Supporting Information - S1

MATERIALS & METHODS SUPPORTING INFORMATION

Measuring 15NFrom Soil N pools

To obtain total dissolved nitrogen (TDN) samples three composited cores were extracted with an SMS® volumetric slide hammer (4.2 cm diameter from four zones along each 30 m transect in each plot) (12 total cores per plot, 4 field replicates). The cores were located within 2 m of resin or hyphal ingrowth bag positions equidistant along belt transects. The depths of the organic, mineral, and total core were recorded, green moss removed, and the horizons bagged separately. Only organic soils were used for isotopic analyses due to logistical limitations and because the majority of roots and fungal biomass in black spruce forest are located in the organic soil (Ruessand others 2003; Ruessand others 2006). Total volumes extracted from each plot averaged 3369 cm3 of wet soil and comprised 124 samples throughout the study. Each composited soil sample was stored at 4oC for approximately 24 h prior to homogenization and extraction. TDN was extracted from 24 g (wet weight) subsamples placed in a pre-cleaned and acid washed 250 mL HDPE plastic cup with 120 mL of 2M KCl in nano-pure (Barnstead Thermo Scientific®) water. For each new batch of 2M KCl prepared, four 60 mL blanks were taken to correct for 15N contamination (Knapp and others 2005). Soil extractions were shaken on a reciprocal table for 20 minutes and left undisturbed for 18 to 24 hours. The resulting supernatant was vacuum filtrated through glass fiber filter papers (Whatman® 1820-070) in HDPE Buchner funnels at the Ecosystem Ecology Laboratory, University of Alaska, Fairbanks, and shipped frozen to UF for subsequent processing.

Replicate 6 mL subsets of the 2M KCl extracted TDN and acidified salt (0.1M HCL / 2M NaCl) extracted NH4+ ions were oxidized to NO3- using a modified peroxodisulfate thermo-oxidation(Cabrera and Beare 1993; Doyle and others 2004). For each oxidation, standard solutions of 1, 5, and 10 mg of KNO3-, glycine, and 6-aminocaproic acid (ACA) were used to assess digestion efficiency and potential isotope fractionation during the digestion procedure. Digestion efficiency of the high concentration (for example, 10 mg/L) of the highly recalcitrant ACA standard was improved to greater than 87% at lower salinity (0.5M) presumably due to the interference of Cl- with free radical generation (Peyton 1993). Accordingly, all digested samples were diluted with nano-pure water prior to oxidation prior to 15N analyses.15N values from the NO3-ions resulting from the oxidations were measured using the bacterial denitrifier technique (Sigman and others 2001; Knapp and others 2005). This method involves the use of naturally occurring denitrifying bacterial strains (Pseudomonas aureofasciens) to convert potentially small (~0.5 µg N) amounts of NO3-to N2O for isotopic analyses.Oxidized NO3-samples were pH adjusted prior to delivery to denitrifying bacteria. Consistent 15N values were found for N3 standards when prepared in H2O, KCl, and the acidified NaCl solutions. All NO3-samples were analyzed on a GC-Pal® gas sampler arm coupled to a ThermoFinnigan® continuous flow isotope ratio mass spectrometer at UF. Following blank and/or reagent corrections, the 15N values of extracted dissolved organic N (DON) was calculated using the following mass balance equation:

15NDON = (15NTDN × [TDN] – (15NNH4 × [NH4+] + 15NNO3 × [NO3-])) / [DON] (1)

where the “15NNH4” and “15NNO3”values were derived from the field-incubated ion exchange resins. 15N measurements from incubated ion exchange resins, unlike single time point salt extractions, integrate episodic production and removal processes (Jones and Kielland 2002; Booth and others 2005) without many potential soil sampling artifacts (Hales and Ross 2008). Previous tests on the exchange resins employed herein indicated no isotope fractionation effects (Lehmann and others 2001; Templer and Weathers 2011).

Statistical Modeling

Linear regressions were constructed on the following response variables: foliar 15N, %N, and %P; the soil 15N values of extractable DON and NH4+; standing aboveground biomass of black spruce; and, belowground biomass of fungi. For each model we wished to reduce our explanatory variables to prevent inflation of Type I error rates (Harrell 2001) yet still address specific hypothetical relationships. In particular, most soil fertility variables were reduced to a single variable with Principle Component Analysis (PCA). This partial variable reduction allowed specific variables of interest to be retained in subsequent models based on the response variable of interest. For instance, when modeling foliar N content, both DON and resin-extractable mineral N were retained as independent predictors and the remaining soil fertility variables were reduced to a single PC axis. Similarly, for foliar P, resin extractable PO4-was withheld from that variable’s fertility PCA, and in the case of foliar 15N, the 15N values of the soil N forms were withheld. Fungal biomass was withheld from all soil fertility PCA’s so that the explanatory power of this variable could be independently assessed. Predictor variables were standardized by subtracting from the mean and dividing by the standard deviation (Schielzeth, 2010) prior to regression analyses in R (2.11.1, The R® Foundation for statistical computing 2010).Independence of explanatory variables was assessed with scatter plot matrices and variance inflation factors to assure a lack of co-linearity and to detect severe outliers. PCA’s were conducted using JMP® 8.0.2 (SAS Institute Inc., Cary, NC, USA).

Second-order bias-corrected Akaike information criterion (∆i= AICci – AICcmin) was used to rank models because it is generally regarded as an unbiased estimator that assesses relative model fit, penalizes over-parameterization, and allows multiple working hypotheses to be simultaneously evaluated (Burnham and Anderson 2004; Andersen 2008). In contrast, fitting all possible models or using stepwise model selection has potential to inflate Type I error rates and to occasionally fail in selecting the most informative model (Whittingham and others 2006; Andersen 2008) but see Murtaugh (2009). Graphical diagnostics of residuals against fitted values and sample quantile against theoretical quantile plots were used to assess underlying distributional assumptions for all high-ranking models.

References

Andersen DR. 2008.Model based inference in the life sciences: A primer on evidence. NY: Springer.

Booth MS, Stark JM, Rastetter E. 2005. Controls on nitrogen cycling in terrestrial ecosystems: A synthetic analysis of literature data. Ecological Monographs75: 139-157.

Burnham KP, Anderson DR. 2004. Multimodel inference. Sociological Methods & Research33: 261-304.

Cabrera ML, Beare MH. 1993. Alkaline persulfate oxidation for determining total nitrogen in microbial biomass extracts. Soil Science Society of America Journal57: 1007-1012.

Doyle A, Weintraub MN, Schimel JP. 2004. Persulfate digestion and simultaneous colorimetric analysis of carbon and nitrogen in soil extracts. Soil Science Society of America Journal68: 669-676.

Hales HC, Ross DS. 2008. Drastic short-term changes in the isotopic composition of soil nitrate in forest soil samples. Soil Science Society of America Journal72: 1645-1652.

Harrell FE. 2001.Regression modeling strategies: with applications to linear models, logistic regression, and survival analysis. New York: Springer-Verlag.

Jones DL, Kielland K. 2002. Soil amino acid turnover dominates the nitrogen flux in permafrost-dominated taiga forest soils. Soil Biology & Biochemistry34: 209-219.

Knapp AN, Sigman DM, Lipschultz F. 2005. N isotopic composition of dissolved organic nitrogen and nitrate at the Bermuda Atlantic Time-series Study site. Global Biogeochemical Cycles19: GB1018.

Lehmann MF, Bernasconi SM, McKenzie JA. 2001.A method for the extraction of ammonium from freshwaters for nitrogen isotope analysis. Analytical Chemistry73: 4717-4721.

Murtaugh PA. 2009. Performance of several variable-selection methods applied to real ecological data. Ecology Letters12: 1061-1068.

Peyton GR. 1993. The free-radical chemistry of persulfate-based total organic carbon analyzers. Marine Chemistry41: 91-103.

Ruess R, Henkdrick RL, Burton AJ, PregitzerKS, Sveinbjornsson B, Allen MF, Maurer GE. 2003. Coupling fine root dynamics with ecosystem carbon cycling in black spruce forests of interior Alaska. Ecological Monographs73: 643-662.

Ruess RW, Hendrick RL, Vogel JG, Sveinbjornsson B 2006. The role of fine roots in the functioning of boreal forests. In: Chapin III FS, Oswood MW, Van Cleve K, Viereck LA, Verbyla DL eds. Alaska's changing boreal forests. New York, NY: OxfordUniversity Press.

Schielzeth H. 2010. Simple means to improve the interpretability of regression coefficients. Methods in Ecology and Evolution.doi: 10.1111/j.2041-210X.2010.00012.x

Sigman DM, Casciotti KL, Andreani M, Barford C, Galanter M, Bohlke JK. 2001. A bacterial method for the nitrogen isotopic analysis of nitrate in seawater and freshwater. Analytical Chemistry73: 4145-4153.

Templer PH, Weathers KC. 2011. Use of mixed ion exchange resin and the denitrifier method to determine isotopic values of nitrate in atmospheric deposition and canopy throughfall. Atmospheric Environment45: 2017-2020.

Whittingham MJ, Stephens PA, Bradbury RB, Freckleton RP. 2006. Why do we still use stepwise modelling in ecology and behavior? Journal of Animal Ecology75: 1182-1189.

1Mayor et al.