TRL levels for all activities start at TRL2. Our aim in PollinERA is to develop all tools to TRL4-5. The final ERA system cannot go beyond TRL5 within the project since making this operational requires acceptance by regulatory authorities and member states. However, we will offer our in silico models to EFSA, to be included in the EFSA Knowledge Junction; this will be facilitated by the fact that partner IRFMN coordinates the sOFT-ERA project, funded by EFSA; which is aiming to develop in silico models to generate predicted values for the EFSA database OpenFoodTox. The ERA framework created based on ALMaSS will also be offered to EFSA through ongoing partner AU & UOS collaborations via EFSA PPR panel and the MUST-B working group.

Our challenge is to select key species and pesticide combinations to test, whilst creating robust testing protocols, models, and frameworks and making these available to stakeholders, and filling relevant knowledge gaps. With this in mind, we take as broad a range of organisms into account in the toxicological tests as possible. We then focus on modelling the three bee species already defined by EFSA guidance, dipterans (hoverflies), and butterflies and moths. These chosen species represent the major groups of pollinators and will be modelled realistically. However, we will also make it possible to alter key characteristics to represent more vulnerable species (e.g., by reducing reproductive potential or dispersal ability). In this way, we can also provide flexibility to test vulnerable species identified in the future. In addition, we will pilot the co-monitoring of pesticide residues in pollinator-relevant exposure sources and pollinator communities across European cropping systems and landscapes to enable real-world validation and expansion beyond the indicator species. The systems ERA framework tools provided here will be developed from the building blocks already being used in Europe and will follow the roadmap project results from EFSA (PERA and IPol-ERA). Hence, the methodology will be familiar to EFSA and national risk assessors, which should ease future take-up of the tools.

Broadly speaking, the project methods structure is that SO1 feeds to SO2 through SO4, whilst SO2 and SO3 feed forward to SO4, which acts as an integrating hub (Fig. 1).

To achieve SO1, we will determine how pesticide exposure, sensitivity, chronic impacts, and mixture toxicity varies among pollinator taxa, from these we will define focal species for ERA and test species for laboratory study. There are ca. 350,000 pollinator species worldwide and ca. 1,000 different commercialised pesticides. Considering these numbers, the test types (chronic vs acute, adults vs larvae) and their possible combinations, a critical issue is the selection of species and compounds to test. We have identified the following selection criteria:

a) Criteria for pollinator test species selection: The potentially most exposed species in the agroecosystems belonging to the three main pollinator groups (bees, dipterans, lepidopterans) will be selected. A special focus will be given to species that are commercially available or easily reared under laboratory conditions and that share ecological traits with or are phylogenetically close to the focal species (see SO3). We will evaluate routes and degree of exposure and pesticide sensitivity for at least seven species, and four species for chronic toxicological tests for single and combined substances. Relevant information on species vulnerability based on their potential pesticide exposure will be obtained from PollinERA results, literature, and IPol-ERA.

b) Criteria for pesticide selection: Pesticides with different modes of action and high occurrence levels in agroecosystems will be selected. A special focus will be given to new potentially dangerous compounds that are considered “safe for pollinators” with the current RA procedure. The mixture experiment will consider the most likely pesticide combinations in fields. We will test four different pesticides and a pesticide combination based on the information from other projects (e.g., PoshBee), stakeholder consultation (including IPol-ERA results) and exposure analysis.

Exposure routes and sources – By reviewing the literature the relevance of sources and routes of exposure for key pollinator groups (wild bees, butterflies, hoverflies, and moths) and sampling and analysing the most pollinator relevant matrices (pollen, nectar, water, plant, soil) will be determined. Sites for environmental sampling will be selected in conjunction with the monitoring in SO2. Samples will be collected in three countries, covering the three regulatory zones within the EU: Italy (southern zone; UNIBO), Poland (central zone; UJAG, INC-PAS) and Sweden (northern zone; ULUND). Pollen and nectar will be collected from honey bee foragers returning to sentinel hives following PoshBee protocols. The other matrices will be sampled in relation to expected important non-Apis routes and sources of exposure: water – small collections of standing water as potential habitat for saprophagous hoverfly larvae, plant – butterfly larvae host plants, and soil – nesting locations of ground nesting solitary bees. One partner, SLU, will screen and quantify pesticide concentrations to reduce variability (e.g., number of screened compounds and limits of detection) and use gas and liquid chromatography coupled with mass spectrometry (Knapp et al. 2023).

Pollinator sensitivity – The important step forward will be testing the range of pollinator species (‘a)’ above) for their sensitivity toward commonly used pesticides using standardised and uniform methods. Although not yet standardised, some protocols for wild bee pesticide testing are currently available in literature (Sgolastra et al. 2017, Mokkapati et al. 2022) and/or under ring testing by some international groups (e.g., COLOSS APITOX TF; ICPBR non-Apis working group), these test methods will be the baseline for their extension to other pollinator groups. Testing a range of pollinator species representing non-Apis bees, butterflies, hoverflies, and moths will enable us to construct species sensitivity distribution (SSD) curves for different pesticides. We will use data from the literature where it is available to augment the SSDs. Four laboratories will perform the tests (UJAG, INC-PAS, UNIBO, ULUND). Single compounds from ‘b’ above will be tested through contact and oral exposure to construct SSD profiles for pollinator species. For non-managed species, individuals will be obtained from the field, maintaining a balance between obtaining robust results and using a minimal number of animals. Specific biological traits of each tested species (e.g., body size, lipid content, hairiness, pH haemolymph, water content) will be characterised and correlated with relevant ecotoxicological endpoints to predict intrinsic sensitivity across species. This trait-based approach will also be useful for understanding the mechanisms of sensitivity concerning the toxicity processes, i.e., the TKTD.

Mixture toxicity – Despite the well-proven significant interactions between many toxic chemicals, mixture toxicity testing has never been introduced into ERA or registration procedures of pesticides (and other chemicals). Pesticides are commonly used in complex formulations, including combined multi-pesticide sprays or sequential exposure. Thus, pesticides representing different types (insecticides, herbicides, fungicides) and chemical groups and different modes of action, will be tested for the occurrence and frequency of significant interactive effects on selected pollinators representing different groups. The exposure to and impact of multi-pesticide exposure on pollinators is poorly studied. More data is needed to identify potential interactions that may lead to deviations from dose addition (potentiation, synergism, antagonism). The choice of interactions tested will be based on actual pesticide use in different crops and pedo-climatic zones as well as based on their co-occurrence and residue data obtained from past and ongoing or new monitoring studies (SO2 monitoring). Different life stages of the pollinator species will be tested where necessary based on exposure evaluation.

Sublethal and chronic laboratory tests – Testing will be done on pollinator species that are easily available and selected as model species. This approach will use oral exposure to mimic the prolonged exposure to pesticide residues in the field. The delayed effects on mortality, fecundity, development time, behaviour, etc., will be measured. Delayed effects are also known as time-reinforced toxicity. They are in the upcoming new bee guidance document for pesticide risk assessment (RA) by EFSA, suggested to be analysed by using a combination of chronic tests and TKTD models. The tests will also include combined treatments for pesticides commonly used in mixtures or sprayed next to each other in short intervals. When possible, pollinator exposure will be carried out according to international conventions for toxicity tests (e.g., OECD). Existing tests for the closest taxon will be adapted for species with no standard tests described.

Phototactic tests will be used to assess the sublethal effects on different pollinator species. Because phototaxy (positive or negative) is widespread in the animal kingdom, we expect to use this test as a standard and common method to compare sublethal effects between different pollinator species. In the laboratory, insect pollinators will be fed ad libitum with a syrup (with or without a sublethal dose of one or two pesticides) and then tested to assess their phototactic behaviour. The phototactic response will be quantified by measuring the time an insect needs to reach the light source inside a dark chamber arena. For recording, we will use an infrared camera connected to a laptop.

As the primary target of many insecticides is the nervous system, another possible sublethal effect can be changes in the respiration rate. The respiration rate may increase or decrease due to the direct toxic effect of an insecticide or may increase due to high detoxification costs. The respiration rates will be measured in selected pollinator species at sublethal concentrations, estimated based on bioassays performed earlier with SO1 toxicity tests.

Links to other SOs – The knowledge about the frequency and magnitude of exposure as well as interactive and sublethal (sub-chronic and/or chronic) effects on pollinators of the pesticides applied in mixtures, as a sequential application, or foraged concurrently in the landscape will allow control of pest populations while protecting the most important pollinators. This information feeds to SO4, cross-compliance. These data will allow the construction of more realistic models for ERA. Data on pesticide residue levels in pollinator relevant matrices and combined and sublethal effects of pesticides on selected pollinators will be further used in pesticide risk indicator development for pollinators (SO2) and predictive modelling using the TKTD and ALMaSS models (SO3, SO4).

Sampling site selection and landscape toxic load mapping – Pesticide and pollinator co-monitoring is initially focused toward areas with high predicted toxic load of pesticides. Site selection for environmental sampling and co-monitoring will thus be based on landscape pesticide toxic load and cover major cropping systems in Europe and the three EU regulatory zones. Within Task 4.a, ALMaSS will produce a grid of the predicted pesticide toxic load and the extent of semi-natural areas over the three sampling regions (Italy, UNIBO; Poland, UJAG, INC-PAS; Sweden ULUND), with 6 sites selected in each region (Fig. 2). The sites will be selected in grid cells with high pesticide toxic load and will be centred on two cropping systems, arable (oilseed rape, annual; 3 sites per country) and fruit (apple, perennial; 3 sites per country), which previously have been identified as potentially of high pesticide risk for bees (Jonsson et al. 2022, Rundlöf et al. 2022). Sites within a crop and country will be distributed along a gradient of semi-natural habitat extent to provide useful landscape context variability for scenario validation in ALMaSS and potential for pesticide exposure and effect mitigation for pollinators (Park et al. 2015). Field inspections will be done to validate in silico outcomes for crops and landscape context.

Figure 2.  

The pesticide and pollinator co-monitoring and sampling covering pollinator relevant matrices, groups and cropping systems in three sampling regions in the EU zones (south, central, and north).

Co-monitoring of pesticides and pollinators (PPCoMS) – Field monitoring for pesticides and pollinators (co-monitoring) borrows methods from current separate monitoring schemes. Following the method of the SPRING Minimum Viable Scheme Pilot within EU-PoMS, protocols will be implemented for transect surveys of 500 m, divided into 10 50-meter segments, to collect all bees, butterflies and hoverflies encountered within 1 m to each side and 2 m in front and up for bees and hoverflies and 2.5 m to each side and 5 m in front and up for butterflies while slowly walking along the transect during satisfactory weather. Non-collected pollinators are noted to the closest possible group on the EU-PoMS Transect Recording Form. The pollinator collection and recording are done to capture the pollinator populations and community at the sampling sites. The transect is walked back while counting and recording flower units in each of the segments using a 0.5 m² square to estimate local flower abundance and collection of plant material for pesticide residue analysis. Moths are surveyed and collected using special moth traps piloted within the Swedish Nat-PoMS. The sampling and survey, coordinated by ULUND, will be conducted by UNIBO (Italy), UJAG and INC-PAS (Poland) and ULUND (Sweden) to cover southern, central, and northern Europe as well as evaluate the inter-team feasibility of the protocols. Sampling and surveys are repeated twice to cover both previously identified high-risk pesticide periods (Knapp et al. 2023) and key pollinator activity periods (Arnberg et al. 2022). Collected pollinators are kept frozen until they are first identified to species in the laboratory and after that, pooled into a taxonomic group (wild bees, butterflies, hoverflies, moths) samples and sent for pesticide residue identification and quantification by SLU.

Pesticide screening and quantification – The laboratory at SLU will process environmental and insect samples from field locations based on their long experience in determining a large number of pesticides in bee-related matrices (Rundlöf et al. 2022, Jonsson et al. 2013). The laboratory is accredited according to Swedac 1208 ISO/TEC 17025 and has been running the national pesticide monitoring since 2002 with the aim to track long-term trends in pesticide residues in surface water and groundwater, later extended to sediment, air and precipitation. SLU will use liquid chromatography-tandem mass spectrometry (LC-MS/MS) and gas chromatography-mass spectrometry with negative chemical ionisation (GC- (NCI)MS) and methods development in SO1 & SO4.

Pesticide risk indicators for pollinators – There is no specific pesticide risk indicator for pollinators from which to indicate risk from information gathered by field monitoring. Hence, PollinERA will develop two such indicators, TWC_mix and PollSPEAR, which could be used along other more general pesticide environmental risk indicators such as pesticide (toxic) load and the EU Harmonised Risk Indicators. The toxicity weighed concentration of the pesticide mixture (TWC_mix) has been used by ULUND and SLU to track pesticide risk for bees among sites, matrices and pesticide active ingredients (Knapp et al. 2023, Jonsson et al. 2022, Rundlöf et al. 2022) and also applied in the PoshBee project. It is based on pesticide concentrations in bee-related matrices weighed by standardised toxicity data for honey bees (average of acute oral and contact lethal dose to 50% of the test population). In PollinERA, TWC_mix will be further used for non-bee-related matrices and related to pollinator species richness and abundance to evaluate its usefulness for pollinators also beyond bees. In addition, data on pollinator species sensitivity and traits (SO1 & SO3) will be combined with monitoring data on pollinator species abundance to develop a pollinator community-based pesticide risk indicator similar to the trait-based SPEAR (Species At Risk) indicator developed for aquatic invertebrate communities (Liess and Ohe 2005). We will attempt to improve on the SPEAR concept by following Rubach et al, 2010’s suggestion to use mean species sensitivity of pollinators as the benchmark. The suggested indicator development is targeted towards pesticide risk and impact tracking which would provide tools for tracking the goal fulfilment of pesticide and pollinator related targets on reduce pesticide risk explicit within the strategies of the EU Green Deal.

Links to other SOs – The prediction of high pesticide toxic load mapping for site selection for the pesticide and pollinator co-monitoring relies on the ALMaSS landscape simulation (SO4). The co-monitoring will in turn provide empirical information on environmental pesticide exposure among pollinator taxa (SO1) and landscape context (SO4) as well as pollinator population densities (SO3). The development of pollinator pesticide risk indicators relies on pollinator exposure, sensitivity and trait information collected from the literature and produced in the laboratory (SO1) as well as in silico (SO3). The indicators are used to track outcomes of scenarios and policy related to pesticides and pollinators specifically (SO4).

SO3 includes development of three types of in silico models and one data analytical process. The model types are models to predict pesticide properties from their chemical structure (QSARs), models to describe the toxicokinetics and toxicodynamics of pesticides in pollinators (TKTD), and third, agent-based population models of pollinators in ALMaSS. The latter two feed as components to SO4, whilst the QSARS provide toxicological input for ERA evaluation. The data analysis produces a classification of pesticides into cumulative assessment groups, a necessary step to reduce the dimensionality of the problem for the ERA framework.

QSAR and associated approaches – To fill data gaps, in silico models that put into relationship the chemical structure and the activity (or property) to be calculated, QSAR models can be used. Their development will be based on these three main orthogonal methodological components:

• Quantitative structure-activity relationship (QSAR) models based on collections of experimental values, such as toxicity values; this method refers to the predictive models;
• Toxicological mechanisms, codified into rules such as structural alerts and physico-chemical properties; this method relies on the theoretical reasoning;
• Similarity tools to link experimental values of related substances within the read-across perspective; this method refers to the experimental evidence.

1. These three components will be combined into a weight-of-evidence (WoE) strategy, following the EFSA Guidance on WoE. Partner IRFMN will exploit data for a series of compounds to develop QSAR models. Multitask methods will be applied, addressing effects towards different species, which will exploit data even in a sparse data matrix for the lack of data. The honey bees' data will be the most numerous. The chemical information will be derived from the structure represented as Simplified Molecular Input Line Entry System (SMILES), in the case of models using the software programs CORAL (based on Monte Carlo statistical approach), SARpy and QSARpy (machine learning in house tools using fragments) (Carnesecchi et al. 2019; Carnesecchi et al. 2020a; Carnesecchi et al. 2020b). All these tools are available from the VEGAHUB website, developed in-house – www-vegahub.eu). Alternatively, molecular descriptors will be calculated using software such as Dragon 7.0 (Kode srl 2016) or PaDEL (Yap 2010) and used to develop models using support vector machines (SVMs), random forests, artificial neural networks (ANNs), deep learning, etc.
2. Using mechanistic reasoning to identify different species’ sensitivity, based on the TK inputs derived from SO1 (Partner UOS), to cover the data gaps. This information will be linked to partitioning consideration and related to the different species biological traits. In this case, physico-chemical data, experimental or predicted using existing models (e.g., in VEGA), will be used to model toxicology among different species. Further features, related to active transport or toxicodynamic properties, will be addressed on a case-by-case basis, depending on the available theoretical information. In the case of the toxicodynamic properties, mode of action (MoA) will be used, extending the model already developed in VEGA for honeybees. Thus, these theoretical considerations will identify different behaviour compared to the honey bee's model; these deviations will be represented by rules based on structural alerts and physico-chemical properties associated with the chemical structure of the substance to be evaluated.
3. Experimental values will be used with software in VEGAHUB such as ToxRead and VERA. These are software programs for read-across to provide toxicity evaluation for substances similar to those represented by the source ones. They use different similarity inputs: the overall structural chemical similarity, as in VEGA; the presence of relevant molecular groups; the physico-chemical similarity; the toxicological similarity, using structural alerts and MoA. A further similarity measurement based on toxicokinetics will be added to PollinERA.

These three components will be combined into a weight-of-evidence (WoE) strategy, following the EFSA Guidance on WoE. Existing models for the environmental degradation products, implemented in JANUS (available from VEGAHUB), will be used by IRFMN to provide information not only on the parental pesticide, but on its degradation products, while the VEGA model on mode of action can be used to group substances in a mixture.

Finally, the resulting models will be implemented in VEGA. Different technological solutions may be adopted to implement the models within other platforms, such as EFSA Knowledge Junction, OECD QSAR Toolbox, Danish (Q)SAR Database, AMBIT, ICE, etc. Docker solutions will be proposed in this case.

Definition of environmental cumulative assessment groups (CAGs) – This is a critical step to reduce the dimensionality of the problem down to several substance groups that can be followed in environmental matrices and simulated concurrently in the ERA framework tools. CAG definition requires a set of steps each refining the groups assigned to the individual CAG and draws on information across PollinERA. The first selection is based on toxicology and mode of action. Within VEGAHUB there is a model for the mode of action of pesticides, which will be useful to cluster substances. Molecules with similar modes of action and toxicological profiles will be combined for each focal species using data from SO1 and the in silico predictive models. These selections will be further refined by defining agricultural use (SO4) and environmental fate (SO1), and exposure of pollinators (SO1 & SO4). The result will be a set of CAGs that may be species and location specific but can be standardised to create comparable RAs across a wide range of pesticides and locations. The target here is to reduce the number of specific chemicals to be independently followed in the simulations from hundreds to 30-50 CAGs. The results of this analysis will be discussed with stakeholders (SO4) and once agreed will be implemented in the ERA framework.

TKTD modelling – To date, almost all TKTD modelling approaches have addressed aquatic species and one exposure route: uptake from the water. Nevertheless, one of the specific advantages of TKTD models in the context of the ERA of chemicals is that they can integrate several separate exposure routes and hence allow RA based on combined exposure routes. One first step towards using TKTD modelling for insects was recently made by suggesting the BeeGUTS model, which can account for two different exposure routes of chemicals for honey bees (Apis), i.e., via contact and oral exposure. Using BeeGUTS, it was possible to consistently describe experimental results of honey bee survival accounting for acute contact, acute oral, and chronic oral exposure with high accuracy using a single set of parameter values, and knowledge about the physiology of honeybees (e.g., honey stomach size). In PollinERA, we will develop robust generalised TKTD models integrating multiple exposure pathways without prior knowledge of detailed physiological characteristics. We will take advantage of ongoing project work within an EFSA-funded tender describing TKTD processes for non-target arthropods (AENEAS project), where the modified GUTS models are screened for their potential to integrate various exposure pathways. We will further optimise the structure of such pollinator-TKTD models and develop and provide robust and efficient methods to calibrate these models based on minimal experimental data while integrating machine learning techniques such as QSAR and read-across. The methodology will be applied to model the uptake and effects of various pollinator species, including hoverflies, butterflies, and moths. The result will be parameter sets for these pollinator species for a range of relevant pesticide compounds, together with the methodology that will allow to apply these models for further pollinator species and pesticides.

Population modelling – We propose to model four main pollinator groups, hoverflies (Syrphids), butterflies and moths (Lepidoptera) and bees (Apoidea) to cover the suggested groups for European pollinator monitoring (Potts et al. 2020). We define seven focal species for modelling using agent-based models in ALMaSS: utilising existing ALMaSS bee models (Apis melifera, Osmia bicornis and Bombus terrestris) and creating new models of species in the other pollinator groups. We plan to develop models for an aphidophagous (Sphaerophoria rueppellii) and a saprophagous (Eristalinus aeneus, Eristalis tenax) hoverfly species, a diurnal butterfly (Maniola jurtina) and a Noctuid moth (Orthosia sp.). However, the final choice to be informed by the recommendations of the EFSA IPol-ERA project when PollinERA starts. These models will include plasticity to represent the groups in Europe by re-parameterisation to represent the more sensitive species in the groups; they will integrate the TKTD models developed by the project into each individual modelled insect.

We propose a structured approach to create models for the new species. Each will first be scoped and designed using the new Formal Model approach (Topping et al. 2022), laying out the salient literature on the species and the intended design of the species models with equations for review. We will then implement the species models within the ALMaSS model in C++ using the OOP paradigm to allow for extensibility of code and easier maintainability, and document them using the MIDox approach that links documentation directly to the program code. Along with the implementation, the models will be calibrated against available ecological and behavioural data and knowledge on each species within the RA context (existing and from SO2). The models will also be parameterised to represent more sensitive species, allowing also to represent the target groups better. For example, the Bombus model could be parameterised to have shorter foraging ranges and smaller colonies, more similar to the common carder bee (Bombus pascuorum), a doorstep forager with small colonies.

The modelling will take a multi-stressor approach, taking account of the similarity of action, but of different exposure routes, to test multiple active pesticides. Single products will be tested but against a level of background exposure to other pesticides and stressors as found by monitoring (SO2). By incorporating TKTD modelling, the individual insect pollinators will be able to respond more realistically to the accumulated body burden of pesticides while also allowing recovery, rather than relying on probabilistic chances of death upon exposure. This will allow for a more realistic, and therefore a more refined, ERA.

Links to other SOs – TKTD modelling will be used as a module within the species models in the ALMaSS simulation system (SO4). Integration of such a new module will require optimising the implementation of the TKTD modelling. We will hence further optimise the TKTD equations and methodologies to allow the most efficient and robust calculations of internal concentrations of a range of pesticides and their impact on pollinator species as modelled in ALMaSS. The QSAR models will be used in the ERA (SO4) too. Data to develop these models (TKTD and QSARs) comes from SO1. Definitions of CAGs feeds forward to SO4 and the ERA framework but also relies heavily on inputs from SO1.

Simulation methods are designed to develop a spatial-temporally dynamic pesticide exposure and toxic-risk modelling tool for the pollinator systems based on ALMaSS. ALMaSS, often used to simulate pesticide effects (Ziółkowska et al. 2021), is a mature landscape-scale simulation system for investigating the effect of landscape structure and management changes on the population size and distribution of animals and environmental, economic, and agricultural consequences in agricultural landscapes. ALMaSS will provide the framework to host the pesticide application and fate simulation, which will be fully integrated into the spatial system providing modelling to toxic loads to support the ERA.

ALMaSS landscape modelling: A major part of the ALMaSS system is the environmental modelling (landscape) in which the agent models are placed. Through collaboration with H2020 projects (EcoStack, B-GOOD, and PoshBee) the fully working ALMaSS landscape models at the national or regional level are or will be soon available for 11 EU countries (Fig. 3) and can be adapted and (if necessary) upgraded to be used within the PollinERA. ALMaSS landscape models consist of spatial and temporal components, where the spatial component is a detailed raster map of spatial resolution of 1-m², and the temporal component allows capture of changes in vegetation patterns, and processes related to, i.e., farming activities, with a time step of 1 day. In these models, cultivable areas are described in more detail by delineating agricultural parcels and grouping these into farm units of different types. Such highly detailed models require data of sufficiently high resolution and quality. The methodology assumes (Topping et al. 2016) the use of detailed topographic databases (usually in scales 1:10,000 – 1:50,000) and IACS / LPIS (Integrated Administration and Control System / Land Parcel Identification System) data, particularly maps of LPIS reference (agricultural) parcels together with an agricultural register of crops cultivated in reference parcels and anonymised information on parcel owners. However, acquiring such data can be very time-consuming and costly for some countries. Therefore, in PollinERA existing landscape capability will be expanded only in those countries where ALMaSS simulation is already underway. These include the three countries used for monitoring and for which toxic load predictions will be used to determine optimal monitoring locations.

Figure 3.  

ALMaSS coverage in Europe. All are now operational.

ALMaSS is an open-source open-science project hosted on GitLab, and is the main tool used and developed by the Centre for Social-Ecological Systems Simulation (SESS) at AU. It is an agent-based model system based on highly detailed simulation models (e.g., all bees in a colony as individuals making decisions every 10 minutes), a software engineering endeavour not to be confused with the typical simple agent-based models encountered in ecology. ALMaSS has seen use in regulatory RA to support policy evaluation for the Danish government, and, most recently, for the Dutch government to assess aspects of their agricultural sustainability policy. It can be used as a multi-criteria decision-making support by combining alternative options and goals in future scenarios, which can be compared and contrasted using objective metrics such as efficacy/cost. Key ALMaSS knowledge advances relate to the context dependency of results and the need to evaluate management/risks using a systems approach, fully dynamically integrating all significant factors. This is the basis of the new approach developed by the European Food Safety Authority (EFSA) for ERA of bees, relying on ALMaSS simulation modelling to provide the honey-bee risk assessment model ApisRAM.

Landscape simulation for toxic load – Apart from the detail often used in ALMaSS models, what sets this system apart from others is that from its conception, it has utilised dynamic landscapes rather than static maps. Therefore, a very substantial part of the system is modelling the environment the pesticides will be applied into and their subsequent environmental fate. ALMaSS models can track the daily growth of vegetation, pollen and nectar resources, and the environmental concentration of multiple pesticides. Spatial relationships can be specified in ALMaSS for applying pesticides and their environmental behaviour in soil and vegetation, thus providing an excellent basis for developing a toxic load simulation. Pesticides are represented as substance concentration per 1m² on vegetation surface, in soil or plant parts (pollen, nectar), which links to the environmental sampling and analysis in SO2. All are modelled with environmental decay rates that take weather into account. Supporting data will come from existing models for the environmental degradation products, implemented in JANUS (available from VEGAHUB), will be used by IRFMN to provide information not only on the parental pesticide, but on its degradation products. When cultivating a crop, initial concentrations are determined by applying the pesticide following normal agricultural practice. This application is determined by the agricultural system in place in a particular landscape and is thus location-specific. It can be overridden, however, for simulation of experimental scenarios (e.g., all farmers apply pesticides).

A key challenge here is expanding the capability to track multiple chemicals at this level of detail. This is only possible through cumulative assessment groups (from SO2), the application of well-designed data structures and the use of parallel processing in software design. Using these methods, we expect to be able to track 30-50 concurrent CAGs/Pesticides. As well as feeding the ALMaSS population models, the landscape concentrations provide a separate output as spatial-temporal maps of toxic load for each landscape (for use in SO2 to guide the choice of monitoring landscapes).

Integration of the pesticide and non-regulated stressors will occur as a consequence of the impacts of management and climate modelled directly in the pollinator models developed. For example, direct mortality by farming actions such as cutting vegetation or sub-models for parasitoids, as appropriate.

Scenario generation methods – Scenarios will be developed to be relevant and replicable to all three traditional regulatory regions of the EU, represented by collaboratively developing scenarios in Sweden / Denmark (northern) Poland / Germany/ The Netherlands (central) and Italy (southern). Scenarios will also cover the Europe’s main agricultural / cropping systems of Europe, i.e., arable, horticulture, fruit and oleaginous crops. The formulation of scenarios will require coordinated guidance between project experts and targeted stakeholder representatives of the project’s two main user groups: namely

1. Risk assessors (in regulatory authorities and industry) - users of the tools and
2. Risk managers (at central EU level and national level). Input and feedback from these targeted end-users will be gained through a series of facilitated workshops / webinars held during the life of the project.

In addition, our end user engagement strategy, through a series of coordination meetings, will link PollinERA to other key projects engaged in the review and assessment of current RA guidance (e.g., PARC etc.). These activities will seek to bring together different viewpoints and explore a variety of ERA approaches, employing sound SSH (social science) approaches regarded as integral for effective transition management. This will further anchor the development of ERA tools and methods, formulation of scenarios, and impacts for policy on aspects of generic applicability, especially for higher-hierarchical-level decision contexts, highlighting the versatility, fitness-for-purpose, and applicability of project results.

Framework methods – This is the generation of the One System framework from the components combining all components of SO1-SO3. The toxic load relates to the hazard identified in SO1 combined with the environmental concentration through time of multiple pesticides. Toxic load provides a common currency for the combination of pesticide risks, allowing the effect of multiple pesticides to be directly combined. The sum of the combinations will be either dose addition or synergistic/antagonistic effects as predicted from SO1. Toxicological knowledge will feed to TKTD and exposure modelling. In silico predictive toxicology and agricultural usage information will predict combination effects and provide CAGs. The population-level RAs will access this information in space and time, driving TKTD models in individual pollinator agents simulated in the landscape. All will be integrated into ALMaSS-based landscape simulations for European countries using the pollinator population models, which will also bring in non-pesticide stressors, such as mortality due to farm management and parasites and diseases. Finally, a feedback loop based on monitoring will be specified to allow the adjustment of models, and to direct monitoring to critical locations. The models and procedures developed will be used with the PollinERA case studies (monitoring landscapes from WP4 and scenarios cases to demonstrate the tools across Europe) to develop and showcase the PollinERA One System. A professional user interface will not be developed in the research project, but web-based user interfaces (e.g., using R and Shiney) will be provided so that users can test and evaluate the tools and provide access to the scenarios. The framework will then be ready to test regulatory and policy scenarios.

Policy impact methods – PollinERA is committed to contribute to the EU goal of reversing the decline of pollinators. This is done by mapping PollinERA key results and tools onto the intersection between the pesticide regulatory framework - Regulation EU 1107/2009 and Sustainable Use of pesticides Directive 128/2009, including its revision - and the pollinator focused biodiversity conservation initiatives – linked to and delivering on the European Green Deal and the fundamental Sustainable Development Goals (SDGs). Identifying these intersection nodes will facilitate the implementation of the revised EU Pollinator Initiative and the Restoration Law, contributing to achieve the targets established by the Biodiversity 2030 and the Farm to Fork Strategies or understanding if those targets has been met. Specifically,

1. an integrated system for pesticide ERA will facilitate transformative change through maintained food production without degrading land or reducing pollinator biodiversity (SDG2/15);
2. pesticide risk indicators specific for insect pollinators and co-monitoring to set baselines, track the goal fulfilment of pesticide risk reduction and halting pollinator decline; and
3. pesticide mitigation actions that are trialled in silico, verified in landscapes and for potential integration as financially supported agri-environmental measures in the Common Agricultural Policy (CAP). Applying SSH approaches, authorities and other stakeholders will be engaged to define policy options that cover these intersection nodes, for possible implementation through the National Action Plans (NAPs).

To ensure the uptake of PollinERA advancements, the developments of the relevant regulatory frameworks will be continuously followed up through the PARC partnership (SLU, AU, UOS, ULUND) and by BEBC through monitoring of the decision-makers’ discussions. See Table 8 for more detailed stakeholder coverage.

Interdisciplinary approach – PollinERA will take an interdisciplinary approach integrated through tool development. It uses multiple modelling approaches from chemical modelling to systems modelling together with laboratory experiments to generate knowledge and tools and uses social science approaches to co-develop these with stakeholders and to maximise impact.

The PollinERA project is closely linked to a number of international and national research and innovation projects and activities due to direct involvement of partners (Table 2). In addition, PollinERA will liaise with other activities and projects in this Horizon Europe Cluster 6 (Table 3) or in any project group suggested by the EC.

The promotion of gender equality and the gender dimension in the Horizon Europe Research and Innovation Programme is addressed within the PollinERA project. The coordinator (AU), university and research institution project partners (UJAG, ULUND, UNIBO, UOS, INC-PAS, IRFMN, and SLU) all have gender quality plans implemented that undergo regular updates and amendments. Our NGO partner (BEBC) is represented by its female Scientific Director and Project Manager in this project. Furthermore, PollinERA confirms gender balance with the researchers of the consortium, comprised of 41% women and 59% men.

PollinERA considers the insights to an inclusive analysis presented in Gendered Innovation 2 (European Commission, Directorate-General for Research and Innovation 2020). The project will include a gender dimension in the project activities, ensuring user acceptance of differences in biological characteristics and the social and cultural features of genders. AU will be in charge of gender considerations within the PollinERA consortium.

Operationalisation of the gender approaches will benefit the performance of the project team and, thus, the project. During the project, AU will ensure the assessment of interactions of all genders with the proposed solutions and potential gender differences according to Gendered Innovation 2. In PollinERA, potential gender differences in the project will be considered, particularly concerning WP5 & 6's interaction with stakeholders. PollinERA ensures that the group of relevant stakeholders will be complementary, considering their added value and their social, cultural, and societal background, including gender differences.

Open Science – all PollinERA protocols and results will be disseminated using an Open Science initiative to maximise impact and end-user engagement. All protocols developed will be described on the PollinERA website. The communication and dissemination WP includes significant representation from all partners to enable continuous scientific feed to the website. All program code will be available as open-source (GitLab) and all discoveries translated to easily understood news articles published on the website. Models relating biological, environmental, and physical-chem properties of chemical substances to their structure will be published on VEGAHUB. Full documentation of all models will also be provided in an easy to navigate format. This initiative is in addition to the normal open data principles of H-Europe. The project will also make use of the highly innovative Open Access journal Research Ideas and Outcomes (RIO) and Food and Ecological Systems Modelling Journal (FESMJ), where a special Open Science Collection for PollinERA is envisaged as well. In addition to conventional research papers, the journals welcome contributions documenting the entire research cycle, including data, models, methods, workflows, results, software, perspectives, and policy recommendations (see section 2.2. for details).

Data management plan – A data management plan will be developed at the beginning of the project conforming to FAIR principles. Its aim is to ensure that the data is acquired, processed, and stored systematically and transparently and that it complies with formal and regulatory requirements. The data management plan will comprise the three major steps – data acquisition, data processing and data storage. It consists of a data governance framework, a plan for metadata management and a definition of the data pipeline. Partner BEBC will manage this process and will assist in the development of the relevant data models.

Data governance framework – A framework for data governance, consisting of a formal framework for the execution and enforcement of authority over the management of data and related assets, will be set up in collaboration with all PollinERA teams involved in data acquisition and processing. Such a framework is currently developed for the EU Pollinator Hub. It could be used as a basis for the development of procedures, dedicated work instructions and guidelines for the PollinERA project. In this framework it will be defined how and with which tools the teams will standardise, collect, and model data. If considered necessary, a standard business glossary with commonly defined and shared terms will be created and a procedure to update this dictionary will be established. Existing data standards will be collected based on data sampling and processing methods in order to solve conflicts such as ambiguity of concepts or different use of terms. Where necessary, legal requirements concerning the data will be addressed in the data governance framework. If the data should contain personal identifiable information, it will be ensured that the required consent to process this data will be tracked and executed throughout the data life cycle. Data ownership will be managed in the same way. Eventually, roles and responsibilities within the project with respect to data stewardship will be defined: responsibilities for data collection, maintenance, and modelling.

Metadata management – A plan for the management of metadata, necessary to describe the nature and purpose of datasets, will be developed for the PollinERA project. The metadata management plan will contain a procedure to collect a procedure to collect information on the origin of the data, the entities (individuals, organisations) responsible for standardisation, acquisition and processing (transformation, modelling, storage) of data, material and methods used for acquisition and processing (transformation, modelling, storage) data, quality standards applied (e.g., regulatory standards, internal standards, protocols and results for method validation).

Data Pipeline – A data pipeline will be defined, establishing how (through an existing upload module or through dedicated procedures) and by whom data will be integrated on the EU Pollinator Hub. Dataset reports will be created by the EU Pollinator Hub as part of the data integration process, containing a catalogue of the metadata and the results from data profiling (technical metadata) such as information of files, fields (names, types, length) and (if necessary) encoding. Unique and persistent identifiers are provided for datasets, providers, and files. Relationships within and between datasets are documented. Information on data integrity (completeness, accuracy, consistency across data files or datasets) as well as uniqueness, is provided. Data quality issues and the resolution of such issues are documented. The team of BEBC will set up and operate the data pipeline.

Open data and data storage – Research data and other outputs (metadata) of PollinERA will be integrated on the EU Pollinator Hub, an open data hub developed by a consortium led by BEBC and sponsored by the European Food Safety Authority (EFSA) which aims, among others, to establish a free-of-charge and automated service specialised in the standardised collection, processing and dissemination of pollinator-related data from validated sources. Some on EU Pollinator Hub will be stored within the EFSA knowledge junction in Zenodo. More structured data will be imported into the EU Pollinator Hub itself, transformed to the data model used by the platform, and stored within the MariaDB database server used by the Hub.

Important links to impacts and outcomes are indicated with references to Expected Outcome numbers (EOX), each linked to the specific objectives. The One System ERA and policy evaluation framework (KER6) entails a significant technological advancement that links complex models with data streams to create a systems tool. This tool is able to evaluate the impact of new and existing pesticides under realistic conditions, as well as the impacts of policy or management changes. It is supported by knowledge creation and scientific tool development and access.

Knowledge Creation – PollinERA creates new knowledge to support understanding the landscape-agriculture-pesticides-pollinator system. SO1 provides an understanding of sources and routes of pollinator pesticide exposure and the effects of such exposure among pollinators, both separately and as mixtures. Ecotoxicological knowledge is expanded on both exposure and effect, particularly in the direction of non-bee pollinators – as this is a major gap. This new ecotoxicological knowledge is used to support the development of TKTD models, for mechanistically linking environmental contamination to effects on pollinators (EO4), and for validating predictions of pesticide environmental fate and landscape toxic load (EO3). SO2 will expand the ecotoxicological insights by linking the SO1 environmental contamination of pollinator-relevant matrices (nectar and pollen as food, water where hoverfly larvae develop, plants that host butterfly larvae, and soil that wild bees nest in) to the contamination in insect pollinators (wild bees, butterflies, hoverflies, moths), their population densities and community. This will enable conclusions on links between pollinator community composition and pesticide exposure levels, as well as the potential for seminatural areas to mitigate pesticide-related pollinator loss and knock-on effects on food production (EO1, EO2). SO3 will generate knowledge on the likely behaviour of novel pesticides through modelling (EO3), and SO4 will generate systems-level knowledge of the effect by integrating toxicology, environmental fate, and exposure, with landscapes, pollinator ecology, and behaviour and management practices (pesticides, farming, biodiversity support, mitigation) (EO5).

Scientific Tool Provision and Accessibility – PollinERA aims at developing science-based tools to support sustainable development. Key tools developed or created by the project include the SO1 development of new protocols for toxicological testing of pollinators (KER1) (EO4). From SO2 the creation of a co-monitoring scheme for pesticides and pollinators is a key result (KER2) (EO1, EO2). From SO3 there will be the creation of a set of models from QSARs to predict chemical impacts based on their structure (KER3), through generic and specific TKTD models (KER4), and to agent-based pollinator population models in ALMaSS (including TKTD) (KER5) (EO3, EO4). All PollinERA tools, comprising the monitoring scheme and systems models will be backed by detailed documentation to provide easy access for researchers to further develop or use these tools. Analysis of where PollinERA tools might be used to bridge different policy instruments will help steer users to impacts (KER8).

The expected outcomes are referred to in sequence below with the contribution of PollinERA highlighted. EO1-3 refer to knowledge creation, which is the main expected outcome of the call, while EO4-5 focus on implementation. EO1: Routes of exposure, which link ecosystem and biodiversity dynamics to chemicals are better understood. From SO1, PollinERA information will be stored in the EU Pollinator Hub and used by the scientific communities and international authorities (e.g. EFSA, ECHA) for developing guidance documents for insect pollinators. This guidance will be facilitated by PollinERA’s partner connections to these authorities. The PollinERA Pollinator and Pesticide Co-monitoring Scheme (PPCoMS) will constitute a tangible resource for several stakeholders (policymakers, decision makers, risk assessors, risk managers) to further understand the link between pesticide exposure and its consequences on functional biodiversity and pollination services. This links to EO2: Issues raised by the contamination of biodiversity in the natural environment are better known (SO2). Here PollinERA will contribute to identifying the most persistent and co-occurring pesticides in the fields and their level of accumulation in the environmental matrices that are relevant to different insect pollinators. PollinERA outcomes will provide a platform to develop a post-registration monitoring scheme to identify the actual impact of pesticides on biodiversity at the landscape level, to be used by policymakers to better target policy, and to evaluate if measures are working. From SO3, through predictive in silico approaches PollinERA will develop tools that predict the fate of new chemicals of emerging concern (EO3). These tools will be used by public authorities for screening pesticide risk, as well as by industry to identify early on whether novel pesticides might raise concern.

Using the information from SO1-3 in SO4, PollinERA will address EO4 (Toxicological and ecological impacts of contaminants are better understood and risk assessments for relevant highly exposed species are strengthened) by identifying the most vulnerable/exposed species, which could act as umbrella species for key pollinator groups provides critical input to the regulatory process and will dramatically affect the pesticide ERA guidance. In addition, the ecotoxicological protocols developed by PollinERA will be included as new tests/model species in regulatory ERA. Using the One System in SO4 we will address EO5 (Prevention and mitigation measures are identified and developed). We will contribute to identifying scenarios at high risk for pollinators and develop suitable mitigation measures by testing and identifying risk and mitigation scenarios useful in preventing ecological impact on insect pollinators. PollinERA outcomes will provide support for farmers, risk assessors and managers in their decisions simulating their consequences for pollinator health.

Although PollinERA will contribute to all destination impacts to some degree we have identified three main areas where impacts will be most significant (Table 4).

Barriers to impacts primarily act on two levels, the implementation of the tools and tests (critical risks see Part A), and secondly to the uptake of the tools or access to resources, which if not successful, will largely prevent societal and economic impacts (Barriers Table 5).

The planning of Dissemination, Exploitation, Communication and Engagement (DECE) activities in PollinERA is based on a methodology starting connected to the PollinERA planned actions and KERs above. These strategic elements aim to ensure that the DECE activities planned will focus on achieving real impact and supporting the project through adequate planning and execution.

Plan for the Exploitation and Dissemination of Results (PEDR) – PollinERA’s activities are policy and technically orientated, and thus ‘Policy Briefs’ (KPI9) will be targeted to disseminate results to the regulatory community related to chemical risk assessment and management from the pesticide, biocide and veterinary production areas, both at EU and national levels. Policy Briefs will outline analyses and recommendations for legislative initiatives associated to activities and results obtained by PollinERA. The dissemination of Policy Briefs is intended to raise awareness, foster the ‘buy-in’ and adoption of the project’s new ERA tools and approaches amongst the wider regulatory community, and will be supported and benefit from project’s engagement activities (KPI5) and training activities (KPI11). PollinERA’s scientific advances (knowledge and tools) will be disseminated to a variety of interested audiences, through the publication of scientific peer-reviewed articles (KPI8), as well as scientific outreach via PollinERA conference session at a major conference (KPI12).

There will be a focus on promoting access, visibility, and longevity of the project results with the project’s strong emphasis on Open Science and usage of various established platforms (e.g., European Bee Hub, VEGAHUB, TKPlate, and publication of formal models, KPI1-4) for publishing all information, data, and results. A dedicated PollinERA topical collection of articles using novel publishing formats (M24) will be established in FESMJ. FESMJ is an innovative open access journal which facilitates the publication of models, datasets and software solutions in several areas (agriculture, food, social-ecological interactions, bioeconomy, natural resources, environmental sciences etc.). Novel types of articles (e.g., Formal Models) help documenting the outcomes of the full research cycle, including data, formal model papers, model validation studies, software, data analytics pipelines and visualisation methods. The topical collection will store and highlight the most notable project results, such as ecotoxicological knowledge gaps data sets, pollinator pesticide risk indicators, models for chemicals and pollinator populations, as well as a systems-based approach to pollinator ERA. Other project outputs (e.g., reports) will be available in the Research Ideas and Outcomes journal (RIO). In addition, the project will be registered / incorporated in the EU Pollinator Information Hive, INSPIRE Knowledge Base, and establish a link and explore collaborations with the Biodiversity Information system for Europe (BISE). These measures will enable exploitation of PollinERA outputs to a selected audience (see Table 8).

Communication Strategy (CS) – The main strategic objective of the CS will be to ensure that the PollinERA activities effectively reaches a wider audience and society as a whole, promoting the project and its results, generating awareness and interest. The communication will occur via appropriate means to translate activities into meaningful and tangible actions. A website will be the project’s communication and dissemination hub, ensuring visibility of project activities and results (KPI10). The website will have intuitive navigation and up-to-date content, and will: i) ensure general dissemination of project goals, structure and results; ii) provide secure storage in online libraries; iii) provide community links with other related EU Horizon projects; iv) offer regular dissemination of news, events, broadcasting an active dialogue through the project’s social network profiles.

The main outreach messages for PollinERA will focus on describing the beneficial impacts that the project results will have on insect pollinators, human health, and the environment. These messages will be targeted and tailored for general audiences, anticipated to be a broad variety of stakeholders potentially interested in our work, ranging from academia, field operators (such as beekeepers, veterinarians, farmers or farming advisors), environmental NGOs and naturalists, industry and their associations or citizens. All in all, we would like to target over 1-2 million EU citizens by the end of the project to improve their knowledge about the risks of chemicals to pollinators' health, become aware of the role and the importance of pollination and insects and ways to improve the risk assessment and management of chemicals on them. Supporting DCE activities will include accounts set-up on LinkedIn, Twitter, YouTube, Facebook with a target of ~50 posts / year / per account, +100 followers / year, >10,000 impressions / year.

Engagement Strategy (ES) – Complementing the project’s PEDR and CS, particular attention will be paid to promote and foster the involvement of key European stakeholders to maximise the acceptance and uptake of project exploitation and engagement. We have identified policy related institutions as main consumers of PollinERA’s outcomes, some of which will play a special role, in the co-development engagement activities (e.g., members of its External Advisory Board). For the ES, a preliminary list of potential stakeholders is provided in Table 8, highlighting (*) those proposed as members of the PollinERA External Advisory Board (EAB), to be contacted when the project starts. A variety of European stakeholders will be actively engaged in the project to improve partners’ expert capacity, oversee, or evaluate calibration/validation the tools proposed and quality of regulatory proposals. This engagement will be done through interactive thematic workshops and training sessions (KPI5 & KPI11). In addition, we will engage with projects having related activities/goals (e.g., PARC, SPRINT, BETTER-B, MUST-B or future Horizon Europe winning projects), aiming to benefit from and maximise the knowledge exchange with these projects throughout the duration of PollinERA project. Direct communication and information exchange with relevant Biodiversa+ projects and JRC Knowledge Centres is envisaged.

PollinERA will develop its Intellectual Property framework (IP) in connection with its Data Management Plan (DMP). Management of intellectual property and foreseen protection measures generated will be managed in compliance with the Consortium Agreement (CA), which will be signed at the beginning of the project. The CA will address background and foreground knowledge, ownership, protected third party components, and protection, use, and dissemination of results and access rights. The principles are:

• Background information and knowledge contributed to the project by each participant will be listed in the CA. This information will be provided royalty-free for the implementation of the project’s tasks.
• Results shall be owned by the participant who generated them. Each participant will be responsible for ensuring fulfilment of their obligations under the Grant Agreement regarding results by planning with any third parties that could claim rights to them.
• Whenever results have been produced jointly, the ownership of the results will be shared among the participants who carried out the work. The terms to be agreed upon via a joint ownership agreement.
• Each participant will be responsible for examining possibilities to protect results that may be commercially or industrially exploited. Participants will ensure that adequate steps towards protection are taken prior to DECE activities, preventing unapproved public disclosure of results, models, tools, and data.
• Access rights to results will be granted on a royalty-free basis for further research and commercial exploitation.

The Fig. 6 Gantt chart collates the PollinERA deliverables (see Part A) together with the task timelines and milestones (see Part A).

Figure 6.  

Gantt chart showing the overall plan for implementation of deliverables and timelines. The coloured bars in the chart depict the timing of the tasks. The timeline (in months) is given on the top x-axis, task numbers on the left y-axis, and WP numbers on the right y-axis. Deliverables (black squares) and milestones (white squares) are indicated by their number in task bars at their due time.

The PollinERA consortium is comprised of 11 partners, of which six are universities, two are research institutions, two are SMEs, and one is an NGO.

The consortium partners were chosen with care each to fulfil key roles in the project addressing the project objectives. Most of the partners have experience from participation in H2020 projects and have worked together on previous or current research projects, which is beneficial to the partner workflow of PollinERA (see Fig. 7).

Figure 7.  

Visualisation of past & present collaborations between PollinERA partners. Collaborating partners are indicated by partner number at activity (dark hexagon). Funding bodies acronyms are indicated with underline, & project names are included below. Task force (TF), panel, & working group (WG) activities are indicated in italics. PollinERA partners in blue are universities, red are research institutions, green is an NGO & brown are SMEs.

Partners work broadly integrated across the project but with key specialised experience grouped to the specific objectives (see Table 10), representing SO1 (UJAG, ULUND, UNIBO, INC-PAS & UOS), SO2 (UJAG, ULUND, UNIBO, INC-PAS & SLU), SO3 (AU, UJAG, UOS & IRFMN), and SO4 (AU, UJAG, ULUND & BEBC). The SME partners (PEN & ZIP) are specialists in communication and dissemination activities, and database development and management, respectively.

PollinERA partners come from a wide geographical area (Fig. 8) and from different areas of expertise.

Figure 8.  

Map of nationality of partners in the PollinERA consortium.

The contributions of partners to the major SO sub-tasks are shown in Table 10. Most partners contribute across the project, and their key expertise is listed below.

Aarhus University (AU) coordinates PollinERA. The coordinator is Professor Christopher John Topping who leads the SESS centre. He is a deputy coordinator for H2020 EcoStack, and in the project management group of H2020 B-GOOD, as well as a partner in H2020 PoshBee, and coordinator of the EFSA bee simulation modelling project, ApisRAM, and the ELONTA project. He is also editor in chief of the FESMJ, which will be used to provide a platform for model dissemination, and vice-chair of the EFSA PPR panel. The coordinator will be supported administratively by the project office, and scientifically by all PollinERA partners on a day-to-day basis. The project office comes from AU’s Research Support Office management team and has considerable experience of Horizon project management. The Research Support Office contains a broad range of administrative, clerical, and financial experience, drawing support from AU Contracts, Finance and Legal Departments. The coordination group is further assisted by the SESS Centre Administrator who is experienced in supporting Horizon projects.

AU also has a major role in SO3 & SO4. The SESS team have extensive experience of developing large-scale simulation models as well as leading multi-disciplinary projects, including the EFSA’s IPol-ERA project. The core of the PollinERA framework for systems ERA will be based around ALMaSS, developed by the coordinator over 25 years, and now forming part of the toolbox used by EFSA for pollinator risk assessment (ApisRAM). The team has very strong programming and modelling and expertise. From a social science and humanities (SSH) perspective, James Henty Williams (AU) also provides key support and management experience for the multi-actor approach to H2020 projects (WP leader in B-GOOD for multi-actor approach), as well as EFSA funded projects (e.g. EFSA PERA and IPol-ERA) and will bring this experience and expertise in developing and facilitating workshops, undertaking interviews and establishing active forums for knowledge exchange, all contributing to WP5’s stakeholder engagement activities. SESS has access to multiple higher performance computing resources available for use by PollinERA.

The project management is complemented by WP6 Communication, Dissemination and Exploitation led by Pensoft Publishers who support communication and dissemination activities for the entire project. Moreover, the team will contribute to the stakeholder mapping process and will assist the organisation of various workshops and other events. In addition to supporting WP6, PEN will also contribute to WP7 Project Management. PEN has been providing science communication for over 30 projects, they are a science communication expert, and also a science publisher. They have published more than 1000 books and e-books. PEN is well-known among academics worldwide with its technologically advanced peer-reviewed Open Access journals. The company is actively developing new tools, workflows, and methods for text- and data publishing, dissemination of scientific information and technologies for semantic enrichment of an articles’ content.

WP6 is supported by BeeLife European Beekeeping Coordination, BEBC. As chair of the EU Bee Partnership (stakeholder group centralising pollinator related data), participant to European Commission stakeholder advisory committees (DGs Agriculture, Santé and Environment) and ecotoxicologist engaged in the evolution and implementation of pesticide RA and management on bees since 2009, Dr. Noa Simon Delso has vast knowledge of the regulatory framework, related institutional and external stakeholders, and policies related to the topic. She also follows the European policy developments impacting pollinators, e.g., F2F, CAP, Natural Restoration Law, SUR. Thus, not only will she strongly support the engagement and exploitation activities in WP6, she will also assist the regulatory and policy-related activities in WP5. She is the coordinator of the EU Pollinator Hub initiative that she develops with colleague Dr. Michael Rubinigg, plant physiologist and data science expert in data curation, standardization, and quality management. Dr. Rubinigg peer reviews the data integrating the EU Pollinator Hub and assists Gregor Sušanj (ZIP) in the architecture of the EU Pollinator Hub, by defining data acquisition standards, procedures, and validation. Given Dr. Rubinigg's expertise, he oversees the Data Management Plan (DMP) (Task 7.d). The Hub will be connected to the PollinERA exploitation activities as it enables the FAIR data principles that governs the project and integrate PollinERA data with data from other projects and initiatives.

Jagiellonian University (UJAG) Team is led by Professor Ryszard Laskowski, an ecologist and ecotoxicologist with broad experience in studying effects of trace metals and pesticides on invertebrates and microbial-mediated soil processes. He has (co)authored 6 books and over 120 peer-reviewed papers and book chapters. In 2012-2018 he was a member of EFSA’s Scientific Panel on Plant Protection Products and their Residues. PI in 20 major research projects, including 5 EU-funded, listed among 2% of most influential scientists in the world (Stanford University & Elsevier 2021, 2022). Researchers engaged published broadly on different aspects of ecotoxicology and, more recently, landscape models, including experience with ALMaSS and pesticide simulation. The studies cover the effects of a variety of inorganic and organic pollutants (especially pesticides) on soil microbial processes, soil-dwelling and epigenic invertebrates as well as pollinators and other arthropods. The team is expert in acute bioassays and chronic tests, using different exposure routes, such as topical, oral, spray, or coated glass assays. The laboratories are equipped with, among others, HPLC MS/MS, GC MS, three AASs, Potter tower, Micro-Oxymax respirometer, and high-quality walk-in climatic chambers.

Lund University (ULUND) is leading the work on developing and trialling a pesticide and pollinator co-monitoring scheme and pesticide risk indicators specific for insect pollinators. The leader is Dr. Maj Rundlöf, with nearly two decades of working and publishing extensively on insect pollinators in agricultural landscapes and exposure and effects of pesticides on bees. She is member of the EFSA working group Bee guidance revision and Biodiversity Theme leader within the Strategic Research Area Biodiversity and Ecosystem services in a Changing Climate (BECC). She has participated in several European projects (e.g., STEP, PoshBee, PARC) and coordinated national projects (e.g., DELETE, on developing landscape ecotoxicology for bees, and MixToxBee, on pesticide mixture exposure and effects on bees for monitoring and ERA). Dr. Lars Pettersson, responsible for the Swedish butterfly monitoring scheme and pollinator scheme piloting and partner in the SPRING project supporting the implementation of the EU-PoMS, and Dr. Lina Herbertsson, with expertise on sublethal effects of pesticides on bees and pollination services and partner in the iENT project on neuroenvironmental toxicity of pesticides to pollinators.

Institute of Nature Conservation of the Polish Academy of Sciences (INC-PAS) has extensive experience in testing the acute and chronic effects of pesticides and their mixtures on solitary bees’ and other NTAs’ survival, physiology and biochemical markers and is leading the work on mixture effects. The leader is Dr Agnieszka Bednarska who specialises in ecotoxicology with the main interest in the effects of stressors (including, but not limited to pesticides) and their interactions on life history traits and physiology of invertebrates (including insect pollinators), and their consequences for the population level. She participated in a number of research projects founded by the main Polish public bodies (NCN projects) being PI in 4 of them and international projects (e.g., MUSBERA, EcoStack, CREAM - Mechanistic Effect Models for Ecological Risk Assessment of Chemicals, NoMiracle - Novel Methods for Integrated Risk Assessment and Cumulative Stressors in Europe), covering different exposure routes and both laboratory- and field-based studies. She has co-authored 40 peer-reviewed papers in international journals and serve as Associate Editor in Ecotoxicology journal. She is employed as an assistant professor at Institute of Nature Conservation, Polish Academy of Sciences (full position) and Institute of Environmental Sciences, Jagiellonian University (part time). Some of the facilities necessary to run the experiments and/or chemical analyses are shared with UJAG, making the two teams highly compatible and interoperable.

Mario Negri Institute (IFRMN), the WP2 leader, is a non-profit research organization with major activities in drug research. Dr. Emilio Benfenati, head of the Department of Environmental Sciences, has a long experience in toxicity and environmental modelling. He coordinated and participated in tens of European and international projects, like sOFT-ERA (an EFSA-funded project with the aim to develop in silico models to generate predicted values for the EFSA database OpenFoodTox) and JANUS (a German UBA-funded project with the aim to develop a prioritization system for environmental assessment), in particular. He is the author or co-author of more than 500 papers (including models to estimate the toxicity towards Apis mellifera) in international journals and has edited a few books. He will be assisted by a team of experienced researchers to develop in silico toxicology and fate models for SO3.

The Bologna University (UNIBO) team is led by Fabio Sgolastra, Associate Professor in General and Applied Entomology at the University of Bologna. He has 20 years of experience in bee research, which mainly focuses on ecotoxicology and conservation of managed and wild pollinators in the agroecosystems and the valorisation of the pollination service. He is/was actively involved in several European projects (e.g., FP7 AMIGA, LIFE4POLLINATORS, PRIMA DREAM) and innovative and policy-oriented initiatives linked with bee health (e.g., EFSA working groups for the preparation of Scientific Opinions and Guidance Documents, EIP-AGRI Focus Group on Bee health and Sustainable beekeeping). Members of UNIBO team are also Professors Giovanni Burgio and Maria Luisa Dindo. Their expertise in insect rearing and management will contribute to developing protocols for the new model species.

University of Osnabrück (UOS), is represented by Professor Andreas Focks with more than 15 years of experience in development and regulatory application of TKTD models. He was member of the EFSA working group on TKTD models and of the EFSA working group on the bee guidance revision, where he is responsible for integration of dynamic effect modelling. His experience includes the development, usage, and evaluation of population models for environmental risk assessment of chemicals, amongst others he is chairing the SETAC working group on Model acceptability criteria and scenario development.

Swedish University of Agricultural Sciences (SLU) main contribution is the pesticide analysis for all samples collected in PollinERA as part of the source and route characterization of pesticide exposure in the key pollinator groups (wild bees, butterflies, hoverflies, moths). SLU has long experience with monitoring pesticides in environmental samples, including expertise in development of analytical methods to detect pesticides residues in complex environmental matrices. SLU was involved in a number of research project about pesticide residues in pollinators, such as BiNeo, MixToxBee, and AirBeeSafe, and currently in PARC 6.4.4.

Združba IP, d.o.o., under brand name ZIP Solutions is a web development SME. ZIP is experienced in the field of digital marketing and website development, with project including the development of the Pollinator Hub from TRL1 up to TRL8. Other noticeable projects include Austrian Varroa Alert System (bienengesundheit.at) as well as Beehive wandering exchange for Niederösterreich and Oberösterreich in Austria. On all these projects Gregor Sušanj, full-stack software engineer, owner and CEO of the company was the lead architecture designer and developer. As well as main contributions to the SO4 framework development, ZIP will also support technical development in WP3, WP5 & WP6.

The partner group as a whole matches the Part-A specific aspects of the call matching each aspect with key experts in the field, covering pollinator ecology and pesticide exposure (ULUND, UNIBO), toxicological testing including mixtures and development of sensitivity distributions (UJAG, INC-PAS), predictive in silico modelling (effects, classification, mixture effects and fate) (IRFMN), TKTD modelling (UOS), pollinator monitoring and risk indicators (ULUND), pesticide monitoring (SLU), pesticide risk assessment (AU, ULUND), landscape simulation modelling (UJAG, AU), and agent-based pollinator population models (AU), policy (BEBC, ULUND) and open databases/hubs (ZIP), stakeholder engagement (AU, BEBC) and communications (PEN).