Efficacy of Copper Based Algaecides for Control of Quagga and Zebra Mussels

Transcription

1 Effcacy of Copper Based Algaecdes for Control of Quagga and Zebra Mussels Prepared by: Renata Claud M.Sc., T.H. Prescott P.Eng., Sergey Mastsky Ph.D. & Heather Coffey M.Sc. January 1, County Road 35, RR#2, Pcton, Ontaro, KK 2T Phone: Fax:

2 Effects of copper based algaecdes on mussels Table of Contents 1. Introducton Scope Methodology Methodology at Davs Dam for dose response curves of quagga mussels Methodology at San Justo Reservor for dose response curves of zebra mussels Methodology for a short term exposure and recovery test to determne post-exposure mortalty Methodology for statstcal modellng Results Effects of Green Clean products on dressend mussels Effects of copper sulfate on dressend mussels Copper sulfate and zebra mussel mortalty durng 96h exposure n July Effect of copper sulfate on zebra mussel mortalty durng 96h exposure n December Effect of copper sulfate on zebra mussel mortalty followng 12h exposure n December Effect of copper sulfate on quagga mussel mortalty durng 96h exposure n November Effect of copper sulfate on quagga mussel mortalty followng 12h exposure n November Modelng the response of zebra mussels to copper sulphate at 96h exposure Modelng the response of quagga mussels to copper sulphate at 96h exposure Effects of Natrx on dressend mussels Effect of Natrx on zebra mussel mortalty durng 96h exposure n July Effect of Natrx on zebra mussel mortalty durng 96h exposure n December Effect of Natrx on zebra mussel mortalty followng 12h exposure n December Effect of Natrx on quagga mussel mortalty durng 96h exposure n November Effect of Natrx on quagga mussel mortalty followng 12h exposure n December Modelng the response of zebra mussels to Natrx at 96h exposure Modelng the response of quagga mussels to Natrx at 96h exposure Effects of Captan lqud copper algaecde on dressend mussels Captan and zebra mussel mortalty durng 96h exposure n December Captan and zebra mussel mortalty followng 12h exposure n December Effect of Captan on quagga mussel mortalty durng 96h exposure n December Effect of Captan on quagga mussel mortalty followng 12h exposure n December Modelng the response of zebra mussels to Captan at 96h exposure Modelng the response of quagga mussels to Captan at 96h exposure Effects of EarthTec algaecde on dressend mussels Effect of EarthTec on zebra mussel mortalty durng 96h exposure n December Effect of EarthTec on zebra mussel mortalty followng 12h exposure n December... 4 p. 2

3 Effects of copper based algaecdes on mussels Effect of EarthTec on quagga mussel mortalty durng 96h exposure n December Effect of EarthTec on quagga mussel mortalty followng 12h exposure n December Modelng the response of zebra mussels to EarthTec at 96 h exposure Modelng the response of quagga mussels to EarthTec at 96 h exposure Dscusson Possble confoundng factors: Temperature and Water Qualty Seasonal varaton n zebra mussel mortalty Dfferences n effect of algaecdes: quagga vs. zebra mussels Role of retenton tme Effect of temperature Conclusons and Management Recommendatons References Appendx 1 Tables of Envronmental Parameters Appendx 2 Algaecde Calculaton Tables p. 3

4 Effects of copper based algaecdes on mussels 1. Introducton Dressend mussels, the zebra and quagga mussels, arrved n the Unted States from Europe n the 198s and quckly spread to many Eastern waterways, rvers, and lakes. These mussels are extremely prolfc and can produce costly mpacts by attachng to and cloggng water ntakes, trashracks, ppes, fre control systems, coolng water systems, fsh screens, and vrtually all types of underwater nfrastructure. Snce 27, dressend mussels have been present n the lower Colorado Rver. The mussel populatons have prolferated and mussels are now adversely affectng the Hoover, Davs, and Parker Dams. More recently, dressend were found n Calforna, Kansas, Nebraska, and Oklahoma and have been detected n New Mexco, and Utah. Varous aquatc herbcde formulatons have been reported as havng a negatve effect on dressend mussels durng aquatc weed and algae control treatments. The objectve of ths study was to determne the effcacy of some frequently used algaecdes on mussel mortalty. Selected for ths tral was Green Clean Sodum Carbonate Peroxyhydrate Powder (27.6% Hydrogen Doxde by weght), Green Clean Lqud (27% Hydrogen Doxde), SePRO Corporaton Natrx and Captan, non-propretary copper sulfate crystal and EarthTec algaecde/bactercde. Copper products (copper sulfate and copper carbonates or chelates) can be used to control mollusks n open water systems, but requre a Specal Local Need Label (also known as a Secton 24-c) ssued by the USEPA. 2. Scope The purpose of the experment descrbed n ths report was to evaluate the effcacy of algaecde formulatons to control dressend mussels at concentratons normally used for the control of aquatc weeds and algae. The experment was done n the followng steps; 1. Determne dose/exposure response curves of adult quagga mussels to selected algaecdes under temperature regmes whch may be encountered durng weed/algae control applcatons. The source water for the tests was the Lower Colorado Rver Water (Arzona). 2. Determne dose/exposure response curves of adult zebra mussels to selected algaecdes under temperature regmes whch may be encountered durng weed/algae control applcatons. The source water for the tests was San Justo Reservor water (San Bento County, Calforna). 3. As most algaecdes are not appled for a perod of 96 hours n the open envronment, addtonal tests were done to determne f a short exposure to any product whch caused at least 6% mortalty n the 96 hour test product wll result n delayed post exposure mortalty of adults. Exposure of eght and twelve hours was followed by a recovery perod of 36 hours. Mussel mortalty was assessed every 12 hours. p. 4

5 Effects of copper based algaecdes on mussels 3. Methodology The experments were carred out n moble flow-through laboratores stuated next to the two sources of water. One laboratory was at Davs Dam (Fgure 1) on the Lower Colorado Rver to test the effects on quagga mussels, and the second laboratory was stuated at San Justo Reservor to test the response of zebra mussels. The same research protocol was followed at both locatons, wth ambent water temperatures between 16.5 and 19.5 ºC. Fve dfferent algaecdes were tested: GreenClean Sodum Carbonate Peroxyhydrate Powder (27.6% Hydrogen Doxde by weght) and Green Clean Lqud (27% Hydrogen Doxde). Both products were tested at the maxmum strength approved for use n open water. Ths translates to the equvalent treatment of 9 lbs/ acre-foot for the powder and 3 gal/acre-foot for the lqud. Ths product dsspates wthn 12 hours n the envronment. On the advce of the manufacturer we lmted the experment to a 12 hour exposure of adult mussels. Copper sulfate s a chemcal compound wth the chemcal formula CuSO 4. Ths salt exsts as a seres of compounds that dffer n ther degree of hydraton. The anhydrous form s a pale green or gray-whte powder, whereas the pentahydrate (CuSO 4 5H 2 O), the most commonly encountered salt, s brght blue. Ths form s commonly used as an aquatc herbcde. The chemcal data sheets suggest a maxmum of 2 ppm of product for algae treatment, and guarantee a mnmum of 25% copper ons n the product. Therefore the maxmum recommended applcaton s.5 mg/l copper equvalent. Natrx (SePro Corporaton) s a copper carbonate ethanolamne complex orgnally lsted as an algaecde. It now has an EPA specal use label n some states for control of nvasve mollusks. For the mollusk control applcatons the recommended applcaton rates are hgher than for algae control. As our purpose was to test collateral damage durng treatment for algae the maxmum recommended dose recommended for algae control was chosen as the upper lmt of applcaton (1 mg/l copper equvalent). Captan (SePro Corporaton) s a double chelated copper compound contanng 9% copper by weght. It s prmarly used for the control of planktonc and flamentous algae. There are no drnkng water or fsh consumpton restrctons wth Captan-treated waters. The maxmum applcaton rate s 1 mg/l of copper. EarthTec algaecde/bactercde, manufactured by Earth Scences Laboratores Inc., contans 2% copper pentahydrate n lqud form from whch 5% metallc copper content s derved. The product has a low ph. It s used for the control of planktonc, flamentous and chara algae n lakes, ponds, fsh farms, fountans and other water systems. EarthTec s regstered by the EPA n all 5 states and certfed to ANSI/NSF Standard 6 for addton to drnkng water. The maxmum allowed applcaton rate s 1 mg/l of copper. Most of the copper based algaecdes have a maxmum suggested applcaton of 1 mg/l of copper (Cu 2+ ). Therefore, two copper levels were tested n ths experment: a hgh level of 1 mg/l of copper and a low level of.5 mg/l of copper. The notable excepton s copper sulfate, whch has a recommended applcaton rate of 2 mg/l of product, whch s.5 mg/l of copper equvalent. p. 5

6 Effects of copper based algaecdes on mussels Ths was the value used for the hgh concentraton whle the low concentraton was.25 mg/l of copper. In ths report, mg/l s always used to refer to the mg/l of copper equvalent n order to be able to drectly compare effects of dfferent products. The copper levels to be used n the experment were calculated n the same manner as would be done when carryng out an algaecde control campagn. In order to obtan the hghest, manufacturer recommended level of copper equvalent, we calculated the weght or volume of each product to be added to 5 gallons of raw water n each drum usng the manufacturer provded product label sheets. Appendx 2 shows the worksheets used to calculate the amounts of product to be used. The second treatment wth each product was carred out at 5% of the maxmum recommended level. For copper sulfate n granular form,.37 mg was dssolved n 5 ml of raw water and added drectly to the drum for the hgh copper concentraton, and the low concentraton was acheved usng.18 mg of dssolved copper sulfate. For the other algaecdes, 1 ml of product was frst dssolved n 1 ml of fltered lake water to mnmze expermental error on product addton as the amounts of product needed at full concentraton were very small. A precson ppette was used to measure the dluted product nto drums contanng 5 gallons of raw water. The amount of dluted product and the resultant copper concentraton s shown n Table 1. Table 1. Amount of product used (mg of product or ml of dluted product soluton n 5 gallons of raw water) for each algaecde and the resultng copper concentraton. Copper Sulfate Natrx TM Captan TM EarthTec amount amount amount Treatment copper copper copper amount of copper of of of Level conc. conc. conc. product conc. product product product Low.18 mg.26 mg/l 8.7 ml.5 mg/l 8.7 ml.5 mg/l 8.3 ml.5 mg/l Hgh.37 mg.52 mg/l 17.4 ml 1. mg/l 17.4 ml 1. mg/l 16.7 ml 1. mg/l Please see Appendx 2 for the calculaton tables used to determne algaecde amounts. After the algaecde was added, a drll mounted pant mxer was used for thorough mxng (Fgure 2). p. 6

7 Effects of copper based algaecdes on mussels Fgure 1. Feld Laboratory at Davs Dam Fgure 2. Mxer p. 7

8 Effects of copper based algaecdes on mussels 3.1 Methodology at Davs Dam for dose response curves of quagga mussels Adult mussels were collected from the docks at Kathernes Landng Natonal Park. The alkalnty at ths locaton was measured as 13 mg/caco 3, the calcum concentraton was approxmately 8 mg Ca/L, and ph vares from 7.9 to 8.6. The mussels were sorted mmedately, dscardng empty shells and crushed ndvduals. Groups of 1 (±1) adults were placed nto ndvdual mesh bags. The bags contanng adults were placed n aerated lake water to acclmate to laboratory condtons over 48 hours. Natural clumps of mussels were kept as ntact as possble to mnmze the stress on the adults and to evaluate f algaecdes would cause de-clumpng. De-clumpng n dressends has been observed as a sub-lethal response to noxous envronments. Fve 5 gallon drums n the laboratory were flled wth raw water from the Colorado Rver and allowed to stablze over 24 hours. The temperature was mantaned by mantanng the ambent ar temperature n the laboratory to match the desred temperature of the water n the drums (Fgure 3). Then the algaecdes were added (see secton 3.). Fgure 3. Drums and assocated coolers for Algaecde Experment. Followng the mxng process, the drp valves nstalled on the bottom of each drum were opened and the soluton from each drum started flowed nto ndvdual coolers (Fgure 4). When coolers were flled to approxmately 5%, 3 mesh bags contanng 1 (± 1) adults each were placed n p. 8

9 Effects of copper based algaecdes on mussels each cooler. Each bag contaned a waterproof label to dentfy the ndvdual bags n each cooler as 1, 2 or 3 (Fgure 5). On ntal exposure, clumps of quagga mussel were observed flterng and ndvduals were movng n the bag. Fgure 4. Cooler for captve adults Fgure 5. Mesh bag contanng captve adults p. 9

10 Effects of copper based algaecdes on mussels The drp valves contnuously passed soluton from the drum nto the assocated 4 L cooler over a perod of 96 hours at a rate of approxmately 2 L/hour. The dscharge from each cooler was collected nto buckets for dsposal at the evaporaton basn at Davs Dam. Each bag was examned every 12 hours to determne f adults were dyng. The dead mussels were removed from the bag, counted and the bag was returned to the cooler. At that tme, the temperature, ph and dssolved oxygen measurements were taken usng a Hach HQ4d multprobe. All mussels remanng n the bags were removed from the test coolers after 96 hour exposure and placed n flow-through recovery coolers. Post-exposure mortalty was montored for up to 12 hours. At the start of the experment, a clump of adult mussels was placed n a glass beaker contanng water wth each concentraton of each test product. The mussels were observed for flterng behavor. The exceptons to the above protocol were the experments usng the GreenClean Sodum Carbonate Peroxyhydrate Powder (27.6% Hydrogen Doxde by weght) and Green Clean Lqud (27% Hydrogen Doxde). Both products were tested at the maxmum strength approved for use n open water. These products dsspate wthn 12 hours n the envronment. On the advce of the manufacturer we lmted the experment to a 12 hour exposure of adult mussels wthn the drums used to mx the chemcals. Three bags of approxmately1 mussels were placed n each drum for 12 hours, then removed and evaluated for mortalty. The expermental temperature was 25ºC. 3.2 Methodology at San Justo Reservor for dose response curves of zebra mussels Adult mussels were collected from varous settlng substrates n the San Justo Reservor, and kept n a large drum wth contnuous flow-through water. The protocol from Davs Dam was then followed for the remander of the experment. The dscharge from the experment was collected nto a soakng pt outsde of the traler. The alkalnty of San Justo Reservor water was between 8 and 85 mg/caco 3 and the calcum concentraton was between 18 and 2 mg Ca/L, ph vared from 7.9 to Methodology for a short term exposure and recovery test to determne postexposure mortalty As most algaecdes are not appled for a perod of 96 hours n the open envronment, we devsed the followng test to determne f a short exposure to a product wll result n delayed post exposure mortalty of adults. Only products whch after 96 hours had caused mortalty above 6% were tested n the short term recovery experments. In these experments, the product was mxed nto 5 gallon drums as descrbed n secton 3.1. Into each drum, we placed 6 bags contanng adult mussels. Each bag conssted of approxmately 5 or 1 adult mussels, dependng on the experment. The groups of mussels were contaned n labelled mesh bags. We wthdrew 3 bags of mussels from the drum wth product beng tested after 8hrs, and 12hrs, rnsed them n raw water, examned the adults for mortalty, removed any dead ndvduals and placed the bags n flow through recovery coolers. p. 1

11 Effects of copper based algaecdes on mussels We examned each bag every 12 hours untl the 36-hour recovery tme was reached followng the longest exposure (12 hours), a total of 48 hours. We montored temperature, ph and dssolved oxygen n the drum and n the recovery coolers. 3.4 Methodology for statstcal modellng Statstcal modelng was used to understand the effects of our experments whle controllng for covarates such as temperature. Whle the effects of the algaecdes may appear to be evdent, there was some varablty n temperature both between and wthn experments (.e. each concentraton tested was n a separate cooler). Whle we attempted to control temperature, statstcal modelng was used to ensure that the apparent effects were ndeed true effects after controllng statstcally for temperature. Statstcal modelng was also used to understand whch covarates played the bggest role n determnng mussel mortalty. Our ntal models ncluded concentraton, month (July or November for some algaecdes), and temperature. We reduced our ntal models to the fnal models presented n the results based on whch varables were statstcally sgnfcant. In other words, the modelng enabled us to say that n some cases not all varables were useful n predctng mussel mortalty. Dependng on the propertes of the mortalty data at hand, we used two dfferent approaches for modelng the response of dressends exposed to algaecdes. If there was no temporal autocorrelaton present n the mortalty values, we used the generalzed lnear models (GLMs) wth the log lnk functon (a.k.a. logstc regresson ; Venables and Rpley 22). When the autocorrelaton was present, we used the generalzed addtve models (GAMs) wth correlated resduals (Wood 26). The two approaches are descrbed below n more detal. GLMs were ftted to the dressend mortalty data n order to derve equatons descrbng the change of mortalty over tme, and also to test whch of the measured covarates (e.g., concentraton, temperature, etc.) nfluenced that change. Example of a typcal ftted model s as follows: exp( 1 exp( The terms of ths model are nterpreted n the followng way: p h h 1 1 C ) C ) p s the th cumulatve observed proporton of dead dressends observed by tme t. The modeled proporton was calculated after poolng the data from three replcate mesh bags together. 1 s the effect of tme (h),.e. a coeffcent showng how much the proporton of dead molluscs ncreases when tme ncrements by 1 hour. 2 s the effect of water temperature (t). The temperature was centered by subtractng ts mean value from each of the measured ndvdual temperature values, makng the mean of the resultng values equal to zero. The man reason for performng such a transformaton was to ease the nterpretaton of the model ntercept, (see below). After centerng, the meanng of the coeffcent 2 s as follows: t ndcates how much the t t p. 11

12 Effects of copper based algaecdes on mussels proporton of dead molluscs changes when temperature devates (.e., ncreases or declnes) from ts mean value by one centgrade degree. 3 s the effect of algaecde concentraton (C), whch was treated as a categorcal predctor and thus was coded n the model as a dummy varable. Thus, wth the two concentratons tested, the varable concentraton takes on only two possble values f t s low, and 1 f t s hgh. When relevant, other categorcal predctors (e.g., month) were coded smlarly. s the model ntercept. Takng nto account the aforementoned meanngs of the other model parameters, the ntercept s nterpreted as the expected dressend mortalty at the hgh algaecde concentraton at tme, wth the water temperature kept at ts mean level. The analysed mortalty values demonstrated an excessve varance as compared to the varance that would be expected from a varable wth bnomal dstrbuton. To account for ths overdsperson, we ftted logstc regresson models of the so called quasbnomal famly usng the maxmum lkelhood methods of parameter estmaton (Venables and Rpley 22). The devance test (Venables and Rpley 22) was employed to decde whch parameters were to be kept n the model and whch of them had only mnor effect on mortalty and should be left out. In the Results secton, we present only the fnal models, whose parameters were found to have sgnfcant statstcal assocaton wth dressend mortalty. Adequacy of the fnal models was checked by examnng the dstrbuton of ther standardsed resduals (.e. dfferences between the actual observatons and model-ftted values, dvded by ther standard devaton). The resduals of the ftted models are assumed to be normally dstrbuted. Ths assumpton was checked usng the quantle-quantle plots (Q-Q plots), whch vsualse the actual quantles of resduals aganst the theoretcally expected normal quantles. If the normalty assumpton s met, data ponts on a Q-Q plot form approxmately a 45 dagonal lne (Venables and Rpley 28). To further support ths vsual check quanttatvely, we also used the Shapro-Wlks test for normalty (Shapro and Wlk 1965). GAMs represent a class of sem-parametrc statstcal models, whch are partcularly sutable for modelng non-lnear relatonshps. An example GAM ftted to the dressend mortalty data from ths study s as follows: p exp( 1 exp( f ( h ) f ( h ) C ), C ) where all parameters have the same nterpretaton as n the GLMs descrbed above, except for the nonparametrc term f(h ). In ths partcular example, f(h ) represents a splne smoother that models a nonlnear response of mortalty to tme. Ths smoothng functon was estmated as a cubc regresson splne (Wood 26). Resduals of the ftted GAMs were allowed to be correlated through the followng submodel ( autoregressve model of order 1, AR-1), whch relates a resdual at tme h to the resdual at tme h 1, along wth nose (Wood 26): t t h h h 1 h The parameter s unknown, and s to be estmated from the data. The AR-1 model corresponds to the followng correlaton structure: p. 12

13 Effects of copper based algaecdes on mussels cor(, ) 1 s h h s f s else h Suppose,.5 and h = s + 1. Then the correlaton between resduals separated by one hour s.5. If the separaton s two unts n tme, then the correlaton s.5 2 =.25. The further apart two resduals are, the less they are correlated. Ths autocorrelaton structure was appled on the mortalty tme seres for each algaecde concentraton separately (e.g., hgh and low ), and then one average estmate of was obtaned for both tme seres (Wood 26). Adequacy dagnostcs of the ftted GAMs was performed usng the same tools as for the GLMs (see above). GLMs were ftted usng standard functonalty of the R v2.15. statstcal software (R Development Core Group 212). GAMs were ftted wth the help of add-on package mgcv for R (Wood 26). 4. Results 4.1 Effects of Green Clean products on dressend mussels The experment started on May 21, 212. The test chemcals were the GreenClean Sodum Carbonate Peroxyhydrate Powder (27.6% Hydrogen Doxde by weght) and Green Clean Lqud (27% Hydrogen Doxde). Both products were tested at the maxmum strength approved for use n open water. Ths translates to the equvalent treatment of 9 lbs/acre-foot for the powder and 3 gal/acre-foot for the lqud. Over the course of the experment, the expermental temperature ranged from 25.3ºC to 28.ºC. Ths product dsspates wthn 12 hours n the envronment. On the advce of the manufacturer we lmted the experment to a 12 hour exposure of adult mussels. No mortalty was recorded n ether of the expermental treatments or n the control. Further, there was no de-clumpng of the adult mussels (de-clumpng s a sgn of mussels under stress from the envronment). None of the shells of the adults were bleached or compromsed n any way. Table 2. Envronmental parameters durng Green Clean experment Start 1am Fnsh-1pm Drum Temp DO ph Temp DO ph A C D Due to lack of effcacy, these products were not tested at San Justo Reservor on zebra mussels. p. 13

14 Precentage of mussels dead (%) Effects of copper based algaecdes on mussels 4.2 Effects of copper sulfate on dressend mussels Two separate exposure experments usng copper sulfate were carred out on zebra mussels. The frst experment was conducted n July 212. The second experment was carred out n December 212. The reason for repeatng the experments n December was the addton of two copper algaecdes to the testng protocol after July 212. To be able to compare results for all four algaecdes tested under the same temperature regme, the experment wth copper sulfate was repeated. It s mportant to remember that the hgh concentraton of copper sulfate s n fact half the g/ml copper equvalent of other algaecdes, based on recommended applcaton rates Copper sulfate and zebra mussel mortalty durng 96h exposure n July The frst experment started on July 9 th, 212 at San Justo Reservor. The results of the July recovery experment are shown n Fgure 6. Temperatures n the drum wth the low copper level averaged 18.5 C, wth a mnmum of 15.7 C and a maxmum of 22.8 C. The drum wth hgh copper level had an average temperature of 18.6 C wth a mnmum of 15.9 C and maxmum of 22.8 C. For the complete record of envronmental parameters see Appendx h end of exposure mg/l Bag 1.26 mg/l Bag 2.26 mg/l Bag 3.52 mg/l Bag 1.52 mg/l Bag 2.52 mg/l Bag Tme (h) Fgure 6. Zebra mussel mortalty when exposed to copper sulfate n July The July experment had an average mortalty of.8% after 96h at the low copper concentraton, and an average mortalty of 25.% after 96h at hgh copper levels. It was noted that mussels were p. 14

15 Precentage of mussels dead (%) Effects of copper based algaecdes on mussels contnung to experence mortalty after they were removed from the expermental coolers and placed n recovery chambers. Recovery chambers contnued to be montored and after 148h mortaltes rose to 6.7% for low and 7.% for hgh copper Effect of copper sulfate on zebra mussel mortalty durng 96h exposure n December The experment usng copper sulfate was repeated n December, 212. After the 96 hour exposure to low copper concentraton the maxmum mortalty vared from 4.5 to 23%. At the hgh copper concentraton the maxmum mortalty vared from 29 to 47%. After an addtonal 6 hours n the recovery cooler the maxmum mortalty for low copper rose to a mnmum of 5.26% and a maxmum of 31%. At the hgh copper level the mnmum mortalty ncreased to 4% and maxmum mortalty reached 72% (Fgure 7). Durng the expermental exposure the ambent water temperature vared from a hgh of 21.5ºC to a low of 15.6 ºC. There was no mortalty recorded n the control h end of exposure mg/l Bag 1.26 mg/l Bag 2.26 mg/l Bag 3.52 mg/l Bag 1.52 mg/l Bag 2.52 mg/l Bag Tme (h) Fgure 7. Zebra mussel mortalty when exposed to copper sulfate n December Effect of copper sulfate on zebra mussel mortalty followng 12h exposure n December A short term exposure/recovery experment was also conducted n December. Zebra mussels were exposed to copper sulfate at hgh copper level for 12h, and then observed durng a 36h recovery perod. There was no mussel mortalty recorded n ths experment or n the controls. p. 15

16 Precentage of mussels dead (%) Effects of copper based algaecdes on mussels Effect of copper sulfate on quagga mussel mortalty durng 96h exposure n November In November, an experment exposng quagga mussels to copper sulfate was conducted over 96h. Ths experment was carred out at Davs Dam, startng on November 14 th, 212. The ambent temperature durng the experment fluctuated between 18.2ºC and 16.7 ºC. The average temperature was 17.2 ºC. The results are shown n Fgure h end of exposure Tme (h).26 mg/l Bag 1.26 mg/l Bag 2.26 mg/l Bag 3.52 mg/l Bag 1.52 mg/l Bag 2.52 mg/l Bag 3 Fgure 8. Quagga mussel mortalty when exposed to copper sulfate The mortalty of the quagga mussels after 96 hours (Fg. 8) ranged from 58 to 74% at low copper concentraton. At the hgh copper concentraton the maxmum mortalty vared from 96.2% to 96.5%. After an addtonal 6 hours n the recovery cooler the maxmum mortalty for mussels exposed to a low copper concentraton rose to a mnmum of 84.5% and a maxmum of 95.6%. At the hgh copper concentraton level, the mnmum mortalty ncreased to 98.3% and maxmum mortalty reached 1%. We contnued to observe the mussels n the recovery tank for an addtonal 5 hours. The mortalty n the adults exposed to low copper contnued to ncrease slghtly. There were no mortaltes recorded n the control group Effect of copper sulfate on quagga mussel mortalty followng 12h exposure n November In November, a recovery experment was conducted where quagga mussels were exposed to copper sulfate at hgh copper levels for 12h, and then observed durng a 36h recovery perod. There was no mussel mortalty recorded n ths experment or n the controls. p. 16

17 Effects of copper based algaecdes on mussels Modelng the response of zebra mussels to copper sulphate at 96h exposure The followng logstc regresson model was found to adequately ft the data on mortalty of the zebra mussels after ther exposure to copper sulfate for 96 h (see also Fgures 9 and 1): p exp( 9.3.4h.19[t 18.4] 3.55C 2.4Month.57Perod 1.83[C Month ]) 1 exp( 9.3.4h.19[t 18.4] 3.55C 2.4Month.57Perod 1.83[C Month ]) All varables ncluded nto ths model consderably contrbuted to predcton of the zebra mussel mortalty (Table 3). As s seen from the equaton, the cumulatve mortalty, n addton to ts ncrease over tme, was postvely assocated wth the concentraton of copper sulfate, water temperature, and overall was also hgher n December compared to July. However, there was a sgnfcant nteracton between the copper sulfate concentraton and month,.e. mussel mortalty at the hgh copper concentraton was lower n December compared to July. In addton, we found that the ncrease n mussel mortalty was generally slower n the post-exposure perod (.e., after 96 h; see the Perod term n the equaton) n December compared to July. Table 3. Analyss of devance table for parameters of the GLM that descrbes the response of zebra mussels to treatment wth copper sulfate for 96 hours. Devance of the null model (.e. model that has no predctors ncluded) s 432.6, wth 45 degrees of freedom. Parameter Devance explaned by a parameter (added nto the model (sequentally) Degrees of freedom Resdual degrees of freedom Devance left unexplaned P-value for the chsquared test h (tme) <<.1 t (temperature) <<.1 C (concentraton) <<.1 Month a <<.1 Perod b C Month <<.1 a Month was coded n the model as a dummy varable that takes a value of for July and 1 for December b Perod was coded n the model as a dummy varable that takes a value of for the perod from to 96 hours (ncludng), and 1 for the remanng duraton of the experment. p. 17

18 Observed quantles Proporton dead Effects of copper based algaecdes on mussels.8 july dec.6 conc.4.2 low hgh Tme, h Fgure 9. Ft of the model whch descrbes the response of zebra mussels to treatment wth copper sulfate for 96 hours. Dots show the actual cumulatve proportons of molluscs found to be dead by a certan tme snce the begnnng of the experment. Sold lnes correspond to the ftted values. Dashed lnes denote 95% confdence ntervals for the ftted values. Pearson correlaton coeffcent for the observed vs. ftted values s.985 (P <<.1), suggestng that the model fts the data very well Expected normal quantles Fgure 1. Q-Q plot for resduals of the model that descrbes the response of zebra mussels to treatment wth copper sulfate for 96 hours. The dots on ths plot form approxmately a dagonal lne, suggestng normalty of the resduals. Ths s further confrmed by the non-sgnfcant Shapro-Wlk s test for normalty (W =.956, P =.83). p. 18

19 Temperature, C low hgh Effects of copper based algaecdes on mussels It should be noted that the results of ths model are to be nterpreted wth cauton as the partcular mechansm behnd the month effect s not clear and could potentally be confounded by temperature. As s seen from Fgure 11, temperature demonstrated mrrored temporal profles n July compared to December. Thus, t s not clear whether t was actually the seasonal effect or the temperature effect that caused the observed dfferences n mortalty rates ncrease between the months july dec Tme, h Fgure 11. Dynamcs of the water temperature over the duraton of experments on mortalty of zebra mussels exposed to copper sulfate n dfferent months Modelng the response of quagga mussels to copper sulphate at 96h exposure As the data from ths experment demonstrated an autocorrelaton (not shown here), we ftted a generalzed addtve model wth correlated resduals to these data. The followng GAM was found to ft the data adequately (see Fgures 12 and 13): p exp( 1 exp( f ( h ) f ( h ) t t 1 1 C ) C ) 1 1 p. 19

20 Proporton dead Effects of copper based algaecdes on mussels All varables ncluded nto ths model consderably contrbuted to predcton of the quagga mussels mortalty (Table 4). As s seen from the equaton, the cumulatve mortalty was nonlnearly related to tme (see Fgure 11), and was postvely assocated wth the concentraton of copper sulfate. F-stat represents the varance wthn groups dvded by the varance between groups. If there s no dfference between the groups of data, then the F-statstc value would be, the hgher the value the more dfference there s between the groups. Therefore tme plays a very mportant role n mortalty of quaggas when exposed to copper sulfate We observed no statstcally sgnfcant relaton of mortalty rate to the water temperature, a fndng that could be related to a low varaton of the recorded temperature values (mn 16.4 C, max 18.2 C). The estmated parameter of resdual autocorrelaton made up ρ =.412, and was statstcally sgnfcantly dfferent from (lower 95% confdence lmt =.36, upper 95% confdence lmt =.686). Table 4. Analyss of varance table for parameters of the GAM that descrbes the response of quagga mussels to treatment wth copper sulfate for 96 hours (n = 3). Parameter Degrees of freedom F-statstc P-value for the F-test C (concentraton) <<.1 f(h) (tme) << conc.5 low hgh Tme, h Fgure 12. Ft of the model whch descrbes the response of quagga mussels to treatment wth copper sulfate for 96 hours. Dots show the actual cumulatve proportons of mussels found to be dead by a certan tme snce the begnnng of the experment. Sold lnes correspond to the ftted values. Dashed lnes denote 95% confdence ntervals for the ftted values. Pearson correlaton coeffcent for the observed vs. ftted values s.996 (P <<.1), suggestng that the model fts the data very well. p. 2

21 f(hour) Observed quantles Effects of copper based algaecdes on mussels Expected normal quantles Fgure 13. Q-Q plot for resduals of the model that descrbes the response of quagga mussels to treatment wth copper sulfate for 96 hours. The dots arrange approxmately nto a dagonal lne, suggestng normalty of the resduals. Ths s further confrmed by the non-sgnfcant Shapro-Wlk s test for normalty (W =.996, P =.443) Hour Fgure 14. Estmated smoothng functon for tme n the model that descrbes the response of quagga mussels to treatment wth copper sulfate for 96 hours. The smoother s shown as a sold lne, along wth ts 95% pont-wse confdence bands. The Y-axs reflects the contrbuton of the smoother to the ftted mortalty values. p. 21

22 Precentage of mussels dead (%) Effects of copper based algaecdes on mussels 4.3 Effects of Natrx on dressend mussels Smlar to the copper sulfate experment, 96h exposure mortaltes for zebra mussels were montored n both July and December Effect of Natrx on zebra mussel mortalty durng 96h exposure n July The frst experment started on July 9, 212 observng zebra mussel mortalty durng exposure to Natrx TM. Temperatures n the drum wth the low copper averaged 17.9 C, wth a mnmum of 15.1 C and a maxmum of 22.8 C. The drum wth hgh copper had an average temperature of 18. C wth a mnmum of 15.2 C and maxmum of 22.8 C mg/l Bag 1.5 mg/l Bag 2.5 mg/l Bag 3 1. mg/l Bag 1 1. mg/l Bag 2 1. mg/l Bag Tme (h) 96h end of exposure Fgure 15. Zebra mussel mortalty when exposed to Natrx n July The July experment (Fgure 15) had an average mortalty of 16.2% after 96h at low copper, and an average mortalty of 63.% after 96h at hgh copper. The experment was contnued as mussels contnue to experence mortalty after beng placed n recovery chambers wth fresh water, and mortaltes rose after 148h to 62.3% for low and 97.2% for hgh copper. p. 22

23 Precentage of mussels dead (%) Effects of copper based algaecdes on mussels Effect of Natrx on zebra mussel mortalty durng 96h exposure n December The experment was repeated n December 212 (Fgure 16). After the 96 hour exposure to low copper as Natrx, the maxmum mortalty vared from 49.6 to 63.8%. At the hgh level of copper as Natrx, the maxmum mortalty vared from 56.4 to 8%. After addtonal 6 hours n the recovery cooler, the maxmum mortalty for low copper rose to a mnmum of 58.7% and a maxmum of 72.4%. At the hgh copper level, the mnmum mortalty ncreased to 73.2% and maxmum mortalty reached 93.9% (Fgure 16). Durng the expermental exposure, the ambent water temperature vared from a hgh of 22.3ºC to a low of 13.1 ºC. The average temperature for the low copper treatment was 19.3 ºC and 18.7 ºC for the hgh copper treatment. There was no mortalty recorded n the control mg/l Bag 1.5 mg/l Bag 2.5 mg/l Bag 3 1. mg/l Bag 1 1. mg/l Bag 2 1. mg/l Bag Tme (h) 96h end of exposure Fgure 16. Zebra mussel mortalty when exposed to Natrx p. 23

24 Precentage of mussels dead (%) Effects of copper based algaecdes on mussels Effect of Natrx on zebra mussel mortalty followng 12h exposure n December In December, a recovery experment was also completed. Zebra mussels were exposed for 12h to hgh copper soluton (1. mg/l) created wth Natrx. Followng exposure, the mussels were placed n a recovery chamber and observed for 36h for mortalty. There was some post exposure mortalty noted durng ths experment, from 1.89% to 5.88% after 36h of recovery (Fgure 17). The temperature averaged 17.4 C, wth a mnmum of 16.4 C and a maxmum of 19. C Bag 1 Bag 2 Bag Tme (h) Fgure 17. Zebra mussel mortalty followng removal from Natrx soluton p. 24

25 Precentage of mussels dead (%) Effects of copper based algaecdes on mussels Effect of Natrx on quagga mussel mortalty durng 96h exposure n November Ths experment was carred out at Davs Dam, startng on November 14, 212. The ambent temperature ranged from 18.2ºC and 16.7 ºC. The average temperature was 17.1 ºC for the low copper exposure and 17.2 ºC for the hgh copper exposure mg/l Bag 1.5 mg/l Bag 2.5 mg/l Bag 3 1. mg/l Bag 1 1. mg/l Bag 2 1. mg/l Bag Tme (h) 96h end of exposure Fgure 18. Quagga mussel mortalty when exposed to Natrx The mortalty of the quagga mussels after 96 hours ranged from 92.2 to 93.9% at low copper as Natrx (Fgure 18). At the hgh copper concentraton all three bags reached 1% mortalty at 96 hours. The survvors from the hgh copper concentraton were observed for an addtonal 48 hours durng whch tme the mortalty contnued to ncrease. At the end of ths 48 recovery perod the mortalty ranged from to 98.9 to 1%. There was no mortalty n the control groups. p. 25

26 Precentage of mussels dead (%) Effects of copper based algaecdes on mussels Effect of Natrx on quagga mussel mortalty followng 12h exposure n December A recovery experment was completed n December. There was no mortalty of the quagga mussels after the 12 hour exposure to hgh copper from a Natrx soluton. However, mortalty was observed after a 24 hour recovery perod. At the end of the experment, 36 hours after exposure, mortalty ranged from 2.% to 13.5% (Fgure 19). The ambent temperature ranged from 18.2ºC and 16.7 ºC. The average temperature was 17.1 ºC. There was no mortalty observed n the control group Bag 1 Bag 2 Bag Tme (h) Fgure 19. Quagga mussel mortalty followng removal from Natrx soluton Modelng the response of zebra mussels to Natrx at 96h exposure The followng logstc regresson model was found to adequately ft the data on mortalty of the zebra mussels after ther exposure to copper sulfate for 96 h (see also Fgures 2 and 21): p exp( h 1 exp( h.21[t 18.5].21[t 18.5] 2.3C 1.18Month 2.3C 1.18Month.82Perod 1.13[C Month ]).82Perod 1.13[C Month ]) All varables ncluded nto ths model consderably contrbuted to predcton of the zebra mussel mortalty (Table 5). As s seen from the equaton, the cumulatve mortalty, n addton to ts ncrease over tme, was postvely assocated wth the concentraton of Natrx, water temperature, and overall was also hgher n December compared to July. However, there was a sgnfcant nteracton between the Natrx concentraton and month,.e. n December the mortalty of molluscs exposed to hgher concentraton overall was lower than at ths same concentraton n July. In addton, we found that the ncrease of the mussels mortalty was p. 26

27 Effects of copper based algaecdes on mussels generally slower n the post-exposure perod (.e., after 96 h; see the Perod term n the equaton). Table 5. Analyss of devance table for parameters of the GLM that descrbes the response of zebra mussels to treatment wth Natrx for 96 hours. The devance of the null model (.e. model that has no predctors ncluded) s 822, wth 45 degrees of freedom. Parameter Devance explaned by a parameter (added nto the model sequentally) Degrees of freedom Resdual degrees of freedom Devance left unexplaned P-value for the chsquared test h (tme) <<.1 C (concentraton) <<.1 Month a <<.1 t (temperature) <<.1 Perod b <.1 C Month <.1 a Month was coded n the model as a dummy varable that takes a value of for July and 1 for December b Perod was smlarly coded as a dummy varable that takes a value of for the perod from to 96 hours (ncludng), and 1 for the remanng duraton of the experment. p. 27

28 Observed quantles Proporton dead Effects of copper based algaecdes on mussels 1. july dec.75 conc.5.25 low hgh Tme, h Fgure 2. Ft of the model whch descrbes the response of zebra mussels to treatment wth Natrx for 96 hours. Dots show the actual cumulatve proportons of molluscs found to be dead by a certan tme snce the begnnng of the experment. Sold lnes correspond to the ftted values. Dashed lnes denote 95% confdence ntervals for the ftted values. Pearson correlaton coeffcent for the observed vs. ftted values s.985 (P <<.1), suggestng that the model fts the data very well Expected normal quantles Fgure 21. Q-Q plot for resduals of the model that descrbes the response of zebra mussels to treatment wth Natrx for 96 hours. The dots on ths plot form approxmately a dagonal lne, suggestng normalty of the resduals. Ths s further confrmed by the non-sgnfcant Shapro-Wlk s test for normalty (W =.965, P =.182). p. 28

29 Temperature, C low hgh Effects of copper based algaecdes on mussels It should be noted that the results of ths model are to be nterpreted wth cauton as the partcular mechansm behnd the month effect s not clear and could potentally be confounded by temperature. As s seen from Fgure 22, temperature demonstrated mrrored temporal profles n July compared to December. Thus, t s not clear whether t was actually the seasonal effect or the temperature effect that caused the observed dfferences n rates of mortalty s ncrease between the months july dec Tme, h Fgure 22. Dynamcs of the water temperature over the duraton of experments on mortalty of zebra mussels exposed to Natrx n dfferent months Modelng the response of quagga mussels to Natrx at 96h exposure The followng logstc regresson model was found to adequately ft the data on mortalty of quagga mussels after 96 h exposure to copper sulfate (see also Fgures 23 and 24): p exp( exp( 7.7.1h.1h 1.19C ) 1.19C ) All varables ncluded nto ths model consderably contrbuted to predcton of the quagga mussel mortalty (Table 6). As s seen from the equaton, the cumulatve mortalty, n addton to ts ncrease over tme, was postvely assocated wth the concentraton of Natrx. p. 29

30 Proporton dead Effects of copper based algaecdes on mussels Table 6. Analyss of devance table for parameters of the GLM that descrbes the response of quagga mussels to treatment wth Natrx for 96 hours. The devance of the null model (.e. model that has no predctors ncluded) s 7911, wth 24 degrees of freedom. Parameter Devance explaned by a parameter (added nto the model (sequentally) Degrees of freedom Resdual degrees of freedom Devance left unexplaned P-value for the chsquared test h (tme) <<.1 C (concentraton) << conc.5 hgh low Tme, h Fgure 23. Goodness of ft of the model that descrbes the response of quagga mussels to treatment wth Natrx for 96 hours. Dots show the actual cumulatve proportons of molluscs found to be dead by a certan tme snce the begnnng of the experment. Sold lnes correspond to the ftted values. Dashed lnes denote 95% confdence ntervals for the ftted values. Pearson correlaton coeffcent for the observed and ftted values s.995 (P <<.1), suggestng that the model fts the data very well. p. 3

31 Observed quantles Effects of copper based algaecdes on mussels Expected normal quantles Fgure 24. Q-Q plot for resduals of the model that descrbes the response of quagga mussels to treatment wth Natrx for 96 hours. The dots on ths plot form approxmately a dagonal lne, suggestng normalty of the resduals. Ths s further confrmed by the non-sgnfcant Shapro-Wlk s test for normalty (W =.976, P =.794). p. 31

32 Precentage of mussels dead (%) Effects of copper based algaecdes on mussels 4.4 Effects of Captan lqud copper algaecde on dressend mussels Captan and zebra mussel mortalty durng 96h exposure n December Zebra mussels were exposed to a low and hgh concentraton of Captan (.5 mg/l and 1. mg/l) for 96 hours n December 212 (Fgure 25). Frst mortalty n adults was noted after 12 hours n the hgh copper soluton. At the end of the 96 hour exposure perod, the low treatment acheved maxmum mortalty between 2.3 and 32.7% whle the hgh treatment acheved mortalty rangng from 66.2 to 78.3%. We contnued to observe the survvng mussels for an addtonal 6 hours n a recovery tank. The mortaltes contnued to ncrease durng the recovery perod. At the end of the 6 hour recovery, the low treatment acheved maxmum mortalty between 31.8% and 55.8% whle the hgh treatment acheved mortalty rangng from 8.28 to 95.3%. The mnmum temperature durng the experment was 13.1 ºC, maxmum was 21.3 ºC, and the average temperature was 17.7 C n the low treatment and 18.2 ºC n the hgh treatment mg/l Bag 1.5 mg/l Bag 2.5 mg/l Bag 3 1. mg/l Bag 1 1. mg/l Bag 2 1. mg/l Bag Tme (h) 96h end of exposure Fgure 25. Zebra mussel mortalty when exposed to Captan as low copper and hgh copper for 96 hours p. 32

33 Precentage of mussels dead (%) Effects of copper based algaecdes on mussels Captan and zebra mussel mortalty followng 12h exposure n December The recovery experment observed zebra mussel survval for a 36h recover perod, after removal from a 12h exposure to hgh copper soluton of Captan. The 12-hour exposure resulted n no mortalty. At the end of the 36-hour recovery perod, the maxmum mortalty was 3% (Fgure 26). The mnmum temperature durng the experment was 15.5ºC, maxmum was 21.ºC and the average temperature was 19.ºC Bag 1 Bag 2 Bag Tme (h) Fgure 26. Zebra mussel mortalty after 12h exposure to Captan as hgh copper equvalent (1 mg/l) p. 33

34 Precentage of mussels dead (%) Effects of copper based algaecdes on mussels Effect of Captan on quagga mussel mortalty durng 96h exposure n December Quagga mussels were exposed to a low and hgh concentraton of Captan (.5 mg/l and 1. mg/l) for 96 hours n December 212 (Fgure 27). Frst mortalty n adults was noted after 12 hours n the hgh copper soluton. Mortalty dd not occur n the low copper soluton untl the 24 hour mark. There contnued to be hgher cumulatve mortalty observed n the group exposed to hgh copper soluton untl the 84 hour mark. After that pont, the cumulatve mortalty n the low copper treatment was essentally equvalent that of the hgh copper treatment. The low treatment acheved 92 to 95% mortalty n 96 hours, whle the hgh treatment acheved 91 to 96% mortalty. The mnmum temperature durng the experment was 16.8 ºC, maxmum was 18. ºC and the average temperature was 17.2 ºC mg/l Bag 1.5 mg/l Bag 2.5 mg/l Bag 3 1. mg/l Bag 1 1. mg/l Bag 2 1. mg/l Bag Tme (h) Fgure 27. Quagga mussel mortalty when exposed to Captan for 96 hours p. 34

35 Precentage of mussels dead (%) Effects of copper based algaecdes on mussels Effect of Captan on quagga mussel mortalty followng 12h exposure n December The recovery experment looked at effects of short term exposure (12h) to a hgh copper Captan soluton on quagga mussels durng a recovery perod of 36h (Fgure 28). Some mortalty was observed followng the ntal 12 hour exposure. The mortalty ncreased durng the 36 hour recovery perod. The maxmum mortalty acheved at the end of the 36 hour recovery perod was 1%. The temperature averaged 17.5 C, wth a mnmum of 17.1 C and a maxmum of 18. C Bag 1 Bag 2 Bag Tme (h) Fgure 28. Quagga mussel mortalty followng 12 hour exposure to Captan soluton, hgh copper equvalent (1. mg/l) Modelng the response of zebra mussels to Captan at 96h exposure The followng logstc regresson model was found to adequately ft the data on mortalty of the zebra mussels after ther exposure to Captan for 96h (see also Fgures 29 and 3): p 1 exp( exp( h.4h.23[t.23[t 17.9] 17.9] 2.6C ) 2.6C ) All varables ncluded nto ths model consderably contrbuted to predcton of zebra mussel mortalty (Table 7). As s seen from the equaton, the cumulatve mortalty, n addton to ts ncrease over tme, was postvely assocated wth the concentraton of Captan and water temperature. p. 35