Background

Pollinators, such as bees, butterflies, moths, birds, and bats, play a crucial role in the reproduction of plants by transferring pollen between flowers, enabling fertilization and the production of fruits and seeds. It has been estimated that about 90% of flowering plant species across the globe rely on animal pollinators [30]. Moreover, about 35% of agricultural crop species depend on animal pollination [25]. Yet, reduced land use heterogeneity and intensive agricultural practices, such as the use of pesticides, have detrimental effects on pollinators, including one of the most efficient pollinators, bumblebees (Bombus spp.) [12, 19].

Some insecticides have been banned in the European Union (EU) because of their potentially detrimental effects on bees [13,14,15,16]. Fungicides, however, are generally considered less toxic for bees. Yet, studies have shown that they can contribute to bee decline, for example by impairing learning performance [11], disrupting nest recognition [2] as well as impairing metabolic functions [10, 45]. Furthermore, fungicides, in combination with other pesticides, can produce synergistic effects, increasing the toxicity of the chemicals to pollinators [23, 40].

Amistar (Syngenta) is a systemic broad-spectrum fungicide commonly used in agriculture in Europe. The active substance in Amistar is azoxystrobin. Amistar was the representative product for azoxystrobin in the previous EU-evaluation [17]. EU regulators assessed the use of Amistar as having a low risk to honeybees based on the current agreed risk assessment which includes assessment of the ratio of the application rate (g active substance/ha) and median acute lethal dose (LD50) (hazard quotient < 50) [17]. Based on the assessed low risk, several Member States have authorized the product to be applied on bee-attractive flowering crops without any mitigation measures to protect bees. In the assessment, the acute oral and contact LD50 values of Amistar were > 200 µg azoxystrobin/honeybee [17]. Straw and Brown [38] challenged the use of the LD50 value by studying the effects of 0.8 µl Amistar/bumblebee (corresponding 200 µg azoxystrobin/bumblebee). The study showed that the exposure level caused 30% mortality, reduction in appetite, weight loss and gut melanization in the treated bumblebees. The effects were caused by a co-formulant, alcohol ethoxylates, not the active substance azoxystrobin. The study stated that the EU risk assessment relies too much on mortality (LD50) and underestimates sublethal effects.

A 10-colour learning paradigm has been used to study the sublethal effects of pesticides on bumblebees, which is designed and shown to be able to detect large variations in learning and memory performance between individuals [27]. Bumblebees foraging for nectar and/or pollen use sophisticated visual learning and discrimination strategies that are mainly based on colour information [9]. These cognitive traits have a strong effect on the foraging success of bees and therefore their fitness [34]. Pesticides, such as neonicotinoids (reviewed in [37]), and glyphosate-based herbicides [22] have been shown to impair bumblebees’ learning and long-term memory. However, the effects of fungicides on bumblebee cognition have not been studied before. This research aimed to examine the cognitive effects of chronic exposure to Amistar fungicide on bumblebees. We studied whether 5 days of oral exposure to 0.015 µl Amistar (3.75 µg azoxystrobin/day) impairs buff-tailed bumblebees’ (B. terrestris) learning and long-term memory. The cognitive abilities were studied by utilizing a 10-colour learning paradigm [27].

Materials and methods

Bumblebees

Seven bumblebee colonies were purchased from Koppert (Berkel en Rodenrijs, The Netherlands). The colonies were transferred into two-chamber wooden nesting boxes [31 × 13.5 × 11.5 (height) cm] the same day they arrived in the laboratory. The nesting boxes were connected to a flight arena [60c × 45 × 25 (height) cm] by a transparent acrylic tunnel [25 × 3.5 × 3.5 (height) cm], where the movement of the bumblebees from the nest to the arena was controlled with white plastic sliding doors. The setup was kept indoors under standardized light (LED, 2700 K, 230 VAC) with temperatures ranging between 19 and 22 °C and a photoperiod of L12/D12. Foragers of each colony were marked with individual number tags (Opalithplättchen, Warnholz & Bienenvoigt, Ellerau, Germany). The tags were attached to their thorax by using Super Glue Gel (Loctite, OH, USA). A mass feeder containing 40% (w/v) sucrose solution was placed inside the nesting box, and the bees had free access to it. In addition, the bees were given approximately 7 g of commercial pollen into the nesting box every second day (Koppert B.V., Berkel en Rodenrijs, The Netherlands).

Pre-training of the foragers

The forager bees were pre-trained to forage sucrose solution from transparent chips. The bees were allowed to move freely to the flight arena where they encountered ten transparent chips, each attached to the top of a 4-cm-high transparent stand with 10 µl of 40% (w/v) sucrose solution on top. Only active forager bees from each colony were selected in the pre-training phase. Each bee was allowed to enter the flight arena when it exhibited signs of willingness to forage and waited in the tunnel. Individual bumblebees were considered ready for the learning phase after they had successfully visited the arena three times and foraged from the transparent chips each visit.

Pesticide exposure

The commercial product Amistar (Syngenta, Switzerland, Finnish Reg.no 1836) containing azoxystrobin 250 g/l was used in the study. A total of 50 pre-trained foragers from seven colonies were assigned to the following treatments: (0) no Amistar (control, 29 bees) and (1) 0.015 µl Amistar/bee (3.75 µg azoxystrobin/bee, 21 bees).

The pre-trained foragers of the treatment group were exposed to Amistar three times a day for five consecutive days (Fig. 1). During each exposure, they received 0.005 µl Amistar (1.25 µg azoxystrobin) diluted in 10 µl sucrose solution (1.25 µg × 3 exposures = 3.75 µg/day). To ensure a minimum one-hour interval between exposures, each bee was exposed to the exact level of Amistar inside the tunnel that connected the nest box and the flight arena. In total during the five days of exposure, the bees received 0.075 µl Amistar, corresponding to 18.75 µg azoxystrobin per bee. The pre-trained control bees were treated similarly, except that they were exposed only to 10 µl pure sucrose solution during each exposure. Each exposure was given to the bee inside the tunnel between the nest box and the flight arena with a minimum of 1-h intervals. The interval was based on the mean duration of foraging bouts in resource-abundant environments (66 ± 4.6 min) as determined by Westphal et al. [44]. During the exposure, each bee was kept in the tunnel for 10 min to ensure that the bee had fully consumed the solution before returning to the colony.

Fig. 1
figure 1

Experimental setup and timeline of the experiment

Full size image

Colour discrimination task

The ten-colour discrimination task developed by Li et al. [27] was employed to test the learning and memory abilities of bumblebees in a way that is ecologically relevant. This task has been previously established, designed, and proven to result in significant variations in individual learning and memory performance [22, 27]. The learning phase was performed two days after the bee received its first Amistar or control exposure. In this phase, each bee was subjected to five learning bouts in the arena. In each learning bout, a single bumblebee was released in the arena, where it encountered 20 colour chips (flowers) of ten different colours (two flowers per colour) disposed of in random order. Five colours were considered as rewarding with 7 µl of sucrose solution (40% w/v), whereas five colours were aversive with 7 µl of a water solution saturated with quinine. Each learning bout lasted a maximum of ten minutes during which the bumblebee was expected to land on the flowers and to learn to dissociate between rewarding and aversive flower colours. The bout was stopped earlier if the bumblebee stopped foraging and attempted to return to the nest. After each learning bout, the bees returned to the nest for a minimum of ten minutes to empty their honey crops and started the next learning bout when they were ready and waited in the tunnel. This approach allowed us to ensure that the bees were in a foraging state and motivated to perform the task. A landing was defined as a bumblebee landing on top of a flower and touching the sucrose/quinine with its antennae or proboscis. The flowers were cleaned with 70% ethanol in water between each bout to ensure no scent marks were used to solve the task.

After the learning phase (learning bouts 1–5), the bees did not have an access to the flight arena for two days, but the Amistar and control exposures in the tunnel were continued (Fig. 1). The memory test was performed on the third day after the end of the learning phase. The experimental setup was the same as in the learning phase except that each flower contained 7 µl of water.

Statistical analyses

The effects of Amistar treatment on bumblebee performance in the learning phase and in the memory tests were analysed using generalized linear mixed models (GLMM). The analyses were conducted using R 4.1.1 software [31], and the models were fitted using the glmmTMB function of the glmmTMB package [6]. The models were fitted with a binomial distribution, and the regression lines were extracted using the effects package [18]. Individual bees were considered as the unit of replication and the colony was used as a random effect in the models to control for potential pseudoreplication. The relative influence of each observation was adjusted in the models by using the ‘weights’ function. We performed residual diagnostics and checked the dispersion of our models using the DHARMa package in R [21].

Model 1 tested whether Amistar exposure affected the bumblebees’ performance (proportion of correct landings) in the learning phase. The following formula was used: glmmTMB (Performance ~ learning bout (bouts 1–5) * treatment + (1|colony/bee identity), family = “binomial”, weights = a total number of landings).

Model 2 tested whether Amistar exposure affected the bumblebees’ memory retention between the final learning bout and the subsequent memory test. The following formula was used: glmmTMB(Performance ~ bout (5th learning bout and memory test) * treatment + (1|colony/bee identity), family = “binomial”, weights = a total number of landings).

Results

Based on Model 1, the bumblebees’ overall performance (i.e. the proportion of correct landings) significantly increased over the learning bouts (glmmTMB; estimate = 0.27, SE = 0.04, z = 7.15, p ≤ 0.01). Control bees and the Amistar-treated bees did not significantly differ from each other in overall performance (glmmTMB; estimate = − 0.34, SE = 0.26, z = − 1.30, p = 0.193, Fig. 2) or in the change of performance over the learning bouts (bouts 1–5) (glmmTMB; estimate = 0.03, SE = 0.07, z = 0.45, p = 0.652, Fig. 2, left side).

Fig. 2
figure 2

The left side of the figure shows the predicted levels of performance (proportion of correct landings) and its 95% confidence band for the sample values of performance of control and treated bumblebees (three days of chronic exposure to Amistar (0.015 µl/day) in the learning phase (bouts 1–5) of the 10-colour discrimination experiment (control n = 29; treated with Amistar n = 21). Circles on the right side of the figure represent model estimates of performance and error bars represent confidence levels at 95% for the sample values of performance in the memory test (control n = 29; treated with Amistar n = 17). The dashed lines between the predicted levels of performance in the learning phase and the memory test represent memory retention between the final learning bout and the subsequent memory test. The label 'ns' indicates a non-significant memory retention trend between the treatments. The horizontal dashed line indicates the chance level (50%). Black bars above each learning bout represent the number of sample values per bout

Full size image

Model 2 did not reveal a statistically significant difference between the Amistar-treated and control bees in memory retention between the final learning bout and the subsequent memory test. However, we did observe a non-significant trend that suggests a potential effect (glmmTMB; estimate = − 0.57, SE = 0.33, z = − 1.70, p = 0.088, Fig. 2, right side).

Discussion

In this research, we studied whether chronic Amistar exposure impairs bumblebees’ learning and long-term memory in the 10-colour discrimination task. Our analysis did not reveal a statistically significant difference in performance between the Amistar-treated and control bees in the learning phase. However, we observed a non-significant trend in memory retention between the final learning bout and the subsequent memory test of the Amistar-treated bees, which was negative when compared to the control bees. These new findings provide further support for our hypothesis that Amistar may have a negative effect on bee cognitive performance. Further studies with larger sample sizes and longer exposure periods are needed to confirm these findings. Nonetheless, our study highlights the importance of evaluating the sublethal effects of pesticides on pollinators, as even non-lethal doses can potentially affect their cognitive abilities and fitness.

Environmental conditions and agricultural practices affect pesticide residue levels in the environment and thus, the residue levels vary across time and geographical location [5, 29, 46]. Azoxystrobin is one of the most widely detected pesticides in bee-relevant matrixes like pollen and nectar [35], though worldwide, pesticide residues in these matrixes remain little studied [3, 46]. Based on the studies reporting azoxystrobin residues in pollen, the residue levels vary from a few micrograms to more than 500 µg/kg [4, 5, 24, 26, 28, 29, 32, 35, 39, 43]. The only information about azoxystrobin levels in nectar we found, were unpublished data by the Finnish Food Authority, where the residue levels were maximum of 38.6 µg/kg and a study by Krupke et al. [26], where the residue levels were maximum of 0.6 µg/kg. Overall, we found it challenging to estimate the field-realistic residue levels in both pollen and nectar, due to the high variability in the reported azoxystrobin levels in pollen, and the lack of studies on the residue levels in nectar.

There is a debate in the science community about whether the pesticide doses studied correspond to the actual exposure of bees in the field [7, 8, 37]. In this study, based on the residues found in nectar (Finnish Food Authority, unpublished data) and pollen [24], the daily exposure level was a hundred times higher than the daily oral exposure of bumblebees in Finnish agriculture (Additional file 1:). However, it is irrelevant whether the tested dose is field-realistic or not, as the sublethal effects of Amistar are not well known. Therefore, it is crucial to provide information about the potential risks of the product on bees, especially since there are gaps in our knowledge about pesticides. Testing higher doses can help establish a dose–response relationship, which can help predict the effects of lower doses in the field. Additionally, testing a range of doses can help identify potential risks and provide a basis for further research, even in cases where the field-realistic dose is unknown.

Amistar has been shown to decrease bumblebees’ foraging performance and pollen deposition [41], though in honeybees (Apis mellifera) no effects on colony development and foraging activity were detected [42]. In the field, pollinators are rarely exposed to individual stressors, but they encounter various stress factors simultaneously and these may act synergistically [20]. Fungicides have been shown to synergize the effects of insecticides in several studies (e.g. [23, 33]). The synergistic effects of azoxystrobin together with other pesticides are studied less and the existing studies concentrate mainly on the synergistic effects of Amistar and Closer (pesticide product containing sulfoxaflor). In semi-field experiments, no synergistic effects of Closer and Amistar were found on bumblebees [41] or on solitary bees [36]. In contrast, Naggar et al. [1] showed that azoxystrobin together with insecticides flupyradifurone and sulfoxaflor reduced honeybee health and caused dysbiosis. Due to the limited number of studies concentrating mainly on the synergistic effects of Amistar and sulfoxaflor, the harmful effects of the studied exposure levels of Amistar on bumblebee cognition cannot be ruled out when bumblebees are exposed to other pesticides.

Better knowledge of pesticide residue levels in the environment is essential in order to further the sustainable use of pesticides. In our opinion, residue monitoring should be regulated in the EU and not rely on separate studies made by academia. The lack of a comprehensive database on the residue levels in the environment leaves too much uncertainty on the field-realistic exposure levels of pollinators and other environments, and thus hinders the critical evaluation of the protectiveness of the EU pesticide legislation. Likewise, the dataset would reveal which pesticides and pesticide combinations are the most relevant ones in different geographical regions. With help of this knowledge, policymakers and academia could prioritize specific pesticides in their actions.

Conclusions

Our experiment provides evidence that chronic exposure to the fungicide Amistar does not seem to impair the learning ability of bumblebees, although a negative trend was observed in memory retention between the final learning bout and the subsequent memory test. The consistent trend from the last of the five learning bouts to the memory test suggests that exposure to Amistar may have a sublethal effect on memory retention in bumblebees. However, further research is needed to confirm this effect and its magnitude. To fully understand the potential sublethal effects of commonly used pesticides, including fungicides, and potential synergies with other pesticides, more information on pesticide residues in different environments is needed. This knowledge is essential to tackle the drivers of pollinator decline and to provide regulatory bodies with information to design more sustainable directives for pesticide use in agroecosystems.

Limitations of the study

This study has limitations that should be considered. Firstly, the laboratory dose used may not reflect the actual exposure of bumblebees in the field. Secondly, exposure levels may vary in the field, affecting actual exposure. Thirdly, the sample size was small, limiting statistical power and generalizability. Finally, the study focused on a single species of bees and cognitive task, while bumblebees are often exposed to multiple pesticides simultaneously in the field. Future research should investigate these questions in greater depth and explore the potential long-term effects of Amistar on bee health and behaviour. Despite these limitations, this study emphasizes the need for sustainable agricultural practices that protect both crop yields and pollinator health.