Article pubs.acs.org/est
Heat-Wave Effects on Oxygen, Nutrients, and Phytoplankton Can Alter Global Warming Potential of Gases Emitted from a Small Shallow Lake Maciej Bartosiewicz,*,†,‡,⊥ Isabelle Laurion,†,‡ François Clayer,† and Roxane Maranger‡,§ †
Centre Eau Terre Environnement, Institut national de la recherche scientifique, 490 de la Couronne, Québec G1K 9A9, Canada Groupe de Recherche Interuniversitaire en Limnologie et en environnement aquatique, Université de Montréal, C.P. 6128 Succ. Centre-Ville, Montréal, Québec H2V 2S9, Canada § Département des Sciences Biologiques, Université de Montréal, C.P. 6128 Succ. Centre-Ville, Montréal H2V 2S9, Canada ‡
S Supporting Information *
ABSTRACT: Increasing air temperatures may result in stronger lake stratification, potentially altering nutrient and biogenic gas cycling. We assessed the impact of climate forcing by comparing the influence of stratification on oxygen, nutrients, and globalwarming potential (GWP) of greenhouse gases (the sum of CH4, CO2, and N2O in CO2 equivalents) emitted from a shallow productive lake during an average versus a heat-wave year. Strong stratification during the heat wave was accompanied by an algal bloom and chemically enhanced carbon uptake. Solar energy trapped at the surface created a colder, isolated hypolimnion, resulting in lower ebullition and overall lower GWP during the hotter-than-average year. Furthermore, the dominant CH4 emission pathway shifted from ebullition to diffusion, with CH4 being produced at surprisingly high rates from sediments (1.2−4.1 mmol m−2 d−1). Accumulated gases trapped in the hypolimnion during the heat wave resulted in a peak efflux to the atmosphere during fall overturn when 70% of total emissions were released, with littoral zones acting as a hot spot. The impact of climate warming on the GWP of shallow lakes is a more complex interplay of phytoplankton dynamics, emission pathways, thermal structure, and chemical conditions, as well as seasonal and spatial variability, than previously reported.
■
INTRODUCTION Small and shallow lakes are abundant water bodies that play major roles in global carbon and nutrient cycles1,2 and in controlling regional climate.3,4 The thermal structure, mixing patterns, and energy fluxes of lakes are strongly influenced by meteorological conditions.5 The meteorological forcing drives turbulence both at the air−water interface and within the water column and can therefore influence the gas-exchange dynamics.6,7 Weather conditions during the ice-free season are also responsible for year-to-year differences in nutrient inputs from the watershed, be it total load or timing of delivery.8 Consequently, changes in weather conditions can affect both biotic and abiotic components of lake ecosystems, with possible implications on water quality and the biogenic production of greenhouse gases (GHG). Lakes of different sizes and mixing regimes have been shown to respond differently to the strength and persistence of meteorological forcing.9 Models simulating the effects of future warming on lakes predict that an increase in average summer temperature and irradiance may lead to greater water-column stability in both large10 and small lakes.11,12 However, although changes in stratification of large deep lakes will be relatively © XXXX American Chemical Society
slow in time, the mixing regimes in shallow lakes will be more strongly influenced by warming over short time-scales (mainly due to their larger surface-to-volume ratios). The possible effects of a stronger stratification in shallow lakes may include higher rates of carbon and nutrient remobilization and a switch from macrophyte to phytoplankton domination.13 The extent to which the pelagic and littoral zones of shallow lakes will respond to warming may differ, particularly when the littoral zone is dominated by macrophytes. For example, the presence of submerged macrophytes can result in stronger stratification and a reduction in mixed layer depth.14,15 Macrophytes can also influence oxygen dynamics and nutrients cycling in both the sediments16,17 and water column,18 changing conditions for GHG production and emission.19−21 Thus, a differential or stronger response in both physical structure and subsequent biogeochemical transformations could Received: December 26, 2015 Revised: May 23, 2016 Accepted: May 24, 2016
A
DOI: 10.1021/acs.est.5b06312 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
determination of chlorophyll-a concentrations (Chl a) following the extraction in methanol32 and quantified by UV−vis spectrophotometry. Time series on the thermal structure of the water column at the pelagic (as in ref 7) and littoral station were obtained with a thermistor chain installed from June to September 2011 and from May to October 2012. The mixed-layer depths were calculated based on density criteria.5 Buoyancy frequency (N, in cycles per hour or cph) describing water-column stability was computed as N2 = g ρ−1 dρ/dz, where g is the gravity constant and ρ the water density. Greenhouse-Gas Concentration and Global-Warming Potential. At each sampling date, lake surface-water concentrations of CO2 and CH4 were collected using the headspace technique through the equilibration of 2 L of water (4 sites in 2011 and 6 sites in 2012) into 20 mL of ambient air during 3 min of vigorous shaking. The headspace gas was injected into helium-flushed preevacuated 5.9 mL sealed Exetainer vials (Labco).33 Gas samples were taken in triplicate (coefficient of variation, CV < 15% for either CO2 or CH4) within 5 min of sample collection and analyzed by gas chromatography (Varian 3800 with a COMBI PAL headspace injection system, a CP-Poraplot Q 25 m × 0.53 mm column and a flame ionization detector). Aqueous N2O concentrations were collected in a similar manner, but the samples were injected into 9 mL pre-evacuated borosilicate vials, and concentrations were determined by gas chromatography using an ECD detector (Schimadzu 2014 GC with a Tekmar 7050 autosampler, and a Poropaq Q 80/100 column to separate gases with 95% argon and 5% CH4 as the carrier). Diffusive GHG flux (in mmol m−2 d−1) at the lake surface was estimated as Fluxd = k (Csur − Ceq), where k is the gas transfer coefficient (cm h−1), Csur is the gas concentration in surface water (mmol L−1), and Ceq is the gas concentration in equilibrium with the atmosphere. The gas transfer coefficient was estimated using a wind-based model,34 where k = 2.07 + 2.15 U1.7 (U is the wind speed at 10 m above the ground) and was corrected via Schmidt (Sc) scaling for a given gas where k = k600 (Sc/600)1/2. Although the Cole and Caraco wind model used in the present study provides a conservative estimate of gas fluxes from the pelagic zone,7 it is probably well-suited to integrate the overall flux of this small and shallow lake,35,36 where around 40% of the surface includes a low-turbulence macrophyte-dominated littoral area.37 The calculated chemical effects38 significantly enhanced CO2 fluxes but only in July and August 2012, when pH was high (up to 8.5). The GHG ebullition flux was measured four times in 2011 and five times in 2012, with four to nine submersible inverted funnels located either in the pelagic or littoral zone of the lake for 4−24 h depending on flux rates. The collected gas bubbles were sampled in triplicate vials (5.9 mL Exetainer) and analyzed by gas chromatography, as described above. The ebullition flux was calculated as Fluxe = C Vg Vm−1 A−1, where C is the concentration of a given gas in the syringe, Vg is the total volume of the gas in the syringe, Vm is the gas molar volume at ambient temperature obtained from the meteorological station, and A is the area of the funnel. The CO2 and CH4 fluxes calculated for the pelagic site 1 (as in Bartosiewicz et al. 2015) as well as pelagial sites 2 and 3, and littoral sites 1−3 (this study only) were used to calculate the open-water GWP index of GHG evading from Lakes Jacques, given in CO2 equivalents. This was done by multiplying fluxes by 1 for CO2, 105 for CH4, and 289 for N2O. These multiplication factors correspond to
be expected in the littoral than pelagic zone of lakes under conditions of warmer climate.22 Summer heat waves have become more frequent over the last few decades23 as a result of global climate change,24 and these events will likely influence the physical structure and subsequent functioning of lakes. For example, a dramatic increase in water-column stability during a heat wave resulted in significantly stronger hypolimnetic oxygen depletion in a large deep lake.25 This observation supports the predictions of climate models that suggest warming will enhance the release of phosphorus (P) from the sediments and result in increased primary production.26 Stronger stratification, higher P availability, and primary production in lakes could also lead to an increased production and storage of more potent GHG such as methane (CH4)27,7 and nitrous oxide (N2O).28 The release of GHG from lakes depends, however, on the patterns and intensity of mixing controlled by wind and heat exchanges.29 Thus, how the increased incidence of summer heat waves will influence stratification and consequently alter nutrient availability, phytoplankton dynamics, and patterns of GHG emissions in small and shallow lakes remains poorly understood. This study consists of a natural experiment that assessed the influence of hot and dry weather during a heat wave30 on the stratification and subsequent impact on nutrient, chlorophyll, and oxygen levels in the pelagic and littoral zones of a shallow productive lake during the open-water season. We compared patterns of CO2, CH4, and N2O concentrations and total GHG flux normalized by the global-warming potential of each gas (hereafter named the GWP index, analogous to the carbon footprint of an ecosystem) in the littoral and pelagial zones during an average rainy cool and an exceptionally hot dry year.
■
MATERIALS AND METHODS Study Site. The small (0.18 km2) and shallow Lake Jacques (mean depth 0.75 m) is located 30 km from Quebec City (Canada; Figure S1). The littoral zone of the lake (0.07 km2 and 39% of the total lake area) is occupied by an invasive macrophyte (Brasenia schreberi, J.F.Gmel). The pelagic zone of the lake is free of floating and submerged macrophytes. Organic matter content of the surface sediment is high in both the littoral (loss-on-ignition: LOI = 45 ± 7%, n = 10, dry sediments combusted at 550 °C for 4 h) and pelagic zones (LOI = 38 ± 5%, n = 10). Water enters from two small inlets, both located in its eastern part, and an underground spring located in its southern part. Physicochemistry. Surface meteorology and physicochemistry was followed in Lake Jacques and described in detail in ref 7. Oxygen concentrations were assessed at the surface and in profile at each sampling site with a YSI probe (600QR, YSI Inc., Yellow Springs, Ohio). Surface water samples were collected at several littoral and pelagic sites in both years (Figure S1). Samples were filtered through prerinsed cellulose acetate filters (0.2 μm pore size; Advantec Micro Filtration Systems, MFS) and analyzed for soluble reactive phosphorus (SRP)31 and dissolved nitrogen (N-NO3−, alkaline persulfate digestion). Nonfiltered water was acidified with H2SO4 (0.15% final concentration) and stored at 4 °C until analysis for total phosphorus (TP, spectrophotometry after persulfate digestion) and total nitrogen (TN, colorimetric method after potassium persulfate digestion in alkaline conditions as nitrate−nitrite by reduction with hydrazine). Water samples (100−500 mL) were also filtered onto GF/F filters (Advantec MFS) for the B
DOI: 10.1021/acs.est.5b06312 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 1. Temporal evolution of dissolved oxygen (DO) stratification in the littoral (upper panels) and pelagic (lower panels) zones of Lake Jacques in 2011 (left panels) and during the heat wave of 2012 (right panels).
Figure 2. Seasonal dynamics of mixed-layer depth (MLD, A-B), rainfall events (arrows indicate rainfall above 10 mm d−1, C-D), and average (±SD) surface water nutrients (E-H) and phytoplankton biomass (Chl-a, I-J) in the littoral (dotted lines) and pelagic (solid lines) zones of Lake Jacques in 2011 and during the heat wave of 2012.
summer heat wave that lasted from June 26 to September 13; for details on surface meteorology in Lake Jacques during the study, see ref 7. Briefly, the air temperature was higher in 2012 than in 2011 (average of 18.5 and 17.5 °C, respectively), with maxima of 33.5 °C compared to 31.0 °C reached in mid July. The lake warmed rapidly early in 2012, with air temperature reaching 28 °C in May as compared to a high of 23 °C for the same month the previous year. The winds over Lake Jacques were generally low, with maximum values of 10 m s−1. The average daily precipitation and wind speeds were both significantly lower during the heat wave of 2012 than in 2011 (respectively, 4.3 versus 8.4 mm d−1 and 2.4 versus 2.9 m s−1; p < 0.01). Despite its shallowness, the lake was stratified in both years, but the hot and dry weather resulted in a much stronger stratification. Indeed the average surface water temperature was 2 °C higher, and the bottom water 2 °C lower during the heat wave versus the average wet year. Buoyancy frequencies (N) were persistently high (50 cph) during summer 2012, indicative of reduced vertical exchange within the water column. In contrast, there was more intensive mixing of the water column
estimations of the climate forcing of, respectively, CH4 and N2O over a 20 year horizon.39,40 Sediment porewater gas samples were also collected in 2012 by in situ dialysis at 1 cm vertical sampling resolution using acrylic peepers prepared following a standard protocol (see the Supporting Information) and deployed 10 cm below and 5 cm above the sediment surface in the littoral and pelagic zones in July and August for 14−21 days. Upon retrieval, the peepers were sampled (3 × 2 mL of porewater from each depth added to evacuated 3.7 mL Exetainers) to obtain three replicate profiles of CH4. Porewater CH4 profiles were used to compute CH4 production and consumption rates in the sediments following numerical methods (see the Supporting Information).41 Fluxes of CH4 at the sediments−water interface were calculated using Fick’s equation42 and temperature-dependent transfer coefficient.43 All statistical analyses were carried out with either Sigma Plot 11 or Prism 6.
■
RESULTS Surface Meteorology and Lake Mixing. The weather in 2012 was hotter and drier than in 2011, particularly during the C
DOI: 10.1021/acs.est.5b06312 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
40 and 60 μg Chl-a L−1 (pelagic and littoral zones, respectively; Figure 2I,J) during the midsummer algal bloom. Thus, significantly more Chl-a was found in surface waters of the lake in 2012 than in 2011 (25.6 compared to 10.2 μg L−1, p = 0.005, Mann−Whitney U-test). Moreover, bloom-forming cyanobacteria such as Anabaena sp. and Aphanizomenon sp. were regularly observed during the hotter summer (up to 40% of the total phytoplankton biovolume; data not shown). GHG Concentrations and Open-Water GWP Index. Patterns in the deviation from saturation (Δgas) for all gases differed between the two years. The most striking difference was the peak in saturation of all three gases at fall overturn in 2012 in both littoral and pelagic zones (Figure S2). Indeed, the values of 172 μM, 11.5 μM, and 26.6 nM for ΔCO2, ΔCH4, and ΔN2O are among the highest surface water concentrations to be observed in temperate lakes. Overall, the average ΔCO 2 of surface waters were significantly higher in 2011 than in 2012 (24.9 and 17.5 μM, respectively; p = 0.001, Mann−Whitney U-test). The lake was continuously supersaturated in CO2 in 2011 but persistently undersaturated during the summer heat wave of 2012. This undersaturation event coincided with strong thermal stratification and high Chl-a concentrations (Figure 2 B, J). In both years, ΔCO2 was negatively correlated to ΔO2, but the relationship was stronger during the heat wave year (Figure S4). On average, ΔCO2 was higher in the littoral zone (littoral = 29.5 μM, pelagic = 5.9 μM), and patterns were more synchronous between zones in 2012. Departure from saturation for dissolved CH4 (ΔCH4) at the lake surface was significantly higher in 2012 than in 2011 (1.61 μM versus 0.74 μM, p = 0.01, Mann−Whitney U-test), but this was due primarily to the much higher concentrations measured at fall overturn. Concentrations were also higher in the littoral than in the pelagic zone in both years (p < 0.0001). The highest dissolved ΔCH4 concentrations were recorded by the end of summer 2012. In both years, ΔCH4 at the lake surface was negatively related to NO3− (Figure S3). Interestingly in 2011, ΔCH4 was negatively related to TP, but in 2012, it was positively related to Chl-a. During the heat wave, CH4 was produced at high rates in both littoral and pelagic shallow sediments (mainly between 0 and 4 cm depth; Figure S5). Profiles suggest that CH4 was partly consumed in situ in the littoral zone during July, whereas it evaded without being oxidized from pelagic sediment. Despite significant increase in sedimentary CH4 concentrations in the pelagial zone between July and September 2012 (p = 0.001, paired t test), there was on average more CH4 in the littoral sediments. For ΔN2O, no significant differences in overall concentrations were observed between years (p = 0.06). However, during the heat wave, N2O concentrations were typically lower, and on occasion, surface waters were undersaturated. Although ΔN2O decreased with increasing NO3− concentrations and increased with surface temperature and oxygen concentrations in 2011 (Figures S3 and S4), no significant relationship with either oxygen or any other environmental variable was observed in 2012. When GHG diffusive fluxes for all three gases were normalized to their GWP, seasonal patterns between zones were more similar in 2011 (CV = 57 and 59% for littoral and pelagic zones, respectively) than in 2012 (CV = 141 and 180%) (Figure 3). Direct and indirect consequences of the heat wave resulted in a major accumulation of GHG over the summer period and their efflux during fall overturn in both the pelagic
in summer 2011, as N often dropped to 30 or 40 cph after a rainstorm. The mixing efficiency (Γ), defined as the ratio of potential energy gain to dissipation, was set at 0.2, and the kinetic energy dissipation rate ε at 10−7 m−2 s−3, allowing us to calculate the vertical distribution of eddy diffusivity44 as Kz = Γ ε/N2. This parameter was used to calculate time scales of mixing, given by τ = L2/ Kz, where L is the mixing depth and Kz is the vertical distribution of eddy diffusivity. In 2011, it took 4 h to mix 0.4 m of water column, whereas it took 15 h for the same amount of mixing in 2012. The mixed-layer depth (MLD) was on average 30% shallower in 2012 (0.46 m) than in 2011 (0.31 m). Oxygen Stratification. Persistent oxygen stratification was observed in both years, but the extent of hypoxic volume was much larger in 2012 (on average 28 339 m3) than in 2011 (7878 m3, p = 0.001, paired Student’s t test). During the heat wave, the oxygen was depleted in the hypolimnion from early May 2012 and persistently low until the mixing event in midSeptember (Figure 1). Given the stability of the water column due to steep temperature gradients, the sporadic rainfall events in July and August 2012 did not disturb this oxygen stratification. Interestingly, macrophyte presence did not result in any apparent oxygen increase in the lower water column of the littoral zone during the summer. There was significantly less oxygen in the hypolimnion during the summer heat wave than during the previous year (respectively, 42 and 72% or 4.8 and 8.2 mg L−1; Wilcoxon test, p < 0.001, n = 28). However, surface oxygen concentrations were higher in 2012 than in 2011 (p < 0.01), with maxima in the pelagic zone reaching about 13.5 mg L−1 (165% saturation). In 2012, surface oxygen levels were also higher in the macrophyte-free pelagic zone of the lake than in its littoral zone (p = 0.03). Nutrients and Phytoplankton Biomass. Seasonal changes in nutrient concentrations appear to be related to the extent of water column mixing, which differed between years (Figure 2 and Table S1). Increases in nutrients tended to coincide with heavy rain events (>10 mm per day, e.g., midsummer 2012). In 2011, TP concentrations followed a similar pattern in the littoral and pelagic zones and were related to water temperature and MLD. Concentrations were higher in summer 2012 (24 compared to 17 μg L−1 in 2011, p = 0.03, Mann−Whitney U-test, n = 28), and no significant correlation with vertical structure was found. TP concentrations varied less in the pelagic zone in 2012 (CV = 33%) than in 2011 (47%) and were more variable in the littoral zone (CV = 54% in 2012 compared to 46% in 2011). Seasonal changes in SRP and NNO3− were markedly different between years and did not clearly follow rainstorm events. In summer 2011, SRP concentrations reached maximum values of 4 or 5 μg L−1 (for pelagic and littoral zones, respectively) and were never depleted below 0.3 μg L−1. In 2012, SRP peaked in the pelagic zone after the initial springtime mixing and then decreased below the detection limit in both littoral and pelagic zones. NNO3− was considerably higher and more variable in 2011 as compared to 2012 (Figure 2 G, H; on average, 0.29 mg L−1 versus 0.09 mg L−1, CV = 105% versus 68%, respectively). Periods of depletion were observed in both years but more so in 2012. In summer 2011, there was no major phytoplankton accumulation in surface waters (Chl-a < 25 μg L−1), and concentrations did not differ between the pelagic and littoral zone. On the contrary, phytoplankton biomass increased steadily after the onset of stratification in June 2012, reaching D
DOI: 10.1021/acs.est.5b06312 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Bubbles were rich in CH4, representing up to 51% of the gas volume, whereas N2O represented only up to 17 ppm; CO2 concentrations were negligible. Given the high CH4 concentrations, this resulted in a significant proportion of overall emissions generated through ebullition (Table 1). Although CH4 contributed most to the GWP index in both years, striking differences in net fluxes and emission pathways were observed. First, estimates for overall emissions were approximately two times higher during the average wet year. Interestingly, we found that ebullition was the dominant CH4 emission pathway in 2011 (64% of total emissions), whereas diffusive fluxes were more important in 2012 (61%). Diffusive CO2 flux contributed more (21%) than diffusive CH4 flux (14%) to GWP index during the average year but not in 2012. Neither diffusive nor ebullitive N2O flux contributed significantly to the GWP index of the lake (
