mimicking daphnia magna bioassay performance by an electronic tongue for urban water quality control

8
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/authorsrights

Upload: independent

Post on 19-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/authorsrights

Author's personal copy

Mimicking Daphnia magna bioassay performance by an electronictongue for urban water quality control

Dmitry Kirsanov a,b,*, Evgeny Legin b,c, Anatoly Zagrebin d, Natalia Ignatieva d,Vladimir Rybakin d, Andrey Legin a,b

a Laboratory of Chemical Sensors, St. Petersburg State University, St. Petersburg, Russiab Laboratory of Artificial Sensor Systems, ITMO University, St. Petersburg, Russiac Sensor Systems LLC, St. Petersburg, Russiad Institute of Limnology, Russian Academy of Sciences, St. Petersburg, Russia

H I G H L I G H T S G R A P H I C A L A B S T R A C T

� Daphnia magna bioassay can besimulated with multisensor system.

� Urbanwater toxicity can be predictedfrom potentiometric ET data.

� Independent test set validation con-firms statistical significance of theresults.

A R T I C L E I N F O

Article history:Received 13 November 2013Received in revised form 14 March 2014Accepted 15 March 2014Available online 20 March 2014

Keywords:BioassayWater toxicityMultisensor systemsElectronic tongue

A B S T R A C T

Toxicity is one of the key parameters of water quality in environmental monitoring. However, beingevaluated as a response of living beings (as their mobility, fertility, death rate, etc.) to water quality,toxicity can only be assessed with the help of these living beings. This imposes certain restrictions ontoxicity bioassay as an analytical method: biotest organisms must be properly bred, fed and kept understrictly regulated conditions and duration of tests can be quite long (up to several days), thus making thewhole procedure the prerogative of the limited number of highly specialized laboratories. This reportdescribes an original application of potentiometric multisensor system (electronic tongue) when the setof electrochemical sensors was calibrated against Daphnia magna death rate in order to perform toxicityassessment of urban waters without immediate involvement of living creatures. PRM (partial robust M)and PLS (projections on latent structures) regression models based on the data from this multisensorsystem allowed for prediction of toxicity of unknown water samples in terms of biotests but in the fastand simple instrumental way. Typical errors of water toxicity predictions were below 20% in terms ofDaphnia death rate which can be considered as a good result taking into account the complexity of thetask.

ã 2014 Elsevier B.V. All rights reserved.

1. Introduction

Water quality and safety is a question of highest importancenowadays. Extensive anthropogenic influence led to massive

hydrosphere pollution all over the world. The concerns of themodern society regarding this issue are reflected in numerouslegislative initiatives in this field, such as European Union WaterFramework Directive [1], Clean Water Act [2]. The number ofanalytical techniques and methods suggested for water qualitycontrol is really huge and growing. It would be hard to name ananalytical method which was not yet suggested for water analysis.Chromatographic, electrochemical, optical, etc. methods can be

* Corresponding author. Tel.: +7 812 328 28 35.E-mail address: [email protected] (D. Kirsanov).

http://dx.doi.org/10.1016/j.aca.2014.03.0210003-2670/ã 2014 Elsevier B.V. All rights reserved.

Analytica Chimica Acta 824 (2014) 64–70

Contents lists available at ScienceDirect

Analytica Chimica Acta

journal homepage: www.elsev ier .com/ locate /aca

Author's personal copy

effectively employed to solve certain particular tasks in this field.One of the integral parameterswhich are of ultimate importance inecological water control is its toxicity. Toxicity is understood as ameasure of a real harm to the living creatures caused by watersample. There is a special group of methods called bioassay whichis intended for direct estimation of water toxicity for biologicalobjects. The general idea behind this type of analysis is tomonitor abiological reaction of a biotest object placed in a water sampleunder study. This reaction is measured in some appropriate wayand compared with that of the same biotest object placed in acontrol sample of pure nontoxic water. Growth rate, mobility,survivorship rate of the test organisms, etc. can be monitored asbiological reactions [3]. Various aquatic biota species can beemployed for testing, such as e.g. algae, luminescent bacteria,infusoria, daphnia, fish, etc. An obvious advantage of this approachis direct information about biological harm that can be caused tothe test object by contaminated water, while most of the othermethods based on physicochemical measurements can onlyprovide information on the content of selected chemical sub-stances. This is, of course, useful but not always directly related tothe danger posed for living beings, especially when severalcontaminants are present in the water simultaneously that isquite common. It is important mentioning that sometimesdifferent substances with dissimilar chemical structures can causethe same degree of biological damage in various organisms [4].Different bioassays and test procedures were suggested and someof them are now accepted as standard analysis legitimized in ISOstandards, e.g. Vibrio fischeri (luminescent bacteria) [5], Daphniamagna [6], green algae [7]. It is noteworthy that none of thesuggested methods can serve as a universal toxicity assessmentinstrument because of the various sensitivity patterns of differentbioassays. Being an indispensable technique for toxicity assess-ment bioassay methodology still has a number of drawbacks: itsimplementation requires highly specialized laboratories; most ofthe methods are quite slow (usually it takes 24–72h to get theresults); biotest objects must be properly fed and kept underthoroughly regulated conditions.

There are numerous publications in literature where attemptswere made to overcome these bioassay drawbacks by developingnew sensor technologies, see e.g. recent reports [8,9]. The U.S.Environmental Protection Agency (EPA) has certified severalRapid Toxicity Testing Systems (RTTS), like e.g. MicroTox1 [10]. Athorough inspection of the protocols of such methods revealsthat all of them require either some biological objects (e.g.freeze-dried bacteria, usually luminescent Vibrio fischeri) or bio/chemical reagents or quite sophisticated preparative procedures.Besides, some of these methods require highly qualified person-nel for operation and in many cases reagent sets include bacteriaor living cells, which makes them costly and imply specialconditions for their storage. This makes such RTTS and the likemethods hardly suitable for continuous autonomous monitoringof water toxicity.

Electronic tongue (ET) methodology proved to be useful incomplex assessment of various liquidmedia [11–13]. The operationprinciple of such systems is quite simple – it suggests the use of anarray of cross-sensitive chemical sensors and subsequent process-ing of unresolved analytical signal from a sensor array by modernmultivariate statistic methods [14]. A cross-sensitivity of sensorsmeans that each sensor of the array has its own sensitivity andselectivity pattern and can respond with different response valueto various substances of interest in a sample. The resulted signalfrom all sensors of such array contains information on overallsample composition (according to sensitivity patterns of thesensors) and can be related to certain quality parameters of thesamples by means of multivariate data processing both inqualitative and quantitative way. The ET research field is rapidly

growing due to a distinct trend in modern analytical chemistrytowards fast and simple instruments [15,16].

Our preliminary research [17] with various aquatic organisms(Daphnia magna, Chlorella vulgaris, Paramecium caudatum) hasshown that potentiometric multisensor system being preliminarycalibrated against bioassay results for a set of samples with knowntoxicity allows for prediction of toxicity in new, totally unknownwater samples in terms of bioassay (e.g. survivorship rate). Thisprocedure does not require complex analytical instrumentation,sophisticated sample preparation and measuring procedures andelaborated biological object maintenance.

The idea of the present study was to extend our preliminarystudies by a full-scale experiment with sufficient number ofsamples to explore the opportunities and limitations of ETtechnology for water toxicity detection. For this purpose weparticipated in ongoing research initiative devoted to thecontinuous monitoring of urban water reservoirs in St. Petersburg(Russia). In the framework of this activity regular hydrochemicaland toxicity analysis of pond water had been performed, thus wewere able to analyze sufficient amount of samples by potentio-metric multisensor system. A full access to the results ofhydrochemical analysis and toxicity assessment permitted properstatistical validation of the ET predictive performance.

2. Materials and methods

2.1. Samples

A series of pond water samples was collected during severalexperimental sessions in July and August 2012. 29 ponds located inSt. Petersburg city area and its closest suburbs were sampled.Water samples from each pond were taken from two horizons –

surface layer (0.5m thick) and near-bottom layer, yielding 58samples in total. The two samples from different horizons of thesame pondwere notmixed together andwere treated separately inall further studies. Each sample was split into three portions: thefirst one for classical hydrochemical analysis, the second one forbioassay with Daphnia magna and the third one for electronictonguemeasurements. All samples had the same history of storageand transportation before various analyses. Due to diverse locationof sampling various pond water samples had various visualappearance, some of them were clear and transparent while theothers were turbid, with brownish color and distinct smell. Allsamples were encoded by a number and a letter: thus 15s stays forthe sample taken from the surface (s) of the 15th pond and e.g. 3bstays for the sample taken from the bottom (b) of the 3rd pond. Thedepth values for the studied ponds varied in the range 1–4.8mwith the average value 2.1m; traffic conditions in the immediateproximity of the ponds were from heavy to zero (for several pondslocated in the park areas). Temperature of the sampled water wasin the range 11–24 �C with the average value 18.7 �C.

2.2. Hydrochemical analysis

15 different hydrochemical parameters were determined for allsamples, these were pH, dissolved oxygen, total phosphorus (TP),inorganic phosphorus (IP), electric conductivity, carbon dioxide,ammonium, nitrite, nitrate, total nitrogen, color, chemical oxygendemand (CODCr), biochemical oxygen demand, suspended solids,oil hydrocarbons. The methods of their assessment and their rangein the studied samples are presented in Table 1. The determinationof these parameters is regulated by normative documentsdeveloped by Russian Federal State Budgetary Institution “StateHydrochemical Institute”.

D. Kirsanov et al. / Analytica Chimica Acta 824 (2014) 64–70 65

Author's personal copy

2.3. Daphnia magna bioassay

One- or two-day-old Daphnia magna Straus were employed in96h toxicity tests. Ten animals were used for a study in eachsample and all studies were performed in three replicate runs.25mL of sample was used for each replica. The sample wasconsidered being toxic when 50% or more of the animals died. Thefact of death was determined visually using optical microscope.The data from three replicas were averaged for further processing.Resulted toxicity values varied from 0 to 100% with the average44.4% and the median 35.0%. Due to technical issues bioassayresults were obtained only for 50 samples.

2.4. Multisensor system analysis

Multisensor system (Electronic Tongue) employed in this studywas based on 19 potentiometric cross-sensitive sensors. Ourprevious trial [17] employed 23 sensors, however during the dataprocessing we found that only 19 sensors have reasonably highregression coefficients in PLS models for prediction of watertoxicity in terms of Daphnia magna bioassay while other wereuseful for infusoria and chlorella prediction. Thus, only these 19sensorswere employed for described experiment. Six of themwerepoly(vinylchloride) (PVC)-plasticized anion-sensitive sensorsbased on various anion-exchangers, seven were PVC-plasticizedcation-sensitive sensitive sensors with neutral ligands described inRef. [18], and six sensors were chalcogenide glass electrodes withvarious RedOx sensitivity patterns. The details on sensor compo-sitions and preparation procedures are widely available inliterature, see e.g. [18–20]. Electrochemical measurements werecarried out in the following galvanic cell:

Cu|Ag|AgCl, KClsat|sample solution|sensor membrane|solidcontact|Cu

EMF values (sensor potentials) were measured with 0.1mVprecision against the standard silver/silver chloride referenceelectrode usingmade-in-house 32-channel digital high impedancevoltmeter connected to a PC for data acquisition. All samples weremeasured in four physically different replicas and the results wereaveraged for further processing. Replicated measurements ofdifferent samples were taken in random order. Water sampleswere measured as is, without any sample preparation procedures,

dilution, etc. The measurement time in each samplewas 3min andthree last readings were averaged and further employed for dataprocessing. After the measurements sensors were washed byseveral portions of distilled water for 2–3min. This procedureallowed for �5mV reproducibility in replicated measurements ofthe same samples. The measurements in pond water samples withmultisensor system were performed in parallel with bioassaymeasurements in physically different portions of the same pondwaters.

2.5. Data processing

The results of potentiometric measurements in the sampleswere combined intomatrix formwith samples in rows and sensorsin columns. Thus, each matrix element was the reading of thesensor k in the sample i. The dimensions of the ET matrix were 58samples�19 sensors. The same was done for the data ofhydrochemical measurements, which yielded the matrix 58samples�15 parameters. The data from bioassay resulted in avector 50 samples�1 parameter. We applied several differenttechniques to relate these two data sets with each other and also torelate multisensor system response with toxicity values producedby Daphnia magna bioassay. Principal component analysis (PCA),canonical correlation analysis (CCA), partial least squares regres-sion (PLS) and partial robust M-regression (PRM) were employedfor data analysis.

PCA is a widely applied method for data dimensionalityreduction and visualization of hidden data structure. The methodis based on projections of initial samples from initial multivariatespace onto a new coordinate space where new coordinate axes(principal components) are located in the direction of maximumvariance and are mutually orthogonal; see e.g. [21] for details.

CCA is aimed for the assessment of the correlation between twodata sets sharing the same row mode i.e. obtained for the samesample set [22]. To apply CCA we preliminary compressed the ETand hydrochemical data with PCA without significant informationloss and after that CCA was run on the resulting PC scores for bothdata sets. This approach was shown to be effective for ET dataprocessing in wine analysis [23].

PLS is a classical tool for multivariate regression, it finds theregression coefficients vector with a condition of covariance

Table 1Summary on hydrochemical methods.

Parameter Analytical method and instrumentation Range min–max(average)

Median

pH, pH units Electrometric determination by glass electrode, pH-meter (25 �C) 6.89–9.57 (8.07) 7.92Dissolved oxygen, mgL�1 Titrimetric determination by Winkler (iodometric method) 0.34–15.42 (6.65) 7.12Inorganic phosphorus, mgP L�1 Colorimetric determination with ammonium molybdate and ascorbic acid, l=880nm;

spectrophotometer0.003–2.492 (0.190) 0.30

Total phosphorus, mgP L�1 Oxidation with K2S2O8 with subsequent IP determination 0.023–2.787 (0.274) 0.097Electric conductivity, mSmcm�1 Conductometric determination in situ (25 �C); conductometer 139–1334 (670) 534Carbon dioxide, mgL�1 Calculative method based on pH and HCO3

� data 0.0–48.4 (8.7) 5.7Ammonium, mgNL�1 Colorimetric method with hypochlorite and phenol,

l =630nm; spectrophotometer0.0–6.271 (0.987) 0.030

Nitrite, mgNL�1 Colorimetric determination with sulphonilamide and N-(1-naphtyl) ethylenediamine,l =520nm; spectrophotometer

0.0–0.090 (0.006) 0.0

Nitrate, mgNL�1 Reduction of NO3� (Cu-Cd) with subsequent NO2

� determination 0.006–6.367 (0.031) 0.01Total nitrogen, mgNL�1 Oxidation with K2S2O8 with subsequent NO3

� determination 0.39–6.44 (1.50) 0.615Color, degrees Pt-Co Visual colorimerty by comparison with artificial standard (Pt-Co) 17–158 (42.3) 36Chemical oxygen demand,mgOL�1

Oxidationwith K2Cr2O7 with subsequent determination of K2Cr2O7 excess by titrationwith FeSO4

(NH4)2SO4�6H2O16.41–117.56 (55.10) 50.62

Biochemical oxygen demand,mgO2 L�1

Determination as a difference of dissolved oxygen content at the beginning of the experiment andafter 5 days incubation at 20 �C (bottle method)

1.03–16.75 (6.11) 4.89

Suspended solids, mg L�1 Gravimetric determination after filtration through 0.45mm membrane 1.2–100.7 (13.11) 7.1Oil hydrocarbons, mgL�1 IR spectrometric determination after extraction with CCl4 and Al2O3-column chromatography,

l =2700–3100 cm�1;IR-spectrophotometer

0.043–0.215 (0.110) 0.103

66 D. Kirsanov et al. / Analytica Chimica Acta 824 (2014) 64–70

Author's personal copy

maximization between the score matrices of reference andindependent data. PLS is widely applied in all fields of analyticalchemistry and described in literature in details, see e.g. [24].

Partial robust M-regression (PRM) is a robust multivariateregression technique which employs M-estimator instead of leastsquares estimator when constructing the regression and theimportant thing about M-estimator is that it uses weights to takeinto account the “leverage” of the samples. The results of PRM areusually muchmore stable against any outlying points compared toPLS. Mathematical introduction of PRM can be found in [25].

PCA and PLS were performed in The Unscrambler1 9.7 (CAMOSoftware AS, Norway). Model validation procedures are describedbelow in the text for each particular case.

CCA and PRM were performed using R 2.15.1 statisticalcomputation software [26]. For PRM calculations we employed‘‘chemometrics” R package by P. Filzmozer and K. Varmuza [27].

As a measure of predictive ability of the quantitative regressionmodelsweused the rootmean square errorof prediction/calibration

(RMSEP and RMSEC correspondingly), which is calculated as:

RMSE ¼ sqrtðsumððyi;ref � yi;modÞ2n

;

where n is the number of samples in the test set, yi,ref is thereference value of the parameter of interest, yi,mod is the valuepredicted by model for RMSEP or the value initially fitted by themodel for RMSEC.

3. Results and discussion

PCA was run separately on the data from hydrochemicalanalysis and ET measurements. The first six score vectorsaccounted for 97% of variance in ET data and the first sevenexplained 96% of variance for hydrochemical data. Correspondingscore matrices were then subjected to CCA.

The extracted values of the three first canonic roots were 0.964,0.851, 0.547, which implies that the reasonable match of the

[(Fig._1)TD$FIG]

Fig. 1. CCA similarity maps defined by canonical variates 1 and 2, for (a) ET analysis, and (b) hydrochemical measurements in pond waters.

D. Kirsanov et al. / Analytica Chimica Acta 824 (2014) 64–70 67

Author's personal copy

variance structure in two data sets was observed. In CCA similaritymaps (Fig. 1) one can compare e.g. the location of samples #53, 55and 17 in the lower part of the plots and #50, 41, 42 in the upperpart. This result assumes that there is a possibility of developingcertain regression models for numerical prediction of particularhydrochemical parameters from ET data. PLS models for estima-tion of CODCr, total nitrogen and conductivity were constructed asexamples of such correlations. Table 2 shows the parameters of“measured vs predicted” lines for these models. Validation wasperformed with independent test set obtained from random splitof samples into calibration (41 samples) and validation (17samples).

A reasonably good prediction performance can be observed forthese three parameters. Of particular practical interest is a goodcorrelation of the ET data with CODCr parameter since thetraditional procedure for its determination is quite long andtedious (sample oxidation with potassium bichromate andtitration of its excess). A correlation with conductivity is notsurprising since potentiometric response of the sensors is due tothe presence of the ions, which are also responsible for conductiveproperties of the sample. The rest of the hydrochemicalparameters, however, were not that well correlated with ETresponse. Typical R2 values in corresponding PLS regressionmodels were around 0.6. This is quite well understood taking intoaccount very complex composition of the surface water samples.Such parameters as e.g. total and inorganic phosphorus and carbondioxide can hardly be correctly measured by potentiometricsensors, due to a high hydrophilicity of the corresponding anions(phosphate and carbonate).

An attempt was made to predict the toxicity values fromhydrochemical data by constructing a PLS1 regression model withDaphnia bioassay data as an independent variable. The parametersof model validation were very poor (R2 = 0.28, RMSEP=30% in fullcross-validation), whichmeans that hydrochemical parameters arepoorly connected to Daphnia response, although they can stillprovide valuable information on water quality. Hotelling’s T2 testfor two-dimensional score plot of samples revealed severaloutliers, however, their subsequent elimination did not improvethe parameters of the PLS regression.

As the next step in this study we constructed PLS regressionmodel for prediction of toxicity in terms of Daphnia response fromET measurements. The first attempt resulted in very poorcorrelation (R2 = 0.25). A thorough inspection of X–Y relationoutliers plot detected several possible outliers, namely the samples4, 8, 27, 28 and 29 both from the surface and the bottom. Afterelimination of these samples from themodel R2 of 0.85 and RMSEPof 14% were obtained in full cross-validation. This confirms ourprevious findings in [17] that bioassay results can be simulatedwith reasonable precision by a multisensor system. The reason forwhich the above mentioned samples did not fit the model was

explored on the basis of hydrochemical data, corresponding pondlocation and other individual features. It was found that pond 4 hadvery strong organic pollution confirmed by high values of totalmicrobial count from additional microbiological analysis. Itappears that ET system exhibits certain sensitivity to theseunknown organic substances but Daphnia do not have it(corresponding toxicity values were moderate: 36.7% for surfaceand 40% for near-bottom layers) and, consequently, 4s and 4bsamples became outliers. The pond 8 is located along the roadwithheavy traffic and was possibly contaminated with heavy metalsand hydrocarbons. Another issue regarding this pond is that it hadsignificant amount of cyanobacteria, although it is not yet clearlyunderstood nowhow it can be related to ET response. The ponds 27and 29were supposedly polluted by organic substances, confirmedby microbial analysis, just like in case with the pond 8. The natureof deviations of pond 28’s water remained unclear.

In the presence of outliers robust regression methods canperform better [28], thus we attempted to use PRM regressioninstead of PLS to see if it can deal with outlying samples withoutthe need to eliminate them from the model. The resulted modelwith all of the available samples also exhibited poor correlation(R2 = 0.55 in cross-validation) although higher than that of ordinaryPLS (see above). Subsequent removal of the data from the ponds 4,8, 27, 28, 29 resulted in a reasonable PRM regression model withR2 = 0.83.

It is widely known that cross-validation of the multivariateregression models can often produce over-optimistic results [29–31]. To avoid any far-reaching conclusionswe performed validationof regression models for toxicity prediction using an independenttest set. This was done both for PLS and PRMmodels. The followingprocedure was employed for splitting the samples into calibrationand test set: for each of the analyzed ponds one sample from thesurface or from the near-bottom layer was randomly chosen forcalibration set while the remaining one for each pond wasemployed in the test set to assess the predictive ability of the ETsystem. The results of predictions are given in the Fig. 2.

It can be seen that both methods allow for reasonable precisionin toxicity assessment from potentiometric ET data. SummarizedRMSEP values for these test sets were 18% for PLS and 20% for PRMregression. Noticeable deviations from the reference toxicityvalues can be observed for samples 18s and 24b in both cases.Taking into account numerous advantages of ET methodology suchas fast and simple measurement protocol, absence of specificreagents, this technique can be suggested as a potential alternativefor traditional bioassay methods. Due to sufficient number ofsamples in this study we were able to validate the regressionmodels in a reliable and statistically significant way, thus theapplicability of potentiometric multisensor systems for Daphniamagna bioassay simulation was reliably proved. What is particu-larly interesting about multisensor approach is that toxicity

Table 2Parameters of ET performance for prediction of total nitrogen, COD and conductivity in pond water samples.

Slope Offset R2 RMSE

Total nitrogen (0.39–6.44), mgNL�1

Calibration 0.82 0.22 0.82 0.71Validation 0.86 0.05 0.81 1.11

Chemical oxygen demand (16.41–117.56), mgO L�1

Calibration 0.91 5.07 0.91 9.23Validation 1.03 -4.74 0.79 11.08

Conductivity (139–1334), mSmcm�1

Calibration 0.88 82 0.88 137Validation 0.84 78 0.91 128

68 D. Kirsanov et al. / Analytica Chimica Acta 824 (2014) 64–70

Author's personal copy

predictions in terms of various bioassays are potentially possiblewith the same sensor system. The conclusion can be done from asingle measurement by a sensor system, using preliminaryobtained calibration models for each bioassay type. Obviously,sensitivity of the sensor array in this case should cover the wholetoxicants range, which the reference bioassays are sensitive to. It isworth of mentioning that the regulatory limits established forparticular pollutants in potable water can be significantly lowerthan sensitivity thresholds both for daphnia and potentiometricsensors, thus proper care should be taken when translating thesestudies into neighboring fields. Another important issue is that wemade the calibrations against the 96-h daphnia tests and theresults of these tests do not take in account contaminants withprolonged and delayed toxicity. This gives a chance for underesti-mation of potential risks related to some herbicides, polyaromaticcompounds, etc. To take this into account a calibration againstprolonged toxicity tests would be required. We see the primarytasks for further development in the improvement of precision ofthis bioassay simulation and in clarification of platform applica-bility limits by examination of other bioassay methods as thesource of reference data for the multisensor array.

4. Conclusion

ET methodology adopted for the toxicity evaluation in terms ofbioassay opens up a perspective of routine water quality andtoxicity evaluation by such systems without using biologicalobjects at all. A multisensor system, being calibrated against theresponse of these various biological objects, can be applied forassessing the potential harm of contaminated water to theseobjects. Toxicity values of urban water samples in this study weredetermined by potentiometric electronic tongue in terms ofDaphnia magna death rate. Predictive models constructed from ETdata (PLS and PRM regression) were validated with independenttest sets. It was found that root mean square prediction errors didnot exceed 20%.We believe this is a very promising result since theanalytical task formulation is quite complex and rather unusual. Amultisensor approach can offer advantages of on-line measure-ments and continuous monitoring, simple and inexpensiveexperimental set-up and reagentless procedure. However, furtherextensive research in this field is required to increase the precisionof the method and to understand the reasons for certain observedoutliers.

[(Fig._2)TD$FIG]

Fig. 2. Prediction of toxicity in test set samples with PLS (a) and PRM (b) regression models.

D. Kirsanov et al. / Analytica Chimica Acta 824 (2014) 64–70 69

Author's personal copy

Acknowledgement

This work was partially financially supported by Government ofRussian Federation, Grant 074-U01.

References

[1] EU Directive, 200. Directive 2000/60/EC of the European Parliament and of theCouncil of 23 October 2000 establishing a framework for Community action inthe field of water policy. Official Journal L 327, 22/12/2000 P. 0001–0073.http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:32000L0060:EN:HTML, last accessed on 12 October 2013.

[2] US Senate, 2002. FederalWater Pollution Control Act, http://www.epw.senate.gov/water.pdf, last accessed on 12 October 2013.

[3] C.J. Keddy, J.C. Greene, M.A. Bonnell, Ecotoxicology and Environmental Safety30 (1995) 221.

[4] C. Sonnenschein, A.M. Soto, Journal of Steroid Biochemistry 65 (1998) 143.[5] ISO 11348-1, 2007. Water quality – Determination of the inhibitory effect of

water samples on the light emission of Vibrio fischeri(Luminescent bacteriatest) – Part 1: Method using freshly prepared bacteria.

[6] ISO 6341, 2012.Water quality –Determination of the inhibition of themobilityof Daphnia magna Straus (Cladocera, Crustacea) – Acute toxicity test.

[7] ISO 8692, 2012. Water quality – Fresh water algal growth inhibition test withunicellular green algae.

[8] R.S.J. Alkasir, M. Ornatska, S. Andreescu, Analytical Chemistry 84 (2012) 9729.[9] J. Li, Y. Yu, Y. Wang, J. Qian, J. Zhi, C. Blaise, J.-F. Férard (Eds.), Electrochimica

Acta (2013).[10] B. Thomas Johnson, Microtox1 Acute Toxicity Test, in: C. Blaise, J.-F. Férard

(Eds.), Small-scale Freshwater Toxicity Investigations, vol. 1 Toxicity TestMethods, Springer, Berlin, 2005 pp. 69–105.

[11] P. Ciosek, W. Wroblewski, Talanta 76 (2008) 548.[12] Z. Wei, J. Wang, Analytica Chimica Acta 694 (2011) 45.

[13] X. Cetó, J.M. Gutiérrez, M. Gutiérrez, F. Céspedes, J. Capdevila, S. Mínguez, C.Jiménez-Jorquera, M. Del Valle, Analytica Chimica Acta 732 (2012) 172.

[14] A. Legin, A. Rudnitskaya, Yu. Vlasov, C. Di Natale, E. Mazzone, A. D‘Amico,Sensors and Actuators B: Chemical 65 (2000) 232.

[15] A. Riul Jr., C.A.R. Dantas, C.M. Miyazaki, O.N. Oliveira Jr., Analyst 135 (2010)2481.

[16] H. Smyth, D. Cozzolino, Chemical Reviews 113 (2013) 1429.[17] D. Kirsanov, O. Zadorozhnaya, A. Krasheninnikov, N. Komarova, A. Popov, A.

Legin, Sensors and Actuators B: Chemical 179 (2013) 282.[18] D. Kirsanov, M. Khaydukova, L. Tkachenko, A. Legin, V. Babain, Electroanalysis

24 (2012) 121–130.[19] P. Ciosek, W. Wroblewski, Talanta 76 (2008) 548–556.[20] G. Yu. Mourzina, M.J. Schöning, J. Schubert,W. Zander, A.V. Legin, G. Yu. Vlasov,

H. Lüth, Analytica Chimica Acta 433 (2001) 103–110.[21] K.H. Esbensen, Multivariate data analysis – in practice, An Introduction to

Multivariate Data Analysis and Experimental Design, 5th ed., CAMO AS Publ.,Oslo, 2001.

[22] J.F. Hair, R.E. Anderson, R.L. Tatham,W.C. Black, Multivariate Data Analysis, 5thed., Prentice Hall Inc., USA, 1998.

[23] D. Kirsanov, O. Mednova, V. Vietoris, P.A. Kilmartin, A. Legin, Talanta 90 (2012)109.

[24] S. Wold, M. Sjöström, L. Eriksson, Chemometrics and Intelligent LaboratorySystems 58 (2001) 109.

[25] S. Serneels, C. Croux, P. Filzmoser, P.J. Van Espen, Chemometrics and IntelligentLaboratory Systems 79 (2005) 55.

[26] R Development Core Team, R: A Language and Environment for StatisticalComputing, R Foundation for Statistical Computing, Vienna, Austria,2010http://www.R-project.orghttp://cran.r-project.org/web/packages/che-mometrics/index.html.

[27] http://cran.r-project.org/web/packages/chemometrics/index.html[28] M. Daszykowski, K. Kaczmarek, Y. Vander Heyden, B. Walczak, Chemometrics

and Intelligent Laboratory Systems 85 (2007) 203.[29] R.G. Brereton, TrAC-Trends in Analytical Chemistry 25 (2006) 1103.[30] P. Filzmoser, B. Liebmann, K. Varmuza, Journal of Chemometrics 33 (2009) 160.[31] K.H. Esbensen, P. Geladi, Journal of Chemometrics 24 (2010) 168.

70 D. Kirsanov et al. / Analytica Chimica Acta 824 (2014) 64–70