We are pleased to share the publication of our new manuscript, which assesses how well telemetry-based SDMs transfer through space and time using marine mammal and seabird tracking data. These findings can inform which model algorithms and pseudo-absence techniques are best to use to support dynamic ocean management. Title: Temporal and spatial transferability in telemetry-based dynamic species distribution models: The effects of algorithms and pseudo-absence techniques Authors: Joshua C Wilson, Philip N Trathan, Hugh J Venables, Horst Bornemann, Rochelle Constantine, Daniel P Costa, Luciano Dalla Rosa, Louise Emmerson, Ari S Friedlaender, Michael E Goebel, Simon Goldsworthy, Mark A Hindell, Mary-Anne Lea, Mônica M C Muelbert, Silvia Olmastroni, Yan Ropert-Coudert, Colin Southwell, Ryan R Reisinger Abstract: Species distribution models using animal tracking data to predict foraging habitat suitability can inform dynamic ocean management techniques that respond to changing environmental conditions. Such models require accurate transferability in time (and sometimes space), often involving extrapolation into new environmental conditions. However, the impacts of modelling configuration on transferability are unclear. Here we built species distribution models using animal tracking studies from the Southern Ocean and projected them in time and space. We tested 24 different model configurations to assess how the choice of pseudo-absence technique and algorithm influences temporal and spatial transferability. Using data from a variety of seabird and marine mammal species with differing ecological traits, we aimed to identify 1) which model configurations produce consistently high transferability scores and 2) whether any tested algorithms or pseudo-absence techniques were better equipped to deal with environmental extrapolation and small sample sizes. Models consistently achieved high temporal transferability scores. Of all tested configurations, background sampling combined with tree-based machine learning algorithms or models accounting for autocorrelation performed best in this context. Conversely, no model configuration consistently attained high spatial transferability scores, with all exhibiting poor predictive capacity in many cases. The impacts of environmental extrapolation and sample size on transferability were also assessed. Most models exhibited greater temporal transferability when built with larger datasets that required lesser environmental extrapolation. We recommend that researchers use data-rich ensembles of the most reliable algorithms built with background sampling when looking to predict near real-time distributions and, where possible, avoid spatial extrapolation when data do not cover every population within the target area. DOI: https://doi.org/10.1016/j.ecolmodel.2026.111702 Please feel free to reach out if you have any questions! All the best, Josh ___________________________________________________________________________________________________ Joshua Wilson (he/him) PhD Student | University of Southampton//British Antarctic Survey | INSPIRE DTP National Oceanography Centre, European Way, Southampton, SO14 3ZH LinkedIn<https://www.linkedin.com/in/joshua-wilson-9b5a69147/> | Bluesky<https://bsky.app/profile/oshuwilson.bsky.social>