Lake sediments store metal contaminants from historic pesticide and herbicide use and mining operations. Historical regional smelter operations in the Puget Sound lowlands have resulted in arsenic concentrations exceeding 200 μg As g^-1 in urban lake sediments. Prior research has elucidated how sediment oxygen demand, warmer sediment temperatures, and alternating stratification and convective mixing in shallow lakes result in higher concentrations of arsenic in aquatic organisms when compared to deeper, seasonally stratified lakes with similar levels of arsenic pollution in profundal sediments. In this study we examine the trophic pathways for arsenic transfer through the aquatic food web of urban lakes in the Puget Sound lowlands, measuring C and N isotopes – to determine resource usage and trophic level – and total and inorganic arsenic in primary producers and primary and secondary consumers. Our results show higher levels of arsenic in periphyton than in other primary producers, and higher concentrations in snails than in zooplankton or insect macroinvertebrates. In shallow lakes arsenic concentrations in littoral sediment are similar to deep profundal sediments due to arsenic remobilization, mixing, and redeposition, resulting in direct arsenic exposure to littoral benthic organisms such as periphyton and snails. The influence of littoral sediment on determining arsenic trophic transfer is evidenced by our results which show significant correlations between total arsenic in littoral sediment and total arsenic in periphyton, phytoplankton, zooplankton, snails, and fish across multiple lakes. We also found a consistent relationship between the percent inorganic arsenic and trophic level (determined by δ¹⁵N) in lakes with different depths and mixing regimes. Cumulatively, these results combine to provide a strong empirical relationship between littoral sediment arsenic levels and inorganic arsenic in edible species that can be used to screen lakes for potential human health risk using an easy, inexpensive sampling and analysis method.

Study area and site characterization

Of the 10 study lakes located in the Puget Sound lowland region of Washington State, USA, 8 lakes are within the predicted deposition field of arsenic-contaminated smelter emissions, and two control lakes are located outside the predicted deposition field (Pine and Bonney Lakes). Of the 8 lakes potentially impacted by aerial fallout of arsenic, littoral sediment concentrations range from 7 to 213 µg As g^-1.

Previous work reported that Angle Lake and Lake Killarney have the greatest concentrations of arsenic in profundal sediment for the 8 sampled lakes in the predicted deposition field (208 μg g⁻¹ and 206 μg g⁻¹, respectively. Lake Killarney is shallow (maximum depth 4.6 m) and experiences frequent, year-round mixing events (termed ‘polymictic’). Angle Lake is deeper (maximum depth 15.8 m) and displays strong seasonal stratification. Each type of organism described below was collected from Angle Lake and Lake Killarney, and a subset of those organisms was sampled from the 8 remaining lakes. A summary of samples collected from each lake and the analyses performed is provided in the S1 Table.

Sample collection

Water and food web sampling

Water samples for dissolved arsenic, and phytoplankton and zooplankton tows, were collected from a boat at approximately the deepest point in each lake as described previously. In June 2019, water samples were collected using a peristaltic pump, filtered (0.45 µm Geotech cartridge filter), acidified with 1% HNO₃ (v/v) in the laboratory, and allowed to stand for 14 days prior to analysis for dissolved arsenic. Duplicate phyto- and zooplankton samples were collected in 2016, 2017, and 2019 using a vertical net tow (20 μm and 153 μm mesh, respectively) from 1-2 m above the lakebed. Phytoplankton were then pre-filtered through a 153 μm sieve to remove zooplankton. Plankton were collected on 5.0 µm polycarbonate filters and either dried at 60 °C (for total As) or stored at -80 °C (for As speciation).

Periphyton was grown in situ on multiple 23 cm² acrylic plates assembled into 3 x 3 plate frames and placed at 0.5 m depth in the approximate center of each lake. Plates were deployed for at least 35 days during the summers of 2019 and 2021. Both sides of each plate were scraped of periphyton with a plastic razor blade and rinsed with ultrapure reverse osmosis de-ionized water into a filtration vessel. Material was captured on 5.0 µm polycarbonate filters and subsamples were either dried at 60 °C overnight or stored at -80 °C. Triplicate grab samples of mixed submerged macrophytes were harvested from nearshore (1-2 m depth) using an aquatic plant rake (Pond & Beach Rake Gen 2) in August and September 2020 and cleaned of attached sediment. Roots were removed and random portions of each rake harvest were either dried at 60 °C overnight or stored at -80 °C.

Insect midge larvae (Chironomidae) were collected from sediments in July and August 2020 at multiple sites in each lake using a dredge (Wildco stainless steel petite ponar). Dredge samples were sifted through a mesh sieve to reveal larvae within. Larvae were transferred into a container of lake water on ice. In the laboratory, larvae were separated into groups of 9-10 individuals per composite sample and either dried at 60 °C overnight or stored at -80 °C. Chinese mystery snails (Bellamya chinensis) were collected in June 2019 and July 2021 from the littoral zone (< 2 m depth) of 6 lakes. The collection involved hand netting via snorkeling or from a boat. B. chinensis were placed on ice in sealed plastic bags on the boat, euthanized in the laboratory, and then later thawed and body tissue removed from the shell. Pumpkinseed (Lepomis gibbosus) and bluegill sunfish (L. macrochirus) were collected by beach seining from 5 lakes in June 2019 and 2021. A scientific collection permit was approved for fish by the Washington Department of Fish and Wildlife (SCP Olden 22-172); no permit is required for non-vertebrate sample collection. Field site access does not require a permit for publicly accessible lakes in Washington. Fish were euthanized and muscle tissue dissected. Fish and snail tissues were either dried at 60 °C overnight or stored at -80 °C.

Laboratory Analyses

Oven-dried samples of periphyton, phyto- and zooplankton, macrophytes, Chironomidae, snail whole soft tissue, and fish tissue were divided into 2 subsamples for total arsenic and δ¹³C and δ¹⁵N analysis. Small portions (< 3 mg) of each food web constituent were transferred into tin capsules for stable isotope analysis. All samples stored at -80 °C were prepared for arsenic speciation by drying at 85 °C. Macrophyte, snail, and fish tissue samples were homogenized using a porcelain mortar and pestle, while all other samples were used in entirety for arsenic analysis.

Total arsenic

Samples for total arsenic analysis were prepared by microwave-assisted (CEM MARS 5) total digestion protocol (modified EPA method 3015a) using trace metal grade HNO₃ in pressurized digestion vessels. After digestion, sample solutions were diluted to 2% (v/v) HNO₃. Concentrations of total arsenic in water and digested sediment, periphyton, plankton, Chironomidae, and B. chinensis and fish tissue samples were determined by inductively-coupled plasma mass spectrometry (ICP-MS; Agilent 7900). Calibration was performed using a certified multi-element standard (Agilent Multi-element calibration standard-2A). Efficacy of the digestion procedure was verified using certified reference material BCR-414 (Trace elements in plankton), NIST 2711a (Montana Soil II), and DOLT-5 (dogfish liver) which yielded a recovery of 90 ± 15% (n = 8), 92 ± 16% (n = 10) and 91 ± 12% (n = 22), respectively. Analytical accuracy of the ICP-MS method was assessed using certified reference material NIST 1640a (trace elements in natural water), which had a recovery of 87 ± 6% (n = 14) for arsenic. The limit of detection (LOD) for arsenic was 0.25 μg L^-1. Samples below the limit of detection were assigned a value of 0.125 μg L^-1 (half the LOD).

Arsenic speciation

Arsenic speciation of phyto- and zooplankton collected in 2016 and 2017 was determined in the Trace Element Analysis laboratory at Dartmouth College using methods based on Taylor and Jackson [42]. Plankton samples were freeze-dried and transferred from polycarbonate filters into sample vials and 10% methanol was added. Samples were sonicated in a 30 °C bath for 2 h, then filtered (0.2 µm) to remove residual solids. The supernatant was eluted with 20 mM (NH₄)₂CO₃ (1.1 mL min⁻¹) on an Agilent LC1120 liquid chromatograph. Arsenic species were separated using an anion-exchange column (Hamilton PRP-X100) at 35 °C and then analyzed on an Agilent 8900 triple quadrupole ICP-MS. Analytical accuracy was verified using a secondary standard with a recovery of 104 ± 2% (n = 3). The extraction efficiency (sum of species/total As by ICP-MS) for phytoplankton was 10 ± 6% (n = 4) and 23 ± 2 % (n = 2) for zooplankton.

The speciation of arsenic in phyto- and zooplankton collected in 2019, periphyton, macrophytes, Chironomidae, B. chinensis, and fish tissue were also determined at the Trace Element Analysis laboratory at Dartmouth College, following a dilute acid heat assisted extraction (modified from [43]). An aliquot of the extract was diluted 1:1 with 200 mM NH₄CO₃ and analyzed for arsenic species by ion chromatography coupled to inductively coupled plasma mass spectrometry (IC-ICP-MS). A Dionex AS14 anion-exchange column was used with an NH₄CO₃ gradient to separate arsenate, arsenite, dimethylarsinic acid, monomethylarsonic acid, and arsenosugars (run time of 7 minutes, flow rate of 1 mL min^-1, and injection volume of 20 μL). The system was calibrated using primary standard solutions of the arsenic species (Sigma Aldrich, Spex certiprep) and NIST 2669a urine level II was used for quality control. The extraction efficiency was 80 ± 16% (n = 6) for fish, 57 ± 27% (n = 6) for B. chinensis, 88 ± 9% (n = 2) for Chironomidae, 55 ± 14% (n = 2) for macrophytes, 124 ± 45% (n = 2) for zooplankton, 87 ± 0.9% for phytoplankton (n = 2), and 60 ± 17% (n = 2) for periphyton. Other studies report similar ranges for extraction efficiencies across different environmental materials using equivalent methods.

Stable isotope analysis

To explore how differences in lake morphometry may influence arsenic trophic transfer, we conducted a food-web investigation in lakes Angle and Killarney, two spatially proximate lakes with similar levels of arsenic in profundal surface sediments but differing maximum depths and mixing regimes. We used ¹³C/¹²C as a proxy for basal C resources and ¹⁵N/¹⁴N as a proxy for trophic position, which together indicate resource use. Periphyton, plankton, macrophyte, Chironomidae, B. chinensis, and fish tissue samples were dried for 24–48 h at 60 °C, homogenized with mortar and pestle, and encapsulated in tin capsules. Tissues were sent to the University of California Davis Stable Isotope Facility and analyzed for ratios of stable isotopes (¹³C/¹²C and ¹⁵N/¹⁴N) using an elemental analyzer (PDZ Europa ANCA-GSL) interfaced with an isotope ratio mass spectrometer (PDZ Europa 20-20; Sercon Ltd., Cheshire, UK). Data are reported as permil (‰) relative differences from standards of Vienna Pee Dee Belemnite for C and atmospheric N, expressed as delta (δ) units. Long-term standard deviations for estimates of natural abundance stable isotope values based on reference material at the University of California-Davis are 0.2‰ for δ¹³C and 0.3‰ for δ¹⁵N.