Following this years’ SETAC Europe meeting in Maastrict, our Senior Ecotoxicologist, Gareth Le Page, reflects on some of the key themes and discussions from across the conference. Drawing on conversations, presentations, and his wider observations from the week, he shares his perspective on the direction of travel for environmental risk assessment and its role in regulatory decision-making.

 

It’s a month on from SETAC Europe in Maastricht and I’ve had a chance to reflect and compile my thoughts. Along with the normal increase in caffeine, long days of trying to hold too much new information and longer evenings drinking with friends, there were some key trends that popped out at me, and one revelation that came to me only on the last night.

The revelation…. Bitterballen. Why have I only just heard about these? Why are they not mandatory in all late-night bars?

The trends and other SETAC observations, I’ll try and explain below (clearly, everyone leaves SETAC having seen different talks and spoken to different people, so these are just my own reflections rather than an attempt at a meeting summary).

While many post-conference summaries have rightly highlighted advances in NAMs, and emerging technologies (including use of AI tools), a broader shift appears to be underway.

One of my strongest impressions from SETAC over the years, but particularly from Maastricht, is that environmental risk assessment is becoming increasingly mechanistic, data-driven and based on the integration of multiple evidence streams rather than just relying on individual studies or summary endpoints. Perhaps not in and of itself a new thing, but across ecotoxicology, environmental fate, exposure assessment and monitoring, researchers are increasingly focused on extracting more insight from large datasets and using more sophisticated models to predict environmental outcomes. Or, for example, utilizing the full time series data from ecotox studies to derive further insight.

However, perhaps the most important discussions were not about the science itself (the approaches and tools seem to be well developed and their scientific basis well supported). It was about confidence and the ‘how/when/where’ we can incorporate them into our regulatory efforts.  As our scientific approaches become more sophisticated, the challenge is no longer simply just generating data. Increasingly, the challenge is determining when new approaches are sufficiently robust, transparent and reliable to support regulatory decisions. Furthermore, how can these data-heavy and complex approaches be aligned with the pragmatic decision frameworks necessary for a risk assessment regulatory framework that is not only accurate, but also practical. I suspect this tension between the increasing ability to describe biological and ecological complexity and the requirement to provide simplified, practical, transparent and reproducible regulatory decisions will become one of the key challenges over the coming years.

For organisations operating at the interface between science and regulation like us here at Enviresearch, this distinction may become one of the defining challenges of the next decade.

Effect Modelling is Becoming Part of Mainstream Ecotoxicology

One of the most striking observations from Maastricht was the widespread use of modelling approaches throughout the whole program. These are no longer confined to specialist modelling sessions but plastered across all ecotoxicologically relevant sessions, with nearly all sessions I attended containing at least one presentation using effects models. Indeed, it seems that it is increasingly unusual not to encounter some form of mechanistic modelling approach.

Perhaps one of the clearest signals of this was the launch of the new book – ‘Mechanistic Effect Models in the Environmental Risk Assessment of Chemicals’, developed through the SETAC MAD (Models for Assessment of Chemicals) Working Group. It provides guidance to move mechanistic effect models from academic development towards practical environmental risk assessment. While these approaches have been discussed within SETAC for many years, their visibility throughout the conference suggests they are increasingly becoming part of the regulatory scientist’s toolbox and there is ever less justification for regulators to dismiss these approaches.

The question is no longer whether mechanistic models have value, but how they should be used and under what circumstances their use leads to better risk assessment decisions.

Bigger Datasets are Driving New Scientific Insights

A second recurring theme was the increasing use of large datasets to answer environmental questions that would previously have been addressed through individual studies.

Across the conference, researchers drew upon environmental monitoring programmes, biomonitoring datasets, regulatory study databases, long-term field datasets, systematic reviews and meta-analyses to identify patterns and relationships that would be difficult or impossible to detect from individual studies alone. These datasets were being combined with statistical, probabilistic and predictive approaches to support environmental decision-making.

This trend was reflected in work presented by us at Enviresearch. Our poster examining zinc ecotoxicity data explored how large collections of existing chemical and biomonitoring data can be analysed to improve understanding of the performance of a bioavailable Zinc EQS. Similarly, our work with Natural England demonstrated how existing datasets can be integrated and re-analysed to build a targeted surveillance program that supports evidence-based decision-making at site level as well as inform national policy conversations.

Additionally, although not quite the same as using ‘large’ datasets (and conveniently linking to the effect modelling observations above), we are now revisiting smaller single study datasets and asking what additional insight can be extracted from the complete time series information available. Historically, regulatory ecotoxicology has often reduced these datasets into a small number of summary endpoints.

Again, the scientific challenge is increasingly shifting from simply generating data to extracting meaningful insight from the substantial datasets that already exist.

Scientific Acceptance, Regulatory Acceptance and Regulatory Implementation

One of the strongest themes emerging from Maastricht may not have been explicitly discussed at all.

Many of the approaches presented at SETAC now enjoy broad scientific acceptance. The scientific community increasingly recognises the value of mechanistic effect models, NAMs, advanced statistical approaches and predictive tools. However, this scientific acceptance is not the same as regulatory acceptance. Regulatory acceptance requires confidence that methods are reproducible, transparent, reliable and fit for purpose within decision-making frameworks. Moreover, even where regulatory acceptance exists in

principle (e.g. TKTD models are explicitly discussed in the EFSA birds and mammals, aquatic and bee guidance documents), a further challenge remains: implementation.

Implementation addresses the practical questions faced by applicants and regulators. How should a new method be applied? What evidence is required? How will different authorities interpret the results? What constitutes sufficient confidence for decision-making? And where does the resource come from to support regulators in evaluating these new tools to help with their evaluation/implementation (a frustration both to the applicants and the authorities who need to deliver the evaluation)? There is a growing bank of supporting documents for the former questions – but the latter seems to be a bigger problem.

I used to talk about acceptance and implementation as though they are synonymous. They are not. Regulatory acceptance means “this can be used” whilst implementation means “this can be routinely applied”.

And for me, this distinction between scientific acceptance, regulatory acceptance and regulatory implementation can be seen to emerge repeatedly when reflecting on the conference as a whole. At the moment, we seem to be straddling over regulatory acceptance and implementation with different tools at different stages; for example, TKTD models are much closer to implementation than population

modelling that sits more within regulatory acceptance and some AI assisted approaches well within the scientific acceptance stage.

An Interesting Absence: Where are the Implementation Discussions?

Another observation from Maastricht was less to do with what was discussed, but what was discussed relatively little. Perhaps I simply attended the wrong sessions, but one thing that surprised me was that despite my expectations of considerable discussions on the potential implementation and implications of the revised EFSA Birds and Mammals Guidance and the forthcoming Bees Guidance, there appeared to be comparatively little focused on practical implementation and impacts.

Does this reflect the fact that understanding of the implications is still in its early stages? Or does it reflect the large areas of contention, uncertainty and a ‘let’s wait and see’ attitude?

Other Notable Themes from Maastricht

  • Terrestrial ecotoxicology received increased attention (likely driven by the forthcoming guidance update), including recovery processes, terrestrial exposure pathways and ecosystem protection goals. Indeed, our poster highlighted the continuing challenge of developing soil risk assessment approaches that are both scientifically robust and practical to implement.
  • Precision application technologies are beginning to reshape discussions around exposure assessment and mitigation. Our poster explored how existing risk assessment approaches may need to evolve to better account for increasingly targeted application techniques and their potential impact on environmental exposure.
  • Although ECHA’s CLP guidance outlines non-EATS modalities as part of the criteria for endocrine disruption hazard classification under EU CLP, there was only a limited discussion of these modalities. However, we presented a poster on this and remain on the page that we need to be considering these mechanisms and the regulatory implications now they are included in EU legislation.
  • Mixture assessment remains a major area of development, particularly probabilistic approaches and realistic combined exposure scenarios.
  • PMT/vPvM and mobility assessment continue to gain momentum, with increasing focus on monitoring data and groundwater relevance.
  • The boundary between monitoring and risk assessment is becoming increasingly blurred as monitoring data are used to support predictive assessment and regulatory decision-making.

A New Phase for Regulatory Science?

I wonder if these observations suggest that environmental risk assessment is entering a new phase? It could be that the future is unlikely to be defined by a single technology, model or regulatory process/framework. Instead, could it be characterised by the integration of diverse evidence streams, advanced modelling approaches and large-scale datasets?

For us regulatory scientists, success will depend not only on understanding these new tools but also on demonstrating how they can be applied transparently, consistently and meaningfully within regulatory frameworks.

Conclusion

SETAC Europe 2026 showed we are increasingly focused on integration and interpretation. Large datasets and advanced modelling approaches are transforming how environmental questions are addressed. We have a much richer evidence base to support our predictions and the challenge is convincing regulators that these new ways of generating predictions are sufficiently robust to use in regulatory decision-making.

However, innovation alone will not drive this regulatory change. Progress will ultimately depend on our ability to build confidence in the conclusions drawn from these approaches and to translate scientific advances into robust, transparent and decision-relevant regulatory outcomes. We need to work out what we need to do in order to increase confidence and move from acceptance to implementation.

Looking back, what strikes me most is that many of these trends are not independent developments. Large, new and revisited datasets are feeding increasingly sophisticated models, which in turn are being used to generate predictions that regulators are now being asked to evaluate and implement. With the scientific basis becoming better developed, the next challenge is ensuring our regulatory frameworks and implementation pathways evolve at a similar pace.