3). Predictably, both alder and spruce leachate treatments showed significant increases in extractable organic carbon as opposed to the control (Table 3), but were statistically indistinguishable from each other. Levels of NH4⁺ depended on the particular leachate addition. While both spruce and alder leachate soils showed significant decreases in NH4⁺, alder leachate soils showed significantly lower concentrations of NH4⁺ than spruce leachate soils (Table 3). No differences were found among treatment categories in NAG or aP enzyme activity and no correlations were found between bacterial community structure and enzyme activity. Correlations between community shifts and extractable NH4⁺ were examined, revealing a significant, albeit weak, correlation between bacterial phylogenetic community structure and soil NH4⁺ (RELATE test, Rho = 0.129, P = 0.026).

Here, we show a directional selection of alder on microbial community structure with colonization and increases in alder dominance across early succession. PERMANOVA analysis reveals that overall phylogenetic composition of pre-colonization, alder-colonizer, and alder-dominant soil bacterial communities are significantly different, while a higher resolution view of particular taxa associated with alder soils demonstrates that changes are consistent with alder driven effects as observed in the past at this site (Knelman et al., 2012). For example, increases in Actinobacteria attributed to alder colonization at this site are again demonstrated (Table 2). Alder-conditioned soils, with increases in the relative abundance of Actinobacteria, are not surprising given the symbiosis between N-fixing Frankia and alder (confirmed at this site) and the fact that alder may support other Actinobacteria as well (Ghodhbane-Gtari et al., 2009). Other strong changes are noted with increases in Alphaproteobacteria, including N-fixing Rhizobiales, which have been shown in past work to correspond with vegetation (including non-legumes) and plant-influenced carbon environments (King et al., 2012; Knelman et al., 2012; Cederlund et al., 2014; Yeoh et al., 2017). Conversely, Betaproteobacteria, which have been found to dominate early in succession before plant arrival (Nemergut et al., 2007; Sattin et al., 2009), strongly decrease with alder colonization in this study. In this way, our work shows how the well-documented early plant colonist, alder, drives directional shifts in bacterial community composition with increasing dominance.

The microcosm experiment was designed to understand the potential for spruce effects on alder-conditioned soil bacterial communities via litter inputs, such as carbon and secondary compounds (Clein and Schimel, 1995; Schimel et al., 1996; Bowman et al., 2004). End communities of each treatment were significantly different from each other indicating the potential for a general leachate effect and heterospecific plant effect (PERMANOVA, P < 0.05). It is important to note that while carbon increased in both leachate-addition treatment soils, % moisture, along with NPOC, TDN, and pH, which can potentially influence microbial community structure, shows no statistically significant differences between the alder and spruce leachate addition soils.

The overall shifts in microbial community phylogenetic structure under alder and spruce leachate treatments were reflected in significant shifts in relative abundance of major taxa from the controls. Increases in the relative abundance of Acidobacteria under both spruce and alder leachate additions, for example, are indicative of a leachate effect rather than interspecific interaction. Beyond this, spruce leachate addition uniquely drove differences from controls and alder treatments with relative increases in Planctomycetes and Chloroflexi and relative decreases in Deltaproteobacteria. In this way, the spruce-specific effects – beyond simply a leachate effect on alder-conditioned soil bacterial communities – demonstrate how late successional spruce may alter bacterial communities during transitions between early and late successional plant species. Further, we point out that the starting soils were from alder-dominated soil samples, so the fact that alder leachate results in no significant effect in the cases of Planctomycetes, Chloroflexi, and Deltaproteobacteria (Rhizobium and Agrobacterium as well), reinforces that the shifts in the reported microbial taxa are spruce-specific, rather than relating to carbon addition in general.

Spruce influence was observed across a variety of bacterial genera (Figure 3). In some cases, these responses occurred in known plant symbionts such as Agrobacterium and Rhizobium, which show accumulation with increasing alder dominance in the natural chronosequence (Table 2) but decrease with Spruce influence (Figure 3). Such changes are of particular interest given that a directional switch in the selection of symbionts can greatly implicate plant-microbe feedbacks and the turnover of early to late successional plants (Van der Putten et al., 1993; Bever, 1994; Van Der Heijden et al., 2006; Bever et al., 2013). The spruce leachate effect specifically drove declines in these genera of Rhizobiales in contrast to the alder-conditioned control soils in the experiment and the observed pattern of increasing abundance in Rhizobiales with increasing alder-influence in the glacier forefield. Although neither plant is leguminous, past work has shown the co-occurrence of Rhizobiales with non-leguminous colonizing plants in stressful environments (King et al., 2012; Knelman et al., 2012) and Rhizobiales may be involved in asymbiotic N-fix (Buckley et al., 2007). Both Agrobacteria and Rhizobium are genera that are found in root-associated bacterial communities at global scales (Yeoh et al., 2017) and our work shows that interspecific effects lead to directional changes in the relative abundance of such symbiotic organisms as well as broader bacterial taxa and community structure. Other genera known to associate with plants endophytically, such as Janthinobacterium (Farrar et al., 2014) and Sphingomonas (Khan et al., 2014), respond to spruce leachate with significant increases and decreases respectfully as compared to alder leachate and alder-conditioned control soils. This work shows that ecologically important plant symbionts may be altered through interspecific plant interactions, though more work is needed to understand the impact on plant-microbe feedbacks.

Interestingly, we find that spruce leachate addition communities show strong and significant decreases in dispersion of bacterial community structure among communities as opposed to both the control and the alder leachate communities, which both reflect soil bacterial communities under alder influence. While we note that we would fully expect to see strong shifts in microbial community composition in response to influence by a different plant species, the change in dispersion may indicate an overall stronger structuring effect of spruce as well. This finding is consistent with research that suggests a stronger relationship between later-successional species and microbial communities and the ability of late successional species to condition early successional soils (Kardol et al., 2006; Abbott et al., 2015; Koziol and Bever, 2015).

While this work elucidates how interspecific interactions between early and late successional plant species may be reflected in belowground bacterial community structure, we also sought to examine functional changes in biogeochemistry associated with these community shifts. Examination of microbial enzyme activity relating to N and P cycling showed no significant change among the microcosm treatments and no connection with bacterial community composition. Levels of NH4⁺, however, depended on the particular leachate addition. Alder-leachate soils showed significantly lower concentrations of NH4⁺ than spruce leachate soils, which were both significantly lower than controls. While it is difficult to suggest a mechanism underlying these patterns due to the potential effects of N immobilization, mineralization, and nitrification (Clein and Schimel, 1995; Schimel et al., 1996), it is interesting that beyond a general leachate effect, species-specific effects drive significant differences in the availability of NH4⁺. This important shift in the cycling of inorganic N may impact transitions from early to late successional plants as has been noted in past work examining transitions of alder to poplar/spruce in ecological succession (Chapin et al., 1994; Clein and Schimel, 1995; Schimel et al., 1996). Such shifts in nutrient cycling can strongly impact plant dynamics and thus offers a mechanism by which alterations of microbial communities by early vs. late successional plants may drive successional patterns. In this study, the changes in patterns of bacterial community structure due to spruce leachate correlate with N availability. The mantel-like RELATE tests reveal a significant correlation between NH4⁺ and bacterial community structure across microcosm treatments. It is important to note that this correlation though significant is weak, but may be obfuscated by the fact that only a subset of the microbial community may be responding spruce leachate or involved with N transformations that could impact ammonium pools, for example. While further research is needed to understand how interspecific plant interactions may alter specifics of microbial community structure with implications for nutrient cycling, this research suggests it may be an important consideration in understanding the role of plant–microbe interactions as drivers of ecosystem succession.