Water temperatures in the enclosures were approximately 18–19 °C at the start of the experiment and during the metiram application period. Temperatures gradually declined from sampling day 17 onwards and the lowest water temperature measured in the enclosures was approximately 12 °C. Data on weather conditions during the experiment can be found in the Supporting Information.
Statistically significant changes in physico-chemical endpoints are presented in Table 3 and temporal trends are presented in the Supporting Information. A significant, but small, treatment-related increase in electronic conductivity was observed at the highest treatment-level on the days 10 and 17. Control enclosure pH values ranged between 7.2 and 9.0 and pH of control enclosures were significantly lower than that of treatment enclosures in the pre-treatment period (day −1) and immediately after first metiram application (day 3). However, on sampling days 10, 17 and 48, pH values showed a small, but statistically significant, treatment-related decrease and on day 59 there was a significant increase in pH although all deviations were less than 1 pH unit.
DO concentrations were relatively low in all enclosures during the application period, but were always higher than 4 mg/L. After day 17, DO levels increased to approximately 10 mg/L. A small but significant decline in DO was observed on day 17 in the 324 μg a.i./L treatment, and on day 48 DO levels were significantly higher in all treated enclosures relative to controls. All treatment-related differences in DO concentration were less than 1–2 mg/L.
Alkalinity values in control enclosures ranged between 1.08 to 1.24 mmol/L. On day 17, a small but significant treatment-related increase in alkalinity was measured while at the end of the experiment (day 59) a small but significant treatment-related decrease was observed.
At the start and the end of the experiment, nitrate/nitrite, ammonium, ortho-phosphate and total phosphate concentrations in depth-integrated water samples from the enclosures were below detection limits. On days −1 and day 59 measured concentrations of total soluble nitrogen ranged between 0.9–1.1 and 0.4–0.7 mg/L, respectively. Treatment-related effects on nutrient concentrations in the water column could not be demonstrated.
Metiram concentrations in the dosing solutions were on average 92.7 % of the intended concentration (range of 79.0–113.4 %), but concentrations in depth-integrated water samples collected approximately 2 h after the first fungicide application were only 36.6 % (range 16.0–65.1 %) of the initial concentration, highlighting the rapid disappearance of metiram from the water compartment. Unfortunately, the water samples collected 2 h after the second and third treatment were lost due to technical problems during metiram analysis (corrosion of the metal tubes of the measurement equipment).
Three days after the last application (day 17), the average concentration in water samples collected from the 324 μg metiram/L enclosures was 0.14 μg metiram/L (0.04 % of the initial concentration) and no metiram was detected in samples from the 108 and 36 μg metiram/L enclosures (<0.05 μg metiram/L). Average concentrations of the metabolites EU and ETU in day 17 water samples from the 324, 108 and 36 μg metiram/L enclosures, were 38.8 μg EU/L and 12.2 μg ETU/L, 15.6 μg EU/L and 0.6 μg ETU/L and 4.3 μg EU/L and 0.13 μg ETU/L, respectively. All other degradates analysed were below detection limits (i.e. <20 μg/L for HY; <0.2 μg/L for EBIS, carbimid and TDIT). At the end of the experiment (day 59 after the first treatment) the concentrations of metiram and all metabolites analysed were below detection limits (metiram < 0.05 μg/L; EU < 1.0 μg/L; HY < 20 μg/L; ETU, EBIS, carbimid and TDIT < 0.2 μg/L). These data illustrate that metiram dissipates very fast (estimated water dissipation DT50 of approximately 1–6 h) and that its metabolites are not persistent in the water compartment.
Of the 30 zooplankton taxa collected during this study, 23 were rotifers, three were cladocerans, three were copepods and one was an ostracod. The most abundant zooplankton taxa in decreasing order were: Anuraeopsis fissa (Rotifera), copepod nauplii (Copepoda), Polyarthra remata (Rotifera), Trichocerca gr. similis (Rotifera), Keratella cochlearis (Rotifera), Lecane gr. luna (Rotifera), Cyclopoida (Copepoda), Ceriodaphnia sp. (Cladocera), Trichocerca gr. porcellus (Rotifera) and Squatinella rostrum (Rotifera). The number of zooplankton taxa was significantly reduced relative to controls at the highest dose (324 μg a.i./L) 10 days after the first application (Table 4). However, a statistically non-significant decline in zooplankton richness was observed in the post-exposure period (days 17–24) in the two highest doses (Fig. 1a).
Multivariate PRC analysis indicated that the zooplankton community was significantly affected by exposure to metiram (Monte Carlo permutation test p = 0.009) with the rotifers Anureopsis fissa and P. remata being particularly negatively affected by the metiram application (Fig. 2a). Significant treatment-related effects on the zooplankton community were detected at the highest concentration (324 μg a.i./L) on day 3 and in the 108 and 324 μg a.i./L enclosures on days 10, 17 and 24 (Table 4).
At the population level, statistically significant differences between treatments and controls could be observed for 13 of the 30 zooplankton taxa, but for two of them these differences occurred in the pre-treatment period (Table 4) and consequently were not treatment-related. Results of univariate analyses of population data (Williams test, p < 0.05) are presented in Table 4 and temporal trends illustrated in Fig. 3.
Treatment-related effects on Rotifera total abundance were observed at the two highest treatment-levels (NOEC = 36 μg a.i./L) and started soon after first application. Full recovery was observed on day 31 (Fig. 3a). Similar treatment-related declines were observed for A. fissa from day 3 to day 24 at the two highest treatment levels (NOECpopulation = 36 μg a.i./L). After day 31 densities in controls declined to zero and differences in abundance between treatments did not show a clear concentration–response relationship (Fig. 3b; Table 4). P. remata was the most sensitive rotifer species with minor but significant declines observed at the 36 μg a.i./L treatment level (days 3 and 24) and clear treatment-related declines observed from day 3 to day 31 at 108 and 324 μg a.i./L, followed by recovery. Note, however, that the decline in the 108 μg a.i./L enclosures was more pronounced than in enclosures that received 324 μg a.i./L (Fig. 3c; Table 4). The abundance of the rotifers T. gr. similis and K. cochlearis declined in the highest treatment level from day 3 to day 17 and in the 108 μg a.i./L enclosures on day 17. After day 17, abundance of T. gr. similis declined in all enclosures, including the controls (Fig. 3d; Table 4). Significant declines in the abundance of K. cochlearis were observed in the 36 μg a.i./L enclosures on day 24 (Table 4). Treatment-related declines in abundance of the rotifer Scaridium longicaudum were observed on day 10 (NOEC = 108 μg a.i./L), day 17 (NOEC = 36 μg a.i./L) and day 31 (NOEC = 108 μg a.i./L) and at the highest treatment level this species was not detected from day 10 to day 31, followed by recovery (Table 4).
Effects of metiram on Copepoda total abundance were consistent, but small. Statistically significant treatment-related declines in abundance were observed from day 3 up to day 24 and on day 48 (NOECs of 108 μg a.i./L, except on day 24 when a NOEC of 12 μg a.i./L was calculated) (Fig. 3e; Table 4). Treatment-related declines in Cyclopoida abundance were observed from day 3 to day 31 at the highest treatment level (NOEC = 108 μg a.i./L), except on day 17 when a NOEC of 36 μg a.i./L could be calculated. Full recovery of Cyclopoida was observed after day 31 (Fig. 3f; Table 4). Copepod nauplii were abundant in all enclosures and minor, but statistically significant, declines were observed at the highest concentration (NOEC = 108 μg a.i./L) on days 3, 10, 24 and 48. On day 24, the calculated NOEC was 12 μg a.i./L (Table 4).
Total abundance of Cladocera was not affected by metiram (Fig. 3g; Table 4) and the only treatment-related response observed was for Alona sp. on day 48 when a NOEC of 108 μg a.i./L could be calculated (Table 4). All other populations of Cladocera did not show a treatment-related response. Ostracoda occurred at low densities in all enclosures and a significant increase in abundance was observed (Table 4; NOEC = 108 μg a.i./L) on an single sampling date (day 17).
Sixty-three macroinvertebrate taxa were collected from the enclosures, the majority of which were Insecta (34 taxa), Mollusca (10), Oligochaeta (6), Hirudinea (5), Turbellaria (5), Crustacea (2) and Hydracarina (1). Several of these taxa occurred in low densities and/or were observed on a limited number of sampling dates only. The most abundant macroinvertebrate taxa in decreasing order were: Dero sp. (Oligochatea), Chaoborus sp. (Insecta), Chironomini (Insecta), Mesostoma sp. (Turbellaria), Lumbriculus sp. (Oligochaeta), Orthocladinae (Insecta), Ceratopogonidae (Insecta), Caenis sp. (Insecta), Zygoptera (Insecta) and Dugesia lugubris (Turbellaria).
A small decrease in the number of macroinvertebrate taxa relative to controls could be observed on day 15 (a day after the third metiram application) in the enclosures that received the highest concentration (324 μg a.i./L) (Fig. 1b; Table 5). Treatment-related effects of metiram on the macroinvertebrate community could not be demonstrated by means of multivariate PRC analysis (Monte Carlo permutation test p = 0.83). Although statistically significant differences between treatments and controls could be observed for 15 of the 63 macroinvertebrate taxa, these deviations predominantly occurred on isolated sampling days (Table 5). The only macroinvertebrate taxon for which statistical significant differences were observed on two consecutive samplings in the post-treatment period (days 43 and 57) was Dytiscidae larvae, but this taxon occurred in low densities (always <5 individuals per sample) and the effect concerned a treatment-related increase. For the ephemeropteran Caenis sp. and the mollusc Gyraulus crista, a statistically significant decline in numbers was calculated on day 15 (immediately after the third application), but densities of both taxa were low in all enclosures (Table 5).
One hundred and nine phytoplankton taxa were collected during this study, the majority of which were Chlorophyta (49 taxa), Desmidiaceae (23), Cyanophyta (14), Diatomeae (10), Euglenophyceae (7), Chrysophyceae (3), Dinoflagellata (2) and Cryptophyceae (1). A limited number of taxa dominated the phytoplankton community and many taxa occurred in low densities and/or were observed on a limited number of sampling dates only. The most abundant phytoplankton taxa in decreasing order were: Volvox (Chlorophyta), Scenedesmus arcuatus (Chlorophyta), Tetraedron minimum (Chlorophyta), Pennales (Diatomeae), Pseudanabaenaceae (Cyanophyta), Phacotus lendneri (Chlorophyta), Aphanocapsa (Cyanophyta), Anabaena (Cyanophyta), Oocystis (Chlorophyta) and Aphanothece (Cyanophyta).
A small decrease in the number of phytoplankton taxa relative to controls was observed on day 17 at the highest concentration (324 μg a.i./L) (Fig. 1c; Table 6). There was little evidence of a treatment-related response in total chlorophyll a biomass with significant reductions only observed on day 31 at the highest concentration (Fig. 1d; Table 6).
PRC analysis demonstrated that metiram treatment did not explain a significant component of the variation in phytoplankton community composition (Monte Carlo permutation test p = 0.544). Nevertheless, statistically significant treatment-related effects could be calculated for 42 of the 109 phytoplankton taxa (not including the abundance of main taxonomic groups), although the vast majority of these taxa (37 out of 42) showed a statistical significant response on an isolated sampling day only and mostly concerned low density taxa (<10 individuals/ml). In addition, statistically significant responses related to both decreases (15 cases) and increases (27 cases) in abundance and were mostly observed in the highest treatment only (NOEC of 108 μg a.i./L) (Table 6).
The blue-green alga Anabaena sp. (Fig. 3h) was one of the few phytoplankton taxa that showed a clear treatment-related decline in abundance on two consecutive sampling days (day 17 NOEC = 108 μg a.i./L and day 24 NOEC = 36 μg a.i./L), followed by recovery (Table 6).
In the Supporting Information more detailed information is provided on the treatment-related responses on total abundance of the main taxonomic groups of algae, and on abundance of individual phytoplankton taxa that showed a statistical deviation on at least two consecutive sampling days, or on a single sampling day during the application period (day 3–17).
Biomass of macrophytes
Prior to the metiram application the above-sediment macrophyte biomass was estimated to be 58.6 ± 13.4 g dry weight per enclosure (geomean ± SD; n = 3). At the end of the study the above-sediment macrophyte biomass in control enclosures had increased to 79.8 ± 6.8 g dry weight (geomean ± SD; n = 4), but no significant treatment-related effects on above-sediment biomass could be observed in the treated enclosures when compared to controls (also see Supporting Information).
Microbial endpoints and alder leaf decomposition
Based on conidia abundance, the dominant aquatic hyphomycetes on pre-conditioned alder leaf material were Angillospora
longissima and Tetracladium setigerum. Whereas in controls the abundance score of A. longissima conidia generally increased during the course of the experiment (Fig. 4a), the abundance of T. setigerum conidia remained relatively low (Fig. 4b). For both species statistically significant treatment-related effects could not be demonstrated (William’s test, p > 0.05) despite the trend in lower abundance for T. setigerum in most enclosures that received metiram (Fig. 4b). A statistically significant effect of metiram on total fungal biomass (increase) associated with alder leaf litter could be observed on sampling day 3 only (William’s test, p < 0.05; NOEC 4 μg a.i./L; Table 7). This effect, however, did not show a clear concentration–response relationship (Fig. 4e). Mass loss of decomposing alder leaves increased during the experiment; mass loss in coarse mesh bags (Fig. 4c) increasing at a faster rate than mass loss in fine mesh bags (Fig. 4d). However, there was no significant treatment effect on mass loss in both types of litter bags (William’s test, p > 0.05).
PRC analysis indicated that sediment bacterial community structure differed significantly between control and metiram-treated enclosures whether expressed in terms of relative band density values of the DGGE profiles (Fig. 2b; Monte Carlo permutation test p < 0.05) or OTUs (Fig. 2c; Monte Carlo permutation test p < 0.05). Given that significant differences between control and treated enclosures were present pre-application (i.e. day −4, Table 7) they cannot be attributed to the metiram treatments. PRC analyses detected no significant effect of metiram application on the sediment fungal community structure (Monte Carlo permutation test p > 0.05).