A compendium of geochemical information from the Saanich Inlet water column

Extensive and expanding oxygen minimum zones (OMZs) exist at variable depths in coastal and open ocean waters. As oxygen levels decline, nutrients and energy are increasingly diverted away from higher trophic levels into microbial community metabolism, resulting in fixed nitrogen loss and production of climate active trace gases including nitrous oxide and methane. While ocean deoxygenation has been reported on a global scale, our understanding of OMZ biology and geochemistry is limited by a lack of time-resolved data sets. Here, we present a historical dataset of oxygen concentrations spanning fifty years and nine years of monthly geochemical time series observations in Saanich Inlet, a seasonally anoxic fjord on the coast of Vancouver Island, British Columbia, Canada that undergoes recurring changes in water column oxygenation status. This compendium provides a unique geochemical framework for evaluating long-term trends in biogeochemical cycling in OMZ waters.

Extensive and expanding oxygen minimum zones (OMZs) exist at variable depths in coastal and open ocean waters. As oxygen levels decline, nutrients and energy are increasingly diverted away from higher trophic levels into microbial community metabolism, resulting in fixed nitrogen loss and production of climate active trace gases including nitrous oxide and methane. While ocean deoxygenation has been reported on a global scale, our understanding of OMZ biology and geochemistry is limited by a lack of time-resolved data sets. Here, we present a historical dataset of oxygen concentrations spanning fifty years and nine years of monthly geochemical time series observations in Saanich Inlet, a seasonally anoxic fjord on the coast of Vancouver Island, British Columbia, Canada that undergoes recurring changes in water column oxygenation status. This compendium provides a unique geochemical framework for evaluating long-term trends in biogeochemical cycling in OMZ waters.

Background & Summary
Marine oxygen minimum zones (OMZs) are widespread, naturally occurring water column features that arise from respiration of organic matter in subsurface waters with restricted circulation. Operationally defined by oxygen (O 2 ) concentrations between 0 to 20 μM, and differential accumulation of nitrite (NO 2 − ) and reduced sulphur compounds, OMZs currently constitute 1-7% of global ocean volume [1][2][3][4][5][6][7] . As oxygen levels decline, nutrients and energy are increasingly diverted away from higher trophic levels into microbial community metabolism 2,8 . As a result, OMZs are hotspots for the biogeochemical cycling of carbon, nitrogen and sulphur with resulting feedback on nitrogen loss processes and climate active trace gas production including nitrous oxide (N 2 O) and methane (CH 4 ) [9][10][11][12][13] . The effects of climate change, including increased stratification and reduced O 2 -solubility in warming waters are resulting in OMZ expansion and intensification 1,8,[14][15][16][17][18] reinforcing the need to monitor changes in water column geochemistry in oxygen-deficient waters.
Oceanographic surveys in OMZ waters rely on a standard suite of measurements including temperature, salinity, density and conductivity. Additional parameters including irradiance, used to measure water column light penetration, fluorescence used to monitor chlorophyll concentrations and dissolved gases including O 2 and carbon dioxide (CO 2 ) provide information on primary production 2,19,20 . Chemical measurements of phosphate (PO 4 3 − ), silicic acid (SiO 2 ), and nitrate (NO 3 − ) are measured as essential nutrients supporting growth and cell division 15 . Nitrite (NO 2 − ) and ammonium (NH 4 + ) are also measured to better constrain nitrogen cycling processes 2,10,11,13 . Because some OMZs can become completely anoxic, hydrogen sulfide (H 2 S) concentrations can be used as an indicator for sulphate reduction driving chemoautotrophic metabolism 3,9 . Measurements of N 2 O and CH 4 can also be used to monitor potential climatological impacts of OMZ expansion [9][10][11][12][13]21 . Collectively, these measurements define geochemical gradients in OMZ water columns that shape the conditions for coupled biogeochemical cycling.
Saanich Inlet is a seasonally anoxic fjord on the coast of Vancouver Island, British Columbia, Canada [22][23][24][25] . Saanich Inlet is an inverse estuary where a glacial sill at the mouth restricts exchange between deep basin and external waters for most of the year. Freshwater is supplied at the inlet mouth predominantly by the Cowichan and Fraser Rivers, producing horizontal density differences that result in an inward flow into the inlet in the surface layer and outward flow at depth 25,26 . During spring and summer months, high levels of primary productivity in surface waters and limited vertical mixing of basin waters below the sill result in anoxia and the accumulation of CH 4 , NH 4 + and H 2 S 27-29 . In late summer and fall, neap tidal flows produce an influx of denser water from the Northeastern subarctic Pacific (NESAP) Ocean that cascade over the sill, resulting in vertical mixing and the re-supplying of deep basin waters with O 2 and nutrients 25,26 . The recurring seasonal development of water column anoxia followed by deep water renewal makes Saanich Inlet a model ecosystem for monitoring biogeochemical responses to changing levels of water column O 2 -deficiency 2,30-32 .
Here we present a compendium of time-series observations encompassing historical O 2 measurements 25,33 (Fig. 1a) and more recent monthly monitoring efforts in Saanich Inlet from 2006 through 2014, representing over 100 independent sampling expeditions (Fig. 1b). This compendium contains physical (temperature, salinity, density, irradiance, and fluorescence), chemical (PO 4 3 − , SiO 2, NO 3 − , NO 2 − , NH 4 + , and H 2 S), dissolved gas (O 2 , CO 2 , N 2 , N 2 O, CH 4 ), and biological (cell counts) parameter data (Fig. 1b,c) useful in comparing to other oceanographic time-series from the Northwest Atlantic to Eastern Tropical Pacific through the Global Ocean Sampling expeditions 34 , the Hawaii and Tara Oceans [34][35][36] and Bermuda Atlantic Time-series 37 and in the development of biogeochemical models. In addition, this geochemical compendium is paired with a cognate compendium of multi-omic sequence information (DNA, RNA, protein) focused on microbial diversity, abundance and function. 38 Combined, these compendiums provide a community-driven framework for observing and predicting microbial community repsonses to changing levels of oxygen deficiency extensible to open ocean OMZs.

Methods
Time-series monitoring in Saanich Inlet was conducted on a monthly basis aboard the MSV John Strickland at station S3 (48°35.500 N, 123°30.300 W) as previously described 32

Environmental sampling
Historical dissolved O 2 concentrations were obtained from station S3 by sampling with Niskin bottles at discrete depths and subsequently analyzing water samples using various modifications of the Winkler method 25,33,39 (Data Citation 1). Historical water column profiles can also be accessed at the Ocean Sciences Data Inventory website hosted by the Institute of Ocean Sciences and Fisheries and Oceans Canada (http://www.pac.dfo-mpo.gc.ca/science/oceans/data-donnees/search-recherche/profiles-eng.asp). Samples collected from February 2006 to February 2008 were processed and analysed for dissolved gases and nutrients as first reported in Zaikova et al. 32 (Fig. 2). Beginning on February 2008, a Sea-Bird SBE 25 CTD (conductivity, temperature and depth), with Sea-Bird SBE 43 dissolved O 2 and Biospherical Instruments PAR sensors attached was used to measure conductivity, temperature, dissolved O 2 , PAR/ Irradiance and fluorescence (Data Citation 1). To minimize the effects of off-gassing, waters were collected in the following order; dissolved O 2 for Winkler titration (from select depths for CTD calibration), dissolved gases (N 2 O and CH 4 ), NH 4 + , H 2 S, nutrients, cell counts (Data Citation 1) and salinity (from selected depths for CTD calibration). A detailed seawater sampling video protocol can be found online (http://www.jove.com/video/1159/seawater-sampling-and-collection).
Chemical data CTD data analysis. CTD data were downloaded, converted and pre-processed in the laboratory using the SeaBirdSeasoft software. Downcast data of the deepest cast (200 m) was extracted and converted from ASCII format into a.cnv file for manual curation. Salinity and density were calculated using the Derive module with the corrected conductivity measurements. Temperature and salinity were exported using an ITS-90 scale. Oxygen sensor measurements collected in millilitre per litre (ml l −1 ) were converted to micromolar (μM) units (Data Citation 1). Discrete winkler analyses from water samples spanning LV depths were used to calibrate the CTD O 2 measurements (Data Citation 1).
Nitrate, phosphate and silicic acid. For each depth, sample water was filtered through a 0.2 μm acrodisc (Millipore) and used to rinse a 15 ml tube three times before filling with 14 ml. Samples were stored on ice and later in the lab at −20°C for up to four months prior to analysis. A Bran Luebbe AutoAnalyser 3 using air-segmented continuous-flow and standard colorimetric methods was used for analysis. In brief, nitrate (NO 3 − ) was reduced to nitrite by a copper-cadmium reduction column. Nitrite was then quantified by a modified colorimetric assay 40 , reading sample absorbance at 550 nm. Orthophosphate (PO 4 3 − ) was quantified based on the colorimetric method for reduced phospho- molybdenum complex, reading samples absorbance at 880 nm 41 . Silicic acid (H 4 SiO 4 ) was quantified by reduction to a molybdenum blue complex, reading sample absorbance at 820 nm. Oxalic acid was added to remove phosphate interference 40 (Data Citation 1).

Ammonium.
A fluorometric measurement protocol for NH 4 + analysis was carried out as previously described in Holmes et al. for marine samples 42 . For each depth, glass amber scintillation bottles were rinsed three times, then filled to overflowing and capped immediately to minimize off-gassing of NH 4 + and stored on ice for 1-3 h before processing. A total of 5 ml of sample water was transferred to vials with 7.5 ml o-phthaldialdehyde (OPA; Sigma) in triplicate. Simultaneously, 7.5 ml of OPA was added to prepared NH 4 + standard curve (0.025-10.0 μM NH 4 Cl) and stored at room temperature for up to 4 h. Fluorescence at 380 ex /420 emm was read using a Turner Designs TD-700 fluorometer (2006-2009) or Varioskan plate reader (2009)(2010)(2011)(2012)(2013)(2014) in triplicate with 300 μl of sample or standard in a 96-well round bottom plate (Corning) (Fig. 3) (Data Citation 1).
Nitrite. The protocol for NO 2 − analysis was carried out as previously described in Armstrong et al. modified for marine samples 40 . For each depth, sample water was filtered through a 0.2 μm acrodisc (Millipore) and used to rinse a 15 ml tube three times before filling with 14 ml filtered sample water and stored on ice for 1-3 h before processing. A total of 2 ml of sample water was transferred to 4 ml plastic cuvettes in triplicate, and 100 μl sulphanilamide and 100 μl nicotinamide adenine dinucleotide (NAD) were added. Simultaneously, reagents were added to prepared standards (0.025-5.0 μM NaNO 2 ). Cuvettes were inverted and stored on ice for up to 4 h. Absorbance at 542 nm was read using a Cary60 spectrometer (Fig. 3) (Data Citation 1).
Hydrogen sulfide. The protocol for H 2 S was carried out as previously described in Cline 43 modified for marine samples. For each depth, 10 ml sample water was collected directly into a 15 ml tube containing 200 μl 20% Zinc Acetate and stored on ice for 4-24 h before processing. A total of 300 μl of sample was transferred into triplicate wells of a 96-well round or flat-bottom plate (Corning), and 6 μl Hach Reagent (Hach) 1 and 2 for sulphide assay were added to each well. After 5 min incubation, absorbance at 670 nm was read using a spectrophotometer (2008)(2009)  Dissolved gases. For each depth, sample water was collected through silicon tubing (~15 cm long and 1/4″ thick, pre-flushed for a few seconds with sample water) into a 30 or 60 ml borosilicate glass serum vial, overflowing three times the volume and taking care to remove air bubbles from the tubing and vial during filling. The vials were spiked with 50 μl saturated mercuric-chloride solution, then crimp-sealed  with a butyl-rubber stopper and aluminium cap. Samples were stored in the dark at 4°C until processing. Dissolved gases were analysed using either headspace for CO 2 , CH 4 , N 2 and N 2 O (2006-2009, samples stored for up to 2 years) or automated purge-and-trap for CH 4 and N 2 O only (2009-2014, samples stored for o3 months) coupled with gas chromatography-mass spectrometry (GC-MS) 44 (Data Citation 1). Samples with >20% s.d. between replicates were excluded from our study to discard any long storage effects.

Data Records
Data record 1 The Saanich Inlet O 2 historical data  is accessible in comma-separated-value format file 'Historical_O2_DATA.csv' on the Dryad Digital Repository [Data Citation 1] containing the data fields outlined in Table 1.

Data record 2
The Saanich Inlet time-series CTD data is accessible in comma-separated-value format file 'Saanich_TimeSeries_CTD_DATA.csv' on the Dryad Digital Repository [Data Citation 1] domain containing the data fields outlined in Table 2.

Data record 3
The Saanich Inlet time-series chemical data is accessible in comma-separated-value format file 'Saanich_TimeSeries_Chemical_DATA.csv' on the Dryad Digital Repository domain containing the data fields outlined in Table 3 [Data Citation 1].

Data record 4
The Saanich Inlet time-series Winkler O 2 data is accessible in comma-separated-value format file 'Saanich_TimeSeries_Winkler_DATA.csv' on the Dryad Digital Repository [Data Citation 1] domain containing the data fields outlined in Table 4.

Technical Validation
Data quality control Data in the Saanich Inlet time series was collected and processed by experienced scientists with extensive training in the sampling methods and data processing steps described above. People interested in becoming part of the scientific crew were invited to participate in training sessions with experienced scientists in the field and laboratory to gain practical experience. Once in the field, trainees were carefully supervised during sample collection for a minimum of 3 months for quality assurance. Following each cruise, the acting chief scientist compiled all chemical and physical data collected and conducted initial quality controls, checking for outliers and verifying standard curves. Data were then entered into an in-house database along with field notes and precise records of volumes of water filtered informing downstream analyses.

CTD and chemical data validation
The SeaBird 43 dissolved O 2 sensor was calibrated by Winkler O 2 measurements 45 . Samples from selected depths were collected into Winkler glass Erlenmeyer flasks using latex tubing, overflowing three times to ensure no air contamination. Oxygen concentration was determined using a Brinkman autotitrator, routinely calibrated with a potassium iodide standard. Stability of CTD O 2 measurements was determined by comparing the high values with Winkler measurements, and low values with sulfidic profiles where the sensors levels off. Where H 2 S is detected we consider O 2 measurements to be 0 μM based on spontaneous auto-oxidation reaction of H 2 S with O 2 . We have estimated our limit of detection for the automated Winkler method at~0.007 ml l −1 or~0.3 μM.
The SeaBird conductivity sensor was calibrated using salinity samples collected at selected depths.

Data field Description Units
Longitude Unique geographical coordinates for sampling station Decimal degrees followed by letter    Salinity glass bottles were rinsed 4 times and filled with water sample, stored at room temperature and analyzed within 4 months on a Guildline Portasal salinometer. For each cruise, standard curves for NH 4 + and NO 2 − were prepared. Stock solutions and reagents for both assays were freshly made every three months and stored in the dark at 4°C and were tested prior to being used for analysis. Stock solution quality and assay validation was carried out using linear regression and calculating the r squared value (r 2 ≥ 0.90) on the absorbance data (Fig. 3). Standard curve stock solutions and reagents for H 2 S assay were evaluated every three months based on manufacturer's instructions. We have estimated our limit of detection for these assays to be 0.

Flow cytometry validation
Concentration of flow cytometry (FL) alignment beads was determined by microscopy using a hemocytometer. Bead counts for each FL run were then used to calculate the volume of sample measured. Two blanks were included in each FL run, and consisted of sterile water bead/dye solution with sterile water in place of sample water, to ensure instrument cleanliness and optics function. Size gates were set to include beads and bacterial and archaeal cell sizes and to reduce noise of any small particulate debris.

Gas analysis validation
A thorough review of the Purge and Trap GCMS (PT-GCMS) method validation has been previously described 44 . Standard curves were run at the start of each batch of 25 samples by injecting precisely measured quantities of a standard gas mixture (CH 4 , N 2 O, CO 2 and N 2 ) calibrated against National Ocean and Atmospheric Administration (NOAA) certified reference gas mixture. Single standards were also measured every~2-hours (5-6 sample per run) to monitor instrument drift. The precision of CH 4 and N 2 O measurements based on replicate measurements of air-equilibrated water samples was o4%. Accuracy was confirmed by measuring dissolved N 2 O and CH 4 in carefully prepared air-equilibrated, temperature-controlled Milli-Q water and comparing this to expected concentrations based on gassolubility equations 46,47 . Detection limits depend on the volume of sample being purged, and were 0.8 nM for CH 4 and 0.5 nM for N 2 O for the samples analyzed in this time-series (2009-2014) (Fig. 3). Samples were run in duplicate or triplicate to ensure reproducible readings. The relative s.d. between replicate samples was calculated and included in the output data. The output data are also carefully inspected to ensure optimal instrument performance during sample analysis before being submitted to the database.

Usage Notes
Oxygen considerations Based on the amount of dissolved O 2 in the water column and the biogeochemical processes associated with it, thresholds for O 2 -defined water column conditions were determined as previously described 2 . As the range of O 2 concentrations is wide and has great impact on biological processes in the lower concentrations, we suggest using the O 2 thresholds described in Wright et al. 2 for analysis.