Roundtables
Len Thomas (University of St Andrews)
Ben Stevenson (University of Auckland)
Paul van Dam-bates (Fisheries and Oceans Canada)
The past, present and future of
spatial capture recapture
In this round table we will briefly review the history of spatial capture recapture (SCR) methods (originally called spatially explicit capture recapture), give an overview of the current state of the art and discuss together possible and promising future directions for methodological development and application. We will also emphasize the links between SCR and other approaches, such as mark recapture distance sampling and modelling animal movement.
Frédéric Barraquand (CNRS & University of Bordeaux)
Publishing statistical ecology: how do we ensure equitable access to the literature and publication opportunities for a global, interdisciplinary community?
The publishing ecosystem is dominated by a few large commercial publishers now promoting the gold open access model, with substantial article processing charges. The gold open access model as most often practiced not only excludes some scientists from publication, but also increases costs to universities and institutions, and is therefore not sustainable. Attempts to produce a more just open access to the literature are varied, including green open access, with preprints / postprints of paywalled content deposited on public servers, and diamond open access, where articles are free to read and publish, and the journals themselves funded by institutions. This round table will discuss open access models from an international perspective, their relative popularity in ecology and in statistics and how they interact with disciplinary cultures, the current challenges they face (rise of AI content, read-and-publish agreements,…), and possible ways forward.
Matthew Weldy (National Audubon Society)
Sarah Saunders (National Audubon Society)
Tim Meehan (National Audubon Society)
Sensor-based observations: building conceptual frameworks for modern ecological data
Recent advances in sensors and machine learning are transforming ecological data collection. Automated recorders, cameras, and other sensors now generate large, continuous, and high-dimensional data streams, which are typically processed by classifiers (e.g., convolutional neural networks, random forests) to infer species detections or other ecological events. These outputs differ fundamentally from traditional, observer-based survey data, offering continuous prediction scores rather than discrete detections, and thus raise new conceptual and analytical challenges. This shift from binary to probabilistic evidence decouples classification from decision-making and represents a fundamental change in how ecological observations are generated and interpreted. This roundtable will explore conceptual and analytical frameworks for integrating sensor-derived data into ecological inference. Discussion will focus on: 1) How continuous classifier outputs differ from observer-based data and what they imply for inference 2) Statistical approaches for calibrating and interpreting prediction scores, including correlated errors across species 3) Methods for analysing classifier data that address false-positive predictions, propagate classification uncertainty, and scale to multi-species contexts The overall goal of this proposed roundtable is to develop shared conceptual foundations and statistical tools to bridge the gap between machine learning outputs and ecological decision-making.
