Supplemental text and figures, analytical code, and full dataset documenting compositional differences and results of incubations for effects on dissolved organic carbon concentration and character.

Data and analysis in this resource describe stream samples collected in late summer of 2016 and 2017 from seven study regions selected to include Arctic, Boreal, and alpine ecosystem types and to represent a range of current and future climatic conditions in the permafrost zone (continuous, discontinuous, and non-permafrost). Water samples were collected from three or more locations within each study region. Selected sites were nested in river networks, except in interior Alaska, where the three sites came from independent streams. The seven sampled regions include broad variation in climate, geology, topography, and vegetation. In all permafrost-affected regions, various types of permafrost degradation have been observed, and other forms of less visible permafrost warming and degradation are also occurring. Though permafrost degradation is present in all the studied permafrost catchments, three of the seven regions were specifically chosen for their proximity to abrupt thaw features. Please see the primary manuscript for site details, complete methods description, statistical analyses, and citations to relevant literature.

Incubations were performed locally by each regional team, and samples were shipped to centralized locations for analysis. Stream water was filtered on site (0.7 m, Whatman GF/F) and refrigerated until laboratory incubations were initiated. We divided the filtered bulk stream samples into 200-mL aliquots and treated each aliquot with one of eight acetate (CH3COO-) and nutrient treatments (Table S1). We used acetate as the priming substrate in these experiments We used ammonium (NH4+), nitrate (NO3-), and phosphate (PO43-) as the inorganic nutrient substrates. Treatments were added only at the start of the incubations to simulate mixing of permafrost thaw products with modern DOM in stream networks.

Inorganic nutrients (NH4+, NO3-, NO2-, and PO43-) in unamended (background) stream waters were determined at µg L-1 levels on a QuAAtro39 continuous segmented flow analyzer (Seal Analytical, Inc.). We calculated dissolved inorganic nitrogen (DIN) as the sum of NH4+, NO3-, and NO2-. Acetate and other dissolved solutes in the treated incubation samples (NO3-, NO2-, Cl-) were measured at mg L-1 levels on an ICS 2100 Ion Chromatograph (Dionex, Thermo Scientific) equipped with an anion column (ASX-18 column). DOC and total nitrogen (TN) in all samples were determined using a V-TOC CSH Total Carbon Auto-Analyzer with a TNM-1 Total Nitrogen Module (Shimadzu Corporation).

We collected additional subsamples at t0 and t28 from a subset of the treatments (CT, A3, and AN) for optical analysis via fluorescence spectroscopy to evaluate indices of DOM composition. These subsamples were filter sterilized (0.22 µm, PES) into 40 mL amber glass vials and stored in the dark at 4˚C during shipment and until analysis. We measured the absorbance and fluorescence of these subsamples with a spectrofluorometer (Aqualog, Horiba Scientific, Edison, New Jersey). Detailed analysis of DOM chemical composition was performed for only the CT and A3 treatments at the t0 and t28 timesteps for a subset of sites (a total of 33 samples) via ultrahigh resolution mass spectrometry with a 21 T FT-ICR MS. These subsamples were filtered to 0.7 m (GF/F pre-combusted at 450oC for 5 hours) and stored frozen until analysis.

To calculate rates of acetate and background DOC consumption, we poured off and froze 15-mL subsamples immediately following the addition of treatments (t0), after 7 days (t7), and after 28 days (t28). We calculated change in background DOC and acetate for each replicate individually and then calculated the mean and standard deviation of ΔDOC and ΔAcetate across the three replicates for each site and timestep. We calculated change in optical properties (ΔOptical) and relative abundance (ΔRA) of molecular composition in the same way as ΔDOC and ΔAcetate. We calculated priming and nutrient effects for each site as the ΔDOC in each treatment minus the ΔDOC in the unamended (control) treatment. This yielded positive values for the nutrient and priming effects when the treatment resulted in greater background DOC consumption (i.e. positive priming) and negative values when the treatment DOC consumption was less than the control (i.e. negative priming).