9
Multicomponent molluscicide mixtures for zebra mussel control Raquel Costa a, b, , 1, 2 , David C. Aldridge c, 3 , Geoff D. Moggridge a, 1 a Department of Chemical Engineering and Biotechnology, University of Cambridge, New Museums Site, Cambridge CB2 3RA, UK b Department of Chemical Engineering, University of Coimbra, Pólo II, Rua Sílvio Lima, 3030-790 Coimbra, Portugal c Department of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, UK abstract article info Article history: Received 21 September 2011 Accepted 14 March 2012 Available online 17 April 2012 Communicated by Paul Helm Keywords: Biofouling Dreissena polymorpha Pallas Invasive bivalves Mixture toxicity Pest control Intuitively, it is reasonable to expect enhanced control of the biofouling zebra mussel through multi- component molluscicide cocktails. In this study, the potential of combined potassium chloride, polyDADMAC, niclosamide ethanolamine salt and 2-(thiocyanomethylthio)benzothiazole (TCMTB) for zebra mussel mitigation was investigated. A series of mixtures of varying compositions was tested. First, the combination was considered in its entirety, and the nature of the biocides' joint toxicity was elucidated by adopting a structured classication system previously dened. Then, a central composite experimental design was employed to detail the contribution of each ingredient to the blend performance and ultimately derive an empirical model of mixture effects to optimise the formulation composition. Whilst the action of some of the toxins was synergised, the blend does not appear promising for zebra mussel control. Overall, the chemicals acted less than additively and, under some circumstances, antagonistic effects were observed. Although these results do not immediately lead to a new approach to pest mitigation, the study highlights aspects that are of practical relevance for the design of combined chemical treatments. In particular, this work recalls the funnel hypothesis from the eld of ecotoxicology (blends tend to be additive as the number of ingredients increases), which may provide key guidance in the mixture design process. Furthermore, the study shows that multiple biocides do not necessarily ensure improved zebra mussel mitigation, and therefore the nature of their combined effects should always be carefully examined. The systematic procedure proposed here to critically design biocide blends is useful in this context. © 2012 International Association for Great Lakes Research. Published by Elsevier B.V. All rights reserved. Introduction The freshwater mussel Dreissena polymorpha Pallas, commonly known as the zebra mussel, is one of the world's most economically and ecologically important pests. The species, native to the basins of the Black and Caspian Seas, has remarkable dispersal potential and an ability to colonise a variety of environments and habitats (Claudi and Mackie, 1994; McMahon, 1996). It is well established in many waterbodies in Western Europe and North America, and it continues to spread (Minchin et al., 2002; Pimentel et al., 2005). In addition to the negative effects on the infested ecosystems, the zebra mussel has a major impact on freshwater-dependent industries. Drinking water treatment facilities and process cooling systems are especially vulnerable to the species' biofouling activity (Claudi and Mackie, 1994; Elliott et al., 2005). The application of lethal toxicants or substances that impair the ability of the animals to attach to hard surfaces is the foremost strategy in most industrial zebra mussel control programmes (Claudi and Mackie, 1994; Post et al., 2000; Mackie and Claudi, 2010). Compared to other control strategies, such as the use of lters, the physical cleansing of infested structures or the coating of vulnerable surfaces with antifouling materials, a chemical control approach tends to be more cost-effective and versatile. It may be implemented in existing facilities without major structural changes and it provides full protection of the system against a range of biofouling agents (Post et al., 2000). In spite of these advantages, a chemical control method raises concerns related to the selectivity and cost of some toxins. Due to these concerns, the need to discover new molluscicides and/or develop improved application strategies has been acknowledged (e.g. Aldridge et al., 2006; Costa et al., 2008). Combining biocides to take advantage of their cumulative and synergistic effects intuitively appears as a promising avenue to explore in this context. Although this strategy has proved successful for the mitigation of several terrestrial and aquatic pests (Ahmad, 2004; Singh et al., 2005), it has not been comprehensively investigated for the case of invasive bivalves (Costa et al., 2011). For the zebra mussels, only a limited number of mainly binary molluscicide mixtures have been examined, and generally the matter has been approached from an empirical rather than Journal of Great Lakes Research 38 (2012) 317325 Corresponding author. Permanent address at: Department of Chemical Engineer- ing, University of Coimbra, Pólo II, Rua Sílvio Lima, 3030-790 Coimbra, Portugal. Tel.: +351 239798700; fax: +351 239798703. E-mail addresses: [email protected] (R. Costa), [email protected] (D.C. Aldridge), [email protected] (G.D. Moggridge). 1 Tel.: +44 1223 334777; fax: +44 1223 334796. 2 Tel.: +351 239798700; fax: +351239798703. 3 Tel.: +44 1223 33660; fax: +44 1223 336676. 0380-1330/$ see front matter © 2012 International Association for Great Lakes Research. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jglr.2012.03.010 Contents lists available at SciVerse ScienceDirect Journal of Great Lakes Research journal homepage: www.elsevier.com/locate/jglr

Multicomponent molluscicide mixtures for zebra mussel control

Embed Size (px)

Citation preview

Journal of Great Lakes Research 38 (2012) 317–325

Contents lists available at SciVerse ScienceDirect

Journal of Great Lakes Research

j ourna l homepage: www.e lsev ie r .com/ locate / jg l r

Multicomponent molluscicide mixtures for zebra mussel control

Raquel Costa a,b,⁎,1,2, David C. Aldridge c,3, Geoff D. Moggridge a,1

a Department of Chemical Engineering and Biotechnology, University of Cambridge, New Museums Site, Cambridge CB2 3RA, UKb Department of Chemical Engineering, University of Coimbra, Pólo II, Rua Sílvio Lima, 3030-790 Coimbra, Portugalc Department of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, UK

⁎ Corresponding author. Permanent address at: Depaing, University of Coimbra, Pólo II, Rua Sílvio Lima,Tel.: +351 239798700; fax: +351 239798703.

E-mail addresses: [email protected] (R. Costa), [email protected] (G.D. Moggridge).

1 Tel.: +44 1223 334777; fax: +44 1223 334796.2 Tel.: +351 239798700; fax: +351239798703.3 Tel.: +44 1223 33660; fax: +44 1223 336676.

0380-1330/$ – see front matter © 2012 International Adoi:10.1016/j.jglr.2012.03.010

a b s t r a c t

a r t i c l e i n f o

Article history:Received 21 September 2011Accepted 14 March 2012Available online 17 April 2012

Communicated by Paul Helm

Keywords:BiofoulingDreissena polymorpha PallasInvasive bivalvesMixture toxicityPest control

Intuitively, it is reasonable to expect enhanced control of the biofouling zebra mussel through multi-component molluscicide cocktails. In this study, the potential of combined potassium chloride, polyDADMAC,niclosamide ethanolamine salt and 2-(thiocyanomethylthio)benzothiazole (TCMTB) for zebra musselmitigation was investigated. A series of mixtures of varying compositions was tested. First, the combinationwas considered in its entirety, and the nature of the biocides' joint toxicity was elucidated by adopting astructured classification system previously defined. Then, a central composite experimental design wasemployed to detail the contribution of each ingredient to the blend performance and ultimately derive anempirical model of mixture effects to optimise the formulation composition. Whilst the action of some of thetoxins was synergised, the blend does not appear promising for zebra mussel control. Overall, the chemicalsacted less than additively and, under some circumstances, antagonistic effects were observed. Although theseresults do not immediately lead to a new approach to pest mitigation, the study highlights aspects that are ofpractical relevance for the design of combined chemical treatments. In particular, this work recalls the funnelhypothesis from the field of ecotoxicology (blends tend to be additive as the number of ingredientsincreases), which may provide key guidance in the mixture design process. Furthermore, the study showsthat multiple biocides do not necessarily ensure improved zebra mussel mitigation, and therefore the natureof their combined effects should always be carefully examined. The systematic procedure proposed here tocritically design biocide blends is useful in this context.

© 2012 International Association for Great Lakes Research. Published by Elsevier B.V. All rights reserved.

Introduction

The freshwater mussel Dreissena polymorpha Pallas, commonlyknown as the zebra mussel, is one of the world's most economicallyand ecologically important pests. The species, native to the basins ofthe Black and Caspian Seas, has remarkable dispersal potential and anability to colonise a variety of environments and habitats (Claudi andMackie, 1994; McMahon, 1996). It is well established in manywaterbodies in Western Europe and North America, and it continuesto spread (Minchin et al., 2002; Pimentel et al., 2005). In addition tothe negative effects on the infested ecosystems, the zebra mussel hasa major impact on freshwater-dependent industries. Drinking watertreatment facilities and process cooling systems are especiallyvulnerable to the species' biofouling activity (Claudi and Mackie,1994; Elliott et al., 2005).

rtment of Chemical Engineer-3030-790 Coimbra, Portugal.

[email protected] (D.C. Aldridge),

ssociation for Great Lakes Research.

The application of lethal toxicants or substances that impair theability of the animals to attach to hard surfaces is the foremoststrategy in most industrial zebra mussel control programmes (Claudiand Mackie, 1994; Post et al., 2000; Mackie and Claudi, 2010).Compared to other control strategies, such as the use of filters, thephysical cleansing of infested structures or the coating of vulnerablesurfaces with antifouling materials, a chemical control approachtends to be more cost-effective and versatile. It may be implementedin existing facilities without major structural changes and it providesfull protection of the system against a range of biofouling agents (Postet al., 2000). In spite of these advantages, a chemical control methodraises concerns related to the selectivity and cost of some toxins. Dueto these concerns, the need to discover new molluscicides and/ordevelop improved application strategies has been acknowledged(e.g. Aldridge et al., 2006; Costa et al., 2008).

Combining biocides to take advantage of their cumulative andsynergistic effects intuitively appears as a promising avenue to explorein this context. Although this strategy has proved successful for themitigation of several terrestrial and aquatic pests (Ahmad, 2004; Singhet al., 2005), it has not been comprehensively investigated for the case ofinvasive bivalves (Costa et al., 2011). For the zebramussels, only a limitednumber ofmainly binarymolluscicidemixtures have been examined, andgenerally the matter has been approached from an empirical rather than

Published by Elsevier B.V. All rights reserved.

318 R. Costa et al. / Journal of Great Lakes Research 38 (2012) 317–325

systematic perspective (van Benschoten et al., 1992; Elzinga andButzlaff, 1994; Wildridge et al., 1998a; Mackie and Claudi, 2010). In arecent work, Costa et al. (2011) evaluated the potential of abinary mixture of poly(diallyldimethyl ammonium chloride)(polyDADMAC) and potassium chloride for zebra mussel mitigationbased on the systematic characterisation of the nature of the biocides'joint effects. A structured approach to perform such characterisation,derived from the ecotoxicology background on mixture toxicity, wasproposed in that work.

It is reasonable to expect increased lethal effects, and henceenhanced control, when multicomponent rather than binary mixturesare applied. In the present study, the potential of mixtures of potassiumchloride, polyDADMAC, niclosamide ethanolamine salt and 2-(thiocya-nomethylthio)benzothiazole (TCMTB) for zebra mussel mitigation isdiscussed. The systematic investigation approach previously proposedfor binarymixtures (Costa et al., 2011) is extended to the analysis of theperformance of multicomponent molluscicide blends.

The quaternary combination above was selected for examinationon the following grounds. From the ecotoxicological experimentalevidence on mixture toxicity it appears that there is a propensity forblends to be additive as the number of components increase (Warneand Hawker, 1995; Warne, 2002) whilst more than additivecombinations are more beneficial from the control perspective(Costa et al., 2011). Furthermore, the possible increased effectivenessof multiple chemicals has to be balanced with potential operationaland effluent purification issues related to the dosage of a large numberof substances into the system. For these reasons, mixtures of amoderate number of molluscicides are most likely to be preferred inpractice, and hence a fairly simplemixture of four biocides was chosenfor investigation. The blend incorporates potassium ions (dosed aspotassium chloride) and polyDADMAC, whose joint toxic effects onthe zebra mussel have previously been shown to be more thanadditive (Costa et al., 2011). To these chemicals, niclosamide ions(applied as niclosamide ethanolamine salt) and TCMTB have beenadded. These chemicals are well-accepted zebramussel control agents(McMahon et al., 1993; Waller et al., 1993; Fisher et al., 1994;Wildridge et al., 1998b; Sprecher and Getsinger, 2000) and act throughdifferent mechanisms, whichmaximises the chances of exerting morethan additive effects. Potassium ions, polyDADMAC and TCMTB arebelieved to cause zebra mussels' death by disrupting gas exchange inthe gills by either provoking the depolarisation of (Fisher et al., 1991;Durand-Hoffman, 1995), adsorbing to (Gloxhuber, 1974; Post et al.,1996), or being generally corrosive to (McMahon et al., 1993; Nikl andFarrell, 1993; Walsh and O'Halloran, 1997; Sprecher and Getsinger,2000) the cell membranes, respectively. Niclosamide ions exert theirtoxic action by interfering with the cellular respiratory processes(Ishak et al., 1970; Andrews et al., 1982; Mallatt et al., 1994).

Systematic evaluation of biocide combinations for zebra musselcontrol

The study of the effects of chemical mixtures on aquatic biota iswell-established in ecotoxicology (e.g. Backhaus et al., 2003; Svendsenet al., 2010). Although the investigation of mixture toxicity in theecotoxicological and pest control contexts involves different issues, thefundamental questions being addressed in both fields are similar, andtools from the formermay be of great use in the design of toxin cocktailsfor the mitigation of aquatic nuisances. However, the links betweenthese two areas of research have not been thoroughly explored.

In short, the previous approach to systematically evaluate thepotential of binary biocide combinations for zebra mussel control(Costa et al., 2011) involves a two-level classification scheme, and ituses the concentration addition concept (Sprague, 1970; Könemannand Pieters, 1996; Faust et al., 2001; de Zwart and Posthuma, 2005) asa reference response. In the present paper, this structured evaluationmethod is extended to the design of complex molluscicide mixtures.

When referring to multiple chemicals, the concentration additionconcept implies that all toxicants are non-interactive and share acommon mode of action, being biologically equivalent at the targetsite. This mechanistic model is mathematically expressed by(Könemann and Pieters, 1996; Backhaus et al., 2000; Faust et al.,2001; de Zwart and Posthuma, 2005):

Xn

i¼1

cx;iLCx;i

¼Xn

i¼1

TUx;i ¼ TUx;mix ¼ 1 ð1Þ

where n denotes the number of chemicals in the mixture; i (i=1,…, n)refers to a generic substance; cx,i is the concentration of the compound iin the mixture that elicits a lethal effect of magnitude x; LCx,i is theequivalent effect concentration of component i; the quotientTUx;i ¼ cx;i

LCx;i

is the number of toxic units of the ingredient i in the mixture; andTUx,mix is the total number of toxic units in the blend.

Similar to binary blends, the first level of the joint toxicityclassification system for multicomponent mixtures is used to refer toa specific biocide i of particular interest and judge whether its toxicaction is synergised (TUx,i below one), antagonised (TUx,i above one)or not affected (TUx,i equals one) in the presence of the remainingchemicals in the mixture. However, the concepts of antagonism,synergism and inexistence of joint effect cannot be strictly applied toa multicomponent mixture as a whole. In such blends, differentcomponents may exhibit different variations of their individualtoxicities, and the overall toxicity of the mixture reflects the balancebetween all these variations. At the second level of the joint toxicityclassification system, mixtures where synergism occurs are treated asa whole and their performance is judged based on the overall toxicload necessary to elicit the lethal response under consideration. Thecategories used in the analysis of binary mixtures – additive (TUx,mix

equals one), more than additive (TUx,mix below one) or less thanadditive (TUx,mix above one) – can be directly extended to thecharacterisation of complex mixtures. However, the convenientgraphical representation of TUx,mix in a isobologram is not feasiblefor blends with more than three components. The ultimate assess-ment of the potential of possible multicomponent molluscicideblends for pest control is performed by employing similar criteria asin the case of binary mixtures (Costa et al., 2011).

The selection of concentration addition as a reference response inthe proposed classification system does not imply that all mollusci-cide blends are expected to comply with this model, whichmechanistically refers to non-interactive chemicals with similarmodes of action (Plackett and Hewlett, 1952; Könemann andPieters, 1996). Instead, this reference response was chosen forconvenience (Costa et al., 2011).

Whilst the analysis of the toxicity of binary mixtures is facilitatedby their simplicity, in the case of multicomponent combinations thecomplexity of the experimental procedures required to examine indetail the chemicals' joint effect increases with the number ofingredients. Such complexity is one of the reasons why the analysisof combined toxicity has been typically reduced to the testing ofequitoxic formulations, following fixed ratio designs (e.g. Faust et al.,2001; Warne, 2002). Equitoxic blends contain all ingredients at thesame fraction of the respective individual toxicity (e.g. median lethalconcentration or no observed effect concentration). In fixed ratiostudy designs, a series of dilutions of a given formulation with varyingconcentrations but constant relative proportions of the constituents istested to experimentally describe the dose–response relationship ofthe formulation.

The main drawback of integrative study approaches involvingequitoxic formulations and fixed ratio designs, where the mixture istreated as a whole, is the fact that they provide limited informationon the individual contribution of each chemical to the mixturetoxicity, which is important for the design of pest control strategies. Thesystematic variation of themixture composition and the testing of non-

319R. Costa et al. / Journal of Great Lakes Research 38 (2012) 317–325

equitoxic blends provide a fuller account of the toxicants' combinedaction. For complex mixtures, this is more effectively achieved throughstatistical experimental designs, integrated with multivariate dataanalysis techniques. Full factorial, fractional factorial and centralcomposite designs as well as response surface, principal componentsand partial least squares regression analysis techniques may beemployed (see, for example, Eide, 1996; Groten et al., 1996; Schoen,1996; Smith, 1998; and more recently, Groten et al., 2001; Lock andJanssen, 2002; Tajima et al., 2002). The choice of the approach toexamine the joint toxicity of multiple chemicals largely depends on thenumber of components in the mixture and the extent to which thenature of the combined action needs to be characterised.

In the present study, the ecotoxicology-inspired view on mixturetoxicity discussed in this section was applied to evaluate the potentialof mixtures of potassium chloride, polyDADMAC, niclosamide ethanol-amine salt and TCMTB for zebra mussel control.

Materials and methods

Experimental design: three-phase study

In the first phase of the study, dose–response data for theindividual toxins were obtained. These were used as reference datafor mixture toxicity analysis.

In the second phase, the combination was considered in its entirety,and the joint toxicity of the four biocides was determined by following afixed ratio design. Because the nature of the joint effects of the chemicalscould depend on the blend's composition (Warne, 2002), two differentformulations were examined. In the first (called LC50 formulation), thechemicals were mixed in the relative proportion of their median lethalconcentrations. In the second formulation (designated as LC10 formula-tion), the toxins were present at the ratio of their individual lethalconcentrations for percentile 10. The dose–response data obtained in thisset of experiments was analysed based on the combined actionclassification system presented in the previous section.

After getting a general appreciation of the potential of thequaternary combination for zebra mussel control, the aim of the thirdphase of the study was to examine the contribution of each ingredientto the blend performance, deriving an empirical model of mixtureeffects. This type of data is essential from the practical applicationstandpoint because it allows the formulation to be optimised. A series ofcombinations were tested in this phase. The test treatments followed acentral composite experimental design and response surface analysiswas employed (Box and Draper, 1987;Montgomery and Runger, 1999).

Test organisms

In mid-summer (July), adult zebra mussels were collected fromthe walls of a filter bed in a water treatment plant in London (UK) byusing a paint stripper blade to carefully cut their byssus threads.Immediately after collection, the animals were transported to thelaboratory in field water, and individuals with shell length in therange 20 to 30 mm were selected and thoroughly rinsed. The meanshell length of the specimens used in the study was 25.9±SE 0.8 mm.The test organisms were held in aerated dechlorinated municipalwater in a temperature-controlled chamber at 18 °C (±1 °C), on a12 h-dark/12 h-light cycle. The toxicity tests, which were conductedin the temperature-controlled chamber under the holding conditions,were initiated within 1 week of collection. Dechlorinated municipalwater was used in all experiments.

Single biocide toxicity tests

Sets of ten mussels were placed into 75 1-litre containers holding500 ml of continuously aerated water. The animals acclimated for48 h. Those that did not attach to the bottom of the containers within

this period (no more than 10% in all vessels) were discarded. As aresult, at the moment of dosing, three out of the 75 test potscontained nine rather 10 mussels, which was taken into account inthe mortality data analysis. The toxins were dosed so that thespecimens were exposed to the following individual treatments: (i)25, 50, 100, 200, 400 and 800 mg/l of potassium chloride, applied aslaboratory grade reagent with purity above 99% in weight (FisherScientific UK Ltd, Loughborough, UK); (ii) 10, 20, 40, 120, 360 and1080 mg/l of polyDADMAC, dosed as a proprietary product containingmore than 85% in weight of polymer (SNF UK Ltd, Normanton, UK);(iii) 0.048, 0.096, 0.192, 0.250, 0.324 and 0.422 mg/l of niclosamideethanolamine salt, applied as Bayluscide® WP 70, which contains83.1% in weight of the toxin (Bayer AG, Leverkusen, Germany); and(iv) 10, 20, 40, 120, 360 and 1080 mg/l of TCMTB, dosed as BULAB®6009, whose weight fraction of active ingredient is 30% (BuckmanLaboratories International, Inc., Memphis, US). The four chemicalswere applied individually by following the dosing scheme intendedfor the mixture bioassays. Potassium chloride was applied for 48 h. Inthe case of the remaining toxins, the treatments lasted for 36 h,initiating 12 h after potassium chloride application. In the controltreatment, organisms were not exposed to toxins. Each of the 25treatments was applied in triplicate, the three replications beingrandomly distributed in the temperature-controlled chamber.Mortalitywas monitored 24 h and 48 h after potassium chloride dosing, whichcorresponded to treatment durations of 12 h and 36 h in the case of theother biocides. Failure to respond to an external tactile stimulusprovoked by a blunt probe was used as the death criterion. Deadspecimens were discarded at each mortality assessment. The results ofthe bioassays were analysed in terms of the dose–response dataobtained at the end of the treatments.

Integral mixture toxicity tests

The individual median lethal concentrations and lethal concentra-tions for percentile 10 after the desired exposure periods, obtained fromthe single biocide toxicity tests, were used to define LC50 formulationand LC10 formulation. The proportions inwhich the four chemicalsweremixed in the two formulations are presented in Table 1.

The two quaternary formulations were tested by following aprocedure similar to that implemented in the single biocide toxicitytests. Each formulationwas tested at 11 different overall concentrations(cmix): 0.75, 1.5, 3, 6, 12, 24, 48, 96, 192, 384 and 768 mg/l. Thespecimens were pre-treated with potassium chloride for 12 h and thenexposed to the four chemicals for 36 h, which resulted in 48-houroverall treatments. The pre-exposure of the test organisms to potassiumchloride potentially allowed for the salt to promote the uptake of theother toxins (Costa et al., 2011). In the control treatment, organismswere not exposed to toxins. Each of the 23 treatments was applied intriplicate. The mortality in the test containers was assessed 24 h and48 h after the commencement of the treatments. The results of thebioassays were analysed in terms of the mortality data obtained at theend of the treatments.

Central composite design toxicity tests

Combinations of the biocides were tested following a centralcomposite design, consisting of 27 experimental points: 16 cubepoints, 8 star points and 3 centre points (Box and Draper, 1987;Montgomery and Runger, 1999). Each toxin was tested at five linearlyequidistant concentrations, coded as −2, −1, 0, 1 and 2 (Table 2).These concentrations were chosen so that the centre pointscorresponded to one-fourth of the median lethal concentrations of theindividual biocides after the desired exposure period, and the lowestdosage levels corresponded to the absence of toxins. By following thisapproach, the four toxins were tested at similar proportions of theirindividual toxicities. At the central dosage levels (0), the chemicals

Table 1Composition of the two formulations tested in the integral mixture toxicity study. The compositions are expressed in terms of the weight fraction of the generic biocide i, pi, given bythe quotient between the concentration of the component in the formulation (ci) and the overall mixture concentration (cmix).

Biocide Weight fraction of the biocide in the formulation

LC50 formulation pi ¼ cicmix

¼ LC50;i

P4i¼1

LC50;i

LC10 formulation pi ¼ cicmix

¼ LC10;i

P4i¼1

LC10;i

Potassium chloride 0.447 0.747PolyDADMAC 0.111 0.0434Niclosamide ethanolamine salt 3.21×10−4 8.24×10−4

TCMTB 0.441 0.208

Table 2Codification of the biocide concentrations tested in the central composite experimental design.

Biocide Concentration associated with each code (mg/l)

−2 −1 0 1 2

Potassium chloride 0 30.9 61.8 92.6 124PolyDADMAC 0 7.67 15.4 23.0 30.7Niclosamide ethanolamine salt 0 2.21×10−2 4.42×10−2 6.64×10−2 8.85×10−2

TCMTB 0 30.4 60.9 91.3 122

320 R. Costa et al. / Journal of Great Lakes Research 38 (2012) 317–325

were applied at the same fraction of their single toxicity. At the dosageslevels−1, 1 and 2 thiswas not exactly the case, because their individualdose–response relationships are not parallel (Fig. 1). However, thedifference between the respective individual toxicity proportions atlevels −1, 1 and 2 was not significant.

Each experimental point was tested in triplicate. In general, thetoxicity tests were conducted as described above. The test organismswere pre-treated with potassium chloride for 12 h before they wereexposed to the combinations containing this salt. In three controlvessels, mussels were not exposed to toxins. The overall treatmentslasted for 48 h, corresponding to an equivalent exposure to potassiumchloride and to treatment durations of 36 h in the case of the otherbiocides. The mortality in the test containers was assessed 24 h and48 h after the commencement of the treatments. The results of thecentral composite design toxicity tests were analysed in terms of theresponses produced at the end of the treatments.

Data analysis

Systematic description of the nature of the biocides' joint actionThe results of the integral mixture toxicity tests were analysed

by employing the proposed structured classification system inorder to elucidate the nature of the combined action of the fourbiocides. The analysis was performed for LC50 formulation and LC10

formulation separately. Lethal responses in the range of 10 to 90%were considered. The magnitude of the response (x) was firstspecified. Then, the equivalent lethal concentrations of theindividual toxins (LCx,i) were determined from the results of thesingle biocide toxicity tests by Probit analysis. The composition ofthe experimental combined treatments was converted into a toxicunit scale by dividing the concentration of each component (ci),related to the overall mixture concentration (cmix) (as detailed inTable 1 and originally expressed in mg/l), by the respectiveequivalent effect concentration (see Eq. 1 for details). The numberof toxic units of the mixture able to elicit the lethal response x (TUx,

mix) was then determined by Probit analysis. The nature of thechemical's joint action was further examined by analysing thenumber of toxic units of the individual components in the mixturesable to cause the lethal effect x (TUx,i), which was determined bythe quotient pi⋅LCx;mix

LCx;i, where the overall mixture concentration

producing the effect x (LCx,mix) was estimated from the results ofthe bioassays by Probit analysis.

Central composite design and response surface data analysisA second-order response surface model (Box and Draper, 1987;

Montgomery and Runger, 1999) was fitted to the data obtained in thecentral composite design toxicity tests. In the regression analysis,arcsine transformed mortality percent values (Zar, 1999) were used.

Statistical analysesThe statistical methods for general data analysis were used as

outlined by Zar (1999) and implemented in STATISTICA (StatSoft, Inc.(2003), STATISTICA (data analysis software system), version 6. www.statsoft.com). Dose–response data were modelled by Probit analysisusing the software StatsDirect (StatsDirect, Ltd. (1990), StatsDirectstatistical software. www.statsdirect.com). A significance level of 5%was used.

Results

Single biocide toxicity tests

No mortality occurred under the control conditions. Thedose–response relationships obtained for potassium chloride,polyDADMAC, niclosamide ethanolamine salt and TCMTB areshown in Fig. 1. Table 3 presents the chemicals' median lethalconcentrations and lethal concentrations for percentile 10 afterthe end of the single biocide exposures, which were involvedin the formulation of the treatments assessed in the integralmixture and central composite design toxicity tests.

Integral mixture toxicity tests

Full survival was observed in the control containers. The mortalityproduced by the combined treatments was significantly related to theoverall mixture concentration (two-factor ANOVA following arcsinetransformation of the mortality percent data; F=73.650; df=10;pb0.001; Fig. 2). No significant differences were observed between theeffects elicited by LC50 formulation and LC10 formulation (two-factorANOVA following arcsine transformation of the mortality percent data;F=0.189; df=1; p=0.666).

The nature of the biocides' joint action was explored by employingthe systematic approach described above. Both LC50 formulation andLC10 formulation were observed to exert less than additive effectsover the entire mortality range (Fig. 3). TUx,mix values depended onthe response under consideration (two-factor ANOVA; F=9.658;

Fig. 1. Dose–response data for the single molluscicides: (a) exposure to potassiumchloride for 48 h; (b) exposure to polyDADMAC for 36 h; (c) exposure to niclosamideethanolamine salt for 36 h; (d) exposure to TCMTB for 36 h. The points refer to theexperimental mortality data (mean±SE); the lines represent models obtained byProbit analysis (p≤0.02).

321R. Costa et al. / Journal of Great Lakes Research 38 (2012) 317–325

df=8; p=0.002), but were always significantly higher than one(pb0.05). Such values were not significantly affected by theformulation ratio (two-factor ANOVA; F=0.188; df=1; p=0.676).In Fig. 4, TUx,mix values producing lethal responses in the range 10 to90% are presented as the sum of the number of toxic units of themixture ingredients (TUx,i) as established by Eq. (1), allowing for thepotential of the formulations to be further assessed by referring to the

variation of the toxicity of the individual chemicals as they werecombined. Fig. 4a shows that, in the case of LC50 formulation, TUx,i

were generally lower than one at any response level. The onlyexceptions occurred in the lower mortality percent range forpolyDADMAC and TCMTB and in the upper mortality percent rangefor niclosamide ethanolamine salt, in which cases TUx,i values wereequal to or slightly above one. Generally speaking, the action of thefour biocides can thus be said to have been synergised as they weremixed in the ratio of their median lethal concentrations. In the case ofLC10 formulation (Fig. 4b), the actions of potassium chloride,polyDADMAC and TCMTB were synergised over the entire responselevel range as they were incorporated in the combination. However,from the comparison between Figs. 4a and b, major differences in theperformance of niclosamide ethanolamine salt are evident. In LC10formulation, the action of this molluscicide was synergised inmixtures producing low to intermediate mortality percents, but itwas observed to undergo significant antagonistic effects at responselevels above 60%, with an individual number of toxic units of almost 2for a percent response of 90%. It is worth nothing that the differencesbetween the two formulations in terms of the variation of theindividual chemicals' toxicity (that is, in the relative contributions ofTUx,i values to the total TUx,mix) reflect the difference between thechemicals' relative proportions in the formulations (Table 1) becausethe blends' dose–response curves (LCx,i) were similar (Fig. 2) and, bydefinition, TUx,i is given by the quotient pi⋅LCx;mix

LCx;i.

Central composite design toxicity tests

The bioassays involving the testing of a series of mixtures ofvarying composition following a central composite experimentaldesign did not provide meaningful results. The coefficient ofdetermination characterising the second-order response surfacemodel fitted to the results was 0.589 (p=0.365). None of themodel coefficients were found to be statistically significant (p valuesbetween 0.088 and 0.889), and therefore it was not possible todefinitely infer the relative contribution of each biocide and that oftheir interactions to the mixture toxicity.

Discussion

Potential of the quaternary mixture for zebra mussel control

The results obtained in the single substance bioassays (Fig. 1and Table 3) are consistent with the toxicity data reported in otherstudies (see, for example, Fisher et al., 1991; McMahon et al., 1993;Waller et al., 1993; Durand-Hoffman, 1995; and more recentlyCosta et al., 2011). Mixtures of the four molluscicides do not seemto have been tested before, and therefore a comparison of theresults obtained in the integral mixture toxicity tests (Fig. 2) withliterature data is not possible.

As far as the overall mixture performance is concerned, both LC50formulation and LC10 formulation were found to be less than additiveover the entiremortality range,with TUx,mix values consistentwithwhathas generally been reported in ecotoxicological studies (Fig. 3). Severalauthors (ECETOC, 2001; Warne, 2002) have pointed out that there arenot many examples in the literature of blends whose toxicity differsfrom concentration addition by a factor greater than three.

The toxic nature of the quaternary combination as a whole did notdepend on the ratio at which the components weremixed (Fig. 3). Notethat this was the case even though the compositions of LC50 formulationand LC10 formulation were substantially different (Table 1). As thechemicals were present at the same proportion of their individualtoxicities in the two blends, it is likely that this behaviour isrepresentative of that ofmixtures inwhichno single ingredient accountsfor an overwhelming proportion of mixture toxicity. In general, the jointaction of chemicals may depend on the mixture's quantitative

Table 3Selected lethal concentrations of potassium chloride, polyDADMAC, niclosamide ethanolamine salt and TCMTB for the duration of the individual treatments (48 h in the case ofpotassium chloride and 36 h in the case of the remaining biocides). The estimated concentrations and respective 95% confidence interval limits were obtained by Probit analysis.

Biocide Median lethal concentration after the end of thetreatment (mg/l)

Lethal concentration for percentile 10 after the endof the treatment (mg/l)

Estimate 95% confidence interval Estimate 95% confidence interval

Potassium chloride 247 133–464 87.0 30.6–217PolyDADMAC 61.4 40.0–92.6 5.06 1.68–9.86Niclosamide ethanolamine salt 0.177 0.153–0.199 0.0962 0.0711–0.116TCMTB 243 89.5–727 24.3 4.72–95.9

322 R. Costa et al. / Journal of Great Lakes Research 38 (2012) 317–325

composition (Warne, 2002). However, several cases of combinationswhose overall toxic characterwas not affected by the relative proportionof their components have been reported in the literature (e.g. Faust et al.,2001).

Fig. 3. Joint toxicity of potassium chloride, polyDADMAC, niclosamide ethanolamine saltand TCMTB, combined at different ratios in LC50 formulation and LC10 formulation, atvarying mortality levels. The solid lines represent the estimated TUx,mix and the dashedlines represent the respective 95% confidence interval limits obtained by Probit analysis.

Fig. 2. Dose–response data for LC50 formulation and LC10 formulation. The points referto the experimental mortality data (mean±SE); the lines represent dose–responsemodels obtained by Probit analysis (p≤0.001).

In addition to characterising the biocides' overall joint action, theresults of the integral mixture bioassays were also analysed in termsof the variation of the chemicals' individual lethality as they werecombined. Such variation was observed to depend on the ratio atwhich the chemicals were blended as well as the magnitude of theresponse under consideration (Fig. 4). Remarkably, even though thesensitivity of the mussels to niclosamide ethanolamine salt seemed todiminish as it was incorporated in LC10 formulation, the two testedblends as a whole produced similar dose–response curves (Fig. 2) andthey were characterised by similar TUx,mix values (Fig. 3). The detailedreasons for this cannot be drawn from this study. However, a possible

Fig. 4. Variation of the individual toxicity of potassium chloride, polyDADMAC,niclosamide ethanolamine salt and TCMTB as they were combined in (a) LC50

formulation or (b) LC10 formulation. This variation is expressed in terms of TUx,i inthe mixtures producing responses in the range 10 to 90%. The width of the shadedregions represents the magnitude of the TUx,i values; the total graph area correspondsto the sum of the TUx,i values, which is, by definition, TUx,mix shown in Fig. 3.

Fig. 5. Comparison of the average toxicity for binary (Costa et al., 2011) and quaternary(this study) mixtures containing potassium chloride and polyDADMAC.

Fig. 6. Assessment of the ability of concentration addition and response additionconcepts to predict the toxicity of LC50 formulation. The points refer to theexperimental dose–response data presented in Fig. 2 and the curves represent theestimated dose–response relationships. For further details on the mechanistic modelsof joint toxicity see, for example, Könemann and Pieters (1996), Faust et al. (2001),Backhaus et al. (2003) and de Zwart and Posthuma (2005).

323R. Costa et al. / Journal of Great Lakes Research 38 (2012) 317–325

interpretation for this result is as follows. It is accepted that the modeof action of toxicants upon an organism may change with varyingexposure concentrations (Könemann and Pieters, 1996; Groten et al.,2001; Barata et al., 2006). As the ratio, and consequently the testrange, of the toxin concentrations varied from LC50 formulation toLC10 formulation, the toxicity mechanisms of the biocides and/or theinteractions between them may also have varied. It is conceivablethat these variations occurred in such a way that the action ofniclosamide ethanolamine salt was hampered in the LC10 formula-tion, but the actions of the other biocides (in particular polyDADMACand TCMTB) were more favoured in this formulation than in LC50formulation (Fig. 4). As a result, the overall combination toxicity wassimilar in both cases. In other words, as the compounds were mixedin the ratio of their lethal concentrations for percentile 10, theirbiological actions and interactions varied in such a way that a unitaryconcentration of niclosamide ethanolamine salt became less lethalthan in the LC50 formulation, but a unitary concentration of the othercomponents became more toxic so that the two formulations as awhole produced similar effects.

The results discussed above seriously compromise the potential ofquaternary mixtures of potassium chloride, polyDADMAC, niclosa-mide ethanolamine salt and TCMTB for practical zebra mussel control,with two points deserving to be highlighted in this context. The firstis the fact that, by being less than additive, the quaternary mixturedoes not provide an effective reduction of the total toxic load requiredfor a certain level of pest mitigation. Formulating less than additivemixtures, in practice, corresponds to totally or partially replacing oneof the ingredients by a less toxic fraction of another. The second pointthat is worth discussing with regard to the potential of the mixturefor pest control concerns the fact that synergistic effects wereobserved under certain circumstances, meaning that a given mortal-ity percent was elicited by a biocide dosage lower than that requiredwhen the remaining chemicals were not present. This effect wasparticularly noticeable for polyDADMAC and TCMTB at high responselevels, which are those of major interest for practical control (Fig. 4).Due to such synergistic action, the quaternary mixture could possiblybe suggested as beneficial for use in situations where the reduction ofthe application requirements of one of these toxins is intended. Note,however, that at the target response range the synergism ofpolyDADMAC and TCMTB actions was achieved at the expense ofslight (Formulation LC50; Fig. 4a) to significant (Formulation LC10;Fig. 4b) antagonism of niclosamide ethanolamine salt activity.Moreover, it is likely that similar or even greater increases ofpolyDADMAC and TCMTB toxicities may be achieved through simplermixtures. Within this context, the reduced potential of the quaternarymixture can be illustrated by comparing the results obtained in thisstudy with those presented by Costa et al. (2011), who showed thatpotassium chloride alone significantly synergises polyDADMACaction. As shown in Fig. 5, the susceptibility of adult zebra musselsto combined potassium chloride and polyDADMAC together de-creased in the presence of niclosamide ethanolamine salt and TCMTB.Pest control through the quaternary mixture instead of the binaryblend would then be disadvantageous, implying not only theapplication of higher potassium chloride and polyDADMAC concen-trations, but also the dosing of additional biocides into the industrialplants.

Individual contributions to the mixture performance and modelling ofthe chemicals' joint action

The results of the integral mixture toxicity study did not provideinformation on the contribution of each biocide to the observedmixture performance nor on the possibility of optimising thecombination effects. In an attempt to obtain this type of information,several mixtures were tested following a central composite experi-mental design, and a second-order response surface model was fitted

to the toxicity data obtained (Box and Draper, 1987; Montgomeryand Runger, 1999). Ideally, this analysis would have resulted in anempirical model enabling the prediction of the mixture toxicity fromthe concentration of its components, which is impossible to obtainmechanistically (as illustrated by Fig. 6). The magnitude of theempirical model coefficients associated with the linear and quadratictoxin concentrations would have provided information on thecontribution of the biocides to the combined effect whilst themagnitude of the coefficient of the product terms would havecaptured the role played by the interactions between pairs ofchemicals. Unfortunately, the experimental results did not allow ameaningful mixture toxicity model to be obtained due to the highbackground variation of the response to the treatments. This type ofvariation is not uncommon, and other authors have experiencedsimilar difficulties in ecotoxicological studies. For example, for thesame reasons, Lock and Janssen (2002) did not succeed in developing

324 R. Costa et al. / Journal of Great Lakes Research 38 (2012) 317–325

an empirical model for the combined toxicity of zinc, cadmium,copper and lead to the potworm Enchytraeus albidus.

General approach, multivariate experiments and empirical modelling inmulticomponent molluscicide mixture design

Although the empirical modelling of the blend toxicity has notbeen successful in this study, the value of using multivariatestatistical techniques in the design of multicomponent mixtures forzebra mussel control should not be neglected as discussed below.

The first main challenge a pest manager faces when designing acombination of biocides is to define its qualitative composition.Preliminary selection of candidate ingredients will usually be basedon both the known chemicals' toxic activity and specific systemrequirements. The number of compounds chosen at this initial stagemay still be prohibitively large to explore in detail their joint action.Furthermore, there are two issues that have to be considered whendeciding on the complexity of the molluscicide formulation. First,ecotoxicological studies on mixture toxicity suggest that there is atendency for blends to be additive as the number of componentsincreases (Warne and Hawker, 1995; Warne, 2002) whilst more thanadditive combinations are more beneficial from the control perspec-tive. Second, potential operational and effluent purification problemsrelated to the dosage of a large number of substances into the systemhave to be outweighed by the enhanced performance of multiplebiocides. For these reasons, less complex biocide mixtures are usuallypreferable in practice. A systematic screening of the preliminary set ofmolluscicide candidates is thus useful in order to identify those thatcontribute the most to mixture performance as well as undesirableantagonistic interactions. In this context, two-level full factorial orfractional factorial experimental designs (Montgomery and Runger,1999) may prove valuable.

Once a promising set of chemicals is identified, the nature of theirjoint action can then be described by following the systematic approachproposed in this paper in order to assess the potential of theformulation. The qualitative composition of the mixture may beiteratively re-adjusted at any stage of the design process as theunderstanding of the behaviour of the biocides in the presence of eachother increases.

Another important challenge in designing molluscicide combina-tions for zebra mussel control is to be able to mathematically relate themixture performance to the concentration of the individual ingredients.A model of this type may be integrated in an optimisation framework,so that the final composition of the formulation is set to provide optimalperformance under system-specific economic, regulatory and environ-mental constraints. Typically, a mechanistic approach will not proveadequate to derive amodel formixture toxicity, because the concepts ofconcentration addition and response addition are often of limited use inpractice (Könemann and Pieters, 1996; van der Geest et al., 2000; deZwart and Posthuma, 2005; Barata et al., 2006). Therefore, the use ofmultivariate statistical tools for empirical modelling, such as responsesurface regression analysis (Box and Draper, 1987; Montgomery andRunger, 1999; Lock and Janssen, 2002), will be the preferred approachto obtain blend toxicity models.

Practical implications of the study

Even though mixtures of potassium chloride, polyDADMAC, niclo-samide ethanolamine salt and TCMTB were observed to synergise theaction of some of the ingredients, the quaternary combination is notrecommended for zebramussel control. Although thismajor result doesnot directly lead to an innovative approach to pest mitigation, thepresent study highlighted some aspects of practical relevance, whichare useful in the general context of designing multicomponentmolluscicide mixtures.

One of such aspects arises from the overview of the ecotoxicolog-ical literature on mixture toxicity. As stated by the funnel hypothesis,there is a propensity for multiple chemicals to act additively as thenumber of ingredients in the mixture increases (Warne and Hawker,1995; Warne, 2002). This provides key guidance in the process ofdesigning a new molluscicide blend. As more than additive combina-tions are those that should be achieved, there is no advantage informulating excessively complex mixtures.

This study also highlights the fact that biocide cocktails do notnecessarily increase the control effectiveness, even though this maybe intuitively expected. The mixture may produce mortalities higherthan those elicited by its individual ingredients but still exert a lessthan additive action, which is not particularly beneficial from thecontrol point of view. The overall mixture toxicity may also resultfrom the synergism of the action of some components at the expenseof the antagonism of others. Under these circumstances, the effectselicited by the mixture could be equally achieved by applying onlyone of the antagonised chemicals at a dosage lower than itsconcentration in the blend.

Finally, as a holistic practical contribution, this paper has alsoproposed a general approach that may be employed to designmulticomponent mixtures for zebra mussel control.

Acknowledgments

Financial support from the Portuguese Foundation for Science andTechnology (PhD scholarship SFRH/BD/18731/2004 and researchproject POCI/EQU/59305/2004) is gratefully acknowledged.

References

Ahmad, M., 2004. Potentiation/antagonism of deltamethrin and cypermethrins withorganophosphate insecticides in the cotton bollworm, Helicoverpa armigera(Lepidoptera: Noctuidae). Pestic. Biochem. Physiol. 80, 31–42.

Aldridge, D.C., Elliott, P., Moggridge, G.D., 2006. Microencapsulated BioBullets for thecontrol of biofouling zebra mussels. Environ. Sci. Technol. 40, 975–979.

Andrews, P., Thyssen, J., Lorke, D., 1982. The biology and toxicology of molluscicides.Bayluscide. Pharmacol. Ther. 19, 245–295.

Backhaus, T., Altenburger, R., Boedeker, W., Faust, M., Scholze, M., Grimme, L.H., 2000.Predictability of the toxicity of a multiple mixture of dissimilarly acting chemicalsto Vibrio fischeri. Environ. Toxicol. Chem. 19, 2348–2356.

Backhaus, T., Altenburger, R., Arrhenius, A., Blanck, H., Faust, M., Finizio, A., Gramatica,P., Grote, M., Junghans, M., Meyer, W., Pavan, M., Porsbring, T., Scholze, M.,Todeschini, R., Vighi, M., Walter, H., Grimme, L.H., 2003. The BEAM-project:prediction and assessment of mixture toxicities in the aquatic environment. Cont.Shelf Res. 23, 1757–1769.

Barata, C., Baird, D.J., Nogueira, A.J., Soares, A.M., Riva, M.C., 2006. Toxicity of binarymixtures of metals and pyrethroid insecticides to Daphnia magna Straus.Implications for multi-substance risks assessment. Aquat. Toxicol. 78, 1–14.

Box, G.E.P., Draper, N.R., 1987. Empirical Model-building and Response Surfaces. Wiley,Chichester.

Claudi, R., Mackie, G.L., 1994. Practical Manual for Zebra Mussel Monitoring andControl. Lewis Publishers, Boca Raton.

Costa, R., Aldridge, D.A., Moggridge, G.D., 2008. Seasonal variation of zebra musselsusceptibility to molluscicidal agents. J. Appl. Ecol. 45, 1712–1721.

Costa, R., Elliott, P., Aldridge, D.C., Moggridge, G.D., 2011. Enhanced mortality of thebiofouling zebra mussel, Dreissena polymorpha, through the application ofcombined control agents. J. Great Lakes Res. 37, 272–278.

de Zwart, D., Posthuma, L., 2005. Complex mixture toxicity for single and multiplespecies: proposed methodologies. Environ. Toxicol. Chem. 24, 2665–2676.

Durand-Hoffman, M.E., 1995. Analysis of physiological and toxicological effects ofpotassium on Dreissena polymorpha and toxicological effects on fish. MSc thesis.Ohio State University, Columbus, US.

ECETOC, 2001. Aquatic toxicity of mixtures. Technical Report 80. European Centre forEcotoxicology and Toxicology of Chemicals, Brussels.

Eide, I., 1996. Strategies for toxicological evaluation of mixtures. Food Chem. Toxicol.34, 1147–1149.

Elliott, P., Aldridge, D.C., Moggridge, G.D., Chipps, M., 2005. The increasing effectsof zebra mussels on water installations in England. Water Environ. J. 19,367–375.

Elzinga, W.J., Butzlaff, T.S., 1994. Carbon dioxide as a narcotizing pre-treatment forchemical control of Dreissena polymorpha. Paper Presented at Fourth InternationalZebra Mussel Conference. ICAIS, Madison.

Faust, M., Altenburger, R., Backhaus, T., Blanck, H., Boedeker, W., Gramatica, P., Hamer,V., Scholze, M., Vighi, M., Grimme, L.H., 2001. Predicting the joint algal toxicity ofmulticomponent s-triazine mixtures at low-effect concentrations of individualtoxicants. Aquat. Toxicol. 56, 13–32.

325R. Costa et al. / Journal of Great Lakes Research 38 (2012) 317–325

Fisher, S.W., Stromberg, P., Bruner, K.A., Boulet, L.D., 1991. Molluscicidal activity ofpotassium to the zebra mussel, Dreissena polymorpha — toxicity and mode ofaction. Aquat. Toxicol. 20, 219–234.

Fisher, S.W., Dabrowska, H., Waller, D.L., Babcock-Jackson, L., Zhang, X., 1994.Sensitivity of zebra mussel (Dreissena polymorpha) life stages to candidatemolluscicides. J. Shellfish Res. 13, 373–377.

Gloxhuber, C., 1974. Toxicological properties of surfactants. Arch. Toxicol. 32, 245–269.Groten, J.P., Schoen, E.D., Feron, V.J., 1996. Use of factorial designs in combination

toxicity studies. Food Chem. Toxicol. 34, 1083–1089.Groten, J.P., Feron, V.J., Suhnel, J., 2001. Toxicology of simple and complex mixtures.

Trends Pharmacol. Sci. 22, 316–322.Ishak, M.M., Sharaf, A.A., Mohamed, A.H., Mousa, A.H., 1970. Studies on mode of

action of some molluscicides on the snail, Biomphalaria alexandrina. 1. Effect ofBayluscide, sodium pentachlorophenate, and copper sulfate on succinate, glutamate,and reduced TMPD oxidation. Comp. Gen. Pharmacol. 1, 201–208.

Könemann, W.H., Pieters, M.N., 1996. Confusion of concepts inmixture toxicology. FoodChem. Toxicol. 34, 1025–1031.

Lock, K., Janssen, C.R., 2002. Mixture toxicity of zinc, cadmium, copper, and lead to thepotworm Enchytraeus albidus. Ecotoxicol. Environ. Saf. 52, 1–7.

Mackie, G.L., Claudi, R., 2010. Monitoring and Control of Macrofouling Mollusks inFresh Water Systems. CRC Press, Boca Raton.

Mallatt, J., Mccall, R.D., Bailey, J.F., Seelye, J., 1994. Effects of lampricides on the gillultrastructure of larval sea lampreys and rainbow trout fry. Can. J. Zool. 72, 1653–1664.

McMahon, R.F., 1996. The physiological ecology of the zebra mussel, Dreissenapolymorpha, in North America and Europe. Am. Zool. 36, 339–363.

McMahon, R.F., Shipman, B.N., Long, D.P., 1993. Laboratory efficacies of nonoxidizingmolluscicides on the zebra mussel (Dreissena polymorpha) and the Asian clam(Corbicula fluminea). In: Nalepa, T.F., Schloesser, D.W. (Eds.), Zebra Mussels:Biology, Impacts and Control. Lewis Publishers, Boca Raton, pp. 575–598.

Minchin, D., Lucy, F., Sullivan, M., 2002. Zebramussel: impacts and spread. In: Leppäkoski,E., Gollasch, S., Olenin, S. (Eds.), Invasive Aquatic Species of Europe: Distribution,Impacts and Management. Kluwer Academic Publishers, Dordreicht, pp. 135–148.

Montgomery, D.C., Runger, G.C., 1999. Applied Statistics and Probability for Engineers,2nd edn. Wiley, Chichester.

Nikl, D.L., Farrell, A.P., 1993. Reduced swimming performance and gill structuralchanges in juvenile salmonids exposed to 2-(thiocyanomethylthio)benzothiazole.Aquat. Toxicol. 27, 245–264.

Pimentel, D., Zuniga, R., Morrison, D., 2005. Update on the environmental andeconomic costs associated with alien-invasive species in the United States. Ecol.Econ. 52, 273–288.

Plackett, R.L., Hewlett, P.S., 1952. Quantal responses to mixtures of poisons. J. R. Stat.Soc. 14, 141–163.

Post, R.M., Mueller, M., Petrille, J.C., Shurtz, W.F., 1996. A decade of macrofoulingcontrol using non-oxidising compounds — an industry review. Paper Presented atEPRI Condenser Conference. EPRI, Boston.

Post, R.M., Petrille, J.C., Lyons, L.A., 2000. A review of freshwater macrobiological controlmethods for the power industry. Paper Presented at the 20th Annual Electric UtilityChemistry Workshop. University of Illinois, Champaign.

Schoen, E.D., 1996. Statistical designs in combination toxicology: a matter of choice.Food Chem. Toxicol. 34, 1059–1065.

Singh, P., Singh, V.K., Singh, D.K., 2005. Effect of binary combination of some plant-derived molluscicides with MGK-264 or piperonyl butoxide on the reproduction ofthe snail Lymnaea acuminata. Pest Manage. Sci. 61, 204–208.

Smith, A.T., 1998. Comparison of information-yield from different experimental designsused in algal toxicity testing of chemical mixtures. Environ. Pollut. 102, 205–212.

Sprague, J.B., 1970. Measurement of pollutant toxicity to fish. II. Utilizing and applyingbioassay results. Water Res. 4, 3–32.

Sprecher, S.L., Getsinger, K.D., 2000. Zebra Mussel Chemical Control Guide. ERDC/EL TR-00–1. US Army Corps of Engineers, Vicksburg.

Svendsen, C., Siang, P., Lister, L.J., Rice, A., Spurgeon, D.J., 2010. Similarity, independence,or interaction for binary mixture effects of nerve toxicants for the nematodeCaenorhabditis elegans. Environ. Toxicol. Chem. 29, 1182–1191.

Tajima, O., Schoen, E.D., Feron, V.J., Groten, J.P., 2002. Statistically designedexperiments in a tiered approach to screen mixtures of Fusarium mycotoxins forpossible interactions. Food Chem. Toxicol. 40, 685–695.

van Benschoten, J.E., Jensen, J.N., Brady, T.J., Harrington, D., Lewis, D.P., Sferrazza,J., 1992. Optimizing the use of chemical oxidants for control of the zebra mussel(Dreissena polymorpha). Final Report. Niagara Mohawk Power Corporation,Syracuse.

van der Geest, H.G., Greve, G.D., Boivin, M., Kraak, M.H.S., Gestel, C.A.M., 2000. Mixturetoxicity of copper and diazinon to larvae of the mayfly (Ephoron virgo) judgingadditivity at different effect levels. Environ. Toxicol. Chem. 19, 2900–2905.

Waller, D.L., Rach, J.J., Cope, W.G., Marking, L.L., Fisher, S.W., Dabrowska, H., 1993.Toxicity of candidate molluscicides to zebra mussels (Dreissena polymorpha)and selected nontarget organisms. J. Great Lakes Res. 19, 695–702.

Walsh, A.R., O'Halloran, J., 1997. The toxicity of leather tannery effluent to a populationof the blue mussel Mytilus edulis (L.). Ecotoxicol. 6, 137–152.

Warne, M.J., 2002. A review of the ecotoxicity of mixtures, approaches to, andrecommendations for their management. Paper Presented at Fifth NationalWorkshop on the Assessment of Site Contamination. EPHC, Adelaide.

Warne, M.J., Hawker, D.W., 1995. The number of components in a mixture determineswhether synergistic and antagonistic or additive toxicity predominate: the funnelhypothesis. Ecotoxicol. Environ. Saf. 31, 23–28.

Wildridge, P.J., Werner, R.G., Doherty, F.G., Neuhauser, E.F., 1998a. Acute toxicity ofpotassium to the adult zebra mussel Dreissena polymorpha. Arch. Environ. Contam.Toxicol. 34, 265–270.

Wildridge, P.J., Werner, R.G., Doherty, F.G., Neuhauser, E.F., 1998b. Acute effects ofpotassium on filtration rates of adult zebra mussels, Dreissena polymorpha. J. Great.Lakes Res. 24, 629–636.

Zar, J.H., 1999. Biostatistical Analysis. Prentice Hall, Upper Saddle River.