Acute and additive toxicity of ten photosystem-II herbicides to seagrass

Photosystem II herbicides are transported to inshore marine waters, including those of the Great Barrier Reef, and are usually detected in complex mixtures. These herbicides inhibit photosynthesis, which can deplete energy reserves and reduce growth in seagrass, but the toxicity of some of these herbicides to seagrass is unknown and combined effects of multiple herbicides on seagrass has not been tested. Here we assessed the acute phytotoxicity of 10 PSII herbicides to the seagrass Halophila ovalis over 24 and/or 48 h. Individual herbicides exhibited a broad range of toxicities with inhibition of photosynthetic activity (∆F/Fm′) by 50% at concentrations ranging from 3.5 μg l−1 (ametryn) to 132 μg l−1 (fluometuron). We assessed potential additivity using the Concentration Addition model of joint action for binary mixtures of diuron and atrazine as well as complex mixtures of all 10 herbicides. The effects of both mixture types were largely additive, validating the application of additive effects models for calculating the risk posed by multiple PSII herbicides to seagrasses. This study extends seagrass ecotoxicological data to ametryn, metribuzin, bromacil, prometryn and fluometuron and demonstrates that low concentrations of PSII herbicide mixtures have the potential to impact ecologically relevant endpoints in seagrass, including ∆F/Fm′.

s-triazine and uracil), all have the same mode of action and their combined effects are considered additive for a variety of freshwater 24,[43][44][45] and estuarine microalgae 23 . The Concentration Addition (CA) model of joint action is valid for multiple PSII herbicides as this combines the concentration and potency of each component to calculate the expected total toxicity of a mixture 46,47 . A common approach to test the applicability of CA is to apply Toxic Unit (TU) values to the herbicide concentrations that induce the equivalent toxicity. For example, the concentration which inhibits Δ F/F m ' by 50% (IC 50 ) = 1 TU and is different for each herbicide 23,45 . Concentration-response curves of herbicide mixtures containing a range of TU values can then be used to validate CA for combinations of herbicides in a mixture (see Methods section).
Since the values of the World Heritage listed GBR are based on the fitness and survival of foundation, habitat-forming species such as seagrass, ecotoxicological data for these species should be included in future risk assessments and for the derivation and assessment of future water quality guidelines 48 . Here we assess the acute toxicity of 10 PSII herbicides (Table 1) individually and in mixtures to the tropical seagrass species, Halophila ovalis. The acute effects of individual herbicides on photosynthetic performance (∆F/F m ') of isolated leaves using a miniature bioassay was assessed in concentration-response experiments over a 24 and/or 48 h exposure period(s) in a static system 8 . The miniature bioassay methodology was also applied to examine the toxicity of PSII herbicides in binary and complex mixtures. Data from this study will broaden the relevant ecotoxicity data to include a range of alternative and emerging PSII herbicides and validate the additive toxicity of PSII herbicide mixtures on seagrass for application in monitoring programs and guideline development.

Herbicide
Chemical class Log K ow Water solubility (mg l −1 ) CAS number  Table 2. After 24 h, diuron was the most potent of the herbicides, exhibiting the lowest IC 50 of 4.3 μ g l −1 ( Table 2). Fluometruon with an IC 50 of 132 μ g l −1 was the least potent of the herbicides tested   (Table 3).

Discussion
Phytotoxicity in non-target plants, such as seagrass, has been documented previously for the PSII herbicide diuron in several studies 8,12,13,49 and its effects in chronic exposures lead to both declines in stored energy in the root-rhizome complex and whole-plant effects, including reduced growth and survival 13 .
Here we extend the toxic threshold (IC 10 ) and comparative toxicity data (IC 50 ) for inhibition of photosynthesis (∆F/F m ') in H. ovalis to a further nine PSII herbicides, and this matched dataset includes the first ecotoxicological information for ametryn, metribuzin, bromacil, prometryn and fluometuron for any seagrass species. Confirmation of additive toxicity of binary and complex PSII herbicide mixtures to H. ovalis further validates the importance of additive ecotoxicological effects (when the mode of action  is the same) for application in field monitoring, water quality guideline development and in ecological risk assessments.

Herbicide potencies.
The PSII herbicides demonstrated a wide range of potencies with diuron being most toxic (IC 50 = 4.3 μ g l −1 ) and all other herbicides exhibiting IC 50 s < 30 μ g l −1 except fluometuron which was four-fold less toxic than all other herbicides after 24 h ( Table 2). All of these herbicides bind to the same site in the D1 protein 19 and differences in potency are likely due to the diverse stearic, and lipophilic properties of the herbicides, where herbicides "fit" and form different covalent attachments with the protein 30 . We previously demonstrated even uptake and binding of diuron through the leaf surface of H. ovalis using Imaging-PAM fluorometry and no flooding of the vascular system via the cut stems of isolated H. ovalis leaves 8 . Herbicides with different structures and hydrophobicity are likely to be transported through the leaf and to and from the binding site at various rates, potentially contributing to less rapid impacts of ametryn, metribuzin, prometryn and hexazinone (Table 2). PSII herbicides must cross the hydrophobic semi-permeable cell membrane of the cell in order to successfully inhibit photosynthetic function, and absorption may be more difficult for less lipophilic herbicides 19 such as hexazinone. These slow acting herbicides here are all related s-triazines or triazinones, but the group exhibits a wide range of water solubilities and lipophilicities ( Table 1). This study provides the first seagrass phytotoxicity data for fluometuron, ametryn, metribuzin, prometryn and bromacil, and builds on limited toxicological data for atrazine, hexazinone, simazine and tebuthiuron to tropical species (Table 4). H. ovalis was generally more sensitive to many of these PSII herbicides when compared to other species groups (Table 4), though with some exceptions. Atrazine for example inhibited ∆F/F m ' at similar concentrations in H. ovalis as for other seagrass species in 3 day exposures of potted plants 12 , but at lower concentrations (i.e. greater sensitivity) than green algae 45,50 or coral 28,51 (Table 4). H. ovalis was also more sensitive to simazine than green algae 45 , coral 28 and diatoms 52 . Despite differences in sensitivity of ∆F/F m ' inhibition between species to the same herbicide, these differences were usually within an order of magnitude due to the well conserved binding site on the D1 protein in PSII 53 . Differences in experimental conditions including temperature 29 , light levels 8 and exposure time 13 between studies are also likely to affect apparent toxicity, highlighting the need for strictly controlled and repeatable experimental procedures in phytotoxicity studies.
Application to water quality guidelines. Ecotoxicity threshold values (ETVs) developed specifically for the GBR are intended to protect 99% of species in the World Heritage Area; however, these were developed from limited toxicity data 18 (Table 5). Inhibition of ∆F/F m ' is directly and quantitatively linked to inhibition of photochemical efficiency 54 and this in turn leads to reduced energy status and/or growth and mortality in seagrass following chronic PSII exposures 13,20,22 . Inhibition of (∆F/F m ') is also well correlated with reduced growth in microalgae 23,24 and energetics and reproduction in corals 25,26 and can therefore be considered ecologically relevant as a basis from which guidelines can be developed or assessed. Five of the herbicides registered for use in catchments of the GBR and tested here have no current guidelines; therefore, the matched IC 10 and IC 50 data (Table 2) provides valuable toxicity data as a contribution to risk assessments, interpretation of water quality monitoring and derivation of future guidelines. For some of the herbicides, greater than 10% inhibition of seagrass photosynthesis occurred at concentrations lower than current and proposed ETVs (Table 5).
Mixture toxicity. The overlapping concentration-response curves of all PSII herbicide mixtures and similarity between IC 50 s (TU atr+diu of 0.85 was only 6% and 11% lower than either TU diu+diu or TU atr+atr ) indicates additivity of herbicide effects on PSII activity in H. ovalis (Table 3). The small but significant difference between IC 50 values for [atrazine + diuron] and [atrazine + atrazine] indicated a potentially weak synergistic effect, but no differences between IC 50 for the 10-herbicide mixture (TU mix ) and either of the controls was evident, supporting overall additivity. These results build upon previous research demonstrating the validity of additive effects of PSII herbicide mixtures on photosynthesis with estuarine microalgae in the laboratory 23 and in microcosms 55 , and on cell division in the freshwater green algae Scenedesmus acuolatus for multiple complex mixtures of up to 18 s-triazines 45 . While Concentration Addition (CA) model of joint action is an appropriate approach for calculating total toxicity in mixtures of toxins with the same mode of action (such as PSII herbicides), alternative approaches should be applied to mixtures containing PSII herbicides and pesticides with other modes of action 56 . Contributions towards total toxicity by multiple PSII herbicides, each acting simultaneously at concentrations below individual guidelines can result in ecologically significant effects on aquatic organisms 45 and water quality guidelines based on single herbicides, even widespread and potent herbicides like diuron, could underestimate the ecological threat posed by herbicide mixtures. Concentration Addition has already been applied to compare the actual and expected additive phytotoxicity of field samples containing more than one PSII herbicide [57][58][59] . CA has also been applied to calculate total toxicity for complex mixtures of PSII herbicides in the field towards guideline reporting and risk assessments 16,17,33 and the current study validates this approach for PSII herbicides and ecologically important seagrass species. Matched ecotoxicity datasets like this one for multiple PSII herbicides are valuable, not only for comparing toxicities of individual herbicides but are critical for direct application in evaluating the total toxicity and risks posed by mixtures that are commonly observed in the environment such as the GBR and its catchments 16  Herbicides. Photosystem II inhibiting herbicides from four chemical classes (Table 1) were tested individually and in combination for their toxicity to seagrass. This selection of herbicides was based on application rates as well as contamination data in Queensland catchments adjacent to the GBR 14,16,33,41,42 .
The herbicide diuron was included as a reference toxicant 24 . All herbicides were purchased in the purest available analytical form (> 95%) from Sigma Aldrich. Individual herbicide solutions were prepared in 0.2 μ m filtered seawater using ethanol as a carrier (< 0.03% v/v). Nominal concentrations are reported as the herbicides are non-volatile, water solubility > 30 mg l −1 and octanol-water coefficient (log K ow ) < 4 making loss to adsorption on test vessels unlikely 61,62 . The measured seawater pH, salinity and oxygen concentrations in tests were 8.1, 34-36 psu and 7.0-8.5 mg l −1 respectively.
Miniature seagrass leaf assay. Assays were conducted in 12-well plates (Nunclon, Thermo Scientific), each containing 5 ml herbicide solution. Herbicide concentrations were randomized across all plates to minimise well cluster and potential plate effects 8 . Experimental light intensity was 100 ± 7 μ E (14:10 h light:dark cycle) and temperature maintained at 26 ± 2 °C for all assays. Fluorescence measurements were made with a MAXI Imaging-PAM (I-PAM) (Walz, Germany). Two fluorescence parameters were used to assess impacts of PSII herbicides on the seagrass leaves 8,20 . The effective quantum yield in an illuminated plant (∆F/F m ') provides an estimate of the efficiency of photochemical energy conversion within PSII under a given light intensity 54    reaction centres are open 54 and reductions in F v /F m indicate inactivation and/or photo-oxidative damage to PSII (chronic photoinhibition) 63 .
To quantify ∆F/F m ' , actinic light (100 ± 3 μ E) was applied within the I-PAM chamber for five minutes prior to the activation of the saturating pulse. Minimum fluorescence (F with illuminated samples) was determined by applying a weak modulated blue measuring light (ML setting of 5; 650 nm, 0.15 μ mol photons m −2 s −1 ). Light adapted maximum fluorescence (F m ') was determined using a short pulse (800 ms) of saturating actinic light (> 3000 μ mol photons m −2 s −1 ) and the effective quantum yield of PSII calculated from ∆F/F m ' = (F m ' -F)/F m ' . To quantify F v /F m , leaves were dark adapted for 30 min and F 0 and F m measured in the same fashion as F and F m ' to derive maximum quantum yields Inhibition of quantum yields (% inhibition relative to solvent control) was calculated from treatment data as

Screening.
A screening process was performed immediately prior to running the assays to ensure the leaves were in optimal condition for the experiment 8 . Second and third leaf pairs from the terminal, apical end of the rhizome were transferred to wells containing uncontaminated seawater. Leaves were dark adapted for 30 min and F v /F m was measured. Only leaves exhibiting F v /F m greater than 0.65 (indicating intact and efficient photosystem II apparatus) were used in the subsequent herbicide assays 8 . Average leaf length was 10.0 mm ± 2.5 and width was 4.8 mm ± 1.2.
Experimental duration and leaf health. F v /F m was measured at 0, 24 and 48 h to assess whether PSII remained intact and active 8 . The maximum fluorescence yield (F v /F m ) in uncontaminated solvent controls reduced by less than 8.5% over 24 and 48 h durations in all experiments, confirming that PSII remained intact and functional over the assay duration (one-way ANOVA p = < 0.05). Maximum inhibition of ∆F/F m ' in H. ovalis leaves by diuron is observed in less than 24 h 8 . Here, range finding exposures were performed for all other herbicides to determine whether maximum inhibition of ∆F/F m ' would be achieved following 24 or 48 h exposures. Leaves were exposed to high concentrations of each herbicide and the exposure duration to reach 95% steady state inhibition was recorded. Maximum inhibition was reached between 12 and 24 h exposure for all herbicides except hexazinone, metribuzin, prometryn and ametryn, which were reached within 48 h.
Concentration-response curves. Concentration-response curves were plotted by fitting four parameter logistic curves to the ∆F/F m ' inhibition data from nine replicate leaves for each concentration (SigmaPlot 11.0 and Graph Pad Prism V 6.0). Herbicide concentrations inhibiting ∆F/F m ' by 10 and 50% (IC 10 and IC 50 ) were determined from each curve by applying standard curve analysis. The probability that midpoints (IC 50 ) generated by the logistic curves were statistically different was tested by applying the F test in Graph Pad Prism V 6.0. IC 50 s were considered different when p < 0.05 and post-hoc results are presented for each comparison in the relevant results sections.
Mixture toxicity. Concentration addition (CA) was tested for (i) a binary mixture of [diuron and atrazine] (each 50% v:v) and (ii) a mixture of all [10 herbicides] (each 10% v-v). The Toxic Units (TU) concentration for each component was based on its IC 50 (= 1 TU) at 24 h calculated from the individual assays (Table 3). The bioassay was prepared and conducted in an identical way to solitary herbicide assays (see above). TU sum was calculated from corresponding TU values within the mixture (see Eq 1).