This article was downloaded by: [Umeå University Library]On: 12 November 2014, At: 17:13Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number:1072954 Registered office: Mortimer House, 37-41 Mortimer Street,London W1T 3JH, UK

Journal of EnvironmentalScience and Health .Part A: EnvironmentalScience and Engineeringand Toxicology: Toxic/Hazardous Substances andEnvironmental EngineeringPublication details, including instructions forauthors and subscription information:http://www.tandfonline.com/loi/lesa19

Comparative analysis ofnearshore contaminatedsites in lake Ontario:Ranking for environmentalhazardRainer Brüggemann a c & Efraim Halfon ba GSF‐Forschungszentrum für Umwelt‐undGesundheit , PUC , Ingolstädter Landstraße 1,Oberschleißheim, D‐85758, Federal Republicof Germanyb Canada Centre for Inland Waters , NationalWater Research Institute , Burlington, ON,L7R 4A6, Canadac Institute of Freshwater Ecology and InlandFisheries , Rudower Chaussee 5, Berlin,12484, GermanyPublished online: 15 Dec 2008.

To cite this article: Rainer Brüggemann & Efraim Halfon (1997) Comparativeanalysis of nearshore contaminated sites in lake Ontario: Ranking forenvironmental hazard, Journal of Environmental Science and Health .Part A: Environmental Science and Engineering and Toxicology: Toxic/

Hazardous Substances and Environmental Engineering, 32:1, 277-292, DOI:10.1080/10934529709376540

To link to this article: http://dx.doi.org/10.1080/10934529709376540

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of allthe information (the “Content”) contained in the publications on ourplatform. However, Taylor & Francis, our agents, and our licensorsmake no representations or warranties whatsoever as to the accuracy,completeness, or suitability for any purpose of the Content. Anyopinions and views expressed in this publication are the opinions andviews of the authors, and are not the views of or endorsed by Taylor& Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information.Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilitieswhatsoever or howsoever caused arising directly or indirectly inconnection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private studypurposes. Any substantial or systematic reproduction, redistribution,reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of accessand use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

J. ENVIRON. SCI. HEALTH, A32(l), 277-292 (1997)

Comparative Analysis of NearshoreContaminated Sites in Lake Ontario:Ranking for Environmental Hazard

Rainer Brüggemann*GSF-Forschungszentrum für Umwelt-und Gesundheit, PUC

Ingolstädter Landstraße 1D-85758 Oberschleißheim

Federal Republic of Germany

Efraim HalfonNational Water Research InstituteCanada Centre for Inland WatersBurlington, ON L7R 4A6 Canada

Abstract

A ranking method, based on partially ordered set theory, is applied todegraded inshore waters of Lake Ontario. This ranking method uses testdata collected by Dutka et al. [1] in bottom sediments. The result of thisranking analysis is displayed by Hasse diagrams. Hasse diagrams avoidthe loss of information that occurs when data are aggregated into a rank-ing index. Both ranking schemes, that of Dutka et al. [1] and ours, haveidentified the same sites as the most degraded: Humber River STP out-fall, Mimico Creek mouth and the harbour in Port Hope. In addition, wealso identify Whitby Harbour and another site in Port Hope Harbour asdegraded. These two sites, together with the Industry's Area in Toronto

* Address correspondence to Dr. Rainer Briiggemann, Institute of Freshwater Ecology andInland Fisheries, Rudower Chaussee 5, 12484 Berlin, Germany.

277

Copyright © 1997 by Marcel Dekker, Inc.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

278 BRUGGEMANN AND HALFON

and Lasco Steel in Whitby Harbour, are degraded in a manner differentfrom all other sites (because samples from these sites are highly toxic).We also note that Cherry St. in Toronto (Site 27) should be considereddegraded because of the high ranking in fecal coliforms (FC). Finally,the ranking method identifies two tests, namely the Microtox test andthe genotoxicity test as important for ordering the sites.

Introduction

Dutka et al. proposed a series of complementary, microbiological, biochem-ical, and bioassay tests to quantify degradation of different sites [1]. Thissystem of tests was called a "battery of tests". The information from thesetests was summarized into an index used to rank the sites. The use of any in-dex, however, has the disadvantage that information from each test is lost be-cause test results are aggregated. Zitko approached the same problem from astatistical viewpoint using Principal Component Analysis to map 12 sedimentquality parameters on three principal components to facilitate the visual iden-tification of similar sites and samples [2]. In this paper, we use a previouslypublished ranking method [3-5] that preserves the information from all testsperformed at each site, and we apply it to sediment samples of Lake Ontario.Our method does not merge results from different tests (e.g., toxicity withquality tests) as it is done in the construction of a ranking index. The methodcan be used to identify tests which are important in ranking and those whichmay be eliminated without substantial loss of information. The ranking canbe graphically displayed by Hasse diagrams.

Data

Dutka et al. collected bottom sediment samples at a depth of 2-3 cm at 50(not 51 as described in their paper) sites along the Canadian shore of LakeOntario, from Kingston to the Niagara River. Six samples were taken fromPort Hope Harbour at sites within an area of about 100 m in diameter, there-fore 55 samples from 50 sites were taken. The other samples were taken from44 nearshore sites in Lake Ontario. Their investigation was intended to be asnapshot (June 1985) of the state of degradation. Therefore, they did notperform any time series or pooling of samples [1].

The "Battery of Tests"

The following measurements were performed:

1. Number of fecal coliforms/£.co// (FC), estimated by the membrane fil-

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

CONTAMINATED SITES IN LAKE ONTARIO 279

tration technique and the technique of most-probable-number (MPN).Medium used was mFC agar.

2. Concentration of coprostanol (COP).3. Concentration of cholesterol (CHO).4. Tbxicity of luminescent bacterium Photobacterium phosphoreum, with a

contact time of 15 minutes (Microtox-test, MT).5. Genotoxicity (GT).

The microbiological and biochemical tests are described in detail byDutka et al. [1]. Two interesting tests which will be explained in depth laterare GT and CHO: GT is a calorimetric assay of en2yme activity; CHO wasmeasured according to the APHA standard method [6].

Scoring of the Test Results

Experimental data were grouped and a score was assigned to each of the re-sulting classes. To allow comparison of our results with those of Dutka et al.,we used their classes and scores (Table 1). All scores are presented in Table 2.

Ranking Sites

Ranking with Aggregated Scores

It is difficult to rank polluted sites using multiple measures of degradation.When multiple tests are used to rank sites, different combinations of resultsare possible. The data from Table 2 can be used to exemplify the three possi-bilities:

1. Some sites, for example, 27 and 47, are degraded to the same extentas was also reported by Dutka et al. since their scores are identical(Table 2).

2. Some sites, for example, 27, 25, and 4, can be ranked unequivocally,because all scores of one site are higher than those of another one. Forexample, Site 27 is more degraded than Site 25, which in turn is moredegraded than Site 4, because the score of FC is decreasing in that orderand the scores of all other tests are identical.

3. Some sites, for example, 31 and 27, can not be ordered (ranked) inany way. Site 31 is less degraded than Site 27 when FC is considered.However, Site 27 is less degraded than Site 31 when COP and CHO aretaken into account. In other words, these two sites, are incomparable,because different processes of degradation occurred.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

280 BRUGGEMANN AND HALFON

Table 1: Scores awarding scheme used to rank samples, based on suspectedhazards (Dutka et al. [1], modified).

Fecal coliform/E. coliSediment lOg/100 mL MPN

<0.10.1-100100-500500-25002500-10,000> 10,000

CoprostanolSediment ppm<0.10.1-11-33-55-7> 7

MicrotoxEC50/g wetwt or /mL<0.10.1-0.20.2-0.30.3-0.40.4-1000>1000

Score

012345

Score

0135710

Score

1086420

CholesterolSediment ppm

<0.10.1-22-44-66-8>8

GenotoxicityEquivalent to

ng/mL4NQO °0

0-200200-400400-600600-800

>800

Score

012345

Score

0246

>80010

°4-nitro-quinoline-oxide

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

CONTAMINATED SITES IN LAKE ONTARIO 281

To avoid the ambiguity or contradiction that arises from incomparablesites, Dutka et al. summed the scores to obtain an overall degradation in-dex [1]. Sites with a high index were identified as degraded sites which couldnot be used for drinking and/or swimming. Note that the use of an index hidesthe fact that sites are degraded in different ways. Thus, information is lost.

Ranking with Hasse Diagrams

This loss of information is avoided if Hasse diagrams are applied [7]. Thetheory of Hasse diagrams has been discussed in text books [8, 9,10].

Halfon and Reggiani [4], Halfon [5], Bruggemann and Halfon [11],Bruggemann and Miinzer [12], Steinberg et al. [13], Bruggemann et al. [14],Munzer et al. [15], and Voigt and Bruggemann [16] applied this method to en-vironmental problems. The aim of this method is to use a formal procedureto order (or rank) polluted sites according to results of a test battery. The re-sults of n tests form a vector with n components, called an "n-tuple". In theranking method by Hasse diagrams, two conditions, antisymmetry and transi-tivity must be fulfilled simultaneously for objects for which an order relationholds true. "Order is not a property intrinsic to a single object", it concernscomparison between two objects: 0 is smaller than 1; "Mars is farther fromthe Sun than Earth..." [10]. An example of antisymmetry is the relation "5 islarger than 3 but 3 is not larger than 5" and transitivity means that from 0 <1 and 1 < 1000 we can deduce that 0 < 1000. Furthermore, there are fourdescriptive terms of ordering relations, strict and non-strict, total and non-total. These relations are as fundamental to our purposes and to ecology ingeneral as are the previous two conditions of antisymmetry and transitivity.The statement "Site x is more polluted by a chemical than Site y" means thatx is strictly more polluted than y. In general, the statement "Site x is pollutedexactly as Site y" cannot be excluded. Thus, a non-strict order arises. If allobjects can mutually be compared, then a total order arises, the objects are"comparable". However, the statement "Site x is polluted by chemicals Aand B and Site y is more polluted by A and less polluted by B than x" meansthat there is an ambiguity between the pollution status of x and y. Sites xand y are incomparable. The presence of such incomparable objects withinan ordering scheme is explicitly accounted for by the term non-total order orpartial ordering. For mathematical details, see Davey and Priestley [10]. Inthe following, nine rules are described, which are inherent to the constructionof Hasse diagrams and important for our purpose of ranking in the field ofecology.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

282 BRUGGEMANN AND HALFON

Table 2: Scores of 55 sediment samples from Lake Ontario. Five tests areused to rank the samples: (1) Fecal coliforms, (2) coprostanol, (3) choles-terol, (4) Microtox and (5) genotoxicity (See Table 1).

Identifier

12345678

9591929394

91011121314151617186019202122

FC

2123332132323131331313113233

COP

0000300050000000000000000000

CHO

0000200020000000000000000000

MT

4220008260400600208406200000

GT

0000000000000200000000440000

Site/Sample

Cataraqui River, KingstonCarruthers Point, KingstonDeserontoNapanee RiverOutfall area, Belleville STPMoira RiverTrent RiverCoburgHarbour - Port Hope (9A)a

Harbour - Port Hope (9D)Harbour - Port Hope (9H)Harbour - Port Hope (91)Harbour - Port Hope (9M)Harbour - Port Hope (9T)Breakwall - Port HopeNewcastleBowmanvilleBowmanville CreekRuby HeadMarina OshawaOshawaCorbett Geek, WhitbyHarbour WhitbyLasco Steel (18A)Duffin CreekRouge RiverHighland CreekScarborough

*Labelused byDutkaetal. [1]

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

CONTAMINATED SITES IN LAKE ONTARIO 283

Table 2: Scores of 55 sediment samples from Lake Ontario. Five tests areused to rank the samples.

Identifier FC COP CHO MT GT Site/Sample

Industry's Area, TorontoBetween Toronto IslandsSTP, TorontoHarbour, TorontoCherry St., TorontoOntario Place, TorontoSunnyside Beach, TorontoHumber River, TorontoSTP outfall, Humber RiverMimico CreekEtobicoke CreekLakeview GeneratorMouth of Gedit RiverOpposite Gulf Oil Plant16 Mile CreekBronte CreekPetro Canada PierSpencer Smith ParkEntrance to Burlington CanalGrimbsy BeachJordan HarbourMouth of Pourth DalhousieInside of Bay, Port DalhousiePort Weller

Mouth of Niagara RiverMouth of Niagara RiverMouth of Niagara River

232425262728293031323334353739404142434445464748495051

124252334352322111111553222

000000005000000000000000000

000000004000000000000000000

000000000800600000000000000

400000000000000000000000000

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

284 BRUGGEMANN AND HALFON

1. Hasse diagrams (Figures 1-3) can be constructed with unmodified ex-perimental data as well as with scores (as is done here). The use ofscores reduces uncertainty because the variation of data within the classinterval does not change the score. (An optimized definition of inter-vals is outside the scope of this paper. To enable comparisons of ourresults with those of Dutka et al. we have to take their classificationscheme. [1])

2. By convention, highly degraded sites are located on the top of the Hassediagram. Less degraded sites are displayed at the bottom.

3. The sites are represented by small circles and identified numbers. Nextto the circles, the corresponding scores (FC, COP, CHO, MT, GT) areprovided to allow an easier understanding of the ranking.

4. Identical ranks are attributed to sites characterized by identical scores.For example, this is true for Sites 27,33,46, and 47, which are equivalentto each other. In the diagram, one of the sites (27 in our example)appears as a representative and the other sites are given at the bottomof the diagram. In mathematical terms, Sites 27, 33, 46, and 47 areelements of the same equivalence class. The equivalence relation is:"Equality of the scores". Therefore, it is sufficient to select one elementof the equivalence class as "representative" of the whole class.

5. Sites that are "comparable" with respect to all tests are connected bylines. For example, Site 27 is more degraded for all tests than Site 25.In the graph, the lines should be followed in one direction only, fromtop to bottom, or (exclusively) from bottom to top.

6. Sites that are not comparable are not connected ("incomparable" sites).For example, Sites 27 and 31 can not be compared.

7. For clarity, circles are arranged into levels. "Incomparable" sites canalso be placed on different levels.

8. The presence of a connection between two circles, either directly orindirectly through other circles, indicates that the site on the superiorlevel has larger scores than the site located on a lower level with respectto all tests. As conclusion: Comparable sites cannot be located on thesame level, because they have to be connected by a line (rule 5). Byrules 2 and 8, they must be located on different levels.

9. If possible, a circle is located at the top of the diagram, to be conser-vative. For example, Site 9 is connected with Site 2 (Figure 1). Eachposition above Site 2 would be allowed, in order to represent the factthat all scores of Site 9 are simultaneously equal or larger than thoseof Site 2. However, Site 9 has no upper neighbors, therefore, Site 9 isplaced on the top level.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

CONTAMINATED SITES IN LAKE ONTARIO 285

(5,0,0,0,0) (4,5,4,0,0) (3,5,2,6,0) (3,0,0,8,0) (1,0,0,6,2) (1,0,0,2,4)

^ r^ rr(1,0,0,0,4)

(1,0,0,0,0)

Equivalent samples:{2,8}{4,6,10,13,19,21,22,29,30,48,94}{11,16,40,41,42,43,44,45}{15,92}{17,35}{20,24,26,28,34,37,39,49,50,51.91,93}{23,60}{27,33,46,47}

Figure 1: Hasse diagram ranking 55 bottom sediment samples taken from 50different sites of Lake Ontario. The ranking was based on five tests (Numberof fecal coliforms, concentration of coprostanol, concentration of cholesterol,Microtox and genotoxicity test). The numbers within each circle identify asite. Next to each circle, its individual tuple of scores are given. At the bottomof the picture, sites that occupy the same position in the Hasse diagram areidentified (equivalent sites). Sampling locations are presented in Table 2. Onthe top of the diagram there are the most degraded sites, on the bottom theleast degraded ones. For the sake of simplicity, the sites are organized inlevels.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

286 BRUGGEMANN AND HALFON

(5.0.0) (3,8,0) _ (1,6,2) (1,2.4)

(1.0,4)

Equivalent samples:{2,8}{4,5,6,10,13,19,21,22,29,30,48,94}{11,16,40,41,42,43,44,45}{15,92}{17,35.95}{20,24.26,28.34,37,39,49,50,51,91,93}{23,60}{25,31}{27,33,46,47}

Figure 2: As in Figure 1, but now only three tests are used for ranking: num-ber of fecal coliforms, Microtox and genotoxicity tests.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

CONTAMINATED SITES IN LAKE ONTARIO 287

Equivalent samples:

{1,3,7,20,24,26,28,34,37,39,49,50,51,91,93}

{2,8,9,11,14,16,18,23,40,41,42,43,44,45,60}

{4,6,10,12,13,15,17,19,21,22,29,30,32,35,48,92,94}

{27,33,46,47}

Figure 3: As in Figure 2, but now ranking was based on the number of fecalcoliforms, coprostanol and cholesterol concentrations.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

288 BRUGGEMANN AND HALFON

These nine rules allow a user to understand a Hasse diagram. The Hassediagrams that follow (Figures 1-3) point out relations among the data.

Results and Discussion

Using All Five Tests Used as Basis for Ranking

The concept of incomparability is the most central point of Hasse diagrams,therefore some examples will be discussed in more detail. Test results be-tween Sites 27 and 31 are undoubtedly different. We have chosen these twosites to make a number of remarks. Even if Site 27 may look on a first sightless polluted than 31 (see below), some ambiguities are present that needto be discussed in detail. Site 27 (Cherry Station, Toronto), together with itsequivalent elements (33,46, and 47) and Site 31 (STP outfall, Humber River)are located on the same level (Figure 3) and are not connected by a line. Anextraction of Table 2 shows:

FC COP CHO MT GTSite 27 5 0 0 0 0Site 31 4 5 4 0 0

Site 27 is one score higher than Site 31 according to the first test, but 5and 4 scores, respectively lower than Site 31, according to the second andthird test. The Hasse diagram (Figure 1) demonstrates this contradiction,because the two sites are not connected. The two sites are "incomparable".Thus, even if Site 31 seems to be more hazardous than Site 27, because threetests show a rather high degradation, Site 27 should nevertheless be regardedas suspect because of the higher ranking in FC. This information would bemasked if a single index was used for ranking. The Hasse diagram does notexclude the assessment that Site 31 may be more hazardous than Site 27, but itreveals that the type of degradation is different. Note that Dutka et al. rankedSite 31 as the second worst but Site 27 as the 17th [1]. No mention was madeof the high level of FC at Site 27. Furthermore, Figure 1 shows that 31 othersites (which are connected with 27 in a downward direction) have the samekind of degradation pattern (rules 2,5 and 8). This degradation is describedby non-zero FC scores, decreasing from Site 25 to Site 45 and zero scores inthe four remaining tests.

Dutka et al. [1] and we identify Sites 31 (Humber River STP outfall), 32(Mimico Creek mouth) and 95 (Harbour, Port Hope) as the most degradedones. These three incomparable sites are located on the top level of the Hasse

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

CONTAMINATED SITES IN LAKE ONTARIO 289

diagram (Figure 1) but they have different degradation patterns. Figure 1shows the reason: Site 31 is degraded due to high scores in FC, COP andCHO; Site 32 is degraded because of the MT score and Site 95 has high scoresfor COP and MT.

Site 9 is more degraded than the Sites 2 and 11 (rule 8) and is located onthe top level of the Hasse diagram because of rule 9. The same argument istrue for the Site 18 which is "incomparable" to all other sites, except Sites 23,2, and 11 and the other sites they represent. Sites 9,18, and 23 are degradedin a manner different from all other sites, because the toxicity test GT and/orMT have scores different from zero. The Hasse diagram points out this factgraphically.

Sensitivity Analysis for Ranking

The ranking of sites may depend on tests which are used. Usually the rankingchanges if some of the tests are omitted. If, for example, GT was omitted,Sites 9, 18, and 23 would appear as almost not degraded (see below). Onthe other hand, the elimination of other tests may not change the ranking asdrastically. If, for example, CHO and COP are ignored, a new ranking and,subsequently, a new Hasse diagram results (Figure 2). Deleting MT and GTyields a third ranking diagrammed in Figure 3. These figures are comparedto assess the importance of tests according to our methodology. Figure 2retains almost all features of the original Hasse diagram (Figure 1). Howeverthe third Hasse diagram (Figure 3 shows only a few contradictions: almostall sites are comparable. Many sites are no more differentiated by the threetests used. Our conclusion is that MT and/or GT are more important thanCOP and CHO because the elimination of MT and/or GT changes the Hassediagram dramatically. For example, Sites 9, 18, and 23 are still indicated ashighly degraded in Figure 2 but not in Figure 3.

The apparent trial and error sensitivity analysis can be systematized. Analgorithm that avoids the drawing and laborious comparative analysis ofHasse diagrams will be published later.

Conclusion

Several benefits derive from this ranking analysis: The most polluted sitescan easily be identified. Typically, the Hasse diagram lists more than one siteas the most environmentally degraded ones. Each of these sites is corruptedby a different combination of pollutants, therefore, the sediment quality may

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

290 BRUGGEMANN AND HALFON

be improved through different control options. Furthermore, this techniqueidentifies the more important and the less important criteria to rank the sites.This leads to a reduction of environmental monitoring costs. It is thus of greatvalue for environmental studies, where uncertainty and sampling costs are of-ten big issues. Both ranking schemes, those of Dutka et al. [1] and ours, haveidentified the same sites as the most degraded ones. In addition, our analysishas pointed out locations not considered by Dutka et al. [1] as being exces-sively degraded. The ranking scheme using Hasse diagrams has also iden-tified different degradation patterns that are not immediately evident whenan index is used. The analysis of Hasse diagrams, therefore, allows the re-construction of the ranking and facilitates the decision making process. Itidentifies tests which are needed in future surveillance projects. Tests that donot have much influence on the ranking of sites, such as COP and CHO inour example, can be identified.

Acknowledgments

We thank Dr. Cook for his improvements to the manuscript. Dr. Dutkakindly discussed his results, Natalie Schito, Dr. Klaus Kaiser, Andreas Kaune,and Kristina Voigt reviewed the manuscript. We thank the German Ministryfor Research and Technology for supporting this work within the Canada-German cooperation program.

Notation

CHO concentration of cholesterolCOP concentration of coprostanolFC fecal coliformsGT genotoxicityMPN most-probable-numberMT Microtox-test

References

[1] B.J. Dutka, K. Walsh, K.K. Kwan, A. El Shaarawi, D.L. Liu K., andThompson. Priority site selection for degraded areas based on micro-bial and toxicant screening tests. Water Poll. Res. J. Canada, 21:267-282,1986.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

CONTAMINATED SITES IN LAKE ONTARIO 291

[2] V. Zitko. Assessment of the significance of chemicals in sediments. J.Env. Sci. Health, A27:273-281, 1992.

[3] E. Halfon. Is there a best model structure? II. Comparing model struc-tures of different fate models. Ecological Modeling, 20:153-163, 1983.

[4] E. Halfon and M. Reggiani. On ranking chemicals for environmentalhazard. Environ. Sci. Technol., 20:1173-1176, 1986.

[5] E. Halfon. Comparison of an index function and a vectorial approachmethod for ranking waste disposal sites. Environ. Sci. Technol, 23:600-609, 1989.

[6] APHA 1985. Standard Methods for the Examination of Water and Wastew-ater. APHA, Washington, D. C., 16 edition.

[7] M.G. Reggiani and F.E. Marchetti. On assessing model adequacy. IEEETrans. Systems Man Cyber., SMC-5:322-330, 1975.

[8] F. Harary. Graph Theory. Addison-Wesley, Reading, Mass., 1969.[9] F.P. Preparate and R.T. Yeh. Introduction to Discrete Structures. Addison-

Wesley, Reading, Mass., 1973.[10] B.A. Davey and H.A. Priestley. Introduction to Lattices and Order, Cam-

bridge Mathematical Textbooks. Cambridge University Press, Cambridge,1990.

[11] R. Brüggemann and E. Halfon. Ranking the environmental hazard ofthe chemicals spilled in the Sandoz Accident in November 1986. Sci.Tot. Environ., 97/98:827-837, 1990.

[12] R. Briiggemann and B. Münzer. A graph-theoretical tool for prioritysetting of chemicals. Chemosphere, 27:1729-1736, 1993.

[13] C.E.W. Steinberg, R. Briiggemann, A. Hartmann, W. Heller, M. Kirch-ner, D. Lienert, K. Muller, G. Pestlin, I. Scheunert, Seiler K.P, O.H.Spieser, and J. Klein. Ökotoxikologie: Bestandsaufnahme und Perspek-tiven für ein ökosystemares bewertungskonzept. GSF-Bericht, 40/93,1993.

[14] R. Briiggemann, B. Miinzer, and E. Halfon. An algebraic/graphical toolto compare ecosystems with respect to their pollution-the German River"Elbe" as an example-I: Hasse diagrams. Chemosphere, 28:863-872,1994.

[15] B. Miinzer, R. Briiggemann, and E. Halfon. An algebraic/graphical toolto compare ecosystems with respect to their pollution II: Comparativeregional analysis. Chemosphere, 28:873-879, 1994.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014

292 BRUGGEMANN AND HALFON

[16] K. Voigt and R. Brüggemann. Scoring approach for environmentaldatabases. In Learned Information Inc., Proceedings of the National On-line Meeting New York, pages 427-439. Learned Information, Medford,May 10-12 1994.

Received: January 17, 1995Accepted: April 23, 1995

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

17:

13 1

2 N

ovem

ber

2014