We installed 12 mesocosms in Lake 239 (L239) at the IISD Experimental Lakes Area in northwestern Ontario, Canada, on 18 June 2012. As described by Furtado et al. (2015)
, mesocosms were cylindrical enclosures approximately 2 m in diameter, 1.6–1.7 m in depth, and open to the sediments (Curry Industries, Winnipeg, Manitoba, Canada). Mesocosms were seeded with zooplankton (≥80 μm, 311 individuals/L) collected from L239 from a depth of 20 m to the surface using vertical net hauls.
We used 1.0 mg/mL stock solutions of 50 nm spherical polyvinylpyrollidone- (PVP) or citrate-coated AgNPs (NanoComposix, San Diego, California, USA) suspended in Milli-Q water (18.2 MΩ·cm) and 2 mmol citrate buffer, respectively. Ag concentrations in stock solutions were verified using inductively coupled plasma mass spectroscopy (ICP-MS) following acid (4% HNO3
) digestion. Mean PVP- and citrate-coated particle diameter (determined by dynamic light scattering by the supplier) was 53.6 and 54.9 nm, respectively (Furtado et al. 2015
). Further details were given by Furtado et al. (2014
Starting on 23 June 2012, we added 0, 0.89, 3.56, and 14.24 mg of PVP-coated AgNPs every other day for 39 d to control, low, medium, and high chronic treatments, respectively, for a total of 0, 17.8, 71.2, and 284.8 mg AgNPs added to each treatment, respectively. Target final nominal concentrations (3000 L estimated mesocosm volume) in control, low, medium, and high chronic treatments were 0, 6, 24, and 95 μg PVP AgNP/L, respectively. The citrate treatment received citrate-coated AgNPs in the same dosing regimen as the high chronic treatment. Mesocosms were mixed after each AgNP addition with care taken not to disturb sediments.
On 11 July 2012, mesocosms assigned to the pulse treatment received a single dose of 240 mg of PVP-capped AgNPs to achieve a target nominal concentration of 80 μg AgNP/L. Treatments were replicated twice. Mesocosms were mixed after AgNP addition as described above.
The study reported here was part of a larger investigation of AgNP fate in natural waters and the effects of AgNP exposure on biotic communities. Detailed descriptions of collection and analysis of many of the variables we correlate to ecosystem metabolism have been published elsewhere (AgNP fate and behaviour: Furtado et al. 2014
; bacterioplankton abundance, production, and stoichiometry: Blakelock et al. 2016
; phytoplankton and zooplankton biomass and community structure: Vincent et al. 2017
). We include brief descriptions of these methods below along with detailed descriptions of our calculations of ecosystem metabolism and Ag accumulation in the food web, which are not published elsewhere.
Total Ag concentration (TAg) in screened water was quantified by ICP-MS after acidification (4% HNO3
) and digestion at 70 °C for 6 h (Furtado et al. 2014
We divided algae into large (0.7–35 μm) and small (0.7–1.2 μm) size fractions. For the large fraction, two subsamples of screened water were filtered onto ashed glass fiber filters (average pore size 0.7 μm). As described by Vincent et al. (2017)
, filters were dried at 60 °C and carbon (C) content was measured using a carbon nitrogen elemental analyzer. Two additional subsamples were filtered onto glass fiber filters, which were frozen and analyzed for chlorophyll a
(chl) using a cold ethanol extraction (Sartory and Grobbelaar 1984
) and quantification via fluorometry. A final subsample was filtered onto a 0.8 μm polycarbonate membrane, which was stored in 4% nitric acid and digested at 70 °C for 2 h. The digested solution was filtered (0.45 μm nylon filter) and analyzed for Ag using ICP-MS. C, chl, and Ag concentrations of the small fraction were quantified similarly using 1.2 μm prefiltered screened water (Blakelock et al. 2016
Bacteria in samples of screened water were counted via flow cytometry after staining with SYBR Green I, and heterotroph production was quantified using 3
H-labelled leucine incorporation assays (Blakelock et al. 2016
). Cell-specific production (fg C/cell/d) was calculated as heterotroph production (fg C/L/d) divided by cell abundance (# cells/L).
Zooplankton sample processing and community responses were described by Vincent et al. (2017)
. Collected samples were split using a plankton wheel. Half of the sample was preserved in 2% sugar-buffered formalin and half was filtered through an 80 μm mesh, dried at 60 °C for 24 h, and acidified with 4% nitric acid for Ag analysis. Individuals in preserved zooplankton samples collected from control and high chronic treatments were identified to genus and species (when possible). Individual lengths were measured, and published length–weight regressions (Persson and Ekbohm 1980
; Malley et al. 1989
) were used to quantify total biomass in units of C, which was scaled to mesocosm volume. We assumed that zooplankton biomass was 45% C (Andersen and Hessen 1991
). Zooplankton Ag was quantified from acidified samples as described for microplankton.
Hourly, water from each mesocosm was sequentially pumped past probes that measured dissolved oxygen concentration (DO), temperature (Fast Response Oxygen Optode, #4330F, Aanderaa Instruments), and chlorophyll a fluorescence (Cyclops-7, Hoskins Scientific), and it was then recirculated back into each mesocosm. A data logger (Campbell Scientific, CR1000) controlled the pump and valve system and recorded the data. The pumping system had a small reservoir volume that resulted in carry-over between adjacent mesocosms. We estimate that <1% of total mesocosm volume was exchanged over the course of the experiment.
Daily ecosystem respiration (ER), gross primary production (GPP), and net ecosystem production (NEP) were calculated from diurnal changes in DO concentration corrected for reaeration (calculated using wind speed and barometric pressure from a nearby meteorological station) using the “book keeping” method reviewed by Staehr et al. (2010)
. All equations used below are adapted from those found in table 2 in Staehr et al. (2010)
DO concentrations are adjusted to account for physical exchange across the lake surface according to the following governing equation (Odum 1956
is atmospheric oxygen exchange, and A
is a term including all other processes affecting DO (assumed to be negligible). Change in DO concentration over time was measured empirically by the optical probe. F
was calculated by comparing DO concentration at saturation with the atmosphere (a function of temperature, salinity, and barometric pressure; detailed equations found in Staehr et al. 2010
) to the measured value after accounting for wind speed measured at a nearby meteorological station.
This method assumes that photosynthesis is not occurring during darkness. Therefore, hourly ER rate (g O2
) was calculated by correcting the mean change in DO concentration observed during darkness for F
according to the following equation:
where Z mix
is the depth to which the water column feely mixes, assumed to be the entire mesocosm depth. This method also assumes that ER rate is constant over 24 h. Therefore, daily ER (g O2
) was calculated by multiplying the ER rate by 24.
Both respiration and photosynthesis are occurring during daylight hours. Therefore, the change in DO concentration observed during daylight is the net result of these two processes. Hourly NEP rate (g O2
) during daylight was calculated by correcting the mean change in DO concentration observed during daylight for F
according to the following equation:
The amount of NEP during the daylight period (g O2 m−3 daylight period−1) was calculated by multiplying the hourly NEP rate during daylight by the number of daylight hours.
Daily GPP (g O2
) was then calculated by subtracting the amount of ER during the daylight period (a negative O2
flux) from the amount of NEP during the daylight period according to this equation:
where ERdaylight period
is equal to ERhourly
multiplied by the number of daylight hours.
Daily NEP (g O2 m−3 d−1) was then derived by the difference between daily GPP and daily ER.
We used logarithm response ratios (LRR) to evaluate community and ecosystem responses to AgNP exposure over time. LLR were calculated as the natural logarithm of the value of the response variable (e.g., chl/L) measured in a treatment mesocosm divided by the mean of the response variable from the two control mesocosms. Negative and positive LRR indicate suppressed and stimulated treatment responses, respectively, compared to the controls. LRR of ER, GPP, and NEP were calculated using weekly means of daily values. Mean LRR, or effect sizes, were compared to zero using single sample t tests. Carbon-specific Ag in each biotic compartment was calculated by dividing Ag concentration (Ag/L) by the carbon content (C/L). We used the Bonferroni-corrected α = 0.0002 to evaluate LRR statistical significance.