Sampling and analysis
Lateral subsurface flow begins mid to late April at the start of snowmelt, increases to the peak in early to mid-May, then declines and halts by mid to late June (Fig 1a, Troendle & Reuss, 1997). We collected hillslope export samples as lateral subsurface flow at the base of each hillslope (i.e., at the piped trenches) weekly in 2016-2018 and analyzed for nutrients, dissolved organic carbon (DOC), total dissolved nitrogen (TDN), and DOM chemical characteristics. These samples were collected in pre-combusted (heated for 3 hours at 500 ⁰C) amber glass bottles then filtered through 0.7 µm pore-size glass fiber filters (Millipore Corp, Burlington, MA) within 24 hours of collection and analyzed for DOC, TDN, and DOM fluorescence spectroscopy. We collected DOM reactivity samples in 250 mL pre-combusted amber glass bottles, then filtered through (GF/C Whatman®, 1.2 µm effective pore size) filters within 3 hours of collection to remove bacterial grazers. DOC and TDN were determined using a Shimadzu TOC-VCPNtotal organic carbon analyzer, with 2M HCl addition before analysis to remove mineral C (Shimadzu Corporation Columbia, MD). Detection limits for DOC and TDN were 50 µg L-1.
The chemical character of DOM exported from each hillslope (i.e., lateral subsurface flow) was analyzed using UV-Visible fluorescence spectroscopy on a Horiba Scientific Aqualog (Horiba-Jobin Yvone Scientific Edison, New Jersey, US). Sample C concentration was standardized to 5 mg C L-1 using deionized water (>18 mΩ). Ultraviolet absorbance was analyzed at the 254 nm wavelength. Excitation emission matrices (EEMs) fluorescence scans were completed from 240 nm-600 nm excitation and emission wavelengths, with 3 nm band-pass, 3 nm increments, and 3 second integrations. Scans were blank corrected using deionized water and corrected for inner filter effects (Kubista, Sjoback, Eriksson, & Albinsson, 1994). First- and second-order Rayleigh scattering were masked (10 nm width masking), and samples were normalized by the area of the deionized water Raman scattering peak (Lawaetz & Stedmon, 2009).
SUVA254 (L mg C-1m-1) was calculated by dividing ultraviolet absorbance at 254 nm by DOC concentration (Weishaar et al., 2003). Humification index (HIX) was calculated as the area of spectra collected at 254 nm emission under 435-480 nm excitation divided by the area under the peak 300-345 nm emission (Zsolnay, Baigar, Jimenez, Steinweg, & Saccomandi, 1999). Fluorescence Index (FI) was calculated as the ratio between emission at 470 nm and 520 nm at 370 nm excitation (McKnight et al., 2001; Cory & McKnight, 2005). Fluorescence regional integration modelling (FRI) was applied to EEMs data to account for areas of increased fluorescence not identified by standard EEMs indices (Chen, Westerhoff, Leenheer, & Booksh, 2003). Briefly, EEMs scans were separated into five regions associated with distinct DOM chemical components. Total fluorescence intensity for each region was divided by total intensity of the entire scan, resulting in percentages of total FDOM derived from each DOM chemical component region.
We sampled the organic (O horizon) and mineral (A horizon; 10 cm depth) soil layers at ten locations distributed across the old-growth and second-growth hillslopes (n=20). O horizon material was sampled within a 30 cm by 30 cm quadrat and total O-horizon depth was measured in each quadrat corner. O horizons were separated into litter and duff layers as follows (FIA 2019): needles and recognizable plant material (< 6 mm diameter) were classified as litter, while unrecognizable, fragmented material between the litter and mineral soil layers was considered duff. There was no measurable duff in the second-growth forest. Mineral soil (0-10 cm depth) was collected using a 7.5 cm diameter corer after the O horizon was removed. Mineral soils had rocks, mosses, and lichens removed and were then sieved to 2 mm. We determined gravimetric soil moisture after drying a subsample at 105 ⁰C for 48 hours. Organic horizon samples were well mixed and moss, lichen, and rocks were removed by hand. Organic horizon samples were dried for 6 days at 60 ⁰C and the total mass was recorded. Total bulk density for mineral and organic soils was calculated by dividing the dry weight of the total soil (including > 2 mm fraction in mineral soils) by the sample volume. A subset of each sample was dried for 48 hours at 60 ⁰C, then ground and analyzed for total C and N by dry combustion (LECO 1000 CHN analyzer, LECO Corporation, St. Joseph, MI, USA). Total C and N pools were calculated by multiplying C and N concentrations by the associated O- and A-horizon masses. Water extractable soil organic matter (WEOM) samples were created by leaching a litter layer sample from each hillslope’s O-horizon to replicate DOM inputs (Sparling, Vojvodic-Vukovic, & Schipper, 1998). Litter samples were air dried for 3 days then 10 g subsamples were steeped in 50 mL of DI (>18 mΩ) at 70 ⁰C for 18 hours, shaken and filtered through 0.7 µm pore size glass fiber filters.
Biological oxygen demand assays were performed to assess the reactivity of DOM inputs (WEOM litter leachates) and DOM exports (lateral subsurface flow). DOM input incubations were diluted and standardized to 30 mg C L-1 to mirror C concentrations of DOM export incubations. Each DOM input sample was inoculated with 88 mL of water from an adjacent stream (Sparling et al. 1998), and export DOM was incubated with microbes that occurred naturally in the subsurface flow. Relationships in reactivity between input and output DOM should be considered tentative correlations, as different microbial cultures were used in litter and subsurface leachate incubations. Experimental controls, which only contained the stream water microbial culture with no additional DOM and analytical controls which only contained DI (>18 mΩ) were incubated simultaneously with the oxygen demand assays. Microbial dissolved oxygen consumption (mg O2 L-1) was continuously measured (15 second intervals) using Oxy-4 probes (PreSens, Precision Sensing GmbH, Regensburg, Germany). Input and export samples were incubated in the dark at 20 ⁰C and the hourly oxygen consumption rate (mg O2 L-1 hr-1) was averaged for each sample over the 24-hour incubation period. No samples approached hypoxic conditions (<4 mg O2L-1). Oxygen consumption rates were normalized per gram of C to remove any bias of C quantity on consumption rate. Post-incubation samples were filtered to remove microbial biomass (Nucleopore®, polycarbonate filter, 0.2 µm). Concentration of C mineralized (Δ mg C L-1) was calculated as the difference between measured pre- and post- incubation DOC concentrations, then converted to loss as CO2(ΔCO2, in mg CO2 L-1), under the assumption all CO2 produced by incubations was removed from the sample with HCl addition during C measurement. The respiratory quotient (RQ), was calculated by dividing C lost to respiration (ΔCO2, in mg CO2L-1) by the change in O during the incubation (ΔO2 in mg O2 L-1).
DOC and TDN concentrations, and FDOM indices in subsurface flow exports were compared between old-growth and second-growth land cover types using Welch-Satterthwaite unequal variance t-test with significance assigned at p < 0.01 (R Core Team, 2019). Regional percentages for FDOM components (FRI modelling) in export samples were logit transformed before statistical analyses to remove biases inherent in proportional data. We compared input nutrient concentrations and pools in organic and mineral horizon soils, along with comparisons of oxygen consumption and respiratory quotients in DOM input and export samples using the Student’s t-tests for parametric data, with significance assigned at p < 0.01 (t.test function, R Core Team, 2019). The data that support the findings of this study are available from the corresponding author upon reasonable request.