| Description: |
correctionsFTIR spectra received automatic baseline, ATR, and atmospheric suppression corrections prior to analysis using Omnic Picta software (ThermoFisher, Waltham, MA). Spectral peak heights (within sample relative abundance) representative of different organic compounds were extracted from the spectra and used to calculate five decomposition indices (DI). Extracted peaks were 2920 cm-1, 2850 cm-1, both representing aliphatic compounds (lipids, fats, and waxes), 1630 cm-1, 1730 cm-1, both representing lignin and other aromatics, 1515 cm-1, representing lignin/phenol backbone, and 1030 cm-1 representing cellulose and other carbohydrates. The DIs were calculated by taking the ratios of each peak height to the 1030 cm-1 carbohydrate peak. These ratios represent the relative decomposability of a sample, where higher values correspond to higher relative abundance of more recalcitrant compounds, and lower values correspond to higher relative abundance of labile compounds.
additionalGeographic Extent: Samples were collected from the Arctic and boreal regions of Alaska, USA. Arctic samples were collected from sites at Sagwon Bluffs and the Lupine-Saganavirktok river confluence. Boreal samples were collected from sites colocated with BNZ LTER plots UP3a-c and WDI6. Instrumentation: BioTek Synergy HT microplate reader (BioTek, Winooski, VT, USA), elemental combustion analyzer (Costech ECS4010, Valencia, CA, USA), microwave plasma atomic emission spectrometer (Agilent 4100 MP-AES, Santa Clara, CA, USA), and Nicolet iN10 Infrared Microscope (ThermoFisher, Waltham, MA). Statistics/Algorithms: Full details can be found in the R code folder. A principal components analysis of the DIs (using the R function prcomp) was conducted on centered data. The first principal component (explaining 82% of variance) was retained and used in subsequent analysis as an indicator of SOM recalcitrance. Differences in DIs between stand types and distances from alder within stand types were compared using multiple response permutation procedure (MRPP) using the function mrpp in the R package vegan (Oksanen et al. 2018). Linear mixed effects models were used to assess the fixed effects of distance from alder (near or away), stand type (expanding or non-expanding in the arctic sites, aspen or spruce in the boreal forest sites), and SOM recalcitrance (PC1) on enzyme activities. The random effects of alder individual, site, and individual nested within site, were included to account for the paired structure of the data using the package nlme. Separate analyses were run with each enzyme and four soil quality indicators (PC1, C:N, C:P, and N:P) as response variables in the arctic and boreal regions. All global model fitting was done according to Zuur et al. (2009). When necessary, modeled variance structures were included to meet assumptions of heteroscedasticity. Boreal phenoloxidase activities were square root transformed to meet normality assumptions. Various random effect structures (random slope and intercept, random intercept only, and no random effect) were fit and compared using AICc. The random effect structure with the lowest AICc was maintained. Global models (all fixed effects) with applicable random effects and variance structures were fit using reduced maximum likelihood. Assumptions were visually assessed on the global model using quantile-quantile plots for normality and plots of residuals against fitted values of the full model and each predictor variable for heteroskedasticity.
StatusCompleted
|
