Environmental Toxicology and Chemistry, Vol. 33, No. 4, pp. 930–936, 2014 # 2014 SETAC Printed in the USA
SYNTHETIC ESTROGEN DIRECTLY AFFECTS FISH BIOMASS AND MAY INDIRECTLY DISRUPT AQUATIC FOOD WEBS € PER HALLGREN,*y ALICE NICOLLE,y LARS-ANDERS HANSSON,y CHRISTER BRONMARK ,y LINA NIKOLERIS,z MURTAZA HYDER,x and ANDERS PERSSONy
yDepartment of Biology, Aquatic Ecology, Lund University, Lund, Sweden zCenter for Environmental and Climate Research, Lund University, Lund, Sweden xCenter for Analysis and Synthesis, Chemistry Department, Lund University, Lund, Sweden
(Submitted 18 October 2013; Returned for Revision 2 December 2013; Accepted 30 December 2013) Abstract: Endocrine-disrupting chemicals are known to alter the fitness of individual organisms via changes in growth, behavior, and reproduction. It is largely unknown, however, whether these effects cascade through the food web and indirectly affect other, less sensitive organisms. The authors present results from a mesocosm experiment whereby the effects of the synthetic estrogen 17a-ethinylestradiol (EE2) were quantified in pelagic communities. Treatment with EE2 at a concentration of 28 ng/L had no large effects on the pelagic communities composed only of phytoplankton and zooplankton. In communities where planktivorous roach (Rutilus rutilus) were also present, however, EE2 caused a significant reduction in fish biomass. Moreover, zooplankton biomass was higher in the EE2 treatments, suggesting that zooplankton may have been released from fish predation. Hence, the direct effect of EE2 on roach may have cascaded down the food web to produce positive indirect effects on zooplankton. This result was supported in complementary foraging experiments with roach, showing reduced foraging performance after exposure to EE2. Despite the observed negative effect of EE2 on roach and the positive indirect effect on zooplankton, these effects did not cascade to phytoplankton, possibly because only copepods, but not cladocerans—the major grazers in these systems—were released from fish predation. The authors conclude that the known reproductive impairment in fish by EE2 in combination with the disturbed foraging performance observed in the present study may be a disadvantage to fish that may result in increasing abundance or biomass of prey such as zooplankton. Hence, EE2 may have consequences for both the structure and function of freshwater communities. Environ Toxicol Chem 2014;33:930–936. # 2014 SETAC Keywords: Endocrine-disrupting compounds
Food chain
Fish indices
Zooplankton
17a-Ethinylestradiol
may indirectly affect other organisms that are insensitive to EDCs [8] and cascade down the food chain to affect other trophic levels. This approach has been used with other anthropogenic pollutants, and the direct effects on consumers or resources have been shown to cause indirect effects on other inhabitants in the community [12], as predicted by food web theory [13]. In contrast to anthropogenic pollutants designed to have lethal effects, such as pesticides, EDCs are designed to, or unintentionally are able to, disrupt the endocrine system but rarely have direct lethal effects at the concentrations found in the environment [14]. Chronic, nonlethal effects of EDCs on individuals may well have consequences for the structure and function of populations [6] and communities, however, and theoretical and empirical studies suggest that nonlethal effects may be just as strong as or even stronger than lethal effects when viewed in a community context [13]. Hence, any risk assessment of EDCs needs to consider both the direct and indirect effects those compounds may have in the environment. In general, those EDCs mimicking vertebrate steroid hormones are expected to have more powerful effects on vertebrates than on invertebrates. For example, the direct effect and implications of synthetic estrogens on fish have been addressed in both whole-lake and laboratory experiments. These studies suggest that synthetic estrogens inhibit fish by reducing their reproductive success [15], viability of eggs and hatching success [16], and fertility of male fish [17], and by altering the sex ratio toward more females [18]. Exposure to EDCs could also result in less distinct or frequent male behavior during reproduction, such as reduced aggression and courtship behavior [4]. Kidd et al. [6] showed in a whole-lake experiment that 17a-ethinylestradiol (EE2) resulted in a population collapse of
INTRODUCTION
Aquatic ecosystems across the globe are recipients of many endocrine-disrupting compounds (EDCs), and there is concern that they may affect organisms negatively [1]. This concern has been corroborated in a plethora of studies quantifying the direct effects of EDCs on aquatic organisms, from molecules and tissues [2,3], to behavior [4], to effects on reproduction [5] and population size [6]. Hence, there is broad support for the direct effects of EDCs on the performance of aquatic organisms. Evidence also shows that the direction [7] and magnitude of responses are highly variable [8] and depend on the endocrine system; for example, neuropeptides mainly control reproduction in invertebrates and steroid hormones in vertebrates [9,10]. However, it is largely unknown how and whether the direct effects of EDCs on individual organisms cascade through the food web via indirect interactions. The complexity of natural communities makes the net effect at the community level difficult to predict. Still, it is imperative to gain an understanding of how EDCs affect these food web interactions to predict the impact EDCs may have on the structure and function of natural aquatic communities. One way to address the consequences of EDCs in a community context is to view EDCs as an active component in the food web that can change the density or the behavior of other inhabitants, similar to the effects a predator may have on its prey [11]. Changes in the density or behavior of an organism * Address correspondence to [email protected]. Published online 25 February 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etc.2528 930
EE2 affects fish biomass and aquatic food webs
MATERIALS AND METHODS
Mesocosm setup
The present study was set up at the Lund University (Lund, Sweden) outdoor experimental facility (528420 N, 138120 E) and exposed to the natural temperature and light regime for that location (Figure 2). The mesocosm experiment was performed in 10 barrels (0.74 m in diameter and 1 m high) with 5 replicates each for the control and EE2 treatments. The barrels were aerated and insulated using 50-mm glass wool to avoid large fluctuations in the water temperature. On 27 February 2009 the barrels were filled with 400 L of water from Lake Krankesj€on (558420 3600 N, 138280 2400 E) that had been filtered through a 55mm mesh to start the experiment with phytoplankton and recruitment to the zooplankton community from eggs and small-bodied zooplanktons. The main zooplankton groups in the pelagic zone of Lake Krankesj€on are rotifers, cladocera (Bosmina and Ceriodaphnia), and copepods (calanoida and cyclopoida) that can be divided into size classes of less than 0.2 mm (copepod nauplii included), 0.2 mm to 0.5 mm, and larger than 0.5 mm, respectively [27]. Roach (Rutilus rutilus) were collected as eggs from the Bj€orkaån stream (Scania, Sweden; 558390 2800 N, 138380 5300 E) within the same catchment area as Lake Krankesj€on (Scania, Sweden). The eggs were reared in the laboratory, and 5 roach larvae (length: control
931
P +
_
C
EE2
+
_
_
R Figure 1. Species interaction in a simplified tritrophic food web between predator (P), consumer (C), and resource (R) and the direct and indirect effects of 17a-ethinylestradiol (EE2). It is hypothesized that EE2 has a topdown effect through direct negative effects on juvenile roach (P) foraging performance; indirect positive effects on the zooplankton community (rotifers, cladocera, calanoida and cyclopoida copepods, and copepod nauplii; C) through released predation; and indirect negative effects on phytoplankton (chlorophyll a; R) through increased grazing. Solid lines and dashed lines represent direct and indirect effects, respectively; þrepresents positive effects, and represents negative effects.
9.70 0.25 mm, EE2 treatment 9.60 0.40 mm; analysis of variance [ANOVA] F1,8 ¼ 0.09, p ¼ 0.77) were added to each of the barrels on 27 May (day 85 of the experiment) to mimic hatching of roach larvae in temperate lakes. Roach is a common planktivorous fish that feeds on rotifers and nauplii in their early larvae stage and later on larger zooplankton such as cladocera and copepods [28]. We used 5 roach per barrel to avoid the high densities that would impede growth. Evaporated water was regularly replaced with distilled water, and barrel walls were scrubbed once a week to prevent growth of periphytic algae.
25 20 15
°C
fathead minnow (Pimephales promelas). However, the same experiment also showed that the response depended on the species-specific sensitivity to EE2 and habitat location, as other fish populations did not respond in the same way as the fathead minnow [19]. Whereas fish may be sensitive and respond to estrogen-like EDCs at low concentrations [8], other organisms below fish in the food chain do not respond to EDCs. Both zooplankton and phytoplankton have been found to be insensitive to estrogen-like EDCs except at high concentrations. For example, 46 mg/L EE2 was found to inhibit naupliar development at 10% of the tested population in Acartia tonsa [20,21]. Hence, the direct effects of estrogen-like EDCs are in general expected to be stronger at the top of the food chain and then diminish at lower levels in the food web; but the direct effects can be strong enough to cascade and indirectly affect less sensitive organisms in the food web. In the present study, we present results from a replicated mesocosm experiment assessing the direct and indirect effects of EE2 on a pelagic food web consisting of zooplankton and phytoplankton in both the presence and the absence of juvenile fish. This was done by mimicking the natural succession in temperate lakes of the growth of a plankton community in early spring and later, hatching and growth of a planktivorous fish. Previous studies have mainly focused on the effects of EE2 on reproductive success, but estrogen is an important hormone for a number of physiological processes apart from those related to reproduction, such as locomotion [22] and predatory avoidance behavior [23]. In this experiment, we used juvenile fish to focus not on effects related to the reproductive cycle but rather on the foraging performance of fish. We therefore expected to find stronger direct effects of EE2 at the top of the food web, with the potential that indirect effects could cascade through the food web (Figure 1). We also assessed the direct effects of EE2 on the foraging performance of juvenile fish in a behavioral study. The nominal concentration of EE2, one of the more potent estrogenic EDCs [24], used in the experiments was 50 ng/L, which can be found in sewage effluents and further transported, diluted, and degraded to streams, lakes, and coastal waters [25,26].
Environ Toxicol Chem 33, 2014
10 5 0 0
28
56
70
84
Days
91
105
119
147
Figure 2. Water temperature development in mesocoms experiment as mean standard error for control (solid line) and 17a-ethinylestradiol (EE2 treatment; dashed line).
932
Environ Toxicol Chem 33, 2014
Mesocosm sampling and analysis
The first sampling was done on 3 March for chlorophyll a and zooplankton and later on days 28, 56, 70, 84, 91, 105, and 119, with the last sampling on day 147. We sampled the water by using a plexiglas tube (0.07 m diameter and 1 m long) lowered into the water column at 3 places (2 of them approximately 10 cm from the side and 1 in the middle of the barrel) along the diameter of the barrel. The sampled water (approximately 11 L) was pooled into a bucket. Zooplankton was collected by filtering 10 L of the pooled water through a 55-mm net, and the zooplankton was preserved with Lugol’s solution. We measured the abundance and length for estimation of zooplankton biomass [29] using an inverted microscope for rotifers, cladocera, copepods (calanoida and cyclopoida copepodites and adults), and copepod nauplii. Phytoplankton were sampled by filtering 50 mL of the pooled water through a GF/C filter (Whatman, 25 mm), and the filter was enveloped in aluminum foil and frozen for later analysis. The chlorophyll, used to estimate phytoplankton biomass, was extracted with 3 mL ethanol (96%) [30] and analyzed with a Turner TD 700 fluorometer. After the zooplankton and water samples for chlorophyll a were collected, the sampled water was returned to the barrels. The juvenile roach were collected on 30 July, weighed on a scale (Precisa junior 3100 CD) to the nearest 0.01 g for biomass, and photographed together with a millimeter scale using a Nikon D80 with AF MICRO NIKKOR 105 mm for length measurement to the nearest 0.01 mm and ImageJ software (Ver 1.42q). Mesocosm EE2 treatment and analysis
For the EE2 treatment, EE2 (Sigma-Aldrich, 98% purity) was dissolved in dimethyl sulfoxide (DMSO; Merck, purity 99%) at a concentration of 20 mg/L, and 1 mL of the solution was added after the first sampling of zooplankton and phytoplankton on Monday, 3 March. The EE2–DMSO solution was then added on a weekly basis at a volume of 0.64 mL until the end of the experiment; this volume was based on literature data on the halflife of EE2 in rivers, lakes, and experimental studies [7,31,32]. No DMSO was added to the control, as it was used as control for another experiment [33]. However, the volume of DMSO added to the EE2 treatment was as low as 2.5 mL/L on the first occasion and then later 1.6 mL/L weekly of the water volume, concentrations not toxic to the organisms used in the experiment and less than onetenth of the maximum concentration recommended [34]. Water (1 L) was sampled during the first week of May from each of the replicates at 0 h, 12 h, 24 h, 74 h, and 168 h after EE2 addition to determine the initial exposure and half-life of EE2. The water samples were directly stored in brown glass bottles in the dark at 5 8C until analysis. Concentrations of EE2 were analyzed by using hollow-fiber microporous membrane liquid–liquid extraction combined with gas chromatography–mass spectrometry according to Zorita et al. [35]. Half-life was calculated as t1/2 ¼ ln (2)/k, where k is the slope of the lnC versus time regression. The initial actual water concentration of EE2 was 27.8 0.40 ng/L, and at 168 h the concentration was 13.2 0.22 ng/L and the halflife (t1/2) was calculated at 144 5.8 h. The lower initial concentration may be an effect of partitioning or degradation before the water samples were frozen. Foraging experiment
We performed a laboratory experiment to assess how EE2 exposure affects the feeding rate of roach. Fertilized roach eggs were collected from the same location as the eggs used in the mesocosm experiment and were exposed to 0 ng/L (control) and 50 ng/L EE2 nominal concentration. The eggs were divided into
P. Hallgren et al.
2 different 50-L aquaria per treatment and laid in aquarium net breeders hanging on the inside of the aquaria with the top above the water surface. When the roach larvae had hatched and were actively swimming, they were released into the aquaria. The 2 aquaria for each treatment were connected to a 50-L aerated container in a closed flow-through system. Water was replaced every week, and EE2 was added at a volume of 250 mL from a stock solution of 30 mg/L EE2; the same volume of DMSO was added to the control up to the time of the foraging experiment. The final concentration of DMSO in the aquaria was 1.7 mL/L of water, which is 10 times less than the recommended concentration [34]. During the water change, when the eggs or newly hatched larvae were still in the net breeders, the net breeders were temporarily placed in 2-L aquaria containing water from the 50-L aquaria. When the roach larvae were freely swimming in the aquaria, 5 L of water remained in the aquaria during the water change. No analysis for EE2 was done in the foraging experiment, but the experiment was set up under the same light and temperature conditions (20 8C and 14:10-h light:dark) as a previous experiment done by Hallgren at al. [7], in which EE2 half-life was analyzed and determined to be 54 h. When roach larvae hatched, they were fed with rotifers and nauplii and later with copepods and cladocera. At the size of 33 0.7 mm (12 wk posthatch), 6 roach from each treatment were transferred to individual 2-L aquaria and acclimatized for 12 h before the foraging experiment was begun. At the start of the experiment, 25 Daphnia magna, 0.5 mm to 0.8 mm in length and representing large-bodied cladocera and small-bodied copepods, were added to the 2-L aquaria; after 15 min, the remaining D. magna were counted. Each roach individual was observed 6 times, and 8 h to 12 h separated each observation to standardize the hunger level. Statistics
Data for phytoplankton and zooplankton biomass were analyzed for 2 periods, before and after addition of roach, using repeated-measures ANOVA with time as a within-subject factor for calanoids, cyclopoids, nauplii, cladocera, rotifers, and phytoplankton (chlorophyll a). Control and EE2 treatments were used as between-subject factors. Length of roach when added to the mesocosms as well as length and biomass at the end of the experiment were analyzed using one-way ANOVA. Foraging by roach on D. magna was analyzed using a one-way ANOVA on mean percentage eaten over the 6 trials. Data for biomass and lengths were log-transformed, whereas percentages were arcsine square root–transformed to meet assumptions for homogeneity of variance or normality of distribution. All data are presented as mean 1 standard error (SE), and effects of time and treatment were considered statistically significant when p < 0.05. Ethics statement
Care for the experimental animals in the present study was conducted under a permit provided by the Malm€o/Lund Ethical Committee (M165-07). A permit to collect roach eggs was issued by the Bj€orka-Asumåns fishery conservation area (Scania, Sweden), and collection of water from Lake Krankesj€on is permitted under Swedish law. RESULTS AND DISCUSSION
During the period before roach were added to the mesocosms, there was no significant interaction effect of time and treatment on zooplankton or phytoplankton (Figure 3A–F and Table 1).
EE2 affects fish biomass and aquatic food webs
Environ Toxicol Chem 33, 2014
development, although it is unlikely at the concentration measured in this experiment (i.e., 28 ng/L) and a half-life of 6 d. After addition of roach to the mesocosms, there was a significant treatment effect on roach biomass (ANOVA F1,8 ¼ 5.47, p ¼ 0.05; no mortality was observed; Figure 4A) but not on length (33.5 mm 0.78 mm for control and 32.3 mm 0.48 mm for EE2 treatment: ANOVA, F1,8 ¼ 1.44, p ¼ 0.26). Earlier studies have recorded a reduction in fish growth (both length and weight [36]) as a response to EE2 that may be explained by a modulation of growth hormone levels [37]. In the present study, roach biomass but not length was significantly lower in the EE2 treatment group, possibly suggesting other mechanisms than modulation of growth hormone, as EE2 may inhibit skeletal development that can be coupled to length of the fish [38]. The lower biomass in the EE2-exposed roach may be explained by either lower food intake (e.g., by reduced foraging activity) or food limitation (e. g., lower prey availability). However, the latter is unlikely, as the biomass of calanoids and cyclopoids were higher in the EE2 treatment group (Figure 3A and B and Table 1). Furthermore, the
However, there was a treatment effect with higher nauplii biomass in the control (Figure 3C and Table 1). This suggests a higher rate of copepod reproduction in the control as a result of more food resources, or an inhibition of copepod development or reproduction induced by EE2. There was a higher, but statistically insignificant, biomass of copepods in the control that may have generated a higher biomass of nauplii during the first part of the experiment (Figure 3A and Table 1). There was, however, no significantly higher biomass of phytoplankton (chlorophyll a) or rotifers in the control, which may have boosted copepod reproduction (i.e., generating a bottom-up effect; Figure 3E and F and Table 1). The higher biomass of copepods together with the higher biomass of rotifers may have increased copepod reproduction; because we did not identify nauplii to the species level, however, we cannot say whether the parents of the nauplii were grazers, omnivores, or predators. At a concentration of 46 mg/L, EE2 has been shown to inhibit nauplii development [20]; but this concentration is approximately 2000 times higher compared with the concentration used in the present study. Hence, we cannot disregard this effect on nauplii
A
B
4
mg/L
933
4
Roach Before After
3
3
2
2
1
1
0
0
C
D
4
4
3
3
2
2
1
1
0
0
E
F
4
25
Roach Before After
20
µg/L
3
2
15 10
1
5
0
0 0
20
40
60
80
100
120
140
0
20
40
60
80
100
12 0
140
Days Figure 3. Biomass as mean standard error for (A) calanoids, (B) cyclopoids, (C) nauplii, (D) cladocera, (E) rotifers, and (F) chlorophyll a at sampling points for control (open circles and solid line) and 17a-ethinylestradiol (EE2) treatment (X and dashed line). The dashed vertical line denotes before and after addition of roach larvae.
934
Environ Toxicol Chem 33, 2014
P. Hallgren et al.
Table 1. Summary of results of repeated-measures analysis of variance (ANOVA) on the effects of time and17a-ethinylestradiol (EE2) treatment (with or without EE2 exposure) on biomass of calanoida, cyclopoida nauplii, cladocera, rotifers, and phytoplankton (chlorophyll a) before and after the addition of roach Before roach
After roach
Within-subjects Time treatment
Time
Calanoida Cyclopoida Nauplii Cladocera Rotifera Chlorophyll a
Between-subjects Treatment
Between-subjects
Time treatment
Time
Treatment
F
df
p
F
df
p
F
df
P
F
df
p
F
df
p
F
df
p
27.7 99.0 246 248 16.2 50.4
3.4,27 4,32 2.4,19 2.2,17 1.4,11 4,32
SYNTHETIC ESTROGEN DIRECTLY AFFECTS FISH BIOMASS AND MAY INDIRECTLY DISRUPT AQUATIC FOOD WEBS € PER HALLGREN,*y ALICE NICOLLE,y LARS-ANDERS HANSSON,y CHRISTER BRONMARK ,y LINA NIKOLERIS,z MURTAZA HYDER,x and ANDERS PERSSONy
yDepartment of Biology, Aquatic Ecology, Lund University, Lund, Sweden zCenter for Environmental and Climate Research, Lund University, Lund, Sweden xCenter for Analysis and Synthesis, Chemistry Department, Lund University, Lund, Sweden
(Submitted 18 October 2013; Returned for Revision 2 December 2013; Accepted 30 December 2013) Abstract: Endocrine-disrupting chemicals are known to alter the fitness of individual organisms via changes in growth, behavior, and reproduction. It is largely unknown, however, whether these effects cascade through the food web and indirectly affect other, less sensitive organisms. The authors present results from a mesocosm experiment whereby the effects of the synthetic estrogen 17a-ethinylestradiol (EE2) were quantified in pelagic communities. Treatment with EE2 at a concentration of 28 ng/L had no large effects on the pelagic communities composed only of phytoplankton and zooplankton. In communities where planktivorous roach (Rutilus rutilus) were also present, however, EE2 caused a significant reduction in fish biomass. Moreover, zooplankton biomass was higher in the EE2 treatments, suggesting that zooplankton may have been released from fish predation. Hence, the direct effect of EE2 on roach may have cascaded down the food web to produce positive indirect effects on zooplankton. This result was supported in complementary foraging experiments with roach, showing reduced foraging performance after exposure to EE2. Despite the observed negative effect of EE2 on roach and the positive indirect effect on zooplankton, these effects did not cascade to phytoplankton, possibly because only copepods, but not cladocerans—the major grazers in these systems—were released from fish predation. The authors conclude that the known reproductive impairment in fish by EE2 in combination with the disturbed foraging performance observed in the present study may be a disadvantage to fish that may result in increasing abundance or biomass of prey such as zooplankton. Hence, EE2 may have consequences for both the structure and function of freshwater communities. Environ Toxicol Chem 2014;33:930–936. # 2014 SETAC Keywords: Endocrine-disrupting compounds
Food chain
Fish indices
Zooplankton
17a-Ethinylestradiol
may indirectly affect other organisms that are insensitive to EDCs [8] and cascade down the food chain to affect other trophic levels. This approach has been used with other anthropogenic pollutants, and the direct effects on consumers or resources have been shown to cause indirect effects on other inhabitants in the community [12], as predicted by food web theory [13]. In contrast to anthropogenic pollutants designed to have lethal effects, such as pesticides, EDCs are designed to, or unintentionally are able to, disrupt the endocrine system but rarely have direct lethal effects at the concentrations found in the environment [14]. Chronic, nonlethal effects of EDCs on individuals may well have consequences for the structure and function of populations [6] and communities, however, and theoretical and empirical studies suggest that nonlethal effects may be just as strong as or even stronger than lethal effects when viewed in a community context [13]. Hence, any risk assessment of EDCs needs to consider both the direct and indirect effects those compounds may have in the environment. In general, those EDCs mimicking vertebrate steroid hormones are expected to have more powerful effects on vertebrates than on invertebrates. For example, the direct effect and implications of synthetic estrogens on fish have been addressed in both whole-lake and laboratory experiments. These studies suggest that synthetic estrogens inhibit fish by reducing their reproductive success [15], viability of eggs and hatching success [16], and fertility of male fish [17], and by altering the sex ratio toward more females [18]. Exposure to EDCs could also result in less distinct or frequent male behavior during reproduction, such as reduced aggression and courtship behavior [4]. Kidd et al. [6] showed in a whole-lake experiment that 17a-ethinylestradiol (EE2) resulted in a population collapse of
INTRODUCTION
Aquatic ecosystems across the globe are recipients of many endocrine-disrupting compounds (EDCs), and there is concern that they may affect organisms negatively [1]. This concern has been corroborated in a plethora of studies quantifying the direct effects of EDCs on aquatic organisms, from molecules and tissues [2,3], to behavior [4], to effects on reproduction [5] and population size [6]. Hence, there is broad support for the direct effects of EDCs on the performance of aquatic organisms. Evidence also shows that the direction [7] and magnitude of responses are highly variable [8] and depend on the endocrine system; for example, neuropeptides mainly control reproduction in invertebrates and steroid hormones in vertebrates [9,10]. However, it is largely unknown how and whether the direct effects of EDCs on individual organisms cascade through the food web via indirect interactions. The complexity of natural communities makes the net effect at the community level difficult to predict. Still, it is imperative to gain an understanding of how EDCs affect these food web interactions to predict the impact EDCs may have on the structure and function of natural aquatic communities. One way to address the consequences of EDCs in a community context is to view EDCs as an active component in the food web that can change the density or the behavior of other inhabitants, similar to the effects a predator may have on its prey [11]. Changes in the density or behavior of an organism * Address correspondence to [email protected]. Published online 25 February 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etc.2528 930
EE2 affects fish biomass and aquatic food webs
MATERIALS AND METHODS
Mesocosm setup
The present study was set up at the Lund University (Lund, Sweden) outdoor experimental facility (528420 N, 138120 E) and exposed to the natural temperature and light regime for that location (Figure 2). The mesocosm experiment was performed in 10 barrels (0.74 m in diameter and 1 m high) with 5 replicates each for the control and EE2 treatments. The barrels were aerated and insulated using 50-mm glass wool to avoid large fluctuations in the water temperature. On 27 February 2009 the barrels were filled with 400 L of water from Lake Krankesj€on (558420 3600 N, 138280 2400 E) that had been filtered through a 55mm mesh to start the experiment with phytoplankton and recruitment to the zooplankton community from eggs and small-bodied zooplanktons. The main zooplankton groups in the pelagic zone of Lake Krankesj€on are rotifers, cladocera (Bosmina and Ceriodaphnia), and copepods (calanoida and cyclopoida) that can be divided into size classes of less than 0.2 mm (copepod nauplii included), 0.2 mm to 0.5 mm, and larger than 0.5 mm, respectively [27]. Roach (Rutilus rutilus) were collected as eggs from the Bj€orkaån stream (Scania, Sweden; 558390 2800 N, 138380 5300 E) within the same catchment area as Lake Krankesj€on (Scania, Sweden). The eggs were reared in the laboratory, and 5 roach larvae (length: control
931
P +
_
C
EE2
+
_
_
R Figure 1. Species interaction in a simplified tritrophic food web between predator (P), consumer (C), and resource (R) and the direct and indirect effects of 17a-ethinylestradiol (EE2). It is hypothesized that EE2 has a topdown effect through direct negative effects on juvenile roach (P) foraging performance; indirect positive effects on the zooplankton community (rotifers, cladocera, calanoida and cyclopoida copepods, and copepod nauplii; C) through released predation; and indirect negative effects on phytoplankton (chlorophyll a; R) through increased grazing. Solid lines and dashed lines represent direct and indirect effects, respectively; þrepresents positive effects, and represents negative effects.
9.70 0.25 mm, EE2 treatment 9.60 0.40 mm; analysis of variance [ANOVA] F1,8 ¼ 0.09, p ¼ 0.77) were added to each of the barrels on 27 May (day 85 of the experiment) to mimic hatching of roach larvae in temperate lakes. Roach is a common planktivorous fish that feeds on rotifers and nauplii in their early larvae stage and later on larger zooplankton such as cladocera and copepods [28]. We used 5 roach per barrel to avoid the high densities that would impede growth. Evaporated water was regularly replaced with distilled water, and barrel walls were scrubbed once a week to prevent growth of periphytic algae.
25 20 15
°C
fathead minnow (Pimephales promelas). However, the same experiment also showed that the response depended on the species-specific sensitivity to EE2 and habitat location, as other fish populations did not respond in the same way as the fathead minnow [19]. Whereas fish may be sensitive and respond to estrogen-like EDCs at low concentrations [8], other organisms below fish in the food chain do not respond to EDCs. Both zooplankton and phytoplankton have been found to be insensitive to estrogen-like EDCs except at high concentrations. For example, 46 mg/L EE2 was found to inhibit naupliar development at 10% of the tested population in Acartia tonsa [20,21]. Hence, the direct effects of estrogen-like EDCs are in general expected to be stronger at the top of the food chain and then diminish at lower levels in the food web; but the direct effects can be strong enough to cascade and indirectly affect less sensitive organisms in the food web. In the present study, we present results from a replicated mesocosm experiment assessing the direct and indirect effects of EE2 on a pelagic food web consisting of zooplankton and phytoplankton in both the presence and the absence of juvenile fish. This was done by mimicking the natural succession in temperate lakes of the growth of a plankton community in early spring and later, hatching and growth of a planktivorous fish. Previous studies have mainly focused on the effects of EE2 on reproductive success, but estrogen is an important hormone for a number of physiological processes apart from those related to reproduction, such as locomotion [22] and predatory avoidance behavior [23]. In this experiment, we used juvenile fish to focus not on effects related to the reproductive cycle but rather on the foraging performance of fish. We therefore expected to find stronger direct effects of EE2 at the top of the food web, with the potential that indirect effects could cascade through the food web (Figure 1). We also assessed the direct effects of EE2 on the foraging performance of juvenile fish in a behavioral study. The nominal concentration of EE2, one of the more potent estrogenic EDCs [24], used in the experiments was 50 ng/L, which can be found in sewage effluents and further transported, diluted, and degraded to streams, lakes, and coastal waters [25,26].
Environ Toxicol Chem 33, 2014
10 5 0 0
28
56
70
84
Days
91
105
119
147
Figure 2. Water temperature development in mesocoms experiment as mean standard error for control (solid line) and 17a-ethinylestradiol (EE2 treatment; dashed line).
932
Environ Toxicol Chem 33, 2014
Mesocosm sampling and analysis
The first sampling was done on 3 March for chlorophyll a and zooplankton and later on days 28, 56, 70, 84, 91, 105, and 119, with the last sampling on day 147. We sampled the water by using a plexiglas tube (0.07 m diameter and 1 m long) lowered into the water column at 3 places (2 of them approximately 10 cm from the side and 1 in the middle of the barrel) along the diameter of the barrel. The sampled water (approximately 11 L) was pooled into a bucket. Zooplankton was collected by filtering 10 L of the pooled water through a 55-mm net, and the zooplankton was preserved with Lugol’s solution. We measured the abundance and length for estimation of zooplankton biomass [29] using an inverted microscope for rotifers, cladocera, copepods (calanoida and cyclopoida copepodites and adults), and copepod nauplii. Phytoplankton were sampled by filtering 50 mL of the pooled water through a GF/C filter (Whatman, 25 mm), and the filter was enveloped in aluminum foil and frozen for later analysis. The chlorophyll, used to estimate phytoplankton biomass, was extracted with 3 mL ethanol (96%) [30] and analyzed with a Turner TD 700 fluorometer. After the zooplankton and water samples for chlorophyll a were collected, the sampled water was returned to the barrels. The juvenile roach were collected on 30 July, weighed on a scale (Precisa junior 3100 CD) to the nearest 0.01 g for biomass, and photographed together with a millimeter scale using a Nikon D80 with AF MICRO NIKKOR 105 mm for length measurement to the nearest 0.01 mm and ImageJ software (Ver 1.42q). Mesocosm EE2 treatment and analysis
For the EE2 treatment, EE2 (Sigma-Aldrich, 98% purity) was dissolved in dimethyl sulfoxide (DMSO; Merck, purity 99%) at a concentration of 20 mg/L, and 1 mL of the solution was added after the first sampling of zooplankton and phytoplankton on Monday, 3 March. The EE2–DMSO solution was then added on a weekly basis at a volume of 0.64 mL until the end of the experiment; this volume was based on literature data on the halflife of EE2 in rivers, lakes, and experimental studies [7,31,32]. No DMSO was added to the control, as it was used as control for another experiment [33]. However, the volume of DMSO added to the EE2 treatment was as low as 2.5 mL/L on the first occasion and then later 1.6 mL/L weekly of the water volume, concentrations not toxic to the organisms used in the experiment and less than onetenth of the maximum concentration recommended [34]. Water (1 L) was sampled during the first week of May from each of the replicates at 0 h, 12 h, 24 h, 74 h, and 168 h after EE2 addition to determine the initial exposure and half-life of EE2. The water samples were directly stored in brown glass bottles in the dark at 5 8C until analysis. Concentrations of EE2 were analyzed by using hollow-fiber microporous membrane liquid–liquid extraction combined with gas chromatography–mass spectrometry according to Zorita et al. [35]. Half-life was calculated as t1/2 ¼ ln (2)/k, where k is the slope of the lnC versus time regression. The initial actual water concentration of EE2 was 27.8 0.40 ng/L, and at 168 h the concentration was 13.2 0.22 ng/L and the halflife (t1/2) was calculated at 144 5.8 h. The lower initial concentration may be an effect of partitioning or degradation before the water samples were frozen. Foraging experiment
We performed a laboratory experiment to assess how EE2 exposure affects the feeding rate of roach. Fertilized roach eggs were collected from the same location as the eggs used in the mesocosm experiment and were exposed to 0 ng/L (control) and 50 ng/L EE2 nominal concentration. The eggs were divided into
P. Hallgren et al.
2 different 50-L aquaria per treatment and laid in aquarium net breeders hanging on the inside of the aquaria with the top above the water surface. When the roach larvae had hatched and were actively swimming, they were released into the aquaria. The 2 aquaria for each treatment were connected to a 50-L aerated container in a closed flow-through system. Water was replaced every week, and EE2 was added at a volume of 250 mL from a stock solution of 30 mg/L EE2; the same volume of DMSO was added to the control up to the time of the foraging experiment. The final concentration of DMSO in the aquaria was 1.7 mL/L of water, which is 10 times less than the recommended concentration [34]. During the water change, when the eggs or newly hatched larvae were still in the net breeders, the net breeders were temporarily placed in 2-L aquaria containing water from the 50-L aquaria. When the roach larvae were freely swimming in the aquaria, 5 L of water remained in the aquaria during the water change. No analysis for EE2 was done in the foraging experiment, but the experiment was set up under the same light and temperature conditions (20 8C and 14:10-h light:dark) as a previous experiment done by Hallgren at al. [7], in which EE2 half-life was analyzed and determined to be 54 h. When roach larvae hatched, they were fed with rotifers and nauplii and later with copepods and cladocera. At the size of 33 0.7 mm (12 wk posthatch), 6 roach from each treatment were transferred to individual 2-L aquaria and acclimatized for 12 h before the foraging experiment was begun. At the start of the experiment, 25 Daphnia magna, 0.5 mm to 0.8 mm in length and representing large-bodied cladocera and small-bodied copepods, were added to the 2-L aquaria; after 15 min, the remaining D. magna were counted. Each roach individual was observed 6 times, and 8 h to 12 h separated each observation to standardize the hunger level. Statistics
Data for phytoplankton and zooplankton biomass were analyzed for 2 periods, before and after addition of roach, using repeated-measures ANOVA with time as a within-subject factor for calanoids, cyclopoids, nauplii, cladocera, rotifers, and phytoplankton (chlorophyll a). Control and EE2 treatments were used as between-subject factors. Length of roach when added to the mesocosms as well as length and biomass at the end of the experiment were analyzed using one-way ANOVA. Foraging by roach on D. magna was analyzed using a one-way ANOVA on mean percentage eaten over the 6 trials. Data for biomass and lengths were log-transformed, whereas percentages were arcsine square root–transformed to meet assumptions for homogeneity of variance or normality of distribution. All data are presented as mean 1 standard error (SE), and effects of time and treatment were considered statistically significant when p < 0.05. Ethics statement
Care for the experimental animals in the present study was conducted under a permit provided by the Malm€o/Lund Ethical Committee (M165-07). A permit to collect roach eggs was issued by the Bj€orka-Asumåns fishery conservation area (Scania, Sweden), and collection of water from Lake Krankesj€on is permitted under Swedish law. RESULTS AND DISCUSSION
During the period before roach were added to the mesocosms, there was no significant interaction effect of time and treatment on zooplankton or phytoplankton (Figure 3A–F and Table 1).
EE2 affects fish biomass and aquatic food webs
Environ Toxicol Chem 33, 2014
development, although it is unlikely at the concentration measured in this experiment (i.e., 28 ng/L) and a half-life of 6 d. After addition of roach to the mesocosms, there was a significant treatment effect on roach biomass (ANOVA F1,8 ¼ 5.47, p ¼ 0.05; no mortality was observed; Figure 4A) but not on length (33.5 mm 0.78 mm for control and 32.3 mm 0.48 mm for EE2 treatment: ANOVA, F1,8 ¼ 1.44, p ¼ 0.26). Earlier studies have recorded a reduction in fish growth (both length and weight [36]) as a response to EE2 that may be explained by a modulation of growth hormone levels [37]. In the present study, roach biomass but not length was significantly lower in the EE2 treatment group, possibly suggesting other mechanisms than modulation of growth hormone, as EE2 may inhibit skeletal development that can be coupled to length of the fish [38]. The lower biomass in the EE2-exposed roach may be explained by either lower food intake (e.g., by reduced foraging activity) or food limitation (e. g., lower prey availability). However, the latter is unlikely, as the biomass of calanoids and cyclopoids were higher in the EE2 treatment group (Figure 3A and B and Table 1). Furthermore, the
However, there was a treatment effect with higher nauplii biomass in the control (Figure 3C and Table 1). This suggests a higher rate of copepod reproduction in the control as a result of more food resources, or an inhibition of copepod development or reproduction induced by EE2. There was a higher, but statistically insignificant, biomass of copepods in the control that may have generated a higher biomass of nauplii during the first part of the experiment (Figure 3A and Table 1). There was, however, no significantly higher biomass of phytoplankton (chlorophyll a) or rotifers in the control, which may have boosted copepod reproduction (i.e., generating a bottom-up effect; Figure 3E and F and Table 1). The higher biomass of copepods together with the higher biomass of rotifers may have increased copepod reproduction; because we did not identify nauplii to the species level, however, we cannot say whether the parents of the nauplii were grazers, omnivores, or predators. At a concentration of 46 mg/L, EE2 has been shown to inhibit nauplii development [20]; but this concentration is approximately 2000 times higher compared with the concentration used in the present study. Hence, we cannot disregard this effect on nauplii
A
B
4
mg/L
933
4
Roach Before After
3
3
2
2
1
1
0
0
C
D
4
4
3
3
2
2
1
1
0
0
E
F
4
25
Roach Before After
20
µg/L
3
2
15 10
1
5
0
0 0
20
40
60
80
100
120
140
0
20
40
60
80
100
12 0
140
Days Figure 3. Biomass as mean standard error for (A) calanoids, (B) cyclopoids, (C) nauplii, (D) cladocera, (E) rotifers, and (F) chlorophyll a at sampling points for control (open circles and solid line) and 17a-ethinylestradiol (EE2) treatment (X and dashed line). The dashed vertical line denotes before and after addition of roach larvae.
934
Environ Toxicol Chem 33, 2014
P. Hallgren et al.
Table 1. Summary of results of repeated-measures analysis of variance (ANOVA) on the effects of time and17a-ethinylestradiol (EE2) treatment (with or without EE2 exposure) on biomass of calanoida, cyclopoida nauplii, cladocera, rotifers, and phytoplankton (chlorophyll a) before and after the addition of roach Before roach
After roach
Within-subjects Time treatment
Time
Calanoida Cyclopoida Nauplii Cladocera Rotifera Chlorophyll a
Between-subjects Treatment
Between-subjects
Time treatment
Time
Treatment
F
df
p
F
df
p
F
df
P
F
df
p
F
df
p
F
df
p
27.7 99.0 246 248 16.2 50.4
3.4,27 4,32 2.4,19 2.2,17 1.4,11 4,32
