Toxicity of atrazine was determined for natural phytoplankton communities in lake water samples from Missisquoi Bay (Lake Champlain, CA, US) at monthly intervals from May to August 2005 in order to assess how this herbicide could affect phytoplankton dynamics. Atrazine was added at 0, 5, 20, and 50 μ g L− 1 in 14 L treatments and incubated under simulated ambient light conditions (43 μ mol m− 2 s− 1). Phytoplankton community composition changes in the treatments were followed using a phytoplankton-specific fluorometer (FluoroProbe) and size fractionated filtration and analysis of extracted chlorophyll-a. Phycocyanin-rich Cyanobacteria decreased with increasing atrazine concentration in June and July and increased with atrazine concentration in August. Further experimentation utilizing a fast repetition rate fluorometer revealed that atrazine impacts on phytoplankton autofluorescence can impair FluoroProbe measurements over the short term (1 h to < 7 days).
Missisquoi Bay is a shallow (3 m average depth) embayment located in the northeast corner of Lake Champlain (CA, US). Phosphorus is a pollutant of concern in Missisquoi Bay, with phosphorus levels (1999 to 2003) averaging 45 μg L− 1 (Potamis et al., 2004). Since increasing intensity and frequency of toxic Cyanobacteria blooms (Microcystis) have become common (Watzin et al., 2003), lake managers seek the factors contributing to the Cyanobacteria dominance in this ecosystem. To date, no single factor has been identified as the primary cause of toxic bloom formation in Missisquoi Bay (Mihuc et al., 2006) or elsewhere (Oliver and Ganf, 2000). Given the high concentration of nutrients and herbicides, especially atrazine, in tributaries of Missisquoi Bay, one hypothesis to explain the dominance of Cyanobacteria is the action of herbicides as selective agents.
In the United States and Canada, atrazine is a widely used pre- and post-emergent broad-leaf herbicide that works by suppressing the growth of target weeds by inhibiting the Hill reaction of electron transport during photosynthesis. Concentrations of atrazine and its metabolites (hydroxyatrazine, deethylatrazine, deisopropylatrazine), some of which are persistent and as potent as the parent compound, have been found in surface waters (Müller et al., 1997) and this may affect phytoplankton community composition (deNoyelles et al., 1982; Hamilton et al., 1988).
The objective of this study was to determine if toxicity of atrazine to the phytoplankton community sampled from Missisquoi Bay could be identified as a selective agent for phytoplankton community composition. In these experiments, we used advanced fluorometric instrumentation to assess changes in phytoplankton composition, using a FluoroProbe (bbe Moldaenke, GmbH), and photosynthetic efficiency, estimated using a fast repetition rate fluorometer (FRRF; Chelsea Technologies Group). The FluoroProbe is capable of differentiating among four groups of phytoplankton (i. Chlorophyta and Euglenophyta, ii. Phycocyanin [PC]-rich Cyanobacteria, iii. Heterokontophyta and Dinophyta, and iv. Phycoerythrin [PE]-rich Cyanobacteria and Cryptophyta) based on the fluorescence characteristics imparted on chlorophyll-a (chl-a) by excited accessory pigments unique to each group of phytoplankton. During the course of experiments we suspected that atrazine was imparting an effect on phytoplankton that compromised the use of a FluoroProbe. The FRRF was used to assess the impact of atrazine on photosynthetic efficiency by measuring the ratio of variable to maximum chl-a fluorescence and assisted in establishing an important caveat regarding the use of the FluoroProbe for toxicological work using atrazine.
Atrazine toxicity tests
Near shore water from Missisquoi Bay (44°58.295′ N, 73°12.657′ W) was collected on May 17, June 21, July 26, and August 11, 2005 between 09h:00 and 10h:00 from a depth of 0.5 m into a 210 liter opaque food-grade polyethylene drum and transported within 1.5 hours to the laboratory. In June and July, 2.5 L of non-filtered lake water was transported on ice in a critically cleaned amber glass bottle to the Vermont Agency of Agriculture, Food, and Markets (VAAFM) to determine herbicide concentrations. Lake water (14 L) was placed into buckets double-lined with clean, clear, food-grade polyethylene plastic bags. Size fractionated chl-a concentrations were determined at the beginning and end of the experiment (see below). Atrazine (2-chloro-4-ethylamino-6-isopropylamino-s-triazine; Chem Service, West Chester, PA, USA; 98% purity, carried in methanol) was added to establish 4 treatments (in triplicate) of 0 (control), 5, 20, and 50 μ g L− 1. The buckets were randomly placed in 3 rows under controlled illumination consisting of an equal mix of white fluorescent bulbs (type F40SP65; General Electric) and wide spectrum fluorescent bulbs (type F40PL/AQ; General Electric) that provided an average photon flux density during the experiment of 43 μ mol photons m− 2 s− 1 and a 15:9 light:dark cycle. Air temperature in the laboratory ranged between 21 and 22°C. During the July experiment, water samples were sent to the VAAFM to measure concentration of atrazine in the treatments.
Missisquoi Bay water collected in June was filtered (< 0.45 μ m) and used to calibrate the FluoroProbe (Model II, Series 7; bbe Moldaenke, GmbH, Kiel-Kronshagen, DE) for colored dissolved organic material characteristic of Missisquoi Bay. All other variables were the factory settings for the instrument. In situ fluorescence measurements took place after mixing the contents of each replicate thoroughly; measurements were made at 3-second intervals over 90 seconds in dim light.
Extracted chlorophyll-a determination.
At the first measurement and again at the end of the 7 days, 200 mL, 150 mL, and 50 mL were filtered in triplicate through 20-μ m, 2-μ m, and 0.2-μ m pore-size 47 mm-diameter polycarbonate membrane filters (Isopore™; Millipore Corporation) under low vacuum pressure (< 15 mm Hg). Chlorophyll-a was measured fluorimetrically (TD-700 fluorometer; Turner Designs, Sunnyvale, CA) using the non-acidification fluorometric technique (Welschemeyer, 1994).
Brandy Brook experiments
Water was collected (02 August 2006) at the mouth of Brandy Brook (44°52.500′ N, 75°09.680 W), a tributary of the St. Lawrence River. This site was selected because of its limnological similarities to Missisquoi Bay (high nutrient levels, eutrophic, brown water). We sought to investigate further certain aspects of this study; namely, the response of photosynthetic efficiency and the response of the FluoroProbe to atrazine addition. Fifteen liters were placed into two separate treatments prepared in the same manner as in the Missisquoi Bay atrazine toxicity tests (above). In addition, a 500 mL subsample was placed inside an acid clean glass BOD bottle, and the bottle was returned to the bucket for incubation. The FluoroProbe was placed in the treatment vessel that was shrouded in an opaque black plastic bag and readings were made continuously every three seconds for approximately 20 minutes, while photosynthetic efficiency was concurrently measured using a FRRF every five minutes on aliquots that were removed from the treatment. For the FRRF measurements, water from the treatment was transferred into the dark chamber of the FRRF. At 20 minutes, measurements paused to allow for the addition of atrazine to achieve a total concentration of 50 μ g L− 1, and the addition of the photosystem (PS) II inhibitor DCMU (3-(3,4-dichlorophenyl)-1,1-dimethylurea; Sigma-Aldrich, St. Louis, MO, USA; 98% purity) to the 500 mL glass bottle (final concentration was 70 mg· L− 1) for use as a control to validate the methods. Measurements made by the FluoroProbe continued for another 100 minutes, while alternating atrazine and DCMU measurements were made using the FRRF every five minutes. The experiment was repeated with an atrazine concentration of 1000 μ g L− 1.
The phytoplankton community varied among the sampling dates; however, throughout the summer the division encompassing the Heterokontophyta and Dinophyta was consistently a major contributor to total in situ chl-a (Figure 1). Atrazine and metolachlor, and their metabolites, were the most prevalent herbicides detected (> 0.020 μ g L− 1; Table 1) in Missisquoi Bay in June and July. On the day of sampling for June, the Vermont Department of Environmental Conservation sampled a mid-bay station (45°00.800' N, 73°10.430° W) that was analyzed along with the near shore sample. Acetochlor, a chloroacetanilide herbicide much like metolachlor, was not found; however, its active metabolite acetochlor ESA was present. In the experimental treatments (June experiments), atrazine levels lowered over the course of the experiments by 0.430 ± 0.305 μ g L− 1 in the 50 μ g L− 1 treatment, but was accounted for as the active atrazine metabolite hydroxy atrazine.
May 2005 – Atrazine toxicity experiment
When the experiment started the total chl-a concentration in Missisquoi Bay was approximately 9 μ g L− 1 (Table 2). A dose-response relationship was evident in this experiment (Figure 2A), due primarily to the large contributions to the total chl-a data from the Chlorophyta, and the Heterokontophyta and Dinophyta. PC-rich Cyanobacteria were not present at detectable levels (< 0.1 μ g L− 1), and phytoplankton in the mixed group of PE-rich Cyanobacteria and Cryptophytes disappeared in all treatments, including the control, within 2 days. Based on size fractionated chl-a determinations, nanoplankton were more abundant at the beginning of the experiment for all treatments including the control, but by the last measurement (day 7) picoplankton dominated over the other two size classes and comprised an average of 51% of the total chl-a for all treatments including the control. There was very little change in the community composition at all treatment levels after 7 days (Figure 3).
June 2005 – Atrazine toxicity experiment
The FluoroProbe detected a considerable fraction of PC-rich Cyanobacteria in June and the concentration of this group decreased slightly with increasing atrazine treatment (Figure 4) whereas the Heterokontophyta and Dinophyta increased (Figure 3). A dose-response was not seen among atrazine concentrations for the Chlorophyta and Euglenophyta.
July 2005 – Atrazine toxicity experiment
As observed in June, the biomass of PC-rich Cyanobacteria decreased with increasing atrazine treatment (Figure 4), whereas the Heterokontophyta and Dinophyta biomass increased (Figure 3). Cryptophyta and PE-rich Cyanobacteria made up a negligible amount of the total in situ chl-a. Size-fractionated chl-a data showed relatively equal amounts of microplankton and nanoplankton, with smaller amounts of picoplankton. At the end of the experiment, picoplanktonic chl-a increased with increasing atrazine concentrations, while nanoplankton showed an inverse relationship with atrazine concentrations (Table 2).
August 2005 – Atrazine toxicity experiment
Total chl-a at the sampling site in Missisquoi Bay was 20 μ g L− 1 (Table 2). All phytoplankton groupings showed a dose-response trend of in situ chl-a in response to atrazine additions yet only the Heterokontophyta and Dinophyta had the treatments (0, 5, 20, and 50 μ g L− 1) significantly different from each other (one-way ANOVA, p < 0.05) at the endpoints (Figure 3). In contrast to the June and July experiments, the PC-rich Cyanobacteria increased with increasing atrazine concentrations (Figure 3), at the apparent expense of the Heterokontophyta and Dinophyta. Size-fractionated chl-a revealed that the microplankton dominated throughout the experiment (Table 2).
August 2006 – Brandy Brook atrazine experiments
The Missisquoi Bay experiments suggest that atrazine caused an increase in phytoplankton autofluorescence at the beginning of assays (data not shown) and this caused the FluoroProbe to over-estimate in situ chl-a concentration. This effect was absent by the end of the experiment, as demonstrated by the high degree of correlation between extracted chl-a and in situ chl-a (slope = 0.94, r2 = 0.88; Table 2). In order to further assess the utility of the FluoroProbe for use in these ecotoxicology tests with PS II inhibitors such as atrazine, water from Brandy Brook was tested.
As illustrated in Figure 5, it appears that in situ chl-a increased upon addition of atrazine. A similar observation was made with a 1000 μ g L− 1 addition (data not shown). Both the 50 and 1000 μ g L− 1 experiments started at approximately 5 μ g L− 1 of in situ chl-a and ended two hours later around 8 μ g L− 1. The increase was attributed to the FluoroProbe-specific phytoplankton groupings of Chlorophyta and Euglenophyta and PE-rich Cyanobacteria and Cryptophyta (Figure 6); the phytoplankton were physiologically affected by atrazine and this was detected as an apparent increase in their in situ chl-a fluorescence. The FluoroProbe interpreted the increased fluorescence as an increase in the biomass (using chlorophyll-a as a proxy) of the particular algae.
The FRRF experiment on the 50 μ g L− 1 atrazine treatment showed that the Brandy Brook phytoplankton community had a photosynthetic efficiency (Fv/Fm) of 0.44 ± 0.01 at the onset of the experiment. Both the atrazine and DCMU treatments significantly (t-test, p < 0.0001) decreased Fv/Fm (Figure 4 and Figure 5; t-test: atrazine t232 = 91.5, p < 0.0001; DCMU t280 = 406.9, p < 0.0001); the photosynthetic efficiency for phytoplankton exposed to atrazine fell to 0.34 ± 0.01, while DCMU caused it to plummet to –0.01 ± 0.01. In the experiment that treated the Brandy Brook water with 1000 μ g· L− 1, the Fv/Fm at the beginning was a comparable 0.45 ± 0.06, with a decrease to 0.36 ± 0.02 for atrazine, and –0.01 ± 0.02 for DCMU (t-test, p < 0.0001; atrazine t265 = 17.9; DCMU t262 = 91.9). Neither the atrazine treatments between the two experiments nor the DCMU treatments were significantly different (t-test, p > 0.05; atrazine t318 = 1.8; DCMU t363 = 1.9), showing that atrazine impact on phytoplankton was saturated at 50 μ g L− 1.
There are seasonal variations in the effects of atrazine on the phytoplankton community from Missisquoi Bay. Atrazine affected the separate divisions and the overall abundance of extracted chl-a to a different extent each month. The August experiment provides evidence of atrazine acting as a selective factor for increasing the concentration of PC-rich Cyanobacteria. However, the concentration of added atrazine that caused this increase in PC-rich Cyanobacteria is an order of magnitude greater than atrazine levels found in Missisquoi Bay. The possibility that other herbicides could be acting in synergy with atrazine is possible; however, laboratory assays show that new generation herbicides (e.g. flumetsulam, dimethenamid, mesotrione, nicosulfuron, rimsulfuron) present in lake water at far lower levels than atrazine, are far less toxic (Zananski and Twiss, unpublished data).
Intra-seasonal changes in phytoplankton community structure may affect the response to atrazine due to variable sensitivities to atrazine amongst species, even within the same phylum (Huber et al., 1993; Solomon et al., 1996; Weiner et al., 2004). For example, using microscopy, we noticed that Asterionella was the most dominant genus of diatom present in August. Single species toxicity tests have determined that this diatom is extremely sensitive to atrazine, with 5 and 10-day EC50 concentrations between 2 and 10 μ g L− 1 (Bérard et al., 1999). The dominance of a sensitive species may explain the extreme population crash in Heterokontophyta and Dinophyta in the 50 μ g L− 1 treatment in August. Another explanation for the monthly changes is that some of the members of the community, if present in the specific month sampled, may be able to gain tolerance over the course of the weeklong experiment, resulting in little to no apparent inhibition due to atrazine. Increased community tolerance and recovery has been recorded in several other studies (deNoyelles et al., 1982; 1989; Fromm, 1986; Detenbeck et al., 1996; Bérard et al., 2003), although in our study we are unable to prove the role of tolerance. Over the course of a season, natural species succession could change the community response to atrazine exposure.
The Brandy Brook experiments showed that the increase in fluorescence lasted for over an hour and a half, with little decay of the effect over the observed time period (Figure 4 and Figure 5). The extracted chl-a measurements made after 7 days in each experiment (Table 2) correlated very well with the levels detected by the FluoroProbe, suggesting that the FluoroProbe was accurately measuring phytoplankton community composition at the end of the experiments.
The primary objective here was to determine whether atrazine addition to a natural phytoplankton community would manifest with certain divisions; in particular the PC-rich Cyanobacteria, being more resistant to the herbicide and thus their abundance would increase in proportion with atrazine additions. With confounding factors attributed to temporal (monthly) fluctuations, we were unable to conclude if any divisions were indeed more resistant. A secondary objective was to evaluate the FluoroProbe as an efficient and effective tool for community toxicity studies. It was found that addition of a PS II inhibitor will increase the fluorescence emitted by certain divisions of phytoplankton following excitation by actinic light. The FluoroProbe was determined to be sensitive to the effect that PS II inhibition had on the Chlorophyta and Euglenophyta and PE-rich Cyanobacteria and Cryptophyta. Because of the tendency for the FluoroProbe to overestimate the chl-a after the addition of a PS II inhibitor, it may not be a suitable tool for acute exposure experiments. However, enhanced chl-a fluorescence can be used as a means to assess toxicity of PS II inhibiting chemicals such as atrazine in phytoplankton (Fai et al., 2007).
This research was funded by the Lake Champlain Basin Program, Grand Isle, VT. We thank Nathaniel Shambaugh at the Vermont Agency of Agriculture, Food and Markets for providing chemicals and conducting the herbicide analysis. This is contribution 351 of the Clarkson University Center for the Environment.