Introduction

Recent changes to the typical climate across Europe have resulted in an increased frequency of certain weed species (Krähmer et al. 2020). In Serbia, several of the most severe weeds are classified as invasive: common ragweed (Ambrosia artemisiifolia L.), common milkweed (Asclepias syriaca L.), field dodder (Cuscuta camprestris L.), and common lambsquarters (Chenopodium album L.). Weedy sunflower (Helianthus annuus L.) is an invasive species in southern Europe and Serbia and a primary contributor to yield losses in row crops (Stojićević and Vrbničanin 2022). The invasion of H. annuus started about 15 years ago, when weeds were found close to roads, field margins, or irrigation channels; however, their expansion has now extended into fields, where H. annuus is a highly competitive species with crops (Kanatas et al. 2021). Stojićević (2018) reported more than 200 populations across Serbia dominant on uncultivable areas, on roadsides and fields, with some having a density of more than 200 plants per square meter. While it is an economic problem only within the field, the spread along roadsides is the mainspring that needs to be targeted. It has been reported that competition from H. annuus has caused significant losses in row crops, especially in maize (Zea mays L.) (Deines et al. 2004) and soybean (Glycine max L. [Merit]) (Allen et al., 2000). Furthermore, according to Deines et al. (2004) H. annuus at a density of just four plants m−2 are enough to reduce maize yield up to 46%. Moreover, recent reports from Serbian fields (Ilić et al. 2022) have indicated high abundance in field and vegetable crops, causing loses from 5 to 33%. Taken together, a post program for controlling this weed is an imperative task for producers.

Such a task is not as straightforward as many other invasive weeds. Domesticated sunflower (also Helianthus annuus L.) is a major crop in Serbia and can be grown with two herbicide resistance (HR) traits, either to imazamox (IMI) or tribenuron (SU); both are herbicides that inhibit aceto-lactate synthase (ALS) (Presotto et al. 2012). Recent adoption of Clearfield® technology in sunflower resulted in heightened awareness of possible gene transfer from ALS tolerant sunflower to wild H. annuus, as natural gene flow can happen from herbicide-tolerant varieties to wild relatives (Presotto et al. 2012). This flow is possible depending on the overlap in flowering period, wind speed and direction, and distance between plants (Bozic et al. 2015). Božić et al. (2019) tested the successive generations of weedy H. annuus progeny which grew close to imazamox and tribenuron-methyl resistant hybrids and found evidence of some increased tolerance to tribenuron-methyl, but not to imazamox.

Still, little research has been focused on possible solutions for H. annuus in row crops other than sunflower. The research in Serbia has been mostly focused on evaluating the possible gene flow from sunflower to weedy relatives, rather than finding practical solutions for weedy sunflower control. There are only a few investigations regarding possible control of weedy sunflower. Ilić et al. (2022) reported nicosulfuron efficacy of two H. annuus populations and found lower susceptibility of the tested population. Furthermore, Vrbničanin et al. (2017) reported weedy sunflower fitness followed by nicosulfuron applications. As maize is the most extensively planted crop in Serbia with an area of 0.9–1.1 million of hectares per year (Anonymous, 2021), and post-emergence (POST) herbicides are most commonly used as correction treatments following pre-emergence herbicides for weed control, it was of particular importance to assess the possible herbicides for POST herbicide treatments to control H. annuus in maize and adjoining areas, such as field margins and irrigation channels, as the most likely source of new field infestations.

With the ambition to reduce herbicide applications across the European Union (Tataridas et al. 2022), the EU Green Deal aims to reduce the herbicide use up to 50% by 2030. As herbicides are registered for application in certain rates, any deviation from application rates might result in herbicide resistance evolution, bearing in mind that weeds could survive those rates, as they might be sub-lethal (Gressel, 2011). Nevertheless, there are still available options to apply reduced rates and maintain high level of efficacy. Adding adjuvants into the tank together with herbicides may lead to increased weed control, hence adjuvants change physico-chemical characteristics of the solution, enabling higher uptake of herbicides (Hazen 2000). Therefore, our research sought to evaluate the response of H. annuus to seven POST applied herbicides for weed control in maize analysing dose-response of each herbicide. The research also included the total, non-selective herbicide glyphosate, given that many populations still grow on field margins and irrigation channels. The second study aimed to evaluate possible reduction of herbicide rates by adding a NIS adjuvant into the tank.

Material and methods

Even near the research station of the MRIZP, there are many instances of H. annuus populations expanding from field margins into fields, therefore many local populations were available for sampling. H. annuus seeds were collected across 10 localities near the MRIZP (Fig. 1) in the October 2021 and combined into one composite collection of sunflower seeds. These locations were selected because MRIZP grows maize on more than 1000 ha and recently, weedy sunflower has started invasion on maize fields, reducing yields significantly. Seeds were cleaned and stored in the refrigerator at 5 °C until sowing.

Fig. 1
figure 1

The locations where H. annuus seeds were collected (the area of 8 km2), Google Maps, accessed 16. Sep 2023

Full size image

Two greenhouse experiments were conducted at the Maize Research Institute “Zemun Polje” (MRIZP), Belgrade, Serbia, during 2022 (a dose-response study) and 2023 year (an efficacy study). Eight herbicides were used in the experiment (seven herbicides labelled for weed control in maize, and a total herbicide glyphosate) (Table 1).

Table 1 The list of tested herbicides for H. annuus control in maize
Full size table

For both experiments, H. annuus seeds were planted and grown in D40H cone-tainer cells plastic cones (6.9 cm in diameter, 35.6 cm depth, the volume of 983 mL) (Stuewe and Sons, Inc., Corvallis, OR 97389, USA) filled with growing medium (Floragard, Oldenburg, Germany). Plants were watered and fertilized as needed. The greenhouse was maintained at 30/20 °C day/night and 16/8 h photoperiod (850 μmol m−2 s−1 photosynthetic photon flux). Initially, 5–10 seeds were planted per cone, and later thinned to one plant per cone, representing one replication. When reached 10–15 cm height (4–6 true leaves) plants were moved to a research spray chamber (Avico Praha, Prague, Czech Republic), and following the application were returned to the greenhouse, and grown for another 21 days. For applications, an AI95015EVS nozzle was used calibrated to deliver 93.5 L ha−1 at 414 kPa. After 21 days, plants were harvested (cut at soil surface) and dried at 60 °C to constant mass. All data were converted into a percentage (%) of reduction compared to untreated control (4.5 g ± 0.31).

Dose-response study

The experiment was conducted as a complete block design with four replications in two experimental runs (the 1st run April-Jun 2022; the 2nd run July–September 2022). One H. annuus plant was considered as one replication. All herbicides were applied in the following doses: 0.125X, 0.25X, 0.5X, 1X, 2X, 4X, 8X, where X matches to the field use rate of each herbicide (Table 1). The experiment contained the untreated check, where plants were grown under the same conditions. The model selection function mselect tool in R software (R Foundation for Statistical Computing, Vienna, Austria) was used to compare models, and Weibull (type 1) was selected as the best-fit model based on Akaike’s information criterion (data not shown) for H. annuus biomass reduction, which was analyzed using the drc package in R software (Ritz et al. 2015) following the Eq. (1):

$$\textrm{Y}=\textrm{c}+\left(\textrm{d}-\textrm{c}\right)\exp \left(-\exp \left(\textrm{b}\left(\log \left(\textrm{x}\right)-\log \left(\textrm{e}\right)\right)\right)\right)$$
(1)

where y represents biomass reduction (%), b is the slope at the inflection point, c is the lower limit of the model, d is the upper limit, and e is the inflection point (distance to 50, 90, and 95 biomass reduction (%)). Data from the two experimental runs were combined, with replications and experimental runs considered random effects.

Efficacy study

The experiment was conducted as a randomized complete block design with four replications in two experimental runs (the 1st run Feb-Apr 2023; the 2nd run May-Jun 2023). Again, one H. annuus plant was considered as one replication. The same herbicides were applied as in the previous study (Table 1), while using reduced doses 0.25X, and 0.5, as well as 1X alone and including an adjuvant - non-ionic surfactant (NIS, 1 L ha−1) (Dash, BASF, Germany). Justification for only including this adjuvant for these trials was confirmed by previous research under field conditions (Brankov et al., 2023a). The experiment also included untreated control plants. The data obtained were processed using the statistical package STATISTICA 8.0 for Windows (TIBCO Software Inc., Palo Alto, CA, USA). The differences between the treatments were determined by two-way analysis of variance (ANOVA), with mean separations made at α = 0.05 level using Tukey’s post hoc test. Since the effects of herbicides and adjuvants were significant, comparisons were made for each herbicide within rate and adjuvant used.

Results

Dose-response study

According to the obtained data, sunflower was the most sensitive to bentazone and tembotrione, where plants initially died 7 days after treatment, at all rates. Under such conditions, the model could not estimate the following values: ED50, ED90, or ED95 (Table 2). Also, the model could not estimate values for foramsulfuron and rimsulfuron, indicating high susceptibility of H. annuus to those herbicides. Less than half of the recommended field rate of glyphosate was needed for 90% biomass reduction. ED95 for nicosulfuron and mesotrione were 25 and 28 g, respectively. H. annuus showed tolerance only to dicamba, where ED50 was close to recommended field rate, while ED95 was 2.5-fold higher than the field recommended rate.

Table 2 Percentage of biomass reduction of H. annuus influenced by herbicides and adjuvants
Full size table

Efficacy study

Efficacy for bentazone, mesotrione, and tembotrione was high in all treatments (93.6% - 96.8%), and the influence of added adjuvants was not clearly seen (Table 3) Efficacy for bentazone, mesotrione, and tembotrione was high in all treatments (93.6% - 96.8%), and the influence of added adjuvants was not clearly seen (Table 3). However, all plants died 3–7 days after bentazone and tembotrione indicated very high susceptibility to those herbicides. Sunflower survived longer after application of mesotrione, 17–21 days following applications. Foramsulfuron and rimsulfuron efficacy was increased when NIS adjuvants were added (up to 9.5% at 0.25X of foramsulfuron and 7.3% using the same rate for rimsulfuron), in all treatments. Among all sulfonylureas, nicosulfuron showed the least efficacy, especially applied at 0.25X (40.1% of biomass reduction). The 0.5 rate of nicosulfuron did not provide satisfactory control (81.0%), while adding the adjuvant improved efficacy in all treatments. Satisfactory control of H. annuus was not obtained using reduced rates or the field recommended rate of dicamba. Furthermore, adding the adjuvant did not influence efficacy.

Table 3 Influence of applied herbicides on 50 (ED50), 90% (ED90), and 95% (ED95) of H. annuus biomass reduction at 21 DAT
Full size table

Discussion

While maize and soybean (Glycine max [L.] Merr.) present large crop area worldwide and H. annuus is an increasingly problematic weed in these crops, no previous research has directly examined the efficacy of POST herbicides of interest for possible H. annuus control in Serbia. This study demonstrates high sensitivity of H. annuus to bentazone, tembotrione, rimsulfuron, and mesotrione, even at 1/8 the recommended rate (Fig. 2). The susceptibility at partial rates should imply that full rate applications should not promote Non-Target Site (NTS) resistance in surviving plants (Suzukawa et al., 2021). This evidence may advise use of these herbicides at the full rates as part of an IPM strategy and herbicide rotation plan (Norsworthy et al., 2012). Furthermore, bentazone can be used in soybean as well.

Fig. 2
figure 2

Biomass reduction curves of H. annuus exposed to dicamba, foramsulfuron, glyphosate, mesotrione, nicosulfuron, and rimsulfuron (the graphs containing bentazone and tembotrione are not shown due to high efficacy and unsuitability for curve fitting)

Full size image

Glyphosate is a non-selective herbicide, approved for 10 years more by the EC,and it can be used for weed control on non-agriculture lands in Europe and in certain HR crops in other parts of the world. In our research, we found that less than ½ of the field recommended rate was needed for 90% of H. annuus biomass reduction. While recent literature has reported the potential of glyphosate resistance across H. annuus populations (Singh et al., 2020), our results of increased efficacy by adding an adjuvant should recommend its use and safeguard the potential for NTS. At this time in Europe, glyphosate could efficiently control H. annuus growing adjacent to fields, directly reducing spreading potential of the species.

Dicamba is a useful mode of action against a variety of broadleaf weeds in maize and has become a popular herbicide addition in HR soybean in North and South America. However, our results indicate that dicamba is not an effective mode of action for this population of H. annuus, even when an adjuvant is added. While other adjuvants may also improve dicamba efficacy, NIS has been shown as an adequate partner for dicamba (Creech et al., 2016); although, coarser droplet size may improve herbicide uptake necessary for a systemic herbicide like dicamba (Creech et al., 2016).

Acknowledging the concern for ALS resistance in H. annuus, either inherent or acquired through cross-pollination with domesticated sunflower (Bozic et al. 2015), we found the ALS herbicide results most interesting. The H. annuus population tested was very susceptible to foramsulfuron and rimsulfuron, where plants died at 1/8 of the field rate. Consequently the model could not estimate the ED values of interest (Table 3). While Ilić et al. (2022) reported lower susceptibility of some H. annuus population in Serbia to nicosulfuron, our results demonstrated 95% biomass reduction at less than ½ of the field rate, even without an adjuvant. Best management practices would still recommend using these herbicides with caution and probably in combination with other effective modes of action.

In our research we tested all herbicides in reduced rates, combining them with non-ionic surfactant. As adjuvants are known to increase herbicide efficacy, adjuvant inclusion could allow for effective control with lower rates of herbicides (Delvin et al., 1991; Bunting et al., 2004). Indeed, previous research in Serbia has also confirmed adjuvants as potential tools for increasing herbicide activity and efficacy (Brankov et al., 2023b). This study also supports the addition of NIS adjuvant as a key factor for increased efficacy in the SU herbicides and glyphosate (Table 3). The effect of the NIS adjuvant could be seen with ALS inhibitors clearly, supporting previous research where non-ionic surfactants increased efficacy when NIS adjuvants were added into the tank (Idziak et al. 2023; Sobiech et al. 2020). NIS adjuvants are water soluble chemical and lipid compounds that are not molecularly charged (positive or negative). They reduce the surface tension of the water molecule, which enable the water droplet to cover a greater leaf surface area. While bentazone, mesotrione, and tembotrione all performed well enough alone that no adjuvant advantage was apparent. Glyphosate applied with an adjuvant provided the best control improvement: from 48% applied alone to 96% applied with NIS.

Dicamba was the only tested herbicide which did not benefit from the inclusion of NIS. Those results are not in the line with the previous published results (Polli et al. 2021). Weedy sunflower plants had the characteristics symptoms of the auxin herbicides although biomass reduction was about 80% of the control (Table 2). As dicamba is an auxin herbicide, its might prolong growth of weedy sunflower up to certain time, indicating that biomass reduction was not on satisfactory level. On the other hand, visual observation of injury (data not shown) indicated that those plants on 21st day were highly damaged. Further tests will be needed using other adjuvants in order to increase dicamba efficacy.

While yield protection in maize is a major focus of this research, limiting H. annuus management to the boundaries of a field where maize is grown will not be enough to contain invasion. As already mentioned, H. annuus began in field margins and non-crop areas and has gradually spread into row crop fields. The most recent literature reported that weeds close to fields may receive sub-lethal herbicide doses, and surviving then may evolve metabolic resistance. Gressel (2011) reported that low pesticide rates may hasten the evolution of resistance. Furthermore, Vieira et al. (2020) reported Amaranth species increased tolerance to glyphosate, dicamba, and 2,4-D following exposure to sub-lethal rates via spray particle drift. Neglected and unmanaged, H. annuus found in the border areas and grassy corridors can be a seed source, reservoir for plant pathogens and pests, and harbour genetic variability for herbicide resistance. Therefore, it is of particular importance to manage the vegetative communities in field margins and areas adjacent to fields as well as promote beneficial species. A holistic strategy should include coordinated management between neighbours and land managers of shared borders.

Conclusions

Based on our research, selected herbicides present several options for treatment against H. annuus. The tested population was highly sensitive to tembotrione and bentazone, which can be recommended, especially for their contribution to diversifying herbicide mode of action. Nevertheless, mesotrione, and other ALS inhibiting herbicides tested in the study (rimsulfuron, foramsulfuron, and nicosulfuron) also indicated successful species control. Dicamba did not show satisfactory weed control with or without NIS, but further testing could be done with other adjuvant partners. At the present time, glyphosate applied with NIS could be a good option in for H. annuus control on non-agricultural areas. Furthermore, the recommendation to add adjuvants into the tank with herbicides might enable using lower herbicide rates for H. annuus control, if used in conjunction with other diverse tactics to reduce the risk of developing herbicide resistance.