红桃视频

Northeast Arctic cod (Gadus morhua) in the Barents Sea and blue whiting (Micromesistius poutassou) in the Norwegian Sea are two marine fish populations of major ecological and economic importance. Both populations, and the fisheries that depend on them, have been monitored for decades. The processes that shape their dynamics have also been the subject of sustained scientific research. The resulting time series and causal understanding provide a strong basis for forecasting population dynamics over multi-year horizons, with potential value for fisheries management.

Recent studies have proposed models to forecast blue whiting recruitment and cod biomass one to seven years ahead.

In this presentation, I assess these forecasts against naïve benchmark forecasts. I discuss how relevant current measures of forecast skill are for fisheries management. Finally, I show how uncertainties in causal understanding and model structure can substantially affect forecast skill.

Dr Benjamin Planque

Institute of Marine Research, Norway

Dr Benjamin Planque

Institute of Marine Research, Norway

Benjamin Planque is a quantitative ecologist at the Norwegian Institute of Marine Research (IMR). His research focuses on integrated ecosystem assessment, participatory modelling, climate-humans-ecosystem interactions, foresights, and on how ecosystem science can support ecosystem-based management in the Norwegian and Barents seas.

In the face of ongoing climate change, nature-based climate solutions (NbCS) are emerging as a potential way to help mitigate greenhouse gas (GHG) emissions through a combination of voluntary markets, Scope 3 reporting, and governmental initiatives aimed at meeting Nationally Determined Contributions and UNFCCC Stocktake reporting. Key to such approaches it the accurate quantification of carbon (C) stocks and GHG fluxes, under the status quo and management alternatives, and with robustly quantification of uncertainties and risks. One popular approach to doing so focuses on the fusion of models and data, with “true-up” monitoring data being used to “correct” inventories and improve subsequent predictions. While not often framed as such, at their heart such efforts are iterative ecological forecasts. Here we look at what NbCS can learn from ecological forecasting, using the PEcAn continental-scale terrestrial GHG reanalysis and forecasting system as a focal point for understanding the current state-of-the-art and opportunities. Specifically, the PEcAn reanalysis uses process-based models, machine learning, and Bayesian data assimilation to harmonise an unprecedented range of bottom-up and remotely-sensed data streams at much higher spatial resolutions than conventional model-only ensembles. We will touch on operationalized applications to GHG inventories and management x climate projections with the California Air Resources Board and the US DoD, efforts with the Environmental Defense Fund to improve NbCS uncertainty reporting, and new work on assimilating disturbance observations and forecasting post-disturbance recovery. Finally, we will conclude by discussing the essential role of iterative ecological forecasting in informing climate adaptation and mitigation efforts in a no-analog, non-stationary world.

Professor Michael Dietze

Boston University, US

Professor Michael Dietze

Boston University, US

Michael Dietze is professor of Earth & Environment at Boston University. He is author of the book, “Ecological Forecasting” and founding director of the Ecological Forecasting Initiative (EFI, https://ecoforecast.org), an international grassroots research consortium aimed at fostering a community of practice around near-term ecological forecasting. His lab uses a combination of field research, remote sensing, novel statistical methods, numerical models, and ecoinformatics tools to gain a quantitative understanding of ecological dynamics across scales from the individual to the globe.

Ecological forecasting is increasingly used to anticipate the consequences of environmental change, yet our understanding of predictability remains fragmented across temporal and spatial scales. Most ecological forecasting research has focused on near-term predictions, evaluating forecast skill over years to decades and emphasising the limits imposed by stochasticity, uncertainty, and nonlinear dynamics. In contrast, climate science distinguishes between two complementary forms of predictability: initial predictability, which arises from knowledge of the system’s current state, and forced predictability, which emerges when external drivers generate directional change that exceeds background variability.

In this talk, I propose a unified framework for ecological and eco-evolutionary forecasting based on these concepts. I review recent developments in intrinsic and realized predictability, including forecast horizons and measures of information contained within ecological time series. I then introduce the concept of forced ecological predictability for long-term forecasting, quantified through the Time of Emergence: the moment when climate-driven ecological change becomes distinguishable from historical variability.

Poor near-term predictability does not necessarily preclude robust long-term forecasts. Instead, long-term predictability depends on the interaction between environmental forcing, biological sensitivity, population variability, life-history characteristics, and adaptive responses. I will present comparative analyses and emerging predictability maps that identify where and when ecological signals are expected to emerge under different environmental and biological conditions.

More broadly, I argue that ecology is now poised to move from individual forecasts toward a macroecological science of predictability, one that seeks general principles explaining why some ecological systems are more predictable than others, where information about the future resides, and what ultimately determines the limits of prediction in a rapidly changing world.

Dr Stéphanie Jenouvrier

Woods Hole Oceanographic Institution, US

Dr Stéphanie Jenouvrier

Woods Hole Oceanographic Institution, US

Stéphanie Jenouvrier is a Senior Scientist at the Woods Hole Oceanographic Institution (WHOI), where she leads the FLEDGE Lab (Forecasting the Long-term Ecological and Demographic Effects of Global Environmental Change). Her research integrates climate science, demography, life-history ecology and evolution, and quantitative modelling to understand and forecast how populations and ecosystems respond to environmental change. She is particularly known for developing climate-dependent demographic and eco-evolutionary models applied to polar and marine species, including emperor penguins and albatrosses. Her work has contributed to international conservation assessments and policy discussions, including the US Endangered Species Act and IUCN Red List assessments of the emperor penguin. More broadly, her research focuses on the predictability and uncertainty of ecological responses to climate change across biological scales.

As environmental change intensifies across the globe, identifying the mechanisms that sustain ecological stability has become increasingly urgent. Ecosystems are exposed to shifting climates, altered disturbance regimes, and growing human pressures, all of which challenge their capacity to maintain structure and function. Response diversity, defined as variation among species in their responses to environmental fluctuations, has emerged as a key mechanism that may buffer ecosystems against such change and enhance their resilience. By promoting asynchronous dynamics among species, response diversity can stabilize aggregate community properties even when individual species fluctuate strongly. 

Emerging evidence from theoretical models, controlled experiments, and observational studies show that the distribution and balance of species’ environmental responses can strongly predict community stability. In many cases, these response-based metrics outperform traditional measures of biodiversity, such as species richness, in explaining variation in stability. This perspective shifts attention from how many species are present to how they respond to environmental variation. 

Nevertheless, several key knowledge gaps currently limit progress. These include the role of species interactions in shaping response diversity, the influence of different disturbance regimes and temporal scales of environmental change, and the challenges of measuring and scaling trait-based proxies across systems. Addressing these gaps will be essential for advancing a predictive understanding of ecological stability and for developing robust strategies to manage and conserve ecosystems in a rapidly changing world. 

Professor Owen Petchey

University of Zurich, Switzerland

Professor Owen Petchey

University of Zurich, Switzerland

Owen Petchey is a Professor of Ecology at the University of Zurich, where he leads a research group in the Department of Evolutionary Biology and Environmental Studies. His work focuses on understanding how biodiversity shapes the stability and functioning of ecological communities, particularly under environmental change. His current work focuses on advancing the concept of response diversity and its role in ecosystem resilience. Petchey’s research combines theory, experiments, and data analysis across scales. He is also engaged in interdisciplinary initiatives on biodiversity and sustainability, and contributes to teaching and academic leadership within the Faculty of Science.

Ecological forecasting relies on dynamically updating model predictions as new data and knowledge become available. Although biodiversity monitoring data are increasing rapidly, these datasets are rarely continuous or regularly updated, limiting their value for forecasting. Open-access data are often biased toward particular taxa or collected at spatial scales that are poorly suited for regional conservation management, while more relevant datasets frequently remain inaccessible. At the same time, environmental policy tends to prioritise biodiversity data collection over the development of dynamic forecasting systems, reducing opportunities to evaluate whether scientific outputs provide actionable knowledge. How, then, can ecological forecasting succeed? In this talk, I draw on my experience establishing a Spanish national distributed scientific centre for ecological forecasting. I discuss these challenges, which extend beyond Spain, and explore strategies to overcome current limitations and develop ecological forecasts as operational tools for conservation management and environmental policy.

Dr Maria Paniw

Spanish National Research Council, Spain

Dr Maria Paniw

Spanish National Research Council, Spain

Maria is a research fellow at the Spanish National Research Council and the national scientific coordinator of LifeWatch ERIC, a European Infrastructure dedicated to enabling the development of complex models to predict ecosystem change. The overarching theme of Maria’s research is to assess how feedbacks between phenotypic and genetic traits, individual behaviour, life-history processes, and biotic interactions can allow natural populations and communities to cope with environmental change. To he, understanding biodiversity change begins with observations of individuals. She working with long-term data on animal and plant populations to generate iterative short-term forecasts and to project population dynamics under global change scenarios - trying to create theoretical principles and empirical applications to anticipate and mitigate emerging threats to natural communities.

Predicting future population dynamics is essential for understanding how populations may respond to environmental change. Mechanistic models that explicitly incorporate demographic processes should provide improved predictive performance in comparison to phenomenological models that ignore demographic processes. In this context, integrated population models (IPMs) are particularly valuable, as they naturally combine data sources from the individual (demographic data) and the population (counts of breeders) levels while ensuring proper propagation of uncertainty. Despite their growing use, however, the ability of IPMs to estimate future population size has not yet been formally evaluated, and it is not clear whether they perform better than simpler phenomenological models based on count data only. Here, we use simulation approaches to address these gaps. Specifically, we aim to identify which assumptions and mechanisms governing demographic rates lead to predictions that are both accurate and associated with minimal uncertainty. We focus on i) temporal variation of demographic rates, ii) the inclusion of density dependence and iii) the inclusion of environmental covariates. The simulated data are analysed both with IPMs (mechanistic model) and with state-space models (phenomenological model). We found that model selection was critically important, because using a model that does not well capture the demographic processes leads to inaccurate predictions of future population sizes. Prediction uncertainty based on IPMs was typically only slightly reduced compared to the simpler state-space models. We apply our findings in a case study on European dippers (Cinclus cinclus).

Dr Michael Schaub

Swiss Ornithological Institute, Switzerland

Dr Michael Schaub

Swiss Ornithological Institute, Switzerland

Michael Schaub is a population ecologist and leads the Population Biology Research Group at the Swiss Ornithological Institute in Sempach. He also serves as an associate lecturer at the University of Bern. His research focuses on the mechanisms underlying population dynamics in birds, as well as on the development and application of statistical models for analysing population data. A major focus of his recent work has been on integrated population models, including the publication of the first textbook on this topic.

The Kunming-Montreal Global Biodiversity Framework is driving efforts to halt and reverse biodiversity loss across the world. To support these efforts, decision-makers need evidence on the scale and combination of interventions that should be implemented now for confidence in meeting agreed targets in the future. Biodiversity models play a key role in providing this evidence in a systematic and comparable way. However, the number of biodiversity models has rapidly expanded in recent years, varying in purpose, assumptions, and methodological choices shaped by the specific decision contexts each model was designed to address. While this diversity brings valuable perspectives, it also introduces variation in how models represent ecological processes, including dynamic processes leading to lagged responses and drivers of change, such as indirect drivers or interactive effects. Moreover, the underlying design choices that drive these differences are often not clearly communicated to end users, making it difficult to interpret results and compare findings across models. In this talk, I will draw on some of our experiences of working with UK government agencies on the Biodiversity Pathways project. Through this project, we have mapped the typical questions that policymakers have about biodiversity futures and have contrasted the modelling options relevant for different questions. I will discuss the needs for a modelling typology that summarises the key aspects of variation among models, which can guide modelling choice and help communicate how predictions from a model should be interpreted to end users.

Dr Diana Bowler

UK Centre for Ecology and Hydrology, UK

Dr Diana Bowler

UK Centre for Ecology and Hydrology, UK

Diana is a Senior Scientist at the UK Centre for Ecology and Hydrology. Her research spans from monitoring design, population modelling to large-scale biodiversity patterns, particularly related to insect conservation. As well as ecological questions about the nature of biodiversity change, she works on statistical approaches for extracting biological signals from heterogenous monitoring data, including integrated modelling. She has a special interest in the power of citizen science as a tool for large-scale monitoring and engaging society in conservation, developing tools to improve data quality and flows. More recently, her work has shifted into predictive modelling through projects working in partnership with government agencies.