- Email: [email protected]
The effect of mercury on the feeding behavior of fathead minnows (Pimephales promelas)
Ecotoxicology and Environmental Safety 55 (2003) 187–198
The effect of mercury on the feeding behavior of fathead minnows (Pimephales promelas) M.A. Grippo* and A.G. Heath Department of Biology, Virginia Polytechnic Institute and State University, Blacksburg, VA 24060, USA Received 7 January 2002; received in revised form 26 September 2002; accepted 4 October 2002
Abstract Fathead minnows (Pimephales promelas) were exposed to mercury (1.69, 6.79, and 13. 57 mg/L HgCl2, 10 days exposure), and afterward their foraging ability was tested in a vegetated habitat for 7 days. Among the foraging metrics used were foraging efficiency, capture speed, and the ability to learn and retain information regarding habitat characteristics. In addition to behavioral tests, muscle tissue acetylcholinesterase activity and brain levels of several neurotransmitters were investigated. Comparisons with control fish and fish from the two highest exposure groups revealed consistent performance deficits in foraging efficiency and capture speed. However, no treatment effects on learning were detected, nor were differences in neurotransmitter levels detected. In determining the underlying proximate cause of the foraging deficits, it is believed that the greater pause time exhibited by treatment fish while foraging was the main cause of treatment differences. In the future, behavioral studies will continue to allow toxicity testing of environmentally relevant variables such as those used by behavioral ecologists. r 2003 Elsevier Science (USA). All rights reserved. Keywords: Mercury; Foraging behavior; Learning; Fathead minnows; Neurotransmitters
1. Introduction Ecotoxicology recognizes that within aquatic systems there are multiple interacting levels of organization, ranging from the cellular, to the individual, to the ecosystem level. At the level of the individual, sublethal toxicant effects are manifested in several ways, but among the most sensitive indicators of pollution stress are behavioral alterations. In particular, feeding behavior has been investigated extensively because feeding impairment could have repercussions on an individual’s survivorship, growth, and reproductive capacity. Much previous work has focused on appetite, foraging efficiency, and predator avoidance (reviewed in Beitinger, 1990; Little et al., 1993). More recent research on foraging behavior has often been in the form of mechanistic studies in which a toxicant’s interaction with a specific component of food acquisition is examined (Atchison et al., 1996). The mechanistic parameters of interest are the same as those used by *Corresponding author. CZR Inc., 4709 College Acres Drive, Suite 2, Wilmington, NC, 28403, USA. E-mail address: [email protected] (M.A. Grippo).
behavioral ecologists (e.g., food consumption rate, feeding attempts, capture efficiency, handling time, reaction distance), and all are potentially altered by toxicants (Atchison et al., 1996). Although behavioral toxicologists have investigated several ways in which pollutants can affect foraging behavior, other potential routes remain unexamined. One example is learning and foraging behavior. Numerous studies have confirmed the importance of experience to improvement in foraging ability (reviewed in Hughes et al., 1992). Learning is also valuable for choosing among habitats with differing food values. By choosing the most profitable habitat, fish not only will increase food consumption per unit of foraging time but also will reduce time and energy traveling between patches (Hughes et al., 1992). Furthermore, studies have shown that contaminants can inhibit a variety of learned responses (Strickler-Shaw and Taylor, 1991; Sun and Taylor, 1983). However, few experiments have examined the impairment of ecologically relevant learned behaviors such as those behaviors related to foraging efficiency, even though such investigations would provide insight into the subtle ways in which pollution can harm ecosystems.
0147-6513/03/$ - see front matter r 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0147-6513(02)00071-4
188
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
Despite the relevance to ecotoxicology, aspects of foraging behavior well studied by behavioral ecologists, such as learning, optimal foraging, and prey and habitat choice, are relatively unexplored by toxicologists. Therefore, the objective of the current study was to investigate behavioral alterations in toxicant-exposed fish using two foraging experiments that test prey search strategy, prey location efficiency, and habitat choice. First, an observational study was made comparing, through time, the foraging ability of toxicant-exposed and -unexposed fish introduced to a novel habitat. Also, although fish species may have characteristic search patterns (e.g., cruisers, salutatory searchers), experience may modify that pattern (e.g., changes in hover time, distance moved between hovers) based on habitat structure (Ehlinger, 1989) or prey type (O’ Brien et al., 1989). Because of the potential importance of these search strategies in maximizing foraging return, search behavior was also characterized. It was hypothesized that exposed fish would exhibit deficits in daily measurements of foraging ability compared with control fish, that treatment fish would show little improvement in foraging success over trial days, and that performance deficits would be related to differences in search behavior. Second, experience influences not only foraging success but also habitat choice. Fish are capable of determining, over time, the relative profitability of different habitats and choosing among them based on this learned information (Ehlinger, 1990; Gotceitas, 1990). Consequently, a second experiment examined whether contaminants inhibit or retard the ability of fish to learn and retain information about relative habitat value. Specifically, it was hypothesized that during foraging trials the treatment fish would show no preference between patches of high and low food value and that, based on various metrics of retention, exposed fish would show reduced retention of habitat information across trial days. Mercury (Hg) was chosen as the toxicant for several reasons. First, Hg is highly neurotoxic. In mammals, mercury commonly passes the blood–brain barrier, causing destruction of visual and hippocampal cells, and disrupts the neurons of the peripheral nervous system (Chang, 1979). Also, Hg is a common aquatic pollutant that bioaccumulates in fish through dietary and waterborne exposure (Wiener and Spry, 1996). Finally, a variety of fish behaviors are influenced by neurotransmitters (reviewed in Weber and Spieler, 1994), making toxicant–neurotransmitter interaction an important area of study. Although researchers have examined the effect of organic and inorganic pollutants on neurotransmitters, few have looked at behavior as well. Therefore, the effect of Hg on several neurotransmitters was investigated to determine a correlation among exposure, neurotransmitter levels,
and behavioral alterations. Specifically, acetylcholinesterase (AChE) activity was measured in axial muscle, while norepinephrine (NE), serotonin (5-HT), dopamine (DA), and its metabolite l-DOPA were measured in brain tissue. All of these neurotransmitters are involved variously with locomotion, conditioned responses, and feeding (reviewed in Smith, 1984), and AChE activity is a commonly used biomarker of pollution stress.
2. Materials and methods 2.1. General 2.1.1. Test organisms Fathead minnows (Pimephales promelas) were used as test organisms because they have a wide distribution throughout the midwestern states and are a common toxicological test organism. Fathead minnows are characterized as benthic and mid-level water column feeders (Jenkins and Burkehead, 1994). Juveniles (3.5– 4.0 cm) were obtained from Aquatox (Arizona, USA) and Aquatic Biosystems Inc. (Colorado, USA). Prior to experiments, they were held in Minnow Cool tanks and fed trout chow. Blackworms (Tubificidae spp) from a local pet store were used as prey items during experimental trials. 2.1.2. Toxicant Fish were exposed to mercury (1.69, 6.79, or 13.57 mg/ L) as mercuric chloride (HgCl2) using static systems (28-L aquarium), with all aquariums receiving a daily 50% water change. 2.2. Foraging trials 2.2.1. Protocol for Experiment 1 An entire 20-day run entailed an initial exposure to one of the three HgCl2 treatments in 28-L holding tanks for 10 days and then transfer to noncontaminated aquariums at which time the 7-day foraging trials began. Foraging trials were followed by a 4-day, posttrial appetite study. A fourth group was unexposed and served as the control. Four total runs (blocks) were made. Each experimental run used a completely randomized design with two tanks (replicates) per treatment (eight tanks total). Within each holding tank, 3 fish (subsamples) were housed singly in three separate, gated removable compartments for a total of 24 fish (6 fish/sample, three treatments and a control). The treatment and control groups used in the trials are referred to by their exposure concentration (T1, T6, T13, or control). Also, each holding tank held 2 additional fish, housed separately. These fish, referred to as ‘‘nonexperimental fish,’’ were used for Hg body
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
burden and biochemistry tests immediately after the initial 10-day exposure and did not undergo the training or foraging experiments mentioned below. The purpose of the nonexperimental fish was to allow a comparison between fish immediately after HgCl2 exposure and after the 10-day trial period in which the fish had been held in noncontaminated water. Before the foraging trials began, it was necessary for each fish to be trained to readily enter the feeding area of the experimental aquarium. Therefore, during the initial 10-day exposure period, each fish was individually removed from its holding tank (it remained in its compartment) and transferred to the experimental aquarium, where the gate was removed and the fish was allowed to feed on 12 dead unburied blackworms for 5 min. The feeding area contained no artificial vegetation, just sand. At the end of the 5 min, the fish was led back into the removable compartment and placed in its holding tank. Daily worm consumption was recorded for each fish, yielding information on the appetite effects, if any, of mercury. After the 10-day exposure, the nonexperimental fish were frozen in liquid nitrogen and stored at 701C for biochemical analysis. The 24 trained fish were transferred to clean water, and the foraging trials were initiated. The daily experimental trials began by equally distributing 12 dead, partially buried blackworms (burying increased the difficulty of prey encounter and lessened visual cues) in the vegetated habitat (Fig. 1a). The habitat consisted of one large patch of artificial vegetation (35 42 cm; feeding area 13 cm in depth). The artificial vegetation consisted of 9-cm plastic sticks attached to a plastic mesh plate, which was covered in white dolomite sand. A single fish was then placed in the experimental aquarium and was allowed entrance to the feeding area for 5 min. All remaining prey were counted after completion of each trial. This procedure was repeated for all fish (24 trials per day) for 7 days. The order in which fish fed was randomized daily. After 7 days, all fish underwent another appetite study (4 days) using the protocol from the first 10 days (fish continued to be held in clean water). All trials were video-recorded. From video analysis, the number and time of prey capture were recorded, as were the number and duration of pauses. Also, to characterize movement during the trial, the video monitor screen was divided evenly into quadrants and the number of visits to each quadrant was recorded. From these data, foraging efficiency over time could be monitored along with differences in movement/search strategy. All fish were initially naive about the habitat structure and prey placement. Therefore, improvement in foraging efficiency for each group of fish could be compared over time.
189
Fig. 1. Diagram of the experimental feeding habitats for Experiment 1(a) and Experiment 2(b).
2.2.2. Protocol for Experiment 2 The foraging trials using the second habitat were designed to test the ability of treatment fish to learn the relative food availability of each patch. This series of experiments consisted of two runs (blocks), with each block consisting of a completely randomized design with two replications in the first block and three replications in the second block. The same four treatment exposure concentrations that were used in Experiment 1 were used in Experiment 2. Fish were exposed for 10 days and simultaneously trained to enter the feeding area in the manner described for Experiment 1 (no vegetation in
190
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
feeding area), with the exceptions that fish were housed and allowed to feed in groups of 3, and that 36 unburied worms were placed in the feeding area rather than 12. Following the 10-day exposure, fish were transferred to clean water and the foraging trials were initiated. The trials began by partially burying 22 blackworms in one patch (high food density) and partially buying 4 in the other (low food density). The habitat contained two patches of equal size (35 12 cm) and density of artificial vegetation separated by a sandy area (Fig. 1b). Fish were allowed to forage for 10 min to ensure that the entire habitat could be sampled. This procedure was repeated for all treatment tanks for 6 days. On the 7th day, the relative food values of each patch were switched and the fish were monitored for changes in habitat use. Hence, the first 6 days are referred to as the preswitch period, and the time period after the switch is referred to as are the postswitch period. Relative food values were switched so as to determine how fish from different treatments responded to changes in environmental conditions. As in Experiment 1, the order in which the fish fed was randomized daily. From video analysis, the following data were recorded: the total time that fish were found in each patch, the first patch sampled, and the number of visits to and exits from each patch. Using the time spent in a patch, inferences were made about relative patch preference. The first choice and visit data were used to determine whether the habitat information learned from previous trials was retained. For example, over the 6-day trial period, an increase in the time spent in the profitable patch was considered indicative of learning. Similarly, because repeated visits to the low food patch would be an inefficient use of time, a decrease in visits to the less profitable habitat was also considered evidence of learning. Entering the high food patch first and not visiting the low food patch was considered evidence of retention. 2.3. Biochemical analysis For Experiments 1 and 2, at the end of each run, experimental fish were frozen in liquid nitrogen. After thawing, brains were removed for neurotransmitter determinations. Samples of muscle tissue were also taken for use in the determination of Hg body burden and AChE activity. All samples were stored at 701C until analysis. 2.3.1. Mercury analytical methods The mercury species was mercuric chloride (0.1 M) obtained from Fisher Scientific. Holding tank Hg levels were measured periodically (approximately every other day) using a Perkin–Elmer MA50 cold vapor atomic absorption spectrophotometer (CVAAS) as described in Environmental Protection Agency (EPA) method 7470
(US EPA, 1990). Mercury body burden was determined by the following: digestion of tissue samples in 5 ml of concentrated H2SO4 and 2.5 ml of concentrated HNO3 for 1 h in a 581C water bath, followed by the addition of 15 ml of KPO4. The samples were then left overnight. Before reading the samples, NaCl hydroxylamine sulfate was added, followed by 5 ml of stannous chloride. The samples were immediately analyzed by the CVAAS. For each batch of 9 samples, a tissue spike and spike blanks were prepared and analyzed for percentage recovery. 2.3.2. Acetylcholinesterase activity Weighed tissue samples were added to pH 8 phosphate buffer solution and homogenized using a Biospecs tissue tearer. After homogenization, the samples were centrifuged at 10,000 rpm for 10 min. AChE activity was then analyzed using a colorimetric method (Elman et al., 1961) with a Bausch and Lomb 710 spectrophotometer. 2.3.3. Neurotransmitter analysis Brains were thawed and homogenized in 0.05 M perchloric acid spiked with dihydroxybenzylamine as an internal standard. Following centrifugation, neurotransmitter concentrations in the supernatant were measured by high-performance liquid chromatography (HPLC) using a modified method (Weber et al., 1991) of Salzman et al. (1985). The mobile phase consisted of citric acid (7.0 mM), dibasic sodium phosphate (11.5 mM), sodium octylsulfate (0.43 mM), EDTA (1.3 mM), and diethylamine (0.12%), with the final solution adjusted to a pH of 3.5 with HPLC-grade phosphoric acid. Using this mobile phase, each brain sample took approximately 40 min to run. Samples were analyzed with a BAS Inc. HPLC system using an LC-18 reversed phase column (7.5 cm 4 mm i.d., with 3-mm particles) and electrochemical detection. From chromatograms the following neurotransmitter concentrations were determined: norepinephrine, l-Dopa, dopamine, and serotonin. 2.3.4. Statistical analysis Data were analyzed using the Statistical Analysis System (SAS version 7, North Carolina, USA). Daily treatment comparisons were performed with the PROC Mixed Function for repeated measures experiments. Nontime-dependent comparisons were made with the Student–Neuman–Keul’s test in PROC GLM. This procedure was used rather than a Tukey test so as to decrease the probability of a Type II error. Tukey tests were used for all biochemical analyses because biochemical data showed lower variances than did behavioral data. For Experiment 2, unless otherwise noted, habitat use was compared using Student t tests with SAS PROC UNIVARIATE. For all tests, significance was reported if Po0:05:
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
191
3. Results 3.1. Experiment 1 The pre- and posttrial appetite studies (i.e., the studies that took place before and after the experimental foraging trials with vegetated habitats) revealed no significant differences between groups in the number of worms eaten per day, suggesting that Hg exposure did not result in appetite suppression. Using the foraging trial data, the numbers of worms encountered by the groups were compared using the 3rd minute of the trial. These metrics gave an indication of the foraging efficiency (prey captured/min) over trial time. For most groups, improvement over time occurred until the 3rd or 4th day, after which there was a leveling off in performance. Depending on day and treatment, significant differences from controls were found (Fig. 2). Across the trial days, control fish showed greater mean foraging efficiencies than did all treatment groups. Significant differences existed on Day 2 (control vs T6, Po0:05; control vs T13, Po0:01), Day 3 (control vs T13, Po0:005), and Day 6 (control vs T6, Po0:05). Prey encounter rates were quite variable within treatments and among days, making it difficult to gauge the potential foraging ability of individuals. Consequently, it is useful to look at a fish’s maximum foraging efficiency across the entire 7-day trial period. A maximum encounter value was determined by averaging the two best performances for a single fish. Treatment comparisons were then performed using the SNK multiple comparisons procedure (Zar, 1996). Using data from the first 3 min of the trial, significant differences between controls and T13 were found (Po0:05; Fig. 3). Another, more precise measurement of foraging ability was the time it took for a fish to encounter a
Fig. 2. Number of worms encountered by control and treatment groups after 3 min. Each data point represents the mean and SE for that group for that day. *Significant difference from the control.
Fig. 3. Maximum number of worms encountered (during the first 3 min) over the 7-day trial period for each treatment. Each bar represents the treatment mean and SE of an individual fish’s two best performances over the 7-day trial period. Bars with different letters are significantly different.
Fig. 4. Time to four encounters for each day of the 7-day trial period. Each data point represents the mean and SE for that group for that day. *Significant difference from the control.
certain number of prey items. This metric is a measure of encounter speed and has better resolution than does foraging efficiency. To include in the analysis as many days and groups as possible, the time to four encounters was used. For all statistical analysis, data were log transformed for equal variance. A significant relationship between time to four encounters and trial day (Fig. 4) was found only for control fish (Po0:05). T1, T6, and T13 fish showed no significant improvement over trial days. Mean differences in the encounter speed existed across the 7-day trial period, with controls showing faster encounter speed compared with exposed groups (Fig. 4). However, a significant treatment effect was found only on Day 3 (T6, Po0:01; T13, P ¼ 0:01). As with foraging efficiency, an average of the ‘‘best performance’’ over the 7-day period was calculated. In this case, the two shortest times to three or four encounters were used. In the time it took to encounter four worms, both T6 and T13 fish took significantly longer than did control fish (P ¼ 0:05 and 0.01, respectively) (Fig. 5).
192
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
Controls paused less than did T6 and T13 fish, but only T13 fish were significantly different (Po0:05) (Table 1). The number of quadrant switches did not differ significantly, although treatment means indicate that controls tended to move about the feeding area more than did the other groups.
For the foraging trials, to minimize satiation and patch depletion, only data from the first 2 min of the 10min trial were used. Treatment means for the learning metrics were calculated using the average of a foraging group’s behavior on the first 2 days and the last 2 days of the pre- and postswitch trial period. As noted above, fish fed in groups of 3 (5 groups/treatment). During the preswitch period, comparisons of habitat use on the first and last 2 days revealed significant increases in the time spent in the high food patch for controls and T13 fish only (two-sample t test, Po0:05 for both) (Table 2). However, on the last 2 days, all groups spent the most time in the high food patch (control, Po0:0001; T1, Po0:001; T6, P ¼ 0:012; T13, P ¼ 0:0001) and no treatment differences in habitat use were found. The first patch chosen was also similar for all treatments (Table 2). The mean percentage of fish choosing the high food patch first did increase over time, but on the end of the first 6 trial days there was still no significant tendency to choose the high food patch. No treatment effects on initial patch choice were found. The number of visits to the low food patch declined across trial days for controls and T1 fish only (twosample t test, Po0:05 for both). However, on the last 2 days, patch visitation was similar for all treatments (Table 2). After switching the relative food values of each patch, the fish showed an increased preference for the new high
3.1.1. Experiment 2 Preforaging trial appetite data revealed no significant differences among treatments, although T13 fish ate fewer worms per day than did the other groups.
Fig. 5. Shortest time to four encounters for each treatment. Each bar represents the treatment mean and SE of an individual fish’s two shortest times to four prey encounters over the7-day trial period. Bars with different letters indicate significant differences.
Table 1 Pause time and quadrant switches per day (means7SEs) during first 3 min of the foraging trials Search behavior
Pause time (s) Number of switches
Exposure concentration (mg/L) Control
1.69
6.79
13.57
43.9876.51 10.4071.71
46.2676.67 7.7970.96
59.7276.90 7.1471.14
66.5776.34n 5.9070.73
Note. Each value represents the treatment average of the average number/day for an individual fish. n Significant difference from controls (Po0.05).
Table 2 Learning metric comparisons of the first 2 days and last 2 days (in parentheses) of the preswitch feeding regime Learning metric
Exposure concentration (mg/L) Control 1.69
6.79
13.57
Percentage time in HF patch
46.4711.5 (79.6710.6)n
59.3711.9 (73.678.2)
61.5716.5 (73.3715.1)
62.5711.0 (91.478.6)n
Percentage of first choices for HF patch
38.377.26 (66.7710.5)
48.3715.5 (65.0710.0)
70.078.6 (48.3715.5)
46.7717.8 (83.3710.5)
Number of entrances into the LF patch
1.3570.43 (0.5070.20)n
1.3870.37 (0.627 0.21)n
0.6170.26 (0.7270.27)
Note. All groups spent significantly more time in the high food patch on the last 2 days. n Significant difference between the first and last 2 days within that treatment. HF, high food patch; LF, low food patch.
0.4570.25 (0.3770.29)
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
193
Table 3 Learning metric comparisons of the first 2 days and last 2 days (in parenthesis) of the post-switch feeding regime Learning metric
Exposure concentration (mg/L) Control
1.69
6.79
13.57
Percentage time in HF patch
34.976.08 (52.7716.0)
39.078.04 (45.579.61)
17.4713.8 (25.574.40)
30.2711.9 (38.7717.6)
Percentage of first choices for HF patch
13.379.72 (40.0718.7)
29.0710.1 (56.7711.3)
13.379.72 (36.7716.8)
28.3711.7 (25.3719.4)
Number of entrances into LF patch n
1.8570.45 (0.9070.34)n
1.2370.28 (0.7370.10)
1.7270.26 (1.7670.27)
1.4570.23 (0.98 7 0.26)
Significant difference between the first and last 2 days within that treatment.
Table 4 Muscle tissue Hg body burden (means7SEs) and acetylcholine esterase activity (means7SEs) after a 10-day waterborne exposure to HgCL2 immediately after exposure (pretrial) and after a 10-day behavioral trial during which time the fish were held in clean water (posttrial) Muscle tissue
Exposure concentration (mg/L) Control
1.69
Pretrial Hg (mg/g) AChE activity (as percentage of control)
0.1070.02 100.0070.44
0.5970.06 125.2370.54
Posttrial Hg (mg/g) AChE activity (as percentage of control)
0.0170.01 100.0070.36
0.5970.06 101.0270.32
food patch by the end of 6 days (Table 3) but, except for the control group, continued to spend more time on average in the low food patch, indicating a retention of previously learned information. T6 fish actually spent significantly more time in the low food patch (Po0:005) on the last two days. The percentage of fish choosing the new high food patch first was higher in the last two days of the postswitch regime compared to the first two days in all groups but T13 (Table 3). However, on the last two days of the postswitch trial, only T1 fish chose the high food patch more than 50% of the time, indicating that the former high food patch continued to be sampled first and that habitat information from the preswitch period was retained. Similarly, all groups continued to make entries into what had been the high food patch (Table 3); however, over the postswitch period, control fish showed a significant decline in visitation to the low food patch (two-sample t test, Po0:05).
3.1.2. Biochemical analysis The posttrial fish had 10 days in clean water but appeared to depurate very little Hg during that time period (Table 4). AChE activity was not found to differ significantly with treatment in pre- or postforaging trial
6.79 4.4070.61 93.82070.59
3.1570.27 102.9670.36
13.57 8.0270.33 109.4670.67
7.2770.63 100.0770.29
tests. T1 fish did have elevated, but nonsignificant, activity in the pretrial measurement (Table 4). Similar results were found for neurotransmitter levels (Table 5). In the pretrial measurements, DA and 5-HT levels were uniform across groups. T13 fish had nonsignificantly elevated NE levels compared with control fish, and l-Dopa levels differed nonsignificantly among treatments. However, the variation in l-Dopa did not appear to correspond to Hg body burden. With the exception of 5-HT, all neurotransmitter levels were similar in the posttrial groups.
4. Discussion The first section of the discussion attempts to explain the results of Experiments 1 and 2 by a mechanistic analysis of foraging behavior. Second, biochemical results are interpreted and compared with related literature. Finally, the results of these experiments are examined with regard to implications for aquatic systems in nature. 4.1. Experiment 1 As hypothesized for Experiment 1, foraging ability was related to Hg body burden. When compared on
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
194
Table 5 Neurotransmitter levels (means7SEs) immediately after a 10-day waterborne exposure to HgCl (pretrial) and after a 10-day trial period during which time the exposed fish were held in clean water (posttrial) Concentration (ng/mg) brain
Treatment concentration (mg/L) Control
1. 69
6.79
13.57
Pretrial NE l-Dopa DA 5-HT
0.41670.065 0.22770.059 0.06870.011 0.10470.021
0.37470.054 0.36570.098 0.08070.015 0.08570.015
0.40370.042 0.18670.049 0.07170.011 0.1107 0.020
0.51370.062 0.27670.104 0.07570.008 0.12270.015
Posttrial NE l-Dopa DA 5-HT
0.57170.047 0.39670.090 0.10770.011 0.15570.015
0.56570.050 0.40570.056 0.09170.010 0.11970.009
0.54970.031 0.39970.080 0.08870.008 0.11970.010
0.53770.036 0.37670.059 0.09770.016 0.11770.008n
n
Significant difference with control.
individual trial days, control fish captured more worms in less time. The results most often follow a dose– response pattern in which body burden is inversely related to test performance, although only comparisons with T6 and T13 fish yielded significant differences. Also, when comparing the best performance of individual fish, controls had significantly better mean performances than did the T13 group. The hypothesis that exposed fish would exhibit no improvement in foraging trials over trial days was confirmed in the case of T13 and, in some instances, T6 fish. Finally, it should be noted that all foraging trials took place while the fish were held in noncontaminated water, suggesting persistent effects after Hg exposure. Possible explanations for the foraging deficits found in this study would include appetite suppression in exposed fish, greater pause time exhibited by exposed fish, sensory impairment and treatment differences in learning to feed efficiently. As noted in the results, appetite suppression was not statistically evident in the pre- or posttrial tests, nor were their differences in the time spent in the feeding area during the foraging trials. Pause time (lack of directional movement), however, did differ among treatments, indicating potential sensory/central nervous system changes (Table 1). Moreover, prey search behavior, of which pause time is an important component, is likely to affect encounter rates. To investigate further, the 7-day average time to four captures was correlated against the 7-day average pause time during that time interval (Fig. 6). A significant inverse relationship between pause time and capture time was found (Po.0001). Furthermore, treatment comparisons revealed significant differences in the pause time until four encounters (control vs T13, Po0:01), suggesting that pause time contributed to treatment differences in foraging success (Fig. 7). Differences in
Fig. 6. The relationship between pause time and the time until four worms were encountered. Each data point represents an individual fish’s 7-day average time to four encounters and its 7-day average pause time until four encounters.
Fig. 7. Average pause times (means and SEs) until four worms were encountered. Bars with different letters are significantly different.
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
pause time could result from nervous system depression (lethargy). McKim et al. (1976) reported sluggishness in trout with body burdens of 5–7 mg/g but not in those with 3 mg/g; however, Olson et al. (1975) observed no lethargy in fathead minnows, although they had up to 11 mg/g Hg in muscle tissue (336 days exposure in 0.247 mg/L methylmercury). Treatment differences in learning did not seem to explain any differences in foraging ability. In Experiment 1, the fish were introduced to a novel habitat in which successful foraging required visual and locomotory accommodations for a vegetated environment. Also, prey items were placed in the same locations every day, making retention of prey location information potentially important to foraging success. However, after reviewing the daily trials, no relationship between pause time and trial day was found for any group, suggesting that pause behavior was not a learned response to the habitat. Nor did fish seem to develop a specific search pattern, as neither movement within the aquarium nor the spatial sequence of captures was similar over days. This would suggest a random search. Similar results were found by Marschall et al. (1989) using bluegill sunfish. 4.2. Experiment 2 The second experiment was designed to focus specifically on learning, and it was hypothesized that, compared with controls, (1) the exposed fish would show no discrimination between the high and low food patch while foraging, (2) over trial days, the exposed fish would exhibit no increase in their tendency to initially choose the high food patch, (3) the treatment fish would show no decline in visitation to the low food patch, and (4) all of those hypotheses would also be true with a change in patch food conditions. None of these hypotheses was confirmed, as the results showed no impairment of learning or retention due to Hg exposure. During the preswitch period, all groups spent more time in the high food patch and exhibited the same first patch choice and patch visitation behavior. After switching the relative food values of each patch, all groups continued to enter the former high food patch first and to spend comparatively more time (although less than during the preswitch period) there. These data suggest that all groups learned and retained habitat information and that, interestingly, no group made dramatic changes in habitat use despite the habitat change. The results are not that unusual in that many investigators report that fish frequently depart from mathematical optimality models of habitat use and patch switching behavior (Pyke, 1984). Such a departure may be even more likely for Experiment 2, in which there was low energy cost (short travel distance) associated with movement between patches.
195
Despite the results of the current experiments, Hg may disrupt the process of memory consolidation. One mechanism could be the inhibition of ACh release from neurons. Hg acts at the neuromuscular junction and the synaptic cleft, producing a weakened postsynaptic potentiation. Mercury, by altering ACh release, may impair memory formation because short- and long-term learning is facilitated by the induction of cellular changes that increase ACh release from the presynaptic terminals of neuronal circuits (Bekkers and Steven, 1990). Another possible mechanism is the destruction of granule cells in the hippocampus (Chang, 1979), a primary center for learning and memory in higher vertebrates. Comparing our results with prior toxicological learning studies with fish is difficult because very few have used Hg as a toxicant or examined learning tasks related to foraging. Most have used conditioned response tests in which fish were trained to associate a light and a shock. After successful training, the treatment fish have been exposed to Hg and retested. Authors have reported impaired retention at low (3–6 mg/L) waterborne exposures (Weir and Hine, 1970) and at body levels greater than 2.85 mg/g (Hartman, 1978). In an operant conditioning test (target strikes with food as reinforcement) involving goldfish, Salzinger et al. (1973) reported impaired retention at 6 and 10 mg/L HgCl2. Also, lead, which has toxicity mechanisms similar to Hg, may affect memory in fathead minnows (Weber et al., 1991). The results of past studies would suggest that Hg could disrupt learning and retention at very low doses, although drawing firm conclusions is difficult due to the small number of studies and their methodological differences. There are some possible explanations for the apparent lack of effect in the current study. First, previous conditioning studies tested fish during the exposure period, whereas in the current study the trials took place while fish were in clean water during a recovery period. Second, the mercury species may have been important. Compared with HgCl2, methylmercury (MeHg) passes the blood–brain barrier more readily (Chang, 1979), making it more neurotoxic, although it should be noted that some older learning studies did use HgCl2. Regardless of whether Hg harms memory, in the future, ecologically relevant tests of learning and decision making will be of much value. 4.3. Biochemical data 4.3.1. Acetylcholinesterase Disturbing a process as central as AChE function could result in widespread effects on the central and peripheral nervous system, including sensory, motor, and autonomic function. Interestingly, in our experiment, AChE activity exhibited no treatment effect, even
196
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
though some studies using laboratory (Suresh et al., 1992) and field collected fish (Shaw and Panigrahi, 1990) do suggest that Hg can alter AChE activity. The proximate cause could be a conformational change brought about by the binding of Hg to a sulfhydryl group on the cholinesterase. Another means could be the destruction of anticholinergic cells. Although the results presented here were incongruous with the limited literature, older studies used much greater exposure levels (often in the hundreds of mg/L). 4.3.2. Brain neurotransmitters In prior studies, although they are few in number, both MeHg and HgCl2 altered monoamine levels in brain tissue of fish (Zhou et al., 1999). However, these studies often used high-dose, low-duration exposures. Also, in the current study, analyses were of whole brain and therefore incapable of distinguishing local changes in neurotransmitter levels. This is crucial because production and metabolism of neurotransmitters vary by brain region in fish (Hornby and Peikut, 1990) just as they do in other animals. In fact, Hg toxicity in the brain does seem to be spatially dependent (Tsai et al., 1995). The lack of effect in the current study may have resulted from low-exposure level and duration and, due to the small size, the inability to examine the fish brain by specific anatomical region. Consequently, any local differences were too small to be resolved. Clearly, more work is needed to elucidate the interaction of Hg and neurotransmitters in fish. 4.3.3. Mercury body burden In understanding the environmental significance of the results reported here it is important to consider tissue levels because the waterborne exposure was greater than that generally found in polluted lotic and lentic systems (typically 10–40 ng/L), (Wiener and Spry, 1996). As would be expected, Hg body burden increased with exposure dose (Table 5). Control fish had between 0.1 and 0.012 ppm, levels typical in fish collected from nonpolluted environments lacking conditions that promote metal bioavailability, that is, low pH and warm temperature (Wiener and Spry, 1996). In aquatic systems that have natural mercury deposits or conditions favorable to availability, levels similar to those found in T1 fish would not be uncommon. In polluted areas, piscivorous or bottom feeding fish can contain mean tissue concentrations from 1 to 7 ppm depending on factors such as age and distance from point source. These values are equivalent to those reported for T6 and, although on the high end, T13 fish. However, mercury accumulation is partially a function of longevity (Ward and Neumann, 1999), therefore, short-lived species such as the fathead minnow would be unlikely to accumulate levels above 1 ppm in natural systems.
It is important to note, however, that most (90–95%) of the total mercury found in field-collected specimens is the most toxic form of mercury, MeHg, rather than the inorganic form used here (AMRL, 1998). Also, for the fathead minnow and species occupying similar trophic levels, benthic detritus often constitutes a significant dietary component. By feeding in substrates, the fathead minnow would potentially increase Hg uptake via gill or dietary routes. Finally, fish were tested during a recovery period in which body stores were the only source of exposure. So, despite the uncertainty inherent in any extrapolation from lab to field, the results reported in this study have potential implications for fish in nature.
5. Conclusion Traditionally, toxicologists approach pollution stress from a clinical reference point that emphasizes lethal dosages and cellular and molecular mechanisms of toxicity. Such studies require a controlled environment and, as a result, are laboratory based. One major drawback to this approach is that environmental relevance is often not clearly defined, and no attempt is made to relate alterations at the cellular/molecular level to higher levels of ecosystem organization (e.g., individual, community). In contrast, toxicity tests that examine foraging behaviors focus attention on variables whose environmental relevance is easily recognized. For example, in this study, Hg was found to decrease foraging efficiency, a component of bioenergetics models. General energy budgets such as Consumption ¼ Growth þ Egestion þ Ingestion þ Metabolism þ Work illustrate the direct relationship between growth and food consumption. The ingestion and work variables can be broken down into a variety of components, among them the foraging ability variables used in this study. For example, decreased foraging efficiency could mean that, for a given time period more energy is expended searching, while actual food consumption over that time period, is reduced. If fish are unable to acquire food efficiently, then the life history, reproductive, and (potentially) ecological implications are profound. Toxicology studies that use foraging bioenergeticsbased end points offer several possibilities for ecotoxicologists. Mathematical modeling of net energy gains from foraging could be performed using toxicantexposed and -unexposed fish. These models could then be used to predict pollution-induced alterations in growth, reproduction, and metabolism in the wild. Other mathematical models related to foraging
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
behavior, such as models of prey choice (e.g., optimal foraging theory) and habitat preference (e.g., ideal free distribution, patch departure modeling), could be also be examined. One very promising approach (Smith and Weiss, 1997) is to perform laboratory comparisons of foraging ability using fish collected from polluted and reference sites. This approach allows a degree of hypothesis testing using wild fish that have undergone realistic pollutant exposures. Finally, the purpose of ecotoxicology is to investigate the effects of environmental pollution on the ecology of natural systems, making laboratory investigations most informative if the results can be placed within an ecosystem context. Therefore, one would ultimately attempt to test predictions about how impaired feeding ability in individuals may influence higher levels of ecosystem organization. Testing of this kind would yield information on how contaminants alter organizational levels and linkages within ecosystems. It is only with an ecosystem level understanding that we can regulate the discharge of pollutants with full knowledge of their impacts.
Acknowledgments The authors thank Paul Angermeier and Thomas A. Jenssen for their suggestions and ideas that were essential to this project. The authors are also grateful to the laboratory of Neal Castagnoli for the use of the high-performance liquid chromatography and to Jaques Petzer for his aid in the neurotransmitter analysis.
References AMRL, 1998. A comprehensive evaluation of the South River and South Fork Shenandoah River for mercury contamination: final report. AMRL Technical Report No. 3079, Old Dominion University, Norfolk, VA. Atchison, G.J., Sandheinrich, M.B., Bryan, M.D., 1996. Effects of environmental stressors on interspecific interactions of aquatic animals. In: Newman, M.C., Jagoe, C.H. (Eds.), Exotoxicology: A Hierachical Treatment. CRC Press, Inc., Boca Raton, pp. 319–345. Beitinger, T.L., 1990. Behaviroral reactions for the assessment of stress in fishes. J. Great Lakes Res. 16, 495–528. Bekkers, J.M., Steven, C.F., 1990. Presynaptic mechanism for longterm potentiation in the hippocampus. Nature 346, 724–729. Chang, L.W., 1979. Pathological effects of mercury poisoning. In: Nriagu, J.O. (Ed.), The Biogeochemistry of Mercury in the Environment. Elsevier/North-Holland Biomedical Press, Amsterdam, pp. 519–580. Ehlinger, J.T., 1989. Learning and individual variation in bluegill foraging: habitat-specific techniques. Anim. Behav. 38, 643–658. Ehlinger, T.J., 1990. Habitat choice and phenotype-limited feeding efficiency in bluegill: individual differences and trophic polymorphism. Ecology 71, 886–896.
197
Elman, G.L., Courtney, D., Andres, O., Featherstone, R., 1961. A new and rapid colorimetric determination of AChE activity. Biochem. Pharmocol. 7, 88–95. Gotceitas, V., 1990. Foraging and predator avoidance: a test of a patch choice model with juvenile bluegill sunfish. Oecologia 83, 346–351. Hartman, A.M., 1978. Mercury feeding schedules: effects on accumulation, retention and behavior in trout. Trans. Am. Fish. Soc. 107, 369–375. Hornby, P.J., Peikut, D.T., 1990. Distribution of catecholaminesynthesizing enzymes in goldfish brains: presumptive dopamine and norepinephrine neuronal organization. Brain Behav. Evol. 35, 49–64. Hughes, R.N., Kaiser, M.J., Mackney, P.A., Warburton, K., 1992. Optimizing foraging behaviour through learning. J. Fish Biol. 41 (Suppl. B), 77–91. Jenkins, N., Burkehead, R., 1994. Freshwater Fishes of Virginia. American Fisheries Society, Bethesda, MD. Little, E.E., Fairchild, J.F., Delonay, A.J., 1993. Behavioral methods for assessing impacts of contaminants on early life stage fishes. Am. Fish. Soc. Symp. 14, 67–76. Marschall, E.A., Chesson, P.L., Stein, R.A., 1989. Foraging in a patchy environment: prey-encounter rat and residence time distributions. Anim. Behav. 37, 444–454. McKim, J.M., Olson, G.F., Holcombe, G.W., Hunt, E.P., 1976. Long term effects of methylmercuric chloride on three generation of brook trout (Salvelinus fontinalis): Toxicity, accumulation, distribution and elimination. J. Fish Res. Board Can. 33, 2726–2739. O’ Brien, W.J., Evans, B.I., Browman, H.I., 1989. Flexible search tactics and efficient foraging in saltatory searching animals. Oecologia 80, 100–110. Olson, G.F., Mount, D.I., Snarski, V.M., Thorslund, I.W., 1975. Mercury residues in fathead minnows, Pimephales promelas Rafinesque, chronically exposed to methylmercury in water. Bull. Environ. Contam. Toxicol. 14, 129–134. Pyke, G.H., 1984. Optimal foraging theory: a critical review. Ann. Rev. Ecol. Syst. 15, 523–575. Salzinger, K., Fairhurst, S.P., Friemark, S.J., Wolkoff, F.D., 1973. Behavior of the goldfish as an early warning system for the presence of pollutants in water. J. Environ. Syst. 3, 27–40. Salzman, S.K., Llados-Eckman, C.L., Hirofuji, E., 1985. Determination of spinal cord monoamines and metabolites using three micron columns and dual electrochemical detection. J. Liquid Chromatogr. 8, 345–360. Smith, G.M., Weiss, J.S., 1997. Predator–prey relationships in mummichogs (Fundulus heteroclitus (L.)): effects of living in a polluted environment. J. Exp. Mar. Biol. Ecol. 209, 75–87. Smith, J.R., 1984. Fish neurotoxicology. In: Weber, L.J. (Ed.), Aquatic Toxicology, Vol. 2. Raven Press, New York, pp. 107–152. Strickler-Shaw, S., Taylor, D.H., 1991. Lead inhibits acquisition and retention learning in bullfrog tadpoles. Neurotoxicol. Teratol. 13, 167–173. Sun, T., Taylor, D.H., 1983. The effects of parathion on acquisition and retention of shuttlebox avoidance-conditioning in the goldfish, Carassius auratu. Environ. Pollut. (Series A) 31, 119–131. Shaw, B.P., Panigrahi, A.K., 1990. Brain AChE activity studies in some fish species collected from a mercury contaminated estuary. Water Air Soil Pollut. 53, 327–334. Suresh, A., Sivaramakrishna, B., Victoriamma, P.C., Radhakrishnaiah, K., 1992. Comparative study on the inhibition of acetylcholinesterase activity in the freshwater fish Cyprinus carpio by mercury and zinc. Biochem. Int. 26, 367. Tsai, C., Jang, T., Wang, L., 1995. Effects of mercury and serotonin concentration in the brain of tilapia, Oreochromis mossambicus. Neurosci.Lett. 184, 208–211.
198
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
US Environmental Protection Agency (USEPA), 1990. Test methods for evaluating Solid Waste (SW-846). USEPA. 1983. Method 7470, Rev. 1, Nov. 1990. Office of Solid Wastes, Washington, DC. Ward, S.M., Neumann, R.M., 1999. Seasonal variations in concentrations of mercury in axial muscle tissue of largemouth bass. N. Am. J. Fish Manage. 19, 89–96. Weber, D.N., Russo, A., Seale, D.B., Spieler, R.E., 1991. Waterbourne lead affects feeding abilities and neurotransmitter levels of juvenile fathead minnows (Pimephales promelas). Aquatic Toxicol. 21, 71–80. Weber, D.N., Spieler, R.E., 1994. Behavioral mechanisms of metal toxicity in fishes. In: Malins, D.C., Ostrander, G.K. (Eds.), Aquatic
Toxicology: Molecular, Biochemical, and Cellular Perspectives. CRC Press, Boca Raton, pp. 421–467. Weir, P.A., Hine, C.H., 1970. Effects of various metals on behavior of conditioned goldfish. Arch. Environ. Health 20, 45–51. Wiener, J.G., Spry, D.J., 1996. Toxicological significance of mercury to freshwater fish. In: Beyer, W.N., Heinz, G.H., Redmon, A.W. (Eds.), Environmental Contaminants in Wildlife. CRC press, Boca Raton, pp. 297–339. Zar, J., 1996. Biostatistical Analysis, 3rd Edtion. Prentice-Hall, Upper Saddle River, NJ. Zhou, T., Rademacher, D.J., Steinpreis, R.E., Weis, J.S., 1999. Comp. Biochem. Physiol C 124, 287–294.
The effect of mercury on the feeding behavior of fathead minnows (Pimephales promelas) M.A. Grippo* and A.G. Heath Department of Biology, Virginia Polytechnic Institute and State University, Blacksburg, VA 24060, USA Received 7 January 2002; received in revised form 26 September 2002; accepted 4 October 2002
Abstract Fathead minnows (Pimephales promelas) were exposed to mercury (1.69, 6.79, and 13. 57 mg/L HgCl2, 10 days exposure), and afterward their foraging ability was tested in a vegetated habitat for 7 days. Among the foraging metrics used were foraging efficiency, capture speed, and the ability to learn and retain information regarding habitat characteristics. In addition to behavioral tests, muscle tissue acetylcholinesterase activity and brain levels of several neurotransmitters were investigated. Comparisons with control fish and fish from the two highest exposure groups revealed consistent performance deficits in foraging efficiency and capture speed. However, no treatment effects on learning were detected, nor were differences in neurotransmitter levels detected. In determining the underlying proximate cause of the foraging deficits, it is believed that the greater pause time exhibited by treatment fish while foraging was the main cause of treatment differences. In the future, behavioral studies will continue to allow toxicity testing of environmentally relevant variables such as those used by behavioral ecologists. r 2003 Elsevier Science (USA). All rights reserved. Keywords: Mercury; Foraging behavior; Learning; Fathead minnows; Neurotransmitters
1. Introduction Ecotoxicology recognizes that within aquatic systems there are multiple interacting levels of organization, ranging from the cellular, to the individual, to the ecosystem level. At the level of the individual, sublethal toxicant effects are manifested in several ways, but among the most sensitive indicators of pollution stress are behavioral alterations. In particular, feeding behavior has been investigated extensively because feeding impairment could have repercussions on an individual’s survivorship, growth, and reproductive capacity. Much previous work has focused on appetite, foraging efficiency, and predator avoidance (reviewed in Beitinger, 1990; Little et al., 1993). More recent research on foraging behavior has often been in the form of mechanistic studies in which a toxicant’s interaction with a specific component of food acquisition is examined (Atchison et al., 1996). The mechanistic parameters of interest are the same as those used by *Corresponding author. CZR Inc., 4709 College Acres Drive, Suite 2, Wilmington, NC, 28403, USA. E-mail address: [email protected] (M.A. Grippo).
behavioral ecologists (e.g., food consumption rate, feeding attempts, capture efficiency, handling time, reaction distance), and all are potentially altered by toxicants (Atchison et al., 1996). Although behavioral toxicologists have investigated several ways in which pollutants can affect foraging behavior, other potential routes remain unexamined. One example is learning and foraging behavior. Numerous studies have confirmed the importance of experience to improvement in foraging ability (reviewed in Hughes et al., 1992). Learning is also valuable for choosing among habitats with differing food values. By choosing the most profitable habitat, fish not only will increase food consumption per unit of foraging time but also will reduce time and energy traveling between patches (Hughes et al., 1992). Furthermore, studies have shown that contaminants can inhibit a variety of learned responses (Strickler-Shaw and Taylor, 1991; Sun and Taylor, 1983). However, few experiments have examined the impairment of ecologically relevant learned behaviors such as those behaviors related to foraging efficiency, even though such investigations would provide insight into the subtle ways in which pollution can harm ecosystems.
0147-6513/03/$ - see front matter r 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0147-6513(02)00071-4
188
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
Despite the relevance to ecotoxicology, aspects of foraging behavior well studied by behavioral ecologists, such as learning, optimal foraging, and prey and habitat choice, are relatively unexplored by toxicologists. Therefore, the objective of the current study was to investigate behavioral alterations in toxicant-exposed fish using two foraging experiments that test prey search strategy, prey location efficiency, and habitat choice. First, an observational study was made comparing, through time, the foraging ability of toxicant-exposed and -unexposed fish introduced to a novel habitat. Also, although fish species may have characteristic search patterns (e.g., cruisers, salutatory searchers), experience may modify that pattern (e.g., changes in hover time, distance moved between hovers) based on habitat structure (Ehlinger, 1989) or prey type (O’ Brien et al., 1989). Because of the potential importance of these search strategies in maximizing foraging return, search behavior was also characterized. It was hypothesized that exposed fish would exhibit deficits in daily measurements of foraging ability compared with control fish, that treatment fish would show little improvement in foraging success over trial days, and that performance deficits would be related to differences in search behavior. Second, experience influences not only foraging success but also habitat choice. Fish are capable of determining, over time, the relative profitability of different habitats and choosing among them based on this learned information (Ehlinger, 1990; Gotceitas, 1990). Consequently, a second experiment examined whether contaminants inhibit or retard the ability of fish to learn and retain information about relative habitat value. Specifically, it was hypothesized that during foraging trials the treatment fish would show no preference between patches of high and low food value and that, based on various metrics of retention, exposed fish would show reduced retention of habitat information across trial days. Mercury (Hg) was chosen as the toxicant for several reasons. First, Hg is highly neurotoxic. In mammals, mercury commonly passes the blood–brain barrier, causing destruction of visual and hippocampal cells, and disrupts the neurons of the peripheral nervous system (Chang, 1979). Also, Hg is a common aquatic pollutant that bioaccumulates in fish through dietary and waterborne exposure (Wiener and Spry, 1996). Finally, a variety of fish behaviors are influenced by neurotransmitters (reviewed in Weber and Spieler, 1994), making toxicant–neurotransmitter interaction an important area of study. Although researchers have examined the effect of organic and inorganic pollutants on neurotransmitters, few have looked at behavior as well. Therefore, the effect of Hg on several neurotransmitters was investigated to determine a correlation among exposure, neurotransmitter levels,
and behavioral alterations. Specifically, acetylcholinesterase (AChE) activity was measured in axial muscle, while norepinephrine (NE), serotonin (5-HT), dopamine (DA), and its metabolite l-DOPA were measured in brain tissue. All of these neurotransmitters are involved variously with locomotion, conditioned responses, and feeding (reviewed in Smith, 1984), and AChE activity is a commonly used biomarker of pollution stress.
2. Materials and methods 2.1. General 2.1.1. Test organisms Fathead minnows (Pimephales promelas) were used as test organisms because they have a wide distribution throughout the midwestern states and are a common toxicological test organism. Fathead minnows are characterized as benthic and mid-level water column feeders (Jenkins and Burkehead, 1994). Juveniles (3.5– 4.0 cm) were obtained from Aquatox (Arizona, USA) and Aquatic Biosystems Inc. (Colorado, USA). Prior to experiments, they were held in Minnow Cool tanks and fed trout chow. Blackworms (Tubificidae spp) from a local pet store were used as prey items during experimental trials. 2.1.2. Toxicant Fish were exposed to mercury (1.69, 6.79, or 13.57 mg/ L) as mercuric chloride (HgCl2) using static systems (28-L aquarium), with all aquariums receiving a daily 50% water change. 2.2. Foraging trials 2.2.1. Protocol for Experiment 1 An entire 20-day run entailed an initial exposure to one of the three HgCl2 treatments in 28-L holding tanks for 10 days and then transfer to noncontaminated aquariums at which time the 7-day foraging trials began. Foraging trials were followed by a 4-day, posttrial appetite study. A fourth group was unexposed and served as the control. Four total runs (blocks) were made. Each experimental run used a completely randomized design with two tanks (replicates) per treatment (eight tanks total). Within each holding tank, 3 fish (subsamples) were housed singly in three separate, gated removable compartments for a total of 24 fish (6 fish/sample, three treatments and a control). The treatment and control groups used in the trials are referred to by their exposure concentration (T1, T6, T13, or control). Also, each holding tank held 2 additional fish, housed separately. These fish, referred to as ‘‘nonexperimental fish,’’ were used for Hg body
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
burden and biochemistry tests immediately after the initial 10-day exposure and did not undergo the training or foraging experiments mentioned below. The purpose of the nonexperimental fish was to allow a comparison between fish immediately after HgCl2 exposure and after the 10-day trial period in which the fish had been held in noncontaminated water. Before the foraging trials began, it was necessary for each fish to be trained to readily enter the feeding area of the experimental aquarium. Therefore, during the initial 10-day exposure period, each fish was individually removed from its holding tank (it remained in its compartment) and transferred to the experimental aquarium, where the gate was removed and the fish was allowed to feed on 12 dead unburied blackworms for 5 min. The feeding area contained no artificial vegetation, just sand. At the end of the 5 min, the fish was led back into the removable compartment and placed in its holding tank. Daily worm consumption was recorded for each fish, yielding information on the appetite effects, if any, of mercury. After the 10-day exposure, the nonexperimental fish were frozen in liquid nitrogen and stored at 701C for biochemical analysis. The 24 trained fish were transferred to clean water, and the foraging trials were initiated. The daily experimental trials began by equally distributing 12 dead, partially buried blackworms (burying increased the difficulty of prey encounter and lessened visual cues) in the vegetated habitat (Fig. 1a). The habitat consisted of one large patch of artificial vegetation (35 42 cm; feeding area 13 cm in depth). The artificial vegetation consisted of 9-cm plastic sticks attached to a plastic mesh plate, which was covered in white dolomite sand. A single fish was then placed in the experimental aquarium and was allowed entrance to the feeding area for 5 min. All remaining prey were counted after completion of each trial. This procedure was repeated for all fish (24 trials per day) for 7 days. The order in which fish fed was randomized daily. After 7 days, all fish underwent another appetite study (4 days) using the protocol from the first 10 days (fish continued to be held in clean water). All trials were video-recorded. From video analysis, the number and time of prey capture were recorded, as were the number and duration of pauses. Also, to characterize movement during the trial, the video monitor screen was divided evenly into quadrants and the number of visits to each quadrant was recorded. From these data, foraging efficiency over time could be monitored along with differences in movement/search strategy. All fish were initially naive about the habitat structure and prey placement. Therefore, improvement in foraging efficiency for each group of fish could be compared over time.
189
Fig. 1. Diagram of the experimental feeding habitats for Experiment 1(a) and Experiment 2(b).
2.2.2. Protocol for Experiment 2 The foraging trials using the second habitat were designed to test the ability of treatment fish to learn the relative food availability of each patch. This series of experiments consisted of two runs (blocks), with each block consisting of a completely randomized design with two replications in the first block and three replications in the second block. The same four treatment exposure concentrations that were used in Experiment 1 were used in Experiment 2. Fish were exposed for 10 days and simultaneously trained to enter the feeding area in the manner described for Experiment 1 (no vegetation in
190
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
feeding area), with the exceptions that fish were housed and allowed to feed in groups of 3, and that 36 unburied worms were placed in the feeding area rather than 12. Following the 10-day exposure, fish were transferred to clean water and the foraging trials were initiated. The trials began by partially burying 22 blackworms in one patch (high food density) and partially buying 4 in the other (low food density). The habitat contained two patches of equal size (35 12 cm) and density of artificial vegetation separated by a sandy area (Fig. 1b). Fish were allowed to forage for 10 min to ensure that the entire habitat could be sampled. This procedure was repeated for all treatment tanks for 6 days. On the 7th day, the relative food values of each patch were switched and the fish were monitored for changes in habitat use. Hence, the first 6 days are referred to as the preswitch period, and the time period after the switch is referred to as are the postswitch period. Relative food values were switched so as to determine how fish from different treatments responded to changes in environmental conditions. As in Experiment 1, the order in which the fish fed was randomized daily. From video analysis, the following data were recorded: the total time that fish were found in each patch, the first patch sampled, and the number of visits to and exits from each patch. Using the time spent in a patch, inferences were made about relative patch preference. The first choice and visit data were used to determine whether the habitat information learned from previous trials was retained. For example, over the 6-day trial period, an increase in the time spent in the profitable patch was considered indicative of learning. Similarly, because repeated visits to the low food patch would be an inefficient use of time, a decrease in visits to the less profitable habitat was also considered evidence of learning. Entering the high food patch first and not visiting the low food patch was considered evidence of retention. 2.3. Biochemical analysis For Experiments 1 and 2, at the end of each run, experimental fish were frozen in liquid nitrogen. After thawing, brains were removed for neurotransmitter determinations. Samples of muscle tissue were also taken for use in the determination of Hg body burden and AChE activity. All samples were stored at 701C until analysis. 2.3.1. Mercury analytical methods The mercury species was mercuric chloride (0.1 M) obtained from Fisher Scientific. Holding tank Hg levels were measured periodically (approximately every other day) using a Perkin–Elmer MA50 cold vapor atomic absorption spectrophotometer (CVAAS) as described in Environmental Protection Agency (EPA) method 7470
(US EPA, 1990). Mercury body burden was determined by the following: digestion of tissue samples in 5 ml of concentrated H2SO4 and 2.5 ml of concentrated HNO3 for 1 h in a 581C water bath, followed by the addition of 15 ml of KPO4. The samples were then left overnight. Before reading the samples, NaCl hydroxylamine sulfate was added, followed by 5 ml of stannous chloride. The samples were immediately analyzed by the CVAAS. For each batch of 9 samples, a tissue spike and spike blanks were prepared and analyzed for percentage recovery. 2.3.2. Acetylcholinesterase activity Weighed tissue samples were added to pH 8 phosphate buffer solution and homogenized using a Biospecs tissue tearer. After homogenization, the samples were centrifuged at 10,000 rpm for 10 min. AChE activity was then analyzed using a colorimetric method (Elman et al., 1961) with a Bausch and Lomb 710 spectrophotometer. 2.3.3. Neurotransmitter analysis Brains were thawed and homogenized in 0.05 M perchloric acid spiked with dihydroxybenzylamine as an internal standard. Following centrifugation, neurotransmitter concentrations in the supernatant were measured by high-performance liquid chromatography (HPLC) using a modified method (Weber et al., 1991) of Salzman et al. (1985). The mobile phase consisted of citric acid (7.0 mM), dibasic sodium phosphate (11.5 mM), sodium octylsulfate (0.43 mM), EDTA (1.3 mM), and diethylamine (0.12%), with the final solution adjusted to a pH of 3.5 with HPLC-grade phosphoric acid. Using this mobile phase, each brain sample took approximately 40 min to run. Samples were analyzed with a BAS Inc. HPLC system using an LC-18 reversed phase column (7.5 cm 4 mm i.d., with 3-mm particles) and electrochemical detection. From chromatograms the following neurotransmitter concentrations were determined: norepinephrine, l-Dopa, dopamine, and serotonin. 2.3.4. Statistical analysis Data were analyzed using the Statistical Analysis System (SAS version 7, North Carolina, USA). Daily treatment comparisons were performed with the PROC Mixed Function for repeated measures experiments. Nontime-dependent comparisons were made with the Student–Neuman–Keul’s test in PROC GLM. This procedure was used rather than a Tukey test so as to decrease the probability of a Type II error. Tukey tests were used for all biochemical analyses because biochemical data showed lower variances than did behavioral data. For Experiment 2, unless otherwise noted, habitat use was compared using Student t tests with SAS PROC UNIVARIATE. For all tests, significance was reported if Po0:05:
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
191
3. Results 3.1. Experiment 1 The pre- and posttrial appetite studies (i.e., the studies that took place before and after the experimental foraging trials with vegetated habitats) revealed no significant differences between groups in the number of worms eaten per day, suggesting that Hg exposure did not result in appetite suppression. Using the foraging trial data, the numbers of worms encountered by the groups were compared using the 3rd minute of the trial. These metrics gave an indication of the foraging efficiency (prey captured/min) over trial time. For most groups, improvement over time occurred until the 3rd or 4th day, after which there was a leveling off in performance. Depending on day and treatment, significant differences from controls were found (Fig. 2). Across the trial days, control fish showed greater mean foraging efficiencies than did all treatment groups. Significant differences existed on Day 2 (control vs T6, Po0:05; control vs T13, Po0:01), Day 3 (control vs T13, Po0:005), and Day 6 (control vs T6, Po0:05). Prey encounter rates were quite variable within treatments and among days, making it difficult to gauge the potential foraging ability of individuals. Consequently, it is useful to look at a fish’s maximum foraging efficiency across the entire 7-day trial period. A maximum encounter value was determined by averaging the two best performances for a single fish. Treatment comparisons were then performed using the SNK multiple comparisons procedure (Zar, 1996). Using data from the first 3 min of the trial, significant differences between controls and T13 were found (Po0:05; Fig. 3). Another, more precise measurement of foraging ability was the time it took for a fish to encounter a
Fig. 2. Number of worms encountered by control and treatment groups after 3 min. Each data point represents the mean and SE for that group for that day. *Significant difference from the control.
Fig. 3. Maximum number of worms encountered (during the first 3 min) over the 7-day trial period for each treatment. Each bar represents the treatment mean and SE of an individual fish’s two best performances over the 7-day trial period. Bars with different letters are significantly different.
Fig. 4. Time to four encounters for each day of the 7-day trial period. Each data point represents the mean and SE for that group for that day. *Significant difference from the control.
certain number of prey items. This metric is a measure of encounter speed and has better resolution than does foraging efficiency. To include in the analysis as many days and groups as possible, the time to four encounters was used. For all statistical analysis, data were log transformed for equal variance. A significant relationship between time to four encounters and trial day (Fig. 4) was found only for control fish (Po0:05). T1, T6, and T13 fish showed no significant improvement over trial days. Mean differences in the encounter speed existed across the 7-day trial period, with controls showing faster encounter speed compared with exposed groups (Fig. 4). However, a significant treatment effect was found only on Day 3 (T6, Po0:01; T13, P ¼ 0:01). As with foraging efficiency, an average of the ‘‘best performance’’ over the 7-day period was calculated. In this case, the two shortest times to three or four encounters were used. In the time it took to encounter four worms, both T6 and T13 fish took significantly longer than did control fish (P ¼ 0:05 and 0.01, respectively) (Fig. 5).
192
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
Controls paused less than did T6 and T13 fish, but only T13 fish were significantly different (Po0:05) (Table 1). The number of quadrant switches did not differ significantly, although treatment means indicate that controls tended to move about the feeding area more than did the other groups.
For the foraging trials, to minimize satiation and patch depletion, only data from the first 2 min of the 10min trial were used. Treatment means for the learning metrics were calculated using the average of a foraging group’s behavior on the first 2 days and the last 2 days of the pre- and postswitch trial period. As noted above, fish fed in groups of 3 (5 groups/treatment). During the preswitch period, comparisons of habitat use on the first and last 2 days revealed significant increases in the time spent in the high food patch for controls and T13 fish only (two-sample t test, Po0:05 for both) (Table 2). However, on the last 2 days, all groups spent the most time in the high food patch (control, Po0:0001; T1, Po0:001; T6, P ¼ 0:012; T13, P ¼ 0:0001) and no treatment differences in habitat use were found. The first patch chosen was also similar for all treatments (Table 2). The mean percentage of fish choosing the high food patch first did increase over time, but on the end of the first 6 trial days there was still no significant tendency to choose the high food patch. No treatment effects on initial patch choice were found. The number of visits to the low food patch declined across trial days for controls and T1 fish only (twosample t test, Po0:05 for both). However, on the last 2 days, patch visitation was similar for all treatments (Table 2). After switching the relative food values of each patch, the fish showed an increased preference for the new high
3.1.1. Experiment 2 Preforaging trial appetite data revealed no significant differences among treatments, although T13 fish ate fewer worms per day than did the other groups.
Fig. 5. Shortest time to four encounters for each treatment. Each bar represents the treatment mean and SE of an individual fish’s two shortest times to four prey encounters over the7-day trial period. Bars with different letters indicate significant differences.
Table 1 Pause time and quadrant switches per day (means7SEs) during first 3 min of the foraging trials Search behavior
Pause time (s) Number of switches
Exposure concentration (mg/L) Control
1.69
6.79
13.57
43.9876.51 10.4071.71
46.2676.67 7.7970.96
59.7276.90 7.1471.14
66.5776.34n 5.9070.73
Note. Each value represents the treatment average of the average number/day for an individual fish. n Significant difference from controls (Po0.05).
Table 2 Learning metric comparisons of the first 2 days and last 2 days (in parentheses) of the preswitch feeding regime Learning metric
Exposure concentration (mg/L) Control 1.69
6.79
13.57
Percentage time in HF patch
46.4711.5 (79.6710.6)n
59.3711.9 (73.678.2)
61.5716.5 (73.3715.1)
62.5711.0 (91.478.6)n
Percentage of first choices for HF patch
38.377.26 (66.7710.5)
48.3715.5 (65.0710.0)
70.078.6 (48.3715.5)
46.7717.8 (83.3710.5)
Number of entrances into the LF patch
1.3570.43 (0.5070.20)n
1.3870.37 (0.627 0.21)n
0.6170.26 (0.7270.27)
Note. All groups spent significantly more time in the high food patch on the last 2 days. n Significant difference between the first and last 2 days within that treatment. HF, high food patch; LF, low food patch.
0.4570.25 (0.3770.29)
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
193
Table 3 Learning metric comparisons of the first 2 days and last 2 days (in parenthesis) of the post-switch feeding regime Learning metric
Exposure concentration (mg/L) Control
1.69
6.79
13.57
Percentage time in HF patch
34.976.08 (52.7716.0)
39.078.04 (45.579.61)
17.4713.8 (25.574.40)
30.2711.9 (38.7717.6)
Percentage of first choices for HF patch
13.379.72 (40.0718.7)
29.0710.1 (56.7711.3)
13.379.72 (36.7716.8)
28.3711.7 (25.3719.4)
Number of entrances into LF patch n
1.8570.45 (0.9070.34)n
1.2370.28 (0.7370.10)
1.7270.26 (1.7670.27)
1.4570.23 (0.98 7 0.26)
Significant difference between the first and last 2 days within that treatment.
Table 4 Muscle tissue Hg body burden (means7SEs) and acetylcholine esterase activity (means7SEs) after a 10-day waterborne exposure to HgCL2 immediately after exposure (pretrial) and after a 10-day behavioral trial during which time the fish were held in clean water (posttrial) Muscle tissue
Exposure concentration (mg/L) Control
1.69
Pretrial Hg (mg/g) AChE activity (as percentage of control)
0.1070.02 100.0070.44
0.5970.06 125.2370.54
Posttrial Hg (mg/g) AChE activity (as percentage of control)
0.0170.01 100.0070.36
0.5970.06 101.0270.32
food patch by the end of 6 days (Table 3) but, except for the control group, continued to spend more time on average in the low food patch, indicating a retention of previously learned information. T6 fish actually spent significantly more time in the low food patch (Po0:005) on the last two days. The percentage of fish choosing the new high food patch first was higher in the last two days of the postswitch regime compared to the first two days in all groups but T13 (Table 3). However, on the last two days of the postswitch trial, only T1 fish chose the high food patch more than 50% of the time, indicating that the former high food patch continued to be sampled first and that habitat information from the preswitch period was retained. Similarly, all groups continued to make entries into what had been the high food patch (Table 3); however, over the postswitch period, control fish showed a significant decline in visitation to the low food patch (two-sample t test, Po0:05).
3.1.2. Biochemical analysis The posttrial fish had 10 days in clean water but appeared to depurate very little Hg during that time period (Table 4). AChE activity was not found to differ significantly with treatment in pre- or postforaging trial
6.79 4.4070.61 93.82070.59
3.1570.27 102.9670.36
13.57 8.0270.33 109.4670.67
7.2770.63 100.0770.29
tests. T1 fish did have elevated, but nonsignificant, activity in the pretrial measurement (Table 4). Similar results were found for neurotransmitter levels (Table 5). In the pretrial measurements, DA and 5-HT levels were uniform across groups. T13 fish had nonsignificantly elevated NE levels compared with control fish, and l-Dopa levels differed nonsignificantly among treatments. However, the variation in l-Dopa did not appear to correspond to Hg body burden. With the exception of 5-HT, all neurotransmitter levels were similar in the posttrial groups.
4. Discussion The first section of the discussion attempts to explain the results of Experiments 1 and 2 by a mechanistic analysis of foraging behavior. Second, biochemical results are interpreted and compared with related literature. Finally, the results of these experiments are examined with regard to implications for aquatic systems in nature. 4.1. Experiment 1 As hypothesized for Experiment 1, foraging ability was related to Hg body burden. When compared on
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
194
Table 5 Neurotransmitter levels (means7SEs) immediately after a 10-day waterborne exposure to HgCl (pretrial) and after a 10-day trial period during which time the exposed fish were held in clean water (posttrial) Concentration (ng/mg) brain
Treatment concentration (mg/L) Control
1. 69
6.79
13.57
Pretrial NE l-Dopa DA 5-HT
0.41670.065 0.22770.059 0.06870.011 0.10470.021
0.37470.054 0.36570.098 0.08070.015 0.08570.015
0.40370.042 0.18670.049 0.07170.011 0.1107 0.020
0.51370.062 0.27670.104 0.07570.008 0.12270.015
Posttrial NE l-Dopa DA 5-HT
0.57170.047 0.39670.090 0.10770.011 0.15570.015
0.56570.050 0.40570.056 0.09170.010 0.11970.009
0.54970.031 0.39970.080 0.08870.008 0.11970.010
0.53770.036 0.37670.059 0.09770.016 0.11770.008n
n
Significant difference with control.
individual trial days, control fish captured more worms in less time. The results most often follow a dose– response pattern in which body burden is inversely related to test performance, although only comparisons with T6 and T13 fish yielded significant differences. Also, when comparing the best performance of individual fish, controls had significantly better mean performances than did the T13 group. The hypothesis that exposed fish would exhibit no improvement in foraging trials over trial days was confirmed in the case of T13 and, in some instances, T6 fish. Finally, it should be noted that all foraging trials took place while the fish were held in noncontaminated water, suggesting persistent effects after Hg exposure. Possible explanations for the foraging deficits found in this study would include appetite suppression in exposed fish, greater pause time exhibited by exposed fish, sensory impairment and treatment differences in learning to feed efficiently. As noted in the results, appetite suppression was not statistically evident in the pre- or posttrial tests, nor were their differences in the time spent in the feeding area during the foraging trials. Pause time (lack of directional movement), however, did differ among treatments, indicating potential sensory/central nervous system changes (Table 1). Moreover, prey search behavior, of which pause time is an important component, is likely to affect encounter rates. To investigate further, the 7-day average time to four captures was correlated against the 7-day average pause time during that time interval (Fig. 6). A significant inverse relationship between pause time and capture time was found (Po.0001). Furthermore, treatment comparisons revealed significant differences in the pause time until four encounters (control vs T13, Po0:01), suggesting that pause time contributed to treatment differences in foraging success (Fig. 7). Differences in
Fig. 6. The relationship between pause time and the time until four worms were encountered. Each data point represents an individual fish’s 7-day average time to four encounters and its 7-day average pause time until four encounters.
Fig. 7. Average pause times (means and SEs) until four worms were encountered. Bars with different letters are significantly different.
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
pause time could result from nervous system depression (lethargy). McKim et al. (1976) reported sluggishness in trout with body burdens of 5–7 mg/g but not in those with 3 mg/g; however, Olson et al. (1975) observed no lethargy in fathead minnows, although they had up to 11 mg/g Hg in muscle tissue (336 days exposure in 0.247 mg/L methylmercury). Treatment differences in learning did not seem to explain any differences in foraging ability. In Experiment 1, the fish were introduced to a novel habitat in which successful foraging required visual and locomotory accommodations for a vegetated environment. Also, prey items were placed in the same locations every day, making retention of prey location information potentially important to foraging success. However, after reviewing the daily trials, no relationship between pause time and trial day was found for any group, suggesting that pause behavior was not a learned response to the habitat. Nor did fish seem to develop a specific search pattern, as neither movement within the aquarium nor the spatial sequence of captures was similar over days. This would suggest a random search. Similar results were found by Marschall et al. (1989) using bluegill sunfish. 4.2. Experiment 2 The second experiment was designed to focus specifically on learning, and it was hypothesized that, compared with controls, (1) the exposed fish would show no discrimination between the high and low food patch while foraging, (2) over trial days, the exposed fish would exhibit no increase in their tendency to initially choose the high food patch, (3) the treatment fish would show no decline in visitation to the low food patch, and (4) all of those hypotheses would also be true with a change in patch food conditions. None of these hypotheses was confirmed, as the results showed no impairment of learning or retention due to Hg exposure. During the preswitch period, all groups spent more time in the high food patch and exhibited the same first patch choice and patch visitation behavior. After switching the relative food values of each patch, all groups continued to enter the former high food patch first and to spend comparatively more time (although less than during the preswitch period) there. These data suggest that all groups learned and retained habitat information and that, interestingly, no group made dramatic changes in habitat use despite the habitat change. The results are not that unusual in that many investigators report that fish frequently depart from mathematical optimality models of habitat use and patch switching behavior (Pyke, 1984). Such a departure may be even more likely for Experiment 2, in which there was low energy cost (short travel distance) associated with movement between patches.
195
Despite the results of the current experiments, Hg may disrupt the process of memory consolidation. One mechanism could be the inhibition of ACh release from neurons. Hg acts at the neuromuscular junction and the synaptic cleft, producing a weakened postsynaptic potentiation. Mercury, by altering ACh release, may impair memory formation because short- and long-term learning is facilitated by the induction of cellular changes that increase ACh release from the presynaptic terminals of neuronal circuits (Bekkers and Steven, 1990). Another possible mechanism is the destruction of granule cells in the hippocampus (Chang, 1979), a primary center for learning and memory in higher vertebrates. Comparing our results with prior toxicological learning studies with fish is difficult because very few have used Hg as a toxicant or examined learning tasks related to foraging. Most have used conditioned response tests in which fish were trained to associate a light and a shock. After successful training, the treatment fish have been exposed to Hg and retested. Authors have reported impaired retention at low (3–6 mg/L) waterborne exposures (Weir and Hine, 1970) and at body levels greater than 2.85 mg/g (Hartman, 1978). In an operant conditioning test (target strikes with food as reinforcement) involving goldfish, Salzinger et al. (1973) reported impaired retention at 6 and 10 mg/L HgCl2. Also, lead, which has toxicity mechanisms similar to Hg, may affect memory in fathead minnows (Weber et al., 1991). The results of past studies would suggest that Hg could disrupt learning and retention at very low doses, although drawing firm conclusions is difficult due to the small number of studies and their methodological differences. There are some possible explanations for the apparent lack of effect in the current study. First, previous conditioning studies tested fish during the exposure period, whereas in the current study the trials took place while fish were in clean water during a recovery period. Second, the mercury species may have been important. Compared with HgCl2, methylmercury (MeHg) passes the blood–brain barrier more readily (Chang, 1979), making it more neurotoxic, although it should be noted that some older learning studies did use HgCl2. Regardless of whether Hg harms memory, in the future, ecologically relevant tests of learning and decision making will be of much value. 4.3. Biochemical data 4.3.1. Acetylcholinesterase Disturbing a process as central as AChE function could result in widespread effects on the central and peripheral nervous system, including sensory, motor, and autonomic function. Interestingly, in our experiment, AChE activity exhibited no treatment effect, even
196
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
though some studies using laboratory (Suresh et al., 1992) and field collected fish (Shaw and Panigrahi, 1990) do suggest that Hg can alter AChE activity. The proximate cause could be a conformational change brought about by the binding of Hg to a sulfhydryl group on the cholinesterase. Another means could be the destruction of anticholinergic cells. Although the results presented here were incongruous with the limited literature, older studies used much greater exposure levels (often in the hundreds of mg/L). 4.3.2. Brain neurotransmitters In prior studies, although they are few in number, both MeHg and HgCl2 altered monoamine levels in brain tissue of fish (Zhou et al., 1999). However, these studies often used high-dose, low-duration exposures. Also, in the current study, analyses were of whole brain and therefore incapable of distinguishing local changes in neurotransmitter levels. This is crucial because production and metabolism of neurotransmitters vary by brain region in fish (Hornby and Peikut, 1990) just as they do in other animals. In fact, Hg toxicity in the brain does seem to be spatially dependent (Tsai et al., 1995). The lack of effect in the current study may have resulted from low-exposure level and duration and, due to the small size, the inability to examine the fish brain by specific anatomical region. Consequently, any local differences were too small to be resolved. Clearly, more work is needed to elucidate the interaction of Hg and neurotransmitters in fish. 4.3.3. Mercury body burden In understanding the environmental significance of the results reported here it is important to consider tissue levels because the waterborne exposure was greater than that generally found in polluted lotic and lentic systems (typically 10–40 ng/L), (Wiener and Spry, 1996). As would be expected, Hg body burden increased with exposure dose (Table 5). Control fish had between 0.1 and 0.012 ppm, levels typical in fish collected from nonpolluted environments lacking conditions that promote metal bioavailability, that is, low pH and warm temperature (Wiener and Spry, 1996). In aquatic systems that have natural mercury deposits or conditions favorable to availability, levels similar to those found in T1 fish would not be uncommon. In polluted areas, piscivorous or bottom feeding fish can contain mean tissue concentrations from 1 to 7 ppm depending on factors such as age and distance from point source. These values are equivalent to those reported for T6 and, although on the high end, T13 fish. However, mercury accumulation is partially a function of longevity (Ward and Neumann, 1999), therefore, short-lived species such as the fathead minnow would be unlikely to accumulate levels above 1 ppm in natural systems.
It is important to note, however, that most (90–95%) of the total mercury found in field-collected specimens is the most toxic form of mercury, MeHg, rather than the inorganic form used here (AMRL, 1998). Also, for the fathead minnow and species occupying similar trophic levels, benthic detritus often constitutes a significant dietary component. By feeding in substrates, the fathead minnow would potentially increase Hg uptake via gill or dietary routes. Finally, fish were tested during a recovery period in which body stores were the only source of exposure. So, despite the uncertainty inherent in any extrapolation from lab to field, the results reported in this study have potential implications for fish in nature.
5. Conclusion Traditionally, toxicologists approach pollution stress from a clinical reference point that emphasizes lethal dosages and cellular and molecular mechanisms of toxicity. Such studies require a controlled environment and, as a result, are laboratory based. One major drawback to this approach is that environmental relevance is often not clearly defined, and no attempt is made to relate alterations at the cellular/molecular level to higher levels of ecosystem organization (e.g., individual, community). In contrast, toxicity tests that examine foraging behaviors focus attention on variables whose environmental relevance is easily recognized. For example, in this study, Hg was found to decrease foraging efficiency, a component of bioenergetics models. General energy budgets such as Consumption ¼ Growth þ Egestion þ Ingestion þ Metabolism þ Work illustrate the direct relationship between growth and food consumption. The ingestion and work variables can be broken down into a variety of components, among them the foraging ability variables used in this study. For example, decreased foraging efficiency could mean that, for a given time period more energy is expended searching, while actual food consumption over that time period, is reduced. If fish are unable to acquire food efficiently, then the life history, reproductive, and (potentially) ecological implications are profound. Toxicology studies that use foraging bioenergeticsbased end points offer several possibilities for ecotoxicologists. Mathematical modeling of net energy gains from foraging could be performed using toxicantexposed and -unexposed fish. These models could then be used to predict pollution-induced alterations in growth, reproduction, and metabolism in the wild. Other mathematical models related to foraging
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
behavior, such as models of prey choice (e.g., optimal foraging theory) and habitat preference (e.g., ideal free distribution, patch departure modeling), could be also be examined. One very promising approach (Smith and Weiss, 1997) is to perform laboratory comparisons of foraging ability using fish collected from polluted and reference sites. This approach allows a degree of hypothesis testing using wild fish that have undergone realistic pollutant exposures. Finally, the purpose of ecotoxicology is to investigate the effects of environmental pollution on the ecology of natural systems, making laboratory investigations most informative if the results can be placed within an ecosystem context. Therefore, one would ultimately attempt to test predictions about how impaired feeding ability in individuals may influence higher levels of ecosystem organization. Testing of this kind would yield information on how contaminants alter organizational levels and linkages within ecosystems. It is only with an ecosystem level understanding that we can regulate the discharge of pollutants with full knowledge of their impacts.
Acknowledgments The authors thank Paul Angermeier and Thomas A. Jenssen for their suggestions and ideas that were essential to this project. The authors are also grateful to the laboratory of Neal Castagnoli for the use of the high-performance liquid chromatography and to Jaques Petzer for his aid in the neurotransmitter analysis.
References AMRL, 1998. A comprehensive evaluation of the South River and South Fork Shenandoah River for mercury contamination: final report. AMRL Technical Report No. 3079, Old Dominion University, Norfolk, VA. Atchison, G.J., Sandheinrich, M.B., Bryan, M.D., 1996. Effects of environmental stressors on interspecific interactions of aquatic animals. In: Newman, M.C., Jagoe, C.H. (Eds.), Exotoxicology: A Hierachical Treatment. CRC Press, Inc., Boca Raton, pp. 319–345. Beitinger, T.L., 1990. Behaviroral reactions for the assessment of stress in fishes. J. Great Lakes Res. 16, 495–528. Bekkers, J.M., Steven, C.F., 1990. Presynaptic mechanism for longterm potentiation in the hippocampus. Nature 346, 724–729. Chang, L.W., 1979. Pathological effects of mercury poisoning. In: Nriagu, J.O. (Ed.), The Biogeochemistry of Mercury in the Environment. Elsevier/North-Holland Biomedical Press, Amsterdam, pp. 519–580. Ehlinger, J.T., 1989. Learning and individual variation in bluegill foraging: habitat-specific techniques. Anim. Behav. 38, 643–658. Ehlinger, T.J., 1990. Habitat choice and phenotype-limited feeding efficiency in bluegill: individual differences and trophic polymorphism. Ecology 71, 886–896.
197
Elman, G.L., Courtney, D., Andres, O., Featherstone, R., 1961. A new and rapid colorimetric determination of AChE activity. Biochem. Pharmocol. 7, 88–95. Gotceitas, V., 1990. Foraging and predator avoidance: a test of a patch choice model with juvenile bluegill sunfish. Oecologia 83, 346–351. Hartman, A.M., 1978. Mercury feeding schedules: effects on accumulation, retention and behavior in trout. Trans. Am. Fish. Soc. 107, 369–375. Hornby, P.J., Peikut, D.T., 1990. Distribution of catecholaminesynthesizing enzymes in goldfish brains: presumptive dopamine and norepinephrine neuronal organization. Brain Behav. Evol. 35, 49–64. Hughes, R.N., Kaiser, M.J., Mackney, P.A., Warburton, K., 1992. Optimizing foraging behaviour through learning. J. Fish Biol. 41 (Suppl. B), 77–91. Jenkins, N., Burkehead, R., 1994. Freshwater Fishes of Virginia. American Fisheries Society, Bethesda, MD. Little, E.E., Fairchild, J.F., Delonay, A.J., 1993. Behavioral methods for assessing impacts of contaminants on early life stage fishes. Am. Fish. Soc. Symp. 14, 67–76. Marschall, E.A., Chesson, P.L., Stein, R.A., 1989. Foraging in a patchy environment: prey-encounter rat and residence time distributions. Anim. Behav. 37, 444–454. McKim, J.M., Olson, G.F., Holcombe, G.W., Hunt, E.P., 1976. Long term effects of methylmercuric chloride on three generation of brook trout (Salvelinus fontinalis): Toxicity, accumulation, distribution and elimination. J. Fish Res. Board Can. 33, 2726–2739. O’ Brien, W.J., Evans, B.I., Browman, H.I., 1989. Flexible search tactics and efficient foraging in saltatory searching animals. Oecologia 80, 100–110. Olson, G.F., Mount, D.I., Snarski, V.M., Thorslund, I.W., 1975. Mercury residues in fathead minnows, Pimephales promelas Rafinesque, chronically exposed to methylmercury in water. Bull. Environ. Contam. Toxicol. 14, 129–134. Pyke, G.H., 1984. Optimal foraging theory: a critical review. Ann. Rev. Ecol. Syst. 15, 523–575. Salzinger, K., Fairhurst, S.P., Friemark, S.J., Wolkoff, F.D., 1973. Behavior of the goldfish as an early warning system for the presence of pollutants in water. J. Environ. Syst. 3, 27–40. Salzman, S.K., Llados-Eckman, C.L., Hirofuji, E., 1985. Determination of spinal cord monoamines and metabolites using three micron columns and dual electrochemical detection. J. Liquid Chromatogr. 8, 345–360. Smith, G.M., Weiss, J.S., 1997. Predator–prey relationships in mummichogs (Fundulus heteroclitus (L.)): effects of living in a polluted environment. J. Exp. Mar. Biol. Ecol. 209, 75–87. Smith, J.R., 1984. Fish neurotoxicology. In: Weber, L.J. (Ed.), Aquatic Toxicology, Vol. 2. Raven Press, New York, pp. 107–152. Strickler-Shaw, S., Taylor, D.H., 1991. Lead inhibits acquisition and retention learning in bullfrog tadpoles. Neurotoxicol. Teratol. 13, 167–173. Sun, T., Taylor, D.H., 1983. The effects of parathion on acquisition and retention of shuttlebox avoidance-conditioning in the goldfish, Carassius auratu. Environ. Pollut. (Series A) 31, 119–131. Shaw, B.P., Panigrahi, A.K., 1990. Brain AChE activity studies in some fish species collected from a mercury contaminated estuary. Water Air Soil Pollut. 53, 327–334. Suresh, A., Sivaramakrishna, B., Victoriamma, P.C., Radhakrishnaiah, K., 1992. Comparative study on the inhibition of acetylcholinesterase activity in the freshwater fish Cyprinus carpio by mercury and zinc. Biochem. Int. 26, 367. Tsai, C., Jang, T., Wang, L., 1995. Effects of mercury and serotonin concentration in the brain of tilapia, Oreochromis mossambicus. Neurosci.Lett. 184, 208–211.
198
M.A. Grippo, A.G. Heath / Ecotoxicology and Environmental Safety 55 (2003) 187–198
US Environmental Protection Agency (USEPA), 1990. Test methods for evaluating Solid Waste (SW-846). USEPA. 1983. Method 7470, Rev. 1, Nov. 1990. Office of Solid Wastes, Washington, DC. Ward, S.M., Neumann, R.M., 1999. Seasonal variations in concentrations of mercury in axial muscle tissue of largemouth bass. N. Am. J. Fish Manage. 19, 89–96. Weber, D.N., Russo, A., Seale, D.B., Spieler, R.E., 1991. Waterbourne lead affects feeding abilities and neurotransmitter levels of juvenile fathead minnows (Pimephales promelas). Aquatic Toxicol. 21, 71–80. Weber, D.N., Spieler, R.E., 1994. Behavioral mechanisms of metal toxicity in fishes. In: Malins, D.C., Ostrander, G.K. (Eds.), Aquatic
Toxicology: Molecular, Biochemical, and Cellular Perspectives. CRC Press, Boca Raton, pp. 421–467. Weir, P.A., Hine, C.H., 1970. Effects of various metals on behavior of conditioned goldfish. Arch. Environ. Health 20, 45–51. Wiener, J.G., Spry, D.J., 1996. Toxicological significance of mercury to freshwater fish. In: Beyer, W.N., Heinz, G.H., Redmon, A.W. (Eds.), Environmental Contaminants in Wildlife. CRC press, Boca Raton, pp. 297–339. Zar, J., 1996. Biostatistical Analysis, 3rd Edtion. Prentice-Hall, Upper Saddle River, NJ. Zhou, T., Rademacher, D.J., Steinpreis, R.E., Weis, J.S., 1999. Comp. Biochem. Physiol C 124, 287–294.