Potential toxicity concerns from chemical coagulation treatment of stormwater in the Tahoe basin, California, USA

ARTICLE IN PRESS Ecotoxicology and Environmental Safety 72 (2009) 1933–1941

Contents lists available at ScienceDirect

Ecotoxicology and Environmental Safety journal homepage: www.elsevier.com/locate/ecoenv

Potential toxicity concerns from chemical coagulation treatment of stormwater in the Tahoe basin, California, USA S.E. Lopus a,, P.A.M. Bachand a, A.C. Heyvaert b, I. Werner c, S.J. Teh c, J.E. Reuter d a

Bachand & Associates, 2023 Regis Drive, Davis, CA 95618, USA Desert Research Institute, 2215 Raggio Parkway, Reno, NV 89512, USA c University of California, School of Veterinary Medicine, Department of Anatomy, Physiology and Cell Biology, Aquatic Toxicology Program, Davis, CA 95616, USA d Department of Environmental Science, Tahoe Environmental Research Center, University of California at Davis, One Shields Avenue, Davis, CA 95616, USA b

a r t i c l e in fo

abstract

Article history: Received 16 January 2009 Received in revised form 1 April 2009 Accepted 4 April 2009 Available online 9 May 2009

Coagulant dosing of stormwater runoff with polyaluminum chlorides (PACs) is used in numerous waterbodies to improve water clarity, but the potential risks of PACs to aquatic organisms in Lake Tahoe, California are not fully understood. To assess these risks, the USEPA 3-species toxicity test and a nonstandard fish test using Japanese medaka (Oryzias latipes) were used to determine the toxicity of PACtreated and non-treated stormwater samples to aquatic species. Stormwater samples were collected from three sites representing runoff from different urbanized areas in May 2004; samples received coagulant dosing using three different coagulants (JC1720, PAX-XL9, Sumalchlor50) at levels optimized with jar testing. Raw stormwaters were toxic to algae and fathead minnows (mortality). Treatment with coagulants increased toxicity to zooplankton (reproduction) and had no consistent effects on the other toxicity metrics. & 2009 Elsevier Inc. All rights reserved.

Keywords: Coagulants Ecotoxicity Medaka (Oryzias latipes) Stormwater Runoff

1. Introduction Phosphorus and fine particles in the range of 0.5–20 mm have been identified as the major pollutants of concern (PoC) negatively affecting water clarity in Lake Tahoe (Heyvaert et al., 2004; Strecker and Howell, 2003; Swift et al., 2006), an oligotrophic alpine lake in Northern California, USA. Several PoC sources have been identified, including runoff from undisturbed areas, runoff from disturbed or urbanized areas, and atmospheric deposition (Reuter and Miller, 2000). Stormwater runoff from urbanized areas around Lake Tahoe has been the focus of recent water quality management efforts, as these areas are more accessible and conducive to treatment than other sources of suspended sediments and nutrients. Efforts have been directed at these areas to control runoff by implementing stormwater treatment best management practices (BMPs) designed to capture and treat surface runoff and increase infiltration (Bachand et al., 2005a; Schuster and Grismer, 2004). However, a review of regional and national data sets (Bachand et al., 2005a; Winer, 2000) suggests that the short hydraulic residence times common among most Lake Tahoe treatment BMPs do not allow effective  Corresponding author. Fax: +1 510 558 8033.

E-mail addresses: [email protected] (S.E. Lopus), [email protected] (P.A.M. Bachand), [email protected] (A.C. Heyvaert), [email protected] (I. Werner), [email protected] (S.J. Teh), [email protected] (J.E. Reuter). 0147-6513/$ - see front matter & 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ecoenv.2009.04.004

settlement of fine particles and associated phosphorus, nor can they accommodate highly variable storm runoff pulses occurring with frequent storms (Bachand et al., 2005a, 2007; Lumos and Associates Inc., 2005). Discharges from these stormwater treatment BMPs are therefore unlikely to meet regional surface water quality standards under typical hydraulic conditions and landscape configurations (Bachand et al., 2005a; Caltrans, 2002). Settling rates of fine particles and dissolved constituents are partially controlled by physiochemical coagulation processes, which can be manipulated to accelerate the removal of entrained PoC. While autochthonous coagulation is an important ecosystem process, it proceeds at relatively slow rates, taking days to weeks depending upon water chemistry, hydraulic mixing, and water depth (Bachand et al., 2007; Weilenmann et al., 1989). Recent studies in the Lake Tahoe area have shown that using lowintensity coagulant dosing (LICD) techniques to treat stormwater runoff with select polyaluminum chlorides (PACs) may effectively decrease phosphorus and turbidity levels in surface waters in conjunction with existing treatment wetlands (Bachand et al., 2005b; Trejo-Gayton et al., 2006). Because coagulant dosing is a technically feasible, cost-effective means of reducing suspended phosphorous and fine particles in surface waters, LICD techniques may be well suited to meet Lake Tahoe water quality goals (Bachand et al., 2005b; Caltrans, 2001). Despite the functionality of LICD techniques, the potential for chemical coagulation to introduce toxicity to aquatic organisms remains in question, particularly in oligotrophic systems such as

ARTICLE IN PRESS 1934

S.E. Lopus et al. / Ecotoxicology and Environmental Safety 72 (2009) 1933–1941

Lake Tahoe. A portion of coagulant remains in the environment in any coagulation treatment process, and coagulant concentration is strongly correlated with toxicity in some wastewater systems (Al-Mutairi, 2006). Investigations of coagulant-related toxicity with alum and alum–polymer combinations have demonstrated mixed results, including reductions in toxicity to zooplankton with PAC treatment (Selcuk et al., 2007) and dose-related increases in aluminum toxicity to zooplankton and algae with increasing polymer levels (Guida et al., 2004). Since urban runoff can produce significant toxicity (Skinner et al., 1999), and toxicity of urban runoff is largely associated with suspended sediments (Caltrans, 2003), the question of whether coagulant treatment will exacerbate or reduce stormwater toxicity remains. This paper addresses that question by evaluating the toxicity of non-treated and coagulant-treated stormwater samples to several aquatic organisms. We first assessed ecotoxicity to algae, zooplankton, and fish using the standard USEPA 3-species toxicity test, and secondly, we measured the effects of stormwater and

Fig. 1. Map of stormwater sampling locations. S ¼ Stag, TC ¼ Tahoe City, and SR ¼ Ski Run.

coagulants on the embryonic development of Japanese medaka, a fish species that was the subject of recent ecotoxicity testing by Patyna et al. (2006) and Rhodes et al. (2005). Specifically, this study was performed to assess whether stormwater runoff entering the Tahoe basin introduces toxicity into the lake and, if so, whether coagulants reduce or increase the toxicity. This study was designed to provide a foundation for more extensive toxicity studies by identifying key issues and addressing these basic hypotheses.

2. Materials and methods Stormwater runoff samples were collected into high-density polyethylene (HDPE) buckets from three Tahoe basin locations representing northern urbanized (Tahoe City), southern urbanized (Ski Run), and general highway (Stag) runoff (Fig. 1) in May 2004. Sampling was timed to a ‘‘first flush’’ event, being the first runoff to occur in the Tahoe basin since the end of winter snow fall approximately 3 months earlier. Stormwater from these sites was dosed with one of three polyaluminum chloride-based coagulants (JC 1720, PAX-XL9, Sumalchlor 50) characterized in Table 1; coagulants were selected based on their slightly different chemical characteristics, their robustness to a range of mixing conditions and dosing levels, and their effectiveness in removing turbidity and phosphorus from stormwater runoff under different conditions (Bachand et al., 2005b; Trejo-Gayton et al., 2006). Dosing levels were determined using standard jar testing methods, in which a dose is determined based upon appearance of the threshold of optimal flocculate formation within a jar (Gnagy, 1994). Stormwater from each site was chilled and homogenized by shaking. One-liter stormwater samples were placed in multiple glass beakers. Each beaker was treated with a coagulant dosage within a range of at least six dosages, spanning under-dosed to over-dosed conditions, with two replicates for each dosage. Jars were agitated with jar stirrers and allowed to settle for 30 min. Turbidity of each jar was measured at pre-dose and after 15 and 30 min of settling, and floc observations were recorded. For each stormwater and coagulant tested, optimal dosing level (Table 1) was determined according to visual assessments of settling and clarity characteristics and/or turbidity assessments of the lowest dose that achieved less than 10 NTU after 30 min of settling. All dosing regimes reduced stormwater turbidity (Table 2). Non-treated and coagulant-treated stormwaters were tested for toxicity to green algae (Selenastrum capricornutum, growth), zooplankton (Ceriodaphnia dubia, reproduction and mortality), and fish (Pimephales promelas, biomass and mortality) by the Aquatic Toxicity Laboratory, University of California at Davis (UCD ATL) using standard USEPA methods (US EPA, 2002). In addition, nonstandard tests using the teleost species Japanese medaka (Oryzias latipes) were conducted to determine toxic effects on reproductive success and embryonal development of fish. Fish (Fathead minnow, P. promelas): This chronic, 7-d test consisted of four replicate 600 ml glass beakers each containing 250 ml of sample and 10 organisms. Tests were initiated with less than 24 h-old P. promelas larvae obtained from Aquatox (Hot Springs, Arkansas). Fish were fed freshly-hatched Artemia nauplii twice daily. Approximately 80% of the test solution was renewed daily. Control water consisted of deionized water amended to USEPA moderately hard specifications. Tests were conducted at 2572 1C, with a 16:8 light–dark photoperiod. Mortality was recorded daily to measure acute toxicity, and after 7 d, surviving fish were anesthetized with MS-222, rinsed with deionized water, dried to a constant weight at 103–105 1C (approximately 16 h), and weighed with a Mettler AE 163 balance to measure chronic toxicity. Zooplankton (Water flea, C. dubia): Less than 24-h-old C. dubia, born within an 8-h period, were obtained from an in-house culture at UCD ATL. The test consisted of 10 replicate 20 ml glass vials each containing one organism. Sample water containing a mixture of S. capricornutum and YCT (a mixture of yeast, organic alfalfa and trout chow) was renewed daily. Control water consisted of Sierra SpringsTM water amended to USEPA moderately hard specifications. Tests were

Table 1 Coagulant specifications and stormwater dosing regimes. Coagulant code

a

J1720 PXXL9 SUM50 a

Vendor

JenChem JC Kemiron Summit Sumalchlor

Coagulant

1720 s PAX-XL9 s 50 s

Designation

Polyaluminum chloride Polyaluminum chloride Aluminum chlorohydrate

Blended with an organic polymer. SG ¼ specific gravity.

% basicity

70 67 83.5

PH

4.3 2.8 4.2

SG

1.29 1.26 1.34

% Al

6.0 5.6 12.4

Dosing regime (mg-Al/L)

Dosing regime (mg-coag/L)

Ski Run

Stag

Tahoe City

Ski Run

Stag

Tahoe City

17.3 17.0 20.6

8.8 16.5 16.0

2.4 5.7 2.6

290.8 303.6 166.1

147.9 294.6 129.0

40.3 101.8 21.0

ARTICLE IN PRESS S.E. Lopus et al. / Ecotoxicology and Environmental Safety 72 (2009) 1933–1941 conducted at 2572 1C with a 16:8 light–dark photoperiod. Mortality (acute toxicity) and reproduction (number of neonates, to measure chronic toxicity) were recorded daily and at test termination (after 3rd brood; days 6–8). Algae (Green algae, S. capricornutum): These 4-d chronic tests consisted of four replicate 200 ml glass flasks containing 100 ml of water and 1 ml of 1.0  106 cell/ ml S. capricornutum, obtained from a culture at UCD ATL. Control water consisted of glass distilled water. Test chambers were incubated at 2572 1C under continuous cool white fluorescent light, on a mechanical shaker in constant orbital motion (100 cycles/min); flasks were randomized twice daily. Growth (final cell count) was measured at test termination.

Table 2 Mean turbidity (NTU), standard deviation, and sample size of non-treated and treated stormwaters, by site and coagulant. Turbidity (NTU)

Non-treated Treated J1720 PXXL9 SUM50

Ski Run

n

Stag

n

Tahoe City

n

81.576.8

36

45.074.7

36

26.872.3

42

2 2 2

3.470.4 3.370.2 10.570.2

2 2 2

5.470.9 3.670.2 9.773.8

5.770.4 1.970.3 9.972.2

2 2 2

Treated turbidity represents measurements of stormwaters receiving optimal dosage after 30 min settling time.

1935

Medaka (O. latipes): In addition, toxic effects on fecundity and embryonal development of the teleost fish species medaka (Oryzias latipes) were quantified. This test used a repeated measures approach in which two adult female and one adult male medaka were exposed to 500 ml of stormwater or control water for 4 days. The test consisted of three replicates per treatment, and 80% of water was renewed daily. Fish were fed twice daily, and water quality parameters (conductivity, pH, temperature, dissolved oxygen) were measured and recorded daily. Across all days of experimentation, conductivity ranged from 158 to 385 mS, pH ranged from 6.2 to 7.8, temperature ranged from 24.5 to 26.0 1C, and dissolved oxygen ranged from 3.5 to 9.2 mg/L. Tests were performed at 25 1C and a photoperiod of 16L:8D. Fertilized eggs from each replicate were collected on days 0–4 of adult exposure and were then exposed to stormwater or control water until hatching. Developing embryos were monitored for abnormalities and mortality daily. Toxicity endpoints measured were adult survival, number of fertilized eggs produced per day, and hatching success. Results of reference toxicant tests with green algae, waterflea and fathead minnow were within the normal range for each test species, and water quality parameters were within the ranges specified by US EPA (2002). All components of the investigation were conducted in accordance with national and institutional guidelines for the protection of animal welfare.

2.1. Statistical analyses Data obtained from the USEPA 3-species tests were analyzed using ANOVA (p-values cited in text) and Tukey–Kramer HSD (significant differences provided in Tables 3, 5 and 6 and Figs. 2–6). Data for the medaka test were analyzed using repeated measures ANOVA (p-values cited in text), correlations analysis (R-values cited in text), and Tukey–Kramer HSD to make comparisons within days

Table 3 Mean values shown for each toxicity metric in USEPA 3-species test. Control Algae Cell count All stormwater Sig. (po0.05) Stormwater site Sig. (po0.05) n Zooplankton Reproduction (#) All stormwater Sig. (po0.05) Stormwater site Sig. (po0.05) n Mortality (%) All stormwater Sig. (po0.05) Stormwater site Sig (po0.05) n

2.24E+06 b Control 2.24E+06 e 12

28.4 b Control 28.4 c 20 5.0% a Control 5.0% ab 20

Fish Survivor biomass (mg/survivor) All stormwater 0.22 Sig. (po0.05) a Stormwater site Control 0.22 Sig. (po0.05) ab n 8 Mortality (%) All stormwater 4.9% Sig. (po0.05) a Stormwater site Control 4.9% Sig. (po0.05) a n 8

Stormwater: no dosing

Ski Run 5.82E+05 ab 4

Ski Run 39.0 d 10

Ski Run 0.0% ab 10

Ski Run 0.16 a 4

Ski Run 72.5% b 4

Stormwater: coagulant dosing

Stag 1.54E+06 cde 4

1.09E+06 a Tahoe City 1.14E+06 abcd 4

Stag 23.4 bc 10

26.3 b Tahoe City 16.5 b 10

Stag 20.0% ab 10

10.0% a Tahoe City 10.0% ab 10

0

0.21 a Tahoe City 0.26 ab 4

Stag 100.0% c 4

62.5% b Tahoe City 15.0% a 4

Stag N/A

Ski Run 4.50E+05 a 12

Ski Run 6.3 a 29

Ski Run 30.0% b 30

Ski Run 0.30 b 5

Ski Run 93.3% c 12

Stag 1.07E+06 bc 12

1.05E+06 a Tahoe City 1.62E+06 d 12

Stag 6.2 a 30

12.0 a Tahoe City 23.5 bc 30

Stag 3.3% a 30

14.4% a Tahoe City 10.0% ab 30

Stag 0.15 a 4

0.26 a Tahoe City 0.29 b 12

Stag 94.2% c 12

66.7% b Tahoe City 12.5% a 12

‘‘Coagulant dosing’’ values are means of values from each coagulant. Biomass of surviving fish at Stag not applicable, due to 100% mortality rate at site. ‘‘All stormwater’’ values are means of values from each stormwater site. Different letters indicate statistical differences (po0.05) within rows between control, non-dosed stormwaters, and dosed stormwaters. Designation of letters begins with ‘‘a’’ assigned to the lowest value. Bold face identifies the most toxic group(s) for metrics with significant differences observed. Significance calculated within rows using Tukey–Kramer HSD.

ARTICLE IN PRESS 1936

S.E. Lopus et al. / Ecotoxicology and Environmental Safety 72 (2009) 1933–1941

Fig. 2. Mean algae cell counts (USEPA 3-species test) for control water, non-treated stormwater, and treated stormwater. Different letters indicate statistical differences (po0.05) according to Tukey–Kramer HSD. Vertical bars represent 71 SD.

Fig. 5. Mean fish biomass (mg per survivor, USEPA 3-species test) for control water, non-treated stormwater, and treated stormwater. No significant differences (a ¼ 0.05) between variables, according to Tukey–Kramer HSD. Vertical bars represent 71 SD.

Fig. 3. Mean zooplankton brood size (USEPA 3-species test) for control water, nontreated stormwater, and treated stormwater. Different letters indicate statistical differences (po0.05) according to Tukey–Kramer HSD. Vertical bars represent 71 SD. Fig. 6. Mean fish mortality (USEPA 3-species test) for control water, non-treated stormwater, and treated stormwater. Different letters indicate statistical differences (po0.05) according to Tukey–Kramer HSD. Vertical bars represent 71 SD.

(significant differences provided in Tables 4–6). Significance was determined with an a-level of 0.05. Significant correlations with RX|0.70| were considered to be highly correlated. Since coagulant dosing regimes were determined with the somewhat subjective method of jar testing, we did not differentiate between coagulants with respect to toxicity. Statistical analyses were performed using Statistica (ANOVA, Tukey–Kramer HSD, correlations analysis) and jmp (repeated measures ANOVA) statistical software.

3. Results 3.1. Effectiveness of coagulants

Fig. 4. Mean zooplankton mortality rate (USEPA 3-species test) for control water, non-treated stormwater, and treated stormwater. No significant differences (a ¼ 0.05) between variables, according to Tukey–Kramer HSD. Vertical bars represent 71 SD.

All coagulants reduced stormwater turbidity, although effectiveness varied (Table 2). Coagulants reduced turbidity with approximately equal effectiveness for Ski Run stormwater (p ¼ 0.141), but SUM50 reduced turbidity less effectively than the other coagulants for Stag (po0.001) and Tahoe City (p ¼ 0.020) stormwaters.

ARTICLE IN PRESS S.E. Lopus et al. / Ecotoxicology and Environmental Safety 72 (2009) 1933–1941

1937

Table 4 Mean values shown for each toxicity metric in Medaka Test, each day. Day 0 Control

Fish Fecundity (eggs per day) All stormwater 7.33 Sig. (po0.05) a n 3 Fish larvae Hatching success (%) All stormwater 83.5% Sig. (po0.05) a n 2 Days to hatch All stormwater 7.50 Sig. (po0.05) a n 2

Day 1

Day 2

Day 3

Day 4

SW: no dose

SW: Coag

Control

SW: no dose

SW: Coag

Control

SW: no dose

SW: coag

Control

SW: no dose

SW: coag

Control

SW: no dose

SW: coag

9.67 a 9

13.44 a 27

0.33 e 3

10.22 ef 9

9.89 f 27

2.00 j 3

13.56 j 9

9.93 j 27

7.50 q 2

8.11 q 9

7.22 q 27

3.00 x 15

8.67 x 45

7.56 x 135

95.4% a 8

75.1% a 22

0% e 1

60.0% ef 9

81.0% f 24

50.0% j 1

69.7% j 9

69.9% j 23

100% q 2

71.9% q 9

72.2% q 21

100.0% x 1

73.8% x 9

71.7% x 21

8.50 a 8

8.37 a 19

N/A

9.00 f 7

7.65 e 23

10.00 j 1

10.44 j 9

10.43 j 21

8.50 q

10.13 q

9.72 q

8.00 x 1

9.67 x 9

9.74 x 19

0

Different letters indicate statistical differences (po0.05) within rows and days between control, non-dosed stormwaters (‘‘SW: no dose’’), and dosed stormwaters (‘‘SW: Coag’’). Letters ‘‘a–c’’ used to indicate statistical differences within rows on day 0, ‘‘e–g’’ indicate differences on day 1, ‘‘j–l’’ indicate differences on day 2, ‘‘q–s’’ indicate differences on day 3, and ‘‘x–z’’ indicate differences on day 4. Designation of letters begins with initial letter (‘‘a,’’ ‘‘e,’’ ‘‘j,’’ etc) assigned to the lowest value. Significance calculated within rows using Tukey–Kramer HSD. Bold face identifies the most toxic group(s) for days with significant differences observed.  Significance calculated with t-test for days to hatch on day 1 because no data available for control group.

Table 5 Ranking the effects of non-dosed stormwaters on each toxicity metric and the toxicity of non-dosed stormwaters compared with controls.

USEPA 3-species: toxicity rankings Algae Cell count Zooplankton Reproduction Mortality Fish Survivor biomass Mortality USEPA 3-species: toxicity vs. control Algae Cell count Zooplankton Reproduction Mortality Fish Survivor biomass Mortality Medaka Test: toxicity rankings Fish Fecundity Fish larvae Days to hatch Hatching success Medaka Test: toxicity vs. control Fish Fecundity Fish larvae Hatching success Days to hatch

Ski Run (S. urbanized)

Stag (highway)

Tahoe City (N. urbanized)

All sites

2

1

1.5

1 1

2 1

2 1

1 2

NA 3

1 1

More

Same

More

More

Less Same

Same Same

More Same

Same Same

Same More

NA More

Same Same

Same More

1

1

1

2 1

1.5 1

1 1

Less

Same

Same

Less

Same More

Same Same

Same Same

Same Same

Ranking is from 1 to 3 with 1 being least toxic. Half numbers identify no statistical difference between higher- and lower-ranked stormwater. Toxicity vs. control indicates whether sites were significantly more, less, or same toxicity as control for each toxicity metric. For medaka metrics, all values (days 0–4) included, without accounting for repeated measures. Bold face and bold italics face highlight significant increases and decreases in toxicity, respectively, compared with control water. All differences in table determined using Tukey–Kramer HSD, a ¼ 0.05.

ARTICLE IN PRESS 1938

S.E. Lopus et al. / Ecotoxicology and Environmental Safety 72 (2009) 1933–1941

Table 6 Comparing the toxicity of dosed stormwaters with control and non-treated stormwater for each toxicity metric. Ski Run (S. urbanized) USEPA 3-species: toxicity of dosed vs. control Algae Cell Count More Zooplankton Reproduction More Mortality Same Fish Survivor biomass Same Mortality More USEPA 3-species: toxicity of dosed vs. non-treated Algae Cell count Same Zooplankton Reproduction More Mortality Same Fish Survivor biomass Less Mortality More Medaka test: toxicity of dosed vs. control Fish Fecundity Less Fish larvae Days to hatch More Hatching success Same Medaka test: toxicity of dosed vs. non-treated Fish Fecundity Same Fish larvae Days to hatch Same Hatching success Same

Stag (Highway)

Tahoe City (N. urbanized)

All sites

More

More

More

More Same

Same Same

More Same

NA More

Same Same

Same More

Same

Same

Same

More Same

Same Same

More Same

NA Same

Same Same

Same Same

Same

Same

Less

Same Same

Same Same

Same Same

Same

Same

Same

Same Same

Same Same

Same Same

Toxicity vs. control indicates whether dosed sites were significantly more, less, or same toxicity as control for each toxicity metric. Toxicity vs. non-treated indicates whether dosed sites were significantly more, less, or same toxicity as non-treated stormwater for each toxicity metric. For Medaka metrics, all values (days 0–4) included, without accounting for repeated measures. Bold face and bold italics face highlight significant increases and decreases in toxicity, respectively, with coagulant dosing. All differences in table determined using Tukey–Kramer HSD, a ¼ 0.05.

3.2. USEPA 3-species test: algae cell counts Across all sites, non-dosed stormwater significantly reduced algae cell counts below control levels (po0.001, Table 3, Fig. 2). Algae cell counts were also significantly lower in dosed stormwater across all sites than in control water (po0.001), but coagulant dosing had no significant effect on algae cell counts compared with non-treated stormwater (p ¼ 0.851, Table 3, Fig. 2). While Tahoe City’s non-treated water was only moderately toxic to algae, and cell counts were not significantly different in stormwater from other sites, the treated Tahoe City stormwater was significantly less toxic than treated stormwater from other sites (Tukey–Kramer HSD po0.05, Table 3). Tahoe City algae cell counts were significantly higher in treated stormwater than in non-treated stormwater (ANOVA p ¼ 0.022, Tukey–Kramer HSD not significant [p40.05], Table 3), despite the fact that the coagulant SUM50 did not remove turbidity entirely. There were no significant differences in algae cell counts between treated and non-treated stormwater from Ski Run or Stag sites (p ¼ 0.464 and 0.254, respectively).

(p ¼ 0.003, Table 3), but Tahoe City stormwater significantly reduced fecundity compared with the control (p ¼ 0.003; no significant difference at Stag: p ¼ 0.198). Coagulant dosing significantly increased toxicity. Across all sites, zooplankton brood size was significantly lower in dosed stormwater than in control water samples (po0.001, Table 3, Fig. 3) or in non-treated stormwater (po0.001). Brood sizes for both Ski Run and Stag stormwaters greatly decreased with coagulant dosing. For Ski Run stormwater, brood sizes dropped to between 0 and 10.4 when treated with coagulants, as compared with a mean of 39.0 for the non-treated stormwater (po0.001, Table 3), and for Stag stormwater, average brood size with dosing dropped to between 2.3 and 9.9, as compared with mean 23.4 for the non-treated Stag stormwater (po0.001). Only in the Tahoe City stormwater was zooplankton brood size not significantly affected by coagulant dosing (p ¼ 0.094, Table 3); treated Tahoe City water showed effects ranging from little change to improved brood size. This may in part be due to Tahoe City’s comparatively low dosing levels (Table 1): mean dosing levels for Stag and Ski Run stormwater were 13.8 and 18.3 mg-Al L1, respectively, while mean dosing level for Tahoe City stormwater was only 3.6 mg-Al L1.

3.3. USEPA 3-species test: zooplankton reproduction 3.4. USEPA 3-species test: zooplankton mortality Non-dosed stormwater had no significant effect on zooplankton brood size compared with the control across all sites (p ¼ 0.529, Table 3, Fig. 3). However, significant site-specific effects were observed at Ski Run and Tahoe City. Zooplankton fecundity was higher in Ski Run stormwater than in the control

Non-dosed stormwater had no significant effect on zooplankton mortality compared with the control across all sites (p ¼ 0.533, Table 3, Fig. 4) or at individual sites (Ski Run: p ¼ 0.489; Stag: p ¼ 0.210; Tahoe City Wetland: p ¼ 0.619).

ARTICLE IN PRESS S.E. Lopus et al. / Ecotoxicology and Environmental Safety 72 (2009) 1933–1941

Coagulant dosing also had no significant effect on zooplankton mortality when compared with controls across all sites (p ¼ 0.256, Fig. 4), compared with untreated stormwater across all sites (p ¼ 0.539) or compared with untreated stormwater at individual sites (Ski Run: p ¼ 0.051; Stag: p ¼ 0.087; Tahoe City Wetland: p ¼ 1.000, Table 3).

1939

beakers more stressful to the fish. There were no significant differences in fecundity between control, non-treated stormwater, and treated stormwater samples after 0 or 2–4 days of exposure (p40.05, Table 4).

3.8. Medaka tests: hatching success 3.5. USEPA 3-species test: fish biomass Non-dosed stormwater had no significant effect on biomass of surviving fish compared with the control across all sites (p ¼ 0.711, Table 3, Fig. 5) or within sites (Ski Run: p ¼ 0.156, Tahoe City Wetland: p ¼ 0.149, no data available for Stag due to 100% mortality). There was also no significant difference in biomass of surviving fish between treated stormwater and the control across all sites (p ¼ 0.186, Fig. 5); fish biomass was significantly higher than the control in treated stormwater from Ski Run (ANOVA p ¼ 0.006, Tukey–Kramer HSD not significant [p40.05], Table 3) and Tahoe City Wetland (ANOVA p ¼ 0.019, Tukey–Kramer HSD not significant [p40.05]), and significantly lower than the control at Stag (ANOVA p ¼ 0.013, Tukey–Kramer HSD not significant [p40.05]). Coagulant dosing did not significantly affect biomass of surviving fish when compared with non-dosed stormwater across all sites (p ¼ 0.136, Fig. 5), but it significantly increased fish biomass compared with non-dosed stormwater at Ski Run (p ¼ 0.036, Table 3). No significant differences were seen in treated versus non-treated stormwater from Tahoe City Wetland (p ¼ 0.392, Table 3, no data available for Stag stormwater).

According to repeated measures analysis, there was no significant difference in medaka hatching success between control, non-dosed, and dosed samples (p ¼ 0.966, Fig. 8). Within treatment groups, there was a significant relationship between days of exposure and hatching success (p ¼ 0.043, Fig. 8), and the relationship between treatment group and hatching success depended on days of exposure (p ¼ 0.007). However, hatching success was not strongly correlated with days of exposure for dosed, non-dosed, or control samples (p40.05). Zero-percent hatching success among control embryos after 1 day of exposure resulted from fecundity levels of zero in this treatment on this day (see above). There were no significant differences between control, untreated stormwater, and treated stormwater samples after 0 or 2–4 days of exposure (p40.05, Table 4).

3.6. USEPA 3-species test: fish mortality Non-dosed stormwater significantly affected fathead minnow survival across all sites (po0.001, Table 3, Fig. 6). Control mean mortality of 4.9% was significantly less than mean mortality of 72.5% at Ski Run (po0.001, Table 3) and 100% at Stag (po0.001) but not significantly less than mortality in stormwater from Tahoe City (p ¼ 0.144). Although treated stormwater was significantly more toxic than control water across all sites (po0.001, Fig. 6), coagulant dosing did not affect fish survival compared with nontreated stormwaters across all sites (p ¼ 0.753). However, coagulant dosing significantly increased toxicity at Ski Run compared with non-treated stormwater (p ¼ 0.002, Table 3). There were no significant differences in mean mortality between non-dosed and dosed samples at Stag (p ¼ 0.272, Table 3) or Tahoe City Wetland (p ¼ 0.752).

Fig. 7. Mean medaka fecundity for control water, non-treated stormwater, and treated stormwater. Vertical bars represent 71 SD.

3.7. Medaka tests: fecundity According to repeated measures analysis, there was no significant difference in medaka fecundity between control, nondosed, and dosed samples (p ¼ 0.541, Fig. 7), and within treatment groups, there was no significant relationship between days of exposure and fecundity (p ¼ 0.517). The relationship between treatment group and fecundity did not depend on days of exposure (p ¼ 0.190). Fecundity was not strongly correlated with days of exposure for dosed (R ¼ 0.243, po0.05, Fig. 7), non-dosed (p40.05), or control samples (p40.05). After 1 day of exposure, fecundity was significantly higher in stormwater treated with a coagulant than in control water (p ¼ 0.015, Table 4); however, this effect is potentially related to nonchemical factors such as light intensity caused by the differences in turbidity and coloration between control water (clear) and stormwater (turbid/intense coloration), with the clarity of control water making the transfer from aquariums to

Fig. 8. Mean medaka hatching success for control water, non-treated stormwater, and treated stormwater. Vertical bars represent 71 SD.

ARTICLE IN PRESS 1940

S.E. Lopus et al. / Ecotoxicology and Environmental Safety 72 (2009) 1933–1941

Fig. 9. Mean medaka days to hatch for control water, non-treated stormwater, and treated stormwater. No data for control water on day 1. Vertical bars represent 71 SD.

3.9. Medaka tests: days to hatch According to repeated measures analysis, there was no significant difference in medaka days to hatch between treated and non-treated stormwater (p ¼ 0.419, Fig. 9, insufficient data for control samples due to 0% hatching success on Day 1). Within treatment groups, there was a significant relationship between days of exposure and days to hatch (po0.001, Fig. 9), but the relationship between treatment group and hatching success did not depend on days of exposure (p ¼ 0.585). There was no strong correlation between days to hatch and days of exposure for dosed (R ¼ 0.425, po0.05, Fig. 9), non-dosed (p40.05), or control samples (p40.05).

4. Discussion This study assessed the aquatic toxicity potentially associated with the chemical coagulation process used to remove phosphorus and fine particles from stormwater. All coagulants achieved the intended goal of increasing stormwater clarity, reducing mean turbidity from ranges of 26.8–81.4 NTU for untreated stormwaters to 1.9–10.5 NTU with treatment, although SUM50 was less effective than J1720 or PXXL9. There was significant toxicity in some of the untreated stormwaters, and the coagulant treatment increased toxicity to some species, but it is difficult to attribute the increase in toxicity to individual coagulants. The observed toxic effects in treated stormwater were more likely a result of stormwater-associated contaminants, possibly in combination with coagulant residues, than of the coagulants alone. Toxicity of stormwater varied by source and affected test species differently. Each stormwater source was less toxic than others in at least one toxicity metric and more toxic than others in at least one other metric (Table 5). Ski Run stormwater, from southern urbanized runoff, was the least toxic site with regard to zooplankton reproduction, moderately toxic for fish (mortality), and the most toxic site for green algae. It also negatively affected embryo development of medaka (days to hatch). Stag stormwater, which is dominated by highway runoff, was the least toxic site for green algae, one of the most toxic sites with respect to zooplankton reproduction, and the most toxic site for fish. Tahoe City stormwater, from northern urbanized runoff, was the least toxic site for fish, but it was one of the most toxic sites with respect to zooplankton reproduction.

These differences in toxic effects on aquatic species are likely due to the different types of toxicants present in each stormwater source. For example, due to their mechanism of action, insecticides are generally more toxic to zooplankton than fish or algae (US EPA, 2007), whereas some heavy metals are more toxic to algae than zooplankton (US EPA, 2007). Roofs and building sidings can be major sources of Cd, Cu, Pb, and Zn to urban stormwater; vehicle brake emissions can contribute Cu; tire wear can contribute Zn; atmospheric deposition can contribute Cd, Cu, and Pb; and car washes can be major contributors of Cd, Cr, Pb, and Zn (Davis et al., 2001; So¨rme and Lagerkvist, 2002). Medaka have been shown to be more sensitive to stormwater toxicity from developed land surfaces than from open space, with stormwater toxicity highly correlated with total metals but poorly correlated with individual chemical pollutants, including Cd, Cr, Cu, Pb, Ni, and Zn (Skinner et al., 1999). Since no clear toxicity trends appear when comparing urbanized runoff (from Ski Run and Tahoe City) with highway runoff (Stag), it is possible that the toxicant composition of the urbanized runoff sources may be quite different as a result of their different locations. A more detailed land use analysis at these sites may reveal the reasons for the observed differences in toxicity. Stormwater toxicity is also expected to vary temporally and seasonally, with higher metal flux in early season than late season storms (Tiefenthaler et al., 2008), but all stormwater samples were collected at the same time and generally represent a first rainfall flush from a late spring event after snowmelt, so temporal and seasonal variations are not expected to be the cause of variance within the different toxicity metrics. Coagulant-dosed stormwaters were more toxic in some cases than the controls but in general did not significantly affect toxicity compared with non-dosed stormwater except with regard to zooplankton reproduction, which increased (Table 6). Table 6 summarizes the toxicity effects for the treated stormwaters compared with the control and the non-treated stormwaters. Unlike stormwater from the other sites, toxicity in stormwater from Tahoe City did not change following coagulant treatment (Table 6). It is possible that this was due to the relatively low coagulant dosing levels used to treat this sample (Table 1) or to the low turbidity of the non-dosed Tahoe City stormwater (Table 2). Since we used only one dosing level per treatment, we could not extrapolate the effects due to dosing level, and we did not investigate correlations between dosing level and toxicity. At Stag, chronic toxicity to zooplankton increased with coagulant dosing, but toxicity did not change significantly with respect to algae growth, zooplankton mortality, fathead mortality (which was 100% in non-dosed samples), or any medaka toxicity metrics (Table 6). At Ski Run, chronic toxicity to zooplankton and acute toxicity to fathead increased with coagulant dosing (Table 6); coagulant treatment at Ski Run did not change algae cell count or acute zooplankton toxicity, and it decreased chronic fathead toxicity. Ski Run was the only site at which decreases in toxicity were observed for any toxicity metrics. The effect of coagulant dosing on toxic contaminants is not consistent across all sites, and the toxicity of treated stormwaters is likely due to the toxicity of the non-dosed stormwater contaminants and the resulting interactions of the stormwaters with the coagulants. For example, if the toxic agent is associated with suspended sediment particles, coagulation may well remove toxicity. If the toxic compounds are dissolved, toxicity will remain the same. However, if the coagulant remains in solution at toxic concentrations, toxicity to aquatic organisms will increase. Due to the subjective nature of jar testing, we were not able to compare the effects of individual coagulants on toxicity levels because it is possible that the observed effects across and within sites are due to the dosing level rather than an inherent tendency

ARTICLE IN PRESS S.E. Lopus et al. / Ecotoxicology and Environmental Safety 72 (2009) 1933–1941

of each coagulant to affect toxicity. Although toxicity effects often varied significantly within sites between coagulants (results not shown by coagulant), we considered these results to be possibly related to human error from determination of dosing level and did not analyze those differences in this paper. Algae cell count and fish mortality were the toxicity metrics most negatively affected by stormwater. Non-treated stormwaters reduced mean algae cell counts by about 50% and increased mean fish mortality by over 1000% (Table 3), but they did not significantly affect mean zooplankton mortality or reproduction, fathead minnow biomass, or medaka hatching success or days to hatch across all sites (Tables 3 and 4). In comparing toxicity between non-treated and treated stormwaters, only zooplankton reproduction showed significant negative changes across all sites with dosing (Table 6), which were most likely due to the coagulant treatment. Mean zooplankton brood sizes decreased by over 50% when stormwaters were treated with coagulants (Table 3). 5. Conclusions The standard USEPA 3-species toxicity test and a non-standard medaka test were used to assess the toxicity of stormwater before and after coagulant treatment aimed at decreasing fine particle and dissolved phosphorus concentrations for improved water clarity of Lake Tahoe, California. It was shown that untreated stormwater contained contaminants acutely or chronically toxic to some test species. The effects on different toxicity metrics varied between stormwaters. In general, non-dosed stormwaters reduced algae cell counts and increased fish mortality, whereas coagulants reduced zooplankton reproduction.

Acknowledgments The authors wish to thank Steve Peck, Russ Wigart, and Tim Delaney for their contributions to this project. Funding for this project was provided by the United States Department of Agriculture Forest Service (USDA Forest Service) through a grant managed by the City of South Lake Tahoe, California, USA. References Al-Mutairi, N.Z., 2006. Coagulant toxicity and effectiveness in a slaughterhouse wastewater treatment plant. Ecotoxicol. Environ. Saf. 65, 74–83. Bachand, P.A.M., Bachand, S.M., Heyvaert, A., 2005a. BMP treatment technologies, monitoring needs, and knowledge gaps: status of the knowledge and relevance within the Tahoe Basin. Technical Report, Office of Water Programs, California State University Sacramento, Sacramento, CA. Bachand, P.A.M., Heyvaert, A., Reuter, J., Werner, I., Bachand, S.M., Teh, S.J., 2007. Chemical treatment methods pilot (CTMP) system for treatment of urban runoff—Phase I. Feasibility and design, Final Report for the City of South Lake Tahoe. South Lake Tahoe, CA.

1941

Bachand, P.A.M., Trejo-Gaytan, J., Darby, J., Reuter, J., 2005b. Final report: smallscale studies on low intensity chemical dosing (LICD) for treatment of highway runoff. CTSW-RT-06-073.13.1. California Department of Transportation. Caltrans, 2001. New Technologies for Stormwater Treatment: Chemical Treatment to Improve Settling (CTIS), Reconnaissance Study. CTSW-RT-01-026. California Department of Transportation. Caltrans, 2002. Caltrans Tahoe highway runoff characterization and san trap effectiveness studies. Report CTSW-RT-02-44. CTSW-RT-02-44. California Department of Transportation. Caltrans, 2003. A review of the contaminants and toxicity associated with particles in a storm water runoff. Report CTSW-RT-03-059.73.15, CTSW-RT-03059.73.15, California Department of Transportation. Davis, A.P., Shokouhian, M., Ni, S., 2001. Loading estimates of lead, copper, cadmium, and zinc in urban runoff from specific sources. Chemosphere 44, 997–1009. Gnagy, M., 1994. Review of jar testing procedures and calculations. In: AWWA Ohio Section Conference. Guida, M., Mattei, M., Melluso, G., Pagano, G., Meric-, S., 2004. Daphnia magna and Selenastrum capricornutum in evaluating the toxicity of alum and polymer used in coagulation–flocculation. Fresenius Environ. Bull. 13, 1244–1247. Heyvaert, A.C., Reuter, J.E., Strecker, E., 2004. Evaluation of Selected Issues Relevant to the Design and Performance of Stormwater Treatment Basins at Lake Tahoe. Tahoe Research Group and Geosyntec Consultants. Lumos and Associates Inc., 2005. East Pioneer Trail Watershed hydrology study. Final Report, City of South Lake Tahoe and California Tahoe Conservancy, South Lake Tahoe, CA. Patyna, P.J., Brown, R.P., Davi, R.A., Letinski, D.J., Thomas, P.E., Cooper, K.R., Pakerton, T.F., 2006. Hazard evaluation of diisononyl phthalate and diisodecyl phthalate in a Japanese medaka multigenerational assay. Ecotoxicol. Environ. Saf. 65, 36–47. Reuter, J.E., Miller, W.W., 2000. Aquatic Resources, Water Quality, and Limnology of Lake Tahoe and its Upland Watershed. PSW-GTR-175. United States Department of Agriculture—Forest Service, Albany, CA. Rhodes, S., Farwell, A., Hewitt, L.M., MacKinnon, M., Dixon, D.G., 2005. The effects of dimethylated and alkylated polycyclic aromatic hydrocarbons on the embryonic development of the Japanese medaka. Ecotoxicol. Environ. Saf. 60, 247–258. Schuster, S., Grismer, M.E., 2004. Evaluation of water quality projects in the Lake Tahoe basin. Environ. Monit. Assess. 90, 225–242. Selcuk, H., Rizzo, L., Nikolaou, A.N., Meric- , S., Belgiorno, V., Bekbolet, M., 2007. DBPs formation and toxicity monitoring in different origin water treated by ozone and alum/PAC coagulation. Desalination 210, 31–43. Skinner, L., de Peyster, A., Schiff, K., 1999. Developmental effects of urban stormwater in medaka (Oryzias latipes) and inland silverside (Menidia beryllina). Arch. Environ. Contam. Toxicol. 37, 227–235. So¨rme, L., Lagerkvist, R., 2002. Sources of heavy metals in urban wastewater in Stockholm. Sci. Total Environ. 298, 131–145. Strecker, E., Howell, J., 2003. Preliminary water quality evaluation for Lake Tahoe TMDL. Draft Technical Report. GeoSyntec Consultants for Lahontan Regional Water Quality Control Board, South Lake Tahoe, CA. Swift, T.J., Perez-Losada, J., Schladow, S.G., Reuter, J.E., Jassby, A.D., Goldman, C.R., 2006. Water clarity modeling in Lake Tahoe: linking suspended matter characteristics to Secchi depth. Aquat. Sci. 68, 1–15. Tiefenthaler, L.L., Stein, E.D., Schiff, K.C., 2008. Watershed and land use-based sources of trace metals in urban storm water. Environ. Toxicol. Chem. 27, 277–287. Trejo-Gayton, J., Bachand, P.A.M., Darby, J., 2006. Treatment of urban runoff at Lake Tahoe: low-intensity chemical dosing. Water Environ. Res. 78, 2487–2500. US EPA, 2002. Short-Term Methods for Estimating the Chronic Toxicity of Effluents and Receiving Waters to Freshwater Organisms. EPA-821-R-02-013. USEPA Office of Water, Washington, DC. US EPA, 2007. ECOTOX User Guide: ECOTOXicology Database System. Version 4.0. Weilenmann, U., O’Melia, C.R., Stumm, W., 1989. Particle transport in lakes: models and measurements. Limnol. Oceanogr. 34, 1–18. Winer, R., 2000. National Pollutant Removal Performance Database for Stormwater Treatment Practices. Center for Watershed Protection for the EPA Office of Science and Technology in Association with Tetratech, Ellicott City, MD.