Validation of the use of blood samples to assess tissue concentrations of organochlorines in glaucous gulls, Larus hyperboreus

Pergamon Chemosphere, Vol. 37, No. 13, pp. 2627-2643, 1998 © 1998 Elsevier Science Ltd. All rights reserved

0045-6535/98/$ - see front matter PII: S01M5-6535(98)00162-3

V A L I D A T I O N O F T H E U S E O F B L O O D S A M P L E S T O A S S E S S

T I S S U E C O N C E N T R A T I O N S O F O R G A N O C H L O R I N E S I N

G L A U C O U S G U L L S , Larus hyperboreus

Espen O. Henriksen 1., Geir W. Gabrielsen ~ and Janneche Utne Skaare 2~

J Norwegian Polar Institute, N-9005 TROMSO, Norway 2 Norwegian College of Veterinary Medicine, PO Box 8146 Dep., N-0033 OSLO, Norway

3 National Veterinary Institute, PO Box 8156, N-0033 OSLO, Norway

(Received in Germany 18 February 1998; accepted 21 April 1998)

ABSTRACT

Fitteen adult glaucous gulls, Larus hyperboreus, were captured near Ny-Alesund, Svalbard. The birds were kept

in captivity for 24 - 41 days and fed a diet of polar cod, Boreogadus saida. A range of organochlorines (OCs) were

quantified in blood, brain, liver, and subcutaneous fat tissue. For more than 80 % of the quantified OCs, r 2 values

>0.75 were found for the blood-liver concentration correlations. Repeated sampling revealed intra-individual

temporal variability in blood OC concentrations. Much of the temporal variability in OC blood concentrations was

associated with changes in nutritional condition. ©1998 Elsevier Science Ltd. All rights reserved

-~ INTRODUCTION

In recent years, disturbingly high levels of anthropogenic persistent organochlorines (OCs) have been found to

bioaccumulate in arctic marine top predators [ 1,2]. The monitoring of OC levels and the studies of their possible

effects depends on the analysis of tissues or body fluids. As noted by Friend and co-workers [3], the use of blood

samples has several advantages compared to sacrificing animals. In endangered species, sacrificing even a few

individuals may be unacceptable. Non-destructive sampling also allows long-term studies with repeated sampling

and monitoring of potential impact of OCs on survival and reproduction.

The use of blood as sample matrix for OC analysis depends on the assumption that the concentrations of OCs in

blood are related to tissue concentrations. This assumption is supported by studies on a number of species

including man (e.g. [4]), polar bear, Ursus maritimus, [1], and various birds [3,5,6]. In the case of the

polychlorinated biphenyls (PCBs), regression analysis has demonstrated that the tissue:blood partition coefficient

can be expressed as a function of molecular structure [7].

The present study focuses on the glaucous gull, Larus hyperboreus, an opportunistic arctic predatory gull which

can accumulate high tissue concentrations of OCs, in particular PCBs [2,8]. All previous studies on OCs in the

' to whom correspondence should be addressed

2627

2628

glaucous gull have been based on analysis of tissue samples or eggs. The purpose of the present study was to

validate the use of blood samples to assess OC levels in the glaucous gull. Our primary objective was to determine

the relationship between concentrations in blood and the corresponding concentrations in selected tissues for a

range of OC compounds. We also examined the temporal variation in blood OC levels under the standardised

conditions of the experiment. This was done because any inherent intra-individual temporal variation would

underlie the additional variation expected under less standardised sampling from free-living birds. In effect studies

on free-living birds, it is desirable to know whether an elevated blood concentration of an OC compound at a

single point in time is related to a chronic load.

MATERIALS AND METHODS

Between May 31 and June 10 1995, 22 adult glaucous gulls were caught in the vicinity of Ny-Alesund, Svalbard

(79°N, 12°E) with a cannon-net, using seal blubber as bait. The birds were kept in captivity at the Norwegian

Polar Institute's research station in Ny-,~lesund. Seven of the birds had to be rele.ased after 1-3 days because they

did not acclimatize to captivity. The remaining 15 individuals (11 females and 4 males) were placed separately in

plastic covered wire mesh cages (0.75m x 0.75m x 0.60m) on an outdoor platform, covered by a plastic roof. The

walls of the platform allowed light to pass through between each board. Thus, the birds were exposed to the

ambient light and air temperatures, while they were protected against direct sunlight and precipitation. The birds

were weighed, using a Mettler PE 16 (±0.1 g) balance, and examined every morning.

The birds were fed polar cod (Boreogadus saida), which is an important species in their natural diet [9,10]. Polar

cod were collected in the Kongsfjorden area in June 1995, and kept frozen at -10°C until thawed. OC

concentrations were not determined in the fish, However, a mean EPCB-concentration of 33.3 ng/g lipid has been

found in whole polar cod from the northern Barents Sea [11]. Apart from 12 hours before each blood sampling,

food was available ad libitum. The birds had free access to water. All birds ate voluntarily during the entire

experiment.

Blood samples were taken every fifth day from the brachial vein on the inside of each wing, altemating between

the two wings. The terminal blood samples were analysed for OCs as described below. In addition, selected OCs

were also quantified in one of the two first blood samples from each bird. The birds were sensitive to handling.

To reduce stress, handling was limited to once a day. Refilling of food and water was done twice a day. Individual

birds were kept in captivity for between 24 and 41 days. Nine birds lost <15% of their initial body mass at the end

of the experiment. These birds were fasted for 2 days before they were killed. Initial body masses were 1240-

1520 g for females and 1620-1860 g for males. Body masses at the end of the experiment were 980-1260 g for

females and 1350-1540 g for males. The birds were killed by decapitation or cervical dislocation.

2629

OCs were quantified in blood and tissue samples using the method described by Brevik [12] with modifications by

Bernhoft et aL [1]. Briefly, samples were homogenized and extracted twice with cyclohexane and acetone using

an ultrasonic homogeniser. Extractable fat was determined gravimetrically. Cleanup was done with sulphuric

acid. The samples were injected on a gas cromatograph with a SB-5 capillary column and an electron capture

detector. The following OCs were determined: ct-, 13-, and T-HCH, oxychlordane, trans-chlordane, cis-chlordane,

trans-nonachlor, p,p '-DDE, o,p '-DDD, p,p '-DDD, p,p '-DDT, HCB, Mirex, and 33 PCB-congeners (IUPAC nos.):

28, 31, 47, 52 (not analysed in brain samples), 56, 66, 74, 87, 99, 101,105, 110, 114, 118, 128, 136, 137, 138, 141,

149, 151,153, 156, 157, 170, 180, 187, 189, 194, 196, 199, 206 and 209. In each batch of 24 samples, recoveries

of added standard mixtures to clean samples were calculated. Percentage recoveries were in the in the range of 70-

130%. Quantification limits for individual OCs were 0.3-8 ng/g in blood lipids, 0.9-17 ng/g in liver tipids, 1.8-31

ng/g in brain lipids and 0.3-58 ng/g in lipid from subcutaneous fat tissue. The laboratory's analytical quality was

approved in several intercalibration tests, including the four steps of the ICES/IOC/OSPARCOM intercomparison

exercise on the analysis of PCBs in marine media.

Statistical methods

JMP and SAS statistical software (SAS Institute, Cary, NC, USA) were used in the statistical analysis. P-values

less than 0.05 were considered statistically significant. In general, the OC concentration data departed significantly

from the normal-distribution, according to the Shapiro-Wilk test. After loglo transformation, the Shapiro-Wilk test

indicated that only 6 out of 152 variables (38 compounds x 4 matrices) departed significantly from normal

distributions. Based on this, underlying normally distributed populations were assumed for the logl0-transformed

OC concentration data. All statistics on OC data were based on logl0-transformed values. For comparison of

concentration ratios (between blood and tissues or between different blood samples), log~0-transformations were

performed on the ratios. The means and confidence-intervals of the ratios were then back-transformed to

geometric means with 95% confidence intervals. Sexes were pooled because OC concentrations did not differ

significantly between the sexes. The data for percentage extracted lipid from subcutaneous fat was consistent with

a normal distribution and was not transformed. Correlations were explored using squared Pearson's product-

moment correlations (r 2) and associated p-values, assuming independent observations from normally distributed

populations. Correlations between blood and tissues were not calculated if the OC was below the quantification

limit in more than half of the tissue or blood samples. In the remaining cases, half the quantification limits were

used for OCs below the quantification limits. OC-concentrations are presented as median and ranges. To facilitate

comparison with other studies, arithmetic means are provided.

To investigate variation in blood concentrations, multivariable linear regression models were fitted to the data for

the 13 individual OC compounds that were quantified twice (SAS General Linear Model). Models were created to

describe the relationship between OC concentrations in the I st and the 2 "d (terminal) set of blood samples. In the

models, the concentration of an OC in the 2 nd blood sample (Y) was assumed to depend on X1, the corresponding

2630

concentration in the 1 st blood sample, and on X2, the change in body mass between the 1 st and the 2 "d sampling.

Body mass change was included in the models because nutritional condition is important for tissue OC

concentrations [13]. The period between the 1 st and the 2 "d blood sample varied between birds (22-35 days).

The model fit was: Y = ,80 + flIXl + ~X2 + E,

Y = cone. of specific OC in the 2 nd blood sample X1 = corresponding OC cone. in the 1 st blood sample X2 = change in body mass between 1 st and 2 "d blood sampling fl0, ,81, ~ = regression coefficients E = error component.

These models partitioned the total variance of Y based on the proportion of the total sum of squares that could be

accounted for by Xt and X2 (Type I sum of squares) (Fig. 4). Significance of the regression coefficients for Xl and

X2 (i.e., fll and ,82) was judged from t-tests (H0: regression coefficient = 0).

RESULTS

Prevalence and levels of OCs in blood at the end of the experiment

The following OCs were above the quantification limit in all the analysed blood samples: chlordanes (sum of oxy-

and cis-chlordane, and trans-nonachlor), HCB, HCHs (sum of cx- and [3-HCH), DDTs (sum ofp,p '-DDT and p,p "-

DDE), Mirex and PCBs (29 individual congeners with 3-10 chloro substituents). PCBs dominated the OCs present

(Table 1). Median ZPCB concentration in blood was 83 lag/g extractable lipid. The PCB pattern was dominated

by PCBs no. 153, 138, 180 and 118, which accounted for >70 % of the PCBs (Fig. 1). DDT related compounds,

mainly p,p'-DDE, were the second most prevalent OCs. The median EDDT concentration in blood extractable

lipids was 26 ~tg/g. Median concentrations of chlordanes, HCB, Mirex and HCHs were 4.9, 2.5, 0.8 and 0.2 ~tg/g

lipid, respectively. Oxychlordane constituted about 90% of sum chlordanes. 13-HCH constituted about 97% of

sum HCHs. The chlordane related compound trans-chlordane, the ~'-isomer of HCH, the DDT compounds p,p'-

and o,p '-DDD, and o,p '-DDT and the PCBs no. 31, 56, 86 and 137 were all below the quantification limits of the

analytical method in all blood samples.

Table 1. Median, mean and range of organochlorine (OC) concentrations and percentage extractable lipids in blood from adult glaucous gulls, Larus hyperboreus from Svalbard. The birds were in captivity for 24- 41 days and fed a diet of polar cod, Boreogadus saida. N=15. Concentrations are in ng/g.

Lipid (%) HCB EHCH EChlordane EDDT Mirex EPCB

median 0.50 11.4 0.8 19.0 135.7 4.4 490.3 mean 0.49 14.8 0.9 31.7 178.3 5.9 660.5 range (0.34-0.64) (4.9-45.0) (0.2-2.9) (6.9-99.6) (36.6-787.1) (I.2-19.0) (103.3-3 173.9)

median (100 %) 2 459 181 4 899 26 093 753 83 225 mean 3 065 195 6 489 37 101 1 188 136 686 range (800 - (66.8 - 654) (1 148 - (7 190 - (191 - (16 927 -

10 000) 22 138) 174 904) 4 223) 705 312)

2631

1.2

1.1

o 1.0

0.9

0.8

= ~ 0.7

~ 0.4 88

0.3

0.2

0.1

~ ® ' & O o o O 0 0 0 0 0 0 0

Figure 1. Pattern oforganochlorines in blood from captive glaucous gulls, Larus hyperboreus, from Svalbard, fed polar cod Boreogadus saida. Blood was sampled after 24-41 days of captivity. For each compound, the mean concentration relative to PCB-153 are plotted (cone. compound x / cone. PCB-153). Error bars indicate 1 S.E.

Relationships between OCs in blood an OCs in tissues

Blood-tissue correlations were not calculated for OCs below the quantification limit jn more than half of the tissue

samples, i.e., ct-HCH (liver and fat), transnonachlor (brain), PCB-151 (liver & brain), PCB-141 (brain), PCB-199

(liver & brain) and PCB-209 (brain). Of the OCs quantified in blood, only a-HCHdid not correlate significantly

with the corresponding level in any of the analysed tissue lipids (Fig. 2). a-HCH was below the quantification

limit in liver, fat, and in one brain sample. PCB-141 was above the quantification limit in 9 of 15 liver samples,

but no significant blood-liver correlation was found (Fig. 2). Of the OCs above the quantification limit in all

samples, blood-tissue correlations were weakest for p,p '-DDT, and the blood-liver correlation was not significant

when based on concentration in blood extractable lipids (Fig. 2). For the blood-liver correlations, r 2 was >0.75 for

29 of 35 OCs. Blood-tissue correlations were calculated both based on whole blood wet weight concentrations and

on blood extractable lipid concentrations. For 34 of 35 correlations between blood and liver, r ~ was higher when

using blood wet weight concentrations than when using concentrations in extractable blood lipid (Fig. 2a). The

blood-brain correlations did not show any consistent dependence on which way blood concentrations were

calculated (Fig. 2b). In contrast, r 2 for 31 of 37 blood-fat correlations were higher when using concentrations in

blood extractable lipids than when using concentrations in whole blood (Fig. 2c).

2632

In all t h ree p lots : mum Squared correlation coefficients based on blood wet weight concentrations

Squared correlation coefficients based on concentrations in blood extractable lipids ........ p (r 2 > 0.264) = 0.05

a) Liver 1.0 0,8

/2 0.6 0.4 0.2 0.0

p=O.05

$ $ ~ 6.6. oo 0. o o o o o o o o o o u o o o o o o o 0. ~0. 13. O o Q. f t . f t . O. I~ n n Ck Q. ft. O_ Q. 13_ ft. Q. ft. n n ~.=_ ==

b) Brain 1.0

°"!!! Illll IIIIHHK,,HKNNultNHN u Buui /2 0 6

o., " U . . . . . . . . . . . " , II N 0.2 . . . . . n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . p=0.05

~ ~ 0 0 ~ ~ m m m m m m m m m m m m m m m m m m ~ 0 o o o o o o 0 o o o o ~ o o o 0 ~

"0

c) Subcutaneous fat 1.0 0.8

/2 0.6 0.4 0.2 0.0

p = O . 0 5

_~ ×°0

O0 ~ m d~ m m co ¢fi ~ d~ m d~ m d~ m m m d~ m g m m~ _ _ ~ ~ oo

~ Q. I~ Q. Q,. I~. 13_ O. I~. I~. I~. O. n I~. O. I~. O. Q. I~ Q.

O O ~0

Figure 2. Squared correlation coefficients between organochlorine concentrations in blood from adult captive glaucous gulls and corresponding concentrations in extractable lipids from: a) liver, b) brain, and c) subcutaneous

fat, Correlations for both blood wet weight concentrations and for blood extractable lipid concentrations are included, Log-transformation was performed prior to calculation of correlations. The reference line indicates the

significance limit for a single correlation.

2633

t : i - i ! o _Bra,,, _

j i , i

10 -1 1 : ~ i . ::

I i!: ::::: :f:! {77!

V-

. . . . L . i t z__==o ° o =" ~.,,-'~a t a t a ° ° ° ~ ° = t a o o o o o o o o o t a o o o o o o = o o o o o o

t a Q. t a r t Q. t a Q . Q . t a D _ t a t a t a t a t a Q.

Figure 3. Ratios between tissue lipid organochlorine concentrations and corresponding concentrations in blood extractable lipids from captive glaucous gulls. Geometric means and 95% confidence intervals are plotted.

For the PCBs with higher degree of chlorination, the blood-liver correlations were stronger than corresponding

blood-brain correlations and blood-fat correlations (Fig. 2). In the case of the blood-fat correlations, a single bird

with extreme fat to blood concentration ratios for the higher chlorinated PCBs weakened the corresponding blood-

fat correlations. Fat concentrations from the above-mentioned extreme individual was excluded in the plot of

average tissue-to-blood concentration ratios in Figure 3. With the exception ofp,p '-DDT, the geometric means of

the OC concentration ratios between fat extractable lipids and blood extractable lipids ranged from 1 to 5 (Fig. 3).

For PCBs, the fat-to-blood concentration ratio increased with the degree of chlorination (Fig. 3). Geometric means

of the liver-to-blood OC concentration ratios (extractable lipid basis) were from 0.9 to 2.0. The corresponding

mean brain-to-blood ratioswere 0.2 to 0.4, with the exception of 13-HCH, p,p'-DDT, and PCB-199. In brain, 13-

HCH and PCB-199 were below or near the quantification limit in most samples. For all tissues, tissue-to-blood

ratios were higher for p,p '-DDT than for other OCs, significantly in fat and brain (Fig. 3).

2634

100

1 Percentage of Total Sum of Squares explained by conc. in 1st blood sample 1 Percentage of Total Sum of Squares explained by body mass change

between 1st and 2 nd blood sample

O0 ,= -=l O'.--.. ¢n09

E ~

I-.-

~ . . c - O

90

80

70

60

50

40

30

20

10

0

Figure 4. Partitioning of the total sum of squares for the linear models that were fitted to the relationship between organochlorine (OC) concentrations in two consecutive sets of blood samples from captive glaucous gulls. In the models, the concentration of an OC in the 2 "d blood sample (]/) was assumed to depend on X1, the corresponding

concentration in the 1 st blood sample, and on X2, the change in body mass that occurred between the 1 st and the 2 na blood sample (see statistical methods in text and Table 2). The figure shows how much of the total variance in the 2 "a blood samples that can be explained by the two parameters in the model. For each compound, w.w. indicates partitioning of variance using wet weight blood concentrations. Correspondingly, 1.w. indicates partitioning of

variance using lipid weight blood concentratiohs.

lntra-individual temporal variability in OC blood levels

Thirteen selected OCs were analysed both in the terminal blood sample (2 nd sample) and in a blood sample taken

on one of the first ten days of captivity (1 st sample). Multivariable linear regression models were used to

investigate the relationship of the OC concentration in the 2 "d sample to the corresponding concentration in the 1 st

analysed blood sample and the body mass change between the two samples. These models were fitted both for

lipid and wet weight blood concentrations. In the models, the explained parts of the total sums of squares (i.e. the

total variance in the 2 "d blood sample) were partitioned between the two independent variables (Fig. 4). The

proportion of the variance that is accounted for by the concentrations in the 1 st sample is equivalent with the

squared correlation coefficient (r 2) for the correlation between the 1 st and the 2 nd blood sample. Thus, the intra-

individual temporal correlation in blood concentrations can be read from the lower part of the stacked columns

2635

(Fig. 4). Intra-individual temporal correlations were stronger when based on wet weight values than when based

on concentrations in extractable lipids (Fig. 4). Even when using the wet weight calculation, the correlation

between the 1 st and the 2 na sample were absent for p,p '-DDT and accounted for less than 25 % of the variance for

HCB and PCB-28. For PCBs, the proportion of variance explained by the 1 st sample seemed to improve with the

degree of chlorination (Fig. 4).

Excluding p,p '-DDT, variation in the 2 nd blood concentration was significantly better explained when the body

mass change was included, when compared to models based only on the initial blood concentration (Fig. 4, Table

2). Regression coefficients for body mass change ranged from -0.0038 to -0.0026 (Table 2). Thus, according to

the model, a 100g decrease in body mass between the 1 st and 2 nd sampling would add an estimated 0.26-0.38 to

[Log (conc. in 2 nd blood sample)] for the different compounds. This is equivalent to multiplicative factors in the

range from 1.8 to 2.4 for the concentration of the different OCs.

With the exception ofp ,p '-DDT, these complete models accounted for 53 % (HCB) to 92 % (PCB-180) of the total

variance in the 2 na blood samples (Fig. 4). According to the multivariable models, the compounds differed with

respect to which parameter that explained most of the variation in the 2 nd blood concentration (Fig. 4). For PCBs-

153, -170 and -180s, the blood concentration in the 1 st sample accounted for most of the variance explained by the

model. In contrast, most of the variance in the 2 "d sample for HCB and PCB-28 was related to change in body

mass (Fig. 4).

Table 2. Estimated regression coefficients for the linear model fitted to describe the relationship between organochlorine (OC) concentrations in two consecutive sets of blood samples from captive glaucous gulls. In the model, the wet weight concentration of an OC in the 2 nd blood sample (Y) were assumed to depend on XI, the

St corresponding concentration in the 1 blood sample, and on )(2, the change in body mass that occurred between the st n o 1 and the 2 blood sample. The model was of the following structure:

Y = ,80 + fllXl + ,82X2 + E (see statistical methods in text for further details, see also Fig. 4).

Compound ~ ,~* ~ * HCB -0.2961 0.9210* -0.0029* oxychlordane -0.5197 1.0798" -0.0028* p,p'-DDE -0.8272 1.1880* -0.0033* p,p'-DDT -0.5615 0.5335 -0.0028 PCB-28 -0.4624 0.9026* -0.0034* PCB-52 -0.6220 0.8570* -0.0033* PCB-101 -0.5297 1.0744" -0.0034* PCB-99 - 1.0130 1.3819 * -0.0036* PCB-118 -1.1291 1.3673' -0.0038* PCB- 153 -1.2638 1.3663" -0.0036* PCB-138 -1.3012 1.4107" -0.0038* PCB-180 -0.5387 1.0707" -0.0026* PCB- 170 -0.5079 1.0609* -0.0029*

* For ]~1 and/~: significantly different from zero, i.e. indicating when the effects of Xl or)(2 on Y are significant (t-test, p<0.05).

2636

As noted above, intra-individual temporal correlations were stronger for wet weight blood concentrations than for

lipid weight concentrations. In contrast, the association of the 2 "d blood concentrations to changes in body mass

were stronger on a lipid weight basis than on a wet weight basis (Fig. 4). Overall, wet weight and lipid weight

concentrations gave roughly the same r 2 for the total models. However, lipid weight concentrations tended to give

higher total r 2 for the compounds for which changes in body mass were more important (HCB and PCB-28). In

contrast, wet weight concentrations gave higher r 2 for the compounds were the 1 st concentration were more

important (i.e., the higher chlorinated PCBs).

The average OC concentrations in blood tended to decline during the period of captivity (Table 3). This decline

was statistically significant for p,p'-DDT and PCBs no. 52, 101,180 and 170 (Table 3). 14 of 15 birds lost mass

between the two blood samples. To estimate the mean change in blood concentration expected with no mass

change, predictions were made using the regression coefficients of the linear model (except p,p'-DDT, Table 2).

The mean of the concentrations in the 1 st blood samples and zero body mass change were inserted into the models,

and predicted mean concentrations in the 2 nd sample were calculated with 95% confidence intervals. These

estimated means were in the range of 0.25 to 0.40 relative to the corresponding means in the 1 st blood samples, and

they were all significantly lower than their corresponding means in the 1 st blood samples (Table 3).

Table 3. Mean ratios between wet weight organochlorine concentrations in two consecutive sets of blood samples from captive glaucous gulls. The second column provides the mean of the observed 2nd:lSt concentration ratios. The third column gives estimates for the same ratios for the special case of no body mass change between the samples. To estimate these ratios, the mean 1 st blood concentrations and zero body mass change were inserted into the multivariable statistical models described in Table 2. All ratios are presented with 95 % confidence-intervals.

Compound (Cone. in 2nd sample) Estimated 2ha: I st sample cone. ratio with (Cone. in 1st sample) no intervening body mass change

HCB 0.80 (0.56-1,14) 0.41 (0.23-0.72)* oxychlordane 0.75 (0.52-1.08) 0,39 (0.21-0.75)* p,p'-DDE 0.81 (0.55-1.19) 0,38 (0.21-0.71)* p,p'-DDT 0.46 (0.28-0.75)* n.c. PCB-28 0.73 (0.50-1.07) 034 (0.20-0.57)* PCB-52 0.54 (0.36-0.81)* 0.25 (0.15-0.44)* PCB-101 0.67 (0.45-1.00)* 0.31 (0.16-0.59)* PCB-99 0.79 (0.55-1.12) 0.35 (0.21-0.57)* PCB-118 0.75 (0.52-1.09) 0.32 (0.19-0.53)* PCB-153 0.78 (0.54-1.13) 0.34 (0.22-0.55)* PCB-138 0.76 (0.52-1.11) 0.32 (0.20-0.51)* PCB-180 0.72 (0.55-0.94)* 0.40 (0.29-0.55)* PCB-170 0.72 (0.55-0.96)* 0.38 (0.27-0.53)* * significantly different from 1 (t-test, p<0.05). n.c. = not calculated due to non-significant effects of the parameters in the statistical model (see Table 2).

2637

~0 10 s o .0

r-

"i lOa o

o

• ~PCB, r2=0.59, p=0.0008 0 I;DDT, rZ=0.53, p=0.002

• • HCB, rZ=0.74, p<0.0001 o v I;HCH, r2=0.70, p<0.0001

4 •

V --

v v ~ v

20 40 60 80 1 O0

% extracted lipid from subcutaneous fat

Figure 5. Organochlorine concentrations in blood from captive glaucous gulls as a function of the percentage of extractable lipid in subcutaneous fat. The scale on the vertical axis is logarithmic and the linear regressions were

performed on log-transformed concentration data.

Relationship between OCs in blood and extractable lipid in fat

OC concentrations in blood were negatively correlated with the percentage extractable lipid from subcutaneous fat

(Fig. 5). The squared correlation coefficient (r 2) was higher for lipid weight concentrations than for wet weight

concentrations (Fig. 6). Correlations were weakest for a-HCH, the chlordanes, PCB-52 and PCB-141 (Fig. 6).

DISCUSSION

Types and levels of OCs

The OC burdens found in this study were similar to those found in previous studies of glaucous gulls in the

Norwegian Arctic [14,15]. It should he noted that the PCB levels in the glaucous gulls in this study are one order

of magnitude higher than the high PCB levels recently found in polar hears at Svalbard [I]. The hepatic

concentrations of mono-ortho chlorinated PCBs (median 18 ~g/g lipid) in the present study were similar to the

mono-ortho PCB concentrations reported in yolksacs of common tem, Sterna hirundo, hatchlings from polluted

areas in the Netherlands and Belgium [16]. It has been demonstrated that the nutritional condition at the time of

sampling can have a great influence on OC measurements [5,13]. In accordance with this, even higher levels of

OCs in glaucous gulls on Svalbard have been found previously in emaciated dead or dying birds [2,8,17, unpubl.

data.]

2638

~ 1 Squared correlation coefficients based on wet weight concentrations 1.0 " ~ Squared correlation coefficients based on lipid weight concentrations

0.9 0.8 0.7

.20.6 0.5 0.4 (1.3 0.2 p~0.05

0.1 0.0

Figure 6. Squared correlation coefficients (r 2) between OC concentrations in blood from glaucous gulls and percentage exmLctable lipids from subcutaneous fat(index of nutritional condition). Correlations for both blood wet

weight concentrations and for blood extractable lipid concentrations are included. Log-transformation of concentrations was performed before calculation of correlations. The reference line indicates the significance limit

for a single correlation.

Meta-para unsubstituted PCB congeners were either below the quantification limit or detected in lower

concentrations than other congeners. This is in accordance with the findings reported for five species of fish-

eating seabirds [18]. In the present study, the exception in this respect was PCB-149. This discrepancy could be

explained by a possible analytical overlap with PCBs -123 and -139, which both have only ortho-meta

unsubstituted positions. Hepatic aldrin epoxidase activities indicatethat omnivore birds, including .gulls, have

better detoxification abilities than birds with more speeialised feeding habits [19]. Moreover, the presence of

proteins cross reacting with anti-rat CYP2B have been demonstrated in the glaucous gull liver [20].

Relationship between blood and tissues

With the exception of p,p '-DDT in blood lipids, absences of significant blood-tissue concentration correlations

were associated with OCs below the quantification limit in at least one of the tissue samples (i.e., ct-HCH in brain

and PCB-141 in liver). The finding of r 2 values >0.75 in more than 80% of the blood-liver correlations indicates

that good estimates for OC tissue concentrations in the glaucous gull can be made based on blood samples.

Due to the poor nutritional condition of several of the birds, only small samples of subcutaneous fat of very

varying consistence and colour could be obtained from some birds. The somewhat weaker blood-tissue

correlations for fat tissue could therefore be related to the inconsistent quality of the fat samples.

The correlations between concentrations in brain and blood were weaker for the higher chlorinated PCBs,

compared both to the corresponding blood-liver correlations and compared to other OCs in brain. Moreover, the

proportions of nona- and decachlorobiphenyls were lower in brain than in blood. In contrast, the relative

concentrations of PCBs in fat, compared to blood, increased with the degree of chlorination. These observations

2639

may relate to differences in lipid composition between tissues. In glaucous gulls, the brain contains more

cholesterol and phospholipids, and less triglycerids, than liver and subcutaneous fat [21]. Thus, the brain contain

more polar lipids than the other tissues. The tissue:blood partition coefficients of PCBs depends on each

congeners molecular structure and the amounts and types of lipids in the tissue [7]. The lipophilicity of PCBs

generally increases with increasing number of chlorines [22]. It is therefore not surprising that the highest

chlorinated PCBs are most skewed towards the least polar compartments. In accordance with this, higher

proportions of the higher chlorinated PCBs were found in blubber when PCB patterns in blubber and serum from

the grey seal, Halichoerus grypus, were compared [23]. In contrast, similar PCB patterns in subcutaneous fat and

plasma were observed in the polar bear, possibly due to a less polar plasma compartment in polar bear compared to

other species [ 1 ].

lntra-individual temporal variability

As illustrated in Fig. 4, the correlation between blood samples taken 3-5 weeks apart were stronger for wet weight

OC concentrations than for the corresponding lipid weight concentrations. A possible reason is that the low

amounts of extractable lipids in blood could result in loss of precision in the gravimetrical determination of

extractable lipids. The available blood samples in the present study were between 5 and 10 ml. For their method

for determination of PCBs in human blood, Mes et al. concluded that 50 ml was the minimum acceptable sample

volume with regard to the blood lipid determination [24]. An intedaboratory study on organochlorine analysis in

fish found that analytical methods need to be improved if OC concentrations are to be standardised to percent lipid

for valid comparison between species and localities [25]. When data are standardised to lipid weight, different

extraction methods may hamper comparison between studies because different lipids are extracted with varying

efficiency, Nevertheless, lipid weight values can have superior informative value when the lipid-content of the

tissue is variable [13,26]. As the lipid content of the tissue is decisive for the tissue:blood partition coefficients

[27], expression on a lipid weight basis gives values that are more directly comparable between blood and tissues,

and between different tissues.

The correlation between the 1 st and the 2 no blood samples in our study should be expected to be highest for the

most persistent compounds. Since all birds were given similar food with low OC levels food during captivity, the

initial variance in OC-concentrations between birds should have greatest retention for the compounds with lowest

clearance rates. Thus, no intra-individual temporal correlation was found for p,p '-DDT, which is metabolised in

birds [28] (Fig. 4). Whole-body clearance rates have been estimated in juvenile herring gulls Larus argentatus for

ten OC compounds [29]. HCB had a higher clearance rate than p,p'-DDE and oxychlordane [29]. In the present

study, the intra-individual temporal correlation for HCB had a r 2 value <0.20, which is about 50% of the

corresponding value for p,p'-DDE and oxychlordane. This suggests that HCB may have a clearance rate higher

than oxychlordane and p,p '-DDE in the glaucous gull. In juvenile herring gulls, clearance ofp,p '-DDE was <50%

of the clearance of oxychlordane [29]. We did not find a similar relationship in the present study. However, the

intra-individual temporal correlations are only indirect measures of clearance rates. The background levels in

264O

study diet may have confounded the results. In particular, HCB is one of the major OCs in polar cod from the

Barents Sea [11 ]. In the case of the higher chlorinated PCBs, good intraindividual temporal correlations suggests

consistent differences in body burdens between individuals. Thus, concentrations of the higher chlorinated PCBs

in a single blood sample integrate exposure over a longer period of time, compared to p,p'-DDT and HCB.

Based on the loss of body mass during the experiment, OC concentrations could have been expected to increase

during the experiment [13]. The results from the multivariable model (Table 2) also indicated an inverse

relationship between mass change and OC concentrations in blood. However, all congeners that were measured

twice decreased from the 1 st to the 2 nd sample. The decrease was significant for 5 out of 13 compounds.

According to the empirical association between mass loss and blood OC levels (Table 2), the mean decreases in

blood OC concentrations would have been larger in absence of mass loss (Table 3). A possible explanation is that

the polar cod consumed during captivity had lower concentrations of OCs than the average diet of free-living

glaucous gulls [11]. The largest concentration decreases were observed for p,p'-DDT, PCB-52 and PCB-101,

which could indicate comparably lower persistence. In contrast to the other PCB congeners that were measured

twice, both PCB-52 and PCB-101 are meta-para unsubstituted. They could therefore be expected to be more

easily metabolised [ 18].

For half of the OCs measured twice in the present study (i.e., HCB, p,p '-DDE, p,p '-DDT, PCBs -28, -99 and -118),

the concentrations in the 2 "0 blood sample correlated stronger to lipid status (see Fig. 6) than to the corresponding

blood levels 3-5 weeks earlier (see Fig. 4, lower bars). Temporal variability could be expected to be even larger in

free-living glaucous gulls, which experience fluctuations in nutritional condition and have a very diverse diet [30].

Repeated sampling of free-living gulls over an extended period could provide information on the consistency of

the interindividual variability in blood concentrations. To relate OC burdens to possible effects, it is of interest to

know whether certain individuals have chronically higher OC burdens over time, or if a point sample is heavily

influenced by intra-individual temporal variability.

Relationship between levels of OCs and nutritional condition

The relationship between tissue concentrations of OCs and nutritional condition has been established previously in

birds [13,31]. For most OCs included in the present study, more than 40% of the,variation in blood concentration

was associated with variation in nutritional condition (measured as percentage extractable lipids from

subcutaneous fat). The weaker correlations of the meta-para unsubstituted PCB-52 and PCB-141 with nutritional

condition may be related to the higher metabolic clearance that is expected for these PCBs [18]. The meta-para

unsubstituted octachlorobiphenyl PCB-199 is, however, not less correlated with nutritional condition than most

other PCBs (Fig. 6).

In the present study, the ranges of measured concentrations of each compound span more than one order of

magnitude, with about 50% of the variation related to nutritional condition. Moreover, in the model for the

2641

relationship between the 1 st and the 2 nd blood sample, the inclusion of body mass change gave an improvement in

the explained proportion of the total variance (see Fig. 4). Regression coefficients in the same model indicated that

a 100 g decrease in body mass was associated with an approximate doubling of blood OC concentrations. Thus, in

seabirds, mere concentration measurements of organochlorines give only a rough indication of the actual OC

burden, as long as the nutritional condition of the sampled individuals are unaccounted for. The more laborious

and not completely non-destructive approach of homogenising whole bird carcasses to measure whole-body

burdens of OCs [32] could possibly give results that are easier to interpret.

Conclusions

Blood concentrations seem to be roughly equivalent to tissue concentrations in assessing body burdens of most of

the investigated OCs. Similar to tissue concentrations, the utility of blood concentrations are limited by their

interrelationship with fluctuations in the size of the body lipid pool, i. e., nutritional condition. Blood and tissue

concentrations are point samples in time of a dynamic process, and great care should be taken to avoid

unwarranted conclusions about differences in OC burdens between individuals or populations. The possible

confounding factor of recent feeding was not investigated and may add noise when sampling from free-living

birds. In this study, we found that in birds on identical diets and with standardised blood sampling, blood levels of

several OCs could only roughly be predicted from corresponding levels in the same birds 3-5 weeks earlier. In

terms of the parameters associated with blood levels, the investigated OCs could be described as spanning a range

f/ore p,p'-DDT, HCB and PCB-28 to the higher chlorinated PCBs. For the former compounds, lipid status was

found most important for blood concentrations, while for the higher chlorinated PCBs, the results indicated that the

chronic body burden was more decisive.

ACKNOWLEDGEMENTS

We would like to thank Bente Brekke for her care 6f captive gulls in Ny-Alesund. We are also grateful for the

technical assistance from Anuschka Polder and Elisabeth Lie. Andrew E. Derocher provided valuable comments

on the manuscript. This study was funded by the Norwegian Research Council (grant I09120/720), the Norwegian

Polar Institute, Tromso, and the National Veterinary Institute, Oslo. This paper is contribution no. 328 from the

Norwegian Polar Institute.

REFERENCES

1. A. Bernhoft, t0. Wiig and J. U. Skaare. Organochlorines in polar bears (Ursus maritimus) at Svalbard.

Environ. Pollut. 95, 159-75 (1997).

2. G.W. Gabrielsen, J. U. Skaare, A. Polder and V. Bakken. Chlorinated hydrocarbons in glaucous gulls (Larus

hyperboreus) in the southern part of Svalbard. Sci. Total Environ. 160/161,337-46 (1995).

3. M. Friend, M. A. Haegle, D. L. Meeker, R. Hudson and U. H. Baer. Correlations between residues of

dichlorodiphenylethane, polychlorinated biphenyl, and dieldrin in the serum and tissues of mallard ducks

(Anas platyrhynchos). In: Animals as monitors of environmental pollutants pp. 319-26, National Academy

2642

of Sciences, Washington D.C., (I 979).

4. J. Mes. Organochlorine residues in human blood and biopsy fat and their ~elationship. Bull. Environ.

Contain. Toxicol. 48, 815-20 (1992).

5. D.E. Capen and T. J. Leiker. DDE residues in blood and other tissues of white-faced ibis. Environ.

Pollut. 19, 163-71 (1979).

6. L. Marsili, M. C. Fossi, S. Casini and S. Focardi. PCB levels in bird blood and relationship to MFO

responses. Chemosphere 33, 699-710 (1996).

7. F.M. Parham, M. C. Kohn, H. B. Matthcws, G DeRosa and C. J. Portier. Using structural information to

create physiologically based pharmacokinetic models for all polychlorinated biphenyls I. Tissue:blood

partition coefficients. Toxicol. Appl. Pharmacol. 144, 340-47 (1997).

8. F. Daelemans. Polychlorinated biphenyls and some selected organochlorine pesticides in seabirds and marine

mammals from the Svalbard archipelago. Dr. thesis, University of Antwerp (1994).

9. K.E. Erikstad, Winter diets of four seabird species in the Barents sea after a crash in the capelin stock. Polar

Biol. I0) 619-27 (1990).

I0. C. Lydersen, I. Gjertz and J. M. Weslawski. Stomach contents of autumn-feeding marine vertebrates from

Hornsund, Svalbard. Polar Record 25, I07-17 (1989).

i I. K. Borg~. Bioaccumulation of organochlorines (OCs) in an Arctic marine food chain. Candidata Scientiarum

thesis, University of Tromso (1997).

12. E.M. Brevik. Gas chromatographic method for the determination of organochlorine pesticides in human

milk. Bull. Environ. Contain. Toxicol, 19, 281-86 (1978).

13. E.O. Henriksen, G. W. Gabrielsen and J. U. Skaare. Levels and congener pattern ofpolychlorinated

biphenyls in kittiwakes (Rissa tridactyla), in relation to mobilization of body-lipids associated with

reproduction. Environ. Pollut. 92, 27-37 (1996).

14. G. Norheim and B. Kjos-Hanssen. Persistent chlorinated hydrocarbons and mercury in birds caught off the

west coast of Spitsbergen. Environ. Pollut. Set. A 33, 143-52 (1984).

15. F. Mehlum and F. F. Daelemans. PCBs in Arctic seabirds from the Svalbard region. Sci. Total Environ.

160/161,441-46 (1995).

16. A.J. Murk, A. T. C. Bosveld, M. van den Berg and A. Brouwer. Effects of polyhalogenated aromatic

hydrocarbons (PHAHs) on biochemical parameters in chicks of the common tern (Sterna hirundo). Aquat.

Toxicol. 30, 91-I 15 (1994).

17. J.A. Bogan and W. R. P. Bourne. Organochlorine levels in Atlantic seabirds. Nature 240, 358 (1972).

18. J.T. Borlakoglu, J. P. G. Wilkins, C. H. Walker and R. R. Dils. Polychlorinated Biphenyls (PCBs) in Fish-

eating Sea Birds -III. Molecular Features and Metabolic Interpretations of PCB Isomers and Congeners in

Adipose Tissues. Comp. Biochem. PhysioL Ser. C97C, 173-77 (1990).

19. M.C. Fossi, A. Massi, L. Lari, L. Marsili, S. Focardi, C. Leonzio and A. Renzoni. Interspecies differences in

mixed function oxidase activity in birds: Relationship between feeding habits, detoxication activities and

2643

organochlorine accumulation. Eitviron. Pollut. 90, 15-24 (1995).

20. E.O. Heoriksen, G. W. Gabrielsen, J. U. Skaare, N. Skjegstad and B. M. Jenssen. Relationships between

PCB levels, hepatic EROD activity and plasma retinol in glaucous gulls, Larus hyperboreus. Mar. Environ.

Res. (in press).

21. I. Halvorsen. Lipidsammensetning i polarmLke (Larus hyperboreus). Candidatus Scientiarum thesis,

University of Tromso (1997).

22. M.D. Erickson. Physical, chemical, commercial, environmental, and biological properties of PCBs. In:

Analytical chemistry of PCBs pp. 5-40, ed. M. D. Erickson, Butterworth Publ., Boston, USA (1986).

23. R.F. Addison and P. F. Brodie. Transfer of organoehlorine residues from blubber through the circulatory

system to milk in the lactating grey seal Halichoerus grypus. Can. J. Fish. Aquat. Sci. 44, 782-86 (1987).

24. J. Mes; L. Marchand, D. Turton and P.-Y. Lau. The determination ofpolychlorinated biphenyl congeners and

other chlorinated hydrocarbon residues in human blood, serum and plasma. A comparative study. Intern. J.

Environ. Anal. Chem. 48, 175-86 (1992).

25. A.G. Miskiewicz and P. J. Gibbs. Variability in organoehlorine analysis in fish: An interlaboratory study and

its implications for environmental monitoring and regulatory standards. Arch. Environ. Contam. ToxicoL 23,

45-53 (1992).

26. A. Bignert, A. G6thberg, S. Jensen, K. Litz6n, T. Odsj6, M. Olsson and L. Reuterg~dh. The need for

adequate biological sampling in ecotoxilogical investigations: a retrospective study of twenty years pollution

monitoring. Sci. TotalEnviron. 128, 121-39 (1993).

27. P. Poulin and K. Krishnan. A biologically-based algorithm for predicting human tissue: blood partition

coefficients of organic chemicals. Hum. Exp. Toxicol. 14, 273-80 (1995).

28. L.J. Blus. DDT, DDD, and DDE in Birds. In: Environmental contaminants in wildlife. Interpreting tissue

concentrations pp. 49-71, eds. W. N. Beyer, G. H. Heinz and A. W. Redmon-Norwood, CRC Press, Boca

Raton, USA (1996).

29. T.P. Clark, R. J. Norstrom, G. A. Fox and H. T. Won. Dynamics of organoehlorine compounds in herring

gulls (Larus argentatus): II. A two-compartment model and data for ten compounds. Environ. ToxicoL

Chem. 6, 547-59 (1987).

30. S.J. Barry and T.W. Barry. Food habits of glaucous gulls in the Beaufort Sea. Arctic 43, 43-49 (1990).

31. A. Sfidergren and S. Ulfstrand. DDT and PCB relocate when caged robins use/'at reserves. Ambio 1, 36-40

(1972).

32. D.W. Anderson, D.G. Raveling, R. W. Risebrough and A. M. Springer. Dynamics of low-level

organochlorines in adult cackling geese over the annual cycle. J. Wildl. Manage. 48, 1112-27 (1984).