A multilevel approach to predict toxicity in copepod populations: Assessment of growth, genetics, and population structure

Aquatic Toxicology 79 (2006) 41–48

A multilevel approach to predict toxicity in copepod populations: Assessment of growth, genetics, and population structure Johanna Gardestr¨om a,∗ , Elena Gorokhova a , Michael Gilek b , Mats Grahn b , Bengt-Erik Bengtsson c , Magnus Breitholtz c a

Department of Systems Ecology, Stockholm University, SE-10691 Stockholm, Sweden School of Life Sciences, S¨odert¨orn University College, SE-14189 Huddinge, Sweden c Department of Applied Environmental Science (ITM), Stockholm University, SE-10691 Stockholm, Sweden b

Received 23 March 2006; received in revised form 2 May 2006; accepted 3 May 2006

Abstract One of the goals of environmental risk assessment (ERA) is to understand effects of toxicant exposure on individual organisms and populations. We hypothesized that toxicant exposure can reduce genetic diversity and alter genotype composition, which may ultimately lead to a reduction in the average fitness of the exposed population. To test this hypothesis, we exposed a copepod, Nitocra psammophila, to a toxic reference compound and assayed resulting alterations in genetic structure, i.e. expected heterozygosity and percent polymorphic loci, as well as other population- and fitness-related measures, i.e. population abundance, demographic structure and juvenile growth. The copepods were exposed to 0.11–1.1 ␮g of the pentabromo-substituted diphenyl ether (BDE-47) mg−1 freeze-dried algae for 24 days (i.e. >1 generation). There was no significant decline in total population abundance. However, there were significant alterations in population structure, manifested as diminished proportion of nauplii and increased proportion of copepodites. In addition, individual RNA content in copepodites decreased significantly in exposed individuals, indicating declined growth. Finally, in the exposed populations, heterozygosity was lower and genotype composition was altered compared to the controls. These results therefore confirm the hypothesized reduction in overall genetic variability resulting from toxicant exposure. Multilevel approaches, such as the one used in the present study, may help unravel subtle effects on the population level, thus increasing the predictive capacity of future ERA. © 2006 Elsevier B.V. All rights reserved. Keywords: Environmental risk assessment; Multilevel; Toxic exposure; Crustacea; Population genetics; RNA

1. Introduction Predicting impacts of anthropogenic chemical substances on ecosystems is one of the main goals of environmental risk assessment (ERA). Usually, ERA is based on a bottom–up methodology, where effects at higher organization levels are extrapolated from results of standardized toxicity tests (European Commission, 2003). These tests often aim to identify effects on individual organisms, most often short-term lethal effects (i.e. LC/EC50 ) but sometimes also chronic no-observed-effect concentrations (NOECs; Forbes et al., 2001). However, extrapolating from the individual to the ecosystem level is a complex task, because the connections between these levels are far

∗

Corresponding author. Tel.: +46 8 16 37 04; fax: +46 8 15 84 17. E-mail address: [email protected] (J. Gardestr¨om).

0166-445X/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.aquatox.2006.05.001

from straightforward and sometimes even misleading (Calow et al., 1997; Belfiore and Anderson, 2001; Forbes and Calow, 2002). Furthermore, it has been suggested that population level responses need to be included in ecotoxicological tests to allow better predictions of effects and risks at the ecosystem level (Bechmann, 1994; Calow, 1996; Calow et al., 1997; Forbes et al., 2001; Forbes and Calow, 2002). This means that developing testing strategies that include scientifically based combinations of species and endpoints is essential for evaluating the complexity of potential biological and environmental effects at a reasonable financial cost (e.g. Escher and Hermens, 2002). Biochemical markers together with biological indicators at higher levels of biological organization can provide good measures of altered state in individual organisms and populations evoked by exposure to toxicants. At the perspective of the individual organism level, the effects of a toxicant can be tackled by behavioral mechanisms (e.g. avoidance) or by cellular mecha-

42

J. Gardestr¨om et al. / Aquatic Toxicology 79 (2006) 41–48

nisms, such as biochemical and physiological responses (Van Straalen, 2003). The latter allows for acclimatization at the cost of important fitness traits such as growth and fecundity (Futuyma, 1997). Measurements of metabolic enzyme activities and RNA-based indices have been especially valuable as indicators of condition in studies of marine invertebrates and fish, i.e. groups for which accurate determination of metabolic and growth rates are difficult (Dahlhoff, 2004). At the population level, tolerance can be developed by genetic adaptation (Van Straalen, 2003), which may lead to alterations in the genetic composition and genetic diversity of the population. Pollution is a well-documented selective force that has been found to, e.g., induce metal tolerance in plants (e.g. Bradshaw and McNeilly, 1981) and pesticide resistance in insects (e.g. Devonshire et al., 1998). More recently, pollution tolerance has been detected in a variety of aquatic organisms (Tanguy et al., 1999; Belfiore and Anderson, 2001). The reduction in overall genetic variability is a frequently observed phenomenon (Ma et al., 2000; Van Straalen and Timmermans, 2002; Ross et al., 2002), even when the population abundance remains unaltered (Street and Montagna, 1996). It can result from a genotoxicant-induced selection at loci important for survival and propagation in polluted environments or from a genetic drift, e.g. as a result of a population bottleneck (Staton et al., 2001), and might ultimately lead to a reduction in the average fitness of the exposed populations (Bickham et al., 2000). Populations of the meiobenthic harpacticoid copepod Nitocra spinipes have been observed to exhibit decreased mean juvenile developmental rate when exposed to brominated flameretardants and synthetic musk fragrances (Breitholtz et al., 2003; Breitholtz and Wollenberger, 2003). In these studies, the pollutant-exposed populations sustained the ability to maintain a stable population growth. It was suggested that the observed phenomenon could be attributed to compensatory effects (i.e. decreased competition among surviving individuals and/or shortened inter-clutch period). This could lead to differential survival and reproduction of tolerant individuals, i.e., genetic adaptation at the cost of the decreased effective population size. As a result, the genetic variability within these toxicant-tolerant populations could have been reduced, through decreased heterozygosity compared to the control populations. The aim of this study was to: (1) test the hypotheses that toxicant exposure can reduce expected heterozygosity and genetically differentiate populations and (2) compare these population genetic endpoints with other population- and growth-related measures, i.e. population abundance, demographic structure and juvenile growth. To investigate this, we exposed Nitocra psammophila to the polybrominated diphenyl ether (BDE-47) in a laboratory test that lasted longer than a generation time. 2. Materials and methods 2.1. Test organisms In general, harpacticoid copepods have short generation times, and thus they can be easily studied for the duration of a full

life cycle, focusing on all crucial ontogenetic stages and life history traits, such as larval development, reproduction, etc.; this is essential for characterization of both lethal and sublethal effects of toxicants (Ingersoll et al., 1999). Harpacticoid copepods have therefore been extensively used as test organisms in chronic ecotoxicity testing (e.g. Hutchinson et al., 1999; Chandler et al., 2004; Breitholtz et al., 2003; Breitholtz and Wollenberger, 2003). The reason for using N. psammophila in this study rather than N. spinipes (which is used in standardized toxicity tests) was that population of the former species has been cultured for a shorter period and therefore presumably exhibits higher levels of heterozygosity compared to the latter species that has been kept in the laboratory for more than 30 years, which implies a high level of inbreeding. It was necessary to collect the N. psammophila almost 1 year before the beginning of this study since it takes time to isolate and identify a single species that can handle laboratory conditions. Ovigerous N. psammophila were isolated at Utansj¨o (62◦ 46 N, 17◦ 57 E) on the East coast of Sweden in August 2002. Cultures of this species have since then been maintained in laboratory at 20 ◦ C, 3.5‰ salinity, in darkness, and fed with a mixture (1:2 on a weight basis) of freeze-dried algae (diatom Thalassiosira weissflogi and cryptophyte Rhodomonas baltica). The experiment was conducted between 17th of June to 11th of July in 2003. 2.2. Test substance ˚ Bergman at the Department of Environmental Professor Ake Chemistry, Stockholm University, kindly provided the polybrominated diphenyl ether 2,2 ,4,4 -tetrabromo-diphenyl ether (BDE-47). This flame retardant had a purity of >99% and was chosen as a model substance since it is a common environmental pollutant worldwide (e.g. De Wit, 2002). Additionally, this substance was suggested to cause compensatory effects in populations of N. spinipes (Breitholtz and Wollenberger, 2003). 2.3. Experimental design For the purpose of the present study, the BDE-47 dose range was intended to comprise a high dose (x) that should give significant effects on population dynamics, and two lower doses (i.e. x/10 and x/100) that should not give any apparent effects on the population dynamics. Another prerequisite for our testing was the recommendation by Breitholtz and Wollenberger (2003), who tested the same diphenyl ether for chronic effects, to not normalize effect levels of very hydrophobic substances to their concentration in the water since they partition almost exclusively to non-water phases; only 0.5% was present in filter-(0.3 ␮m)passing test medium after 15 days exposure although 70% of the test medium was renewed every second day. We therefore decided to administer the diphenyl ether as a single dose normalized to the amount of freeze-dried algae. Further, renewal of test medium containing substances that partition to non-water phases will result in increased load of the substance in the test system over time (Breitholtz and Wollenberger, 2003). Hence,

J. Gardestr¨om et al. / Aquatic Toxicology 79 (2006) 41–48

we decided not to change test medium in the present study. Since all available ecotoxicity data from testing with this particular diphenyl ether were based on concentrations normalized to the concentration in the water, comparisons were difficult to make and the setting of the highest dose failed in the sense that all test animals died before day 24. However, in the two lower doses no significant effects on the number of animals were observed as compared to the control. Due to the difficult partitioning behavior of BDE-47 (Breitholtz and Wollenberger, 2003) and since it was only a model substance, we did not measure total BDE47 concentrations or potential metabolites in water, animals or remaining food. About 300 ovigerous females of N. psammophila were isolated from the continuous culture. Sixteen 500-ml spherical glass beakers were divided between one control group (No. BDE-47) and three treatments: 0.11, 1.1 and 11 ␮g BDE-47 mg−1 food; all in four replicates. However, since all animals died in the treatment with 11 ␮g BDE-47 mg−1 food, this treatment was omitted from the analysis. The treatments were prepared by adding freeze-dried algae (6 mg R. baltica and 3 mg T. weissflogi) followed by 25 ␮l acetone stock solution of the diphenyl ether (25 ␮l of pure acetone was added to the controls). When the acetone had evaporated, 100 ml of brackish water (salinity 3.5‰) and 15 randomly chosen ovigerous female N. psammophila from the isolated pool of 300 ovigerous females were added to each beaker. After 24 days (i.e. >1 generation) the following samples were taken from each replicate: (1) 10% of the beaker content were preserved with Lugol’s solution for estimating abundance and demographic structure of the population, (2) five individual copepodites in the third developmental stage (CIII) for quantifying individual RNA content were transferred to Eppendorf tubes containing 20 ␮l of RNAlater and stored at 4 ◦ C until nucleic acid analysis (Gorokhova, 2005) and (3) nine males were preserved in 95% ethanol for analysis of genetic variability after they had been incubated in clean brackish water for 4–6 h to allow for emptying gut content. The number of ovigerous females (i.e. 15) that were introduced to each beaker at the start of the experiment was based on a compromise to: (1) have a sufficient gene pool and (2) keep the population size below the carrying capacity of the test system. The latter prerequisite was tested in a preliminary experiment were the carrying capacity in the test system (i.e. with 100 ml test medium and unlimited access of food) was found to be in the range of 1500–2000 animals. From this experiment, we also determined that 9 mg of freeze-dried micro algae would be sufficient to have exponential growth and that more than one generation would be generated in the control populations over 24 days exposure in the actual experiment. Nauplii, copepodites and adults in a 10% sub-sample taken from each beaker after 24 days were counted using light microscopy after staining with Lugol’s solution. The robustness of the sub-sampling technique was tested in a pilot experiment, in which triplicates of sub-samples (10 vol.% at 22 day of incubation) from three individual beakers showed low mean variability: nauplii, CV% = 12.9, copepodites/adults, CV% = 20.8, and total number of copepods, CV% = 12.9.

43

2.4. Nucleic acid analysis The rationale of using RNA quantity as a measure of condition and growth is that total RNA content is primarily a function of ribosome number and is correlated with protein synthesis, while DNA exists in a quasi-constant quantity in a somatic cell and therefore might be used as an index of the number of cells (Buckley et al., 1999). Quantities of RNA and DNA were measured in individual copepodites CIII; the choice of this stage was based on lowest individual variability for this stage assayed on specimens from the culture (Gorokhova, unpubl.). Microplate fluorometric high-range assay using RiboGreen was performed to quantify RNA and DNA in individual copepodites after extraction with N-laurylsarcosine followed by RNase digestion as described in detail elsewhere (Gorokhova and Kyle, 2002). Measured RNA and DNA concentrations were expressed as ␮g ind−1 . The following working reagents were used: RiboGreenTM RNA Quantitation Kit (Molecular Probes, cat. # R11490); RNA (16S and 23S from E. coli, Component C of the RiboGreen Kit); RNase, DNase-free (Q-biogene, cat. # RNAS0500), working solution 5 ␮g ml−1 ; Nlauroysarcosine (sarcosyl, Sigma, cat. # L-5125); TE buffer (Qbiogene, cat. # TE1X0001). Fluorescence measurements were performed using fluorometer FLUOstar Optima (BMG Labtechnologies, microplate reader, filters: 485 nm for excitation and 520 nm for emission) and black solid flat-bottom microplates (COMBO; Labsystems, cat. # 9502067). The plate was scanned with 0.2 s well measurement time, with 10 measurements per well. 2.5. Genetic analysis DNA was extracted and purified from individual copepods (males) according to Laird et al. (1991). This was done on 5–9 individuals from each replicate beaker (n = 4) from the control and the 1.1 ␮g mg−1 BDE-47 treatment. Copepods are very small and there is a risk of loosing individuals during handling when extracting DNA. This is probably what happened since we did not succeed in extracting DNA from all individuals; hence not all nine individuals could be analyzed in some replicates. The AFLP (amplified fragment length polymorphism) analysis was done according to Vos et al. (1995) with minor modification as described by Bensch et al. (2002). Genomic DNA was digested with EcoRI and MseI, followed by a preamplification step before the selective amplification. Five sets of selective amplification primers were used (EcoRI + TCT/MseI + CTA, EcoRI + TCT/MseI + CCA, EcoRI + TAG/MseI + CCA, EcoRI + CGT/MseI + TAG, EcoRI + CGT/MseI + TCT) with the EcoRI selective primers labeled with flourescein (Life Technologies). The fragments were separated in polyacrylamid gels and electrophoresed at 40 W for 80–90 min. One individual sample was electrophoresed on all gels as an internal allelic standard. The bands were visualized with a Fluoroimager, FLA-3000 from Fujifilm. The bands were scored manually as 1 (present) or 0 (absent), creating a presence/absence matrix. Only bands that were distinct (as absent or present) in all individuals were scored and included in the matrix.

44

J. Gardestr¨om et al. / Aquatic Toxicology 79 (2006) 41–48

2.6. Statistical analysis Differences between groups, i.e. controls and different nominal doses of BDE-47, in population abundance, demographic structure, and RNA–DNA content were compared using oneway ANOVA followed by Dunnett’s post hoc tests (GraphPad Prism 4.01, GraphPad Software, USA). Data were tested for homogeneity of variances using Bartlett’s test. Unless specified otherwise, data are presented as arithmetic means along with standard errors. In all cases significance was accepted when p < 0.05. All data on proportions were arcsine square root transformed prior to statistical analysis. The present/absence matrix was analyzed using the Bayesian program HICKORY v 1.0.3 (Holsinger and Lewis, 2003) to obtain values of expected heterozygosity across loci (Hs) i.e. genetic diversity and the partitioning of the genetic diversity (FST ). It was carried out on the individual level, i.e. using the individuals within the control (n = 28) and the 1.1 ␮g mg−1 BDE-47 treatment (n = 23) respectively as replicate samples. The Bayesian method uses standard Monte Carlo Markov Chain (MCMC) methods to approximate the posterior distribution of θ (refers to FST ), and F (an estimate of the inbreeding parameter FIS ) (Holsinger and Wallace, 2004). It does it without treating dominant markers as haplotypes and it does not assume that genotypes are in Hardy–Weinberg proportions within populations (Holsinger et al., 2002). The sampler parameters were; burn-in (70.000), sampling (100.000), and thinning (5). We used the default prior, beta (1.1) for F, which is equivalent to a uniform (0, 1) non-informative prior. The data were run five times to assure that the results obtained were consistent. The output of the run gave us four different models. We used deviance information criterion, DIC as a model choice criterion, which is described in more detail in Ellison (2004). Data on Hs, θ, and F were taken from the “full model” since this model had the lowest DIC value (as suggested by Holsinger and Lewis, 2003). These data are presented as mean with 95% credible intervals, which are similar to confidence interval in classical statistics (Holsinger and Lewis, 2003). In order to test if the heterozygosity differed between the controls and the high exposure treatment, the mean and 95% credible interval for the difference (Hscontrols − Hsexposed ) was calculated according to Berry (1996). The 95% credible intervals * for the difference was calculated as ±1.95 of the √ the square root sum of the squared standard error i.e. ((S.E. controls)2 + (S.E. high exposure)2 ). Genetic diversity was also described as percent variable bands, calculated as the proportion of all bands that were variable. To evaluate the genetic differentiation (comparing genotype compositions) both within and between the control and the 1.1 ␮g mg−1 BDE-47 treatment respectively, a constrained principal coordinate analysis (cPCoA) was performed using the add-on package VEGAN (Oksanen, 2004) within R 2.0.0 (R Development Core Team 2004). The mean frequency of each AFLP band within each replicate was used when testing the effect of treatment (Fig. 3C), while all AFLP bands from all individuals were used when testing the effect of replication (Fig. 3B). Both tests were done using a permutation test. It was then illustrated graphically with principal coordinate analyze using S plus.

Fig. 1. Population abundance of Nitocra psammophila after 24 days of exposure to 0.11–1.1 ␮g pentabromodiphenyl ether (BDE-47) mg−1 algae and control populations. Life-stages: vertical stripes-nauplii; white-copepodites; horizontal stripes-adults; black-total number of animals. Error bars indicate standard errors of means (n = 4).

3. Results 3.1. Population abundance and demographic structure Abundance of N. psammophila after 24 days did not differ between the controls and the two BDE-47 treatments (Fig. 1). However, there was a significant effect of treatment on population demographic structure: Compared to the controls, the proportion of nauplii decreased (p < 0.05; Table 1), whereas the proportion of copepodites increased (p < 0.05; Table 1). For the proportion of copepodites, the difference between exposed animals and controls were significant in both BDE-47 treatments (p < 0.05). 3.2. RNA-based indices The effect of treatment on individual RNA content was significant (p < 0.01; Fig. 2); the differences between the exposed animals and controls were significant in both BDE-treatments (p < 0.05 and p < 0.01 for 0.1 and 1.1 ␮g mg−1 , respectively). By contrast, both DNA levels and RNA:DNA ratio were not affected (Fig. 2).

Table 1 Proportion of Nitocra psammophila nauplii and copepodites to the total population size after 24 days exposure to 0.11–1.1 ␮g pentabromodiphenyl ether (BDE-47) mg−1 algae Treatment variable

Mean

S.E.

Control Nauplii/total population Copepodites/total population

34.6 27.1

±3.2 ±2.3

0.11 Nauplii/total population Copepodites/total population

21.1 46.4

±3.1 ±3.5

1.1 Nauplii/total population Copepodites/total population

20.6 42.4

±3.6 ±4.5

F

pa

p < 0.05* F2,9 = 4.335 F2,9 = 6.134

p < 0.05* p < 0.05*

All data were arcsine square root normalized prior to statistical analysis. Significant differences from the control are indicated with F- and p-values (* p < 0.05).

J. Gardestr¨om et al. / Aquatic Toxicology 79 (2006) 41–48

45

Fig. 2. Individual RNA–DNA content and RNA:DNA ratios in N. psammophila copepodites CIII after 24 days exposure to 0.11–1.1 mg pentabromodiphenyl ether (BDE-47) mg−1 algae. Asterisks denote significant differences from the controls (* p < 0.05; ** p < 0.01). Error bars indicate standard error of means (n = 4).

3.3. Genetic variability Results of the Hickory analysis (Holsinger and Lewis, 2003) gave a FST (θ) value of 0.095 (95% CI 0.068–0.126). Since the 95% credible interval did not overlap with 0, there were differences in allele frequencies between the controls and the populations in the 1.1 ␮g mg−1 BDE-47 treatment. The estimate of FIS (F) was 0.246 (95% CI 0.070–0.440) and did also deviate from 0, indicating some inbreeding within these two treatments. The value of the expected heterozygosity (Fig. 3A) was 0.384 for the controls (95% CI 0.363–0.403) and 0.339 for the populations in the 1.1 ␮g mg−1 BDE-47 treatment (95% CI 0.310–0.367), producing a difference of 0.045. The 95% credible interval of the difference was 0.0379–0.0521. Since the interval does not include 0, there was a significant difference in heterozygosity between these two treatments. Five AFLP primers generated 129 resolvable bands of which as many as 128 were polymorphic within all populations. The controls had 123 of 129 variable bands (i.e. 95%). This means that, similarly to the heterozygosity pattern, band variability was higher in controls than in the high BDE-47 treatment, which had 93 of 129 variable bands (i.e. 72%). Fig. 3B and C are based on a constrained principal coordinate analysis (cPCoA) and illustrate the genetic composition of the individuals (Fig. 3B) and the replicates (Fig. 3C) within the control and the 1.1 ␮g mg−1 BDE-47 treatment, respectively. Both figures demonstrate that the N. psammophila in the 1.1 ␮g mg−1 BDE-47 treatment and the control were genetically differentiated from each other (F = 2.37, p = 0.05: model 3 in Table 2). The constrained principal coordinate analysis (cPCoA) also showed that the control replicates were not genetically differentiated after 24 days (F = 1.18, p = 0.10; model 1 in Table 2), indicating that no drift had occurred within this treatment. Contrarily, there might have been some drift in the replicates in the 1.1 ␮g mg−1 BDE-47 treatment since these showed genetic differentiations (F = 1.61, p = 0.05; model 2 in Table 2).

Fig. 3. The results in (A–C) are all based on AFLP marker in N. psammophila exposed to seawater (i.e. control; filled circles) and 1.1 ␮g pentabromodiphenyl ether (BDE-47) mg−1 (open triangles) algae for 24 days. (A) Mean heterozygosity; values are given as mean ± 95% credible interval. The asterisk (*) denotes that the 95% credible interval of the difference does not include 0 and is hence significant. (B–C) Constrained principal coordinate analysis of individuals (B) and of replicates (C) within the two treatments.

Table 2 Analysis of the effect of replication and treatment of N. psammophila

1a

Model Model 2b Model 3c Residual 1a Residual 2b Residual 3c

d.f.

Variance

F

3 2 1 20 17 5

10.32 2.95 1.6 58.24 15.52 3.4

1.18 1.61 2.37

No. perm 300 200 10000

p 0.10 0.05 0.05

The three different models (degree of freedom, variance, F, number of permutations, and p-values) and the residuals for respective model (degree of freedom and variance). a Testing no effect of replication within the control. b Testing no effect of replication within the 1.1 ␮g mg−1 BDE-47 treatment. c Testing no effect of treatment.

46

J. Gardestr¨om et al. / Aquatic Toxicology 79 (2006) 41–48

4. Discussion 4.1. Population effects, genetics and RNA Our results corroborate the hypotheses that toxicant exposure can reduce genetic diversity (measured as mean expected heterozygosity and percent polymorphic bands) and cause genetic differentiation in exposed populations. Interestingly, there was no significant decline in total abundance of the BDE-47 treated populations. However, alterations in population structure manifested as increased proportion of older individuals together with lower individual RNA content indicate negative effects at the organism and population levels. The effect of genetic drift is random, which generally makes populations more differentiated. If drift alone caused the genetic differentiation between the controls and the populations in the 1.1 ␮g mg−1 BDE-47 treatment (Fig. 3C) one would expect to find substantial differences between the replicates as well. This was true for the replicates in the 1.1 ␮g mg−1 BDE-47 treatment but not for the controls. Since genetic drift may have a more significant impact on small populations (Street et al., 1998), a plausible explanation to the differentiation in the BDE-47 replicates could be that they had smaller effective population sizes compared to the controls. The effective size of a population (Ne ) is not equivalent to absolute population size. Instead, it is a measure of how many individuals able to carry a particular allele to the next generation (Newman, 2001). Hence, a smaller Ne suggests reduced allele diversity in future generations. The fact that the control replicates in the present study were not genetically differentiated indicates that drift was not per se a problem in our experimental design but was most likely a result of BDE-47 exposure, which suggests that toxic exposure was the causative agent behind the suggested reduced effective population sizes in the 1.1 ␮g mg−1 BDE-47 replicates. Further, there has likely been a selection towards individuals with the right genotype for the stress caused by the BDE-47 treatment. Consequently, these selected individuals contributed to the following generation more substantially as compared to the individuals with a less successful genotype. However, due to the relatively short exposure time, we cannot conclude whether there were individuals that were more effective at reproducing. Since the generation time of N. psammophila at experimental conditions is 15–20 days, the males that were sampled for genetic diversity analysis could only have been from the F1 generation. The decreased heterozygosity and the genetic differentiation observed in the 1.1 ␮g mg−1 BDE-47 treatment (compared to the controls), could consequently not have been an effect of uneven reproduction efficiency since the females from the P generation (15 egg-carrying females/replicate) already were carrying the eggs before the initiation of the treatment. The estimated F is (F) value indicated presence of inbreeding in the experiment. However, since the males that were sampled for genetic analysis were from the F1 generation, inbreeding must have occurred in the laboratory culture, from which the egg-carrying females used to initiate the experiment originated. Consequently, this inbreeding cannot explain the difference in genetic diversity, partitioning of genetic diversity (FST ) between

the populations from the 1.1 ␮g mg−1 BDE-47 treatment and the control or the fact that these populations were genetically diverged from each other. Despite the fact that the N. psammophila had been in the laboratory for almost a year, 24 days of BDE-47 exposure still was enough to cause differences in genetic diversity. However, the genetic diversity in the control, assayed as percentage of polymorphic bands, was high after 24 days. Nevertheless, it is likely that the genetic diversity was even higher in the early laboratory population as compared to the population that the experiment was started with. Street and Montagna (1996) also found quite high level of genetic diversity in harpacticoid copepods sampled from an area proximal to off-shore platforms. The differences in DNA sequence frequencies in populations of harpacticoid copepods have been described as equivalent to interspecific differences in other species of animals (Bucklin et al., 1992; Street and Montagna, 1996) and possible explanations include their high metabolism, short generation time and fast response to disturbance (Street and Montagna, 1996). How then could the total abundance be upheld in the BDE47 treatments? Possibly females with the resistant genotype had higher survival rate and therefore could propagate more successfully than other genotypes. Unfortunately, the experimental design does not allow us to draw any certain conclusions about this. However, the demographic structure of the control and experimental populations could be indicative of birth-, developmental- or mortality-related changes. In particular, the low proportion of nauplii observed in the BDE-47 treatments could reflect ontogenetic differences in sensitivity (e.g. Lotufo and Fleeger, 1997) with higher mortality in nauplii compared to the copepodite stages. On the other hand, exposure could have caused progressively lowering egg production and/or hatching in females that might indicate: (1) early senescence, (2) increased energetic costs for maintenance and consequent allocation of resources to maximize survival (Barata and Baird, 2000), and/or aberrations in embryonic development. Indeed, in the exposed population significantly lower percentage and marginally significant (p > 0.0523) absolute abundance of nauplii indicate decreased birth rates that are likely to result in lower population stocks if the experiment was a few weeks longer. Moreover, the observed low RNA levels in copepodites (i.e. index of somatic growth) suggest that the development rate of copepodites was probably diminished in the BDE-47 treatment leading to the increased proportion of copepodites in the exposed populations and a larger copepodite number (163 versus 107 individuals in control), although the difference was not statistically significant due to the high variability between the replicates. If the development of N. psammophila in the BDE-47 treatments was suppressed compared to the control, then a lower number of F1 in the exposed populations could contribute to the production of F2 nauplii compared to the control. Thus lower nauplii production and/or retarded development might have been the mechanisms for the observed changes in the populations exposed to BDE47 treatment. Although changes in total abundance were not apparent within the experimental period that was less than two generations, alterations in demographic structure suggest lower fitness and productivity in exposed populations. This is likely

J. Gardestr¨om et al. / Aquatic Toxicology 79 (2006) 41–48

to result in population decline after a few generations, which is also supported by the fact that all animals died in the 11 g mg−1 BDE-47 treatment (i.e. the treatment that was excluded from the analyses). The variation in RNA content was unlikely to be related to differences in body size of the animals between the treatments as suggested by relatively constant individual DNA content. In contrast to RNA levels, RNA:DNA ratio was not affected by the treatment. As pointed out by Suthers et al. (1996), the sequential determination of RNA and DNA from the same sample implies that the RNA:DNA ratio depends on the difference in fluorescence between total nucleic acids and DNA to calculate RNA. The numerator is thus dependent on the denominator, and measurement error may be compounded in the ratio, magnifying error and thus potential variability in the index. Our results therefore support the earlier findings suggesting that RNA content alone might be a more sensitive endpoint (Gorokhova and Kyle, 2002; Rosa and Nunes, 2003; Gorokhova, 2005). 4.2. Implications for ERA Environmental risk assessment aims at predicting impacts of anthropogenic chemical substances on ecosystems. Since environmental stress is a multidimensional entity, it can affect various levels of biological organization differently. Therefore, it might be necessary to widen the perspective of the risk assessment process and to elucidate complex and multilevel effects of toxicants. The links between the levels are complicated but crucial, and responses at each level may provide information that helps to unravel the multifaceted relationships between exposure and observed effects. Future ecotoxicological testing should therefore also include biochemical and genetic responses that may be either a cause or a result of toxic effects observed at higher levels of organization. This would increase the sensitivity of this testing and also provide a mechanistic understanding of complex effects at the population level (Belfiore and Anderson, 2001; Escher and Hermens, 2002; Breitholtz et al., 2006). The present study demonstrates some of the advantages of such integrated multilevel risk assessment. All endpoints measured in concert, complemented each other and generated a broader and a more comprehensive picture of the consequences of chemical exposure. For instance, by analyzing population size, we found that abundance did not differ between the controls and the BDE-47 treatments. Hence, the difference in heterozygosity and genotype composition was rather a result of genotype-dependent differences in survival and/or competitive abilities. It has been common among ecologists to use abundance changes as a basis to assess impacts of disturbances. Population genetic change, however, may be more difficult to detect in the sense that alterations are manifested at lower organizational levels (e.g. the genome). Not the least the present study has shown that despite the lack of dramatic responses on abundance following toxic exposure, more subtle effects on the genetic level, may still be present. In conclusion, our results show that the use of the multilevel approach focusing on effects on different levels in the biological organization may improve the predictive capacity of future envi-

47

ronmental risk assessments. However, more work needs to be done to determine if changes in population genetic structure are related to potential losses in population functionality as viewed by the resilience concept (Holling, 1973). This includes capacity to handle (additional) stress, recover from disturbance and adaptive/learning capacity, which are of major concern since the genetic diversity of a population is related to its ability to evolve (Frankham et al., 2002; Futuyma, 1997; Belfiore and Anderson, 2001; Bickham et al., 2000; Van Straalen and Timmermans, 2002). This implies that in wild populations, experiencing environmental fluctuations, decreased genetic diversity may lower phenotypic plasticity and affect their adaptive potential. Since the likelihood of toxic substances to alter the genetic composition of a population, increases when exposure lasts over several generations, a first step to increase our understanding of potential loss of functionality, is to conduct controlled laboratory studies to determine the long-term pollution effects on population genetics (i.e. exposure over multiple generations). We are at present using this approach in experiments with an exposure time of 6–7 generations and with more environmentally realistic exposure scenarios (e.g. natural sediments). Such experiments will hopefully generate information whether also less drastic toxicant exposures over several generations may cause similar genetic effects as observed in the present study. This could in turn provide important information for future ERA of single substances as well as for remedial activities of contaminated sediments. 5. Conclusion The present study shows that toxicant exposure can reduce genetic diversity and alter genotype composition although population abundance remains unaffected. We suggest that multilevel approaches, e.g. including genetic endpoints and populationand fitness-related measures, may be needed to be able to unravel subtle effects on the population level. Such information could improve the scientific basis for chemicals control but perhaps also provide important information for adequate priority setting of environmental remedial activities of polluted sites. Acknowledgements We would like to acknowledge the Swedish scientific programme NewS (A New Strategy for the Risk Management of Chemicals), which is founded by the Swedish Foundation for Strategic Environmental Research (MISTRA), for its financial support to the present project. References Barata, C., Baird, D.J., 2000. Determining the ecotoxicological mode of action of chemicals from measurements made on individuals: results from short-duration chromic tests with Daphnia magna Straus. Aquat. Toxicol. 48, 195–209. Bechmann, R.K., 1994. Use of life-table and LC50 tests to evaluate chronic and acute toxicity effects of copper on the marine copepod Tispe-furcata (Baird). Environ. Toxicol. Chem. 13 (9), 1509–1517. Belfiore, N.M., Anderson, S.L., 2001. Effects of contaminants on genetic patterns in aquatic organisms: a review. Mut. Res. 489 (2/3), 97–122.

48

J. Gardestr¨om et al. / Aquatic Toxicology 79 (2006) 41–48

Bensch, S., Helbig, A.J., Salomon, M., Siebod, I., 2002. Amplified fragment length polymorphism analysis identified hybrids between two subspecies of warblers. Mol. Ecol. 11 (3), 473–481. Berry, D.A., 1996. Statistics. A Bayesian Perspective. Wadsworth Publishing Company, Belmont, California, USA. Bickham, J.W., Sandhu, S., Hebert, P.D.N., Chikhi, L., Athwal, R., 2000. Effects of chemical contaminants on genetic diversity in natural populations: implications for biomonitoring and ecotoxicology. Mut. Res. 463, 33–51. Bradshaw, A.D., McNeilly, T., 1981. Evolution and Pollution. Edward Arnold, London, UK. Breitholtz, M., Rud´en, C., Hansson, S.O., Bengtsson, B.-E., 2006. Ten challenges for improved environmental risk assessment. Ecotoxicol. Environ. Saf. 63, 324–335. Breitholtz, M., Wollenberger, L., Dinan, L., 2003. Effects of four synthetic musks on the life-cycle of the harpacticoid copepod Nitocra spinipes. Aquat. Toxicol. 63 (2), 103–118. Breitholtz, M., Wollenberger, L., 2003. Effects of brominated flame retardants on development, reproduction and population growth rate (rm ) of the harpacticoid copepod Nitocra spinipes. Aquat. Toxicol. 64 (1), 85–96. Buckley, L., Caldarone, E., Ong, T.L., 1999. RNA–DNA ratio and other nucleic acid-based indicators for growth and condition of marine fishes. Hydrobiologica 401, 265–277. Bucklin, A., Frost, B.W., Kocher, T.D., 1992. DNA sequence variation of the mitochondrial 16S rRNA in Calanus (Copepoda: Calanoida): intraspecific and interspecific patterns. Mol. Marine. Biol. Biotechnol: 1, 397– 407. Calow, P., 1996. Ecology in ecotoxicology: some possible ‘rules of thumb, 5–12. In: Baird, D.J., Maltby, L., Greig-Smith, P.W., Douben, P.E.T. (Eds.), ECOtoxicology: Ecological Dimensions. Chapman and Hall, SETAC Press Series, Cambridge, UK, ISBN 0-412-75470-3. Calow, P., Sibly, R.M., Forbes, E.F., 1997. Risk assessment on the basis of simplified life-history scenarios. Environ. Toxicol. Chem. 16 (9), 1983–1989. Chandler, T.G., Cary, T.L., Bejarano, A.C., Pender, J.L., Ferry, J.L., 2004. Population consequences of fipronil and degradates to copepods at field concentrations: an integration of life cycle testing with Leslie matrix population modeling. Environ. Sci. Technol. 38 (23), 6407–6414. Dahlhoff, E.P., 2004. Biochemical indicators of stress and metabolism: applications for marine ecological studies. Annu. Rev. Physiol. 66, 183–207. Devonshire, A.L., Field, L.M., Foster, S.P., Moores, G.D., Williamson, M.S., Blackman, R.L., 1998. The evolution of insecticide resistance in the peach-potato aphid: Myzus persicae. Philos. Trans. R. Soc. Lond. B 353 (1376), 1677–1684. De Wit, C.A., 2002. An overview of brominated flame retardants in the environment. Chemosphere 46 (5), 583–624. Ellison, A.M., 2004. Bayesian inference in ecology. Ecol. Lett. 7, 509–520. Escher, B.I., Hermens, J.M.M., 2002. Modes of action in ecotoxicology: their role in body burdens, species sensitivity, QSARs and mixture effects. Environ. Sci. Technol. 36, 4201–4217. European Commission, 2003. Technical Guidance Document in Support of Commission Directive 93/67/EEC on Risk Assessment for New Notified Substances and Commission Regulation (EC) No. 1488/94 on Risk Assessment for Exiting Substances. European Commission, Luxembourg, ISBN 92-827-8011-2. Forbes, V.F., Calow, P., Sibly, R.M., 2001. Are current species extrapolation models a good basis for ecological risk assessment? Environ. Toxicol. Chem. 20 (2), 442–447. Forbes, E.F., Calow, P., 2002. Population growth rate as a basis for ecological risk assessment of toxic chemicals. Philos. Trans. R. Soc. Lond. B 357, 1299–1306. Frankham, R., Ballou, J.D., Briscoe, D.A., 2002. Introduction to Conservation Genetics. Cambridge University Press, Cambridge, UK, p. 617. Futuyma, D.J., 1997. Evolutionary Biology, 3rd ed. Sinauer Associates, Inc., Sunderland, Massachusetts, p. 763. Gorokhova, E., 2005. Effects of preservation and storage of microcrustaceans in RNA later on RNA and DNA degradation. Limnol. Oceanogr. Meth. 3, 143–148.

Gorokhova, E., Kyle, M., 2002. Analysis of nucleic acids in Daphnia: development of methods and ontogenetic variations in RNA–DNA content. J. Plankt. Res. 24, 511–522. Holling, C.S., 1973. Resilience and stability of ecosystems. Ann. Rev. Ecol. Syst. 4, 1–23. Holsinger, K.E., Lewis, P.O., 2003. A Package for Analysis of Population Genetic Data v1.0. University of Connecticut. Holsinger, K.E., Wallace, L.E., 2004. Bayesian approaches for the analysis of population genetic structure: an example from Platanthera leucophaea (Orchidaceae). Mol. Ecol. 13 (4), 887–894. Holsinger, K.E., Lewis, P.O., Dipak, K.D., 2002. A Bayesian approach to inferring population structure from dominant markers. Mol. Ecol. 11 (35), 1157–1164. Hutchinson, T.H., Pounds, N.A., Hampel, M., Williams, T.D., 1999. Lifecycle studies with marine copepods (Tisbe battagliai) exposed to 20hydroxyecdysone and diethylstilbestrol. Environ. Toxicol. Chem. 18 (12), 2914–2920. Ingersoll, C.G., Hutchinson, T.H., Crane, M., Dodson, S., DeWitt, T., Geis, A., Huet, M.C., McKenney, C.L., Oberd¨orster, E., Pascoe, D., Versteeg, D.J., Warwick, O., 1999. Laboratory tests for evaluating potential effects of endocrine disrupting compounds. In: de Fur, P.J., Crane, M., Ingersoll, C.G., Tattersfield, L.J. (Eds.), Endocrine Disruption in Invertebrates: SETAC Technical Publication. Brussels, Belgium, ISBN 1-880611-27-9, pp. 107–109. Laird, P.W., Zijderveld, A., Linders, K., Rudnick, M.A., Jaenisch, R., Berns, A., 1991. Simplified mammalian DNA isolation procedure. Nucl. Acids Res. 19, 4293. Lotufo, G.R., Fleeger, J.W., 1997. Effects of sediment-associated phenanthrene on survival, development and reproduction of two species of meiobenthic copepods. Mar. Ecol. Prog. Ser. 151 (1–3), 91–102. Ma, X.L., Cowles, D.L., Carter, R.L., 2000. Effects of pollution on genetic diversity in the bay mussel Mytilus galloprovincialis and the acorn barnacle Balanus glandula. Mar. Environ. Res. 50, 559–563. Newman, M.C., 2001. Population Ecotoxicology, vol. 1, Hierarchical Ecotoxicology Mini-series. John Wiley & Sons. Chichester, UK, ISBN 0471988189. Oksanen, J., 2004. Community Ecology Package. Version 1.6-5. URL: http://www.cc.oulu.fi/∼jarioksa/. Rosa, R., Nunes, M.L., 2003. Seasonal changes in nucleic acids, amino acids and protein content in juvenile Norway lobster (Nephrops norvegicus). Mar. Biol. 143, 565–572. Ross, K., Cooper, N., Bidwell, J.R., Elder, J., 2002. Genetic diversity and metal tolerance of two marine species—a comparison between populations from contaminated and reference sites. Mar. Pollut. Bull. 44, 671–679. Staton, J.L., Schizas, N.V., Chandler, G.T., Coull, B.C., Quattro, J.M., 2001. Ecotoxicology and population genetics: the emergence of “phylogeographic and evolutionary ecotoxicology”. Ecotoxicology 10, 217– 222. Street, G.T., Montagna, P.A., 1996. Loss of genetic diversity in Harpacticoida near offshore platforms. Mar. Biol. 126 (2), 271–282. Street, G.T., Lotufo, G.R., Montagna, P.A., Fleeger, J.W., 1998. Reduced genetic diversity on a meiobentic copepod exposed to a xenobiotic. J. Exp. Mar. Biol. Ecol. 222, 93–111. Suthers, I.M., Cleary, J.J., Battaglene, S.C., Evans, R., 1996. Relative RNA content as a measure of condition in larval and juvenile fish. Mar. Freshwat. Res. 47 (2), 301–307. Tanguy, A., Fernandez Castro, N., Marhic, A., Moraga, D., 1999. Effects of an organic pollutant (Tributyltin) on genetic structure in the pacific oyster Crassostrea gigas. Mar. Pollut. Bull. 38 (7), 550–559. Van Straalen, N.M., 2003. Ecotoxicology becomes stress ecology. Environ. Sci. Technol. 1, 324–330. Van Straalen, N.M., Timmermans, M.J.T.N., 2002. Genetic variation in toxicant-stressed populations. An evaluation of the “genetic erosion” hypothesis. Hum. Ecol. Risk Assess. 8 (5), 983–1002. Vos, P., Hogers, R., Bleeker, M., Reijans, M., van de Lee, T., Hornes, M., Frijters, A., Pot, J., Peleman, J., Kuiper, J., 1995. AFLP: a new technique for DNA fingerprinting. Nucl. Acids Res. 23, 4407–4414.