Weed composition and herbicide use strategies in sugar beet cultivation in Germany

Unkrautzusammensetzung und Herbizidstrategien im Zuckerrübenanbau in Deutschland

2 Material and Methods

The NEPTUN-survey (Netzwerk zur Ermittlung der Pflanzenschutzmittelanwendung in unterschiedlichen, land­wirtschaftlich relevanten Naturräumen Deutschlands) was established and organised by the Julius Kühn-Institut (JKI former Biologische Bundesanstalt für Land- und Forst­wirtschaft) in 2000 to provide data on pesticide use, and thus, to increase the transparency of its use in Germany (Rossberg, 2006). The NEPTUN-surveys sugar beet were carried out in 2005, 2007 and 2009 in cooperation with the IfZ (Rossberg, 2006; Rossberg et al., 2008; Rossberg et al., 2010). The NEPTUN-survey 2009 contained 15 regions called ERA (Erhebungsregionen Ackerbau) in which pesticide use was surveyed from at least 30 farms per ERA (Rossberg et al., 2010). The parameters treatment time, pesticide product, application rate and treated area were recorded. More detailed information is given in the NEPTUN-reports (Rossberg, 2006; Rossberg et al., 2008; Rossberg et al., 2010).

Generally, the use of non-selective herbicides was classified into three groups differing in date and development stage of sugar beet when applied. Three categories were determined: Pre-sowing treatments in autumn, pre-sowing treatments in spring, and pre-emergence treatments after sowing in spring. In the NEPTUN-survey, all glyphosate applications after 1st September of the year previous to sugar beet cultivation were included. In the SBCS, glyphosate application was considered for the first time in 2000, but no classification was made. Additio­nally, the acreage with plough and mulch tillage systems was recorded and set into relation to the use of non-selective herbicides.

In the SBCS, herbicide treatments were classified into two groups, post-emergence treatments only and a combination of pre-/post-emergence treatments. In the NEPTUN-survey, the date of sugar beet emergence was not recorded. Therefore treatments in a period of up to seven days after sowing were regarded to be pre-emergence. All pre-sowing and pre-emergence treatments were aggregated to pre-emergence treatments in order to improve comparability between both surveys. In Fig. 2, glyphosate-containing herbicide products were included in NEPTUN-survey, only.

Fig. 2. Post- and pre-/post-emer­gence herbicide treatments in sugar beet cultivation, estimated (SBCS) and surveyed (NEPTUN) herbicide treated acreage in % of total acreage. Sugar Beet Cultivation-survey (SBCS), Germany 1996–2010 and NEPTUN-survey, Germany 2005–2009.

The number of herbicide treatments was analysed for the two categories of post- and pre-/post-emergence treatments as defined above (NEPTUN and SBCS).

The analysis of the number of herbicide products used per treatment (NEPTUN) based on 11,900 treatments resulting from 30,796 herbicide applications. Approximately 99% of all applications resulted from the first five treatments. Hence, data analysis and presentation included the first five treatments only, which led to more compact figures, as well as the use of the 5th/95th percentile. To illustrate differences in the rate of occurrence of herbicide treatments, proportions of fields with the corresponding number of treatments were shown. The number of herbicide products used was calculated from the number of different herbicide products used per field and treatment, while different products were counted separately, even when the active ingredients were identical.

The analysis of the number of active ingredients per treatment (NEPTUN) based on 11,900 treatments resulting from the application of added up 46,419 active ingredients. It was calculated by summing up identical active ingredients per field and treatment, while different herbicide products with identical active ingredients were counted once.

The treatment index (TI) describes the intensity of pesticides use in the period from harvest of the preceding crop till harvest of the sugar beet (Rossberg et al., 2002). In this study, the Treatment index_H refers to herbicides only, and was calculated separately for each field (1).

The treatment frequency (TF) is defined by the number of pesticide treatments conducted on a certain field, whereas site-specific applications were considered by acreage-coefficient. In this study Treatment frequency_H refers to herbicides only (2).

Herbicide use was assigned to four different herbicide dosage reduction classes which were classified by frequently used reductions of herbicide products referring to the authorised application rate (NEPTUN). Application rates of identical herbicide products were summed up for each field to 8,713 dosages and were assigned to the classes > 75%, 50–75%, 25–50% and < 25% of dosage-reduction relative to the authorised application rate.

The relative importance of each active ingredient was analysed by calculating both the proportion of applications including the respective active ingredient in relation to all applications, and the acreage treated with each active ingredients percentage of the total treated area (NEPTUN).

Herbicide costs were calculated according to the NEPTUN-data and the pesticide price list of BayWa AG, 2009 (Anonymous, 2009b).

Statistical analysis was conducted by the procedure 'proc glm' for analysis of variance followed by a multiple comparison of means according to Tukey (SAS Version 9.2, SAS Institute Inc., Cary, NC, USA).

Generally, weed infestation on regional scale was hete­rogeneous in occurrence and intensity (Table 2). Within exemplary region 1, common weeds dominated clearly over difficult-to-control weeds, whereas in exemplary region 2 and 3 the distribution was more balanced between common and difficult-to-control weed species.

Acreage of plough tillage system decreased from 89% in 1994 to 55% in 2010 (Fig. 1). Acreage of mulch tillage system increased proportionally from 11% to 45%, corres­pondingly the acreage of non-selective herbicides increased from 17% in 2000 to 45% in 2010.

Fig. 1. Tillage systems applied for sugar beet cultivation and use of non-selective herbicides, estimated distribution in % of total acreage. Sugar Beet Cultivation-survey (SBCS), Germany 1994–2010.

The proportion of acreage with post-emergence treatments estimated by the SBCS increased continuously from 80% in 1996 to 89% in 2010, while by the NEPTUN-survey post-emergence treatments were between 70% in 2007 and 80% in 2005 (Fig. 2).

By the SBCS, the number of post-emergence treatments remained relatively constant at approximately 3 (Fig. 3). Pre-/post-emergence treatments increased slightly from 2.9 in 1996 to 3.3 in 2010 (no glyphosate-containing herbicide products were included). In the NEPTUN-survey, post-emergence treatments averaged nearly 3.5 and pre-/post-emergence treatments approximately 4.9, where almost all pre-emergence treatments consisted of glyphosate-containing herbicide products.

Fig. 3. Number of herbicide post- and pre-/post-emergence treatments per field in sugar beet cultivation, estimated (SBCS) and surveyed (NEPTUN) values. Sugar Beet Cultivation-survey (SBCS), Germany 1996–2010 and NEPTUN-survey, Germany 2005–2009.

Across years, the 1^st treatment showed a significantly lower number of herbicide products compared to the 2^nd of about 2.5 and 3.0, respectively, while the 3^rd treatment remained at a similar level as the 2^nd (Fig. 4). The 4^th and 5^th treatments had a significantly lower number of herbicide products of about 2.2 compared to the other treatments. The differences in herbicide products per treatment between years were not significant.

Fig. 4. Number of herbicide products per treatment for the first five post-emergence treatments in sugar beet cultivation, mean/median and the 5th/95th percentile. Different upper and lower case letters indicate significant differences between years, and treatments within each year, respectively. (Tukey-test, p ≤ 0,05). No. = 11,684 treatments, NEPTUN-survey, Germany 2005–2009.

Across years, the 2^nd treatment showed the highest number of active ingredients per treatment of approximately 4.4 compared to the others, while the 1^st treatment remained on a similar level as the 3^rd of about 4.0 (Fig. 5). The 4^th and 5^th treatments showed a significantly lower number of active ingredients of approximately 3.0 compared to the other treatments. The differences in active ingredients per treatment between years were not significant.

Fig. 5. Number of active ingre­dients per treatment for the first five post-emergence treatments in sugar beet cultivation, mean/median and the 5th/95th percentile. Different upper and lower case letters indicate significant differences between years, and treatments within each year, respectively. (Tukey-test, p ≤ 0,05). No. = 11,684 treatments, NEPTUN-survey, Germany 2005–2009.

The Treatment index_H showed an increasing tendency for the 1^st, 2^nd and 3^rd treatments of about 0.5, 0.6 and 0.7, respectively (Fig. 6). The differences were significant in the years 2007 and 2009. Treatment index_H of the 4^th and 5^th treatments remained on a similar level as of the 1^st treatment. The differences in Treatment index_H per treatment between years were significant in 2005 and 2007.

Fig. 6. Treatment index_H for the first five post-emergence treatments in sugar beet cultivation, mean/median and the 5th/95th percentile. Different upper and lower case letters indicate signi­ficant differences between years, and treatments within each year, respectively. (Tukey-test, p ≤ 0,05). No. = 11,684 treatments, NEPTUN-survey, Germany 2005–2009.

On average of all dosages, 37% could be assigned to the dosage-reduction class 50–75% and 29% to the class 25–50%, followed by the class > 75% and < 25% with 20% and 14%, respectively, the differences were significant between all classes (Table 3). From 2005 to 2009 the proportion of the class 50–75% significantly decreased, however, the class < 25% and 25–50% significantly increased, while differences within the class > 75% were not significant.

Ethofumesate, phenmedipham and metamitron were applied on nearly the total acreage and desmedipham on about 70%, representing about 3/4 of all herbicide applications (Table 4). Approximately 1/2 of the acreage was treated with triflusulfuron, chloridazon and quinmerac. Proqaquizafop, fluazifop-P and quizalofop-P were used on < 20% of the acreage in order to control monocotyledons.

The treatment frequency_H of fields with pre-/post-emergence treatments differed between the observed ERA from 2.5 to 6.3, on average 5 treatments were done per field (Table 5). 2.9–4.5 treatments were done per field with post-emergence treatments only. The 1^st treatment was applied on average 15 days after sowing, varying from 12 to 21 days between the ERA followed by a mean treatment interval from approximately 12 days varying from approximately 8 to 15 days. On average, 2.6 herbicide products and 4.1 a.i. were applied per treatment, varying from 2.2 to 3.5 herbicide products and 3.6 to 4.8 a.i., respectively. This resulted in a treatment index_H of 2.05 on average, varying from 1.47 to 2.51. Herbicide costs were on average 213 €/ha, varying from 177 €/ha–251 €/ha. 35% of all dosages in 2009 could be assigned to the dosage-reduction class 50–75% varying from 27% to 43%, 33% to the class 25–50% varying from 25% to 51%. The dosage-reduction class > 75% and < 25% were assigned to 19% and 14%, varying from 8% to 26% and 10% to 20%, respectively.

Methodological critique concerning the quantitative analysis of surveys refers to size, distribution and mode of taking the sample. Experts estimation in SBCS leads to a good overview of many aspects with a more regional character e.g. weed infestation, for which an enormous amount of data is compulsory when evaluated field-specifically. Otherwise, precision might be lower for specific aspects like e.g. treatment frequency or treatment interval of herbicide use. These aspects were generated precisely by using results of the NEPTUN-survey on farm scale. For regional analysis of NEPTUN-data at least 30 samples were used compared to the SBCS where data was estimated by local sugar factory advisers. Estimation of experts might be of low reliability, particularly if catchment areas are relatively large. However, comparative analysis of many aspects of both surveys resulted in high coincidence (Buhre et al., 2011).

In the last decades, herbicide strategies were evolved towards split-applications and post-emergence treatments. Post-emergence treatments increased from 6% in 1980 to 62% in 1992 (Zink et al., 1994) and in the SBCS from approximately 80% in 1996 to 90% in 2010 (Fig. 2). The NEPTUN-survey even indicated post-emergence proportion > 98% of the acreage in 2005–2009 (Fig. 2) if gly­phosate use is not considered as it was done in SBCS. The use of pre-emergence treatments resulted in a higher total number of herbicide treatments per field of approxi­mately 5.0 (including glyphosate) compared to approximately 3.5 when applied solely post-emergence (Fig. 3). According to Schäufele (2000) and Kudsk (2002), on average 3–3.5 treatments were applied predominantly to control annual broad-leave species and to compensate less efficacy of post-emergence herbicides. Usually, 3 treatments led to a sufficient weed control efficacy > 95% if they were adapted and timed precisely to field-specific weed population (Bruns et al., 2008; Vasel et al., 2011).

Across the last 15 years, treatment patterns of herbicides were similar although spread of weeds like goosefoot (CHESS) and knotweed (POLSS) increased from 47 and 35% to 79 and 86%, respectively. The 1^st treatment was usually applied approximately 15 days after sowing, varying from 12–21 days, depending on weed species, weed emergence and weather conditions. Between the years, a similar trend was observed in number of herbicides and active ingredients used per treatment for the first five treatments. The 1^st treatment consisted of approximately 2.5 herbicides followed by 3, 3, 2, and 2 to the 5^th treatment (Fig. 4). 4 active ingredients were used in the 1^st treatment followed by 4.5, 4, 3, and 3 to the 5^th treatment (Fig. 5). Meanwhile the treatment index_H of the first three treatments (Fig. 6) increased from 0.5 to 0.7, which could be explained by more weed species in later developmental stages of the sugar beet as well as less sensitiveness of more developed sugar beet and weeds (Winner, 1981). The treatment index_H decreased in the 4^th and 5^th treatments, each, to approximately 0.5 in rela­tion to 2 herbicide products and 3 active ingredients used per treatment (Fig. 4, Fig. 5). The treatment index_H showed significant differences between the years (Fig. 6). Consequently, the adaption of changing weed infestation was encountered by different herbicide use intensity but consistent treatment patterns on national scale (Fig. 4, Fig. 5). This could be an adaption to seasonal effects like drought, coldness as well as efficacy and selectivity of active ingredients etc. which influence the time of canopy closure and thus the competitiveness of the sugar beet against weeds (Winner, 1981; Kobusch, 2003).

Considering the generally low treatment index per treatment and the proportional dosage reductions, it becomes obvious that only few dosages were applied at authorised rate. Therefore, emphasis of weed control was not put on the exhaustion of authorised application rates but on the combination of various herbicide products with their altering active ingredients. This process of dosage-reduction and split-applications over the past three decades was also driven by reducing costs of herbicide treatments.

The active ingredients ethofumesate, phenmedipham and metamitron as well as desmedipham were most widespread in weed control which represent approximately 75% of all herbicide applications, and were applied on almost the total acreage. These active ingredients could control almost every common or standard weed infestation in sugar beet crop. Difficult-to-control weeds or specific weed species like annual mercury (MERAN), three-cleft bur-marigold (BIDTR) or volunteer rape (BRSNN) require additional active ingredients like triflusulfuron or chloridazon & quinmerac which were used on approximately one half of the acreage. Efficacy and the composition of these active ingredients are mainly influenced by weather conditions. At dry conditions a higher proportion of contact acting active ingredients e.g. phenmedipham, clopyralid or triflusulfuron are required. However, favourable conditions give an advantage to residual acting active ingredients e.g. chloridazon, dimethenamid-P, ethofumesate or metamitron. Generally, efficacy and therefore the potential decreasing application rates increase with better environmental conditions and thus the treatment index_H per treatment decreases. Regulating weed beet in sugar beet is limited to non-selective a.i. like glypho­sate (Longden, 1993). Compared to this, selective a.i. like sulfonylurea or growth promoters could be used to control weed beet in cereals more easily.

Thresholds for weed control in sugar beet were exa­mined but could not be used for most weeds because of low efficacy and high cost of herbicides to control most tall weeds (Bräutigam, 1998; Wellmann, 1999). Hence, split weed control in the very early stage of weed development could reduce application rates > 75% compared to the authorised application rate. Reduced dosages minimise physiological stress for sugar beet plants (Beissner, 2000) and could have economic and ecological benefits by using fewer amounts of herbicides. These split-applications of herbicide products are unique in sugar beet cultivation compared to other crops.

Mechanical weed control by tractor hoeing is an alter­native measure in weed control reducing herbicide intensity. Using this technique, the adverse side effects e.g. damage of plant leaves and increasing risk of soil erosion have to be taken into account (Gummert et al., 2012), while tractor hoeing had a low efficacy of intra-row weed control (Kouwenhoven et al., 1991). However, reducing herbicide input leads to lower herbicide costs of up to 100 €/ha, but increasing costs for labour, even hand-labour and machinery in total.

Today’s and future topics in sugar beet cultivation concerning pesticide use and its public acceptance may be realised, inter alia, by the National Action Plan on Sustainable Use of Plant Protection Products (BMELV, 2008) and the introduction of the respective Guidelines for Integrated Pest Management in Sugar Beet Cultivation (Gummert et al., 2011). The implementation of the general principles of Integrated Pest Management has to be followed for it, but not all principles are relevant in sugar beet cultivation. Relevant in weed control are preventive measures, monitoring, direct control, reduction to necessary minimum and documentation. However, anti-resistance strategies, non-chemical measures and threshold values are of lower importance. A tool to increase public acceptance towards pesticide use is to monitor the status quo of its application. This was realised by NEPTUN-surveys for sugar beet cultivation till 2009 and it will be done by the Panel of Plant Protection (PaPa) from 2010 to 2014 in connection with monitoring of the particular weed infestation. By focusing on risk reduction instead of application rates of pesticides in general, both profitability and ecological impact could be improved in the sense of a sustainable development.