Responses of Fathead Minnows (Pimephales promelas) During Life-Cycle Exposures to Pulp Mill Effluents at Four Long-Term Receiving Water Study Sites

Transcription

1 Integrated Environmental Assessment and Management Volume 5, Number 2 pp Ó 2009 SETAC 275 Responses of Fathead Minnows (Pimephales promelas) During Life-Cycle Exposures to Pulp Mill Effluents at Four Long-Term Receiving Water Study Sites Dennis L Borton,*À Diana L Cook,` W Kenneth Bradley,À Raymond E Philbeck,À Monique G Dubé, Nancy J Brown-Peterson,jj and William R StreblowÀ ÀNational Council for Air and Stream Improvement, PO Box 12868, New Bern, North Carolina 28562, USA `National Council for Air and Stream Improvement, PO Box 458, Corvallis, Oregon 97339, USA University of Saskatchewan, Toxicology Centre, Saskatoon, Saskatchewan S7N 5B3, Canada jjdepartment of Coastal Sciences, The University of Southern Mississippi, 703 East Beach Drive, Ocean Springs, Mississippi 39564, USA (Received 15 July 2008; Accepted 13 January 2009) EDITOR S NOTE: This is 1 of 8 papers reporting on the findings of an ongoing study conducted by the National Council for Air and Stream Improvement (NCASI) to understand the relationship between pulp and paper mill effluents and ecological conditions in 4 receiving streams in the United States. The data reported in this Special Series were collected between 1998 and ABSTRACT We exposed fathead minnows (Pimephales promelas) to 7 concentrations of effluents from pulp mills at 4 Long-Term Receiving Water Study (LTRWS) sites. The primary objective of these investigations was to determine the potential for toxicity, particularly on fish reproduction, of the pulp mill effluents using laboratory tests. These tests were performed as LTRWS fish community assessments were being completed, thus results of the laboratory fish reproduction tests could be compared to in-stream fish community measurements. In general, bioindicators measured during the life-cycle tests, including gonadosomatic index (GSI), hepatosomatic index, condition factor, numbers of tubercles on heads of males and females, and gonadal histology did not show consistent patterns or dose response and did not predict effects on egg production. Gonadosomatic indexes and tubercles also did not indicate estrogenic or androgenic responses to the effluents during the life-cycle tests. The most consistently sensitive test endpoint showing a dose response was the 25% inhibition concentration (IC25) for egg production. Based on this endpoint all 4 effluents had effects on fish reproduction from 8% by volume to 100% effluent. However, in-stream effects on fish reproduction would not be expected based on these 4 life-cycle tests for any of the LTRWS stream sites from these tests. The mean effluent concentration in Codorus Creek was approximately 32%, and the IC25 for the life-cycle test was 100% effluent, providing a margin of safety of approximately 3 times. The margins of safety at the other sites are much greater: 34 times for Leaf River, Mississippi, USA (IC25 ¼ 69%, 2% mean receiving water concentration), 36 times for the McKenzie River, Oregon, USA (IC25 ¼ 18%, 0.5% mean receiving water concentration), and 40 times for the Willamette River, Oregon, USA (IC25 ¼ 8%, 0.2% mean receiving water concentration). Effects on fish numbers, diversity, and community structure due to the effluent were also not found during the LTRWS, which is consistent with these laboratory results. These findings indicate that in this case, when laboratory results combined with in-stream effluent concentrations suggest in-stream effects on fish population are not expected, the laboratory results are consistent with the in-stream observations. However, inferences about situations where laboratory results predict in-stream effects cannot be made from these data. Review Special Series Keywords: Pulp mill effluent Fathead minnow Reproduction Life-cycle Bioindicators INTRODUCTION Concerns have arisen regarding chronic sublethal effects of pulp and paper mill effluents, with emphasis on reproduction or reproductive fitness. Hewitt et al. (2006), McMaster et al. (2006), and Parrott et al. (2006) summarized and evaluated Canadian studies on pulp mill effluent effects and more recently provided a similar review of the global literature on reproductive effects of pulp mill effluents and sources of those effects (Hewitt et al. 2008). Laboratory studies to determine effects of pulp mill effluents on reproduction have frequently used either complete life-cycle studies with fathead minnows (Pimephales promelas) (Kovacs et al. 1995, 1996; Borton et al. * To whom correspondence may be addressed: dlborton@wildblue.net Published on the Web 1/15/ , 2004; Parrott et al. 2003; Parrott and Wood 2004) or shorter adult exposures (NCASI 1998; Martel et al. 2004; Rickwood et al. 2006; Rickwood and Dubé 2007). Several studies have begun with in-stream evaluations of fish populations followed by additional laboratory studies to determine if similar results would be observed. Results of those studies have been variable to date (Hewitt et al. 2008). The life-cycle studies described herein all took place within the same time period as the ongoing Long Term Receiving Water Study (LTRWS), which evaluated in-stream fish, macroinvertebrate, and periphyton communities (Hall, Fisher, et al. 2009; Flinders, Minshall, Hall, et al. 2009; Flinders, Minshall, Ragsdale, et al. 2009; Flinders, Ragsdale, et al. 2009). Few of the previous life-cycle tests with effluents took place during the same time period as in-stream studies, and thus these investigations are unusual in that respect. The primary

2 276 Integr Environ Assess Manag 5, 2009 DL Borton et al. Table 1. Dilution water source and characteristics during fathead minnow life-cycle bioassays Source, compound, or Codorus Creek Willamette River McKenzie River Leaf River parameter Units Mean 6 SD a Mean 6 SD Mean 6 SD Mean 6 SD Source Well water River water b River water c River water d ph ph units Conductivity lmhos/cm Hardness mg/l Alkalinity mg/l Total cadmium lg/l ND e [0.02] ND [0.02] ND [0.02] 0.1 Total chromium lg/l ND [0.2] Total copper lg/l Total lead lg/l ND [0.02] Total nickel lg/l ND [0.2] Total zinc lg/l a SD ¼ standard deviation. b Willamette River water modified for hardness and alkalinity. c McKenzie River water modified for hardness and alkalinity. d Leaf River water modified for hardness and alkalinity. e ND ¼ not detected [calibration limit]. objective of these investigations was to determine the potential for toxicity, particularly on fish reproduction, of the pulp mill effluents using laboratory tests. Since these tests were performed as the LTRWS evaluations were being completed, results of the laboratory fish reproduction tests could be compared to in-stream fish community measurements and used to determine if laboratory fish reproduction test results are consistent with observations of fish communities during parallel in-stream sampling. We completed a life-cycle bioassay with fathead minnows at each of 4 LTRWS streams to determine the effects of each effluent on fish reproduction. We also compared the laboratory results to in-stream effluent concentrations to determine if receiving water effects on reproduction would be expected based on the life-cycle results. METHODS Dilution water and effluent source and characterization Hall, Ragsdale, et al. (2009) provide a description of each mill and effluent used in the fathead minnow life-cycle tests and the timing of in-stream evaluations. Each laboratory lifecycle test was completed during at least one of the in-stream evaluations of the LTRWS. Laboratory life-cycle tests were completed in a laboratory trailer on-site using a continuous flow of final effluent after secondary biological treatment in aerated lagoons at the McKenzie River, Oregon, USA; Willamette River, Oregon, USA; and Leaf River, Mississippi, USA sites. An initial study at the Codorus Creek site failed because of a disease problem in the control water, therefore 4000 to 6000 L of the activated sludge secondary treated effluent were trucked every 10 to 15 d to the National Council for Air and Stream Improvement (NCASI) Southeastern Aquatic Biology Facility, and well water was used as the dilution water. We completed the fathead minnow lifecycle test with effluent from the McKenzie River mill between May and September McKenzie River water, with additions of 48 mg/l of NaHCO 3, 30 mg/l of CaSO 4 2H 2 O, 30 mg/l of MgSO 4, and 2 mg/l of KCl was the dilution water. The addition of these chemicals brought the dissolved solids to levels similar to those known to support fathead minnows in our cultures (Table 1). The Leaf River life-cycle test was completed from May to September 200l. The dilution water was taken from the mill water supply line after chlorination of water from the Leaf River. The water was dechlorinated with 1 mg/l Na 2 S 2 O 3 and we added 60 mg/l of CaSO 4 5H 2 O and 60 mg/l of MgSO 4. The Willamette River life-cycle test was completed from January to May Willamette River water with additions of 72 mg/l of NaHCO 3, 45 mg/l of CaSO 4 2H 2 O, 45 mg/l of MgSO 4, and 3 mg/l of KCl was the dilution water. The lifecycle test with Codorus Creek effluent was completed from July to December Characteristics of the control waters for each life-cycle test are shown after modification on Table 1. Control water was also the dilution water for each concentration. Conductivity, hardness, alkalinity, and ph were measured weekly over the exposure period. The control waters were also measured once during each life-cycle test for 27 metals including the 6 shown on Table 1. We measured the ph, hardness, alkalinity, and conductivity of the effluent weekly. Mill personnel measured biochemical oxygen demand (BOD) 3 to 5 times weekly (APHA 1998) (Table 2). Total organic carbon (TOC), chemical oxygen demand (COD), and color were measured 3 times during the life-cycle test, when short-term bioassays were performed on the effluents (Hall, Ragsdale, et al. 2009). Effluents were analyzed biweekly for resin acids, fatty acids, polyphenols, phytosterols (NCASI 1986, 1997), condensable tannins, and neutral semivolatile compounds. The polyphenol test is based on a folin phenol reagent reaction and provides an approximation of tannin and lignin content as well as other materials containing aromatic hydroxyl groups (APHA 1998). One

3 LTRWS: Fathead minnow life-cycle studies Integr Environ Assess Manag 5, Table 2. Effluent chemical characteristics during fathead minnow life-cycle bioassays Codorus Creek Willamette River McKenzie River Leaf River Compound or parameter a Units Mean 6 SD b Mean 6 SD Mean 6 SD Mean 6 SD TOC mg/l ph ph units Hardness mg/l Alkalinity mg/l BOD mg/l COD mg/l Color mg/l Conductivity lmhos/cm Polyphenols mg/l Condensable tannins mg/l TSS mg/l Neutral semivolatiles c lg/l Fatty acids d lg/l Resin acids e lg/l ND f [1.0] ND [1.0] Phytosterols g lg/l Total cadmium lg/l 0.2 ND [0.1] Total chromium lg/l Total copper lg/l Total lead lg/l Total nickel lg/l Total zinc lg/l a TOC ¼ total organic carbon; BOD ¼ biochemical oxygen demand; COD ¼ chemical oxygen demand; TSS ¼. b SD ¼ standard deviation. c Total of 16 neutral semivolatiles. d Total of 3 fatty acids. e Total of 7 resin acids. f ND ¼ not detected [calibration limit]. g Total of 4 phytosterols. effluent sample during each test was analyzed for 27 metals including the 6 shown in Table 2. Fathead minnow life-cycle test We completed fathead minnow life-cycle tests using proportional solenoid-valve diluters, modified with a solenoid switch to discharge the entire volume of the effluent, control water, or mixtures of each test concentration alternately into the A or B replicate test chamber. Chambers used for initial hatching and juvenile grow-out were each 33 cm 3 33 cm 3 16 cm deep and held 15 L of solution. Concentrations were 0%, 3%, 6%, 12%, 24%, 50%, and 100% effluent, and flow rates to each test chamber averaged approximately 1 L each 10 to 16 min cycle or about 7 turnovers per chamber per day. A 16-h light, 8-h dark photoperiod was maintained throughout the life-cycle test, and the light intensity at the water surface of each chamber was approximately 100 ft-c using broad-spectrum fluorescent lights mounted over the tanks. Temperature of effluent and dilution water was maintained at C in 1000 L holding tanks within the lab and the holding tanks were aerated to maintain DO. Temperature in each chamber was also maintained at C and dissolved oxygen was maintained above 6 mg/l by aeration. The temperature, dissolved oxygen, ph, conductivity, hardness, and alkalinity were measured in each tank and chamber once each week. Discharge volumes from each solenoid chamber were also measured once each week and adjusted as needed. All tanks were monitored at least twice daily for normal operation of the dilutor system and aeration and tanks were checked for any mortality at that time. Tanks were also cleaned weekly or as needed during the life-cycle test to remove extra food or waste. We have continuously maintained a fathead minnow culture at NCASI Southeastern Aquatic Biology Facility for over 20 y from stock originally obtained from US Environmental Protection Agency (USEPA), Duluth, USA, and periodically infused with additional stock from the North Carolina Department of Environment and Natural Resources or private

4 278 Integr Environ Assess Manag 5, 2009 DL Borton et al. Table 3. Results from a fathead minnow life-cycle bioassay with effluent from the Codorus Creek mill. Endpoint a Units Egg hatchability1 d % Day survival e % * 82* 76* 28-Day avg. weight f mg Day avg. biomass g g Eggs/female/day N Egg hatchability2 h % * * * Adult survival? % Adult survival / % Avg. final weight? mg Avg. final weight / mg Avg. GSI? * Avg. GSI / Avg. LSI? Avg. LSI / * Avg. K? Avg. K / Head tubercles? N Head tubercles / N POF / % Atresia i / % Mid maturation i? % Late maturation i? % * Significantly different from the control. a GSI ¼ gonad somatic index; LSI ¼ liver somatic index; K ¼ condition factor; POF ¼ histological assessment of postovulatory follicles. b SD ¼ standard deviation. c IC25 ¼ effluent concentration projected to cause of reduction of 25%. d Eggs from culture to start test. e Average survival of 2 groups of 50 fish. f Average weight of all fish weighed individually. g Average of the total weight of 2 groups of fish. h Eggs from fish exposed during the life-cycle test. i Based on histological assessments. suppliers. We transported eggs from that culture and placed the hatching chambers suspended in exposure tanks to initiate a fathead minnow life-cycle bioassay. A total of 100 eggs per replicate with 2 replicates per concentration was added to the chambers. Hatching occurred by day 4 or 5, when we counted the live larvae and calculated the percent hatch. We moved 50 of the larvae to grow-out chambers placed in the exposure tanks with an exchange rate of 6 to 8 times daily. The larvae were fed newly hatched (,24 h old) Artemia nauplii 3 times a day at the initial rate of 3 ml/chamber of a 1600/mL suspension. This amount was increased on a weekly basis to a maximum of 12 ml by day 23. On day 28 the fish were not fed and the juveniles were individually weighed. The same procedures as outlined for the first 28 d continued until day 56. The food was supplemented with 2 ml of 250 g/l frozen brine shrimp slurry on day 29, and increased to 5 ml by day 55. We moved the fish to a second exposure system of 61 cm 3 31 cm 3 31 cm deep glass aquaria on day 56 for the reproduction portion of the life-cycle test. The turnover rate of solution in each of the aquaria was 4 to 6 times per day. We continued to feed the fish frozen brine shrimp for the remainder of the test. When males began to show spawning coloration, usually at 75 to 85 d post hatch, we placed 1 male and 2 females in each of 4 spawning areas in each aquarium that were separated by stainless steel screens. Thus, 8 spawning groups in 2 replica aquaria were exposed to each concentration. We fed 3 ml/group of the 250 g/l frozen brine shrimp slurry twice daily to each spawning group. An 8-cm half section of 10-cm stainless-steel pipe was placed in each chamber as spawning substrate. Spawning usually started

5 LTRWS: Fathead minnow life-cycle studies Integr Environ Assess Manag 5, Table 3. Extended IC25 c % % % % % % % % % * *.100% * * * * * between 80 and 90 d post hatch and continued for 6 to 7 weeks before the test was terminated. We recorded the numbers of eggs and spawns, and mortality, daily throughout the spawning period. We measured total length (TL, mm) and weight (W, 0.1 g) of each adult fish, excised and weighed gonads (GW, 0.1 g) and livers (LW, 0.1 g), and confirmed the sex of each fish at termination by macroscopic inspection of the gonads. Several spawns from each test concentration were analyzed for egg hatchability. We collected spawns from as many different spawning chambers as possible and the collection of eggs continued throughout the reproduction period. Eggs were checked twice daily for viability and dead eggs were removed. The larvae were counted after the eggs hatched, and the percent hatch was computed. BIOINDICATORS We calculated the gonadosomatic index (GSI) of male and female fathead minnows at the termination of life-cycle testing with the formula GSI ¼ [GW/(W GW)] We calculated the hepatosomatic index (LSI) using the same formula but substituting LW for GW. We calculated condition factors (K) at the termination of the life-cycle exposure for the male and female fathead minnows using the formula: K ¼ (100)(W)/SL 3, where SL represents standard length. We preserved the heads in neutralized buffered formalin and counted the numbers of tubercles on the heads of males and females. Gonad histology Gonadal histological evaluations of male and female fathead minnows were completed at the termination of the life-cycle tests for Codorus Creek, Willamette River, and Leaf River effluents. We preserved the whole gonads in 10% neutral buffered formalin at the end of the life-cycle exposure. Tissues were rinsed in running tap water overnight, dehydrated in a series of graded ethanols, processed, and embedded in paraffin following standard histological protocols. Tissues were sectioned at 4 lm, placed on slides and stained with hematoxylin and eosin. Ovarian tissue was examined microscopically for reproductive class (early maturation, mid maturation, late maturation, and regression) according to Brown-Peterson (2002) and stages of oocyte development (primary growth, cortical alveolar, early vitellogenic, and late vitellogenic), post-ovulatory follicles (POF), and oocyte atresia according to Jensen et al. (2001). The percentage of each oocyte stage, as well as POF or atresia, was calculated for three 3100 fields of view and a mean percentage for each specimen was determined. Testicular tissue was examined microscopically for reproductive class (early, mid, or late maturation regression) and stages of spermatogenesis following Brown-Peterson et al. (2002). Statistical evaluations Inhibition concentrations (IC) were calculated at the 25% reduction concentration (IC25) for several endpoints during the life-cycle bioassays using a linear interpolation model (Efron 1982). The IC25 is a point estimate of the concentration of effluent that causes a 25% reduction from the controls and is frequently used to estimate effects for whole effluent toxicity tests (USEPA 1990). The IC25 from the linear interpolation model is reported for all of the endpoints of the fathead minnow life-cycle test except those described herein as bioindicators or gonadal histology. In addition, results of hypothesis testing with Fisher s exact test for egg hatchability and all survival results, and Dunnet s test for significant differences between control and treatments for 28 d growth, egg production, adult weights, and numbers of tubercles are reported (SAS Institute 2001). Endpoints such as 28 d total biomass, GSI, LSI, and K could not meet assumptions for parametric analyses, thus the Kruskal Wallis test was used for comparisons of treatments versus the control (SAS Institute 2001). Differences in the percentages of oocyte stages, POF, and atresia among effluent treatments were examined for females using ANOVA with a Bonferroni post hoc test. All percentage data were arcsine square root transformed prior to analysis. Correlations between the percentage of males in reproductive classes and the percentage of males containing spermatogonia and effluent treatments were examined with Spearman rho and Kendall tau-b. Differences in the percentage of males in reproductive classes among treatment groups were tested with a Kruskal Wallis ANOVA. Analyses of the gonad histology data were performed with SYSTAT 9.0 for Windows (SPSS version 11). All analyses were considered significant at p 0.05.

6 280 Integr Environ Assess Manag 5, 2009 DL Borton et al. Table 4. Results from a fathead minnow life-cycle bioassay with effluent from the Willamette River mill Endpoint a Units Egg hatchability1 d % 86 75* Day survival e % Day avg. weight f mg * * Day avg. biomass g g Eggs/female/day N * * Egg hatchability2 h % Adult survival? % Adult survival / % Avg. final weight? mg * Avg. final weight / mg Avg. GSI? * Avg. GSI / Avg. LSI? * Avg. LSI / * Avg. K? Avg. K / Head tubercles? N Head tubercles / N POF / % Atresia i / % Mid maturation i? % Late maturation i? % *SD ¼ significantly different from the control. a GSI ¼ gonad somatic index; LSI ¼ liver somatic index; K ¼ condition factor; POF ¼ histological assessment of postovulatory follicles. b SD ¼ standard deviation. c IC25 ¼ effluent concentration projected to cause of reduction of 25%. d Eggs from culture to start test. e Average survival of 2 groups of 50 fish. f Average weight of all fish weighed individually. g Average of the total weight of 2 groups of fish. h Eggs from fish exposed during the life-cycle test. i Based on histological assessments. RESULTS Effluent characterization Table 2 shows the characteristics of each effluent measured during the fathead minnow life-cycle tests. The Codorus Creek effluent had the lowest concentrations of components arising from pulping tree furnish (black liquor losses), including TOC, COD, color, polyphenols, condensable tannins, and phytosterols. The relatively low amounts of resin acids and fatty acids reflect the low liquor losses, although these and the low BOD also indicate excellent biological treatment in the activated sludge system. The highest amounts of many of these same components, including TOC, COD, color, conductivity, condensable tannins, and resin acids were found in the Leaf River effluent, and the Willamette River and McKenzie River effluents generally had intermediate levels of these components. One exception was the amounts of phytosterols, which were highest in the Willamette River effluent followed by the McKenzie River effluent. Biochemical oxygen demand was lowest in the Leaf River effluent, but the BOD levels indicate all effluents were well treated. Fathead minnow life-cycle results Tables 3 to 6 show the results of life-cycle tests with the 4 mill effluents. Effluents from Codorus Creek (Table 3), the McKenzie River (Table 5), and the Leaf River (Table 6) did not significantly reduce the hatchability of eggs from wellwater cultures placed in the effluents. Significant reductions were found at 3%, 50%, and 100% (v/v) of Willamette River

7 LTRWS: Fathead minnow life-cycle studies Integr Environ Assess Manag 5, Table 4. Extended IC25 c 75* 68*.100% 88* 93*.100% *.100% % * * 8% % % % * % % * * * effluent (Table 4) although a dose response is not apparent until effluent concentrations of 24% and higher. The hatch in 100% effluent was 79% of the control, resulting in IC25s for Willamette River effluent as well as the other 3 rivers that were.100% effluent. The survival of juveniles at 28 d was also significantly reduced in 50% and 100% of the Willamette River effluent (Table 4), possibly because of effluent exposure, although these survival rates are within the range of control survival for the 4 tests. Juveniles in 6%, 12%, and 24% of the Codorus Creek effluent had significantly reduced survival compared to controls, but there was no difference in juvenile survival at the higher effluent concentrations (Table 3). Interestingly, all juveniles exposed for 28 d to the McKenzie River effluent had significantly greater survival than controls (Table 5). The controls during the test with McKenzie River effluent had the lowest survival of the 4 control groups while survival of the fish in effluent approximated that of the other tests (Tables 3 to 6). The reduced survival of the McKenzie River controls probably also reduced competition and allowed those fish to grow larger than any other control group during the 28 d of exposure (Tables 3 to 6). This in turn was probably the primary reason the average weights of fish in all McKenzie River effluent groups were lower than the controls (Table 5), although a dose response was not evident. The 28 d average biomass shows the combined effect of survival and growth. The combination of low control survival, but high average weight in the controls (Table 5), resulted in average biomass being significantly higher at the 50% and 100% effluent concentrations during the McKenzie River life-cycle bioassay. There was no dose response for 28 d survival, weights, and biomass that could be related to effluent from Codorus Creek and Leaf River (Tables 3 and 6), and observed differences were probably not due to effluent exposure. A dose response was apparent for the Willamette River, with lower levels of all 3 parameters at the highest effluent concentration (Table 4). Effluent from the Codorus Creek mill had the least effect on fish reproduction, which was measured as eggs per female per day. Significant differences in egg production were not found for any Codorus Creek effluent concentration, although an IC25 of 100% effluent was calculated (Table 3). Leaf River effluent also did not have significant reduction of egg production in any effluent concentration, but the IC25 (69% effluent, Table 6) was lower than the Codorus Creek effluent. Reproduction effects of the McKenzie River and Willamette River effluents were greater than Codorus Creek or Leaf River, with significant effects at 50% and 100% effluent for the McKenzie River effluent (Table 5) and all concentrations of 12% and higher for the Willamette River effluent (Table 4). The IC25s were 18% and 8% effluent for the McKenzie River and Willamette River effluents, respectively. Eggs produced in effluent concentrations during each of the life-cycle tests generally hatched equally well to those in the control (Tables 3 to 6). An exception was the hatch rate of eggs from the Leaf River bioassay, which was significantly lower for eggs incubated in all but 12% and 24% effluent (Table 6). However, these lower hatch rates are similar to those in control groups from Codorus Creek and McKenzie River (Tables 3 and 5), and given the overall variability of this endpoint an effect of the effluent on hatchability is doubtful. Additionally, all of the IC25s for egg hatchability were greater than 100% effluent. All of the adult males survived during each of the 4 lifecycle bioassays (Tables 3 to 6). While there was female mortality in all tests, adult females survived at similar rates compared to controls during 3 of the tests (Tables 3 to 5). During the Leaf River bioassay, survival rates of females in 5 of the 6 effluent concentrations were significantly greater than the controls (Table 6). Survival of control females was lowest during this test compared with the other 3, which may explain the significantly higher survival in effluent-exposed females in Leaf River. Weights of adult males were significantly reduced in 50% and 100% effluent during the life-cycle bioassay with McKenzie River effluent and the IC25 for growth of males was 78% effluent (Table 5); this was the only test showing reduced male weights. A tendency toward higher but not significantly different weights of males in higher effluent concentrations occurred in the other 3 tests. Male fathead minnows had significantly increased weight in 6% and 50% effluent during the bioassay with effluent from the Willamette River mill (Table 4). Weights of males and females in the Codorus Creek effluent were about 50% greater than fish during other tests (Tables 3 to 6). The fish in control and Codorus Creek effluent exposure began spawning 3 to 4 weeks later than those during other tests and spawned less frequently which probably contributed to the increased

8 282 Integr Environ Assess Manag 5, 2009 DL Borton et al. Table 5. Results from a fathead minnow life-cycle bioassay with effluent from the McKenzie River mill Endpoint a Units Egg hatchability1 d % Day survival e % 72 91* 90* 95* 93* 28-Day avg. weight f mg * * * * 28-Day biomass g g * Eggs/female/day N * Egg hatchability2 h % * * * Adult survival? % Adult survival / % Avg. final weight? mg Avg. final weight / mg Avg. GSI? * Avg. GSI / Avg. LSI? * Avg. LSI / Avg. K? Avg. K / * Significantly different from the control. a GSI ¼ gonad somatic index; LSI ¼ liver somatic index; K ¼ condition factor. b SD ¼ standard deviation. c IC25 ¼ effluent concentration projected to cause of reduction of 25%. d Eggs from culture to start test. e Average survival of 2 groups of 50 fish. f Average weight of all fish weighed individually. g Average of the total weight of 2 groups of fish. h Eggs from fish exposed during the life-cycle test. weights. Final weights of females were never reduced compared to controls in all 4 tests, and the weights of females exposed to 50% and 100% of the Codorus Creek effluent were significantly greater than the controls. Like the males, there was a tendency toward higher final weight of females with increasing effluent concentrations although differences were not usually statistically significant (Tables 3 to 6). The GSIs of both males and females were never reduced compared to the controls during any test. The tendency was toward greater GSI for both males and females with increasing effluent concentration, and significant increases were found for males and females exposed to 100% of the Willamette River effluent (Table 4). Significant increases in GSI were also found for males in 6%, 50%, and 100% Codorus Creek effluent (Table 3), and for females from 6% through 100% of the Leaf River mill effluent (Table 6). Control GSI of Leaf River females was the lowest of the control groups for the 4 tests, but exhibited the least individual variation (Tables 3 to 6). The LSI of males tended to increase with increasing effluent concentration during 3 of the life-cycle bioassays. The exception was fish exposed to Leaf River effluent where LSI of males was approximately equal in all but 6% effluent, which was significantly reduced from the control (Table 6). The LSI of females tended to be reduced in effluent concentrations of 12% and higher of the Willamette River effluent and was significantly lower in 12% and 50% effluent (Table 4). The LSI of females from the other 3 life-cycle tests did not show any trends with the only significant change being a reduction in 3% Codorus Creek effluent (Table 3). Condition factor of both males and females was usually not significantly different from controls. Females exposed to Leaf River effluent were an exception to this observation, with K significantly greater in concentrations of 12% effluent and higher (Table 6). The numbers of tubercles on heads of male and female fatheads were evaluated as secondary sexual characteristics at the termination of all but the McKenzie River life-cycle tests. There were always from 16 to 20 tubercles on the males, and numbers were never significantly different from controls. Tubercles were not found on any females (Tables 3, 4, and 6). Based on gonadal histological analysis, all fish of both sexes in all concentrations in the Leaf River, Willamette River, and Codorus Creek tests were capable of spawning. All examinations found normal female reproductive development with continuous oocyte recruitment, a high percentage of late vitellogenic oocytes, and no abnormal oocytes. The percentage of POF varied among tests and treatments and was significantly higher in controls in the Leaf River test compared to the 50% and 100% effluent treatments (Table

9 LTRWS: Fathead minnow life-cycle studies Integr Environ Assess Manag 5, Table 5. Extended IC25 c % 94* 99*.100% * *.100% * *.100% 4 6 6* 3 6 4* 18% * % % % * * 78% % * 6), suggesting more control females spawned immediately prior to sacrifice than those in the higher effluent concentrations. Leaf River control females also had a significantly higher percentage of atretic oocytes than those exposed to 50% effluent. In contrast, females from Codorus Creek had a higher percentage of atretic oocytes at the highest effluent concentrations (Table 3). Overall, histological results for females indicate little effect of the effluents on reproductive potential, although high levels of Codorus Creek effluent appeared to increase atresia of vitellogenic oocytes while moderate levels of the Willamette River effluent resulted in lower numbers of POF and higher atresia rates. Males from all 3 tests exhibited normal spermatogenesis at all effluent concentrations and all males had spermatozoa in the sperm ducts. However, 3 control males in the Leaf River test and 1 male in the 50% effluent treatment in the Willamette River test had a few primary oocytes in the testis; the cause for this is unknown. In all tests, a higher percentage of males were in the late maturation histological class at the highest effluent concentrations (Tables 3, 4, and 6). Effluent concentration was negatively correlated with the percentage of males in the mid maturation class, but positively correlated with the late maturation class in all tests; these relationships were significant for the Leaf River and Willamette River tests. DISCUSSION During these life-cycle tests most of the measures of effects of the effluent such as egg hatchability, juvenile survival, juvenile growth, or juvenile biomass either did not show a dose response or were less sensitive than the egg production endpoint. Egg production is both the best direct estimator of effects of effluent on fish reproduction of the life-cycle test, and also is nearly always the most sensitive endpoint when pulp mill effluents are evaluated (Parrott et al. 2006). The IC25 for egg production was consistently the most sensitive estimator for reproduction effects during these tests, and this parameter indicates that no effects on fish reproduction would be expected in each of the receiving waters. The mean effluent concentration in Codorus Creek is about 32% (Hall, Ragsdale, et al. 2009), and the IC25 for the life-cycle test was 100% effluent, providing a margin of safety of approximately 3 times. The margins of safety at the other sites are much greater: 34 times for Leaf River (IC25 ¼ 69%, 2% mean receiving water concentration), 36 times for the McKenzie River (IC25 ¼ 18%, 0.5% mean receiving water concentration), and 40 times for the Willamette River (IC25 ¼ 8%, 0.2% mean receiving water concentration). The life-cycle tests were also performed during at least one of the in-stream fish and community sampling times of the LTRWS. During the in-stream sampling effects due to effluent were not found in any of the 4 rivers on community structure or density measurements likely to be related to reproduction, including species type and abundance (Flinders, Ragsdale, et al. 2009). Thus, the laboratory life-cycle tests results that found no effects on fish reproduction at the in-stream waste concentrations were consistent with those that found no effects on in-stream fish population measurements at each of the 4 LTRWS sites. These studies also provide information for an NCASI database on the effects of pulp mill effluents on fish reproduction, which we have used to evaluate relationships between effluent components or characteristics and reproduction during fathead minnow life-cycle bioassays. That database consists of over 25 life-cycle tests with several types of effluents, wood leachates, or individual effluent compounds. Within the data set of mill effluents, 2 of the effluents from the LTRWS, Codorus Creek with an IC25 of 100% effluent, and Leaf River with an IC25 of 69% effluent are among the upper 25th percentile with the least effect on egg production (NCASI unpublished). The life-cycle tests with the McKenzie River effluent (IC25 ¼ 18%) and the Willamette River effluent (IC25 ¼ 8%) resulted in the 2 lowest IC25s of the entire database of modernized mills. We have evaluated some of the possible reasons for the differences in effects among mill effluents. Pulping and black liquor components have been suggested as possible sources within mills that may alter fish reproduction (Hewitt et al. 2008). Measurements taken during these tests indicate that Codorus Creek effluent has the least amount of components originating from pulping or black liquor (Table 2), and the least amount of effects on reproduction of these 4 effluents. The low amounts of these components may explain the nearly negligible effect of the Codorus Creek effluent. We also note that the reproduction of the controls was the poorest of these 4 groups and the database as a whole. The extent this contributed to the low effect on reproduction of this effluent is unknown. However, the Leaf River effluent was highest of the 4 LTRWS mill effluents in pulping-derived components such as TOC, COD, color, and condensable tannins and was about equal to the Willamette River and McKenzie River effluents in polyphenols. The one component measured in these 4 effluents that parallels effects on fish reproduction is the amount of phytosterols. Phytosterols have been suggested as causing changes in fish reproduction in other studies as reviewed by Hewitt et al. (2008), although results have been variable and inconclusive. During studies with leachates from

10 284 Integr Environ Assess Manag 5, 2009 DL Borton et al. Table 6. Results from a fathead minnow life-cycle bioassay with effluent from the Leaf River mill Endpoint a Units Egg hatchability1 d % Day survival e % * 28-Day avg. weight f mg Day avg. biomass g g Eggs/female/day N Egg hatchability2 h % 83 69* 68* Adult survival? % Adult survival / % 79 94* 93* 94* 100* Avg. final weight? mg Avg. final weight / mg Avg. GSI? Avg. GSI / * * * Avg. LSI? * Avg. LSI / Avg. K? Avg. K / * * Head tubercles? N Head tubercles / N POF / % Atresia i / % Mid maturation i? % Late maturation i? % * Significantly different from the control. a GSI ¼ gonad somatic index; LSI ¼ liver somatic index; K ¼ condition factor; POF ¼ histological assessment of postovulatory follicles. b SD ¼ standard deviation. c IC25 ¼ effluent concentration projected to cause of reduction of 25%. d Eggs from culture to start test. e Average survival of 2 groups of 50 fish. f Average weight of all fish weighed individually. g Average of the total weight of 2 groups of fish. h Eggs from fish exposed during the life-cycle test. i Based on histological assessments. pine chips, Borton et al. (2006) reported no effect on fish reproduction when the same phytosterols were at least 4 times higher than concentrations at the IC25s in these bioassays. Also, NCASI (1999), using a similar fathead minnow life-cycle test, evaluated the effects of stigmasterol, a phytosterol found in pulp mill effluents, and found no effects on fish reproduction at much higher concentrations than those occurring during these tests (nominal concentration of 200 lg/l, mean measured exposure concentration of 74 lg/l). Thus, phytosterols are unlikely to have caused the effects on egg production found during these tests. As noted earlier, pulping sources and liquor losses are possible sources of components altering reproduction, but specific sources are unknown. If pulping is a source, then different furnishes (tree species) may have differing components that alter fish reproduction. Hall and LaFleur (1997) summarized literature indicating that naturally occurring compounds, including high molecular weight components, in leachates from forests could have a variety of effects on exposed biota. Hewitt et al. (2008) also suggests that although Canadian EEM studies provide strong evidence that responses found are not entirely dependent on variables such as types of wood feedstock, dilution, production, and biotreatment, wood furnish remains a possible source. The McKenzie River mill and the Willamette River mill have furnish that is primarily Douglas fir (Pseudotsuga menziesii) and western hemlock (Tsuga heterophylla), while Leaf River furnish is about 90% pine and 10% hardwood, and Codorus Creek furnish is approximately 55% hardwood and 45% pine (Hall, Ragsdale, et al. 2009). Borton et al. (2006)

11 LTRWS: Fathead minnow life-cycle studies Integr Environ Assess Manag 5, Table 6. Extended IC25 c % 100* 98*.100% * *.100% % % 59* 69*.100% % *.100% % % * * * * * * 0* reported that leachates from Douglas fir and western hemlock had greater effects than pine leachates on several marine and freshwater species during laboratory toxicity tests including fathead minnow reproduction during life-cycle tests. These results suggest the fir and hemlock furnish may have components with greater toxic effects. The wood was not pulped during the leachate studies, nor was the leachate biologically treated as would be the case in a mill. Even so, this suggests a possible source of the effects noted and should be further examined. The LSIs of male fathead minnows tended to increase with increasing mill effluent, for 3 of the life-cycle tests, but not during the test with Leaf River effluent. Borton et al. (2003) found a weak but significant positive correlation between the LSI of male fathead minnows and effluent concentrations during 4 fathead minnow life-cycle tests. Similar to these findings, significant differences usually only occurred at high concentrations and LSI was not a useful predictor of actual reproduction during the tests. Male K was never significantly different in any treatment, and K of females was significantly greater in 100% McKenzie River effluent and all concentrations from 12% through 100% of the Leaf River effluent. Lowell et al. (2004) and Hewitt et al. (2008) indicated that fish exposed to pulp mill effluents tended to have decreased gonad size and increased K. During these studies these 2 factors did not occur together, nor did they correlate well to reproduction effects as measured in egg production. Both K and GSI were greater at effluent concentrations of 12% or higher of Leaf River effluent. Also, since the Leaf River effluent had an IC25 for egg production of 69% effluent, changes in K and GSI were not predictive of reproduction. Similar to an earlier study by Borton et al. (2003) none of the bioindicators measured, including GSI, LSI, and K, numbers of tubercles, or gonad histology were reliable indicators of effects on reproduction during these tests. Also, since the number of tubercles was never significantly different between controls and effluent-exposed groups, neither GSI nor numbers of tubercles indicated androgenic or estrogenic effects of the effluent. Many studies have reported that the size of gonads of fish exposed to pulp and paper mill effluents have been reduced and these studies were also summarized by Hewitt et al. (2008). Fathead minnows have been shown to be capable of this response, both by specific estrogenic or androgenic substances (Giesy et al. 2000; Ankley et al. 2001) or by pulp mill effluents (Parrott and Wood 2004). During the 4 lifecycle tests reported here, GSIs of both male and female fathead minnows were never reduced compared to the controls and there was a tendency toward larger gonads of males and females with increasing effluent concentrations. The reasons for larger gonads are unknown; however, less recent spawning activity or reduced overall spawning activity can result in larger gonads in fathead minnows, at least in females (Jensen et al. 2001). Histopathological assessment of gonadal tissues is important for understanding the reproductive health of fishes, and changes in developmental stages, oogenesis, spermatogenesis, and atresia can be useful for identifying the reproductive health of fishes as well as evaluating sublethal stressors (Blazer 2002). Overall, males appear to be slightly more affected by pulp mill effluents than females based on histological findings in these tests. For all tests, a high percentage of males at the highest effluent concentrations were in the late maturation class. Males in the late maturation reproductive class are typically found during the latter portion of the reproductive season, when active spermatogenesis is reduced (Brown- Peterson et al. 2002). This suggests that life-long exposure to effluent concentrations below the margin of safety found using egg production may result in earlier testicular maturation than in unexposed fish. For females, differing percentages of POF and atresia in control versus fish exposed to effluents may be useful histological indicators. Changes in the percentage of oocyte atresia have been demonstrated for fathead minnow exposed to a variety of chemicals (Miles- Richardson et al. 1999; Lange et al. 2001; Ankley et al. 2005; Kunz et al. 2006), and significant differences in oocyte atresia were found in 2 of the 3 tests examined here. A higher percentage of POF suggests a more recent spawning of that group and is commonly used as an indicator of higher spawning frequency than other groups of fish (Brown- Peterson 2002). For instance higher percentages of POF were found in control females during the Leaf River study. The actual number of spawns in the 2 highest treatment groups during the Leaf River study was slightly but not significantly reduced compared to controls or other treatments, indicating a trend in numbers of spawns that may be consistent with the POF observations. Furthermore, GSI values are lowest in