Arch. Environ. Contain. Toxicol. 25, 95-101 (1993)

A R C H I V E S

OF

Environmental Contamination n d Toxicology © 1993 Springer-Verlag New York Inc,

Effects of Diquat on Freshwater Microbial Communities A. L. Melendez, R. L. Kepner, Jr., J. M. Balczon, and J. R. Pratt 1 School of Forest Resources, The Pennsylvania State University, University Park, Pennsylvania 16802, USA

Abstract. A static microcosm system was used to evaluate effects of the herbicide diquat (0.3-30 mg/L) on the structure and function of naturally derived microbial communities on polyurethane foam substrates. Microbial communities were exposed to a single application of diquat and were monitored for 21 days. Effects on community structure included changes in algal cell density at diquat concentrations/> 0.3 mg/L (after an initial decrease in net productivity), bacterial cell density (1 mg/L diquat), and increased biomass accumulation (10 and 30 mg/L diquat). The species richness of protozoa was reduced at diquat concentrations > 0.3 mg/L; protozoan species composition was progressively more dissimilar with diquat treatment. Effects on community function included inhibition of net productivity and community respiration (10 and 30 mg/L diquat), and decreased enzyme activities [alkaline phosphatase (1, 10, and 30 mg/L diquat), electron transport system (i> 0.3 mg/L diquat), and [3-glucosidase (i> 0.3 mg/L diquat)]. Both photosynthetic and nonphotosynthetic organisms were affected by diquat. Most structural and functional responses were sensitive indicators of stress. Estimated chronic toxicity values ranged from 0.3 mg/L (day 3) to 5.5 m g / L (day 21). Most microbial responses indicated that microbial community structure and function did not recover within the 21-day exposure period.

The bipyridylium herbicide diquat (6,7-dihydropyrido-[1,2a:2', 1 'c]pyrazinediium dibromide) is used widely in the control of aquatic weeds. Diquat interferes with electron transport in photosystem I (Jager 1983). In natural waters, diquat dissipates from the water column largely due to adsorption by sediments, suspended particles, and aquatic plants and through absorption by aquatic plants and algae (Summers 1980). Ultimately, most of the diquat applied to the water accumulates in the sediment either directly or after release from decaying plant and animal matter (Birmingham and Colman 1983). Adsorption tends to reduce the phytotoxicity of diquat (Simsiman et al. 1976). Diquat appears not to be directly toxic to many species of fish and invertebrates when applied at recommended rates for the control of aquatic weeds (0.5-3 mg/L, Simsiman et al. 1976).

IAddress correspondence to Dr. J. R. Pratt, 7 Ferguson Building, The Pennsylvania State University, University Park, PA 16802.

At these levels, however, microbial communities may be affected (Pratt et al. 1990; Cragg and Fry 1984; Birmingham and Colman 1983; Ganstad 1982; Cullimore 1975; Breazeale and Camper 1972). This study examined the effects of diquat on the structure and function of naturally derived microbial communities collected on polyurethane foam substrates (PFs). The diquat concentrations used (0.3-30 mg/L) are toxic in singlespecies toxicity tests and include the range of field use concentrations. The communities comprised bacteria, algae, fungi, protozoa, and micrometazoa. Microbial communities on PFs were studied for 21 days in a static microcosm system following a single application of diquat. The objectives were to evaluate diquat effects on community structure and function by maximizing exposure to the herbicide, and to determine to what extent recovery of microbial responses occurred after a single application of diquat. The structural responses measured were algal cell density, bacterial cell density, and total biomass (protein content). The functional responses measured were, alkaline phosphatase activity (APA), electron transport system activity (ETSA), and [3-glucosidase activity ([3-glcA). These enzymes are ubiquitous in microbial communities, are important in nutrient cycling, and have previously been used in the evaluation of toxic stress (Burton and Lanza 1984). Other functional responses measured in microcosms were dissolved oxygen (DO), pH, net productivity, and community respiration.

Methods Microcosms

A laboratory toxicity experiment was conducted under static conditions in high density polyethylene tanks (35 × 28 × 15 era). Polyethylene was used to minimize diquat loss by adsorption to glass. Water (7 L) from a local pond (Linden Hall, Centre County, PA) were added to each of 18 replicate microcosms. Pond water was passed through a plankton net (63 Ix mesh) to remove large detritus and some plankton. Mean pH of the pond was 9.1, and alkalinity and hardness were 37 nag CaCO3/L and 39 mg CaCO3/L, respectively. Microcosms were covered with clear plastic sheeting to minimize evaporation. Water lost to evaporation during the experiment was replaced by the addition of distilled-deionized water. Lighting was provided by daylight-equivalent lights (Vita-Lites, Durotest, Inc.) placed above the test system to provide an intensity of approximately 5,000 lux. Lights were set on a 12 h light: 12 h dark photoperiod. Water temperature was not controlled

A. L Melendez et al.

96

but was measured daily. The mean water temperature during the 25day acclimation period was 23.8°C (range 18.7-28.6°C). The mean water temperature during the 21 day post-treatment period was 22.4°C (range 18.9-27.5°C). The source of organisms for the experiment was a natural microbial community collected on PFs (4 × 5 × 6 cm). The PFs were placed in the same pond used as a source of water for the microcosms. Pond organisms were allowed to colonize the PFs for 4 weeks. The PFs were placed in pond water in a clean container, brought to the laboratory, and transferred to the microcosm units. Each PF substrate was attached with a loop of cotton line to hooks on the bottom of each microcosm. Five PFs were placed in each microcosm. Triplicate microcosms were randomly assigned to five diquat treatments and control for subsequent dosing (18 microcosms total). The dissolved oxygen (DO) content in the water of each microcosm was monitored daily at lights-off (1900) using a YSI Model 58 meter and oxygen probe which was calibrated prior to each use. Temperature and pH measurements were also made at the time of DO readings. Both DO and pH readings were used as the criteria for microbial acclimation to laboratory conditions. Acclimation proceeded for 25 days until similar and stable readings were obtained among microcosms. Measurements of DO, pH, and temperature were taken daily throughout the posttreatment portion of the experiment. On day 26, one substrate was removed from each microcosm. Three PFs were randomly selected from the 18 PF samples. These PFs were destructively sampled to remove as much of their contents as possible into separate, sterile, plastic cups. The three samples were analyzed for structural and functional responses to obtain an estimate of microcosm conditions at the time of dosing. Triplicate microcosms were dosed (day 26) with a single application of diquat (Valent Diquat H/A, Valent USA Corp., 35.3% active ingredient) at nominal concentrations of 0, 0.3, 1, 3, 10, and 30 mg/L diquat. Concentrations of diquat in microcosms were determined spectrophotometrically after reduction with sodium dithionite (Baker 1984; Pratt et al. 1990). Diquat concentrations in microcosms were determined immediately after dosing and approximately every other day thereafter. Following dosing, one PF from each microcosm was destructively sampled after 3, 7, 14, and 21 days. Structural and functional microbial community responses were assayed as follows.

Structural Responses Biomass: Protein content was used as an estimate of total biomass in PF substrates. Protein was extracted (Rausch 1981) and determined by the method of Bradford (1976). Chlorophyll could not be determined because of interference by diquat.

et al. (1979). Subsamples of live material were removed from samples and systematically scanned using brightfield microscopy. Taxa encountered were identified to the lowest practical taxonomic level, usually genus and often species. Subsampling continued until the number of identified taxa became asymptotic. The number of taxa was used as a measure of community richness, and species identities were also used in subsequent analyses.

Functional Responses Enzyme Assays: Subsamples (0.5-1.0 ml) were concentrated by centrifugation at 2,700 × g for 10 min. Alkaline phosphatase activity was determined by a modification of the method of Sayler et al. (1979). Subsamples were not sonicated, and incubation was done at room temperature. Activity was expressed as nmole p-nitrophenol/ml sample/h. [3-glucosidase activity was determined by a modification of the method of Morrison et al. (1977). Subsamples (2-4 ml) were concentrated by centrifugation, resuspended in 1 ml of 25 mM NaHCO 3, and incubated at room temperature with the substrate p-nitrophenyl-[3-Dglucoside (1 mM). Chloroform (50 p~l) was added to the assay as a bacteriostatic agent. After six hours, the reaction was stopped with 0.25 ml of 1 N NaOH. Electron transport system activity was measured by a modification of the method of Johnson (1986). Subsamples (8-12 mL) were concentrated by centrifugation. The reaction mixture consisted of 2.5 ml concentrated subsample, 0.5 ml of 0.1% sodium citrate, and 1 ml of 0.2% 2-(p-iodophenyl)-3-(p-nitrophenyl)-5-phenyl tetrazolium chloride (INT). The mixture was incubated in the dark at room temperature for 1-2 h. The reaction was stopped with 0.5 mL of 37% formaldehyde. Samples were centrifuged, and the resulting red colored INTformazan in the pellet was extracted with 5 ml methanol. INT-formazan produced by ETSA was determined from a standard curve of INT-formazan dissolved in methanol (5-42 nmol/ml). Activity was expressed as nmol INT-formazan/ml sample/h.

System-Level Responses: Dissolved oxygen and pH readings were taken daily at lights-off. Dissolved oxygen readings were taken weekly at both lights-off and lights-on to estimate net productivity and community respiration by the three-point DO method of McConnell (1962). Because microcosms were covered with plastic sheeting, an assumption was made that minimal gas diffusion occurred between the atmosphere and microcosms. Therefore, productivity and respiratory values were not corrected for diffusion.

Dam A n a ~ s ~ Algal and Bacterial Enumeration:A 2 ml subsample was fixed in 3% buffered glutaraldehyde and refrigerated until analysis. Separate subsamples were used for the enumeration of algae and bacteria. Algal cells were counted in a Palmer-Maloney cell (0.1 ml) as (1) green algae (subdivided into unicellular, colonial, and filamentous, (2) cyanobacteria (subdivided into colonial and filamentous), and (3) diatoms. At least 250 algal units (i.e., cells, colonies, filaments) were counted in at least 20 random fields at a magnification of 250×. This number of algal units and random fields was considered as the minimum needed to obtain repeatable estimates of algal density after analysis of saturation curves (units/mt vs. number of counts). Bacteria were enumerated by a modification of the epifluorescent direct count method (Porter and Feig 1980) following dispersion with sodium pyrophosphate in an ultrasonic bath (Velji and Albright 1986). At least 200 cells were counted in at least 20 random fields at a magnification of 1,250×.

Protozoan Species Richness: Protozoan species richness was determined prior to dosing and on days 7 and 21 using the method of Cairns

Response variables were tested for equality of variances by residual analysis and Hartley's test (Sokal and Rohlf 1983). Total algal and unicellular green algal counts showed unequal variances; therefore, the data from these variables were log-transformed. All microbial community responses were compared for each sampling data using analysis of variance (ANOVA). When significant differences were detected by ANOVA (c~ = 0.05), Dunnett's test (Dunnett (1955) was used to determine which treatments differed significantly from the control (a = 0.05). Differences in species composition (presence-absence) were tested using a randomization method and Jaccard's coefficient of similarity (Pratt and Smith 1991). Temporal changes in system-level responses measured in microcosms (DO, pH, net productivity, and community respiration) were monitored by plotting responses against time. The 95% minimum significant difference (MSD) was plotted around mean control values, after calculation using Dunnett's procedure. Treatments with mean values outside this confidence band were interpreted as significantly different from controls.

Diquat Effects in Microcosms

97

Table 1. Concentration of diquat in the water column of microcosms (mg/L). For the initial and final measured concentrations, means of three replicates are shown with standard deviations in parentheses. The slope and intercept of the decay curve are also shown Nominal 0 0.3 1 3 10 30

InitiaP

Finalb

Slope

NDc 0.46 (O.O9) 1.36 (0.12) 4.28 (0.13) 13.97 (O.2O) 40.61 (1.11)

ND 0.18 (O.O6) 0.59 (0.10) 2.89 (0.07) 10.39 (O.O7) 33.95 (0.45)

--0.008

Unicellular Green Algae 10 (1 0 6 cells/ml)

Intercept -0.351

-0.031

1.21

-0.043

3.83

-0.135

13.40

-0.242

39.70

1 Day 3 Day 7 -8-

Day 14 Day 21

0.1

0

a Concentration measured immediately after application b Concentration measured 22 days after application c Non-detectable at 0.05 mg/L

For each measured response, a no-observable-effect concentration (NOEC) and a lowest-observable-effect concentration (LOEC) were determined. The NOEC was the highest diquat concentration not significantly different from control, and the LOEC was the lowest diquat concentration significantly different from control. The maximum allowable toxicant concentration (MATC, analogous to a chronic toxicity value) was calculated as the geometric mean of the NOEC and LOEC. Diquat concentrations used in estimating NOEC, LOEC, and MATC values were obtained from the linear regression of degradation curves.

Results and Discussion

Diquat Concentration Diquat dissipated linearly from the water column of microcosms (Table 1). Herbicide in the water was persistent in the absence of major sinks such as plants and sediment. The concentration of diquat in the water from PFs did not differ from that in the microcosms (data not shown). For simplicity, the nominal diquat concentrations are used in subsequent discussions, except when discussing MATCs.

Structural Responses Algal Cell Density: The algal community in controls was dominated by green algae (82% average relative numerical abundance), followed by diatoms (12%), and blue-greens (6%). Unicellular green algal density significantly increased relative to control in the 10 and 30 mg/L treatments on days 14 and 21 (Figure 1). This increase corresponded to a bloom of Oocystis spp. which, on day 14, accounted for > 90% and > 98% of total algal counts in the 10 and 30 mg/L treatments, respectively. On day 21, Oocystis spp. accounted for > 98% of total algal abundance in both treatments. Unicellular green algae were relatively resistant to diquat. The dominant genera were Oocystis, Characium, Ankistrodesmus, Cosmarium, and Staurastrum. These remained dominant

0.1

1 Diquat (mg/L)

10

100

Fig. 1. Effect of diquat on density (cells/ml) of unicellular green algae sampled from PF substrates in laboratory microcosms

in all treatments except at 10 and 30 mg/L diquat, on days 14 and 21, where only Oocystis predominated. Although no attempt was made to enumerate each algal species present, results in this study compare, at least qualitatively, with those of single species toxicity tests. Cullimore (1975) found that some unicellular green algae were resistant to diquat. For example, the growth of Chlorella pyrenoidosa, C. ellipsoidea, Coccomyxa subellipsoidea, and Stichococcus bacillaris was unaffected by 10 mg/L diquat. Chlorella vulgaris had a chlorophyll a-based 3-d ECso > 2.94 mg/L (Philips et al. 1992). Diquat affected other algal forms. By day 21, green filamentous algal density had significantly increased (13-19-fold) from control values at 1-10 mg/L diquat, after an initial decrease in numbers. Numbers of green colonial algae had significantly decreased from control values (41-100%) on days 14 and 21. Similarly, by day 14, diatom densities had decreased significantly from controls (54-74%) at 0.3, 10, and 30 mg/L diquat. Previous work has shown that the diatom Navicula sp. is strongly inhibited by diquat; Philips et al. (1992) showed a 75% reduction in chlorophyll a after 3 d at 0.03 mg/L. Their predicted ECso of 0.019 mg/L contrasts with the lack of inhibition observed for N. pelliculosa at 0.3 mg/L by Birmingham and Colman (1983). Cyanobacteria were also sensitive to diquat. By day 3, the concentration of filamentous cynaobacteria was significantly lower than control at 3 mg/L diquat. By day 14, cyanobacteria were either absent or few ( < 400 units/ml; >t 83% reduction) in all treatments. Birmingham and Colman (1983) also found that cyanobacteria were sensitive to diquat. The growth of Anabaena flos-aquae and Anacystis nidulans was inhibited after a 2-7 day growth period in the presence of 0.3 mg/L diquat. In contrast, the growth of Chlorella vulgaris, a unicellular green alga, was unaffected by 3 mg/L diquat. Ganstad (1982) also found that, for the species tested, cyanobacteria were more sensitive to diquat than green algae. Both green algae (i.e., Chlorella pyrenoidosa and Chlamydomonas reinhardi) and cyanobacteria (i.e., Nostoc musorum, Anacystis nidulans, and Anabaena variabilis) were completely inhibited by 0.276 mg/L diquat. However, the growth of cyanobacteria was inhibited

A. L Melendez et al.

98 Taxa

Table 2. Effect of diquat on protozoan community similarity (Jaccard's coefficient) in microcosms treated with diquat

5O

Diquat (mg/L)

30'

Day 7 - ~ - Day 21

20 i i

-~-

Pretreatment

Control 0.3 1.0 3.0 10.0 30.0

Within community similarity

Similarityto control

day 7

day 21

day 7

day 21

0.52 0.55 0.44 0.48 0.49 0.42

0.54 0.59 0.52 0.53 0.35 0.20

-0.34 0.31 0.24 0.18 0.11

-0.36 0.20 0.27 0.17 0.12

1

~0/~ i iJiJll~ i JlmJliH i iiJllJll J t Jl~l~l= 0.1 1 10 100 Diquat (mg/L)

Fig. 2. Effect of diquat on protozoa species richness in laboratory microcosms. Data are shown for pretreatment samples and for samples taken on days 7 and 21 of the experiment

16"

Protein (ug/mL)

14 t2 - e - Day 3 Day 7 Day 14 Day 21

10

5-15% by 0.05 mg/L diquat. Philips et al. (1992) also found that the cyanobacteria Anabaena flos-aquae, Microcystis aeruginosa, and Lyngbya wallei were sensitive to diquat, all having 3-d ECso values < 0.15 mg/L. Their results indicated a wide range of susceptibility to diquat among the ten species tested.

8

O 4

2 - / / i

Bacterial Cell Density: Bacterial cell density at 1 mg/L diquat significantly increased from control values by 1.6-fold on day 7. By day 21, bacterial numbers in this treatment had decreased 35% below control values. A concentration-response effect [i.e., decrease in bacterial cell numbers with increasing diquat concentration] was not observed, despite diquat's antimicrobial properties (Summers 1980). Increased bacterial density may have resulted from increased nutrient supplies after diquat addition. Changes in [3-glcA and APA indicated the alteration of nutrient cycling, because the activity and synthesis of these inducible enzymes affect and are affected by nutrient levels. Significant amounts of nutrients are probably released from microorganisms affected by diquat. Others have reported increases in bacterial density after application of low levels (around 1 mg/L) of diquat or paraquat to microcosms. Cragg and Fry (1984) treated microcosms containing sediments and macrophytes with 1 mg/L diquat. They observed a significant increase in bacteria, which was maximal 7 days after herbicide addition. Ferebee and Guthrie (1973) observed an increase in bacteria after paraquat addition (0.7 and 1.4 rag/L) to a laboratory culture system containing lake water. They speculated that the increase in bacterial density was a result of an increased nutrient supply due to algal death. Protozoan Species Richness: Approximately 40 species of protozoa were present on pretreatment samples. On both days 7 and 21, there was a significant linear relationship between increasing diquat and decreasing protozoan species richness (Figure 2), with protozoan species richness falling to about 10 species in the highest treatment. There was no evidence of recovery between day 7 and day 21 in all diquat treatments > 0.3 mg/L.

0

~

~lll,lt

~

0.1

i

iJlllll

i

i

1 Diquat (mg/L)

PIINH

J

10

I

JIIHI

100

Fig. 3. Effect of diquat on protein content (ixg protein/ml) of PF substrates in laboratory microcosms Previous studies have shown that diquat reduces protozoan richness. Pratt et al. (1990) showed that similar diquat treatments (0.1-38 mg/L) caused a linear decrease in protozoan richness and that only bacterivorous species remained in the highest treatment. As in this experiment, the reduction in species richness was approximately 75% at 38 mg/L. The similarity of communities to controls decreased with diquat treatment (p < 0.001, Table 2). Control communities had similarities (Jaccard coefficient) of 0.52 on day 7, and similarity decreased to 0.42 in the highest treatment. By day 21, control community similarity was 0.54, but similarity in the highest treatment was only 0.20. Multiple comparisons of community similarity showed that the two highest diquat treatments (10 and 30 mg/L) were consistently different from all other treatments.

TotalBiomass: On day 21, the protein content in the 10 and 30 mg/L treatments increased significantly from control by approximately two fold (Figure 3). This increase in protein was most likely due to the bloom of Oocystis spp. that occurred in these treatments. A concurrent increase in bacterial numbers was not observed.

Functional Responses Enzyme Activities: On days 14 and 21, APA at 1, 10, and 30 mg/L diquat was significantly reduced by 35-74% compared to

Diquat Effects in Microcosms

99

ETSA (% control)

A P A (% control)

110-

140

100 120 90100 -e-

Day 3 Day 7

8O

-E}-

Day 14

--W--

Day 21

80"

-~-

Day 3 Day 7

70" Day 14 --W--

60"

Day 21

60 5040

20

40"

--~/~ 0

~ i llllN 0.1

J I I I~H. 1

J J I JlIHI 10

I

P I Jllll 100

30"-H 0

i

i ii;11, 0.1

i iJllNi

i 1

i illlll~ 10

J J ~11111 100

Diquat (mg/L)

Diquat (mg/L)

Fig. 4. Effect of diquat on alkaline phosphatase activity (APA) from PF substrates in laboratory microcosms. APA units are nmol p-nitrophenol/ml sample and are expressed as percent of control response

i

Fig. 5. Effect of diquat on electron transport system activity (ETSA) from PF substrates in laboratory microcosms. Data are expressed as percent of control response

BglcA (% control)

control (Figure 4). Inhibition of APA at 10 and 30 mg/L diquat coincided with the bloom of Oocystis. Actively growing algae excrete photosynthetic products including readily utilizable dissoved organic matter (UDOM) that may inhibit the activity and repress the synthesis of inducible, extracellular enzymes such as alkaline phosphatase and 13-glucosidase (Chrost 1990). In addition, high amounts of polymeric DOM (e.g., polysaccharides, protein, etc.) are released from the lysis of senescent algae (Chrost 1989). These substrates may induce the synthesis of extracellular enzymes (Chrost 1990). Leakage of polymeric DOM from dying algae probably occurred at/> 3 mg/L, resulting in APA stimulation in the 3 mg/L treatment. However, stimulation of APA at 10 and 30 mg/L may have been opposed by high levels of UDOM excreted by the blooming Oocystis. Inhibition of APA at 1 mg/L remains unexplained unless some UDOM was released from cells even though no mortality was apparent. However, the regulation of extracellular enzyme activity is complex, being a function of environmental conditions (e.g., pH, temperature, oxygen level) and substrate concentration. Enzyme activity may also be influenced by growth stimulation or inhibition of microorganiams that produce extracellular enzymes. In all treatments, electron transport system activity was significantly inhibited by 35-66% compared to control (Figure 5). Unlike respiration estimated by the dissolved oxygen method, respiration estimated by ETSA did not recover to control levels. Because INT is an artificial substrate, it may be preferentially assimilated by specific populations of microorganisms, and utilized by a selected population of dehydrogenases. This may bias the estimation of respiration by ETSA. In addition, DO measurements taken in microcosms measured the activity of the entire microcosm, but ETSA was determined only on the microbial community of the PFs. The pattern of 13-glcA in microcosms (Figure 6) resembled that of ETSA. On days 7 and 21, enzyme activity in all treatments significantly decreased from control (24--61%). Although no significant differences were detected on day 14 (p = 0.072), 13-glucosidase activity was lower in all treatments compared to control. Increases in microbial 13-glcA are directly

110 1

100] 9O -e-

Day 3

-~-

Day 7

70 Day 14 -W--

60

Day 21

50 40 30

-I1

0

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

0.1

1

,

10

......

100

Diquat (mg/L)

Fig. 6. Effect of diquat on [3-glucosidase activity ([3-glcA) from PF substrates in laboratory microcosms. Data are expressed as percent of control response

proportional to microheterotrophic activity (Chrost 1989). Therefore, similarities in 13-glcA and ETSA, which reflect respiratory activity, may be expected. Like APA, 13-glcA can be regulated by stimulation-inhibition of enzyme activity and by induction-repression of enzyme synthesis. Inhibition of [3-glcA indicated changes in carbohydrate cycling dynamics and indicated that nonphotosynthetic organisms (micro-heterotrophs) were affected by diquat.

System-Level Responses: Dissolved oxygen in treated microcosms decreased from pretreatment levels (Figure 7). By day 3, DO in the 10 and 30 mg/L treatments had significantly decreased from control values by 25% and 31%, respectively. Following dosing, pH measurements followed a pattern similar to that of DO. By day 3, the pH in the 10 and 30 mg/L treatment had decreased significantly from control values (16% and 19%, respectively), but by the second week of treatment

100

A. L Melendez et al. Table 3. Maximum allowable toxicant concentration (MATC, mg/L) for selected microbial responses significantly affected by exposure to diquat. The lowest observable effect concentrations (LOEC) are shown for responses that were significant at the lowest treatment. MATCs are based on mean measured concentrations of diquat

Dissolved Oxygen (mg/L) 13

11109' 8 7"

-8

-6

-4

-2

0

2

4

6

8

Treatment

10

12

14

16

18

20

22

Day

aiquat concentration -o-- Control

~

0.3 mg/L

--e-- 10 rng/L

--z-- 30 mg/L

-~-

1 mg/L

--

Confidence Band

~

3 rng/L

Fig. 7. Effect of diquat on afternoon dissolved oxygen (mg/L) in microcosms. Confidence band is 95% minimum significant difference from control based on Dunnett's procedure

Response

Day 3

Day 7

Day 14

Day 21

Alkaline phosphatase Electron transport system 13-Glucosidase Protein Protozoa species richness Filamentous cyanobacteria cell density Total algal cell density Bacterial cell density

NS" 0.33 b NS NS -2.04

NS 0.29 b 0.29 b NS 0.29 b 0.54

0.43 0.24b NS NS -0.24 b

0.32 0.18 b 0.18 b 5.55 0.32 0.188

NS NS

NS 0.54

NS NS

5.55 0.32

"NS = not significantly different from controls b Lowest-observable-effect concentration

Maximum Allowable Toxicant Concentration

the pH at 10 and 30 mg/L diquat were not significantly different from control. The pH values at 0.3-3 mg/L diquat were significantly lower than control 1 day after application, but quickly recovered to control levels by day 4. Net productivity decreased with increasing diquat treatment, and was significantly lower than control at 10 and 30 mg/L diquat. Negative values for net productivity (10 and 30 mg/L diquat) indicated that photosynthetic activity was lower than daytime respiration. Community respiration followed a pattern similar to that of net productivity. Respiration decreased significantly from control at 10 and 30 mg/L diquat, and remained depressed in these treatments through day 6 (falling below zero at 30 mg/L diquat). Although microcosm tanks were covered with plastic sheeting, the assumption of minimal gas diffusion can be questioned, because negative values for community respiration were obtained. Negative values for respiration are not possible because nighttime productivity is minimal. However, since relative values were needed for comparison, productivity and respiration estimates were not corrected for diffusion. Net productivity inhibition was expected given that diquat interferes with photosynthesis. The concomitant decrease in pH reflected an increase in CO2 due to reduced uptake by photosynthetic organisms and increased respiration. Dissolution of CO2 leads to the formation of H2CO3, which in turn lowers water pH (Stumm and Morgan 1980). Inhibition of community respiration was probably a result of diquar s interaction with the respiratory electron transport chain of microorganisms; diquat inhibits respiration in bacteria (Wallnoefer 1968). Diquat has a variable effect on algal respiration, depending on experimental conditions (Summers 1980). Diquat inhibited respiration of Chlorella in the light (Turner et al. 1970), but stimulated respiration in the dark (Stokes and Turner 1971). Diquat stimulated respiration in Scenedesmus (Van Rensen 1969). By the second week of diquat treatment, DO, pH, net productivity, and community respiration were not significantly different from control values. Recovery most likely involved the selective enrichment of resistant species and acclimation adaptation of microorganisms not sensitive to diquat (Blanck et al. 1988). Oocystis spp. were resistant to diquat and bloomed in the 10 and 30 mg/L treatments, whereas the growth of other algae was inhibited.

Electron transport system activity was the most consistently sensitive indicator of stress (Table 3). However, on certain sampling days other functional (13-glcA) and structural responses (blue-green filamentous and bacterial densities, protozoan species richness) also showed evidence of diquat-related stress. Least sensitive were protein biomass, total algal densities, and unicellular green algal densities. Exposure of microbial communities to diquat was maximized in microcosms given the lack of natural sinks such as plants, and sediment. These sinks may also serve as sources of nutrients. Therefore, microbial responses in this study may not necessarily be representative to those that may be obtained under field conditions. However, diquat may be persistent in the water when the sediment in a water body is low in clay and organic matter. For example, diquat was persistent in microcosms amended with a sandy sediment (J.R. Pratt unpublished). The findings in this study can be summarized as follows. Diquat treatment resulted in changes in the structure and function of the naturally derived microbial communities. Effects on community structure were observed as changes in algal and bacterial cell density, and increased biomass accumulation. Effects on community function included inhibition of net productivity and community respiration, and decreased enzyme activities. Structural and functional responses were sensitive indicators of stress. Least sensitive were total biomass (protein) and total algal cell densities. Changes in net productivity and algal density indicated that diquat affected photosynthetic organisms. Changes in bacterial density and 13-glcA indicated that nonphotosynthetic organisms were also affected. Diquat had direct and indirect effects on structure and function of the microbial communities. Examples of direct effects are the inhibition of net productivity and community respiration. Examples of indirect effects, possibly caused by an increase in nutrient supply, are the stimulation of bacterial growth, the bloom in Oocystis spp., and the inhibition of APA and [3-glcA. Most responses measured in microcosms did not return to control levels within the 21-day diquat exposure, especially at 10 and 30 mg/L diquat. The lack of recovery was probably a result of diquat's persistence in microcosms. Only system-level functional responses returned to control levels in

Diquat Effects in Microcosms

all treatments, and these variables were expected to be the most robust indicators of microcosm condition (Odum 1985). The experiments suggest that in the absence of environmental sinks for diquat, the herbicide has broad and long-lasting effects on microbial community structure and function at concentrations at or below typical field concentrations.

References Birmingham BC, Colman B (1983) Potential phytotoxicity of diquat accumulated by aquatic plants and sediments. Water Air Soil Pollut 19:123-131 Btanck H, W/ingberg S-A, Molander S (1988) Pollution-induced community tolerance--a new ecotoxicological tool. In: Cairns J Jr, Pratt JR (eds), Functional testing of aquatic biota for estimating hazards of chemicals. ASTM, Philadelphia, pp 219-230 Bradford MM (1976) A rapid sensitive method for the quantitation of microgram quantities of protein utilizing the principle of proteindye binding. Analyt Biochem 72:248-254 Breazeale FW, Camper ND (1972) Effect of selected herbicides on bacterial growth rates. Appl Microbiol 23:431-432 Cairns, J Jr, Kuhn DL, and Plafkin JL (1979) Protozoan colonization of artificial substrates. In: Weitzel RL (ed), Methods and measurement of periphyton communities: A review. ASTM, Philadelphia, pp 39-54 Chrost RJ (1990) Microbial ectoenzymes in aquatic environments. In: Overbeck J, Chrost RJ (eds), Aquatic microbial ecology. Springer-Verlag, NY, pp 47-74 --(1989) Characterization and significance of [3-glucosidase activity in lake water. Limnol Oceanogr 34:660-672 Cragg BA, Fry JC (1984) The use of microcosms to simulate field experiments to determine the effects of herbicides on aquatic bacteria. J Gen Microbiol 130:230%2316 Cullimore DR (1975) The in vivo sensitivity of some species of Chlomphyceae to a selected range of herbicides. Weed Res 15:401-406 Dunnett CW (1955) Multiple comparison procedure for comparing several treatments with a control. J Am Stat Assoc 50:1096-1121 Ferebee RN, Guthrie RK (1973) The effects of selected herbicides on bacterial populations in an aquatic environment. Water Res Bull 9:1125-1135 Ganstad EO (1982) Weed control methods for recreational facilities management. CRC Press, Boca Raton, FL Jager G (1983) Herbicides. In: Buchel KH (ed), Chemistry of pesticides. John Wiley & Sons, NY, pp 322-392 Johnson TB (1986) Potential impact of selected agricultural chemical contaminants on northern prairie wetland: A microcosm evaluation. Environ Toxicol Chem 5:473-485 JT Baker (1984) Baker-10SPE Application Guide. Vol. II. JT Baker Chemical Co, Philipsburg, NJ McConnell W (1962) Productivity relations in carboy microcosms. Limnol Oceanogr 7:335-343

101

Morrison SJ, King JD, Bechtold RE, White DC (1977) Evidence for microfloral succession on allochthonous plant litter in Apalachicola Bay, Florida, USA. Mar Bio141:229-240 Odum, EP (1985) Trends expected in stressed ecosystems. BioScience 35:419-422 Philips EJ, Hansen P, and Verlardi, T (1992) Effects of the herbicide diquat on the growth of microalgae and cyanobacteria. Bull Environ Contam Toxico149:750-756 Porter KG, Feig Y (1980) The use of DAPI for identifying and counting aquatic microflora. Limnol Oceanogr 25:943-948 Pratt JR, Bowers NJ, Cairns J Jr (1990) Effects of sediment on estimates of diquat toxicity in laboratory microcosms. Water Res 24:51-57 Pratt JR, Smith EP (1991) Significance of change in community structure: A new method for testing differences. In: Biological criteria: Research and regulation, EPA 440/5-91-005. US Environmental Protection Agency, Washington, DC, pp 91-103 Rausch P (198 I) The estimation of micro-algal protein content and its meaning to the evaluation of algal biomass I. Comparison of methods for extracting protein. Hydrobiologia 78:237-251 Sayler GS, Puziss M, Silver M (1979) Alkaline phosphatase assay for freshwater sediments: application to perturbed sediment systems. Appl Envir Microbio138:922-927 Simsiman GV, Daniel TC, Chesters G (1976) Diquat and endothall: Their fates in the environment. Res Rev 62:132-174 Sokal RR, Rohlf FJ (1983) Biometry. WH Freeman, NY Stokes DM, Turner JS (1971) The effects of the dipyridyl diquat on the metabolism of Chlorella vulgaris; III. Dark metabolism. Effects on respiration rate and the path of carbon. Aust J Biol Sci 24:433447 Stumm W, Morgan JJ (1981) Aquatic Chemistry: An introduction emphasizing chemical equilibria in natural waters, 2nd ed. John Wiley & Sons Inc, NY Summers LA (1980) The bipyridinium herbicides. Academic Press, London Turner JS, Stokes DM, Gilmore LB (1970) The effects of the dipyridyl diquat on the metabolism of Chlorella vulgaris I. Gas exchange in the light. Aust J Biol Sci 23:43-61 Van Rensen JJS (1969) Effects of diquat on photosynthesis in Scenedesmus. Meded Landbouwhogesch Wageningen 69-4:1-11 Velji MI, Albright LJ (1986) Microscopic enumeration of attached marine bacteria of seawater, marine sediment, fecal matter and kelp blade samples following pyrophosphate and ultrasound treatments. Can J Microbiol 32:121-126 Wallnoefer VP (1968) Untersuchungen tiber die antimikrobielle Wirkung von Deiquat und Paraquat. Z Pflanzenkr Pflanzenschutz 75:218-224

Manuscript received October 24, 1992 and in revised form February 4, 1993.