Study site and sampling scheme
Three sites were selected from the Danish tree species trials that were established in 1973 as monoculture 0.25-ha stands of each tree species, across a pH and soil nutrient gradient (Vesterdal et al. 2008). Mattrup had an agricultural land-use history, whereas both Vallø and Viemose were previously forested sites (beech forest).
Site Mattrup is located in Ring Skov on the peninsula of Jutland in the western part of Denmark (55°N57′36″, 9°E37′10″). The elevation is 100–110 m asl. and the site gently slopes eastward. The soil is a loamy till and is characterized as a well-drained, mixed, acid, mesic, coarse-loamy Lamellic Hapludalf (Soil Survey Staff 2014). The mean annual temperature in the period 2005–2015 was 8.5 °C with mean annual precipitation of 829 mm (Danish Meteorological Institute 2005). Mattrup was agricultural land from 1842 to 1973, thus a period of at least 130 years and probably longer (Christiansen et al. 2010).
Site Vallø is located on the island of Zealand (55°N25′23″, 12°E3′47″) 40 km southwest of Copenhagen at an elevation of 50 m asl. and gently sloping westward. The soil is a glacial till and is characterized as a moderately well-drained, mixed, nonacid, mesic, coarse-loamy Oxyaquic Hapludalf (Soil Survey Staff 2014). The mean annual temperature in the period 2005–2015 was 8.8 °C with a mean annual precipitation of 635 mm (Danish Meteorological Institute 2005). Vallø has been a managed beech forest for the last 250 years (Christiansen et al. 2010).
The third site, Viemose, is located at the coast in the southeast corner of the island Zealand (55°N1′2″, 12°E9′27″) at an elevation of 10 m asl. The soil is a glacial till and is characterized as a well-drained, fine-loamy, mixed, acid Typic Hapludalf (Soil Survey Staff 2014). The mean annual temperature in the period 2005–2015 was 9.2 °C with a mean annual precipitation of 618 mm (Danish Meteorological Institute 2005). Viemose has been a managed forest for at least 250 years (Danish Cadastre 2017), and the tree species was beech for more than a century before the experiment was established.
Monoculture stands (0.25 ha, within a total area of 1.6 ha) of common ash, European beech, pedunculate oak, small-leaved lime, sycamore maple and Norway spruce were planted in 1973 as part of a national tree species experiment (Vesterdal et al. 2008). There was no replication of tree species within each site so the sites serve to ensure replication. The original beech trees at Vallø and Viemose were felled at the establishment of the plots. The understory was not managed (i.e. removed) at the sites and was most abundant in the more open ash and oak stands where ground vegetation (herbs, grasses and some seedlings and saplings) covered the forest floor. Ground vegetation was almost non-existent in spruce, beech and lime stands apart from transient spring flora. Individual tree species plots had been thinned approximately every fourth year since 1987 and had been managed according to common practice (spacing and thinning) of the forest districts (Christiansen et al. 2010). Ash was absent at Vallø due to establishment failure.
In June 2015, four mineral soil cores were taken from 0 to 10 cm depth in the cardinal points of three 5-m radius subplots within each plot (see Fig. S1 for sampling scheme). Samples from within each subplot were composited in the field yielding three replicate soil samples for each tree species per site (n = 3). Soil samples were placed in coolers in the field and transported back to the University of Copenhagen for subsequent analyses where they were stored at 4 °C for 8 h. Soils were sieved through a 2-mm mesh sieve and six 15-g subsamples were immediately removed for the 15N pool-dilution analyses. A portion of each composite sample was stored at − 20 °C for DNA extraction, and tools were sterilized with ethanol between samples to minimize cross-contamination.
Soil C, N and pH
Total organic C and N concentrations were determined on ground samples by dry combustion based on the Dumas method (Matejovic 1993) using a FLASH 2000 EA NC Analyzer (Thermo Fisher Scientific, Waltham, MA, USA) with a quantification limit of 0.07%C. Soil pH (< 2 mm fraction) was determined in solution of 0.01 M CaCl2 (1:2.5 soil-solution ratio) and analysed with a combination electrode GK2401 (Radiometer, Copenhagen, Denmark). No major differences in soil bulk density, water holding capacity, and stone content were observed in previous studies within these same plots (Vesterdal et al. 2008).
Gross rates of N ammonification and nitrification
The 15N pool-dilution method, modified through the use of plastic Nalgene jars in place of glass mason jars used by Drury et al. (2008) and Hart et al. (1994), was used to determine gross rates of N ammonification and nitrification, with each soil sample analysed in triplicate. This method adds a fixed quantity of 15N for a specific soil, and does not constitute a major fertilization effect at our sites in which the inputs of N with throughfall are higher than in the location where this method was developed in North America. Six 15 g subsamples from each soil sample were transferred to 250 ml Nalgene jars and sealed with Parafilm (n = 3 subsamples for both 15NH4+ and 15NO3−). The Parafilm seal was punctured to enable gas exchange and maintain aerobic conditions throughout the experiment, and samples were then incubated at room temperature for 24 h prior to initial 15N fertilization treatments. Four ml of 15NH4Cl solution (99 atom %; Sigma Aldrich) or 4 ml of K15NO3 (99 atom%; Sigma Aldrich) was added to the soil samples in each respective jar, with an equivalent application rate of 12 μg N g−1 soil. This rate would be equivalent to addition of around 1 kg N ha−1 to the top 1 cm of mineral soil. For comparison, the annual mean N throughfall flux (kg N ha−1 year−1, SE, n = 4) at Vallø and Mattrup sites for spruce (28 ± 2) was significantly larger than for maple (12 ± 1), beech (11 ± 1) and oak (9 ± 1) stands but not different from that of lime (15 ± 3) (Christiansen et al. 2010). Labelled N was injected into the samples in 1 ml intervals four times, and gently homogenized by hand-mixing to ensure isotopic labelled N was applied uniformly throughout the soil sample, and the Parafilm seal was replaced.
A 5 g soil subsample was removed from each Nalgene jar and placed into a 50 ml falcon tube, immediately after 15N fertilization, 24 h after 15N fertilization, and 48 h after 15N fertilization. Fifty ml of 1.0 M KCl was added to each falcon tube, shaken for 1 h, and filtered through glass fibre filters in a vacuum syringe filtration system. A 15 ml subsample of each extract was reserved for ammonium and nitrate concentration analyses, which were determined colorimetrically using a flow injection analyser (FIA PE FIAS 300, Perkin-Elmer, Waltham, MA, USA) using the indophenol-blue and cadmium reduction methods for NH4–N and NO3–N, respectively. Due to high anticipated levels of N we prepared 10 × dilutions of each KCl extract. The remainders of the KCl extracts were used for microdiffusion acid trapping of 15NH4–N and 15NO3–N, with the sequential addition of MgO and Devarda’s alloy following the International Atomic Energy Agency protocol (IAEA, 2001). Acid traps were dried, packaged in tin cups, and analysed by elemental flash combustion analysis (EA 1110, Thermo Scientific, Bremen, Germany) in combination with stable isotope ratio mass spectrometry (Delta PLUS, Thermo Scientific, Bremen, Germany).
Net rates of ammonification and nitrification were calculated as the difference in inorganic N between the incubated samples and the initial soil extractions. The gross rates of ammonification, nitrification and microbial consumption were calculated following Hart et al. (1994), such that gross ammonification is equal to: ((Initial NH4+ concentration- Final NH4+ concentration)/extraction time) × [(log(APE 15N Initial/APE 15N Final)/log(Initial NH4+ concentration/Final NH4+ concentration)], where APE = Atom Percent Excess. Gross consumption was calculated as: Gross ammonification—((Final NH4+ concentration- Initial NH4+ concentration)/extraction time). Net ammonification was calculated as Gross ammonification—Gross consumption. Nitrification rates were similarly calculated. Based on the strong covariance with soil organic matter concentrations, which varied among tree species, N transformation rates were expressed per g C to evaluate the differences in transformation rate dynamics among tree species, and gene copies were similarly expressed per g C.
DNA extraction, amplification, and quantification
DNA was extracted from 0.25 g of field-moist mineral soil using the MoBio Power Soil DNA isolation kit (MoBio Laboratories, Inc., Carlsbad, CA). DNA quality and concentration was measured using a nanodrop 2000 spectrophotometer (Thermo Fisher Scientific Inc., Wilmington, DE) and electrophoresis in 1% agarose gels, then stored at − 20 °C prior to amplification using the primers listed in Table 1. All qPCR runs were completed in duplicate using 20 μl reactions consisting of: 10.0 μl of SYBRGreen (2×) PCR Master Mix (Life Technologies Corp., Carlsbad, CA, USA), 0.25 μl of each forward and reverse primer, 1 μl of DNA template, and 8.5 μl of nuclease-free water. All reactions were run on a Stratagene Mx3005P Real-Time PCR system (Stratagene, La Jolla CA, USA), with fluorescence measured during extension. The specificity of the PCR amplification was tested via the inspection of the melting curves that were prepared at the end of each PCR run. PCR products were also run on a gel to verify the presence of a single band of the correct size.
qPCR conditions for fungal ITS were 95 °C for 10 min followed by 40 cycles of: 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 30 s. Standard curves for fungal ITS were constructed using ten-fold serial dilutions of purified PCR products containing the ITS1 region of Fusarium avenaceum genomic DNA, which ranged from 103 to 109 gene copies. qPCR conditions for bacterial and archaeal 16S were 95° C for 10 min followed by 40 cycles of: 95 °C for 15 s, 53 °C for 30 s, and 72 °C for 20 s, and we used ten-fold serial dilutions of purified PCR products containing the 16S region of Pseudomonas putida genomic DNA, which ranged from 102 to 107 gene copies.
We quantified Cd-nitrite reductase (nirS) and Cu nitrite reductase (nirK) with qPCR conditions of 10 min at 95 °C and 40 cycles of: 95 °C for 1 min, 60 °C for 1 min and 72 °C for 1 min. The standard curve for nirS and nirK used tenfold serial dilutions of 101–107 gene copies from Pseudomonas putida. All gene copies were calculated using exact soil extraction weights, and are presented in analyses as log gene copies/ul/g of dry weight soil.
We quantified the number of amoA gene copies for bacterial ammonia oxidizers (amoA AOB) and archaeal ammonia oxidizers (amoA AOA). These primers generate amplification products of 491 bp for bacterial amoA and 440 bp for archaeal amoA, respectively. The qPCR amplification was performed at 95 °C for 10 min followed by 40 cycles of: 95 °C for 30 s, 57 °C/58 °C for 30 s and 72 °C for 60 s/45 s (AOA/AOB), and concluded with a melting curve analysis. The standard curves for AOA and AOB amoA were prepared from the fosmid clone 54d9 and the pCR 2.1-TOPO plasmid carrying AOB amoA from Nitrosomonas europaea ATCC19718 (Feld et al. 2015), using the primers amoA-1F and amoA-2R. Standard curves were constructed from the extracted plasmids using tenfold dilutions of 101–107 gene copies AOB amoA copies per microliter, and 101–106 gene copies AOA amoA copies per microliter.
We used a two-way analysis of variance (ANOVA), to test for differences in soil properties (pH, C%, N% and C:N) between tree species and sites (using replicate sample averages, such that there were 3 soil replicates and further two 2 technical replicates for all instrumental analyses of response variables within each of the three sites). We did not include an interaction term, as there was no replication of tree species plots within sites (see Supplementary Material for R code). We used Tukey tests and least square means to make paired comparisons between the six species. Overall effects of tree species on N transformation rates and soil microbial abundances were tested using linear mixed modes. We ran separate models for each response variable and included tree species as fixed factor and site as random factor. We used Tukey HSD for pairwise comparisons between tree species.
We explored relationships between site and soil properties, soil microbial gene abundances and N transformation rates in two ways. First, we used principal component analyses (PCA) to visualize the relationships between the three groups of data: gross and net N transformation rates, all six gene abundances, soil pH and soil C:N ratios. Another PCA was constructed to examine relationships between all six gene abundances, soil pH, and soil C:N ratios. Second, to examine whether tree species effects on microbial abundances were dependent on pH (based on the strong correlation with PC1 in the PCA), we used linear mixed models including tree species, pH and their interaction as fixed effects and site as random. We also explored similar relationships with C:N ratios and N transformation rates as well as gene abundances, which were less significant than pH relationships and are shown in the Supplementary Material section (and all versions of these figures with data presented at per g soil in Supplementary Material). Because of the large number of response and explanatory variables; and because not all relationships between them have a theoretical support, we only ran these analyses to test certain biologically meaningful relations (e.g. AOA and AOB with gross nitrification; focusing on C:N ratio and pH effects on N transformation rates and gene abundances in subsequent analyses). All analyses were completed in R version 3.3.1 (R Foundation for Statistical Computing 2016) using the vegan (Oksanen et al. 2013), MASS (Venables and Ripley 2002) and lmer (Bates et al. 2014) packages for the mixed effects models and principal components analyses, the ggplot2 (Wickham 2009) package for all bar charts, lsmeans (Lenth 2016) for Tukey’s post hoc comparisons. For linear mixed models all solid lines indicate significant relationships, and dashed lines indicate non-significant relationships.