More field relevant ecological assessments, apart from single species tests using standard species, are needed to better predict agrochemical effects at higher ecosystems levels. Therefore, an experiment using a non-target aquatic community was used to test the hypothesis of the negative effect of a single pulse of copper sulfate on plankton abundance, structure, richness and diversity endpoints. Microcosms (20 l volume) were established during 21 days of experimentation, using six replicates for controls and with two concentrations of copper sulfate (High treatment, H: 20 µg Cu l−1; and Low treatment, L: 2 µg Cu l−1), both within the copper legal threshold following the Water Framework Directive (2000). The general lineal model found significant differences at the phytoplankton abundance endpoint at the end of the experiment, with an increase of phytoplankton abundance in L treatments dominated by the smaller cell size class. The principal response curve on zooplankton data, despite being insignificant, pointed out some dissimilarities in abundance and structure between treatments and controls: treatments showed lower abundances and were richer in cladocerans and copepods than the control microcosm where rotifers and nauplii were dominant. This indicated that trends change in community structure due to the effects of copper sulfate, and that even if the copper concentrations under study were within legal limits, they showed potential to induce changes in planktonic communities.
Agriculture provides food resources through what is considered an ecosystem service that only a healthy ecosystem can provide; therefore, both economic and ecological impacts derived from agriculture practices should be considered and ecological risks prevented. The Water Framework Directive (WFD) faces this concern in relation to water bodies, so it intends to assess, monitor and manage the ecological and chemical status of all surface and groundwater bodies. Nonetheless, small freshwater systems are often ignored and undervalued (Downing, 2010). This preconception should change in light of freshwater systems, from large lakes to small ponds, contributing both to regional diversity (Oertli et al., 2002; Downing, 2010; Gilbert et al., 2014) and to global cycles playing an important role, for instance, in carbon cycling (Downing, 2010). In the context of this study, aquatic systems can be impacted by agrochemicals such as copper sulfate (CuSO4) that is used to control the fungi Cycloconium oleaginum in olive tree cultivation.
Even though copper can be found naturally in different forms, it can be toxic in aquatic systems as Cu2+ (Lenwood et al., 1998). Water Quality Criteria (WQC), following the European WFD, established a maximum “legal” limit in waters of 40 µg l−1 of copper. However, copper WQC may not be safe enough for not-target species since it is mainly based on single species test, but little information about the effects on other taxa is used in the regulation process. In this sense, previous studies have also found that 40 µg l−1 of copper is not a safe limit for amphibians as Bufo bufo, B. calamita and Pelodytes ibericus (García-Muñoz et al., 2011) or for the plankton community (Del Arco et al., 2014); neither for marine fishes as Sparus aurata (Oliva et al., 2007). Impacts at lower hierarchical levels at individual levels (i.e. thought changes in morphological, physiological or biochemical aspect) can translate into impact at higher hierarchical levels at community levels (i.e. driving diversity loses ending on the loss of ecosystem services; Montes and Sala, 2007). So then, community studies aiming to increase knowledge of the toxic effects of concentrations, even within legal limits on a wider range of species, should be encouraged in order to detect sub-lethal, direct and indirect effects upon the aquatic systems.
The present study is the second in a series of experiments aiming to assess the effect of copper sulfate on plankton communities, both above and below legal limits. In the first study, we explored copper concentrations of 200 and 20 µg l−1, and a direct negative effect on the plankton community structure was found (Del Arco et al., 2014). Therefore, a further step is to understand how lower but still “legal” copper concentrations would affect the community. The present experiment was conducted to study copper sulfate pulses under 20 µg l−1 to test the hypothesis that treatments, within legal limits, would affect the aquatic community owing to both, sub-lethal and indirect effects that are not detected in single species tests used for legislation purpose.
Material and methods
Eighteen outdoor microcosms (n = 6, circular plastic bucket of 20 l volume) were assembled based on an adapted protocol from OECD (2006). Microcosms were filled with 18 liters of water and 5 cm of sediment. The filtered water came from a supply artificial pond (HUMEXPUJA, experimental wetland infrastructure in the University of Jaén, Spain). The homogenized sediment came from a natural freshwater wetland (Casillas wetland, UTM 30SVG1083 with a surface area of 2.7 ha [Ortega et al., 2006]). The microcosm experiment lasted for 3 months, from March to May 2012. Before treatment addition, a stabilization period of 7 weeks was used to develop the plankton and benthic communities from resistant structures present in the sediment. After that phase, the experiment started, consisting of three weeks with copper addition on day 7 (D7).
Microcosms were exposed to two concentrations of copper sulfate, High treatment (H): 20 µg Cu l−1 and Low treatment (L): 2 µg Cu l−1. Following the WFD, the Spanish national legislation established a copper WQC level of 40 µg Cu l−1 (BOE, 2011). Therefore, both treatments, H and L, fall within legal limits.
There were both controls and two treatments (n = 6). Nominal dosages of copper sulfate were added above the water surface in to the microcosms as a one-off pulse on day 7 (D7). After stirring, water samples were taken for direct analysis by inductively coupled plasma (ICP) mass spectrometry in order to confirm the target nominal concentrations of copper. In addition, water samples from the controls were analyzed to ensure no copper was present.
Physical chemical variables
Each microcosm was sampled on days 0, 7, 14 and 21 (D0, D7, D14 and D21), after the 7 weeks of stabilisation. Data on physical-chemical variables: temperature, pH, dissolved oxygen and conductivity were obtained using a field probe (YSI-556 MPS). At the same time, water samples (100 mL) were taken and transported in cold and dark conditions to the laboratory for alkalinity analysis which was measured using an 848 Tritino Plus device.
Phytoplankton and zooplankton endpoints
Phytoplankton and zooplankton were sampled weekly. Phytoplankton samples were analyzed using flow citometry and chlorophyll-a was measured (for methods description, see Del Arco et al., 2014). The filtered water was returned to the microcosm. Zooplankton was identified into different taxonomic practical levels (TPL) (Van Wijngaarden et al., 2005): Ostracoda, Copepoda, Cladocera and Rotifers. Moreover copepods were divided in two groups, nauplii (that include together calanoids and ciclopoids) and adults plus copepodites. The cladocerans were dominated by Ceriodaphnia sp., Alona sp. and Macrothrix sp. The abundance and richness of each TPL and the biodiversity (modified Shannon-Wiener diversity index: H´ = −∑pi log2 pi, using TPL instead of species) in all of the experiment were evaluated.
The generalized linear models analysis (GLM) was used for test differences between treatment and controls for all physical-chemical variables and taxon levels. Prior to analysis, data were tested for normality and homoscedasticity using the test of Shapiro-Wilk and Levene, respectively. Abundance data of zooplankton and phytoplankton were log (x+1) transformed. The analyses were carried out with the SPSS 19 computer program.
Moreover, in order to evaluate the community response to treatments, a Principal Response Curve (PRC) analysis was used to assess the main community response to the treatments along the experiment (CANOCO software package, version 4.5) (Van den Brink and Ter Braak, 1999). PRC is a multivariate technique recommended to analyse complex changes in community structure over time under a treatment exposure in micro/mesocosms experiments (European Commission, 2002; Sanderson et al., 2009). It is based on the abundance response of TPL along the experiment in each treatment. The null hypothesis implies that the PRC analysis does not show the treatment effects on the community (Frampton et al., 2000).
Average exposure concentrations (AEC) fit target nominal copper concentrations of 2.63 ± 0.83 µg l−1 for L treatment, and 28.10 ± 2.46 µg l−1 for H treatment.
Physical-chemical average values along the experiment were highly similar among all microcosms (Table 1). GLM (univariate analysis) of physical-chemical parameters did not show significant differences either, control and treatments or between treatments (p > 0.05).
Figure 1 shows the phytoplankton abundance dynamics throughout the experiment in each sampling day. General lineal models (GLM) shows that there is a significant difference in the last week (D21) in terms of the phytoplankton abundance mean values between H, L and C treatments (F = 36.403; p < 0.001) (Figure 1; Table 2). A post hoc Tukey test denotes that the abundances of phytoplankton in L treatments were different to controls and H treatments. In phytoplankton size classes, marginal statistical significant differences were found in the ultraplankton size class at the end of the experiment between C and L treatment (F = 3.161; p = 0.072; Figure 2). However, no differences were detected in chlorophyll a concentrations.
Zooplankton abundance and community structure were similar on D0 before exposure (Figure 3), so then any differences after the exposure would be related to the treatments. TPL richness and diversity (modified Shannon-Wiener index) values were not statistically different among controls and treatments; neither along nor at the end of the experiment (F = 2.259, p-value = 0.9620) even though total abundance decreased as concentrations increased (Table 2). PRC diagram (Figure 3) shows the overall response of community structure and TPL abundance changes owing to each treatment along the experiment. 1D-plots represent the weight of the TPL in the overall response of the community: Positive TPL weight mean that the TPL are likely to follow the PRC response, negative TPL weight show the opposite trend; and, the response of TPL weight near zero is not related to the main response shown by PRC (Van den Brink and Ter Braak, 1999).
The experimental design intended to expose microcosms to equal outdoor environmental conditions. It was achieved since there were not significant statistical differences in physical-chemical variables; consequently, the copper treatments were the only discriminating factor.
Phytoplankton and its size distribution are recognized as an important ecological attribute of aquatic ecosystems (Rodríguez and Li, 1994; Guerrero and Castro, 1997). The results obtained in this study showed that significant statistical differences were found in phytoplankton abundance at the end of the experiment (D21) between C and H treatments in respect to L treatments; resulting in an increase of phytoplankton abundance in L treatment. It is surprising that phytoplankton abundance in L treatments differ from the control but not the H treatments; this could be explained by the intermediate disturbance hypothesis (Connell, 1978). Phytoplankton size classes shows marginally statistical significant differences in the ultraplankton size class between C and L treatment that supports the intermediate hypothesis, phytoplankton size classes based on Figure 2 shows that treatments differ, in this case based on size classes, specifically on the ultraplankton size class abundance. It is important also to note the quasi-extinction of the largest size class (nanoplankton) in H treatment (Figure 2). This is a typical characteristic of perturbed ecosystems, the general pattern of reduced size in stressed communities (Kerr, 1974; Pérez et al., 2010) that in our case is developed towards an increase in the pico and ultraplankton size class and a reduction in the abundance of the nanoplankton. Similar results have been obtained in previous studies, indicating the effects of a wide range of abiotic factors on the relative contribution of nanoplankton (Reynolds, 1984; Rojo and Rodríguez, 1994). By contrast, chlorophyll-a concentrations have not shown statistical differences between control and treatments (p > 0.05). It is well known the important role of the nanoplankton in the chlorophyll-a concentration in many aquatic systems, with values over 50% of the total chlorophyll-a concentration (Rodríguez and Guerrero, 1994). As nanoplankton varies in the experiment between treatments by a factor of 16, it was expected to obtain chlorophyll-a statistical differences between them. Effects of copper on chlorophyll-a cell concentrations have been reported (Rosko and Rachlin, 1977; Sunda et al., 1995). Sunda et al. (1995) assessed the effect of copper on different phytoplankton species (Thalassiosira pseudonana, Thalassiosira oceanica and Emiliania huxleyi) finding a decrease in cell chlorophyll-a concentration under copper exposure. It could partially explain the absence of chlorophyll-a differences in our experiment despite the high different factor of nanoplankton. In fact, even if L treatment has a higher nanoplankton abundance than H treatment by the end of the experiment (Figure 2), its chlorophyll-a content was lower (Table 2) although not statistically different. Chl-a may not be the most appropriate endpoint for studies using natural assemblages of phytoplankton (Pérez et al., 2010). In addition to these results, it is important to point out the relevance of seasonal variability in the sensitivity of natural communities assemblages to a toxicant exposure (Winner and Owen, 1991; Pérez et al., 2010). It may explain the different phytoplankton abundance response observed under the treatment of 20 µg l−1 in the winter experiment of Del Arco et al. (2014) and the present study. In the first, the 20 µg l−1 treatment resulted in a lower abundance respect to the controls, while in the present study no differences were found in comparison to the controls. Such differences may be driven by two main factors: different phytoplankton community assemblages of different seasons and by different irradiation intensity between winter and summer. Twiss et al. (2004) studied the sensitivity of phytoplankton under different Cu enrichment treatments at community levels (Lake Superior, Canada) finding the influence of light levels on phytoplankton responses through the photon flux mechanisms of the photosynthetic activity under the presence of copper.
Zooplankton abundance and the community structure (Figure 3, PRC) have not shown significant statistical differences between control and treatments. However, it gives information about community structure change signals which can be used to preview potential taxa shifts and indirect effects on the phytoplankton. Similar results were obtained when the effects of the antibiotic Monensin on zooplankton communities in aquatic microcosms were studied by Hillis et al. (2007). Even though the PRC was not significant, it was considered obvious a negative effect of the antibiotic at the greatest treatment concentration. In the present study, the PRC analysis showed a tendency of the treatments towards a different community structure compared to the controls that were related with the observed changes at TPL endpoints as abundance, richness and diversity (Table 2). The abundance of rotifers and nauplii decreased as the copper treatment concentration increased; while adult copepods, cladocera and ostracoda increased (Table 2). Those three TPL mark the differences on community structure, richness and diversity between controls and treatments.
Focusing on richness and diversity, surprisingly, both are higher in the treatments than in the controls at the end of the experiment (Table 1). This kind of response agrees with the intermediate disturbance hypothesis (Connell, 1978; Menge and Sutherland, 1987; Hanazato, 2001). This temporal increase of diversity is disturbance-dependent and would reverse (Connell, 1978, Huston, 2014); therefore it cannot be seen as a positive result prompting stable coexistence in the long-term. In fact, it can be interpreted as the result of copper disturbing the internally-driven progression of a community structure driven mainly by competitive-exclusion mechanisms (Connell, 1978; Reynolds, 1995). In the control microcosms, under this specific environmental and phytoplankton availability, rotifers and nauplii dominated the community. However, the disturbance in the treatments, resulting on phytoplankton community structure changes, may have disrupted competitive exclusion prompting the reported increase of diversity and the first signals of community shift. Therefore, even if exposure concentrations are within legal limits and the impact is not drastic, there is the negative effect of copper concentrations.
The observed community shift could be mainly related to food edibility. The above-mentioned reduction in the nanoplankton size class in H treatment coincides with the lower abundance of rotifers and nauplii compare to C and L treatment. It may indicate a community structure change related to grazing pressure disruption. Therefore, it raises two important aspects: a) the potential occurrence of long term effects owing to an indirect effect impacting trophic relationships and, b) the need to perform prolonged experiments in order to catch such potential effects. Indirect grazing pressure disruptions have been found in previous mesocosms studies (Van den Brink et al., 2000; Stampfli, 2011). For instance, Van den Brink et al. (2000) studied the effect of carbendazim on zooplankton and phytoplankton communities and found an increase of phytoplankton owing to a reduced grazing pressure from zooplankton depletion by the fungicide. This indicates that the addition of one pollutant may cause indirect top-down effects (Baird and Burton, 2001) that cannot be detected in single species tests and short-term studies.
In addition, changes in the abundance of rotifers and nauplii marked the decrease of zooplankton in the treatments, while cladocerans and copepods abundance were higher in the copper treatments. It contradicts the most common general order of sensitivity from lower to higher sensitivity of rotifers > copepods > cladocerans (Hanazato, 1998; Relyea, 2005). It also leads us to think of an indirect effect related to changes in phytoplankton structure and grazing pressure that may temporarily favor the most sensitive taxa to the toxicant. For instance, Gui and Grant (2008) studied the combined effect of food availability and chemical exposure on Drosophila melanogaster, and concluded that a release from competition on food resources could counterbalance the chemical impact. This community structure disruption may intensify over time. Therefore, a zooplankton abundance decrease may follow two main paths in the long term: (a) it may recover and have similar values to control. This support that copper causes no effect as GLM and PRC suggested owing to its lack of significance, or (b) it may keep a decreasing tendency suggesting copper long term effect related to community structure changes. A longer experimental period may have helped to discern among those two potential paths. A post-treatment of 21 days could not have been enough to capture the effect on the complete life cycle of the organisms; nevertheless it has been enough to be able to warn about agrochemical induced changes.
Taking into account the richness and diversity modified index, the slight increase in the treatments highlights the importance of direct and indirect effects. Increases of diversity after a chemical pulse have been previously observed. For instance, Hanazato (1997) reported a species richness increase of zooplankton in ponds treated with insecticides as a consequence of competition interactions alterations. In our experiment, the increase of diversity is mostly linked to the survival of ostracoda and the better performance of copepods and cladocerans. The last mentioned, could be potentially related to a release of competition pressure due to the decrease of other dominant competitors taxa in the treatments as rotifers and nauplii. It may suggest that at taxa levels, apart from the initial diversity, an important role to face perturabions is also played by the indirect effects. In this case: due to changes in phytoplankton size classes abundance, due to the intensity of perturbation and by the timing of population and ecosystem responses (Downing and Leibold, 2010).
The first experiment results of this series of studies (Del Arco et al., 2014) agree with the current results. In both experiments, concentrations within legal limits of 2 µg Cu l−1 were explored. There are two main common changes: (a) a zooplankton community changes together with a decrease of abundance as copper concentration increases, and (b) a phytoplankton community changes towards small size classes. A direct comparison of both experiments is inappropriate owing to differences in seasonality and in the developed community's structure. However, the identification of common patterns suggests that these results are robust.
Considering our hypotheses, even treatments within legal limits could negatively affect the aquatic community; that it is observed in phytoplankton community. The zooplankton community has been negatively affected at structural levels suffering a shift tendency in community composition and an abundance decrease was clearly visible in the H treatments. Impacts in L treatments are less detectable but raising concern about long term effects. In addition, as expected, indirect effects have taken place as the phytoplankton community structure reflected; phytoplankton abundance increased owing to low herbivore pressure. This highlights the importance of more complex tests than single species ones to be able to detect these indirect impacts on the community structural changes.
The legal threshold of toxic substances in aquatic systems is established based on single species test using standard species, little information is known about how such legal concentrations will affect the rest of the taxa in more complex ecological scenarios. This study highlights that even a single copper sulfate pulse within WQC would have negative impacts on the plankton aquatic community.
We would like to thank Francisco Márquez for providing fieldwork devices and José Robles and Baltasar Deutor for their valuable technical advice regarding copper analysis. Thanks to Consejería de Medio Ambiente de la Junta de Andalucía for permissions to take samples in Casillas wetland.
This research has been partly supported by a grant from the University of Jaén (Spain) to Ana Isabel Del Arco Ochoa and the research group of Ecología y Biodiversidad de Sistemas Acuáticos (RNM–300, Spain).