[Frontpage] [Contents] [Previous] [Next]

4 Comparison between different herbicide strategies in oilseed rape

4.1 Introduction

One of the keystones of these models is selection pressure from the herbicide on individual weed species/biotypes as a function of dose. The work presented here will later be used as input for a model that predicts herbicide use in a crop rotation with non-transgenic oilseed rape (the reference scenario), glyphosate tolerant oilseed rape and glufosinate tolerant oilseed rape (future scenarios). A similar study, conducted on weeds common to sugarbeet was presented in Madsen and Jensen (1995); however, these curves were not integrated into the presented model of a sugarbeet rotation (Chapter 6). This model uses only one rate per spraying per herbicide.

Traditional herbicides Traditional herbicides in Danish oilseed rape are currently under pressure. One reason is that herbicides commonly used in this crop are being withdrawn (benazolin), restricted in dosage (propyzamide) or banned for environmental reasons. The mentioned two herbicides accounted for 45% of the broad spectrum dicotyledon weed control in this crop (measured as treated area) in 1995 (Miljøstyrelsen, 1996). Oilseed rape is a rather competitive crop (with respect to weeds) and is therefore considered to be one of the less herbicide intensive crops which is reflected in a treatment frequencies (number of times a field is treated with the recommended dose) of 0.82-1.12 in spring varieties and 0.95-1.25 in winter varieties (Miljøstyrelsen, 1996; 1997).

Bioassay Here we present data from greenhouse (bioassay) experiments showing selection pressure from glyphosate or glufosinate on the individual weed species as a function of dose relative to a traditionally used herbicide mixture of benazolin+clopyralid. This mixture was chosen because it currently is the only foliar acting herbicide in oilseed rape (until 1997) which makes comparison of weed control strategies with the exclusively foliar acting glyphosate and glufosinate possible. Data are evaluated with the logistic dose-response model (Streibig et al., 1993) which allows for a comparison of dosages to obtain a certain efficacy level. A similar methodology is used to estimate the efficacy of the weed control on volunteers of oilseed rape in a succeeding cereal crop.

4.2 Materials and methods

Weed control - A

Four weed species were sown into square 637 ml pots (milk cartons cut of 13 cm from bottom) filled with a peat-sand mixture on 8 February 1996 and grown in the greenhouse. The pots were sub-irrigated when needed. Plants per pot were thinned 14 februar to 20 februar to 10 plants of Stellaria media, 6 of Chenopodium album, 10 of Viola arvensis and unknown of Poa annua (no thinning due to poor germination). Pots were sprayed on 26 February in a pot sprayer with a Hardi: 4110-14 nozzle applying 105 l ha^-1. Nine pots of each weed species in four replications were sprayed with glyphosate in a Roundup formulation (360 g a.i. l^-1) with 720, 360, 180, 90, 45, 23, 11, 6 and 0 g ha^-1. The remaining nine pots per weed species in four replications were sprayed with glufosinate in a Basta formulation (200 g a.i. l^-1) with 600, 300, 150, 75, 38, 19, 9, 5 and 0 g a.i. ha^-1. Plants were harvested on 11-12 March. Fresh and dry weight were recorded.

Weed control - B

Spring oilseed rape (variety: Global) and two weed species were sown into 8 l pots filled with soil-peat mixture on 9 May 1996 and grown in the greenhouse. Pots were automatically watered from above. The number of plants per pot was adjusted on 22-30 May to 20 plants of oilseed rape, 20 of Chenopodium album and 28 of Stellaria media. Plants were kept at 5° C from 3-5 June. Pots were sprayed on 6-7 June in a pot sprayer with a Hardi: 4110-16 nozzle applying 200 l volume ha^-1. Seven pots in four replications were sprayed with a commercial mixture of benazolin (500 g a.i. l^-1 formulation) and clopyralid (80 g a.i. l^-1 formulation) in a Benasalox SC formulation with 812, 406, 203, 102, 51, 25 and 0 g a.i. ha^-1, furthermore penetration oil (Agrirob) was added to the spray solution with 500 ml per ha. Another seven pots in 3 replications were sprayed with glyphosate in a Roundup formulation (360 g a.i. l^-1) with 720, 360, 180, 90, 45, 23 and 0 g ha^-1. The remaining seven pots in 4 replications were sprayed with glufosinate in a Basta formulation (200 g a.i. l^{- 1}) with 600, 300, 150, 75, 38, 19 and 0 g a.i. ha^-1. Plants were harvested and counted on 17-19 June. Fresh and dry weight were recorded.

Weed control - C

Three weed species were sown into 2 l pots filled with a soil-peat mixture on 25 juli 1996 and grown in the greenhouse. Pots were manually watered from above. Plants per pot were thinned on 5 August to 10 plants of Chenopodium album, Myosotis arvensis and Capsella bursa-pastoris. Pots were sprayed on 20 August in a pot sprayer with a Hardi: 4110-16 nozzle applying 200 l volume ha^-1. Seven pots of each weed species in four replications were sprayed on 20 August 1996 with glyphosate in a Roundup formulation (360 g a.i. l^-1) with 720, 360, 180, 90, 45, 22.5 and 0 g ha^-1. The remaining seven pots per weed species in four replications were sprayed with glufosinate in a Basta formulation (200 g a.i. l^-1) with 600, 300, 150, 75, 37.5, 18.8 and 0 g a.i. ha^-1. Plants were harvested and counted on 4 September. Fresh weight and dry weight were recorded.

Volunteer oilseed rape

Spring oilseed rape (cultivar: Global) was grown in monoculture, with spring barley (Alexis) or with winter wheat (Texana). The seeds were sown into 8 l pots with a soil-peat mixture 12 June 1996 and grown in the greenhouse. Pots were automatically watered from above. On 21-30 June plants were thinned to 9 oilseed rape plants per pot, and 12 plants per pot of the cereals, but half of the spring barley pots were thinned again due to regrowth and the rest were allowed a higher biomass. The plants were kept at 5° C from 26 June to 2 July. Pots were sprayed 4-5 July in a pot sprayer with a Hardi: 4110-16 nozzle applying 200 l volume ha^-1. Seven pots from each combination in 4 replications were sprayed with tribenuron-methyl in a commercial Express formulation (7.5 g a.i. tablet^-1) with 30, 15, 7.5, 3.8, 1.9, 0.9 and 0 g a.i. ha^-1. The remaining 7 pots from each combination in 4 rep. were sprayed with a commercial mixture of ioxynil (200 g. a.i. l^-1 formulation) and bromoxynil (200 g a.i. l^-1 formulation) in a Oxitril formulation with the following doses: 800, 400, 200, 100, 50, 25 and 0 g a.i. ha^-1. Plants were harvested and counted 18-19 July. Fresh and dry weight were recorded.

Statistical analyses

Fresh weight per pot (greenhouse) was analysed with the logarithm of the dose as predictor. The regression model of fresh weight (y) on dose (z) was a logistic curve:

where D denotes the upper limit of the curve, b is proportional to the slope around ED_50, which is the point of inflection (Streibig et al., 1993). Data were then standardized with the D-parameter to obtain relative fresh weight, and the regression was run again, now resulting in a D-parameter of approximately 100%. To stabilize the variance the Transform-Both-Sides technique was used with the optimal power of transformation ( ) (Carrol & Ruppert, 1988).

For practical weed control, the ED₈₅-value is of greater interest than the ED₅₀. The logistic model was reparameterized to calculate the ED₈₅-value as a parameter in the model (only greenhouse data):

LOF Lack-Of-Fit test (Weisberg, 1985) was used to determine if the model could adequately describe data.

Adjustment

The ED-values from the three greenhouse experiments were adjusted according to the ED₅₀-value from glyphosate treated S. media or C. album, to make the ED-values from the three experimental runs comparable, this resulted in almost identical curves for glufosinate treated counterparts which justifies the adjustment.

4.3 Results

Weed control

The model fitted reasonable to the data from the greenhouse, which is reflected in the confidence limits for the parameters (Table 4.1). However, the LOF-test was significant for M. arvensis and V. arvensis sprayed with glyphosate at the 5% level but not at the 2.5% level, furthermore data were very variable for S. media sprayed with benazolin and clopyralid and for oilseed rape sprayed with glufosinate. Glyphosate was more potent at the ED₅₀ and ED₈₅ level than glufosinate for C. bursa-pastoris, C. album, P. annua, however, this superiority is not statistically significant. For M. arvensis and V. arvensis, glufosinate was more potent than glyphosate; this difference was statistically significant at the 5% level for V. arvensis. The mixture of benzolin and clopyralid was less potent than glyphosate and glufosinate, however, the difference was only statistically significant for C. album at the ED₅₀ level (Figs 4.1, 4.2 and 4.3).

Ranking after sensitivity

The weed species can be ranked after their sensitivity to glyphosate or glufosinate in the greenhouse. At the ED₅₀ value of glyphosate C. bursa-pastoris > S. media > M. arvensis > C. album > P. annua > oilseed rape >V. arvensis. The ranking is almost identical at the ED₈₅ value C. bursa-pastoris > S. media > P. annua > C. album > oilseed rape > V. arvensis. For glufosinate the ranking at the ED50 value is as follows, M. arvensis > C. bursa-pastoris > V. arvensis > S. media > C. album > P. annua > oilseed rape, and at the ED₈₅ value V. arvensis > M. arvensis > C. bursa-pastoris > S. media > C. album > P. annua > oilseed rape.

Volunteer oilseed rape The LOF-test did not reject the hypothesis that the model could be used for evaluation of dose-response relationships where oilseed rape was growing in competition with different biomass of cereals. The only exception was the combination with oilseed rape in a normal biomass of barley, where LOF was significant at the 5% level but not at the 2.5% level. Tribenuron-methyl showed a significantly higher efficacy than the mixture of ioxynil and bromoxynil (Table 4.2), which is not surprising because tribenuron-methyl is considered to be a low-dose herbicide. However, there is no significant difference between different cover crop or cover crop biomass at the ED₅₀ or ED₈₅ level for tribenuron-methyl. The 85% control level of volunteer oilseed rape with tribenuron-methyl was obtained with doses from 0.53 to 1.28 g a.i. ha^-1 which is considerable less than the recommended rate of 7.5 g a.i. ha^-1 (Statens Planteavlsforsøg, 1997). This indicates that volunteer oilseed rape is highly unlikely to pose a problem in a rotation were the succeeding cereal crops are sprayed with a sulfonylurea herbicide, even though extrapolating dose rates from the greenhouse to the field is often misleading. The oilseed rape plants sprayed with the mixture of ioxynil and bromoxynil required 98.1 g a.i. ha^-1 for 85% control, this is also low compared to the allowed dose of 400 g a.i. ha^-1 for broad leaved weed control.

Integrated into model

The parameter values listed in Tables 4.1 and 4.2 are used in the oilseed rape model and these experimental data thus determine the selection pressure posed on the individual weed species present in the model.

4.4 Discussion

LOF Despite the fact that the LOF test showed lack of fit at the 5% level for three of the curves but not significant at the 2.5% level, we chose to apply the same logistic dose-response model to all data in order to obtain comparable parameters.

Field data

It would be useful to validate the relative potencies of the herbicides on different weed species with field data, however, at the time of the writing these data were not ready for being integrated in the model.

Potency

Glyphosate and glufosinate were both more potent than the traditional mixture of benazolin+clopyralid, which was expected from practical experience. In these experiments glyphosate and glufosinate seem equally potent on a mixed weed population. However, the plants were harvested after maximum effect at 2-3 weeks after the spraying. This means that regrowth of the weeds after the spraying was not included in the study, and a systemic herbicide (glyphosate) will suppress regrowth for a longer period than a contact herbicide (glufosinate). Therefore, glyphosate is probably more potent in controlling the weeds if the entire growing season is evaluated, however, this was evaluated in this study.

Greenhouse experiment. Adjusted regression parameters from regression of relative fresh weight on dose of glyphosate, glufosinate or a mixture of benazolin and clopyralid (95% confidens intervals in parenthesis). All results are based on Transform-Both-Sides transformation with =0.25.