Biomarkers in perch (Perca fluviatilis) used in environmental monitoring of the Stockholm recipient
and background areas in the Baltic Sea
Tomas Hansson
A
Doctoral Thesis in Applied Environmental Science Department of Applied Environmental Science
Biomarkers in perch (Perca fluviatilis) used in environmental monitoring of the Stockholm recipient and background areas in the Baltic Sea
Doctoral Thesis in Applied Environmental Science 2008
Tomas Hansson
Department of Applied Environmental Science Stockholm University
SE-106 91 Stockholm Sweden
tomas.hansson@itm.su.se
Copyright © Tomas Hansson, 2008
Permission is hereby granted, free of charge, to any person to use, copy, reproduce, publish, and/or distribute pp. 1-33, provided that the source is stated:
Hansson, T. 2008. Biomarkers in perch (Perca fluviatilis) used in environmental monitoring of the Stockholm recipient and background areas in the Baltic Sea. PhD thesis, Department of Applied Environmental Science, Stockholm University, Stockholm, 33 pp.
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the text.
The copyright of the previously published papers and book chapters included in this thesis are the property of the respective publishers and are reproduced here with their permission.
ISBN 978-91-7155-711-7, pp. 1-33
Printed by Universitetsservice US-AB, Stockholm, Sweden, 2008
Cover photo: One-year old perch (Perca fluviatilis), Aquaria Vattenmuseum, Stockholm, Sweden. Photo by Lennart Balk, 2008.
2
Abstract
This thesis reports the results of biomarker measurements in three environmental monitoring projects. In the first project, which was part of the Swedish national environmental monitoring, biomarkers were measured annually in female perch (Perca fluviatilis) in two background areas in the Baltic Sea during 1988–2000, resulting in a unique 13-year series of measurements. The most important results were a strong decreasing temporal trend in the gonadosomatic index (GSI) and a strong increasing temporal trend in the hepatic ethoxyresorufin O-deethylase (EROD) activity in the Baltic Proper. In the second project, biomarkers and concentrations of classic pollutants were measured in female perch in the Stockholm recipient 1999–2001. This was the first time a large city was investigated as a point source of pollution, and the gradient was longer and included more stations than customary. Severe pollution conditions in central Stockholm were indicated by the poor health status of the perch: retarded growth, decreased frequency of sexually mature females, low GSI, disturbed visceral fat metabolism, increased hepatic EROD activity, decreased muscle acetylcholinesterase activity, increased frequency of hepatic DNA adducts, and a high concentration of biliary 1-pyrenol. Muscle ΣDDT and ΣPCB were measured as pollution indicators and were 10–28 respectively 12–35 times higher than the background levels in perch from the Baltic Proper. In the Stockholm archipelago two superimposed gradients were found. Whereas the response of several biomarkers consistently decreased with increasing distance from central Stockholm, the response of others first decreased from Stockholm to the middle archipelago and then increased to the open Baltic Sea. The latter biomarkers included the frequency of sexually mature females, GSI, hepatic EROD activity, and hepatic DNA adducts. In the third project, potential toxicity from munitions on the seafloor, at a dumpsite in the Stockholm archipelago, was analysed by the nanoinjection of sediment extracts into newly fertilised rainbow trout (Oncorhynchus mykiss) eggs, followed by the measurement of biomarkers in the developing larvae. No biological effects of the dumped munitions were found. The same stations in the Stockholm archipelago as in the second project were investigated as a positive control. The results of the three projects agreed well, which demonstrated the continuous pollution of the Baltic Sea and the severe pollution conditions and adverse biological effects in central Stockholm. Further investigations are urgently needed to identify which pollutants or other factors are causing the observed biological effects, both in the background areas in the Baltic Sea and in the Stockholm recipient.
Sammandrag
I denna avhandling beskrivs resultaten av biomarkörmätningar i tre miljöövervakningsprojekt.
I det första projektet, som är en del av den svenska nationella miljöövervakningen, gjordes årliga biomarkörmätningar i abborrhonor (Perca fluviatilis) i två bakgrundsområden i Östersjön 1988–2000, vilket givit upphov till en unik 13 år lång mätserie. De viktigaste resultaten var en starkt minskande tidstrend för gonadosomatiskt index (GSI) och en starkt ökande tidstrend för etoxyresorufin O-deetylasaktivitet (EROD) i levern hos abborrar i Egentliga Östersjön. I det andra projektet mättes biomarkörer och koncentrationer av klassiska miljögifter hos abborrhonor i Stockholmsrecipienten 1999–2001. Detta var första gången som en storstad undersöktes som punktkälla för miljögifter, och gradienten var längre och innehöll fler stationer än vad som är brukligt. Resultaten visade på en allvarlig föroreningssituation i centrala Stockholm med uppenbara sjukdomstecken hos abborrarna:
hämmad tillväxt, minskad frekvens av könsmogna honor, lågt GSI, störd bukfettsmetabolism, ökad EROD-aktivitet i levern, minskad acetylkolinesterasaktivitet i musklerna, ökad frekvens av DNA-addukter i levern samt hög koncentration av 1-pyrenol i gallan. ΣDDT och ΣPCB i musklerna mättes som föroreningsindikatorer och förekom i 10–28 respektive 12–35 gånger högre koncentration än bakgrundsnivåerna hos abborrar i Egentliga Östersjön. I Stockholms skärgård upptäcktes två överlagrade gradienter. I flera fall minskade biomarkörsvaret med ökat avstånd från Stockholm ända ut till öppna Östersjön, men i vissa fall minskade biomarkörsvaret först från Stockholm till mellanskärgården och ökade sedan igen ut till öppna Östersjön. Bland de senare biomarkörerna återfanns frekvensen av könsmogna honor, GSI, EROD-aktiviteten i levern och DNA-addukterna i levern. I det tredje projektet undersöktes potentiell giftverkan från krigsmateriel på havsbotten vid en dumpningsplats i Stockholms skärgård. Sedimentextrakt nanoinjicerades i nybefruktade regnbågsägg (Oncorhynchus mykiss) varefter biomarkörer mättes i de växande ynglen. Inga effekter av den dumpade krigsmaterielen upptäcktes och som positiv kontroll undersöktes samma stationer i Stockholms skärgård som i det andra projektet. Överensstämmelsen mellan de tre projekten var mycket god, och de påvisar tillsammans den kontinuerliga föroreningen av Östersjön och den höga föroreningsgraden, med sammanhängande allvarliga biologiska effekter, i centrala Stockholm. Det är mycket angeläget att ytterligare undersökningar görs för att identifiera vilka miljögifter eller andra faktorer som orsakar de observerade biologiska effekterna, både i bakgrundsområdena i Östersjön och i Stockholmsrecipienten.
4
Contents
Abstract ... 3
Sammandrag... 4
Contents... 5
List of papers... 7
Abbreviations ... 9
Contribution by the author ... 13
Introduction ... 15
The papers and the projects... 17
Perch as an indicator organism... 19
Statistics ... 21
Discussion of the main findings ... 25
Conclusions ... 29
References ... 31
Tack... 33
List of papers
This thesis is based on the following papers:
I. Long-term monitoring of the health status of female perch (Perca fluviatilis) in the Baltic Sea shows decreased gonad weight and increased hepatic EROD activity
Tomas Hansson, Eric Lindesjöö, Lars Förlin, Lennart Balk, Anders Bignert, and Åke Larsson
Aquatic Toxicology 79 (2006) 341–355
Reprinted with permission from Elsevier Limited
II. Basic physiological biomarkers in adult female perch (Perca fluviatilis) in a chronically polluted gradient in the Stockholm recipient (Sweden)
Maria Linderoth, Tomas Hansson, Birgitta Liewenborg, Henrik Sundberg, Erik Noaksson, Marsha Hanson, Yngve Zebühr, and Lennart Balk
Marine Pollution Bulletin 53 (2006) 437–450 Reprinted with permission from Elsevier Limited
III. Biochemical biomarkers in adult female perch (Perca fluviatilis) in a chronically polluted gradient in the Stockholm recipient (Sweden)
Tomas Hansson, Doris Schiedek, Kari K. Lehtonen, Pekka J. Vuorinen, Birgitta Liewenborg, Erik Noaksson, Ulla Tjärnlund, Marsha Hanson, and Lennart Balk Marine Pollution Bulletin 53 (2006) 451–468
Reprinted with permission from Elsevier Limited
IV. Buried waste in the seabed – acoustic imaging and bio-toxicity – results from the European SITAR project
Philippe Blondel and Andrea Caiti (editors) Springer Praxis, Berlin, Germany (2007)
i. Biotoxicity measurements: the nanoinjection technique
Tomas Hansson, Gun Åkerman, Ulla Tjärnlund, Kerstin Grunder, Yngve Zebühr, Henrik Sundberg, and Lennart Balk
Chapter 5, pp. 31–381
ii. The sea trial site in Möja Söderfjärd: biological sampling Per Jonsson and Tomas Hansson
Chapter 10, pp. 83–861
iii. Results of the biotoxicity measurements
Tomas Hansson, Gun Åkerman, Ulla Tjärnlund, Kerstin Grunder, Yngve Zebühr, Henrik Sundberg, and Lennart Balk
Chapter 15, pp. 127–1411
Reprinted with permission from Springer
1 Page numbers in the book, which differ from the page numbers in the thesis.
Abbreviations
AChE acetylcholinesterase
ACTC acetylthiocholine iodide
AHH aryl hydrocarbon hydroxylase AHR aryl hydrocarbon receptor ANCOVA analysis of covariance ANOVA analysis of variance
ATP adenosine triphosphate
B[a]P benzo[a]pyrene
B[a]PDE-dG-3´p 7R,8S,9S-trihydroxy, 10R-(N2-deoxyguanosyl-3´-phosphate)- 7,8,9,10-tetrahydro-benzo[a]pyrene
BEEP Biological Effects of Environmental Pollution in Marine Coastal Ecosystems
BSI brain somatic index
BSI n(s.weight) somatic weight normalised brain somatic index C control
13C carbon-13
CaCl2 calcium chloride
CDNB 1-chloro-2,4-dinitrobenzene CRS coordinate reference system
CYP1A cytochrome P450 enzyme 1A CYP2B cytochrome P450 enzyme 2B D deuterium 2D two-dimensional 3D three-dimensional
DDD dichlorodiphenyldichloroethane DDE dichlorodiphenyldichloroethylene DDMU 1-chloro-2,2-bis(4-chlorophenyl)ethene DDT dichlorodiphenyltrichloroethane DMF dimethylformamide
DNA deoxyribonucleic acid
DTNB 5,5´-dithio-bis-(2-nitrobenzoate) ECOD ethoxycoumarin O-deethylase
EDTA ethylenediaminetetraacetic acid
EROD ethoxyresorufin O-deethylase
ERODn(GSI) GSI normalised EROD activity
ERODnorm normalised EROD activity EROD-SI EROD somatic index
EROD-SIn(GSI,date) GSI and sampling date normalised EROD somatic index
EU European Union
GIS geographical information system
GSH reduced glutathione
GSI gonadosomatic index
GSIn(date) sampling date normalised gonadosomatic index
GSIn(date,age) sampling date and age normalised gonadosomatic index
GST glutathione S-transferase GST-SI GST somatic index
GST-SIn(date) sampling date normalised GST somatic index HCB hexachlorobenzene
HCl hydrochloric acid
HCH hexachlorocyclohexane HKSI head kidney somatic index
HPLC high performance liquid chromatography
H3PO4 phosphoric acid
H2SO4 sulphuric acid
IA inner archipelago
IUPAC International Union of Pure and Applied Chemistry
KOH potassium hydroxide
LiCl lithium chloride
LiOH lithium hydroxide
LM Lake Mälaren
LSI liver somatic index
LSIn(date) sampling date normalised liver somatic index
lw lipid weight
MA middle archipelago
MC macrophage center
MgCl2 magnesium chloride
MN micrococcal endonuclease
NaCl sodium chloride
NADPH reduced nicotinamide adenine dinucleotide phosphate NaH2PO4 sodium dihydrogen phosphate
Na2HPO4 disodium hydrogen phosphate
Na2SO4 sodium sulphate
(NH4)2SO4 ammonium sulphate
o ortho
OA outer archipelago
p para
32P phosphorus-32
PAH polycyclic aromatic hydrocarbon
PBB polybrominated biphenyl
PCB polychlorinated biphenyl
pH base-10 logarithm of the reciprocal of the hydrogen ion concentration PLSD protected least significant difference
PMSF phenylmethylsulfonyl fluoride
PVC polyvinyl chloride
RNA ribonucleic acid
S9 supernatant obtained by centrifugation at 9,000g S10 supernatant obtained by centrifugation at 10,000g SCI somatic condition index
SCIn(age) age normalised somatic condition index
SDS sodium dodecyl sulphate
SEPA Swedish Environmental Protection Agency
SG somatic growth
SGn(length) length normalised somatic growth
SIM sexually immature
SITAR Seafloor Imaging and Toxicity: Assessment of Risk Caused by Buried Waste
10
SM sexually mature
SMHI Swedish Meteorological and Hydrological Institute
SPD spleen phosphodiesterase
SSI spleen somatic index
TCDD 2,3,7,8-tetrachlorodibenzo-p-dioxin TDI tolerable daily intake
TEQ TCDD equivalents
TLC thin-layer chromatography
TOC total organic carbon
UV ultra-violet VFSI visceral fat somatic index WGS World geodetic system
ww wet weight
ZnCl2 zinc chloride
Contribution by the author
I have participated in the samplings and analyses of the investigations, and have performed the statistical analysis and evaluation, except for the power analysis in Paper I (Table 5). I have also carried out most of the literature research on which the discussion part of each paper is based, and have written the major part of the four papers.
Introduction
Modern society expends much effort on environmental monitoring to gain the necessary knowledge for safe management of the environment and environmental risks. As a result, a considerable amount of new monitoring data is produced every year. Still, the link between measurements in the field and measures taken by society is often problematic. Part of the problem is scientific. It is difficult to produce useful knowledge from data, and even to produce useful data. Another part of the problem is educational. The full importance of environmental issues is still not understood and acknowledged by the general public and decision makers. Erroneous or oversimplified explanations of environmental changes often prevail, and as a consequence, measures are often not taken, even when alarming changes in the environment are detected. Yet another part of the problem is inherent in regular monitoring. Slow, gradual changes in the environment tend to be neglected just because the change from year to year is small, although the accumulated change is large. Finally, society often fails to identify those responsible for taking measures, even when knowledge is available.
The work presented in this thesis differs somewhat from the traditional approach with formulation of a hypothesis, testing of the hypothesis, and presentation of the result. Instead, already available scientific methods were used in ongoing environmental monitoring projects to determine if there were any biological effects, or pollution, in the investigated areas. These projects were team projects, where my main contribution was to evaluate data and write papers. I was dependent on previous knowledge, the state of the art, other people’s measurements, available computer software, etc., but I was also compelled to understand the whole process from the planning of the project to the delivery of the results. The primary focus of the work was to set a good example of environmental monitoring. Applied environmental science was adopted as an academic discipline at Stockholm University in 2005, and this thesis exemplifies the versatility of contemporary applied environmental science.
The papers and the projects
Paper I is an evaluation of data generated within a fish environmental monitoring programme, which is a part of the National Swedish Environmental Monitoring Programme2 pursued by the Swedish Environmental Protection Agency3. More precisely, data from biochemical, physiological, and histopathological investigations of perch (Perca fluviatilis) in two background areas in the Baltic Sea during 1988–2000 were evaluated.
Paper II presents results obtained in an EU project4 intending to demonstrate that biomarkers are useful in marine costal ecosystems. Here we chose to investigate biological effects and concentrations of classic pollutants in perch in the Stockholm recipient 2000–2001. The results were combined with those of a similar, but less extensive, investigation of perch 1999, requested by the Environment and Health Administration of the City of Stockholm5.
Paper III presents additional results from the same investigations as in Paper II.
Paper IV consists of three book chapters describing an investigation that was part of another EU project6, where our primary task was the analysis of potential toxicity from munitions on the seafloor at a dumpsite in the Stockholm archipelago. This was done by nanoinjection of sediment extracts into newly fertilised rainbow trout (Oncorhynchus mykiss) eggs followed by the measurement of biological effects in the developing embryos and larvae. As a positive control, we investigated the same stations in the Stockholm archipelago as in the other EU project (Papers II and III). The sediments used in this investigation (Paper IV) were formed during 2000–2002.
2 Den svenska nationella miljöövervakningen
3 Naturvårdsverket
4 Biological Effects of Environmental Pollution in Marine Coastal Ecosystems (BEEP), EC contract no. EVK3- CT2000-00025
5 Miljöförvaltningen i Stockholm
6 Seafloor Imaging and Toxicity: Assessment of Risk Caused by Buried Waste (SITAR), EC contract no. EVK3- CT-2001-00047
Perch as an indicator organism
Perch has been used as an indicator organism in numerous investigations of biological effects of environmental pollution, but perch have certain properties that are seldom discussed explicitly.
Firstly, biomarker responses often differ naturally between males and females. The sexes should therefore be separated, or an investigation should be confined to either sex. One reason to choose females, as in the projects presented in this thesis, is that males are difficult to catch. Typically, net fishing yields 90 % females. To the best of our knowledge, the reason for this is unknown.
Secondly, perch is a relatively sedentary species, which is absolutely necessary for the results of the biomarker measurements to reflect the conditions in the area where the perch were caught. That perch is relatively sedentary has been demonstrated explicitly in three investigations (Craig, 1974; Kipling and Le Cren, 1984; Böhling and Lehtonen, 1985), but there is also a large number of studies of perch in pollution gradients (including Papers II and III), where a graded response was obtained, which would be impossible if the perch migrated widely. The compilation of all biomarker studies of perch with a graded response in pollution gradients, to further demonstrate its sedentary nature, is an issue for future work.
Thirdly, it has been demonstrated in laboratory experiments that the perch maintains homeostasis up to a salinity of the surrounding water of at least 10 (Lutz, 1972). A common objection to the use of perch in a salinity gradient is that the differences in salinity would have a confounding effect on the biomarkers. Papers II and III indicate that any such confounding effect is, in practice, negligible, at least for a salinity gradient from 0 to 5.4.
Statistics
During my time as a PhD student I have had the privilege of continually learning more about statistical methods. Today, I would use somewhat more accurate and stringent methods, which conform better to conventional procedures, than certain methods used in Papers I–III.
Most probably, however, the conclusions drawn from the data in this thesis would be just the same. Instead of normalising the data before the statistical analysis, I would include the confounding variable in an analysis of variance (ANOVA) or regression model. In the investigations included in this thesis, only reference material was used to define the relationship between the confounding variable and the dependent variable, whereas, in models including the confounding variable, this relationship will be based on all available data. Also, this relationship will be assumed to be the same in all investigated groups, provided that there are no interactions between the confounding variable and the other independent variable(s).
Moreover, methods are available for evaluating the relative importance of independent variables and for model selection. In more sophisticated models, like those I would prefer today, the assumptions of normality and homoscedasticity would be tested for the residuals, rather than the original values of the included variables. Instead of the Kolmogorov-Smirnov normality test (Kolmogorov, 1933; Smirnov, 1939), which has a relatively low power, I would use the more powerful Shapiro-Wilk normality test (Shapiro and Wilk, 1965), and instead of the Breusch-Pagan/Cook-Weisberg test for heteroscedasticity (Breusch and Pagan, 1979; Cook and Weisberg, 1983), I would use the more appropriate and powerful Bartlett’s test (Bartlett, 1937), which is a generalisation of the F-test to more than two groups. Instead of the Dunn-Šidák method for adjustment of the α-level for multiple hypothesis testing (Sokal and Rohlf, 1995), I would use Holm’s sequentially rejective Bonferroni method (Holm, 1979). The Dunn-Šidák method overprotects the null hypothesis when the number of comparisons exceeds about five, while Holm’s method is less conservative (i.e. more powerful) and still gives full protection against an inflated experimentwise Type-I error rate at a given α-level (Ludbrook, 1991, 1998; Shaffer, 1995).
The statement that “there is always some subjectivity in the decision of when and how the Dunn-Šidák method should be used” found in Papers I–III needs an explanation. The purpose of p-value or α-level adjustment for multiple hypothesis testing (multiple comparisons) is to
protect against an inflated experimentwise Type-I error rate. Ludbrook (1991) defined the comparisonwise and experimentwise Type-I error rates as follows:
Number of wrong inferences Comparisonwise Type-I error rate = –––––––––––––––––––––––––––––
Total number of comparisons made
Number of experiments with at least one wrong inference
Experimentwise Type-I error rate = –––––––––––––––––––––––––––––––––
Total number of experiments performed
An example will demonstrate the relationship between comparisonwise and experimentwise Type-I error rate. If six samples (each with sample size = n) are drawn from the same population in one experiment, the null hypothesis that the samples come from the same population is indeed true. Ideally, none of the samples should differ significantly from any of the other samples, but by chance there will be significant differences at a rate called the comparisonwise Type-I error rate. There is a direct relationship between the comparisonwise Type-I error rate and how large a test statistic must be to be considered significant. The larger the test statistic must be, the smaller the comparisonwise Type-I error rate will be. The investigator sets both these parameters simultaneously by the choice of α-level. If all possible comparisons between the six samples are made there will be 15 comparisons, and if the α- level is set to 0.05, corresponding to a comparisonwise Type-I error rate of 5 %, the experimentwise Type-I error rate will be 54 %. This means that, if additional similar experiments are performed, on average every two experiments will contain at least one wrong inference. Alternatively, if we want the experimentwise Type-I error rate to be 5 %, then we have to adjust the α-level, and if the Dunn-Šidák method is used, we will obtain an α´-level of 0.0034. This means that we only accept a comparisonwise Type-I error rate of 0.34 %. So far, everything is just statistical theory. The subjectivity of when and how the Dunn-Šidák method should be used simply refers to the question of how to define an experiment in the sense used here.
The combined use of Fisher’s protected least significant difference (Fisher’s PLSD) and the Dunn-Šidák method needs a comment. Fisher’s PLSD is, in fact, another procedure to protect against an inflated experimentwise Type-I error rate (Hochberg and Tamhane, 1987). It is a simple two-step procedure in which the null hypothesis is tested at the first step by an α-level
22
F-test (ANOVA). If the outcome of the F-test is not significant, then the procedure terminates without making detailed inferences on pairwise differences. On the other hand, if the outcome of the F-test is significant, then pairwise differences are tested by a sequence of α-level t-tests, and in the software StatView 5.0 (SAS Institute Inc., Cary, NC, USA) the respective t statistics for this are based on the error mean square of the previous F-test. As a consequence, Fisher’s PLSD protects against inflated experimentwise Type-I error rate only when the null hypothesis is true, and the p-values of the post hoc test are not adjusted in any way. Therefore, Fisher’s PLSD can be combined with the Dunn-Šidák method, and although this combination is a bit unusual and overprotective, it will give sufficient protection against an inflated experimentwise Type-I error rate.
Paper IV was written without any statistical hypothesis testing. This is because the samples were originally collected to generate a 3D map, illustrating the variation of variables in the environment. This map was to be included in a computer program intended as an evaluation tool for decision makers. We decided that the best way to present the results was to transform the 3D view into a 2D view. Paper IV is a good example showing that graphical presentation may be sufficient to make the overall pattern clear, and that inferences may well be made just by comparison of the means and confidence intervals.
Discussion of the main findings
The 13-year series of biomarker investigations in perch in Paper I is unique. The most important results were a strong decreasing temporal trend in the gonadosomatic index (GSI) and a strong increasing temporal trend in the specific hepatic ethoxyresorufin O-deethylase (EROD) activity in the Baltic Proper. There is probably a mechanistic relationship between these two variables since estradiol, produced by the ovary, has an inhibiting effect on the EROD activity. The statistical analysis showed, however, that the GSI as a confounding variable explained only a part of the increasing trend in the EROD activity. The high pollution load in the Baltic Sea is strongly suspected to cause the remaining part of this trend. There were also indications of a decreasing temporal trend in the frequency of sexually mature (SM) females.
The biomarker investigation of perch described in Papers II and III is the first fish biomarker study performed in the Stockholm recipient. As far as we know, this is also the first time a large city, without heavy industry, has been investigated as a point source of pollution, and the gradient was longer (84+46 km) and included more stations (10) than customary. The results indicated severe pollution conditions in central Stockholm, demonstrated by the poor health status of the perch: retarded growth, decreased frequency of SM females, low GSI, disturbed visceral fat metabolism, increased specific hepatic EROD activity, increased hepatic EROD somatic index, decreased muscle acetylcholinesterase (AChE) activity, increased amount of hepatic DNA adducts, and a high concentration of biliary 1-pyrenol. Muscle ΣDDT and ΣPCB were measured as pollution indicators and were 10–28 respectively 12–35 times higher than the background levels in perch from the Baltic Proper. A bit surprising was to find two superimposed gradients in the Stockholm archipelago. Whereas the response of several biomarkers consistently decreased with increasing distance from central Stockholm, the response of certain biomarkers began increasing again from the middle archipelago to the open Baltic Sea (Table 1). Accordingly, the lowest response for the latter biomarkers was found in the middle archipelago.
Table 1
Biomarkers with a graded response in the Stockholm recipient
Biomarker Responsea from
Stockholm to the middle archipelago
Responsea from the middle archipelago to the Baltic Sea
Somatic growth (SG) Decreasing Decreasing
Brain somatic index (BSI) Decreasing Decreasing
Frequency of sexually mature (SM) females Decreasing Increasing
Gonadosomatic index (GSI) Decreasing Increasing
Visceral fat somatic index (VFSI) Decreasing Decreasing Specific hepatic EROD activity Decreasing Increasing Hepatic EROD somatic index (EROD-SI) Decreasing Increasing
Hepatic DNA adducts Decreasing Increasing
Biliary 1-pyrenol Decreasing Decreasing
a The response refers to the biological effect and not to the actual biomarker values. A decreasing response may be associated with increasing values for certain biomarkers (e.g. GSI) and decreasing values for other biomarkers (e.g. hepatic DNA adducts) and vice versa.
An important observation is that three of the biomarkers with increasing response from the middle archipelago to the open Baltic Sea also showed temporal trends in the background area in the Baltic Proper (Paper I). These biomarkers were frequency of SM females, GSI, and specific EROD activity (Table 1). During the work with Paper I, we had to consider other possible causes of the observed biological effects than just environmental pollutants, for example, temperature and algal toxins. Increased temperature as a possible cause of the temporal trends in the GSI and the specific EROD activity (Paper I) may, however, be refuted by the observed increasing response from the middle archipelago to the open Baltic Sea (Papers II and III), since the temperature was relatively constant during the investigations in the Stockholm archipelago.
The use of the Stockholm archipelago gradient as a positive control in Paper IV yielded valuable information that confirmed the results in Papers II and III, although the investigation found no biological effects of the dumped munitions.
The EROD activity gradient was demonstrated in three ways with concordant results: in perch collected in the field (Paper III), in sea trout (Salmo trutta) exposed to sediments (Paper III),
26
and in rainbow trout larvae exposed to sediment extracts nanoinjected in the newly fertilised eggs (Paper IV). Moreover, the EROD activity gradient was paralleled by a polycyclic aromatic hydrocarbon (PAH) gradient in the sediments, and the EROD activity in the rainbow trout larvae was induced mainly by the PAH containing fraction obtained from the sediment extract (Paper IV). Using more than one method to demonstrate a pollution gradient is rather unusual. The good agreement between the three methods used here confirms the validity of these methods, and shows that any of the methods may be used alone in future investigations.
Muscle AChE was inhibited in perch in central Stockholm (Paper III), but the investigations of sea trout (Paper III) and rainbow trout larvae (Paper IV) showed that no AChE inhibition was induced by the sediments. This is a plausible result, since many AChE inhibitors are readily soluble in water.
The statement that “data on healthy perch showed that individuals >19 cm were sexually mature with only occasional (0–5%) occurrence of individuals not developing eggs certain years” in Paper II needs a comment. In fact, occurrences of individuals >19 cm not developing eggs were found only in recent investigations (Noaksson et al., 2001, 2004, 2005), whereas in older investigations the frequency of such individuals was zero (Le Cren, 1958;
Alm, 1959). Hence, it may well be that all healthy female perch >19 cm should develop eggs.
In Papers I and II we report definitely abnormal frequencies of SM females >19 cm. Similar reports are found in other recent investigations of perch in the Baltic Sea (Sandström et al., 1995; Sandström and Neuman, 2003; Roots et al., 2004). It is suggested that the frequency of SM females should be more widely acknowledged as a biomarker today.
This thesis contains several examples of methods for obtaining a homogeneous material despite confounding variables. In the first place, homogeneity should be obtained by the proper selection of specimens. We confined our investigations to adult females within a specified length interval and, when necessary, in our statistical analysis we separated the females with maturing eggs from those without maturing eggs. Remaining inhomogeneities were adjusted for by normalisation. Some confounding variables to be aware of in the statistical analysis of biomarkers in perch are listed in Table 2. Somatic growth (SG) is confounded by size, since the yearly weight increments are greater for larger individuals.
Liver somatic index (LSI) and GSI both have an annual cycle. GSI is also confounded by age and size, since older and bigger individuals allocate more resources to gonad growth than
younger and smaller individuals. Age and size also confound the somatic condition index (SCI), since the shape of the perch changes as it grows. Brain somatic index (BSI) is confounded by size, since the head grows proportionally slower than the rest of the body. The relationship between specific EROD activity and GSI is still just empirical, although the EROD inhibiting effect of estradiol, produced by the ovary, may provide a mechanistic explanation during at least a part of the reproduction cycle. It is strongly suspected that in perch, as in other species, specific EROD activity has a complete annual cycle. In the juvenile sea trout, the AChE activity was confounded by weight, but the reason for this is unknown.
Table 2
Confounding variables in adult female perch
Variable Confounding variable(s)
Somatic growth (SG) Length Liver somatic index (LSI) Time
Gonadosomatic index (GSI) Time, age, length Somatic condition index (SCI) Age, length
Brain somatic index (BSI) Somatic weight, age, length Specific EROD activity GSI
28
Conclusions
The investigations presented in this thesis give a picture of the Baltic Sea as a chronically polluted inland sea, with pollution from both the western and eastern countries. Pollution effects are observed even in background areas, indicating that pristine areas with unaffected specimens, that might be suitable as controls, may no longer exist. Contrary to a prevailing notion of “the clean waters of Stockholm”, we demonstrated that these waters are highly polluted, producing adverse biological effects in a sedentary fish species. Further investigations are urgently needed to identify which pollutants or other factors are causing the observed biological effects, both in the background areas in the Baltic Sea and in the Stockholm recipient. Clues can be provided by toxicity mechanisms, for example whether the decreased GSI is caused by delayed maturation or inhibited growth. The investigations presented in this thesis have also contributed with several examples of the absolute necessity to consider, and adjust for, confounding variables in order to make reliable inferences from data.
References
Alm, G. 1959. Connection between maturity, size, and age in fishes. Institute of Freshwater Research, Drottningholm, Fishery Board of Sweden. Report No. 40, pp. 5–145.
Bartlett, M.S. 1937. Properties of sufficiency and statistical tests. Proceedings of the Royal Society, Series A 160, 268–282.
Böhling, P., Lehtonen, H., 1985. Effect of environmental factors on migrations of perch (Perca fluviatilis L.) tagged in the coastal waters of Finland. Finnish Fisheries Research 5, 31–40.
Breusch, T., Pagan, A. 1979. A simple test for heteroscedasticity and random coefficient variation. Econometrica 47, 1287–1294.
Cook, R.D., Weisberg, S. 1983. Diagnostics for heteroscedasticity in regression. Biometrika 70, 1–10.
Craig, J.F., 1974. Population dynamics of perch, Perca fluviatilis L. in Slapton Ley, Devon. I.
Trapping behaviour, reproduction, migration, population estimates, mortality and food.
Freshwater Biology 4, 417–431.
Hochberg, Y., Tamhane, A.C. 1987. Multiple comparison procedures. John Wiley & Sons, New York, USA, p. 3.
Holm, S. 1979. A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics 6, 65–70.
Kipling, C., Le Cren, E.D., 1984. Mark-recapture experiments on fish in Windermere, 1943–
1982. Journal of Fish Biology 24, 395–414.
Kolmogorov, A.N. 1933. Sulla determinazione empirica di una legge di distribuzione.
Giornale dell’ Istituto Italiano degli Attuari 4, 83–91. (In Italian)
Le Cren, E.D., 1958. Observations on the growth of perch (Perca fluviatilis L.) over twenty- two years with special reference to the effects of temperature and changes in population density. Journal of Animal Ecology 27, 287–334.
Lutz, P.L., 1972. Ionic and body compartment responses to increasing salinity in the perch Perca fluviatilis. Comparative Biochemistry and Physiology 42A, 711–717.
Ludbrook, J. 1991. On making multiple comparisons in clinical and experimental pharmacology and physiology. Clinical and Experimental Pharmacology and Physiology 18, 379–392.
Ludbrook, J. 1998. Multiple comparison procedures update. Clinical and Experimental Pharmacology and Physiology 25, 1032–1037.
Noaksson, E., Tjärnlund, U., Bosveld, A.T.C., Balk, L., 2001. Evidence for endocrine disruption in perch (Perca fluviatilis) in a remote Swedish lake in the vicinity of a public refuse dump. Toxicology and Applied Pharmacology 174, 160–176.
Noaksson, E., Gustavsson, B., Linderoth, M., Zebühr, Y., Broman, D., Balk, B., 2004. Gonad development and plasma steroid profiles by HRGC/HRMS during one reproductive cycle in reference and leachate-exposed female perch (Perca fluviatilis). Toxicology and Applied Pharmacology 195, 247–261.
Noaksson, E., Linderoth, M., Gustavsson, B., Zebühr, Y., Balk, B., 2005. Reproductive status in female perch (Perca fluviatilis) outside a sewage treatment plant processing leachate from a refuse dump. Science of the Total Environment 340, 97–112.
Roots, O., Järv, L., Simm, M., 2004. DDT and PCB concentrations dependency on the biology and domicile of fish: an example of perch (Perca fluviatilis L.) in Estonian coastal sea. Fresenius Environmental Bulletin 13, 620–625.
Sandström, O., Neuman, E., 2003. Long-term development in a Baltic fish community exposed to bleached pulp mill effluent. Aquatic Ecology 37, 267–276.
Sandström, O., Neuman, E., Thoresson, G., 1995. Effects of temperature on life history variables in perch. Journal of Fish Biology 47, 652–670.
Shaffer, J.P. 1995. Multiple hypothesis testing. Annual Review of Psychology 46, 561–584.
Shapiro, S.S., Wilk, M.B. 1965. An analysis of variance test for normality (complete samples). Biometrika 52, 591–611.
Smirnov, N.V. 1939. Estimate of deviation between empirical distribution functions in two independent samples. Bulletin of the Moscow University 2, 3–16. (In Russian)
Sokal, R.R., Rohlf, F.J., 1995. Biometry, 3rd ed. W.H. Freeman & Co, New York, USA, p. 239.
32
Tack
Jag vill framför allt tacka de personer som gjort denna avhandling möjlig. Eric Lindesjöö, min första chef, tack vare dig fick jag in en fot på ITM. Lennart Balk, min nuvarande chef och handledare, utan din erfarenhet och fingertoppskänsla som forskare skulle det ha varit omöjligt att få ut så mycket information ur våra undersökningar. Ulla Tjärnlund, Gun Åkerman, Birgitta Liewenborg, Erik Noaksson, Maria Linderoth, Henrik Sundberg, Halldóra Skarpheðinsdottir, Marsha Hanson, Kerstin Grunder och Yngve Zebühr, mina kära medarbetare i Biologiska effektgruppen, ni har alla bidragit både till avhandlingen och till arbetsglädjen. Jan-Olov Persson och Anders Bignert, av er har jag fått tålmodig statistisk vägledning, utan vilken det hade varit omöjligt att på bästa sätt tolka våra resultat.
Naturligtvis vill jag även tacka alla på ITMx och då inte minst alla mina rumskamrater genom åren, ingen nämnd ingen glömd.
