This data set is a report of chemistry for surface waters since 1986 draining the S2 peatland and surrounding uplands (the S2 catchment) at the Marcell Experimental Forest (MEF) in Itasca County, Minnesota. Surface water was collected every week or two, and sometimes more often, starting during the 1970s. Sporadic sampling occurred as early as 1966. We only report values since 1986 when metadata are more complete. We report a subset of analytes from 1986 to 1992. A broader suite of analytes is reported after 1992. Some analytes were only measured for several years, were interrupted for periods, and were added at various points throughout the period. As metadata are better described and more data become available through ongoing sampling and analysis efforts, we will update this data product.
Samples are routinely measured for a core suite of pH, specific conductivity, anions (chloride, sulfate), cations (calcium, magnesium, potassium, sodium, aluminum, iron, manganese, strontium), silicon, nutrients (ammonium, nitrate+nitrite, soluble reactive phosphorus, total nitrogen, total phosphorus), and total organic carbon (TOC). Occasionally, stable water and mercury isotopes as well as concentrations of dissolved organic carbon (DOC), bacterial respiration of dissolved organic matter (DOM), biodegradable DOC (BDOC), ferrous and ferric iron, total mercury (filtered or unfiltered), methylmercury (filtered or unfiltered), and lead were measured. Ultraviolet (UV) absorbance, a measure of water color or DOM optical properties, was also measured for some samples.
The MEF is operated and maintained by the USDA Forest Service, Northern Research Station. The S2 catchment is located on US federal government land that is part of the Chippewa National Forest.
We request courtesy notification to stephen.sebestyen@usda.gov if the data are used in a publication, that this data release be fully cited, and that the Northern Research Station of the USDA Forest Service be acknowledged for funding the long-term monitoring program at the MEF. Contact the lead author for more information or when seeking guidance on research regarding these data. The Forest Service MEF research team welcomes collaborative engagement regarding these and other MEF data.
SITE DESCRIPTION:
The S2 catchment has a 6.5-ha deciduous upland forest and a natural, undrained 3.2-ha peatland (raised-dome ombrotrophic bog with a surrounding lagg). A stream forms in the lagg and flow is intermittent throughout the year (Verry et al. 2011). A lagg is the transitional peatland area between a bog and surrounding mineral soil uplands. Streamflow occurs during and after snowmelt and rainfall events. In some years, there is streamflow from snowmelt to freeze-up during the following winter. In most years, there is a period of no streamflow during summer that may extend into fall or winter, and in most years, streamflow does not persist through winter.
Surface elevation ranges from 420.5 m a.s.l at the outlet to 430 m a.s.l. in the uplands. In the uplands, a Warba sandy clay loam developed in glacial till atop 50-m deep outwash sand deposits. The Warba soil series is a fine-loamy, mixed, superactive, frigid Haplic Glossudalfs (an Alfisol; Nyberg 1987). Peat depth has been surveyed across the bog (Verry and Janssens 2011). Peat is less than 1 m deep around the perimeter of the bog to about 7 m at the deepest location. The Loxley peat (Dysic, frigid Typic Haplosaprists; a Histosol; Nyberg 1987) has accumulated in the last 10,000 yr since Wisconsin glaciation (Verry and Janssens 2011). The peatland surface has hummock and hollow microtopography. Hummocks are uneven, elevated areas that rise various heights, up to about 50 cm, above adjacent hollows. Hollows have a relatively uniform elevation within a localized area, with an overall raised-dome profile across the bog surface and a peat surface that gently slopes towards the outlet stream. The peatland water table fluctuates from about 0.10 m above the hollow surface to as much as 0.30 m below during a typical year.
The upland forest was last harvested during the 1910s. The upland forest currently is dominated by aspen (Populus tremuloides), white birch (Betula papyrifera), red maple (Acer rubrum), and balsam fir (Abies balsamea), with some red oak (Quercus rubra), basswood (Tilia americana), and jack pine (Pinus banksiana). Jack pine is limited to a small patch adjacent to the southern edge of the peatland. There has been an observable shift in the dominance of aspen over the last 10 to 15 years as aspen have died. Gaps in aspen have opened to growth of red maple, balsam fir, red oak, and basswood into the canopy.
The peatland has a black spruce (Picea mariana)-tamarack (Larix laricina)-Sphagnum community. Below the overstory tree canopy, there is variable coverage of ericaceous shrubs (Rhododendron groenlandicum, Chamaedaphne calyculata, and Vaccinium angustifolium), cotton grass (Eriophorum spissum), Sphagnum moss, or haircap moss (Polytrichum spp.) across the bog. Three-leaved false Solomon’s seal (Maianthemum trifolium), common pitcher plant (Sarracenia purpurea), and pink moccasin flower orchid (or stemless ladyslipper; Cypripedium acaule) are found throughout the bog. The lagg has most of the same species but is richer in species than the bog (Verry and Janssens 2011). The more noticeable additional species include speckled alder (Alnus incana), paper birch (Betula pumila), ferns, and various sedges (Carex species).
The climate is continental with warm summers, cold winters, and a mean annual air temperature since 1961 of 3.5 deg C (1961 to 2019, Sebestyen et al. 2021b). Air temperature ranges as low as -46 degrees Celsius to as high as 38 degrees Celsius. Mean annual precipitation since 1961 is 787 mm. Most precipitation occurs as rainfall during summer. A winter snowpack starts to accumulate during November or December and fully melts during March or April. There are limited days of mid-winter thaw and those melt events rarely induce much or any streamflow.
At the S2 catchment, snow depth, snow water equivalent, precipitation, ground frost, upland runoff volume, streamflow, and water levels have been monitored, with some measurements as early as 1960 (Sebestyen et al. 2021b). Some chemistry (mostly unpublished) was measured as early as 1966. Surface water draining the S2 peatland is acidic (mean pH = 3.7 at the S2 LAGG POOL from 1992 to 2021) and highly colored with low concentrations of ions and nutrients and high concentrations of DOM (mean TOC concentration +/- 1 standard deviation = 67.6 +/- 19.3 mg/L at S2 LAGG POOL from 1992 to 2021).
LOCATIONS OF WATER SAMPLING:
Water has been repeatedly sampled from two locations. Although the protocol is to only sample when the stream flows, samples were rarely, but sometimes collected when there was no streamflow over the v-notch weir. While the long-term sampling has been at least every two weeks, some samples were collected for separate projects and multiple samples may have been collected on the same day or throughout a week.
Samples have been collected for the longest period (starting 1986 in this data release) from a location about 100 m upstream of a weir. These samples are named S2 LAGG POOL, which is upstream of a discernable stream channel in the general area where the lagg coalesces into the stream. The lagg pool was excavated in peat sometime prior to 1981 (Urban 1983) and a 0.1-mm stainless-steel mesh box was placed there to maintain standing open water for sampling. The pool has infilled over the years as dead biomass has accumulated. It is currently about 30 cm deep and about 50 cm across and the stainless-steel mesh is no longer visible. Because the peatland has a compressible, organic soil that is prone to disturbance, and there are extended periods of standing water on the approach to and surrounding the pool, an elevated boardwalk was constructed during 2011 to access the lagg pool for sampling. Prior to that, the pool was approached by walking on the lagg surface. The area is surrounded by black spruce, tamarack, ericaceous shrubs, Sphagnum, and other lagg plant species (Verry and Janssens 2011). The lagg has hummock and hollow microtopography. The immediate area around the pool is relatively flat and low lying. Samples from the S2 LAGG POOL have occasionally been supplemented with grab samples of stream water at the S2 WEIR to maintain long-term biweekly sampling when the stream was flowing but the S2 LAGG POOL was ice covered.
Beyond those occasional samples during ice cover at the S2 LAGG POOL, stream water at the v-notch weir has been collected for various studies starting during 2001 for various analytes and routinely each week since 2008 for the core suite of analytes. The weir is used to measure stream stage and represents the outlet of the entire S2 catchment. Streamflow is calculated from stream stage (Hertzler 1938) and reported in units of cm/d. Between the S2 LAGG POOL and the weir, the stream flows over mineral soil. Stream water is exposed to sunlight when the deciduous canopy has no leaves and in the weir pool. Samples from the weir are named S2 WEIR and are collected when the stream flows. These samples are either grab samples of water falling over the v-notch (2001 to ongoing) or were autosampled (2008 to 2013) from the weir pool. From about October or November until high flow during snowmelt, an insulated cover is placed over the weir. This shelter is heated with a propane lamp to avoid ice accumulation in the v-notch and to allow sampling during freezing conditions.
From April 2001 to June 2011, grab samples were collected about every two weeks at the S2 WEIR to document reference conditions in the S2 catchment relative to a sulfate and mercury cycling experiment at the S6 catchment (Jeremiason et al. 2006, Coleman Wasik et al. 2012, Coleman Wasik et al. 2015). Samples were collected at higher frequency (one to several days between samples) corresponding to three times each year (2001-2008) when sulfate was added to the S6 bog to mimic atmospheric deposition levels of the late 1970s and from 2009 to 2011 to monitor recovery from the experimental sulfate addition. These filtered samples were analyzed for total mercury and methylmercury concentrations from 2001 to 2011, dissolved organic carbon from 2005 to 2008, chloride and sulfate concentrations from 2006 to 2011, and the core suite of analytes from 2009 to 2011. There were some years with few or no samples during December even if the stream was still flowing.
Other grab samples were collected for studies of DOM biodegradability (sampling from 2009 to 2012; Sebestyen et al. 2021), lead mobilization and transport (2009 to ongoing; Jeremiason et al. 2018), Hg transport and cycling (1993 to ongoing with some gap years; Kolka et al. 2011), and DOM optical properties (2009 to 2012 and 2014 to ongoing; unpublished research).
Stormflow samples were collected from water pooled behind the V-notch using an Isco 3700 portable sampler (Teledyne Isco, Lincoln, Nebraska, USA) from April 2008 to August 2013. From April to June of 2008, automated sample collection was triggered every two hours on some days when stormflow was expected based on precipitation forecasts. From July 2008 to July 2013, automated sample collection was triggered by threshold changes in absolute stream stage, as recorded and actuated from a program running on a datalogger. The threshold value was changed over time, but the general goal was to collect several samples over the rise and fall of stream stage. For example, during 2010, a change in stage from 0 to 8 cm resulted in the actuation of sampling six times as stream stage was rising. Samples may have been collected anywhere from minutes to days apart depending on the magnitude of stormflow and rate of change in streamflow. Not every sample was kept. While the streamflow record (Verry et al. 2018) was derived from stage measured with a Leupold and Stevens Inc. (Beaverton, Oregon, USA) Type A-35 stripchart recorder (0.3-cm precision), the Isco sampler was triggered by a Campbell Scientific (Logan, Utah, USA) CR1000 datalogger with a float-tape-counterweight driven shaft encoder (Vaisala, Louisville, Colorado, Handar 436b encoder, 100 measurements per rotation of a 30.5 cm diameter pulley). A 5:1 reduction gear was used to increase the precision of stage measurements to 0.06 cm. Stage was measured every 10 seconds. These samples were analyzed for the core suite of analytes and occasional samples were analyzed for water isotopes.
Sampling location, whether at the S2 LAGG POOL or at the WEIR, is important to consider, especially for some solutes. Sample location, temporal resolution, and the period of sampling all need to be weighed relative to the research questions that are being asked. While not the catchment outlet, it is logical to use samples from the S2 LAGG POOL and to include occasional samples from the S2 WEIR when the S2 LAGG POOL was ice covered for research that requires the longest and most complete record. Accordingly, a chemistry value will be available when the stream was flowing despite the inaccessibility of the S2 LAGG POOL. However, dates and times of samples need to be assessed by data users to identify samples that are relevant to any particular research objective. It is otherwise not likely advisable to mix samples from the LAGG POOL and WEIR for research on temporal trends. Differences in some solute concentrations between the LAGG POOL and WEIR could lead to a false assessment of patterns and trends, especially during low streamflow. The record of chemistry from the S2 WEIR should likely be used for research that is focused on periods that begin after weekly sampling started during 2008. The 2008 to 2013 period also deserves consideration relative to what level of temporal resolution is most appropriate to address particular research questions. For example, emphasis was placed on sampling during stormflow and snowmelt events using grab sampling up to several times a week and higher temporal resolution with autosampling. Comparing periods with only weekly sampling to periods with autosampling may limit direct comparisons if the chemistry was skewed by additional representation of chemistry during stormflow. Since storm samples have not been collected at other MEF catchments, sub-weekly samples should likely be excluded from direct comparisons of chemistry to other catchments. Given these considerations, the chemistry values are presented in three separate comma separated value (CSV) tables: one table for the S2 LAGG POOL (S2_streamwater_chemistry_LAGG_POOL.csv), one for grab samples from the S2 WEIR (S2_streamwater_chemistry_grabSamples.csv), and another for autosampled waters from the S2 WEIR (S2_streamwater_chemistry_autoSamples.csv). Regardless of intended data use, it is advisable to only use a subset of values that are directly relevant to specific research objectives. Since each sample receives a unique serial identifier (ID), we also advocate that data users specify in a list that accompanies a research publication the serial IDs of samples that were used in any particular data analysis.
WATER SAMPLING AND STORAGE PRIOR TO ANALYSIS:
Unfiltered water was used for most chemistry analyses. If an aliquot was filtered for analysis of a particular solute, the sample may have been filtered later in the field or in a laboratory.
At the S2 LAGG POOL, samples were dipped with a plastic kitchen ladle and poured into bottles. After about April 2010, the ladle or a polyethylene dipper (CXBA00, Global Water Instrumentation, Phoenix, Arizona) with an approximately 0.5-m handle was used. The ladle or dipper was rinsed with 18 megaohm deionized water before the start of sampling. Samples are collected from S2 and other MEF catchments, usually with seven to twenty samples collected on any particular day. At the S2 LAGG POOL, the ladle or dipper was rinsed three times with water from the pool before sampling. The rinse water was discarded outside of and away from the sampling pool.
For S2 WEIR grab samples, stream water flowing over the v-notch was collected in sample bottles.
The autosampler was placed on a wooden platform that was about 1.5 m higher in elevation than the weir pool surface. A standard Isco weighted polypropylene strainer on the end of the tube (1-cm or 3/8 inch internal-diameter reinforced vinyl tubing) was suspended from a float in the pool about 30 cm behind the v-notch and about 10 cm beneath the water surface. Suction was drawn along an approximately 1-m section of Silastic (medical-grade silicon rubber) tubing, connected to a 5-6-m long vinyl tube, by the Isco peristaltic pump. The silicon and vinyl tubing were connected with an approximately 5-cm stainless steel adaptor (0.6 cm diameter). The base of the Isco held twenty-four 1000-mL polypropylene bottles (standard wedge shaped Isco 3700 bottles that fit inside the perimeter of the circular base). At the start of a sampling sequence, sampling time was recorded and air was backflushed through the line to purge any water. A sample of about 800 mL was then pumped into a bottle, followed by another backflush cycle to purge water from the tube. During sampling, the light environment was similar to exposure in the weir pond because about 2 m of tube was underwater in the weir pool before passing underground through schedule 80 PVC for about 3 m before entering the Isco. Pump tubing was changed as needed if dirty or worn from use. The Isco bottles were returned to the Grand Rapids Laboratory and replaced with clean, acid-washed bottles every two to three months, or if noticeably dirty. Water was poured from an Isco bottle into a sample bottle upon retrieval after 3x rinsing of sample bottles. Usually six or fewer bottles filled between retrievals, and the used bottles were sometimes replaced with other unused bottles in the base before all bottles were retrieved and washed. Samples were retrieved on weekdays. Samples usually remained in a sampler if collected after business hours on Friday, over a weekend, or on a holiday until the next business day. Samples were typically transferred from Isco bottles to sample bottles in the field. Samples that froze inside the autosampler were returned to the Forestry Sciences Laboratory in Grand Rapids to thaw and were then transferred to sample bottles. Once maximum air temperatures consistently remained below freezing, the autosamplers were turned off.
Unfiltered water was collected in a new 250-mL low-density polyethylene (LDPE) sample bottle after rinsing 3x with sample water. Samples were refrigerated in the dark until analysis. For a period, samples were collected in new Whirl-Pak (Madison, Wisconsin, USA) bags, but the range of dates is not currently known but was prior to 2000. From here on, we refer only to the 250-mL bottles, but a sample may have been collected in, stored in, and an aliquot transferred from a Whirl-Pak bag for the core suite of analytes rather than from 250-mL bottles. With the start of sampling for water isotopes during 2007, a 16-mL glass scintillation was filled with water (no rinse) and stored at room temperature. Starting during 2019, a new 60-mL high-density polyethylene (HDPE) bottle was filled after rinsing 3x with sample water, and chilled until frozen upon transport to the Grand Rapids Laboratory. Sometimes, other aliquots were collected, as described below, for particular periods and studies.
Unfiltered water was used for most laboratory analyses. It is important to keep in mind that surface waters in peatlands are free of inorganic particulates due to flow paths through peat and slow transit times due to low hydraulic gradients that allow for deposition and retention of particulates on Sphagnum mosses or peat. For that reason, we consider solutes in unfiltered surface water samples to be dissolved. The samples are likely to include colloids, but no inorganic particulates and rarely peat particles. Attempts are made to avoid or eliminate aquatic organisms (mostly mosquito larvae during late spring when abundant) or plant leaves and needles.
Unless otherwise noted, samples were placed inside chilled coolers within minutes of collection. When returned to the Marcell Field Laboratory (before 2006), Marcell Research Center (after 2006), or the Forestry Sciences Laboratory in Grand Rapids, samples were chilled to 4 degrees Celsius in a refrigerator until analyzed, unless otherwise noted.
When collected, date/time of retrieval, sample location, and associated notes were recorded on field data sheets. A unique serial ID number was assigned to all aliquots of the same sample for tracking purposes in the laboratory and data reporting. The serial ID (at a minimum), date/time, and sample location were also written on label tape on each sample bottle or vial.
Occasionally, a field duplicate was collected. These duplicates can be identified by two consecutive serial ID values that have the same timestamp (date, hour and minute).
Time was recorded as hour (Central Standard Time) and minute. Minutes were oftentimes recorded to the closest 5 minute interval (e.g., 0, 5, 10, 15, etc. minutes after the hour) for grab samples. Sampling and recording on field data sheets generally took several minutes, but no longer than 5 minutes to complete. Occasionally, time was not recorded which results in a 0:00 timestamp even though grab samples were never collected at midnight. The autosampler occasionally collected a sample with a legitimate 0:00 timestamp. Otherwise, 0:00 timestamps should be disregarded and treated as if the time is unknown.
Sample ID numbers generally are five-digit integers, six-digit integers, or seven-digit values with decimal places. Samples IDs are not necessarily consecutive because water from other sites at the MEF are interspersed in the numbering series. Since samples were sometimes collected for multiple research projects, several grab samples may have been collected on the same day.
In general, the long-term grab sampling of the S2 LAGG POOL and the S2 WEIR by the Forest Service was every two weeks and samples were in five-digit or six-digit ID series. From 1986 to 2018, a new ID series was started each year and these IDs were used for samples from S2 and other sites across the MEF. For example, sample IDs 63000 to 63310 were assigned during 1986 and IDs 72000 to 72236 were assigned during 1987. After 1988, the first value in a series was usually 1000 plus the first ID value of the previous year. That is, samples 160000 to 160273 were sequentially collected during 1988, and the first sample of 1989 was assigned the ID 161000. Incrementing by 1000 continued until 2003 (i.e., series 160 to 175). During 2004, the series jumped to 213 (i.e., first sample 213000 to last sample 213308) and continued from there (series 213 during 2004 to series 227 during 2018). When weekly and automated samples were added during 2008, a separate series (346) was started. During 2009, samples were added in the 347 series (347000 to 347395). Since 2010, samples have been consecutively numbered from 348000 without starting a new series at the beginning of each calendar year. By 2018, these consecutively numbered samples had reached into the 354 series. During 2018, all sample IDs were merged into the consecutively numbered 354 series to avoid redundant weekly sampling and IDs no longer advance to a new series each year.
For long-term sampling at the S2 LAGG POOL, stream stage at the weir is usually (but not always) recorded prior to or after collection. These measurements are usually read from a manual point gage that is made within about 5 minutes of the recorded sample collection time. Ice cover on the weir pool precludes point gage measurements. When ice-covered, a measurement of water height in the v-notch may have been made using a ruler. When both S2 LAGG POOL and S2 WEIR samples are collected, stream stage may be reported with either location or both.
For most grab samples, water temperature is measured after water sampling, but at about the same time. Alcohol or perhaps mercury thermometers were used exclusively before 2011 and read to the closest degree. Extech (Model 39240, Nashua, New Hampshire, USA) waterproof digital stem thermometers have oftentimes been used since 2011 and temperature is read to 1 decimal place.
LABORATORY ANALYTICAL METHODS:
Most weekly and every other week samples were analyzed for the core suite of analytes at the Forestry Sciences Laboratory in Grand Rapids, Minnesota. From 1992 to 1994, samples were analyzed at the Research Analytical Laboratory at the University of Minnesota (St. Paul, Minnesota). From 1994 to 1997, the core suite of analyses were incrementally shifted back to the Grand Rapids laboratory.
Other laboratories have been used for analytes in addition to the core suite on weekly and biweekly samples, and sometimes on samples collected at higher than biweekly frequency.
Forestry Sciences Laboratory in Grand Rapids, Minnesota:
For each type of laboratory measurement (except water isotopes), every tenth to twentieth sample is run in duplicate (analytical duplicate) followed by two references. References are chosen to be within the range of calibration standards and optimized for particular solutes or suites of solutes to be within the range typically observed for waters at the MEF. Some references and reagents were made in-house from stock solutions or anhydrous reagents. For anion, cation, silicon, nutrient, and TOC analyses, standard solutions were made in volumetric flasks with deionized water (18.0 megaohm/cm). All glassware was acid-washed in a 10 to 15 percent hydrochloric acid, or washed in an automated labware washer (currently a Miele Professional G 7883 CD and previously a VWR Symphony unit prior to 2011) and rinsed with deionized water before drying. Over time, standard solutions have been transitioned from in-house preparation to purchase of commercially prepared solutions. The vendors and concentrations have changed over time and that information is maintained in unpublished laboratory records for each batch of samples. For each instrument and sample, we record the date of analysis and that information is stored in our unpublished laboratory records. Time of analysis is recorded when an instrument is automated via computer and software operation. These protocols have been in place since about 2000, with similar to identical procedures expected before that. The location of some records before 1992 are not currently known.
Several criteria need to be met or a sample is reanalyzed. For concentration values, analytical duplicates and the preceding samples are acceptable for reporting when the relative error is less than 5 percent between duplicates. When certified references differ by more than 10 percent from actual values, a batch of samples is reanalyzed. When a particular sample is higher in concentration than the highest calibration standard, that sample is diluted and re-run until within the range of the calibration standards.
While the methods and quality assessment/quality control procedures (QA/QC) are largely consistent over time in the Forestry Sciences Laboratory, instruments have been updated several times and the method detection limits have changed. New analytes have been added over time. When new instruments were acquired, there was oftentimes little effort to directly compare samples run on both old and new instruments, and sometimes failure of a critical instrument component hindered further analysis prior to the acquisition of a replacement instrument. Nonetheless, certified references spanned the transition periods and provided evidence of consistent results with a previous instrument. Whenever possible, transitions were timed to occur so that an entire year of samples was analyzed by one instrument. When transitions were known to occur sometime during the calendar year, that information is provided.
Information on instruments and methods is more complete and detailed since 2000, largely due to limited staff changes during this recent period. Before 2000, we sometimes have limited info due to the passage of time, turnover of staff, and missing details in the documentation that is available.
Instruments are operated in accordance with Standard Methods (APHA 2017). However, holding times of samples oftentimes do not meet those standards (as described below).
During partial US Federal Government shutdowns lasting more than several days, field sampling was maintained, but laboratory analyses were interrupted until Government operation resumed. During these periods, holding times may have been prolonged. The three government shutdowns that affect laboratory operation occurred from December 6, 1995 to January 1, 1996; October 1 to October 17, 2013; and December 22, 2018 to January 25, 2019.
Typically, refrigerated samples are discarded within two to three years of analysis completion due to limited storage space. For the foreseeable future, there is plenty of space to store frozen aliquots if any sample is left. We have not discarded frozen aliquots and may be able to retain those samples for a considerably longer period.
Specific conductivity:
Conductivity has been measured in a laboratory since 1987. Samples were warmed to room temperature and an aliquot was poured from a 250-mL sample bottle at the time of analysis
Conductivity was measured on a Yellow Springs Instruments (YSI; Yellow Spring, Ohio, USA) Model 35 meter for samples collected from 1987 to 1991 and 1994 to May 14, 2010. The conductivity probe had a 1.0/cm cell constant, but the model is unknown. On each day of operation after verifying instrument response with a standard, blank, and a reference, samples were measured. After about every 15 to 20 samples or at the end of a smaller batch, a reference was measured. An analytical duplicate was measured after about every 10 to 20 samples, or at the end of a smaller batch.
Specific conductivity was measured in the University of Minnesota Research Analytical Laboratory (St. Paul, Minnesota, USA) for samples collected from 1992 to 1994. We have no other metadata, but we do have the original paper reporting documents.
For samples collected from May 17, 2010 to 2019, conductivity was measured on a YSI Model 3100 meter. A YSI 3403 probe (cell constant = 1.0/cm) was used from samples collected until March 2017 and a YSI 3253 probe (cell constant = 1.0/cm) thereafter. After verifying instrument response with a blank and a reference, samples were measured. After about every 10 samples, an analytical duplicate and reference was measured.
For both YSI instruments, the manually loaded cell of the conductivity probe was twice rinsed with sample water and then conductivity was measured on the third poured aliquot (1 cubic cm). Conductivity values were recorded on paper. More than 100 samples could be measured each day, but smaller batches of samples were oftentimes measured within several days of collection. Specific conductivity (conductivity at 25 degree Celsius) was calculated from conductivity measured at 21 degrees Celsius when values were transferred to spreadsheets.
For samples collected during and after 2020, conductivity is measured with a Mettler Toledo Inlab 710 conductivity probe connected to a T7 Titration Excellence titrator with LabX 2019 Software (v. 10.0.0). The Inlab 710 is a 4 platinum poles conductivity cell with a chemical resistant glass body and integrated probe. On each day of operation, the millivolt meter output versus conductivity relationship is calibrated with a 46.7 microSiemen/cm standard, and periodically checked with 23.8, 84.0, or 150 microSiemen/cm references. Eighty mL of sample is poured into a sample beaker (polypropylene; pre-rinsed with deionized water) in an InMotion Pro Autosampler. With this instrument, specific conductivity, pH, and acid neutralizing capacity (ANC; not reported) are sequentially measured from the same aliquot of water. For specific conductivity measurement, the probe is dipped into a sample beaker in the autosampler rack before water is withdrawn for pH measurement. Samples, the standard, and references are measured in a laboratory maintained at 21 degrees C. Specific conductivity is calculated and reported by the T7 instrument. The conductivity probe is rinsed with deionized water between samples.
Samples typically are analyzed within days of collection. Although uncommon, samples sometimes are held for weeks to several months while awaiting maintenance on the meter, for a replacement probe, or full instrument replacement.
pH:
Since 1986, pH has been measured on an aliquot that was poured from a 250-mL sample bottle at the time of analysis. Samples were warmed to room temperature before transferring a sample to a sampler beaker for pH measurement.
At the Forestry Sciences Laboratory in Grand Rapids, autotitrators have been used to measure pH according to Standard Method 4500-H+ B (APHA 2017). Samples were only analyzed if reference values were accurate to within 10 percent and pH is reported to the nearest tenth decimal place. Sample pH is the initial pH prior to acid titering for ANC.
Sample beakers (polypropylene) are pre-rinsed before filling with sample and rinsed (and dried) before reuse with deionized water. When samples were pipetted (DL20 and DL53 titrators), a sample first was warmed to room temperature and a new disposable pipette tip was then used to transfer 50 mL of each sample to a sample beaker.
Samples, standards, and references are measured at room temperature in a laboratory maintained at 21 degree C. Commercial buffer solutions are used for pH calibration on each day of instrument operation.
A Mettler Instruments (Hightstown, New Jersey, USA) DL20 CompactTitrator with a ST20 Sample Changer, DV401 1-mL burette, ME-23955 Data Interface, and DL20 Controller software was used for samples collected from 1986 to 1991 and 1995 to 2001. A two-point calibration (pH 4.00 and 7.00 buffer solutions) was performed on each day of operation. A 40-mL aliquot of each sample was manually pipetted into a sample beaker in the autosampler, which held up to 14 unique samples. Up to three batches of samples were loaded and analyzed each day.
For samples collected from 1992 to 1994, pH was measured in the University of Minnesota Research Analytical Laboratory. We have no other metadata on these analyses, but we do have the original paper reporting documents.
A Mettler Toledo GmbH Analytical (Columbus, Ohio, USA) DL53 Autotitrator with ROND060 autosampler, 10 mL burette, and LABX PC titration software was used for samples collected from 2001 to 2019. A four-point (pH = 4.0, 6.0, 7.0, and 10.0) calibration was performed on each day of operation. A 50-mL aliquot of each sample was manually pipetted into a sample beaker in the autosampler rack, which held 15 sample beakers. Each sample batch included 10 samples, a duplicate sample, and four more samples for 14 total unique samples per batch. The pH probe was automatically rinsed with deionized water between each sample. Up to three batches of samples were loaded and analyzed each day.
A Mettler Toledo T7 Titration Excellence titrator with an InMotion Pro Autosampler, DGi11-SC combined glass pH electrode, DV1010 interchangeable 10-mL burette, and LabX 2019 Software (v. 10.0.0) is used for samples collected during and after 2020. An 80-mL aliquot of a sample is poured into a sample beaker (pre-rinsed with deionized water) that is placed into an InMotion Pro Autosampler that holds 69 beakers (standards, samples and reference solutions). A four-point calibration (pH = 4.0, 6.0, 7.0, and 10.0) is performed each day of operation. Between samples, tubing and the flow-through cell are automatically rinsed with deionized water and then 30 mL of deionized water followed by 20 mL of sample to rinse tubing. The titrator then meters 50-mL of a sample into the cell for initial pH measurement and then titering of acid for measurement of ANC (not reported). Throughput per batch (50 samples plus five duplicates) is higher than the previous DL20 and DL53 titrators, and checks are run after every 10 samples.
Samples for pH analysis typically were analyzed within days of collection. Although rare, samples sometimes were held for weeks to several months while awaiting maintenance on the meter, a replacement pH probe, or full instrument replacement.
Anions:
An aliquot was poured from a 250-mL sample bottle at the time of analysis.
Anion (chloride and sulfate) concentrations are measured using electronic suppression of carbonate-bicarbonate eluent conductivity and conductimetric detection on ion chromatographs. The methods are consistent with Standard Method 4110-C (APHA 2017).
At the Forestry Sciences Laboratory in Grand Rapids, standards were made in-house from separate chloride and sulfate certified stock solutions for samples collected through 2020. The instruments were calibrated with a deionized water blank and five analytical standards. Deionized water blanks and commercial, certified references are analyzed with each batch of samples. Samples are poured from refrigerated 250-mL LDPE sample bottles into 5-mL PolyVials (new, not rinsed) that fit autosampler racks. Four different, successive instruments were manufactured by Dionex or Thermo Scientific Dionex (both located in Sunnyvale, California, USA). Samples are injected through 20 micrometer filter caps (Dionex 038009 or equivalent product from another vendor) into 5-mL PolyVials (Dionex 038008 or equivalent product).
A Dionex 2000i/SP Ion Chromatograph with an ASM-2 autosampler, AMMS Micromembrane Suppressor, a Spectra Physics integrator (model unknown), and automation interface (unknown manufacturer and model) was used for samples collected from 1986 to 1991 and 1995 to July 1999. Dionex IonPac AG1 guard columns and AS4A columns were used for isocratic separation of anions. The detection limits are 0.1 mg chlorine/L and 0.14 mg sulfate/L.
Anion concentrations were measured at the University of Minnesota Research Analytical Laboratory for samples collected from 1992 to 1994. We have little metadata, but samples were analyzed on an ion chromatograph. Samples were shipped several times each year. The detection limits are 0.2 mg chlorine/L and 0.3 mg sulfate/L. We have original paper reporting documents.
A Dionex DX-500 with an AS40 Autosampler, LC10-2 load module, CD20-1 conductivity detector, IP25 isocratic pump, ASRS Ultra II 4 mm Self-Regenerating Suppressor, and PeakNet Software was used for samples collected from August 1999 to 2011. IonPac AG14 pre-columns and AS14 columns (DX500) were used for isocratic separation of anions. The detection limits are 0.01 mg chlorine/L and 0.02 mg sulfate/L.
A Thermo Scientific Dionex ICS-2100 with an AS-DV autosampler and Chromeleon Chromatography Data System Software was used for samples collected from 2012 to 2019. IonPac AG22 pre-columns and AS22 columns were used for isocratic separation of anions. The detection limits are 0.01 mg chlorine/L and 0.02 mg sulfate/L.
A Thermo Scientific Dionex Aquion with an AS-DV autosampler and Chromeleon Chromatography Data System Software is used for samples collected during and after 2020. IonPac AG22 pre-columns and AS22 columns are used for isocratic separation of anions. The detection limits are 0.01 mg chlorine/L and 0.02 mg sulfate/L.
In the past 10 or so years, an ion chromatograph was typically operated every business day, but analyses lagged as samples were collected faster than samples could be analyzed. Sometimes instrument failure led to exceptionally long hold times before anion analysis. For that reason, samples for anion measurement were sometimes analyzed within several days of collection, but sometimes held for months to >1 year before analysis. For samples collected from 1995 to 2009, analysis was completed within 200 days of collection. Samples collected from June to December 2010 were held from 200 to 348 days prior to analysis. Most samples collected from August 2012 to November 2016 were held from 200 to 530 days before analysis. Most samples collected from January 2017 to November 2020 were held from 200 to 850 days before analysis. Due to the currently higher throughput on the Aquion instrument, hold times have decreased to only several weeks after an initial backlog of samples was analyzed.
Concentrations of nitrate were analyzed by ion chromatography on samples collected since 1992. However, samples for nitrate analyses were refrigerated from 1992 to 2019, not frozen. Because we now know that nitrate concentrations were unstable during refrigerated storage, we do not report nitrate concentration values. Some of the refrigerated samples may have been analyzed with short enough holding times to provide environmentally relevant and valid concentrations. However, at this time, we are not reporting any nitrate values from ion chromatography. Past assessments of MEF data should be considered with caution unless short-holding times, freezing, or validation were specified in any particular publication. We no longer determine nitrate concentrations using ion chromatography because the detection limit of 0.05 mg/L is much greater than the 0.002 mg/L detection limit of nitrate+nitrite measured using flow injection analysis.
Nitrite was not separately measured using ion chromatography (or flow injection analysis) and nitrite standards have not been included in calibration standards. Nonetheless, nitrite in appreciable amounts would appear in sample chromatograms and it has been rare in our experience to observe nitrite in MEF water samples, even when holding times are reasonable.
Though not currently included, fluoride and bromide have been included in analytical standards in the past. Most samples did not have detectable (<0.05 mg/L) fluoride or bromide concentrations. Therefore, we have not continued to include fluoride or bromide in standards, and we rarely observe fluoride or bromide peaks in chromatograms.
Cations and silicon:
Cation concentrations have been analyzed by atomic absorption spectroscopy (Standard Method 3111; APHA 2017) or inductively coupled plasma optical emission spectroscopy (ICP-OES; Standard method 3120; APHA 2017). Concentrations of calcium, magnesium, potassium, and sodium have always been measured. Iron was added during 1992. Concentrations of aluminum and manganese were measured for samples collected from 1992 to 1996 and onwards from 2009. Concentrations of silicon and strontium were added during 2009. Sampling and concentration measurement for mercury and lead are separately described.
At the Forestry Sciences Laboratory, unfiltered water is poured from a 250-mL LDPE sample bottle into a borosilicate or polypropylene test tube (new, not rinsed) that is placed in an autosampler rack. Samples were not acidified prior to or for analysis. Samples are refrigerated until analyzed. The instruments were calibrated with a deionized water blank and at different concentrations of each solute across five analytical standards. Analytical duplicates were measured every 10 (in the last decade or so) to 15 (until sometime during the early 2000s) samples.
For samples collected from 1986 to 1991 and after July 26, 1995, samples have been analyzed at the Forestry Sciences Laboratory in Grand Rapids. Standards were made in-house from individual cation certified stock solutions for samples collected through 2020. Commercial, certified references are analyzed with each batch of samples.
A Perkin-Elmer (Norwalk, Connecticut, USA) Model 5000 Atomic Absorption Spectrometer with a Model 3600 Data Station and an auto sampler (no product info available) was used for samples collected from 1986 to 1991 and from July 26, 1995 to 2002. Samples were analyzed for calcium, magnesium, potassium, and sodium concentrations. For samples collected from 1995 to 2002, iron concentration was also measured. Detection limits are 0.022 mg calcium/L, 0.009 mg magnesium/L, 0.006 mg potassium/L, 0.003 mg sodium/L, and 0.03 mg iron/L,
Cation concentrations were measured at the University of Minnesota Research Analytical Laboratory for samples collected from 1992 to 1995. We have little metadata, but we do have the original paper reporting documents. An Applied Research Laboratories (Bausch and Lomb, Sunland, California) 3560 Simultaneous ICP was used. Samples were acidified to 1 percent nitric acid for analysis. Samples were analyzed for calcium, magnesium, potassium, sodium, aluminum, iron, and manganese concentrations. Detection limits are 0.05 mg calcium/L, 0.2 mg magnesium/L, 0.8 mg potassium/L, 0.2 mg sodium/L, 0.2 mg aluminum/L, 0.03 mg iron/L, and 0.003 mg manganese/L. Samples were shipped several times each year. Holding times were two weeks to four months before analysis.
A Thermo Electron Corporation (Waltham, Massachusetts, USA) Iris Intrepid ICP-OES with a Thermo Elemental Timberline IIL Autosampler, iTEVA iCAP software, and radial view of plasma was used for all samples collected from 2003 to 2015. Samples were analyzed for calcium, magnesium, potassium, sodium, and iron. Concentrations of silicon and strontium were measured for samples collected starting 2009. Measurements of aluminum and manganese concentrations resumed for samples collected that year and onward. Detection limits are 0.05 mg calcium/L, 0.05 mg magnesium/L, 0.5 mg potassium/L, 0.1 mg sodium/L, 0.01 mg aluminum/L, 0.05 mg iron/L, 0.01 mg manganese/L, 0.05 mg silicon/L, and 0.01 mg strontium/L.
A Thermo Scientific (Waltham, Massachusetts) ICAP 7600 Duo with an ASX-520 Autosampler and Qtegra Intelligent Scientific Data Solution Software is used for samples collected during and after 2016. Analytes are the same as the Iris Intrepid instrument. This instrument has the option for radial or axial view of plasma. For samples collected from 2016 to 2020, solutes were measured using only radial view of plasma. In axial view, detection limits are 0.05 mg calcium/L, 0.05 mg magnesium/L, 0.5 mg potassium/L, 0.1 mg sodium/L, 0.01 mg aluminum/L, 0.05 mg iron/L, 0.05 mg silicon/L, and 0.01 mg strontium/L.
Regardless of laboratory, samples were typically run one to four times a year in large batches of samples. Analysis occurred within several days of collection for some samples to more than a year from collection for those that were held longest. For samples collected from 1991 to 2007, the hold time was about 200 days or less, with most being <100 days. Some samples collected during 2008 and 2009 were held for up to 1 yr. Some samples collected during 2010, 2011, 2013, and 2016 were held for up to 528 days. Since then, samples were held for about 300 or fewer days before analysis, except one sample from 2017 (661 day holding period, likely due to reanalysis after becoming aware of a failed QA/QC check).
Nutrients:
Prior to 2019, an unfiltered sample was collected in a 250-mL LDPE sample bottle after rinsing 3x with surface water. Samples were refrigerated until analyzed.
Since 2019, unfiltered water has been collected in a 60-mL high-density polyethylene bottle (HDPE) after 3x rinsing. Samples are frozen until thawed for each successive nutrient analysis. Samples were refrozen for weeks to months between sequential analyses of the various nutrients.
In the Forestry Sciences Laboratory, a sample from either bottle type is poured into a borosilicate (new, unrinsed) test tube that is placed into an autosampler rack. Samples for nitrogen and phosphorus chemistry are typically run once or twice a year in large batches of samples. Analysis occurs within several days of collection for some samples to a year from collection for those that are held longest. An analytical duplicate was analyzed after every 10 samples.
Commercial, certified references are analyzed with each batch of samples.
Ammonium and nitrate+nitrite:
Concentrations of ammonium and nitrate+nitrite have been measured on samples collected since 1986. However, samples collected from 1986 to 2018 for ammonium and nitrate+nitrite analyses were refrigerated, not frozen. Because we now know that ammonium and nitrate+nitrite concentrations were not stable during long periods of refrigerated storage, we only report ammonium and nitrate+nitrite methods and concentration values for frozen samples. Some of the refrigerated samples may have been analyzed with short enough holding times to provide environmentally relevant and valid concentrations. However, at this time, we are not reporting any pre-2019 values. Past assessments of S2 surface water based on pre-2019 values should be considered with caution unless short-holding times, freezing, or validation were specified in any particular publication.
Ammonium and nitrate+nitrite concentrations are measured on the same instrument: a QuickChem (Hach Company, Loveland, Colorado) 8500 Flow Injection Analysis System with ASX-260 Series XYZ AutoSampler, Ismatec SM933 IPC High Precision Multichannel Pump, A85132 Heater Apparatus including 175CM and 650CM Inserts, and Omnion software. Throughput is up to 180 samples per day.
Ammonium is measured according to the Lachat QuikChem 10-107-06-1-F method. The Lachat methods are equivalent to the flow injection analysis method to form indephenol blue for colorimetric analysis (Standard Method 4500-NH3 H; APHA 2017). Ammonium is reported as the amount of nitrogen in ammonium. The detection limit is 0.01 mg nitrogen/L for ammonium. Frozen samples were held for 370 or fewer days before analysis.
Nitrate+nitrite is measured according to Lachat QuikChem 10-107-04-1-B. Nitrate is reduced to nitrite using the flow injection analysis method with cadmium reduction, sulfanilamide diazotization coupled to N-(1-naphthyl)-ethylenediamine dihydrochloride to produce colored azo dye, and colorimetric detection as the amount of nitrogen (Standard Method 4500-NO3- I; APHA 2017). The detection limit is 0.002 mg nitrogen/L for nitrate+nitrite. Frozen samples were held for 398 or fewer days before analysis.
Total nitrogen, total phosphorus, and soluble reactive phosphorus:
Concentrations of these three solutes were usually measured using flow injection analysis (FIA). Analysis of total nitrogen started during 1997. Prior to that concentration of total Kjeldahl nitrogen (TKN; not reported) was measured. Total phosphorus concentration has been measured since 1986. Though sometimes measured before 1992, we report soluble reactive phosphorus concentration starting 1992. Aliquots for measurement of total nitrogen, total phosphorus, and soluble reactive phosphorus were poured from a refrigerated 250-mL LDPE sample bottle prior to 2019 and poured from 60-mL HDPE bottles with frozen storage from then onward.
Metadata are incomplete, but a Technicon AutoAnalyzer II with a Sampler IV autosampler was used for total phosphorus concentration measurement from 1986 to 1991. This instrument was at the Forestry Sciences Laboratory in Grand Rapids. Though details are incomplete, the Technicon Autonalyzer II system included a data handler, integrator, recorder, autovalve, pumps, cartridges, and flow cells. Throughput was 50 samples per day and the detection limit is 0.01 mg phosphorus/L.
At the University of Minnesota Research Analytical Laboratory, total phosphorus concentration was measured for samples collected from 1992 to July 1997 and soluble reactive phosphorus (ortho-phosphate) concentration was measured for samples collected from 1992 to July 1997. Total phosphorus was measured by ICP (same ARL 3560 ICP that is listed for cation measurement) for samples collected from 1992 to July 20, 1995. Soluble reactive phosphorus was measured by flow injection analysis, as was total phosphorus for samples collected on and after July 25, 1995. The total phosphorus detection limits are 0.04 mg phosphorus/L for samples collected from 1992 to July 20, 1995 (when ICP was used) and 0.02 mg phosphorus/L for samples collected from July 25, 1995 to July 1997. Otherwise, we have little metadata, but we do have the original paper reporting documents. The soluble reactive phosphorus detection limits are 0.2 mg phosphorus/L for samples collected during March and April 1992 and 0.01 mg phosphorus/L from May 1992 to July 1997.
In the Forestry Sciences Laboratory for samples collected during and after 1997, concentrations of the three analytes were measured using:
* A Lachat (Milwaukee, Wisconsin) QuickChem 8000 with RAS Sampler, A82000 12 Channel Pump, and A85100 Heating Unit, and Omnion FIA Software (RAS) for samples collected from September 1997 to 2014.
* A QuickChem 8500 Flow Injection Analysis System with ASX-280 Series XYZ AutoSampler, Lachat RP-150 Series Reagent Pump, A30111 Lange Lachat QuikChem In-Line Module with heater block and ultraviolet lamp (total nitrogen), A30000 Lachat In-Line Sample Prep Module with heater block and ultraviolet lamp (total phosphorus), and Omnion software for samples collected during and after 2015.
Total nitrogen (TN) concentration is measured colorimetrically after in-line automated persulfate oxidation catalyzed by ultraviolet irradiation to nitrate (Standard Method 4500-N B; APHA 2017). Concentrations were measured according to the Lachat QuikChem 10-107-04-1-P method for samples collected before 2016 and Lachat QuikChem E10-107-04-3-D method thereafter. The Lange Lachat QuikChem In-Line Module with heater block and ultraviolet lamp is used in conversion of organic nitrogen to inorganic nitrogen. The detection limit is 0.05 mg nitrogen/L. Throughput is up to 120 samples per day and samples were oftentimes held for 200 to 570 days before analysis.
Total phosphorus (TP) concentration is measured using automated persulfate oxidation catalyzed by ultraviolet irradiation and flow injection analysis with ascorbic acid reduction for colorimetric detection (Standard Method 4500-PI; APHA 2017). Concentrations were measured according to the Lachat QuikChem 10-115-01-3-A method for samples collected from September 1997 to 2015. Lachat QuikChem E10-115-01-3-A method was used thereafter. The Lachat In-Line Sample Prep Module with heater block and ultraviolet lamp is used to convert organic phosphorus to inorganic phosphorus. The detection limit is 0.02 mg phosphorus/L for samples collected from July 1997 to July 10, 2003 and is 0.05 mg phosphorus/L for samples collected since then. Throughput is up to 120 samples per day and samples were oftentimes held for 200 to 570 days before analysis.
Soluble reactive phosphorus concentration (referred to as ortho-phosphate in all laboratory records) was measured according to the Lachat QuikChem 10-115-01-1-B method for samples collected during and after September 1997. The Lachat methods are equivalent to the flow injection analysis method with ascorbic acid reduction and colorimetric detection (Standard Method 4500-P F; APHA 2017). The detection limit is 0.001 mg phosphorus/L. Throughput is up to 200 samples per day and hold times were <300 days for samples collected from 2004 to 2015 and 2018. For samples collected during 2015, 2016, 2017, 2019, and 2020 had hold times mostly between 200 and 568 days before analysis.
Total organic carbon (TOC):
Total organic carbon concentration was first measured during 1992.
Concentration of TOC was measured at the University of Minnesota Research Analytical Laboratory for samples collected from 1992 to October 1995. We have little metadata other than the method detection limit (0.5 mg carbon/L) on these analyses. We have original paper reporting documents.
After that 1992 to 1995 period, all samples have been analyzed at the Forestry Sciences Laboratory in Grand Rapids. Unfiltered sample water was poured from a 250-mL LDPE sample bottle into an acid-washed 20-mL or 40-mL borosilicate glass vial after rinsing 3x with sample water. Samples were refrigerated until analyzed.
Concentration of TOC has been measured by sodium persulfate wet oxidation catalyzed by:
* Platinum with high-temperature combustion (a Dohrmann instrument) and non-dispersive infrared (NDIR) detection (Standard Method 5310 B, APHA 2017).
* High-temperature combustion with ultraviolet irradiation (two Shimadzu instruments) and NDIR detection (Standard Method 5310 C, APHA 2017).
A Tekmar Dohrmann (Rosemount Analytical, Santa Clara, California) DC-190 High-Temperature TOC Analyzer with autosampler was used for samples collected from November 1995 to 2004. Concentrations were measured as total carbon minus inorganic carbon (TOC by difference, TC-IC). The instrument was calibrated using a deionized water blank and a 100 mg carbon/L sucrose standard (prepared in-house). Every tenth sample was followed by an analytical duplicate and every twentieth sample was followed by a 20-mg carbon/L sucrose reference.
A Shimadzu Corporation (Columbia, Maryland, USA) TOC-V CPH with External Sparge Kit, an ASI-V Auto-Sampler, and TOC Control-V Software was used for samples collected from 2005 to 2012. Concentration was measured as TC-IC for samples collected before June 2010 in the 219 and 348 series, and before January 2011 for the 311 series. Concentration was measured as non-purgeable organic carbon (NPOC) for samples collected after those specified periods. Sebestyen et al. (2020) show that the TC-IC and NPOC are comparable to within 10 percent relative error as a measure of TOC concentration (linear regression, p << 0.001, Pearson correlation coefficient = 0.98, n = 62 including other MEF stream samples, and plotting along a 1:1 relationship). We consider the TC-IC and NPOC methods to be equivalent for S2 surface waters.
A Shimadzu TOC-L with ASI-L Auto-Sampler and TOC Control-L Software is used for samples collected from 2013 onward. Concentration is measured as NPOC on all samples. To continue comparison of the NPOC and TC-IC results, TOC concentration was measured as TC-IC on a subset of samples during 2012, 2016, 2017, 2018, and 2019 for both the S2 LAGG POOL and S2 WEIR locations.
For both Shimadzu instruments, samples were poured from 250-mL LDPE bottles into 40-mL borosilicate glass vials with a cap (polytetrafluoroethylene, PTFE lined silicone septa) that fit an autosampler carousel that held 68 vials or into 20-mL vials that fit an autosampler carousel that held 94 vials. The instruments were calibrated with deionized water and, since about 2015, a certified 100 mg carbon/L potassium hydrogen phthalate (KHP) analytical standard. Before that, a 100 mg carbon/L KHP analytical standard was prepared in-house. The auto dilution feature of the instruments was used to calibrate with three different concentrations diluted from the 100 mg carbon/L KHP standard. Every tenth sample was followed by an analytical duplicate, a deionized water blank, a sucrose reference, and perhaps an additional certified reference. The sucrose reference rotated among 2, 5, 10, 25, 50, or 100 mg carbon per L within a batch of samples.
For the Dohrmann and Shimadzu instruments, a certified reference was analyzed at the beginning and end of a daily batch. Results were acceptable when references were within 10 percent of the actual value. Sucrose references were prepared in-house. The detection limit is 1 mg carbon/L for TOC measured as TC-IC and 0.5 mg carbon/L for TOC measured as NPOC, regardless of instrument.
Samples collected from November 1995 to October 1999 were held from several days to 170 days, except three samples from October and November 1996 were held until January 1998 (447 to 475 days) before analysis. Samples collected from October 1999 to June 2005 generally had longer hold times (100 to 426 days) before analysis. Hold times are below 136 days for most samples collected since October 2005, except samples from July 2010 to November 2010 when hold times ranged between 166 and 272 days. The longest hold times are likely associated with reanalysis after failed QA/QC or an instrument awaiting maintenance or full replacement.
Dissolved Organic Carbon (DOC):
Samples for DOC analysis are not routinely collected. When DOC concentration was measured, there was oftentimes a corresponding sample for TOC concentration measurement at the Forestry Sciences Laboratory.
Concentration of DOC has been measured by sodium persulfate wet oxidation catalyzed by:
* Ultraviolet irradiation (a Dohrmann and Tekmar-Dorhmann instrument) and NDIR detection (Standard Method 5310 B, APHA 2017).
* Platinum with high-temperature combustion (an Oceanography International Analytical instrument) and NDIR detection (Standard Method 5310 B, APHA 2017).
* High-temperature combustion with ultraviolet irradiation (several Shimadzu instruments) and NDIR detection (Standard Method 5310 C, APHA 2017).
Additional aliquots of the biweekly samples of the S2 LAGG POOL were collected from 1993 to 1998 as part of mercury source and transport studies (Kolka et al. 1992, 2001; Fleck 1998; Grigal et al. 2000). Samples were collected in 250-mL polyethylene sample bottles and labelled in the 165 to 170 series that were also analyzed for the core suite of analytes. Samples were transported by overnight courier to the University of Minnesota (St. Paul). Samples were frozen upon arrival at the University of Minnesota Carbon Laboratory. After thawing and before analysis, samples were filtered through 0.7 micrometer glass fiber filters. The filters were precombusted to remove that potential source of carbon contamination. A Dohrmann DC-80 Total Organic C Analyzer was used for analysis. It is not known if the concentration was measured as TC-IC or NPOC. Comparisons of samples from other research sites have shown that DOC concentrations can significantly decrease after sample freezing (Fellman et al. 2008). For that reason, and there being greater differences between TOC and DOC concentrations during the period when samples were frozen than any subsequent periods when samples were refrigerated, we caution on the use of the 1993 to 1998 DOC concentration values. We provide those values in this data release because they have been used in past publications and we have not verified the effect of freezing on DOC concentration for S2 samples. The detection limit is 1 mg carbon/L.
From 2005 to 2008, samples were collected at the S2 WEIR as part of the sulfate addition study (see a following section on mercury sampling and analysis). Samples were labelled with a six-digit serial identifier that started with 296 during 2005, 295 during 2006, 307 during 2007, and 308 and 309 during 2008. One sample from 2005 was labelled 2005.001. For samples from 2005, no time was recorded when a sample was collected. Samples were pumped from the weir pool using a Geopump Peristaltic Pump (Geotech Environmental Equipment, Denver, Colorado, USA) with 0.64 cm internal-diameter Teflon tubing. Stream water was pumped through the entire length of tubing as a rinse. Then, a Perfluoroalkoxy alkane (PFA) filter holder (Savillex 401-31-47-10-31-2 Single Stage Filter Assembly, Eden Prairie, Minnesota) with a 47-mm diameter 0.7-micrometer glass fiber filter (GF/F) was attached to the tubing and rinsed for 3 seconds. A new 250-mL LDPE sample bottle was then filled after 3x rinsing with filtrate. Prior to filtration, filters were combusted for 4 hours at 500 degrees Celsius. These samples were transported to the St. Croix Watershed Research Station – Science Museum of Minnesota in Marine on St. Croix, Minnesota within 48 hr of collection. For samples collected during 2006, an aliquot was poured from a sample bottle into a borosilicate vial to measure DOC concentration as NPOC on a Tekmar-Dohrmann (Cincinnati, Ohio, USA) Phoenix 8000 Carbon Analyzer with autosampler. Information of standards, references, holding times, and other operating procedures are incomplete. The detection limit is assumed to be 0.5 mg carbon/L. For samples collected during 2007 and 2008, the samples were then sent to Gustavus Adolphus College (St. Peter, Minnesota). An aliquot was poured from a bottle into a borosilicate vial to measure DOC concentration as NPOC on a Shimadzu TOC-V with ASI-V autosampler. The auto dilution feature of the TOC analyzer was used to calibrate using three different concentrations from a KHP analytical standard. A sucrose and KHP reference were run after every 10 samples. Samples may have been held for up to 3 months after acidification. The detection limit is 0.5 mg carbon/L.
Aliquots of many samples from the S2 WEIR in the 348 to 351 series were collected from 2009 to 2013 as part of a study of DOM sources, transport, and biodegradability (Sebestyen et al. 2021a). Samples were collected in 1-L acid-washed HDPE bottles or new 250-mL LDPE bottles. Either bottle type was first 3x rinsed with sample water before being filled. Samples were chilled on ice, transported, and received one day after sampling at the Aquatic Ecology Laboratory at the University of Minnesota, St. Paul. Several samples were filtered at the Marcell Research Center within several hours of sampling. However, most samples were filtered after receipt in the laboratory through Whatman (now Cytiva/Global Life Sciences, Marlborough, Massachusetts, USA) GF/F 0.7 micrometer glass fiber filters (precombusted) using vacuum filtration. Filters were pre-rinsed with sample water before filtration. Filtered samples were acidified to pH = 2 and stored in pre-combusted and pre-rinsed borosilicate glass vials until analysis. A Shimadzu TOC-V CPH with External Sparge Kit and ASI-V Auto-Sampler was used and concentration was measured as NPOC. Potassium hydrogen phthalate was used as an analytical standard. In addition to the research described in Jacobson (2012) and Sebestyen et al. (2021a), correlation between DOC concentration measured at the Aquatic Ecology Laboratory and TOC concentration measured at the Forestry Sciences Laboratory were used to provide evidence (linear regression, p << 0.0001, Pearson correlation coefficient = 0.87, and n = 53, and plotting along a 1:1 relationship) that TOC and DOC concentrations are equivalent (Sebestyen et al. 2020).
Some samples were collected at the S2 WEIR during 2010 as part of a study of mercury transport (unpublished research). An aliquot of each weekly sample (399 ID series with a corresponding timestamp in the 348 to 350 series for the core suite of analytes) was collected in an acid-washed 0.5-L, 1-L, or 2-L Teflon bottle. Since the samples were primarily collected for mercury analysis (described below), a Teflon bottle was filled and stored with 1 percent high-purity hydrochloric acid prior to sampling and sample bottles were double bagged. Bottles were drained in the field and then rinsed three times with surface water before filling. Samples were shipped overnight to the University of Minnesota Mercury Laboratory. Upon receipt, the samples were filtered through 45-mm diameter Whatman 0.45 micrometer cellulose nitrate filters in disposable Nalgene (Rochester, NY) filter units and acidified with 0.5 percent trace-metal grade hydrochloric acid. Once preserved with acid, the samples were then sent to Gustavus Adolphus College. Analytical instrumentation and procedures are identical to other DOC measurements made at Gustavus Adolphus College.
Some samples were collected at the S2 WEIR during 2011, 2016, 2017, and 2018 as part of a study of lead transport (Jeremiason et al. 2018). These samples were analyzed at Gustavus Adolphus College. While these samples have an accompanying sample with an identical timestamp in the 350 series (2011), 352 or 353 series (2016), 353 series (2017) and 353 or 354 series (2018) for the core suite of analytes, all of the samples have six-digit serial identification numbers that start with 399 or 400. These samples were collected into 125-mL PETG sample bottles. Samples were shipped overnight. Upon receipt, the samples were filtered through 45-mm diameter Whatman 0.7 micrometer glass fiber filters in acid-washed Savillex teflon filter holders (401-31-47-10-31-2 Single Stage Filter Assembly) and acidified by adding 0.5 percent (volume/volume) 12 mol/L hydrochloric (HCl) acid. Filters were precombusted at 450 degrees Celsius for 4 hr. Analytical instrumentation and procedures are identical to other DOC measurements made at Gustavus Adolphus College.
Some samples from the S2 WEIR were collected during 2013, 2014, and 2015, and analyzed at the US Geological Survey Organic Carbon Laboratory in Boulder, Colorado. There were relatively few samples and the serial identifiers are 351653, 351697, 351780, 352011, 352020, 352033, 352182, 352245, 352264, 352285, 352338, 352355, 352370, 352586, and 352651. In the field, a sample was pumped with a peristaltic pump (Global Water Instrumentation, Phoenix, Arizona, USA, SP100 Water Sampler with silicon pump tubing) through Geotech Environmental Equipment, Inc 0.45 micrometer High Capacity Disposable Filter and collected in a 1-L amber glass bottle with PTFE lined cap. Tubing was rinsed and then a filter was rinsed with at least 20 to 30 mL of surface water. A bottle was 3x rinsed with filtered water prior to filling. Chilled samples were shipped in coolers for next day delivery to the Boulder laboratory. Concentration of DOC was measured using an OI Analytical (College Station, Texas, USA) Model 700 Carbon Analyzer (Aiken 1992) with an autosampler. A sample was poured into a combusted borosilicate tube that fit the autosampler rack. The instrument was calibrated with eight standards, with a 30-mg carbon/L KHP standard as the highest concentration and a blank after calibration. A reported value is the mean of two analytical replicates of each sample. Since only one sample was shipped at a time and concentration was measured within days of receipt, S2 samples were interspersed with other sets of water. Two to four blanks were analyzed between sets of different waters. An inorganic carbon, caffeine, KHP, and a lab fortified blank (US EPA 1995) were analyzed after calibration as references. A caffeine reference may also have been analyzed midway through a run. A sample that was higher in concentration than the highest calibration standard was diluted and re-run until within the range of calibration standards. The detection limit is 0.2 mg carbon/L.
Some samples from 2018 were analyzed at the Forestry Sciences Laboratory in Grand Rapids. These samples, from the S2 WEIR and S2 LAGG POOL, have serial identification numbers that start with 353 or 354. A sample was either field collected in Norm-Ject 60-mL Luer Lock Syringe (polypropylene barrel and polyethylene plunger; Henke Sass Wolf, Tuttlingen, Germany) or poured from a 250-mL LDPE sample bottle into the syringe barrel in the laboratory. For field filtration, a syringe was 3x rinsed with deionized water before a day of sampling and 3x rinsed in between samples with water that was to be collected. In the laboratory, a syringe was 3x rinsed with deionized water between samples, then 3x rinsed with sample water. Samples were filtered through Whatman (Cytiva/Global Life Sciences, Marlborough, Massachusetts, USA) Puradisc 25 GF/F 0.7 micrometer disposable (polypropylene housing) glass fiber filters. A filter was rinsed with about 5 mL or more of sample water prior to 3x rinsing of a vial with filtered water. Filtered water was then collected into a 40-mL borosilicate glass vial with a cap (PTFE lined silicone septa) that fit into a Shimadzu autosampler rack (same instrument and methods as listed above for TOC analyses on the TOC-L instrument). Concentrations of TOC were measured on the same timescale as mentioned for TOC measurements made at the Forestry Sciences Laboratory. The detection limit is 0.5 mg carbon/L.
Ultraviolet Absorbance of DOM, Multiple Wavelength-Spectrophotometry:
A subset of S2 LAGG POOL and S2 WEIR grab samples have been measured for absorbance in the ultraviolet (UV) and visible light ranges, either on whole (samples collected 2014 to ongoing) or filtered (2009-2012 and 2013-2015) waters, as a measure of water color or an optical property of DOM (Weishaar et al. 2003). Data are reported to 3 decimal places with an uncertainty of 0.002 and are unitless decadal (log10) absorbance coefficients at a specified light wavelength (254 and 360 nanometers, nm, are reported; absorbance at other wavelengths may be available upon request).
We report absorbance at these three wavelengths because:
1) Absorbance at 254 nm is commonly used to calculate specific UV absorbance, (SUVA) which is the absorbance divided by DOC concentration (Weishaar et al. 2003).
2) Absorbance at wavelengths <400 nm to be linearly proportional to color values in Platinum-Cobalt units (a reference is needed for this general assertion) and 360 nm was correlated with color. Color values, measured during the 1980s, and corresponding absorbance 360 nm on a fixed-wavelength spectrometer will be provided in a future version of this data release.
Most of the same samples from the S2 WEIR and S2 LAGG POOL in the 348 to 351 series that were sent to the University of Minnesota Aquatic Ecology Laboratory for DOC analysis were also measured for UV absorbance. Samples were collected in 1-L acid-washed HDPE bottles or collected in new 250-mL LDPE bottles. Either bottle type was first 3x rinsed with sample water before being filled. Samples were chilled on ice, transported, and received one day after sampling at the Aquatic Ecology Laboratory. Several samples were filtered at the Marcell Research Center within several hours of sampling. However, most samples were filtered after receipt in the laboratory through Whatman GF/F 0.7 micrometer glass fiber filters (precombusted) using vacuum filtration. Filters were pre-rinsed with sample water before filtration. These aliquots were refrigerated at 4 degree Celsius until measured using a Cary 50 spectrophotometer with 1-cm quartz cuvette. Absorbance was measured at 1-nanometer (nm) intervals from 200 nm to 600 nm. We report UV absorbance at 254 nm. The full spectra are not currently available to us. Samples were analyzed within days of receipt in the laboratory.
The same samples from the S2 WEIR that were sent to the US Geological Survey Organic Carbon Laboratory in Boulder were also analyzed for UV absorbance. There were relatively few samples and the serial identifiers are 351653, 351697, 351780, 352011, 352020, 352033, 352182, 352245, 352264, 352285, 352338, 352355, 352370, 352586, and 352651. A sample was pumped with a peristaltic pump (Global Water Instrumentation SP100 Water Sampler with silicon pump tubing). A sample was field-filtered through a Geotech Environmental Equipment, Inc 0.45 micrometer High Capacity Disposable Filter and collected in a 1-L amber glass bottle with a PTFE lined cap. The tubing was rinsed and then a filter was rinsed with at least 20 to 30 mL of surface water. A bottle was 3x rinsed with filtered water prior to filling. Chilled samples were shipped in coolers for next day delivery to the Boulder laboratory. Absorbance at 254 nm was measured on an Agilent Technologies (Santa Clara, California) 8453 UV-visible Spectroscopy System with photodiode array, combination deuterium discharge and tungsten lamps, and a 1-cm pathlength quartz cuvette. The full spectra are not currently available to us. Samples were analyzed within days of receipt in the laboratory.
Absorbance in both UV and visible light ranges has been measured at the Forestry Sciences Laboratory in Grand Rapids on every sample collected during and after 2014 for samples collected from the S2 LAGG POOL or S2 WEIR. Absorbance every 1 nm from 200 to 700 nm is measured on an Agilent Technologies 8453 UV-visible Spectroscopy System with photodiode array and combination deuterium discharge and tungsten lamps, Sipper System (1FS peristaltic pump), G1811A XY-Sampler, and UV-Visible ChemStation software version 8.04.02(64). A sample is poured from a 250-mL LDPE bottle into a new, unrinsed borosilicate tube. A tube is placed in an autosampler rack. About 4 mL is pumped through a flow-through 1-cm quartz cuvette (Agilent 0100-1225 flow cell UV, 3 mm round aperture, 10 mm pathlength, and 80 microliter capacity). The first 3 mL aliquot rinses the cuvette, the pump stops, and absorbance is measured on the next 80 microliter aliquot. After each scan, the cuvette is rinsed for 20 s at a 1 mL/min flow rate between samples with deionized water. Each spectra is reviewed to identify aberrant spectra due to an air bubble or particulate, or for samples that need to be diluted and reanalyzed. Those samples were rerun within several days until a realistic spectrum within the absorbance range was recorded. We report absorbance at 254 and 360 nm for samples. Absorbance at other wavelengths can be provided upon request. More than 100 samples can be analyzed each day. Since aliquots for TOC concentration and UV-absorbance measurement were almost always poured at the same time, the holding times are identical for TOC and UV-absorbance, except for the rare occurrence of instrument maintenance.
Ultraviolet Absorbance of DOM, Single-Wavelength Spectrophotometry:
Some S2 WEIR grab samples in the 354 series were also measured using a stand-alone, tunable, single-wavelength ThermoSpectronic Instruments (Rochester, New York) Spectronic 20 Genesys (model 4001/4) Spectrophotometer. 4010 10-mm round optical glass cell. The Tungsten-halogen lamp was warmed for a minimum of 30 minutes prior to absorbance measurement. Absorbance at 360 nm was zeroed to a blank of deionized water before the start of measurements and after every hour of operation. A sample was poured from a 250-mL sample bottle into a cell, which was then wiped to remove condensation or water drops before placing the cell into the spectrophotometer. A mark on the cell was always aligned with a mark on the compartment to assure the same path of light. The cell was rinsed with deionized water.
Bacterial Respiration of DOM:
Bacterial respiration (BR) of DOM was measured on two S2 WEIR samples collected during 2009 (347 serires) and five during 2010 (348 series) for a study of DOM biodegradability (Sebestyen et al. 2021a). These samples were collected at the S2 WEIR as grab samples.
An aliquot of a sample was collected in a 1-L acid-washed HDPE bottle or a new 250-mL LDPE bottle. Either bottle type was first 3x rinsed with sample water before being filled. Samples were chilled on ice, transported, and received within one day of sampling at the Aquatic Ecology Laboratory at the University of Minnesota, St. Paul. Mercuric chloride (1% by volume) was added to three vials after 0, 1, and 2 days to stop bacterial activity and preserve samples until analysis. During 2009, the time 0 stopping of bacterial respiration was done at the Marcell Research Center within hours of collection. Bacterial respiration rates were calculated as the total dissolved oxygen concentration loss over the first 48 hours during dark incubations of whole water in sets of nine 6-mL septa vials without headspace at 21 degrees Celsius. Dissolved oxygen concentrations were measured on a membrane-inlet mass spectrometer (Bay Instruments, Easton, Maryland). Ultrapure water (< 18.2 megaohm per centimeter) in gaseous equilibrium with an air-saturated headspace at 4 degree Celsius was used as a reference standard for measurements of oxygen to determine bacterial respiration rates. The detection limit is 0.3 mg oxygen/L.
Biodegradable DOC:
Aliquots of two S2 WEIR samples (348 series, 2009) were used to measured biodegradable DOC (BDOC) concentration for a study of DOM biodegradability (Sebestyen et al. 2021a). These samples were collected in 1-liter (L) high-density polyethylene (HDPE) bottles that were triple rinsed with sample water and then filled. Samples were chilled on ice and transported that day to the Aquatic Ecology Laboratory at the University of Minnesota (St. Paul). These samples were collected at the S2 WEIR as grab samples.
Biodegradable DOC was determined by filling 1-L muffled, dark glass Pyrex bottles with 750 mL of filtrate (muffled 0.7-micrometer glass fiber filters). Bottles were incubated in the dark at room temperature (22 degrees Celsius) for 1 year with 20-mL subsamples removed at time points of 0, 3, 7, 21, 58, 154, and 370 days to measure changes in DOC concentration over time. At each time point, a bottle was gently inverted to re-aerate the water and prevent oxygen limitation of DOM biodegradation. We defined the BDOC concentration as the amount of DOC consumed during these 1-year bottle incubations. Decreasing DOC concentration over time was fit as a negative exponential decay relationship, and the BDOC pool was measured as the initial DOC concentration (day 0) minus the remaining DOC concentration. The initial concentration was the highest DOC concentration during the first three days of the incubation since there was minor variability. The remaining DOC concentration was defined as the horizontal asymptote of the decay relationship. Concentrations of DOC were measured as NPOC on a Shimadzu TOC-V analyzer. Calibration and operation of this instrument is identical to that described for DOC analysis in the University of Minnesota Aquatic Ecology Laboratory. The detection limit is 0.5 mg carbon/L.
Liquid Water Isotopes:
The natural abundances of water stable isotopes (deuterium and oxygen-18) were measured using laser absorption spectroscopy (Lis et al. 2008) for select samples collected during and after 2008. These samples were collected at the S2 LAGG POOL or S2 WEIR (grab or automated sampling).
An aliquot of some samples from October to December 2007 and most samples onward from April 2008, was collected in a 16-ml scintillation vial with a Wheaton (DWK Life Sciences, Millville, New Jersey) Polyseal cap (phenolic with polyethylene cone liner) for liquid water isotope analysis. These samples were labelled in the 216 to 227 series and 346 to 355 series). Vials for water isotope aliquots were completely filled in the field, with no headspace or bubbles. Water isotopes samples are stored at room temperature in the Grand Rapids Forestry Sciences Laboratory before being analyzed there or shipped to other laboratories for analysis. At the time of analysis, a sample was pipetted from a scintillation vial into a 2-mL glass vial that fit into an autosampler rack. The 2-mL vials were sealed with PTFE caps having a silicon septa backed with a PTFE liner and injected into an isotope analyzer within 12 hr.
Some samples have been analyzed on Los Gatos Research (Mountain View, California) DLT-100 Water Isotope Analyzers at Plymouth State University (Center for the Environment Analytical Laboratory; Plymouth, New Hampshire), the University of California (Stable Isotope Facility, Davis), the University of Minnesota (Biometeorology Laboratory, St. Paul), or the University of Toronto (Integrated Watershed Hydrology and Biogeochemistry Research Facility, Scarborough, Ontario, Canada); or a Los Gatos Research T-LWIA-45-EP liquid water isotope analyzer at the Grand Rapids chemistry laboratory. Further information can be found in Stelling et al. (2021a, b). We provide a list of which samples were analyzed by which laboratory in an accompanying spreadsheet (S2_streamwater_waterIsotopeLabs.csv).
All laboratories used similar procedures and certified water isotope standards. Six or seven injections (0.5-1.2 microliter) of a sample were analyzed. Isotopic values are reported relative to the Vienna Standard Mean Ocean Water (VSMOW)-Standard Light Antarctic Precipitation (SLAP) scale. In each laboratory, a series of secondary standards were calibrated to VSMOW and SLAP. Machine raw data were post-processed to account for machine drift and between-sample memory (Wassenaar et al., 2014). Values for deuterium (D) and oxygen-18 (O-18) are reported in delta-notation (permil or per mil relative to VSMOW; Craig 1961). All laboratories used similar procedures to operate the instruments and post-process the isotopic ratios (Stelling et al. 2021). Each laboratory had slightly different analytical precisions for delta deuterium and delta oxygen-18 values: 0.8 permil for delta-D and 0.1 permil for delta-O-18 at Plymouth State University; 2 permil for delta-D and 0.25 permil for delta-O-18 at the University of California; 1 perml for delta-D and 0.25 permil for delta-O-18 at the University of Minnesota; 0.8 permil for delta-D and 0.25 permil for delta-O-18 at the University of Toronto; and 0.5 permil for delta-D and 0.1 permil for delta-O-18 at the Grand Rapids Forestry Sciences Laboratory.
Some samples have been held for weeks to more than a decade before analysis, which is not a concern as long as a vial remained sealed with no headspace. Samples are disposed of after analysis and quality assurance/quality control checks are completed. Unanalyzed samples in our archive of vials are available for eventual analysis.
Ferrous and Ferric Iron:
An aliquot of a weekly sample from September 2016 to November 2020 was collected for determination of ferric and ferrous iron concentration. Samples were collected at both the S2 LAGG POOL and S2 WEIR for a study of iron-mediated carbon cycling. These samples had IDs in the 227, 353, 354, and 355 series.
New sample bottles for ferrous (iron-II) and ferric (iron-III) iron analysis were not rinsed because each bottle received 0.5 ml (30-ml bottle) or 1 ml (60-ml) of 12 mol/L high-purity trace-metal grade hydrochloric acid in advance of sampling to preserve iron speciation in samples. Bottles for ferric/ferrous iron analysis were either refrigerated or stored at room temperature, and later shipped (up to several times a year) to Iowa State University for analysis.
Total iron and ferrous iron concentrations were colorimetrically measured using the high-throughput ferrozine method (Huang and Hall 2017) on a microplate spectrophotometer (Biotek Synergy HT, Winooski, Vermont, USA). Samples were run in triplicate on 96-well microplates and were re-run if there was a large disparity (>15 percent) among the triplicate wells. Standards were run with each set of 24 samples on the microplates. Ferric iron concentration was calculated as the difference between total iron and ferrous iron concentrations (Huang and Hall 2017). The calculation of ferric iron concentration sometimes resulted in small negative values for samples with extremely low ferric iron concentrations. These values were set to 0 mg/L. The detection limit is 0.04 mg/L for ferrous iron and 0.07 mg/L for ferric iron.
When a particular sample was higher in concentration than the highest calibration standard, that sample was diluted and re-run until within the range of calibration standards.
The sum of ferrous and ferric iron concentrations from colorimetric analysis occasionally exceeded the total iron values measured by ICP-OES at the Grand Rapids Forestry Sciences Laboratory. This discrepancy is likely a result of dissolution of particulate iron from occasional peat fragments in the acidified (pH < 2) samples used for ferrous and ferric iron analysis. Samples for colorimetric analysis were collected in pre-acidified bottles to inhibit oxidation of ferrous iron to ferric iron during sample storage. Samples for ICP-OES analysis were not similarly acidified for storage and analysis.
Acidified samples were held for weeks or months to several years before analysis. Samples were analyzed between 2016 and 2021.
Total mercury, methylmercury, and lead:
For various studies since 1993, an aliquot for total mercury (THg) and methylmercury (MeHg) determination was collected for particular samples and projects. From 1993 to 1998, samples were only analyzed for total mercury. During 1999, 2000, 2003 and 2004, there are no mercury analyses to report. Sampling every 1 to 2 weeks resumed during 2001 and both total mercury and methylmercury concentrations were measured. Both unfiltered and filtered samples have been analyzed, though rarely on the same sample except for weekly samples from 2017.
Sampling techniques have been fairly consistent among studies. All samples were collected using US EPA method 1669 (US EPA 1996) for trace metal sampling of ambient waters, though the method was modified to oftentimes allow collection by a single person. Sample bottles were double bagged. Ultra-clean trace metal protocols were used during sampling and for sample processing equipment.
Since 2008, many samples collected using ultra-clean sampling techniques were also analyzed for lead concentration to assess long-term ecosystem responses to this legacy pollutant (Jeremiason et al. 2018).
Numerous laboratories have been used to analyze samples with various instruments. Methods for total mercury were consistent with US EPA Method 1631 Revision E (US EPA 2002). Total mercury samples were oxidized overnight at 60 degrees Celsius with potassium bromine monochloride and neutralized with hydroxylamine hydrochloride prior to analysis. Then, inorganic mercury (Hg-II) was converted to elemental mercury (Hg-0) with stannous chloride reduction. Elemental mercury was then purged from solution and captured in dual gold trap amalgamation. Mercury was removed from a gold trap using thermal desorption in a stream of argon gas. Concentration was measured using cold vapor atomic fluorescence spectroscopy (CVAFS). Methylmercury samples were prepared using distillation, aqueous ethylation, and purge and trap. Methylmercury concentration was usually measured with CVAFS detection (US EPA 1998; Method 1630). Two laboratories used isotope dilution and detection by inductively-coupled plasma mass spectrometry (ICP-MS; e.g., Hintelmann and Ogrinc 2002). A water sample was distilled with 8 mol/L persulfate and 20 percent potassium chloride (KCl; weight/volume) in an acid-cleaned Teflon extraction manifold. Distillates were ethylated by addition of sodium tetraethylborate and purged from solution in a stream of mercury-free nitrogen gas. Volatile mercury species were captured on Tenax traps, followed by thermal desorption in a stream of argon gas. Species were separated using a chromatography column. Concentration was measured either by pyrolytic conversion to elemental Hg with detection by CVAFS or hyphenation to an ICP-MS.
From 1993 to 1998, an aliquot of a sample from the S2 LAGG POOL was collected, except one sample from March 27, 1996 that was collected at the S2 WEIR when S2 LAGG POOL was inaccessible due to ice. A mercury analysis aliquot was collected in an acid-washed 30-mL or 125-mL Teflon (PTFE, polytetrafluoroethylene) bottle that was rinsed three times with surface water before being filled. Prior to sampling, Teflon bottles were filled and stored with 1 percent high-purity hydrochloric acid, which was disposed of at the time of sampling. Unfiltered samples were collected from April 1993 to June 1998. Samples were acidified to 0.5 percent (volume per volume) with trace metal grade nitric or hydrochloric acid within hours of collection and then transported by overnight courier to the University of Minnesota (St. Paul). Samples were either stored in a dark refrigerator or frozen prior to analysis. Total mercury was measured using a Brooks Rand (Seattle, Washington, USA) Model III Total Mercury System with CVAFS detection, Mercury Guru 2.0 Software. The instrument was calibrated with analytical standards directly traceable to National Institute of Standards and Technology (NIST) reference materials. A 6-point calibration curve was made at the start of each run of 40 to 50 samples. Samples with concentrations exceeding the highest standard were diluted and rerun until within range of standards. Each analytical run also included two to three duplicate analyses, one or two blanks, and two or more check standards. References were prepared from National Research Council of Canada (NRCC) or US Department of Standards National Bureau of Standards (Washington, District of Columbia, USA) anhydrous standard reference materials. More details on the field sampling, laboratory, and analysis are provided by Kolka (1996). These data were used in studies of total mercury transport (Kolka et al. 2001; Grigal et al. 2000). The detection limit is 0.05 ng THg/L.
From 2001 to 2008, samples were collected at the S2 WEIR to serve as a reference for an experiment at the S6 catchment to assess mercury cycling and transport responses to sulfate pollution (Jeremiason et al. 2006). We refer to this study as the sulfate addition study. Samples for this study were labelled in seven-digit serial identification series that started with the year of sampling followed three decimal places from 2001 to 2004, with several samples in 2005 identified that way. For example, 2001.001 for the first sample collected in 2001. Other samples were labelled with a six-digit serial identifier that started with 296 during 2005, 295 during 2006, 307 during 2007, and 308 and 309 during 2008. Three samples had a seven-digit identifier with 1 decimal place (i.e., 307309.1, 308007.1 and 308657.1). Samples during 2009 were labelled in the 310 and 311 series when two samples were collected: one for mercury analyses (310 series) and another for the core suite of analytes (311 series). For samples collected from 2001 to 2005, no time was recorded when a sample was collected. Samples were pumped from the weir pool using a Geopump Peristaltic Pump (Geotech Environmental Equipment) with 0.64 cm internal-diameter Teflon tubing. Stream water was pumped through the entire length of tubing as a rinse. Then, a Perfluoroalkoxy alkane (PFA) filter holder (Savillex 401-31-47-10-31-2 Single Stage Filter Assembly) with a 47-mm diameter 0.7-micrometer glass fiber filter (GF/F) was attached to the tubing and rinsed for 3 seconds. Two new 125-mL PETG sample bottles were then filled after 3x rinsing with filtered sample water. Prior to filtration, filters were combusted for 4 hours at 500 degrees Celsius. Samples were acidified to 0.5 percent (volume per volume) with trace metal grade hydrochloric acid within 1 hour. Field duplicates and equipment blanks were collected every 10 samples. While several laboratories were used for total mercury and methylmercury analysis, procedures were similar among those laboratories. Instruments were calibrated daily and deionized water blanks, sample duplicates, and matrix spikes were included with each analytical run. Results were acceptable when references were within 15 percent of the actual value for both total mercury and methylmercury concentrations.
Samples collected from 2001 to 2003 were analyzed at the University of Minnesota Mercury Laboratory. Samples for total mercury were analyzed using a Brooks Rand Model III Total Mercury System with CVAFS detection and Mercury Guru software. A Tekran Instruments Corporation (Toronto, Ontario, Canada) Model 2400 with CVAFS detection was used to measure methylmercury concentrations. The detection limits are 0.2 ng THg/L and 0.04 ng MeHg/L.
All samples collected from 2005 to 2008 for total mercury analysis were measured at the University of Toronto at Mississauga Mercury Laboratory using a Tekran 2600 Automated Sample Analysis System with CVAFS detection. The detection limit is 1.1 ng THg/L. Samples collected during 2005 for methylmercury analysis were measured at the same laboratory using a Tekran 2500 with CVAFS detection. The detection limit is 0.065 ng MeHg/L.
Samples collected during 2006 for methylmercury analysis were measured at the Gustavus Adolphus College Mercury Laboratory using A Tekran 2500 Automated Sample Analysis System with CVAFS detection. The detection limit is 0.03 ng MeHg/L.
Samples collected during 2007 to 2008 were analyzed at the Metropolitan Council Environmental Science Services Division Laboratory (St. Paul, Minnesota). A Brooks-Rand Model III with CVAFS detection was used to measure methylmercury. The detection limit is 0.03 ng MeHg/L.
Samples collected from 2009 to 2011 were analyzed at the University of Toronto at Scarborough Mercury Laboratory. A Tekran 2600 Automated Sample Analysis System with CVAFS detection was used to measure total mercury concentration. For methylmercury analysis, water samples were first distilled with addition of copper sulfate, potassium chloride, sulfuric acid and a known quantity of enriched methylmercury-199 isotope. Distillates were analyzed for methylmercury content by buffered ethylation and Tenax trapping, and then thermal desorption of Tenax traps into a hyphenated gas chromatography-ICP-MS (Agilent 7700x ICP-MS). Isotope dilution calculations were used to derive concentrations of methymercury using peak areas derived via Agilent MassHunter software. The detection limits are 0.15 ng THg/L and 0.03 ng MeHg/L.
Independent of the sulfate addition study, samples in the 399 ID series were also collected from 2009 to 2011 and after.
For samples collected from 2009 to 2010, an aliquot of each weekly sample (399 ID series with a corresponding timestamp in the 348 to 350 series for the core suite of analytes) was collected in an acid-washed 0.5-L, 1-L, or 2-L Teflon bottle. Prior to sampling, a Teflon bottle was filled and stored with 1 percent high-purity hydrochloric acid. Sample bottles were double bagged and rinsed three times with surface water before filling. Samples were shipped overnight to the University of Minnesota Mercury Laboratory. Upon receipt, the samples were filtered through 45-mm diameter Whatman 0.45 micrometer cellulose nitrate filters in disposable Nalgene filter units. Samples for total mercury were preserved and digested by addition of 1 percent bromine monochloride and samples for methylmercury were acidified with 0.5 percent trace-metal grade hydrochloric acid. A Brooks Rand Model III with CVAFS detection was used for total mercury analysis while a Tekran Model 2400 with isothermal gas chromatography, pyrolysis, and CVAFS detection was used for methylmercury analysis. Both instruments were calibrated with 6 standards at the start of each analytical run. Every sample was measured in duplicate or triplicate, and at least one blank followed every 10 samples. References were prepared from NIST reference materials. Analytical standards and certified references were measured two to three times with each batch of samples. The detection limit is 0.2 ng THg/L and 0.04 ng MeHg/L.
Samples from the S2 WEIR were occasionally collected during 2010 and weekly during and after 2011. These samples were shipped to the Gustavus Adolphus Laboratory for lead and mercury concentration measurements. One sample from the S2 LAGG POOL was analyzed for lead. While these samples have an accompanying sample with an identical timestamp in the 348 to 355 series for the core suite of analytes, all of the samples have six-digit serial identification numbers that start with 399 or 400 (the same as some DOC samples, also analyzed at the Gustavus Adolphus laboratory). These samples were collected into 125-mL PETG sample bottles, immediately chilled, and shipped by overnight courier. Upon receipt, the samples were filtered through 45-mm diameter Whatman 0.7 micrometer glass fiber filters in Savillex teflon filter holders and acidified by adding 0.5% (volume/volume) 12 molar hydrochloric acid. Filters were precombusted at 450 degrees Celsius for 4 hr. Samples for lead concentration measurement were diluted 1:100 with 0.32 mol/L nitric acid prior to addition of an internal standard (Inorganic Ventures, Christiansburg, Virginia, USA, ICPMS-71D) followed by analysis on an Agilent Scientific Instruments (Santa Clara, California) 7700x Series ICP-MS with ShieldTorch System, MassHunter Software and 3rd generation Octopole Reaction System. For lead quantification, the ICP-MS was calibrated with multiple levels of Inorganic Ventures analytical standards to span the range from 0 to 10 microgram/L. Analytical standards, a check standard (Inorganic Ventures 1643 Trace Elements in Fresh Water) and a certified reference material (SLRS-5 riverine water, National Research Council Canada, Halifax, Nova Scotia, Canada) were measured after every 15 samples. Lead recovery from a reference material (SLRS-5) was always within the range of certified values. Total mercury concentration was measured using a Brooks Rand MERX-T Automated Mercury System with CVAFS detection. Distillates were analyzed for methylmercury concentration by buffered ethylation and Tenax trapping, and then thermal desorption of Tenax traps into a gas chromatograph (Brooks Rand MERX-M Automated Mercury System) hyphenated to the Agilent 7700x ICP-MS. Aliquots for methylmercury determination were spiked with about 40 picogram of mercury-199 for isotope dilution. Isotope dilution calculations were used to calculate methylmercury concentrations using peak areas derived via Agilent MassHunter Software. Both the MERX-T and ICP-MS were calibrated with a blank and a single analytical standard. The detection limits are 0.2 microgram lead/L, 0.2 ng THg/L and 0.02 ng MeHg/L.
Samples from the S2 WEIR were collected weekly from 2014 to 2015 and shipped to the University of North Carolina Mercury Laboratory. These samples were part of a natural-abundance mercury isotopic tracing study (Woerndle et al. 2016). While these samples have an accompanying sample with an identical timestamp in the 348 to 355 series for the core suite of analytes, all of the samples have six-digit serial identification numbers that start with 399. These samples were collected into 0.5-L, 1-L, or 2-L PTFE sample bottles. Prior to sampling, a PTFE bottle was filled and stored with 1 percent trace metal grade hydrochloric acid. Sample bottles were double bagged and rinsed three times with surface water before filling. Samples were shipped overnight. For analysis, samples were digested using a different approach than the traditional bromine monochloride method, which has recoveries from 70 to 90% of mercury in a sample (Gu et al. 2011; Wang et al. 2015). A water sample was digested overnight in an acidic (nitric and sulfuric acids in a 20 to 1 ratio) mixture of permanganate and persulfate at 95 degrees Celsius (Balogh et al. 1996; Woerndle et al. 2018). The improved mercury recovery with permanganate and persulfate digestion (90 to 105 percent; Balogh et al. 1996) was needed to avoid artificial fractionation of mercury isotopes during sample digestion (Woerndle et al. 2018). Aside from substituting permanganate and persulfate, the total mercury method was the same as US EPA Method 1631 with detection using a Brooks Rand Model III CVAFS. The instrument was calibrated using 0 to 1 ng THg/L solutions prepared from the NIST Standard Reference Material 3133 (NIST-3133) working standard and verified using NIST-1641d as a secondary standard. For methylmercury concentration measurement, unfiltered water was transferred to acid-washed PTFE bottles and sample water was preserved with 0.4 percent (volume per volume) trace metal grade hydrochloric acid and stored at 4 degree Celsius in the dark. A 50-mL aliquot of sample was prepared using distillation, aqueous ethylation, and purge and trap. Methylmercury concentration was measured using a Brooks Rand Model III CVAFS with isothermal gas chromatography and pyrolysis. The instrument was calibrated using 0 to 0.5 ng MeHg/L solutions prepared from the CEBAM Analytical, Inc (Bothell, Washington, USA) working standard that was verified against a standard prepared from NIST-3133. The detection limits are 0.2 ng THg/L and 0.04 ng MeHg/L. For mercury natural-abundance isotope measurement, samples digested with the permanganate and persulfate approach were neutralized with aliquots of 30 percent hydroxylamine. About 1 L of a neutralized sample was weighed and placed in an acid-washed borosilicate glass media bottle with 3 mL of hydroxylamine and 100 mL of 50 percent trace-metal grade sulfuric acid. The media bottles were capped with a 3-hole Pyrex (Corning Inc., Corning, New York, USA) closure and continuously stirred. One tube was used to deliver air from which mercury was stripped with a syringe filter and a gold trap. A second tube was used to add 10% stannous chloride at 1 mL/min. A third tube was used to route elemental mercury to a trap solution containing 1 percent potassium permanganate in 10% sulfuric acid. The water sample was purged by a glass sparger for about 3 hr. A standard (NIST-3133) was added to a trap solution and isotope ratios were measured using a Nu Instruments (Wrexham, United Kingdom) multicollector ICP-MS according to the method described in Blum and Bergquist (2007). Mass-dependent fraction (MDF) of mercury isotopes is reported as the abundance of Hg-202 in a sample relative to NIST-3133 in permil. The mass-independent fractionation (MIF) is the difference between the measured relative abundance of Hg-199, Hg-200, Hg-201, and Hg-204 and the expected relative abundance based on mass dependence of Hg-202. Analytical uncertainties are 0.04 to 0.12 permil for MDF, and 0.07 to 0.16 permil for Hg-199 MIF as determined from replicate analyses of NRC Tort-2 (lobster hepatopancreas reference material) at different concentrations from 0.7 to 5.0 ng/g.
REPORTED VALUES:
To document when a sample was collected, we include a laboratory ID, sample name, and date/time of collection. Sometimes chemistry values are assigned NA for individual solutes or for all analytes (i.e., pH, specific conductivity, and solute concentrations), which may have resulted from insufficient sample volume to complete all analyses, contamination that affected individual solutes or suites of analytes that were simultaneously measured on a single instrument for a particular sample, or contamination that affected all solutes for a particular sample. Samples pending analysis are also assigned a value of NA.
Data values below the detection limit are reported in the data file and are not flagged. Detection limits, as listed above, have changed over time and must be considered when using these data. In recent years, rather than setting negative concentration values to 0, values for any analyte from the Grand Rapids laboratory have been reported as the instrument generated value between the detection limit and 0. However, for most analytes prior to 2010, values less than detection limits were only reported as < the detection limit value in spreadsheets or paper records. Those values, when entered into a database that was restricted to numerical values, were set to 0 when ostensibly the value is a continuous variable between 0 and the detection limit. We acknowledge so-called insider censoring (Helsel 2005, 2009) that is present in the time series of concentration data, and realize that the complexities of shifting detection limits and differential reporting of values less than detection limits complicates analysis and interpretation of data.
MARCELL EXPERIMENTAL FOREST sites and data collection are described in further detail in:
Sebestyen, S.D., C. Dorrance, D.M. Olson, E.S. Verry, R.K. Kolka, A.E. Elling, and R. Kyllander (2011). Chapter 2: Long-Term Monitoring Sites and Trends at the Marcell Experimental Forest. In R.K. Kolka, S.D. Sebestyen, E.S. Verry, and K.N. Brooks (Ed.). Peatland Biogeochemistry and Watershed Hydrology at the Marcell Experimental Forest (pp 15-71). CRC Press, Boca Raton, FL. https://www.fs.usda.gov/treesearch/pubs/37979.
RELEVANT PUBLICATIONS
REFERENCES
Aiken, G. R. (1992), Chloride interference in the analysis of dissolved organic carbon by the wet oxidation method, Environ. Sci. Technol., 26(12), 2435-2439. https://doi.org/10.1021/es00036a015
APHA. (2017). Standard methods for the examination of water and wastewater (23rd ed.). Washington, DC: American Public Health Association
Balogh, S. J., Meyer, M. L., and Johnson, D. K. (1996), Mercury and suspended sediment loadings in the lower Minnesota River, Environ. Sci. Technol., 31(1), 198-202. https://doi.org/10.1021/es960327t
Blum, J. D., and Bergquist, B. A. (2007), Reporting of variations in the natural isotopic composition of mercury, Analytical and Bioanalytical Chemistry, 388(2), 353-359. https://doi.org/10.1007/s00216-007-1236-9
Coleman Wasik, J. K., Mitchell, C. P. J., Engstrom, D. R., Swain, E. B., Monson, B. A., Balogh, S. J., Jeremiason, J. D., Branfireun, B. A., Eggert, S. L., Kolka, R. K., and Almendinger, J. E. (2012), Methylmercury declines in a boreal peatland when experimental sulfate deposition decreases, Environ. Sci. Technol., 46(12), 6663-6671. https://doi.org/10.1021/es300865f
Coleman Wasik, J. K., Engstrom, D. R., Mitchell, C. P. J., Swain, E. B., Monson, B. A., Balogh, S. J., Jeremiason, J. D., Branfireun, B., Kolka, R. K., and Almendinger, J. E. (2015), The effects of hydrologic fluctuation and sulfate regeneration on mercury cycling in an experimental peatland, Journal of Geophysical Research: Biogeosciences, 120(9), 1697-1715. https://doi.org/10.1002/2015JG002993
Craig, H. (1961). Isotopic variations in meteoric waters. Science, 133(3465), 1702-1703. https://doi.org/10.1126/science.133.3465.1702
Fellman, J. B., D'Amore, D. V., and Hood, E. W. (2008), An evaluation of freezing as a preservation technique for analyzing dissolved organic C, N and P in surface water samples, Sc. Total Environ., 392(2-3), 305-312. https://doi.org/10.1016/j.scitotenv.2007.11.027
Fleck, J. A. (1999), Mercury transport through northern forested watersheds: dissolved and particulate pathways, MS thesis thesis, 100 pp, University of Minnesota, St. Paul, MN.
Grigal, D. F., Kolka, R. K., Fleck, J. A., and Nater, E. A. (2000), Mercury budget of an upland-peatland watershed, Biogeochemistry, 50(1), 95-109. https://doi.org/10.1023/A:1006322705566
Gu, B., Bian, Y., Miller, C. L., Dong, W., Jiang, X., and Liang, L. (2011), Mercury reduction and complexation by natural organic matter in anoxic environments, PNAS, 108(4), 1479-1483. https://doi.org/10.1073/pnas.1008747108
Hintelmann, H., and Ogrinc, N. (2002), Determination of stable mercury isotopes by ICP/MS and their application in environmental studies, in Biogeochemistry of Environmentally Important Trace Elements, edited, pp. 321-338, American Chemical Society. https://doi.org/10.1021/bk-2003-0835.ch021
Huang, W., and Hall, S. J. (2017). Optimized high-throughput methods for quantifying iron biogeochemical dynamics in soil. Geoderma, 306, 67-72. https://doi.org/http://dx.doi.org/10.1016/j.geoderma.2017.07.013
Jacobson, M. M. F. (2012), Biological and photochemical degradation of dissolved organic carbon in peatland ecosystems, PhD dissertation thesis, 89 pp, University of Minnesota, St. Paul.
Jeremiason, J. D., Engstrom, D. R., Swain, E. B., Nater, E. A., Johnson, B. M., Almendinger, J. E., Monson, B. A., and Kolka, R. K. (2006), Sulfate addition increases methylmercury production in an experimental wetland, Environ. Sci. Technol., 40(12), 3800-3806. https://doi.org/10.1021/es0524144
Jeremiason, J., Baumann, E., Sebestyen, S. D., Agather, A., Seelen, E., Carlson-Stehlin, B., Funke, M., and Cotner, J. B. (2018), Contemporary mobilization of legacy Pb stores by DOM in a boreal peatland, Environ. Sci. Technol., 52(6), 3375-3383. https://doi.org/10.1021/acs.est.7b06577
Hertzler, R. A. (1938), Determination of a formula for the 120-degree V-notch weir, Civil Engineering, 8(11), 756-757.
Kolka, R. K. (1996), Hydrologic transport of mercury through forested watersheds, PhD dissertation thesis, 264 pp, University of Minnesota, St. Paul, MN.
Kolka, R. K., Grigal, D. F., Verry, E. S., and Nater, E. A. (1999), Mercury and organic carbon relationships in streams draining forested upland/peatland watersheds, J. Environ. Qual., 28(3), 766-775.
Kolka, R. K., Grigal, D. F., Nater, E. A., and Verry, E. S. (2001), Hydrologic cycling of mercury and organic carbon in a forested upland-bog watershed, Soil Sci. Soc. Am. J., 65(3), 897-905.
Lis, G., Wassenaar, L. I., and Hendry, M. J. (2007). High-precision laser spectroscopy D/H and 18O/16O measurements of microliter natural water samples. Analytical Chemistry.
Nyberg, P. R. (1987). Soil survey of Itasca County, Minnesota. USDA Soil Conservation Service, St. Paul, MN.
Sebestyen, S. D., Funke, M. M., and Cotner, J. B. (2021a), Sources and biodegradability of dissolved organic matter in two peatland catchments with different upland forest types, northern Minnesota, USA, Hydrol. Process., 35(2), e14049. https://doi.org/10.1002/hyp.14049
Sebestyen, S. D., Funke, M. M., Cotner, J., Larson, J. T., and Aspelin, N. A. (2020), Water chemistry data for studies of the biodegradability of dissolved organic matter in peatland catchments at the Marcell Experimental Forest: 2009-2011, 2 ed., Forest Service Research Data Archive, Fort Collins, CO. https://doi.org/10.2737/RDS-2017-0067-2
Sebestyen, S. D., Lany, N. K., Roman, D. T., Burdick, J. M., Kyllander, R. L., Verry, E. S., and Kolka, R. K. (2021b). Hydrological and meteorological data from research catchments at the Marcell Experimental Forest, Minnesota, USA. Hydrological Processes, 35, e14092. https://doi.org/10.1002/hyp.14092
Sebestyen, S. D., and Verry, E. S. (2011), Water chemistry responses to watershed experiments at the Marcell Experimental Forest, in Peatland biogeochemistry and watershed hydrology at the Marcell Experimental Forest, edited by Kolka, R. K., et al., pp. 401-432, CRC Press, Boca Raton, FL.
Sebestyen, S. D., Verry, E. S., and Brooks, K. N. (2011), Hydrological responses to forest cover changes on uplands and peatlands, in Peatland biogeochemistry and watershed hydrology at the Marcell Experimental Forest, edited by Kolka, R. K., et al., pp. 433-458, CRC Press, Boca Raton, FL.
Stelling, J. M., Sebestyen, S. D., Griffiths, N. A., Mitchell, C. P. J., and Green, M. B. (2021a), The stable isotopes of natural waters at the Marcell Experimental Forest, Hydrol. Process., 35(10), e14336. https://doi.org/10.1002/hyp.14336
Stelling, J. M., Sebestyen, S. D., Griffiths, N. A., Mitchell, C. P. J., Green, M. B., and Lany, N. K. (2021b), Marcell Experimental Forest stable isotopes of water, 2008 - ongoing, 1 ed., USDA Forest Service, Environmental Data Initiative. https://doi.org/10.6073/pasta/50a287716f80d2b1b602dcceac5a6e5c
Urban, N. R. (1983), The nitrogen cycle in a forested bog watershed in northern Minnesota, MS thesis, 359 pp, University of Minnesota, St. Paul, MN.
Verry, E. S., Brooks, K. N., Nichols, D. S., Ferris, D. R., and Sebestyen, S. D. (2011). Watershed hydrology. In Kolka, R. K., Sebestyen, S. D., Verry, E. S., and Brooks, K. N. (Eds.), Peatland biogeochemistry and watershed hydrology at the Marcell Experimental Forest (pp. 193-212). Boca Raton, FL: CRC Press.
US Environmental Protection Agency (1995), Method 525.2: Determination of organic compounds in drinking water by liquid-solid extraction and capillary column gas chromatography/mass spectrometry, US Environmental Protection Agency, Washington, DC.
US EPA (1996), Method 1669: Sampling ambient water for trace metals at EPA water quality criteria levels, 35 pp., US Environmental Protection Agency, Washington, DC.
US EPA (1998), Method 1630: Methyl mercury in water by distillation, aqueous ethylation, purge and trap, and cold vapor atomic fluorescence spectrometry, 46 pp., US Environmental Protection Agency, Washington, DC.
US EPA (2002), Method 1631, revision E: Mercury in water by oxidation, purge and trap, and cold vapor atomic fluorescence spectrometry, US Environmental Protection Agency,Washington, DC.
Verry, E. S., and Janssens, J. (2011). Geology, vegetation, and hydrology of the S2 bog at the MEF: 12,000 years in northern Minnesota. In Kolka, R. K., Sebestyen, S. D., Verry, E. S., and Brooks, K. N. (Eds.), Peatland biogeochemistry and watershed hydrology at the Marcell Experimental Forest (pp. 93-134). Boca Raton, FL: CRC Press.
Wang, Y., Li, Y., Liu, G., Wang, D., Jiang, G., and Cai, Y. (2015), Elemental mercury in natural waters: Occurrence and determination of particulate Hg(0), Environ. Sci. Technol., 49(16), 9742-9749. https://doi.org/10.1021/acs.est.5b01940
Wassenaar, L. I., Coplen, T., and Aggarwal, P. K. (2014). Approaches for achieving long-term accuracy and precision of d18O and d2H for waters analyzed using laser absorption spectrometers. Environmental Science and Technology, 48(2), 1123-1131. https://doi.org/10.1021/es403354n
Weishaar, J. L., Aiken, G. R., Bergamaschi, B. A., Fram, M. S., Fujii, R., and Mopper, K. (2003), Evaluation of specific ultraviolet absorbance as an indicator of the chemical composition and reactivity of dissolved organic carbon, Environ. Sci. Technol., 37(20), 4702-4708. https://doi.org/10.1021/es030360x
Woerndle, G. E., Tsz-Ki Tsui, M., Sebestyen, S. D., Blum, J. D., Nie, X., and Kolka, R. K. (2018), New insights on ecosystem mercury cycling revealed by stable isotopes of mercury in water flowing from a headwater peatland catchment, Environ. Sci. Technol., 52(4), 1854-1861. https://doi.org/10.1021/acs.est.7b04449
