Ecological Risk Assessment of Tebuthiuron Following Application on Tropical Australian Wetlands

ABSTRACT

The present study assessed the ecological risks of the herbicide tebuthiuron to freshwater fauna and flora of northern Australia's tropical wetlands. Effects characterization utilized acute and chronic toxicity data of tebuthiuron to local freshwater species (three animals and two plants) as well as toxicity data derived from northern hemisphere species. Species sensitivity distributions (SSDs) for four effects scenarios—plant chronic toxicity (NOEC data), plant chronic toxicity (EC/IC₅₀ data), invertebrate and vertebrate chronic toxicity (NOEC data), and vertebrate acute toxicity (LC₅₀ data)—were used to characterize effects and calculate 10, 5, and 1% hazardous concentrations (HCs). Tebuthiuron concentrations affecting 5% of species (i.e., HC₅s) for the earlier scenarios were 0.013, 0.093, 9.0, and 97 mg L⁻¹, respectively. Exposure characterization involved the use of historical field monitoring data of tebuthiuron concentrations following application of tebuthiuron to a large infestation of the wetland weed Mimosa pigra (Mimosa). Tebuthiuron concentrations in surface water ranged from below detection to 2.05 mg L⁻¹ and were still measurable up to 10 months following application. A breakpoint regression model was fitted to the field monitoring data, providing a time-dependent estimate of exposure to tebuthiuron. Risk characterization involved the comparison of the SSDs and associated HCs for each of the effects scenarios, with the time-dependent model of tebuthiuron exposure. Modeled tebuthiuron concentrations over the first 12 days post-application were in excess of concentrations required to cause major (i.e., 50% reductions in population numbers) effects to over 85% of freshwater plant species (based on data for phytoplankton and floating macrophytes). Beyond this period and up to 300 d post-application, 10–20% of species were still predicted to be affected. To quantify the probability of prolonged effects, the plant SSDs were compared to a cumulative probability distribution of tebuthiuron measured from 70 d to 293 d post-application. The probability of at least 5% of freshwater plant species experiencing chronic effects due to tebuthiruon at ≥70 d post-application was 58% based on NOEC data and 8% based on EC/IC₅₀ data. Overlap of the 95% confidence limits of the exposure distribution and plant SSDs indicated substantial uncertainty in the risk estimates. Risks of effects to freshwater invertebrates and vertebrates were generally < 1%. It was concluded that tebuthiuron appears to represent a significant and prolonged risk to native freshwater plant species, particularly phytoplankton and floating macrophytes, whereas the risks to freshwater invertebrates and vertebrates appear low. However, from a management perspective, the risks of tebuthiuron (and other herbicides) must be weighed against the known, serious environmental and economic impacts of the target weed, Mimosa. Overall, the outcomes of the risk assessment support the various management options that have been implemented with regard to the use of tebuthiuron to control Mimosa.

Keywords:

• risk assessment
• tebuthiuron
• toxicity
• tropical
• wetlands
• Australia.

ACKNOWLEDGMENTS

The technical assistance and/or information and advice provided by the following colleagues is gratefully acknowledged: Dave Walden and Dr. Max Finlayson (eriss), Michael Storrs (Northern Land Council), Ian Brown, Graham Schultz, and Ian Miller (NT Department of Primary Industries and Fisheries), and Dr. Keith Solomon (University of Guelph). Finally, credit must also be given to the late Mr. Rod Ansell, a former Top End pastoral station manager who drew our attention to the potential environmental impacts of herbicides used for Mimosa control.

The Australian government owns the copyright to this manuscript. Reprintings in any form or medium are authorized by the copyright holder.

Notes

^a Data from Loh et al. (1980) and Caux et al. (1997).

^a Summarised from Caux et al. (1997).

^b ND: Not determined.

^a Summarized from Camilleri et al. (1998) and van Dam et al. (in press).

^b Represents an LC₅₀.

^a Geometric mean of three LC₅₀ values (115, 144, 126 mg L⁻¹; see Table 2)

^b LC₅₀ value actually reported as > 160 mg L⁻¹ (see Table 2).

^a Regression based on log-transformed “observed” vs. “model-predicted” NOEC values

^b Anderson-Darling (adjusted) Goodness of Fit statistic.

^a Values in parentheses represent 95% confidence limits (CLs).

^a Values in parentheses represent the model uncertainty around the risk estimates based on comparison of the confidence limits of the exposure and effects distributions (see text in CONSIDERATION OF UNCERTAINTY for details)