The 6/04 standard of the European and Mediterranean Plant Protection Organisation (EPPO) on the safe use of biological control is a decision-support scheme (DSS) for the import and release of biological control agents in Europe. It was recently developed by the Joint EPPO/International Organisation of Biological Control (IOBC) Panel on Biological Control Agents. The DSS can be used to assess the potential environmental impacts of biological control agents. It is valid for different types of biological control: classical and augmentative biological control of invertebrates, pathogens and weeds. However, the DSS is not yet widely implemented in Europe and, during preliminary trials, it was found that its broad range of usages could lead to some confusion or misunderstandings, as well as requiring unnecessary information in some cases. Thus, the scheme was modified to specifically assess classical biological control against plant pests, i.e. the introduction of exotic natural enemies of plant pests for establishment and long-term control. The new version of the scheme was then tested on two parasitoids that are presently being released in Europe, the figitid Ganaspis cf. brasiliensis against the spotted wing drosophila Drosophila suzukii and the mymarid Cleruchoides noackae against the Eucalyptus bronze bug Thaumastocoris peregrinus. Both parasitoids were successfully assessed with the new version of the DSS. No major issues were encountered during the assessments and most questions were answered with low levels of uncertainty. Both assessments concluded that the parasitoids were safe to release in the impact assessment areas, with positive impacts exceeding negative impacts. Suggestions for potential improvements are provided.

bronze bug, Drosophila suzukii, importation, invasive species, Thaumastocoris peregrinus

The number of invasive plant pests, in particular herbivorous insects and plant pathogens, is increasing dramatically around the world, despite improved quarantine measures and border security (Santini et al. 2013; Roques et al. 2016; Seebens et al. 2017). This increase is largely due to the exponential growth of the transcontinental trade of fruits, vegetables, live plants and other plant products, such as wood packaging material, seeds and cut flowers, on which plant pests are travelling inconspicuously (Kenis et al. 2007; Santini et al. 2018). Once established, invasive plant pests are difficult to manage because of their uncontrolled development in regions where natural enemies are lacking and host plants have not developed resistance mechanisms against them (Desurmont and Pearse 2014; Garnas et al. 2016; Lovett et al. 2016; McLaughlin and Dearden 2019).

One method for long-term management of invasive plant pests is classical biological control (CBC), i.e. the introduction of natural enemies of exotic origin to control invasive pests, aiming at permanent control of the pest (Van Driesche and Bellows 1996). CBC has been implemented since the 19^th century, mostly against invasive insects using parasitoids and predators. Over 700 insect pests have been targeted, with significant successes (Cock et al. 2015, 2016; Kenis et al. 2017). However, in some cases, negative effects have been observed on non-target species, especially when polyphagous parasitoids or predators were used (Hajek et al. 2016). Therefore, nowadays, the introduction of CBC agents usually follows strict procedures, including a rigorous assessment of their potential non-target effects (Van Driesche and Hoddle 2016). Even so, regulatory frameworks and procedures for the assessment and release of CBC agents vary significantly amongst countries and continents, according to their own legal and societal context and their history of using biological control practices (Afonso et al., in press).

In Europe, the use of classical biological control is hampered by the heterogeneity of national legislation (when existing) and practices and by the lack of common regulations. The European and Mediterranean Plant Protection Organisation (EPPO) has recently published the standard PM 6/04 on the safe use of biological control, presented as a decision-support scheme (DSS) for the import and release of biological control agents of plant pests (EPPO 2018). It was written mainly for decision-makers and assessors of biological control applications to harmonise assessment procedures within EPPO countries. It provides detailed instructions for environmental impact assessment for biological control agents of plant pests for augmentative biological control (ABC) and classical biological control. Although its title states it is for biological control agents of plant pests, it can also be applied to biological control agents of weeds.

However, the EPPO PM 6/04 standard is not yet widely implemented in Europe. In this paper, we tested it to assess the potential as well as safety of two exotic parasitoids, Ganaspis cf. brasiliensis (Ihering) (Hym., Figitidae) and Cleruchoides noackae L in and Huber (Hym., Mymaridae), for release against Drosophila suzukii Matsumura (Dipt., Drosophilidae) and Thaumastocoris peregrinus Carpintero and Dellapé (Hem., Thaumastocoridae), respectively. In preliminary trials, we noticed ambiguities due to the fact that the DDS was made to cover both augmentative and classical biological control agents (including native species) in addition to invertebrate and weed biological control. Therefore, we modified the scheme and retained only questions relevant for CBC against invertebrates using exotic natural enemies.

The EPPO standard PM 6/04 “Decision-support scheme for import and release of biological control agents of plant pests” (EPPO 2018) was prepared by the Joint EPPO/IOBC Panel on Biological Control Agents. It follows the recommendations of the International Standard for Phytosanitary measures 3 (ISPM 3) “Guidelines for the export, shipment, import and release of biological control agents and other beneficial organisms” (FAO 2017) and complements the earlier EPPO Standard PM 6/2 “Import and release of non-indigenous biological control agents” (EPPO 2014) by converting the guidelines into a question-based decision-support scheme.

The DSS is based on a sequence of questions that aim at deciding whether the introduction of a biological control agent (BCA) could cause unwanted environmental impacts and to compare the likelihood and impact of such negative effects with potential positive environmental effects. The DSS is a comprehensive document and rather unique amongst protocols and standards to assess the potential environmental impacts of biological control agents. To the best of our knowledge, no other DSS exist for environmental impact assessments (EIA) of biological control agents. It is valid for different types of biological control: classical and augmentative biological control of invertebrates, pathogens and weeds.

The document consists of two main parts: (I) an express scheme which may produce a rapid result and (II) a comprehensive scheme for certain cases of biological control, for which the express scheme does not lead to sufficiently clear recommendations. Within part I, the following steps are included: Step 1 Initiation; Step 2 BCA categorisation; Step 3 Impact assessment (four questions); Step 4 Decision taking. Part II consists of: Step 1 Pre-assessment (18 questions); Step 2 Assessment of probability of establishment (20 questions); Step 3 Assessment of probability of spread (three questions); Step 4 Assessment of potential environmental consequences (37 questions); Step 5 Recording the degree and types of uncertainty (one part); Step 6 Conclusion of the EIA. The latter is subdivided into the categories: establishment; spread; environmental impact; overall conclusion.

The DSS includes questions about any potential earlier assessments (section 1.2 in the Express assessment and 1.04 in the Full assessment), which may save resources. If the BCA, or a very similar organism (e.g. another population of the same species), may have been subjected to an EIA process before, nationally or internationally, this may partly or entirely replace the need for a new assessment. These questions could be enhanced by requesting information about previous releases of the same BCA, for example, in similar regions/conditions (if yes, how long ago?) and what was the outcome in terms of establishment, impact on target species and non-target species? For both case studies, G. cf. brasiliensis and C. noackae, this would be necessary and useful information to render the decision process for or against releases in additional countries more efficient. For example, information of the outcome of prior releases of C. noackae in South Africa, South America and Israel were instrumental for the release application in Portugal. Similarly, for the G. cf. brasiliensis release application in Switzerland, results from studies on releases or adventive distributions in North America and Italy were considered important information for the risk assessment.

The DSS provides the opportunity to include undescribed cryptic species as potential BCAs, as it includes consideration of taxonomic levels higher or lower than species, including those characterised using molecular methods (Step 2 BCA categorisation of the Express assessment and step 1.02 of the Full assessment). This is important in high priority cases like the CBC of Drosophila suzukii, where only one genetic group of the parasitoid G. cf. brasiliensis (a probable cryptic species) is specific to D. suzukii larvae feeding in fresh fruits.

The DSS starts with the Express assessment for invertebrates and pathogens whereas, for weed biocontrol, it is recommended to start directly with the Full assessment. This is based on the higher risk of weed BCAs becoming pests of important crops. While the Express assessment may be convenient in some cases and save resources such as time and money, especially for augmentative BCA, few countries, if any, in the EPPO region would allow an Express assessment for new CBC agents of plant pests. This is especially the case in countries like France, Israel, Italy or Switzerland, where national authorities require comprehensive risk assessments for the release of any non-native species. Therefore, any new CBC, whether it is introduced against weeds or plant pests, should only be evaluated using the Full assessment.

The sections about agent establishment (Step 2) and spread (Step 3) in the Full assessment are very detailed and comprehensive. While for the assessment of agent establishment, a distinction between different types of biological control (ABC vs. CBC) is possible, this is not the case for the assessment of spread. In ABC, agents are released in the field or in greenhouses to augment the numbers of a species that are native or already established. In the case of ABC for protected crops, an agent is regularly released for temporary control. Establishment in the wild and spread of the released populations of an agent are, in most cases, undesired in ABC, because the aim is to minimise unwanted effects on the environment and non-target species. In contrast, CBC is the intentional introduction of a species into an area where it is not indigenous and, to be successful, it must establish and spread on its own after release. Therefore, for CBC, the section dealing with establishment and spread of a BCA is relevant to assess the feasibility, efficiency and economic risk of a CBC programme, but not for an EIA because the establishment and spread to all areas where the host/prey is present has to be assumed. However, this means that the EIA of CBC agents must very thoroughly assess possible non-target impacts (especially including host specificity) in the entire geographic area to which the impact assessment applies. In contrast, ABC is applied in a defined place such as a cropland and it frequently involves generalist agents. The spread of released BCAs to areas outside the IAA can, therefore, have a detrimental effect on native species and their populations. In the current DSS, the same questions about establishment and dispersal are asked for both ABC and CBC and it remains unclear from the document if the answer is actually a benefit (e.g. high probability of spread and establishment for CBC agents) or a risk (e.g. high probability of spread and establishment for ABC agents). A solution to this problem would be to have a separate DSS for the different types of BC. In a DSS for CBC, many questions about establishment and dispersal would then not be necessary and could be simplified into just a few questions to ensure that establishment is likely. As for the spread, the questions could be totally suppressed, since an established CBC agent will spread anyway, no matter the rapidity of spread. If the natural spread is too slow, the BCA can be distributed through a release programme. Additionally, other questions, for example, whether the BCA is already present or indigenous in the IAA or may cause transient effects, would then become obsolete. Applied to the case studies, a first introduction of the CBC agents into a European country would imply that establishment and spread into neighbouring countries, where the target and its host plant are present and where climate is suitable has to be assumed and thus, the IAA needs to include these countries. In the case of C. noackae, all southern European countries where Eucalyptus trees are planted would be included and, for G. cf. brasiliensis, all European countries.

Several questions are strictly related to CBC of weeds. While it is clearly mentioned in the DSS that such questions have to be answered only in the case of CBC of weeds, these questions needlessly lengthen the scheme under which CBC agents of invertebrates are assessed. Furthermore, the potential impact of weed BC agents (herbivores and plant pathogens) are different from those caused by invertebrate BC agents (parasitoids, predators and entomopathogens), so that a full separation from CBC of weeds and pathogens would allow more precise responses to questions on impact mechanisms, establishment and spread.

One of the clear strengths of the DSS is the consideration of positive environmental impacts of BCA releases. This is a great step forward compared to most processes that solely consider risks. While the selection of the risks and benefits that are included in the evaluation provide an objective baseline, the decision whether the positive impacts exceed the negative ones, remains a subjective expert opinion. In CBC, decision-making processes should weigh both the risks and potential benefits. These benefits should include environmental and socio-economic considerations, such as direct and indirect benefits for farmers/producers and the general public. For example, in the case of G. cf. brasiliensis releases against D. suzukii, environmental benefits (e.g. reduction of pesticide use and reduction of infestation of wild fruits), as well as socio-economic benefits (e.g. reduction of costs for protecting crops), would be expected and these reduce the high losses fruit growers have through this invasive species (Burrack 2015; De Ros et al. 2015; Knapp et al. 2020). However, if the goal of the DSS is to be purely an environmental risk-benefit assessment, this latter suggestion might not be applicable.

Another strength of the DSS is the recording of the degree and types of uncertainty (Step 5 in the Full assessment) when it comes to potential environmental consequences. As some aspects of the risks and benefits may be based on subjective expert opinions, limited data or only on available data from closely-related species, the level and type of uncertainty helps to identify weaknesses in the assessment, research needs or even unacceptable risks for approving releases. What is unclear in this step is the definition of the consequences if the uncertainty is high. A probable consequence in the case of high uncertainty could be the requirement to gather more data through literature reviews or additional tests. Thus, there should be clear directions on what steps should be taken if uncertainties are high.

The overall conclusion of the DSS (Step 4 Decision taking in the Express assessment, Step 6 of the Full assessment) should be a clear acceptance or rejection of the application to import and release a BCA. While this is sufficiently clear in the Express assessment through the binary outcome in sections 4.3 (recommendation for releases) and 4.4 (no recommendation for releases), it is not clear in the Full assessment. However, the Full assessment allows the decision-maker to comment on and justify the conclusion in each of the three major sections: establishment, spread and environmental impact.

Based on the considerations above, we have modified the EPPO Standard PM 6/04 (1) Decision-support scheme for import and release of biological control agents of plant pests (EPPO 2018) for specifically assessing the potential for releases of CBC agents of plant pests (Suppl. material 1). Only modifications essential for this specific type of assessment were made. Other changes, in the format and the phrasing, to be precise, have been kept to a minimum. As for the original standard, the decisions are based only on environmental impacts (negative and positive), not on socio-economic considerations.

The main modifications are as follows:

• The express scheme has been deleted.
• All questions specifically related to weed biological control have been deleted.
• All questions are now unambiguously concerned with classical biological control, i.e. the introduction of an exotic BCA (usually parasitoid or predator) against a plant pest, for a permanent establishment of the BCA leading to long-lasting control of the target pest. Some questions have been modified and all questions referring more specifically to augmentative biological control have been suppressed.
• The section on the assessment of probability of spread has been deleted.

The new scheme is presented in Suppl. material 1. The modified DSS now starts with a short pre-assessment (Step 1) to present the BCA and to ensure that it is a priori suitable for classical biological control in an IAA. Step 2 involves assessing the probability of establishment in the IAA. Step 3 consists of an assessment of potential environmental consequences. The degree and types of uncertainty are considered in Step 4 and Step 5 provides the conclusion of the EIA.

Both G. cf. brasiliensis G1 and C. noackae were successfully assessed with the modified version of the DSS for CBC of plant pests. No major issues were encountered during the assessments, most questions being answered with low levels of uncertainty. For both species, the outcome of the assessment was that the parasitoids were safe to release in the IAA, with the positive impacts exceeding the negative ones. The scheme would need to be further tested on agents that are less specific; however, there is evidence from unpublished assessments with polyphagous natural enemies where no major issues were apparent and the important risks were clearly highlighted.

The two parasitoids assessed in this study were well known due to previous releases in other regions in the case of C. noackae and because of extensive studies in Europe in the case of G. cf. brasiliensis G1. Potential CBC agents that are less studied would be more difficult to assess, resulting in higher uncertainties. The DSS could mention that species that have been well studied and are ready to be released are better suited to this assessment. On the other hand, the scheme could also be used at an earlier stage of a CBC programme to identify priorities for research, such as host range testing or climate matching models.

The goal of modifying the DSS was to build a version that is specifically and unambiguously made for assessing classical biological control agents of plant pests, but without changing the general aim and structure of the scheme and keeping other changes, for example, in the wording, to a minimum. However, the scheme could be further developed in various ways, outlined below.

1. Step 2. Assessment of probability of establishment could be further shortened. Several criteria for assessing establishment of the agent are still considered in our revised version in an effort to find a compromise, but most of them are more relevant for ABC or for CBC of weeds. In CBC of plant pests, the most relevant criteria are factor 1 about the presence of the hosts or prey and factor 3 about a suitable climate. Questions on other abiotic factors, on competition and natural enemies preventing establishment or on the adaptability of the BCA agent may be important in specific cases, but are often difficult to answer. Questions on management practices and control measures that may prevent establishment may be relevant, but all CBC practitioners will first release CBC agents for establishment at sites that are not managed with pesticides. The question on the presence or need of alternative hosts and other essential species is ambiguous and could be rephrased. For CBC, attacks of alternative hosts, i.e. those that occur at the same time as the target species, would be undesired, unless it is another pest that needs to be controlled. For some agents, the presence of an alternate host, i.e. a host that can be attacked before or after the occurrence of the target pest, may be crucial for the establishment in a new area and, thus, important for the success of a CBC programme. On the other hand, if the alternate host is native, it implies that non-target effects will occur. In most CBC programmes, no alternative nor alternate host is required for its success.
2. Step 3. Assessment of potential environmental consequences is based on a comparison and balancing between the negative (risks) and the positive (benefits) environment impacts. However, they are not assessed equally. In the positive impact assessment, only environmental criteria are included. In contrast, in the negative impact assessments, there are also four questions on negative impact on ecosystem services (3.02.12–3.02.15), based on the EFSA (2016) criteria. Impacts on ecosystem services implies socio-economic impacts. For example, provisioning ecosystem services includes provision of food, raw materials etc. In addition, impacts on cultural ecosystem services also clearly imply socio-economic impacts. Often, a decision to release a CBC agent takes into consideration the balance between environmental risks and socio-economic benefits. Thus, the questions of the DSS should explore both the negative and positive impacts for the environment, as well as from the socio-economic point of view. If possible, a positive socio-economic impact component should be included in the DSS, for example, a set of questions determining the socio-economic benefits of BCA releases and the expected reduction in pest impact. If the scheme has to remain purely an environmental risk assessment, the questions on negative impacts on ecosystem services (3.02.12–3.02.15) should be removed, since the negative environmental impacts are sufficiently covered by other questions.
3. Finally, a well-defined rating system could be developed for possible negative and positive impacts by the CBC agent to facilitate the overall conclusion and make the final decision transparent and clear. In such a rating system, the level of uncertainties could be used to facilitate the overall conclusion derived from the DSS. For example, if uncertainty about potential non-target effects is high, the rating system could suggest declining the release application until the uncertainties are resolved. We acknowledge that this is a controversial subject, which has previously been tackled by using risks versus benefits matrices (e.g. Moeed et al. (2006); van Lenteren and Loomans (2006)). The main difficulty with such scoring systems is that the assignment of values is subjective and required total scores for decision-making may vary between jurisdictions.

Besides the use of the DSS by decision-makers, other proponents of a biological control agent release, researchers, for example, can use the document to identify and gather the information that needs to be provided in an application for releases to regulatory authorities. It is important to point out here that the DSS is not a substitute for a release application, but it could be used to develop a standard template for preparing an application for CBC of a plant pest.

We thank the Joint EPPO/IOBC Panel on Biological Control Agents for fruitful discussions on the decision-support scheme. The research leading to these results received funding from the European Union’s Horizon 2020 Programme for Research & Innovation under grant agreement no. 771271 (HOMED). MLS and MK were also supported by CABI with core financial support from its member countries (see https://www.cabi.org/about‐cabi/who‐we‐work‐with/key‐donors/). CA was supported by a PhD grant (SFRH/BD/135845/2018) from Fundação para a Ciência e a Tecnologia (FCT). MB was supported by Forest Research Centre (CEF), (UID/AGR/00239/2019 and (UIDB/00239/2020), Laboratory for Sustainable Land Use and Ecosystem Services – TERRA (LA/P/0092/2020), research units funded by FCT, Portugal.