The São Domingos AMD is expected to eliminate the most sensitive genotypes of aquatic organisms living in the Chança River Reservoir near the affluent entrance (hereafter designated as impacted site), especially after rain events promoting runoff of mine tailings. Indeed, a simulation of a 3% AMD pulse on D. longispina laboratory populations at the carrying capacity, comprising five differentially AMD-resistant (multilocus allozyme) genotypes with similar frequencies, cultured in the laboratory for more than 500 generations, resulted in the elimination of the two most sensitive genotypes, and the most resistant one increased its frequency from 20 to approximately 50% 32.
Genetic erosion at the impacted site could also be enhanced by the emigration of the most sensitive genotypes. The avoidance response of 12 clonal lineages of D. longispina was quantified in a constant gradient of copper (3–87 µg/L) 33. Copper, instead of AMD, was chosen due to various reasons. First, the pH of AMD dilutions is very unstable. Second, among all measured metals, levels of copper showed the greatest differences (also zinc) between reference and impacted sites, which during the main rain season, were one to two orders of magnitude greater at the impacted site (188 vs 2.8 µg/L 25). Third, the resistance to lethal levels of AMD and of copper was found to be correlated (median clonal lethal concentration of copper vs median clonal lethal time at 3% AMD: r = 0.85, p < 0.001, n = 24 34). Avoidance from copper by sensitive clonal lineages was found to occur much faster and at lower concentrations than by resistant lineages 33. The median clonal avoidance time from the highest copper concentration (87 µg/L; range: 24–594 min) and the median clonal avoidance copper concentrations after a 12-h exposure (range: 65–110 µg/L) versus the median clonal lethal copper concentrations after a 48-h exposure (range: 60–369 µg/L) were found to be correlated (r = 0.63, p < 0.05, n = 12 and r = 0.74, p < 0.01, n = 12, respectively) 33. Taking into consideration the literature on the spatial active avoidance from contaminants by zooplankton 33, 35, 36, there are no reasons to doubt the possibility of emigration by D. longispina individuals from the impacted site to less contaminated areas of the Chança Reservoir.
A sensitivity to lethal levels of AMD by D. longispina living in the Chança River Reservoir near the affluent entrance (impacted site) lower than the sensitivity of those from nearby reference sites was first demonstrated by Lopes et al. 37. The authors showed that time-to-death of reference organisms exposed to several AMD contaminated water samples was significantly lower than that of those collected at the impacted site. Furthermore, it was demonstrated that resistance (i.e., the genetically determined component of tolerance) to lethal levels of AMD was responsible for the differences in sensitivity. Therefore, AMD was proven to act as a directional selective pressure resulting in microevolution, despite the counteraction of two processes of genetic diversification. First, the continuous gene flow (immigration from adjacent nonimpacted sites at the Chança River Reservoir) could have reestablished the original genotypes' frequencies. Second, this reestablishment could also have been achieved through the hatching of new genotypes from ephippial eggs, which are produced by sexual reproduction. Although Lopes et al. 37 remarked cautiously that “it is not possible to conclude if this (microevolution) took place through the elimination of sensitive individuals (genetic erosion) or through the spread of new genes (through mutation) or new genotypes (through sexual recombination),” their results appear to indicate that the most sensitive D. longispina genotypes were absent at the impacted site. This may be the case because when exposed to AMD-contaminated samples, reference individuals started to die well before any mortality was observed among those collected at the impacted site.
The genetic erosion hypothesis was further supported in a study explicitly addressing the possible elimination of sensitive genotypes 38. The sensitivity to lethal levels of copper for 265 clonal lineages, 136 from a reference site and 129 from the impacted site, was characterized 38. This characterization was done after removing environmental influences on tolerance (i.e., previous acclimation to elevated levels of metals and/or phenotypic plasticity) by keeping laboratory cultures for at least 15 generations under noncontaminated conditions. The most sensitive clonal lineages to lethal levels of copper were absent at the impacted site, but extremely resistant lineages were present at both the reference and impacted site 38. Therefore, microevolution at the impacted site took place through the elimination of sensitive individuals (genetic erosion) and apparently not through the spread of new genes or new genotypes conferring an enhanced resistance. In a later study with the same species, approximately 360 individuals from each of the two references and the impacted site were collected, and their sensitivity (time-to-death) to a lethal concentration of copper (250 µg/L) was evaluated 39. The 1,072 organisms were distributed into five categories of sensitivity, with the least tolerant organisms almost absent at the impacted site (0.6%), whereas the highly tolerant were represented at both reference sites (43 and 60%). Again, the most tolerant organisms were not exclusively found at the impacted site 39. However, the near absence of very sensitive organisms at the impacted site could also be accounted for by processes other than genetic erosion, namely acclimation and/or phenotypic plasticity, because individuals were maintained under laboratory conditions, without the contaminant pressure, for just 6 h.
After removing environmental influences on tolerance, another study on microevolution, looking at the same species and locations, investigated the time-to-death of at least 20 clonal lineages from each site—reference and impacted—exposed to lethal levels of copper (25 and 75 µg/L, respectively) 25. Only sensitive and resistant lineages to Cu were present at the reference and contaminated site, respectively, thus supporting the genetic erosion hypothesis.
Another daphnid species—Ceriodaphnia pulchella—was also targeted for evolutionary ecotoxicology studies at the same location 27. Originating from 15 egg-bearing females from each of two references and from the impacted site, 20 neonates from each site, after being kept in noncontaminated medium for five generations, were exposed to a locally collected AMD-contaminated water sample and time-to-death was recorded. The existence of genetically determined sensitivity differences to AMD between references and the impacted site was attested. Genetic erosion was considered the probable cause for those differences because the frequency of the most sensitive organisms at the impacted site (16%) was much lower than at the references (73 and 100%) 27.
Beside copper, the comparison of D. longispina resistance to lethal levels of zinc (median clonal lethal time values in a 96-h exposure to 1.5 mg/L) between a reference and the impacted site was also investigated 25. No evidence of genetic erosion was found because both populations had a similar distribution pattern of sensitivities to zinc. Differences in resistance to lethal levels of copper (addressed above) but not of zinc, between the reference and the impacted site, were considered likely to be related to differential selective pressures previously experienced 25.
Given the above, evidence of genetic erosion in the present case study has been gathered using resistance to lethal levels of AMD and copper as selectable traits. However, the AMD is expected not only to be lethal to the most sensitive organisms but also to be a sublethal directional selective pressure. Several of the aforementioned studies addressed genetic erosion looking at sublethal markers in D. longispina, but decisive results could not be obtained. Supporting evidence was found with the reproductive output and the body growth in copper exposures, after removing environmental influences (at least five generations under controlled conditions) by Lopes et al. 37. Specifically, after an exposure to a 60 µg/L copper solution in artificial water and to the latter as control, until the release of the fourth brood (12 replicates per concentration), organisms originating from egg-bearing females collected at the impacted site released as many neonates as the controls, while this copper concentration significantly inhibited the reproductive output of organisms originating from the reference site 37. In the same experiment, both 15 and 30 µg/L copper solutions inhibited body growth of reference organisms but not of impacted site organisms 37. On the contrary, no support was obtained either for reproduction in an AMD exposure 37 or for feeding in copper exposures 25, 38. The former study found that a moderately toxic AMD-contaminated water sample (T6 in Lopes et al. 37) depressed reproduction both of organisms originating from two references and from the most impacted site with an equal intensity (40–43% inhibition) 37. The resistance to lethal copper levels of 136 and 129 clonal lineages from a reference and the impacted site, respectively, was investigated in the latter study 38, as explained above. Feeding rates of each clonal lineage exposed during 24 h to the respective highest nonlethal copper concentration were quantified. In none of the categories of resistance to lethal levels of copper, organisms from the impacted site ate more than those from the reference. Years later, feeding rates of clonal lineages from the impacted site in 24-h exposures to 0.5 and 5 µg/L of copper were found to be no higher than those from the reference 25. Opposite results would be expected if genetic erosion at the impacted site had eliminated those individuals most sensitive to sublethal levels of contamination.
Metallothionein basal levels and their induction by copper as potentially selectable traits were also quantified, but differences between resistant and sensitive genotypes were not found 40. Furthermore, the variation in 20 putatively polymorphic enzyme systems was studied in 24 clonal lineages to evaluate the possible relationship of allozyme profiles with the differential resistance to lethal levels of copper and AMD 36. The lack of such a relationship and the fact that the clones were not representative of each site (reference vs impacted) disallows any speculation on possible genetic diversity differences between sites 36.
In the present case study, D. longispina individuals inhabiting the impacted site are a very small fraction of the Chança River Reservoir population; therefore, microevolution through random genetic drift would not be expected. Furthermore, even large AMD amounts (produced after very strong rainstorms) would not be enough to drastically reduce the population size through a (genetic drift) bottleneck effect. These improbable situations would be the ones for which the biomonitoring of neutral markers has been shown to be an effective tool to detect and quantify genetic differentiation 17, 21, 41. Furthermore, mutations due to genotoxicity of contaminants can mask a possible decrease in genetic variation evaluated using neutral markers 23.
To evaluate the impact of AMD on the genetic diversity and structure of the D. longispina inhabiting the impacted site using amplified fragment length polymorphisms (AFLP) was the main objective of the study by Martins et al. 39. From 1,072 organisms, 31 from each of two references and the impacted site were selected, for AFLP analysis in a stratified random sampling design. Nei's genetic diversity observed heterozygosity and Shannon's information index revealed no differences between each pair of sites (Kruskal-Wallis: p ≥ 0.7), even though differences in time-to-death were evident (Nei´s index and median lethal time values of the two references vs the impacted site: 0.38 ± 0.08 and 0.33 ± 0.16 vs 0.37 ± 0.16, and 1 h 30 min and 6 h 17 min vs 22 h 58 min, respectively) 39. More recently, genetic diversity at the reference and impacted sites using 22 clonal lineages of D. longispina and a panel of eight microsatellites was preliminarily evaluated 42. The results from this investigation did not even slightly point to a reduction at the impacted site. Advantages and limitations of using AFLP, microsatellites, and other molecular techniques within the framework of assessing genetic erosion were addressed by Van Straalen and Timmermans 23.
In summary, the present case study illustrates a situation where directional selection is a main factor of evolution; therefore, genetic erosion is detected by monitoring suitable selectable phenotypic markers, but where genetic drift is probably irrelevant or masked by other microevolution factors, especially gene flow. Therefore, neutral markers may fail to reveal the loss of genotypes in the present AMD-impacted case study, which may also be due to the possible genotoxicity of the AMD leading to mutations. However, some caution should be taken when choosing the selectable traits to be monitored. For example, sublethal responses may be insensitive to detect loss of genotypes.