Dataset: Carbon and Nitrogen

To tease apart the effects of seasonal variation in light availability and nutrients on the response of high-latitude kelp species to pH and temperature, the investigators conducted two separate studies: a "winter" experiment from February 12 to March 18, 2020 (35 days), and a "summer" experiment from August 15 to September 16, 2020 (32 days). In the experimental design, analysis, and reporting, the investigators endeavored to follow best practices for OA research with macroalgae (Cornwall et al., 2012; Cornwall & Hurd, 2016). Both experiments took place at the Sitka Sound Science Center in a flow-through seawater system drawing source water from 20 meters (m) depth (MLLW) in Sitka Sound, Alaska. Incoming seawater was filtered to 20 micrometers (μm) and routed through a UV filter (Smart UV®, Pentair) before diverging into two temperature-controlled (TITAN® heat pump and Optima compact heaters, AquaLogic) recirculating tanks representing treatments for "current" or control temperatures (7° Celsius (C) in winter; 14°C in summer) and "future" ocean warming (OW) projections (11°C in winter; 18°C in summer) (IPCC, 2018) by season. From here, temperature-regulated seawater was pumped into eight header tanks where pH was maintained at setpoint levels for control conditions (pHT 7.6 in winter; pHT 7.9 in summer) and "future" ocean acidification (OA) projections (pHT 7.2 in winter; pHT 7.5 in summer) (Mathis et al., 2015) through a relay system (N = 2 header tanks per pH/temperature treatment). In both seasonal experiments, achievable pHT setpoints for control treatments were constrained by the ambient pH of incoming seawater and were, therefore, lower than the typical seasonal in situ pHTminima observed on local rocky reefs by ~0.1 - 0.2 pH units (Kroeker et al., 2021). That said, the lower-than-average pH values of the control treatments did still fall within the observed pHs captured across all years of in situ environmental data. The investigators chose to maintain the projected end-of-century pH offset for this region (~0.4 pH units) to define the OA treatment setpoints relative to achievable control pH levels. A DuraFET sensor (Honeywell) in each header tank communicated real-time pH measurements to a controller (UDA 2152, Honeywell) that regulated injection of pre-equilibrated low pH seawater through solenoid valves into the headers to maintain pH at treatment set points. The low pH (~6) seawater was produced by bubbling pure CO2 gas into two tanks of seawater flowing from each temperature-controlled tank. Once in each header tank, the CO2 and temperature-equilibrated seawater was continuously mixed before delivery to 24 experimental aquaria (N = 3 aquaria per header) at an average flow-through rate of 2-2.5 liters per minute per aquaria (L min-1 aquaria-1). The "Header pH and Temperature" Supplemental File (2020_kelpGCexps_headerdata.csv) contains a summary of the calibrated pH and temperature data recorded by the Durafet sensors in each header during the experiments.

Seawater nutrient concentrations were not manipulated and thus reflected what was delivered through source water inflow to the system during each experiment. Terrestrial outflow from heavy precipitation over Southeast Alaska’s temperate rainforests and wind stress dynamics in the Gulf of Alaska control nutrient supply onto the Northeast Pacific shelves (Hermann et al., 2009; Hood & Scott, 2008; Ladd & Cheng, 2016; Stabeno et al., 2016). The complexities of how climate change may impact these drivers in tandem with altered phytoplankton productivity (Ji et al., 2010) means that there is little consensus on how seasonal nutrient supply into Sitka Sound may change. Therefore, the investigators chose to assume that nutrient availability, like seasonal light availability, would not differ significantly in this region in the future. All aquaria were fitted with a full-spectrum light (Aqua Illumination) that provided seasonally relevant regimes of photosynthetically active radiation spectra and photoperiod within the aquaria based on observations during overcast days in Sitka Sound (Bell et al., 2022). The entire experimental system was shielded from external light sources, and aquaria positions were randomized by treatment and location in the laboratory to minimize spatial variation among the random factors aquaria and header. The "Nutrient Concentrations with Experimental Aquaria" Supplemental File (2020_kelpGCexps_nutrients.csv) contains the data on nutrient concentrations within experimental aquaria.

At the beginning, middle, and end of each experiment, discrete water samples were collected for the determination of pHT, TA, and nutrient concentrations in each aquarium and header tank. Water samples were collected without aeration and poisoned with saturated HgCl2 (0.025%) in glass bottles within 20 minutes. Airtight samples were transported to the University of California Santa Cruz (UCSC) for analysis within 9 months of collection. pH was measured spectrophotometrically (Shimadzu, UV-1800) using m-cresol purple following best practices (Dickson et al., 2007). Total alkalinity (TA) was measured using open cell titration (Metrohm, 905 Titrandro) and corrected against certified reference materials of CO2 in seawater (Dickson laboratory, Scripps Institute of Oceanography). Water chemistry samples from each tank had a mean standard error of 0.0013 pH units and 0.87 micromoles per kilogram seawater (μmol kg-1 SW-1) among sample triplicates. To calculate in situ pH on the total hydrogen ion concentration scale (pHT; mol kg-1 SW-1)(Dickson, 1993), the investigators used their laboratory measurements of spectrophotometric pH and TA, measurements of temperature and salinity recorded with a handheld meter (YSI) concurrently with discrete water sample collection, and stoichiometric dissociation constants (Dickson & Millero, 1987; Mehrbach et al., 1973) as inputs to the program CO2SYS (Lewis & Wallace, 1998; Pierrot et al., 2006). Corrections were applied to the continuous time series of pH values recorded by durafets in each header during each experiment by calculating an average offset from pHT values calculated from discrete samples.

Kelp used in both winter and summer experiments came from 4.5-7.5 m depth at Talon Island (57.073 N, 135.414 W), Sitka Sound. These experimental "individuals" were collected as whole thalli (Neoagarum fimbriatum and Hedophyllum nigripes), or as single blades with their attached pneumatocysts that were cut from young sporophytes at approximately 1 m above their holdfasts (Macrocystis pyrifera). During transport to the laboratory and prior to the start of the experiments (less than 2 days), all algae was held continuously in ambient flow-through seawater (winter experiment: ~6°C, pHT 7.8; summer experiment: ~13.5°C, pHT 8.0). Individuals were removed briefly only to clean off epiphytes and to record initial morphometrics (maximum blade length, total wet mass) after trimming all blades to 10 centimeters (cm) total length. The investigators took pictures of each trimmed blade to estimate total surface area using ImageJ (NIH v1.8.0).

In both the winter and summer experiments, 3 individuals of each kelp species were randomly assigned to each experimental aquaria (N = 18 individuals per species per treatment). The investigators affixed individuals upright in aquaria by placing their stipes or pneumatocysts through three-strand line suspended over the open ends of 5 cm tall PVC stands. After all seaweeds were processed for initial morphometrics, pH and temperature were gradually changed in treatment tanks stepwise over the course of 3 days to reach final setpoints. During the experiment, kelps were visually checked daily for necrosis and were lightly brushed biweekly during aquaria cleaning to remove diatoms.

At the end of each experiment, individuals were measured and photographed for final morphometrics. Due to the difficulty in capturing three-dimensional tissue growth and the error inherent in wet mass measurements, kelp growth rates were estimated using three different metrics: wet mass in grams (g), maximum blade length in centimeters (cm), and total blade surface area in square centimeters (cm²). The initial (Ginitial) and final (Gfinal) measurements of each metric were used to calculate three relative growth rates (RGR; percent per day (% d-1)) for each individual using Equation 1 (see the attached Supplemental File), where Δt (days (d)) represents the total number of days elapsed between the beginning and end of the experiment. Relative growth rates were used for subsequent statistical analyses of experimental results. Absolute blade length extension rates were used to compare experimental growth to in situ kelp growth measurements (Bell & Kroeker, 2022).

From each individual, the investigators excised new blade tissue grown during the experiment adjacent to the intercalary meristem and pooled this tissue for all species replicates in each aquarium. A portion of this tissue was frozen at -20°C for use in feeding assays. The other portion of this tissue was dried at 60°C for >24 hours and analyzed for nitrogen (N) content (% dry mass) and δ13C values by the UCSC Stable Isotope Laboratory using a CE Instruments NC2500 elemental analyzer coupled to a Thermo Scientific DELTAplus XP isotope ratio mass spectrometer via a Thermo-Scientific Conflo III (routine measurement error ≤ 1.0 %C and ≤ 0.2 %N). The investigators also analyzed blade tissue from non-experimental kelp individuals collected at Talon Islands in each season ("field controls"; N=6 species-1 season-1) for elemental and isotopic analysis.

To compare in situ nutrient and light data with aquaria conditions during the experiment, environmental data was collected at the Talon Island experimental collection site. Benthic seawater was collected for the determination of nutrient concentrations in February and August 2020 (N=3 samples-1 season-1). Seawater for nutrient samples was immediately filtered through a 0.2 µm filter and frozen until analysis for dissolved inorganic nitrogen content as NOx (NO3 + NO2) and ammonium (NH4+) on a Lachat QuikChem 8000 Flow Injection Analyzer at the University of California Santa Cruz Marine Analytical Laboratory (detection limits: < 0.28 µM NOx, < 2.40 µM NH4; average run measurement error < 0.1 µM NOx < 0.8 µM NH4). A Diving-PAM-II (Heinz Wlz GmbH) MINI-SPEC was used to haphazardly record the photosynthetic photon flux density (PPFD; micromoles per square meter per second (µmol m-2 s-1) reaching the benthos at more than 10 locations along the ~5 m depth contour on two clear days in winter (February 28) and summer (September 19).

Tissue from H. nigripes and N. fimbriatum individuals grown in the laboratory was used to investigate whether future ocean conditions affect the palatability of these understory kelp species in either season. In April 2021, methods used by Hay et al. (1994) were modified to create "gels" of homogenized kelp tissue suspended in agar and enmeshed in squares of window screen. Each 30 cm² gel was formed from 0.1547 ± 0.0004 g (mean ± SE) of freeze-dried (FreeZone, Labconco) H. nigripes or N. fimbriatum tissue growth in either the control treatment or the combination OW and OA treatment from each seasonal experiment. The total number of gels used for the feeding assays was limited by the available kelp tissue grown during each experiment, and was consequently lower for gels made from tissue grown in the winter experiment (H. nigripes: N = 11 gels per treatment, N. fimbriatum: N = 12 gels per treatment) versus the summer experiment (H. nigripes: N = 24 gels per treatment, N. fimbriatum: N = 23 gels per treatment). Palatability assays were run by feeding these seaweed gels to the common kelp forest grazer, Strongylocentrotus droebachiensis (green urchin). Urchins with a test diameter of 24 ± 3 millimeters (mm) were collected from the intertidal, starved for 48 hours, and then placed in a flow-through chamber with a single gel in ambient seawater conditions (~7 °C, ~8.0 pH) for 48 hours. Photographs were taken of each gel before and after the assay and the relative consumption of seaweeds grown under different treatments was determined using Image J (NIH v1.8.0).