Estimating the Impact of Pesticide Use Reduction Policies on Irish Cereal Yields Using an Iterative Expert Panel Methodology

Abstract

The European Union’s (EU) Farm to Fork strategy seeks to reduce agricultural pesticide use by 50%, both of total pesticide use and of the most hazardous chemicals. While Ireland has achieved the goal of overall pesticide use reduction, more progress is needed regarding more hazardous substances. Ireland’s cool damp climate is unique within the EU, and with little empirical data on the possible impacts of achieving this goal on Irish farming, we sought to estimate these in cereal crops using a ‘Delphi’ style iterative expert panel methodology, conducted over two rounds, rather than until consensus was reached, to allow for knowledge gaps that may exist to become apparent. A total of 17 crop production experts with at least five years of relevant experience were surveyed anonymously, and then their answers were compiled and fed back to them, allowing participants to revise their responses based on the opinion of the group. Participants estimated that reduced use of more hazardous fungicides and insecticides could both reduce yields by 10–15%, while reduced use of herbicides would reduce yields of winter cereals by up to 30%. These impacts are substantially higher than those predicted in other Europe-wide studies. Application of additional Integrated Pest Management measures was estimated to reduce yield loss but not entirely mitigate it. These findings highlight the economic and food security trade-offs that may be required to achieve the Farm to Fork strategy’s goals.

1. Introduction

The use of chemical pesticides to control fungal pathogens, weeds and insects that impact agricultural production has played a key role in the growth of crop yields since the 1950s [1], allowing modern farming systems to feed a global population of over eight billion people. Estimates by both Oerke [2] and Savary et al. [3] project that, without pesticide use, pest outbreaks would lead to annual global losses in staple crops of 55–85%—a significant rise compared to the typical losses of 15–55% currently attributed to pest impacts.

However, despite this clear benefit of pesticide use, these chemicals come with many drawbacks. Pesticides are believed to be one of the leading causes of large declines in the biomass of insects [4] and other fauna across the globe [5], along with having impacts on water and air quality and human health [6,7] and requiring high levels of fossil fuel use in their production [1]. There is a high level of general awareness of these impacts and thus increasing public concern about pesticide use [8].

Given these concerns, many governments have moved to enact policies to reduce pesticide application through measures such as bans or use limits on certain chemicals [9]. One of the largest of these is the European Union’s (EU) ‘Farm to Fork’ (F2F) strategy, which includes, amongst other goals related to agricultural environmental performance, an ambition to reduce pesticide use by 50%, both overall and of the most hazardous chemicals (either via absolute use reduction or substitution with less hazardous substances), by 2030, relative to a baseline period of 2015–2017 [10].

As an EU Member State, Ireland is seeking to meet these targets and as of 2022 (the most recent year for which data is available at time of writing), had made substantial progress in the goal of overall pesticide use reduction, having reduced use by 45% from the baseline [11], as a number of substances that were extensively used previously had their registrations withdrawn [12]. However, with regard to the use of the more hazardous pesticides, defined as active ingredients listed as candidates for substitution by the European Commission, less progress has been made, with only a 26% use reduction having been achieved [11].

Given the important role that pesticides play in protecting crops and underpinning yield and quality, significant public concern has been raised about the impact that achieving these F2F targets may have on Irish farming [13]. This reflects the concerns raised in studies that have attempted to predict the impacts of the F2F targets on farm outputs (e.g., [14,15,16]). However, these studies either examine the entirety of the EU in aggregate (in the case of Barreiro-Hurle et al. [16] and Beckman et al. [14]) or examine a specific group of member states that does not include Ireland (in the case of Bremmer et al. [15]). Given that Ireland occupies its own unique agro-ecological environment amongst EU member states [16], with a climate that is favorable to high cereal yields but also to high levels of infection by fungal pathogens [17], it is not appropriate to simply assume that these figures, or figures contained in global reviews of pesticide use reduction, can be accurately applied to the Irish context. Thus, Irish-specific data is required to accurately predict how crop yields may be affected by this reduced pesticide usage; however, little data of this kind exists either for Ireland or for the UK (the only nation with a comparable climate).

Another area where Irish-specific projections would be useful is in regard to the efficacy of Integrated Pest Management (IPM) measures. IPM is the practice of using a range of techniques, such as physical, cultural and biological control, to manage pests, with pesticide use being considered the technique of last resort [18]. The increased use of IPM is a goal of the Irish Government’s National Action Plan for the Sustainable Use of Pesticides [19], and increased uptake of IPM would be a likely outcome of farmers having reduced access to chemical pesticides. However, the efficacy of IPM techniques varies substantially as a result of local factors such as crop type and surrounding ecosystem [20], and thus international examples are once again not appropriate for predicting what measures may be effective in Ireland’s unique environment.

A third area related to reduced pesticide use, where Irish-specific data would be valuable, is the risk of resistance development. Pesticide resistance develops in a population when a gene producing resistant phenotypes becomes favored. This often occurs as a result of over-reliance on a small number of pesticide modes of action, which causes high selection pressure in favor of resistance genes [18]. In circumstances where access to some of the suite of pesticides currently available to farmers is reduced, the potential for resistance development is a concern, as farmers would have fewer pesticides available in their repertoire.

Ideally, field trials carried out under Irish conditions would be used to gather the data discussed above; however, only a handful of empirical studies examining the impacts of reduced pesticide use on Irish farms have been published. One of these, Forristal and Grant [21], examines the effects of changes in crop rotations, along with reduced inputs of herbicides, fungicides and nitrogen fertilisers, on cereal crop yields. A second study is that of McNamara et al. [22], who compared yields of barley under standard treatment and insecticide-free conditions. A pair of studies by Burke and Dunne [23,24] examined how yields in Irish winter wheat changed when standard fungicide application programs were replaced with programs based on decision support tools, which typically resulted in less overall fungicide being applied. Earl et al. [25] examined the efficacy of ‘beetle banks’ (an IPM measure where field margins are managed to promote the presence of beetles able to serve as biological pest control agents) adjacent to cereal crops and found them to have no significant effect. Finally, Byrne et al. [26] reviewed a range of IPM techniques that may be relevant to the control of crop weeds in Ireland, but based on their findings from international studies, they concluded that a large amount of additional research was required in an Irish context, including research that specifically accounted for local abiotic factors.

While these studies provide valuable data, none of them examine scenarios directly comparable to the changes in pesticide use that would be expected if Ireland were to meet the F2F goal of reducing the use of candidates for substitution by 50%, and the situation is similar with studies from the UK. The same can be said for studies comparing organic and conventional cropping systems, as pesticide use in organic crops is reduced beyond the F2F targets and is accompanied by changes in nutrient management.

In the absence of more relevant empirical data, a structured approach to gathering expert opinion may be useful for filling some of the knowledge gaps outlined above. This paper seeks to make such estimates using a ‘Delphi’ style expert survey methodology. A Delphi survey is a technique for collecting subjective data from experts in a particular field and adding a level of robustness through iterative feedback [27]. Consequently, they can provide forecasts in situations where more objective data are not available [28]. Such surveys typically involve an initial round of data collection using a survey distributed to participants who are non-randomly selected based on their relevant expertise. Following this first round, results are compiled and fed back to the participants, who are then asked to answer the same questions as in the first round, with the opportunity to change their answers based on the feedback provided [29].

The Delphi survey technique has been used for predictive studies as far back as the 1960s (e.g., [30]), and while for much of that time it was primarily used in the social sciences, since the beginning of the 21st century, its use within the life sciences has markedly increased. This expansion in use in this field is reviewed by Mukherjee et al. [31] and illustrated by a search for the keywords ‘Delphi’ and ecolog* OR agric* OR farm* in the ‘topic’ field of the Web of Science database, which yielded 971 results in June 2024, 94% of which had been published since 2008. Most relevant to this study, expert survey methodologies have been used a number of times to estimate how changes in pest management will affect farming. Examples of this include the work of Creissen et al. [32] who used systematic surveys to investigate the uptake (though not the efficacy) of IPM practices by farmers in the UK and Ireland, and Knutson et al. [33] and Mack et al. [34] who used expert surveys to predict the impacts of changed pesticide usage practices on yields in farms in the USA and Switzerland, respectively.

The goal of this study was to use a Delphi-style expert panel methodology to estimate, in the absence of other empirical data, how meeting the F2F targets for more hazardous pesticides may impact yields in Irish farming. We also sought to use this method to gain insights into the efficacy of IPM techniques and the risk of additional pesticide resistance developing in Ireland in the context of achieving the F2F targets. We focused on cereal crops as this sector accounts for a disproportionally large amount of Ireland’s agricultural pesticide use and is thus more dependent on the availability of pesticides than other enterprises. Cereals are grown on just 11% of Ireland’s agricultural land, but approximately 60% of all agricultural pesticides applied in Ireland are used on this crop type (as measured by weight of active substances) [35]. In 2023, Ireland produced a little over 2 million tonnes of cereals, with an estimated farm-gate value of EUR 344 million, with most of this produce being used as livestock feed [36]. Here, we specifically examined wheat, barley and oats as these three crops are the dominant cereals grown in Ireland, with other crops making only a negligible contribution to total cereal production [36].

2. Materials and Methods

A Delphi-style analysis is distinguished from more basic survey methods by its iterative nature; after the initial questions are posed, participants are provided with feedback summarizing the overall response and are given the opportunity to adjust their initial answers, if they deem this appropriate, based on that feedback. The specific steps undertaken are outlined in the following sections.

All research involving human subjects was approved by the Teagasc Social Research Ethics Committee, and all participants provided their informed consent in writing.

2.1. Expert Selection and Study Rationale

Participants in the Delphi survey were selected to ensure each had applied knowledge of crop pests and the role of pesticides on cereal production under Irish growing conditions. They included applied cereal production and protection researchers (n = 4), cereal field trial managers (n = 8), cereal specialist advisors (n = 7) and cereal production and protection lecturers (n = 2). Each had a minimum of five years of experience in the field, as this was deemed the necessary time to develop the requisite understanding. We avoided potential participants who may have a bias in finding certain results, such as employees of chemical companies. Participants were contacted directly by the authors, inviting participation. Twenty-one potential participants were contacted, of which seventeen agreed to take part, with the remaining four never responding to the invitation to participate after two attempts at contact.

While Delphi surveys often seek to find consensus amongst their participants, posing questions multiple times until a pre-defined level of agreement is reached, others recognise that a lack of consensus can also be a valid outcome [37]. In this study, we considered that if such a lack of consensus were to occur, it would help identify where knowledge gaps existed and, consequently, may be a valuable finding. Thus, rather than seeking to achieve specific consensus criteria, we determined from the outset that our survey would run over only two rounds of questioning—an initial round to pose questions and a single follow-up round to allow participants to change their initial answers in response to feedback from Round 1.

Surveys were sent to participants by email, with Round 1 surveys sent in April 2024 and Round 2 surveys sent in June 2024. All 17 participants completed both rounds of the survey. Throughout the process, participants remained anonymous from one another and did not directly communicate with each other, with all feedback being provided only as aggregated data from the whole group. This was both to preserve individual privacy and to reduce the possibility of answers being biased by social pressure [28].

2.2. Survey Round 1

Participants were surveyed by means of a questionnaire, which was accompanied by instructions for its completion. All questions focused on ‘more hazardous’ pesticides, which, for the purposes of the F2F targets, are defined as chemical active ingredients listed as candidates for substitution by the European Commission. For this study, we focused on candidates for substitution that were registered for use in Ireland on wheat, barley or oats as of February 2024 (

Table 1. More hazardous chemical active ingredients used in Irish cereal crops.

Pesticide Type	Active Ingredients
Fungicides	Benzovindiflupyr, Cyprodinil, Difenoconazole, Fludioxonil, Fluopicolide, Metalaxyl, Metconazole, Paclobutrazol, Tebuconazole
Insecticides	Cypermethrin, Esfenvalerate, Gamma-Cyhalothrin, Lambda-Cyhalothrin
Herbicides	Diflufenican, Flufenacet, Metribuzin, Pendimethalin, Propyzamide, Tri-allate

). Participants were provided with a list of trade names of all products registered for use in Irish cereal crops that contain the active ingredients listed as candidates for substitution. Participants were asked about impacts on yields, the development of pesticide resistance and about alternative pest management techniques. The complete Round 1 survey document is included in Supporting Information (S1) and its contents are summarized below.

2.2.1. Yield Impacts

We asked participants about yield impacts for wheat, barley and oats, using variations on the following question:

“How would you expect the yield of [the crop] (% change in tonnes per ha) to change if farmers were required to reduce use of the listed [chemicals] by 50%, assuming [management scenario] across [timeframe]?”

Participants were also asked for their reasoning in arriving at the answer. An example of a yield impact question is shown in Figure 1.

Values for ‘the crop’ used were wheat, oats and barley. In order to reduce the time burden required of participants, questions related to fungicides and insecticides did not distinguish between these crops grown in different seasons; however, for questions related to herbicides, we asked separately about winter and spring plantings, as weed impacts can vary substantially between cropping seasons in Irish cereals [38].

Values for ‘chemicals’ were fungicides, insecticides and herbicides.

Values for ‘management scenario’ were “assuming no other changes in farmer management practices” (hereafter the ‘Standard Management Scenario’) and “assuming farmers adopted appropriate alternative management practices” (hereafter the ‘Additional IPM Scenario’). We included a note to participants in our instructions for completing the survey that, while the Standard Management scenario would not see farmers change management practices beyond reducing use of the pesticides in question, they should assume that farmers would still apply the listed pesticides strategically, i.e., at full label rates but to a reduced area, rather than applying chemicals to the same area to which they are currently applied but at a 50% rate, where such a rate is not effective. While the Standard Management Scenario may be an unrealistic oversimplification, as all farmers would likely change their pest management tactics with less access to key chemicals, these questions were included to allow participants to provide an indication of what they felt the ‘raw’ value of pesticides to these systems was, and to provide a counterpoint to the Additional IPM Scenario and quantify how effective participants believed the application of additional IPM techniques would be in mitigating the impacts of reduced pesticide use.

For questions related to fungicides and insecticides, only a single value of ‘timeframe’ was included, with participants asked to consider the impacts of reduced use of the listed chemicals over a typical 5-year period, starting the year after pesticide use was reduced (hereafter ‘short term’). Given that weeds can form persistent seed banks that allow impacts to build up over time [39], we included a second timeframe (6–10 years after pesticide use was reduced, hereafter ‘long-term’) in questions related to herbicides (this build up of propagules is less of a concern in regards to fungal and insect pests).

In all cases, participants were asked to consider these questions against a background of a normal range of weather and pest pressure variation. These permutations resulted in 36 variations on this yield impact question being posed.

2.2.2. Development of Pesticide Resistance

Participants were also asked to rate the likelihood, on a scale of 1 to 5 (1 being very unlikely, 5 being very likely), of new pesticide resistance issues developing as a result of reduced use of the listed pesticides. This rating was sought separately for fungicides, insecticides and herbicides and for both the Standard Management and Additional IPM scenarios. Participants were also asked to provide the reasoning behind their answers.

2.2.3. Integrated Pest Management

For each group of pesticides, participants were provided with a list of integrated pest management techniques that could potentially serve to mitigate the impacts of reduced pesticide use. Lists were developed based on a search of the literature and the authors’ knowledge of integrated pest management techniques currently applied, or being considered by researchers, in cropping systems throughout the world. As the substitution of hazardous chemicals with less hazardous chemicals may be a valid way to achieve the F2F goal, the option of ‘alternative chemicals’ was included for each chemical group. Participants were asked to rate on a 1–5 scale (1 being not at all important, 5 being very important) the importance that each technique would have in a situation where the use of the listed pesticides was reduced. Participants were also given the opportunity to add additional techniques and asked to name what they thought the single most important alternative management technique was.

2.2.4. General Impacts

Questions on each pesticide type concluded with a general impact question, where participants were asked the following:

“Do you anticipate any other positive or negative impacts of the outlined reduction in [this group of chemicals] use not covered in the above questions?”

This section was intended to allow participants to express any opinions they felt were important that were not captured elsewhere and, unlike other sections, was not fed back to participants for reassessment in Round 2.

2.3. Survey Round 2

Upon receipt, the results from Round 1 were collated and summarized to provide feedback to participants in Round 2. The Round 2 survey was identical to the Round 1 survey, with all questions except for the general impacts questions being repeated. It was accompanied by summaries of overall responses and reasoning offered (from Round 1), with participants asked to reassess the answers they had given based on this feedback. An example of a Round 2 survey with included feedback is shown in Supporting Information (S2) and summarised as follows.

For each yield impact question, the results of each question were presented as boxplots to illustrate the distribution of answers from Round 1. These were accompanied by summaries of reasoning provided by participants, which were divided into themes (as per Braun and Clarke [40]). Each answer provided was represented by a theme in the feedback, with the exception of generic statements (e.g., “Herbicides can help control weeds.”) or statements that simply served to illustrate a participant’s confidence (e.g., “I’ve seen results of this level in trials.”). We made no attempt to gauge the accuracy of these answers and noted this to participants in our instructions for the Round 2 survey.

For questions related to pesticide resistance, we provided the mean rating (1 to 5 score) from Round 1, along with summaries of participants’ reasons, as in the yield impact questions. For questions related to alternative management techniques, the mean rating for each technique was provided, along with a listing of how many times a particular technique was chosen as most important. Techniques added by participants in Round 1 were added to the list of techniques, and participants were asked to provide an importance rating to these, as with the techniques listed in Round 1.

The above information was sent to each participant, with that participant’s responses from Round 1 highlighted (S2).

2.4. Statistical Analysis

All statistical analysis was carried out using R software v4.4.0 [41].

We assessed differences in participant responses between Rounds 1 and 2. For each of the 36 permutations of the yield impact question, we used Wilcoxon tests to determine whether or not the collective response given by participants to that question changed significantly between rounds. We also calculated the standard deviation of the results of each question and compared the standard deviations of the complete set of questions between the two rounds using the Wilcoxon tests. Similarly, we used Wilcoxon tests to determine if the overall importance rating assigned to each IPM technique listed in Round 1 changed significantly between Rounds 1 and 2, and to determine if the risk rating for resistance development for each pesticide group changed significantly between rounds. Wilcoxon tests were used throughout this process as the data were not normally distributed.

To determine if the predicted impacts varied significantly between the Standard Management and Additional IPM scenarios, we used t-tests and Wilcoxon tests (for normally and non-normally distributed data, respectively) to assess the differences in Round 2 responses between questions regarding each pesticide type in the two scenarios. Wilcoxon tests were also used to assess differences in the predicted risk of pesticide resistance developing between the two IPM scenarios, while we assessed differences in predicted resistance risk between the three groups of pesticides within the same scenario using Kruskal–Wallis tests with post-hoc Dunns test.

3. Results

With the exception of data presented in Section 3.1 to illustrate a change in responses between Rounds 1 and 2, and Section 3.5 (general impacts), all results presented below are based on answers provided by participants in Round 2.

3.1. Change in Predictions Between Rounds

Between rounds, there was a trend towards greater agreement amongst participants. The standard deviation of responses for each of the 36 yield impact questions decreased between Rounds 1 and 2, with the overall result being a statistically significant reduction in variation (W = 987, p < 0.001). However, this change was mostly the result of participants who provided extreme values for answers in Round 1 providing more moderate answers in Round 2. The median value changed between rounds for just six questions, and in all 36 cases the difference in answers between rounds was not statistically significant (all p > 0.05, full results shown in Supporting Information S3).

A similar response was observed with the rating of various IPM techniques, with the average change in the absolute value of the ratings across all techniques being just 0.08 points (on a 1 to 5 scale) between Rounds 1 and 2. The change in individual technique ratings was not statistically significant in any case (all p > 0.05, full results shown in Supporting Information S4). In only one case did a participant change the technique that they considered the most important between rounds. The situation was also similar in the case of questions regarding pesticide resistance development, where the predicted impact did not change significantly between rounds in any case (all p > 0.05, full results shown in Supporting Information S5).

3.2. Predicted Yield Impacts

3.2.1. Fungicides

The median impacts predicted on wheat, barley and oat yields over a five-year period as a result of reduced access to the listed fungicides were 15, 10 and 15%, respectively, in the Standard Management Scenario (Figure 2). The most commonly listed reason for the predicted impacts was Rust (fungal pathogens of the order Pucciniales), with five and six participants noting this would have a substantial impact on wheat and oats, respectively, where access to the listed chemicals was reduced. Two participants mentioned Rust in their responses relating to barley, but noted that it is less of a problem in this crop. Four participants also suggested that difficulties controlling fungal pathogens of the genus Septoria would contribute to the impacts on wheat.

A number of participants (five, ten and one for wheat, barley and oats, respectively) suggested that impacts would be somewhat moderated by the fact that chemicals other than the candidates for substitution could adequately control the major fungal pathogens on the listed crops. However, these participants still suggested median impacts of 15, 10 and 5% yield loss in the Standard Management Scenario for wheat, barley and oats, respectively. Three, two and two participants noted in their respective responses for wheat, barley and oats that the Irish climate makes cereal cropping especially vulnerable to fungal diseases compared to other countries, and that weather conditions were the biggest factor affecting levels of disease impact in any particular year.

The summary of the full list of themes provided for yield impacts for all crops is shown in Supporting Information S6.

When the predicted median yield impacts were applied to yield data of wheat, barley and oats averaged across the years 2019–2023, resulting yield losses as shown in

Table 2. Yield losses resulting from predicted yield reductions.

	Fungicides, Standard Management	Insecticide, Standard Management	Fungicide, Additional IPM	Insecticide, Additional IPM
	Yield Loss (t/Ha)	Yield Loss (t/Ha)	Yield Loss (t/Ha)	Yield Loss (t/Ha)
Wheat	1.20	0.80	0.80	0.50
Barley	0.77	1.54	0.46	0.77
Oats	0.99	0.66	0.66	0.33
	Herbicide, Short Term, Standard Management	Herbicide, Long Term, Standard Management	Herbicide, Short Term, Additional IPM	Herbicide, Long Term, Additional IPM
	Yield Loss (t/Ha)	Yield Loss (t/Ha)	Yield Loss (t/Ha)	Yield Loss (t/Ha)
Winter Wheat	1.45	2.90	0.97	1.93
Winter Barley	1.30	2.59	0.86	1.73
Spring Barley	0.34	0.37	0.34	0.37
Winter Oats	0.68	1.13	0.47	0.76
Spring Oats	0.28	0.41	0.20	0.28

were predicted.

Yield loss figures were obtained by multiplying median impacts (Figure 2, Figure 3 and Figure 4) by average weighted yields over five years (2019–2023) [42]. No figures are provided for Spring Wheat, as the sample size used in Dillon et al. [42] is too small to accurately assess yields in this crop.

3.2.2. Insecticides

The predicted impacts resulting from reduced access to the listed insecticides were median yield reductions of 10, 20 and 10%, respectively, for wheat, barley and oats in the Standard Management Scenario across a five-year period (Figure 3). All participants who provided a reason for the predicted impacts identified the pathogen Barley Yellow Dwarf Virus (BYDV) as being responsible, noting that the listed chemicals are important for the control of aphids, which serve as vectors of this yield-limiting virus. The larger predicted impact on barley, compared to wheat and oats, was suggested to be a result of that crop’s greater vulnerability to this disease. Even participants who thought impacts would be low provided their reasoning in terms of BYDV (suggesting that it can still be adequately controlled with less chemical application, or noting the low susceptibility of wheat and oats), highlighting the importance of this virus as a factor affecting Irish cereal crops. Three participants also suggested that anthropogenic climate change, especially milder winters, would increase the risk of BYDV impacts in wheat, while one suggested this may result in new insect pests becoming a problem in Ireland. No insect pests other than aphids were suggested to contribute to the predicted impacts.

3.2.3. Herbicides

While the predicted yield impacts of reduced access to the listed herbicides in spring crops and all crops in the short-term (five years) were similar to those for fungicides and insecticides, predicted impacts were substantially higher in winter wheat and winter barley over the longer term (6–10 years), with median predicted yield reductions of 30% for both crops in the Standard Management scenario (Figure 4). Additionally, two participants suggested that for both wheat and barley, winter cropping may become entirely non-viable in the long-term due to weed impacts (these predictions are not included in Figure 4 as such predictions imply that farmers will not plant these crops and would put the land to other uses). The larger interquartile ranges shown for long-term yield impacts on winter crops in Figure 4 also highlight that there was a greater level of disagreement amongst participants in this situation compared to either fungicides or insecticides.

Participants noted that grass weeds would be the main cause of yield loss if access to the selected herbicides were reduced, with six participants highlighting this for wheat, three for barley and four for oats. Many participants cited that the listed products include the only herbicides not prone to rapid resistance development and suggested that weeds would quickly develop resistance to the remaining herbicides without access to alternative modes of action (this reason was given in relation to wheat by seven participants and in relation to barley by four).

The lower predicted impact on oats, as compared to wheat and barley, was attributed by seven participants to the fact that oats are more competitive against weeds than these other crops and that the listed chemicals are little used in oats. However, three participants noted that, while the listed chemicals may rarely be used on oats themselves, they are often used on other crops grown in rotation with oats, and reduced access may contribute to impacts on oat crops grown at other times in rotation with wheat or barley.

3.3. Integrated Pest Management

3.3.1. IPM Techniques

The mean value of the importance rating provided by participants for each suggested IPM technique is presented in rank order (highest first) in

Table 3. Mean Importance Ratings of Integrated Pest Management Techniques.

Fungicides		Insecticides		Herbicides
Technique	Mean Importance	Technique	Mean Importance	Technique	Mean Importance
Resistant/tolerant cultivars	4.65	Resistant/tolerant cultivars	4.82	Changes in crop rotation	4.67
Changes in sowing dates	3.18	Changes in sowing dates	4.32	Decision support services/monitoring	4.13
Changes in crop rotation	3.12	Decision support services/monitoring	3.56	Changes in sowing dates	4.03
Seed health (enhanced testing for seed-borne diseases)	2.78	Reducing green bridge	3.4	Change in cropping season (winter vs. spring)	3.94
Decision support services/monitoring	2.65	Economic threshold-based control	3.04	Improved hygiene practices to prevent weed entry	3.75
Improved soil health and residue management	2.37	Changes in crop rotation	2.85	Stale seed bed	3.6
Cultivar mixes	2.31	Avoiding susceptible crops, e.g., winter barley	2.4	Stubble cultivation	3.4
Nutrient management	2.26	Field margins (natural enemy conservation)	2.12	Changes in crop establishment system	3.38
Alternative chemicals, including biologicals	1.68	Change in establishment practices	2	Harvest weed seed control	3.21
Heat/steam treatment of seeds	1.64	Strategic stubble cultivation	1.93	Pre-sowing glyphosate	3.17
Change in establishment practices	1.62	Introducing biocontrol agents	1.53	Resistance testing of weeds	2.87
Spot spraying	1.41	Alternative chemicals, including biologicals	1.19	Mechanical weeding	2.69
		Physical barriers (e.g., netting)	1.12	Seeding rate increase	2.43
		Spot spraying	1.12	Companion crops	2.2
		Improved hygiene practices to prevent insect entry to farms	1.03	Cover crops	2.14
				Nutrient management	1.63
				Thermal weeding	1.44
				Alternative chemicals, including biologicals (please list)	1.2
				Irrigation management	1

, while the technique that each participant considered most important is shown in Figure 5. Both illustrate that the majority of participants considered varietal tolerance to fungal pathogens and BYDV to be the most effective technique to mitigate the impacts of reduced access to fungicides and insecticides, respectively.

In the case of managing impacts of reduced herbicide use, a wider range of IPM techniques received an average importance rating above three (ten techniques, compared to five for insecticides and three for fungicides). Three of the four highest-ranked techniques (changes in rotation, sowing date and cropping season) are all related to system-wide measures to avoid the build-up of weed impacts or to remove crops from direct competition with weeds.

Decision support services/monitoring (DSS), which could potentially allow farmers to make strategic choices about pesticide application and thus manage pests with lower quantities of chemicals, was in the top three most important techniques for both insecticides and herbicides.

3.3.2. Effect of IPM on Yield

In responses regarding the Additional IPM scenario for fungicides, a large number of participants (ten, eight and five for wheat, barley and oats, respectively) expressed doubt in the ability of IPM techniques to mitigate the issues noted in the Standard Management Scenario. Statements by these participants suggested that farmers typically already employ IPM techniques and that additional techniques are not employed because they are non-viable. This is consistent with the fact that, for all three crops, predicted yield impacts as a result of reduced fungicide use were lower in the Additional IPM Scenario than the Standard Management Scenario, but this difference was not statistically significant (Figure 2).

With regards to insecticides, yield reductions were predicted to be significantly lower for all three crops in the Additional IPM Scenario compared to the Standard Management Scenario (Figure 3). However, despite this, some participants expressed the same doubt regarding IPM viability as noted above for fungicides (this view was expressed by five, four and four participants, respectively, for wheat, barley and oats). Amongst participants who expressed greater optimism for the efficacy of IPM measures to substitute for insecticide use, many suggested this was due to the current or near-term availability of BYDV-resistant or tolerant crop cultivars (with six, eight and three participants expressing this view in relation to wheat, barley and oats, respectively). However, some participants expressed caution, noting that these cultivars can be lower yielding than standard cultivars and that with less access to key insecticides, there would be greater evolutionary pressure for BYDV to overcome varietal tolerance.

Unlike in the case of fungicides and insecticides, for herbicides, the effects of the Additional IPM Scenario were inconsistent in terms of their statistical significance compared to the Standard Management Scenario (Figure 4). When considering the Additional IPM Scenario, some participants (five and three relating to wheat and barley, respectively) noted that changes in cropping season or sowing date may mitigate impacts somewhat but could potentially cause their own reductions in yields.

3.4. Risk of Resistance Development

Participants considered the risk of new pesticide resistance issues developing as a result of reduced access to the listed chemicals to be relatively high, with average ratings of likelihood (out of a maximum value of 5) being shown in

Table 4. Mean predicted risk of new pesticide resistance developing. Risk was rated by participants on a 1–5 scale, with 1 meaning “very low risk” and 5 meaning “very high risk”. Values in columns that do not share a superscripted letter are significantly different from one another (Dunns test, α = 0.05).

	Standard Management	Additional IPM	Test Statistics (Standard Management vs. Additional IPM Scenarios)
Fungicide	4.09 ab	3.65 a	W = 185.5, p = 0.140
Insecticide	3.59 a	2.59 b	W = 188, p = 0.058
Herbicide	4.56 b	4.03 a	W = 209, p = 0.017
Test Statistics (Difference between pesticide types)	χ2 = 7.39, p = 0.025	χ2 = 9.68, p = 0.007

Test results shown in bold are statistically significant.

. In both the Standard Management and Additional IPM scenarios, the risk was significantly higher in herbicides than in insecticides, while in the Additional IPM Scenario, the predicted risk in fungicides was also significantly greater than in insecticides (risk for fungicides and herbicides was not significantly different in either scenario).

The range of reasons provided by participants was relatively narrow, with the majority of participants expressing the view that having fewer chemical options to rotate between would encourage the development of pesticide resistance to the chemicals, which remained readily available (this reason was mentioned by 12, 10 and 12 participants in relation to fungicides, insecticides and herbicides, respectively). A somewhat contrary reason was provided regarding insecticides by three respondents, who noted that there are few other chemicals available for the control of aphids in Irish cereal crops, and thus, with less access to the listed chemicals, there would be little for aphids to develop resistance to. These three participants all provided below-average scores in answer to this question (3, 1 and 1).

Average ratings were lower for all three chemical groups in the Additional IPM Scenario, compared to the Standard Management Scenario, but this decrease was only significant in the case of herbicides.

3.5. General Impacts

Participants mostly used the ‘general impacts’ section of the survey to reiterate statements made when providing reasons to accompany the questions discussed in the previous sections, often highlighting the importance of the specific chemicals considered compared to those more generally available to Irish cereal farmers. Novel ideas expressed in this section included the positive impact reduced pesticide use may have on beneficial insects or ecosystem health more generally (expressed by five participants in relation to insecticides and one in relation to fungicides). Two separate participants expressed concern about the impacts on crop quality that may result from reduced fungicide and herbicide use, in addition to the yield quantity impacts that were the focus of the rest of the survey. Two participants also suggested that reducing allowable pesticide usage may impact the commercial decision-making of pesticide manufacturers and suppliers, making them less likely to invest in the development and registration of new pesticides.

4. Discussion

4.1. Reliability of Findings

The predictions of our expert panel are that the impacts on yields of Irish cereal crops as a result of reduced access to the pesticide candidates for substitution would be substantial. Given the large impacts that such predicted yield changes would have on farm incomes and food security, it is important to critically review the reliability and accuracy of these predictions.

The sample size of a study may be seen as an indicator of reliability, and our panel of 17 experts puts this study’s sample size in line with many similar surveys. A review by Mukherjee et al. [31] of 31 Delphi-style studies in the ecology and life sciences field found that 19 used fewer than 20 expert participants. The published study most comparable to this work, that we are aware of, is that of Mack et al. [34], who used a Delphi-style methodology to predict the impact on Swiss cereal cropping of moving from a fungicide and insecticide-free ‘extenso’ farming system to one in which herbicides are also excluded. That study’s use of a panel of 18 experts illustrates that there is precedent for asking questions of this nature with a sample size similar to ours.

The accuracy of a future prediction is harder to judge. While some studies have used Delphi-style surveys to predict future events with high levels of accuracy (e.g., [43]), the performance of this technique in specific cases does not necessarily vouch for its effectiveness in others. One aspect of the data reported here that may, however, allow us to have some confidence in its accuracy is the level of consensus amongst the participants. While reaching a consensus is often a goal of a Delphi study, this was not the case here. We made no attempt to drive participants towards a particular level of consensus, and by having participants remain anonymous from one another, we reduced the likelihood of participants agreeing based on social pressure [44]. Yet despite this, a relatively high level of consensus is in fact apparent in the data.

Beiderbeck et al. [45] suggested that a threshold for defining when consensus has been achieved in a Delphi-style study is when the interquartile range (IQR) of a result is less than a quarter of the maximum range of the scale on which the result is measured. By this metric, an IQR of less than 25 percentage points would be considered a consensus finding in our yield impact questions. This was achieved in all but one of these 36 yield questions (the exception being the long-term impact of herbicide use reduction on winter barley in the Standard Management Scenario, which had an IQR of 30 percentage points—Figure 4). Using this same threshold, questions answered on a 1–5 scale would be considered to have a consensus response with an IQR of 1.25 or less, and this was the case for 35 out of the 45 IPM techniques participants were asked to rate and for four of the six questions related to the development of pesticide resistance (S4 and S5). This measure of consensus shows that our 17 expert participants were effectively in agreement about the magnitude of impacts of the proposed pesticide use reductions in the majority of situations.

4.2. Comparison with Other Studies

While the predicted yield losses shown in Table 2 are substantial, it is also worth noting that, while we asked participants to predict the impact of reduced use of the three groups of pesticides separately, the ambitions of the EU’s F2F Strategy do not distinguish between pesticide types and any policy to bring about a reduction in use would likely affect all three groups together. While the interactions between different types of crop pests can be complex and their impacts are not necessarily directly additive (e.g., [46,47]), it is reasonable to assume that the cumulative impact on a given cropping scenario of reductions in the use of fungicides, insecticides and herbicides combined would likely be greater than those predicted for any one pesticide group individually. However, as organic farming systems in Ireland, which use no synthetic pesticides, may typically see a 45% yield gap compared to conventional systems [16], it is reasonable to assume this is an upper limit on yield impacts that would be experienced as a result of meeting the F2F targets.

Many of the predicted impacts of reduced use of individual chemical groups are close to the overall impacts previously predicted by other examinations of the F2F targets, meaning that regardless of the method used to calculate combined impacts, the estimates of our expert panel will likely be larger than those estimated by previous studies. Bremmer et al. [15] used expert surveys to predict the impacts of the pesticide use reduction goals on a range of crops by 2030 under a scenario roughly comparable to our Additional IPM Scenario. They predicted that wheat yields across seasons in Finland, Germany, France and Romania would be on average 6% lower as a result of the targets. Barrieo-Hurle et al. [16] modeled that cereal yields across the EU would be reduced by 11% as a result of a number of the F2F targets, including pesticide goals. Beckman et al. [14] modeled the impact of a number of actions included in F2F and other EU environmental proposals and predicted that, across all EU agriculture, yields would be reduced by 12% (assuming alternative management generally equivalent to our Additional IPM Scenario).

While one interpretation of the difference between our participants’ predictions and the work of Bremmer et al. [15], Barrieo-Hurle et al. [16] and Beckman et al. [14] could be that our participants overestimated the impacts of reduced pesticide use, another interpretation is that Ireland will experience greater impacts than the EU as a whole. This is supported by data reported by Barrieo-Hurle et al. [16] that Ireland sees a larger gap between conventional and organic yields of wheat and other cereals than any other EU nation, highlighting the greater importance of synthetic pesticides (and fertilizers) in maintaining yields under Irish conditions. The similarity of our predictions with the results of McNamara et al. [22] lends some support to this latter interpretation, with their study finding in field trials of winter barley that plots treated with no insecticide had yields 15% less than treated plots (compared to a 10% median predicted loss in this study for reduced use of the most hazardous insecticides).

However, other Irish studies push against this interpretation, with Burke and Dunne [23,24] finding much smaller yield impacts than those predicted here with total fungicide use reductions of up to 54% and Forristal and Grant [21] finding similar impacts to those predicted here, resulting from a greater reduction in pesticide use, accompanied also by a reduction in fertilizer application. Furthermore, in a study from the UK, the only other nation with a similar climate to Ireland, Webster [48] predicted that reducing pesticide use by 75% would only result in a 10% reduction in wheat yield. However, in each of these cases, the mix of pesticides examined was not the same as those considered in this study, and many of the responses provided by our participants highlighted how these chemicals play a particularly crucial role. Indeed, the classification as a ‘candidate for substitution’ acknowledges that a chemical is hazardous, but it remains registered in part because it has an important role to play that cannot be easily replaced by other currently available substances [49]. Thus, it may be reasonable to assume that reduced use of the chemicals considered here would have a disproportionate impact compared to reduced use of pesticides more generally.

This is consistent with the poor ranking that participants provided the IPM option of ‘Alternative chemicals, including biologicals’, which was listed as a potential IPM strategy for all three chemical groups, and in all cases, the respondents provided it a mean importance rating of less than two (Table 3). As well as being reflective of the limited alternative options currently available to replace the candidates for substitution, this is also consistent with research illustrating that the discovery and production of new pesticide active ingredients is becoming increasingly slow and expensive [50]. Two participants also made an observation that, if correct, would exacerbate this issue, noting in their general responses that policies reducing pesticide use may discourage chemical companies from developing or registering new products, a view also expressed by Byrne et al. [26]. Additionally, a large body of research has examined the pesticidal properties of naturally occurring substances, such as plant extracts and microbial metabolites, and while such studies often show promising results in lab trials, efficacy is often far lower in more realistic field tests (e.g., [51]) or requires impractically large volumes of active ingredients (e.g., 10s–100s of kg per Ha) to achieve (e.g., [52]).

4.3. Efficacy of Integrated Pest Management

Looking beyond chemical control measures, participants also expressed low confidence in the potential for IPM to prevent the predicted yield losses, with the Additional IPM Scenario having high yield losses that were, in many cases, not significantly different from those predicted in the Standard Management Scenario (Figure 2, Figure 3 and Figure 4).

This prediction has mixed support within the literature. Some studies have reported integrated pest management techniques to be highly effective at mitigating the impacts of reduced pesticide use. An example of this is the work of Jørgensen et al. [53], who examined the impacts on wheat crops of Rust infections under a range of IPM scenarios across nine European countries (not including Ireland). They found that, through utilizing resistant crop cultivars and DSS (both IPM techniques that were considered important by participants in this study), yields were maintained close to baseline levels with 75% less pesticide usage (including reduced usage of chemicals that are candidates for substitution).

However, as is the case with many IPM studies, the work outlined by Jørgensen et al. [53] was carried out by researchers on research farms, not on commercial farms. It is possible that methods used by researchers to achieve a result in the course of a scientific study may not be economically or logistically viable when applied to commercial farms due to issues such as higher labor requirements [54] or IPM implementation being highly knowledge-intensive [55]—the ‘socio-technical’ barrier to implementation noted by Wyckhuys et al. [56]. An understanding of this issue may have influenced the ratings given by our participants. A bibliometric analysis by Zhou et al. [57] of 870 papers on IPM published between 1993 and 2022 found just 3.6% of them focused on economics, highlighting that the issue of economic viability of IPM techniques is currently a major gap in the literature.

Other studies suggest that the impacts of reduced pesticide use may be somewhat mitigated by more passive measures. The work of Schneider et al. [58], for example, criticised Barreiro-Hurle et al. [16], Beckman et al. [14] and Bremmer et al. [14] for failing to incorporate ‘ecological feedbacks’ in their analyses, citing the idea that improved provision of ecosystem services, such as pollination and biological pest control, may help offset the impacts of reduced pesticide use. While such a criticism may be valid in certain crops and ecosystems, it is likely to be of limited relevance to Irish cereals in the short term. Cereals are not reliant on animal pollinators, and Ireland’s biodiversity-poor environment means biological control processes that may be valuable in other contexts are of limited effectiveness [25], a fact that is reiterated in our data by the low ratings participants provided to biological control measures as an IPM technique (Table 2). However, it is possible that over the long term, reduced pesticide use may help towards improving Ireland’s biodiversity and thus allow for some greater provision of this ecosystem service.

A situation more closely resembling our predictions is seen in a review by Lázaro et al. [59]. This work found that, across 20 published studies, when reducing fungicide use, disease incidence increased more slowly in fields that were treated according to DSSs than in reference fields, but on the whole still increased. This illustrates that, while IPM techniques can somewhat mitigate the yield impacts of reduced pesticide use, the available techniques are often not a panacea that can entirely prevent yield losses in most situations.

Participants in this study predicted that the risk of additional pesticide resistance developing as a result of reduced access to the listed chemicals would be relatively high across all three chemical groups. As with yield impacts, the reduction in risk as a result of the application of IPM techniques was predicted to be only relatively minor. The assertion made frequently by participants that resistance risk would be higher with a smaller range of chemicals to rotate between is well supported in the literature for all three chemical groups (e.g., [60]).

However, also supported is the idea, expressed by just two participants in relation to fungicides and one in relation to herbicides (S6), that resistance can be effectively managed with strategic use of lower doses of pesticides [18]. Furthermore, with regard to fungal pathogens, resistant/tolerant cultivars can be used synergistically with lower fungicide doses to prevent the development of pathogens of both pesticide resistance and virulence to overcome varietal tolerance [61,62].

The negative impacts of pesticide use are being increasingly studied, with findings showing the role they have played in large global insect biomass decreases [4] and human mortality [7]. Using a hazard-based approach to managing pesticide impacts, the chemicals listed in Table 1 have been deemed to be of concern by EU and Irish policy makers because of their potential to contribute to problems such as these. However, in the opinion of the expert panel consulted in this study, substantial reductions in yield will result from reduced use of these more hazardous pesticides, impacts that can only be partially mitigated by the use of IPM techniques.

The insights provided by our expert panel also highlight ideas regarding potential paths forward in a future where access to more hazardous pesticides is reduced. One is the importance of crop varieties resistant to fungal pathogens and BYDV, considered the most important IPM measure for managing impacts from reduced usage of both fungicides and insecticides (Table 2). Increased research and development in producing such varieties may therefore be able to improve the yields that can be expected in cropping situations where pesticide use is reduced. While our participants were pessimistic about yields under reduced pesticide scenarios, even with the application of IPM, their estimations were based on existing solutions, and in a future where available solutions have expanded, the yields that can be achieved with lower pesticide application may improve.

DSSs were also considered an important factor in protecting yields with reduced pesticides. DSSs can take a range of forms, including in-person advisory services, written materials and devices and software utilizing artificial intelligence to make recommendations or autonomously perform tasks [55]. All such services are only as good as the existing knowledge and technical resources on which they are based. Thus, it may be realistic to imagine that, as with resistant varieties, in a future scenario where these resources are subject to increased investment, they may also result in farm yields with reduced pesticide use exceeding the yields predicted by our expert panelists. Such services could also potentially assist farmers in planning and implementing crop rotations optimized for weed control, and thus also allow the performance of this IPM technique to exceed the expectations of our expert panel.

4.4. Limitations

While this research is intended to contribute to filling the knowledge gap that exists in relation to the Irish-specific impacts of reduced pesticide use, as an expert panel-based prediction, it does so only partly. Field trials will be required to allow policy makers to be truly confident in the accuracy of the predictions made here; however, such trials would take several years to complete and be highly resource intensive. While in the interim the data presented here serves as a valuable first step, a coordinated research effort to gauge both the temporal and spatial impact of the predictions at a field level needs to be prioritised by the research community. It should also be reiterated that the Standard Management Scenario is a somewhat artificial scenario that likely would not eventuate under real-world circumstances, and while the Additional IPM Scenario should be considered more likely, IPM is not without costs, and it is necessary to acknowledge that costs will be incurred in producing yields more in line with this scenario.

Furthermore, this work does not include a cost-benefit analysis of the pesticide reduction targets contained within the Farm to Fork strategy. The starting point of this paper is that, as the EU has adopted these targets on the basis that they will provide environmental and/or health benefits, quantification of these and weighting them against the yield losses found here is outside the scope of this work, but would be a valuable subject for future research.

In attempting to provide answers to questions surrounding the potential impacts of a reduction in the use of the more hazardous pesticides on Irish cereal yields, this research highlights that a number of other important questions remain unanswered. These include questions regarding the overall economics of reduced pesticide use and of IPM techniques that may be used in these reduced pesticide situations. The economics of IPM is a neglected area of research [57], and studies examining the costs and benefits of IPM, as implemented by farmers on commercial farms, as opposed to by scientists on research farms, will be needed to determine if the viability concerns expressed by participants in this study are valid or not. Such data will also allow for an understanding of which IPM techniques are currently viable, which may be viable with additional research or assistance for farmers and which do not have a realistic prospect of widespread implementation.

5. Conclusions

This research sought to estimate the impacts of achieving the pesticide use reduction targets in the Farm to Fork strategy on Irish cereal cropping. The estimated impacts are greater than those that have been predicted in other Europe-wide studies, reinforcing the point that Ireland’s unique climate and landscapes mean that data from other countries cannot simply be applied to the Irish context. While this novel dataset is derived from expert opinion, rather than empirical research, the use of the Delphi method adds an element of robustness to these predictions, and the level of consensus seen in participant responses underlines a degree of confidence in their likely accuracy. In the absence of more comprehensive data from field trials or other primary studies, it now falls to policy makers to judge whether the possible environmental benefits of measures to reduce pesticide use are worth the trade-offs in terms of yields and, if so, what supports can be provided to mitigate these impacts and to ensure that cereal farming remains a viable enterprise in Ireland.