THE POSSIBLE RISKS OF GENE TECHNOLOGY TO THE ENVIRONMENTAL HEALTH - THE IMPACT OF HERBICIDE RESISTANCE ON THE HERBICIDE USE IN SUGAR BEET CULTIVATION

Project leader: Juha Kämäri, Finnish Environment Institute (FEI) P.O.Box 140,
FIN-00251 Helsinki, Finland, tel. +358-9-4030 0474, e-mail: Juha.Kamari@vyh.fi
 
 

PUBLICATIONS

TIIVISTELMÄ SUOMEKSI

Researchers:
Sari Autio (FEI), Liisa Eronen (SBRC), Aarne Kurppa (MTT), Pirkko Laitinen (MTT), Leona Mattsoff (FEI), Jarmo Meriläinen (IER), Keijo Mäntykoski (IER), Seija Oinonen (MTT), Kyösti Raininko (SBRC), Seppo Rekolainen (FEI), Marja Ruohonen-Lehto (FEI), Sari Rämö (MTT), Katri Siimes (FEI) Leena Welling (IER) and Raili Venäläinen (NPHI)
FEI, Finnish Environment Institute, tel.+358-9-4030 00, e-mail: firstname.familyname@vyh.fi (after 15.4.2002 firstname.familyname@ymparisto.fi)
MTT, Agrifood Research Finland, tel.+358-3-4188 1, e-mail: firstname.familyname@mtt.fi
SBRC, Sugar Beet Research Centre, tel.+358-2-7708 200, e-mail: firstname.familyname@danisco.com
IER, Institute for Environmental Research, tel.+358-14-2603 830, e-mail: firstname.familyname@ymtk.jyu.fi
NPHI, National Public Health Institute / Department of Environmental Medicine, tel. +358-17-2013 51, e-mail: Raili.Venalainen@ktl.fi

Financing SYTTY organisation: The Academy of Finland
Funding from SYTTY / Total funding of project (€): 165059 / 354523
Person-months of work funded by SYTTY / Total person-months of work: 54 / 130,4

KEY WORDS: herbicide resistance, sugar beet, environmental impact, gene technology, mathematical models
 

EXTENDED ABSTRACT

1 Introduction

Process of deciding whether genetically modified, herbicide resistant (HR) plants can be introduced to the market includes careful evaluation of potential risks posed on human health and the environment. Some risk factors have been recognised but lack of experimental data acquired in northern conditions has left us with insufficient basis for objective risk assessment. Cultivation of HR sugar beet will evidently lead to major changes in the treatment practices and dosage of two non-selective herbicides, glyphosate and glufosinate-ammonium. This may lead to changes in the herbicide concentrations in ground and surface waters. The potential pathways of herbicides include the surface runoff and erosion directly to surface waters or via percolation to groundwater.

The main objective of this project was to identify and quantify the risks to human health and the environment resulting from the major change in the herbicide use due to the cultivation of HR plants. The main health risk is caused by the possible contamination of surface and ground waters used as drinking water reservoirs. In addition, possible accumulation of herbicide degradation products in a crop plant may be harmful for the human health. Sugar beet was used as a model plant crop.

Field trials were carried out using both glyphosate and glufosinate-ammonium resistant sugar beet cultivars. Weeds were controlled using both conventional (metamitron, ethofumesate, phenmedipham) and non-selective (glyphosate, glufosinate-ammonium) herbicides. Persistence of herbicide residues in different depths was followed for two years. The adsorption coefficients for all five herbicides were determined for both field trial and five additional soil types. Moreover, hydrological parameters essential for simulations were defined. Mathematical simulations are presently underway and will be finalized in 2003. A comprehensive risk assessment will be carried out during year 2003.

2 Methods

2.1 Field trials and soil sampling
Field trials were carried out at two different sites with different soil types. Two plots were sown with glufosinate-ammonium (GA) resistant and two plots with glyphosate (GLY) resistant sugar beet. Three herbicide application schedules were used: one of the two GLY resistant sugar beet plots was treated twice in 1999 and three times in 2000 with GLY and one of the two GA resistant sugar beet plots was treated twice in 1999 and three times in 2000 with GA. The remaining two plots were treated three times per summer with a mixture of metamitron (MET), ethofumesate (EFU) and phenmedipham (PMP).

Soil samples for herbicide residue analysis were taken from six depths (0-3 cm, 3-8 cm, 8-30 cm, 0-30 cm, 30-50 cm and 50-70 cm) one to eight times per growing season. Soil samples were taken with augers and metallic cylinders were used to prevent blending of different soil layers. Climatic conditions were recorded at weather stations near (<2 km) the field trials. In addition, daily rainfall was measured manually at the plot sites. Soil moisture was recorded weekly at each plot during summer months. Number of weeds was calculated in each plot in July and visually estimated in August-September. Sugar beet yield was determined by harvest.

2.2 Herbicide residues and adsorption coefficients
To analyse the residues of MET, its metabolite desaminometamitron (DES),  and EFU in soil and sugar beet, samples were extracted with a mixture of ethylacetate and acetone. The extracted samples were dried, evaporated and diluted into acetonenitrile and quantified with UV-detector in liquid chromatography as described in Ezzell et al. (1995) and Richter et al. (1994).

Soil samples for residue analyses of GLY and its metabolite AMPA (amino-methyl phosphonic acid) were extracted as described in Spann and Hargreaves (1994). Soil samples for GA residue analyses were extracted with water according to Sancho et al. (1996). Plant samples were extracted and purified, with minor modifications, according to Alferness and Iwata (1994). The fluorescent derivatives of GLY and AMPA were prepared with o-phthalaldehyde (OPA) using a post-column instrument and were detected with high performance liquid chromatography according to Pickering Laboratories (1995/1996). Pre-column derivatives of GA were prepared with 9-fluorenylmethyl-chloroformiate (FMOC-C1) according to Sancho et al. (1996) and were detected with high performance liquid chromatography (Sancho et al. 1994).

Moreover, parallel analyses of GA and in addition, analyses of MPP (3-methylphosphinico-propionic acid) and MPA (2-methylphosphinico-acetic acid), metabolites of GA, were carried out by Aventis CropScience.

The adsorption coefficients for all five herbicides (GLY, GA, MET, PMP, EFU) were determined for five different soil types in 1998 and for field trial soils in 1999 with two herbicide concentrations according to the OECD guidelines (1981) with minor modifications for GLY and GA. Samples of MET, PMP and EFU were purified and concentrated with solid-phase extraction (Michaels et al. 1991; Schlett 1991). MET, PMP, EFU and fluorescent derivatives of GLY and GA (Sancho et al. 1994) were detected with high performance liquid chromatography and diode array and fluorescence detector, respectively.

2.3 Modelling
A priori simulations were carried out in the beginning of the project using two available models: GLEAMS (Leonard et al. 1987) and PESTLA (van den Berg and Boesten 1998). Three pesticide applications were simulated for each herbicide. The dates and applied amounts were obtained from preliminary field trials in 1998.

Model selection showed that none of the existing models was sufficient to be used alone (Siimes and Kämäri 2002). The best models for leaching can not be used in estimation of surface losses. The main weakness of the models was inadequate description of winter hydrology and herbicide adsorption. Soil freezing was not described in any of the models. Most of the models assumed that all pesticides adsorb into soil organic matter only. This is not the case with the five studied herbicides.

For final simulations two to five models have been selected. The Swedish MACRO 4.1 model (Jarvis and Larsson 1998) was chosen for estimation of leaching and drainage losses of herbicides. The American model Groundwater Loading Effects of Agricultural Management Systems GLEAMS version 3.0 (Knisel and Davis 2000) was chosen for surface loss and leaching estimations. The other highly regarded models were RZWQM, PEARL and PELMO (see Siimes and Kämäri 2002).

3 Results and discussion

3.1 Field trials
Weather conditions differed significantly during trial years 1999 and 2000. Summer 1999 was extremely dry while conditions in summer 2000 were average when compared to long-term measurements. This evidently had an effect on herbicide behaviour in soil and efficacy as well.

In autumn 1999 weeds covered as much as 90%/90% of GA treated, 70%/60-70% of MET, PMP and EFU mixture treated and 10%/40% of GLY treated plots of clay soil and sandy soil, respectively. In sandy soil annual weeds were controlled more effectively than in clay soil and weed coverage in autumn was mainly caused by perennial weeds. As expected, this had an overall effect on sugar beet yield. Yields were lowest in GA treated and highest in GLY treated plots, respectively. A third herbicide treatment would have been necessary in these dry conditions.

In summer 2000 annual weeds were controlled more effectively than the year before, especially in clay soil. However, perennial weeds were increased prominently in sandy soil plots treated with the traditional herbicides. In autumn 2000, weeds covered 10%/21% of GA treated, 10-13%/33-37% of MET, PMP and EFU mixture treated and 15%/2% of GLY treated plots of clay soil and sandy soil, respectively. Highest root and sugar yields were achieved with GLY treatments.

3.2 Herbicide residues and adsorption coefficients
MET and its metabolite DES were detected only in the top 8 cm soil in both soil types. It was not detected after harvest. During summer months EFU was more mobile in sandy than clay soil. However, after harvest it was detected in the tillage layer (0-30 cm), but not below, of both soil types. Also PMP was more mobile in sandy soil and was detected in 50–70 cm depth after harvest in autumn 1999, apparently resulting from preferential flow in the dry conditions, but not in autumn 2000. GA and its metabolites MPP and MPA were only detected in the top 8 cm soil in both soil types.

GLY was the most mobile of the five studied herbicides. During both trial years GLY and its metabolite AMPA were detected in the tillage layers (0-30 cm) of both soil types already in July. In the autumn 2000 GLY was detected in 50-70 cm depth of both soil types but AMPA was detected only in clay soil. The reason for this could be e.g. the translocation of GLY in the plant from leaves to root tips as described e.g. Geiger et al. (1999).

The values of adsorption coefficients (Kd) of GLY, GA, EFU and PMP were on average smaller in field trial soils than in soils determined in 1998. This indicates that the herbicides have less adsorption sites and are more mobile in the soils of the field trials. The Kd values of MET did not vary between different analysed soils. The studied herbicides can be ordered according to their adsorption coefficients: GLY > PMP > EFU > GA ~ MET, GLY having the highest Kd value.

The adsorption coefficients of the five studied herbicides correspond quite well with the values available for these herbicides in literature, despite the typical properties of Finnish soils deviating to some extent from the more commonly tested Central European field soils, e.g. especially in the rather high content of organic matter and low pH, these properties being typical to Finnish soils. It is obvious that different sorption mechanisms were responsible for the adsorption of different herbicides, and therefore the applicability of Koc (adsorption coefficient corrected with the organic carbon content in soil) is questionable in the case of GLY and GA.

The possible effects of soil phosphorus, clay and organic matter on the behaviour of GLY and GA were studied in a sub-project.

3.3 Modelling
In all a priori simulations the main part of herbicides remain in the top soil during the first summer. In PESTLA simulation GLY and PMP stayed in the top 5 cm soil layer and GA in the top 15 cm soil layer during the whole simulation. EFU and MET stayed in tillage layer (30 cm) for the first simulation year. According to a priori simulation results EFU and MET had the highest leaching risk.

In the final simulations the degradation and leaching of herbicides were simulated by e.g. using the residue and adsorption data obtained in the field trials. So far only the leaching of EFU has been simulated using two models, the PEARL model and the MACRO 4.1 model. Both of these models could successfully be calibrated to observed herbicide concentrations. The MACRO 4.1 model showed simulated herbicide profile near observed with first leaching simulated in the end of year 2000, i.e. after the second year of cultivation. Leaching was estimated to exceed 0.10 µg/ha. In field trials GLY (highly sorptive compound) was found in deeper soil layers than EFU. This can not be simulated by any existing model suggesting that there are certain processes for GLY that need to be modified in the models to be used for final simulations. These adaptations of models are presently underway and are expected to be finalised during the year 2002.

3.4 Risk assessment
Environmental and agricultural assessment of the risks and benefits of genetically modified, HR sugar beet will be performed after empirical data has been analysed and simulation studies have been carried out. Also risks to human health will be assessed. This will be done in year 2003 and needs further funding.

4 Conclusions

For simulation model calculations there is a need for validation data in actual conditions, but such data have rarely been available from Finnish conditions. Therefore, our results provide a first set of measured data to be used in simulations.

In the sugar beet cultivation the sorption properties of the herbicides are very important especially as the first herbicide applications are necessary in the early growth stage of the sugar beet, when the crop interception is minimal and therefore major part of the  applied dose actually enters the soil directly. Furthermore, understanding the sorption properties of the herbicides in actual Finnish conditions makes it possible to compare the environmental impact of cultivation of conventional versus GM sugar beet varieties.

GLY was identified as the least mobile of the five herbicides according to its Kd value and GA as one of the most mobile, together with MET. Strikingly, results from field studies were quite the opposite. GLY was found in 50-70 cm depth but GA and its metabolites MPP and MPA were only detected in the top 8 cm soil. Possible reasons for this finding could be either the colloidal leaching of GLY by preferential flow after a few rainfall events or its translocation to root tips in HR sugar beet. However, this phenomena is not fully understood yet and needs to be confirmed with further studies.

Studies of leaching and fate of GLY and GA used in HR sugar beet cultivation still need considerable research effort.

5 Co-operation

Co-operation and information exchange have been started with the developers of HR products. Both GA residues and the residues of MPP and MPA, metabolites of GA, were analysed from soil and plant samples by Aventis CropScience company in Germany during 2000-2001

6 References

Alferness P & Iwata Y (1994). Determination of glyphosate and (aminomethyl)-phosphonic acid in soil, plant and animal matrices and water by capillary gas chromatography with mass-selective setection. J. Agric. Food Chem. 42: 2751-2759.

Berg F van der and Boesten JJ (1998). PEsticide Leaching and Accumulation model (PEST-LA) version 3.4. Wageningen, Netherlands, Winand Staring Centre for Integrated land, Soil and Water Research, Agricultural Research Department: 150.

Ezzell J et al. (1995). A Comparison of accelerated solvent extraction with conventional solvent extraction for organophoshorus pesticides and herbicides. LC-GC, vol. 13, issue 5.

Geiger DR et al. (1999). Causes of self-limited translocation of glyphosate in Beta vulgaris plants. Pesticide Biochemistry and Physiology 64: 124-133.

Jarvis N & Larsson M (1998). The MACRO model (version 4.1). Technical description. Http://130.238.110.134:80/bgf/Macrothm/document.htm.

Knisel WG & Davis FM (2000). GLEAMS, Groundwater Loading Effects of Agricultural Management Systems, Version 3.0, User manual. USDA, Southeast Watershed Research Laboratory, South Atlantic Area, Publication SEWRL-WGK/FMD-050199, revised 081500, Tifton, GA, USA, 191 p. Available at http://sacs.cpes.peachnet.edu/sewrl/.

Leonard A et al.  (1987). GLEAMS: Groundwater Loading Effects of Agricultural Management Systems, ASAE Paper No 86-2511. American Society of Agricultural Engineers.

Michaels K et al.  (1991). Bestimmung von Herbiziden in Böden und Wässer. Agribiol. Res. 44:37-48.

OECD Guideline for testing of chemicals 106. 1981. Adsorption / Desorption.

Pickering Laboratories: ”HPLC Ion-exchange (isocratic) for glyphosate using 5 mM KH2PO4 (K200)” Product catalog 1995/1996, Method 04.1.

Richter B et al. (1994). Single Laboratory Method Validation Report: Extraction of TCL/PPL (Target Compound List/Priority Pollutant List) BNAas and pesticides using Accelerated Solvent Extraction (ASE) with analytical validation by GC/MS and GC/ECD. Document 116064.A, Dionex Corporation. US.

Sancho J et al.  (1996). Rapid determination of glyphosate residues and its main metabolite AMPA in soil samples by liquid chromatography. Intern. J. Environ. Anal. Chem. 62: 53-63.

Sancho J et al. (1994). Rapid determination of glufosinate in environmental water samples using 9-fluorenylmethoxycarbonyl precolumn derivatization, large volume injection and coupled-column liquid chromatography. J. Chromatogr. A. 678:59-67.

Schlett C (1991). Multi-residue-analysis of pesticides by HPLC after solid phase extraction. Fresenius J. Anal. Chem. 339:344-347.

Siimes K & Kämäri J (2002).  A review of available pesticide leaching models: Selection of models for herbicide fate simulation in Finnish sugar beet cultivation. Boreal Environmental Research (in press).

Spann K & Hargreaves P (1994). The Determination of glyphosate in soils with moderate to high clay content. Pestic. Sci. 40:41-48.
 

[ Projects | Main Page ]