*Increasing the utility of predictive models: Understanding model transferability*
Combined symposium and focus group held in association with the 4th International Marine Conservation Congress (IMCC4) <http://conbio.org/mini-sites/imcc-2016> St. John’s, Newfoundland and Labrador (YYT), Canada Location: Delta Conference Centre, Salon G Date: Sunday July 31, 2016 Time: 08.30am - 10.30am (SY) & 11am - 1pm (FG) We would like to invite all interested IMCC4 attendees to a combined symposium and focus group dedicated to improving the usefulness of predictions derived from statistical models of marine biodiversity (including marine mammals). In the face of global environmental change and scant conservation resources, it is critical that models of biota distribution, abundance or species richness be broadly applicable. The session will therefore concentrate on the issue of spatial and temporal transferability, i.e. the process of building a model in a given area (or time period) and projecting it into another for inference. In the first half of the session (symposium SY42 <http://conbio.org/mini-sites/imcc-2016/program-events/symposia/#July%2031>), a series of keynote presentations will be given to summarise the current state of knowledge on model transferability, illustrating this with practical examples of both transferred models that have performed well and others that have not done so well. The second half (focus group FG43 <http://conbio.org/mini-sites/imcc-2016/program-events/focus-groups/#10>) will take the form of a round-table discussion during which participants will be able to share their experience, recent findings, and formulate questions to identify the fundamental obstacles and opportunities to making transferability a central part of model development and testing. Results from the focus group will provide the foundation for a written publication in the open-access journal Frontiers in Marine Science, to which participants will be encouraged to contribute. There are no registration fees, however light refreshments and finger food will be provided so please RSVP to [email protected] <mailto:[email protected]> so that we can cater for the right number of participants. For further information, please contact Dr. Katherine Yates <mailto:[email protected]> / Dr. Ana Sequeira <mailto:[email protected]> Looking forward to seeing you there! Symposium and focus group organisers. Dr. Katherine Yates, Salford University; Dr. Ana Martins Sequeira, University of Western Australia; Dr. Phil Bouchet, University of Western Australia; Dr. Julian Caley, Queensland University of Technology and Australian Research Council Centre of Excellence for Mathematical and Statistical Frontiers; Prof. Kerrie Mengersen, Queensland University of Technology. Abstract (symposium SY42) Effective planning and prioritisation of conservation actions requires an understanding of where conservation features of interest occur and how management actions may affect them. All too often, however, information on the distribution of biotic features is sparse or lacking. This is particularly true for marine environments, where the vastness of the oceans and the prohibitive costs associated with sampling limits data collection. For many locations, only abiotic and spatial data exist. In these situations transferable models, i.e., models developed for a particular place but which can provide useful information in other locations, could be of great utility. Despite transferability studies in terrestrial systems being relatively common, the model features that may enhance or detract from transferability are still not well understood. This symposium will focus on how best to build predictive models that are highly transferable and how to robustly assess transferability while showing new applications to marine systems. Researchers working on different aspects of transferability will show examples of where transferred models have performed well, even across large distances, and others where they have not. They will also present evidence for which factors seem to affect the predictive performance of transferred models in the marine environment. Joint session is to allow for more in-depth discussion, which will explore recent findings, highlight which are the immediate gaps, and discuss future research avenues. Abstract (focus group FG43) Large areas of the oceans are poorly sampled and are likely to remain so for the foreseeable future. However, planning conservation and management actions requires an understanding of the spatial distribution of features of interest. Thus, it is often desirable to make predictions for areas in which data is lacking. In these cases transferable models would be of substantial value; that is if a model developed for a particular location could be used to make useful predictions at other locations. Little research has focused on model transferability in the marine environment and the features that may enhance or detract from model transferability are still not well understood. Following on from the symposium of the same title, this focus group will explore model transferability. Organised around a series of key questions, this session will be divided into a mix of small group discussions and open debates. Participants will be offered the opportunity to contribute to a written output from the focus group. Light refreshments will be provided.
_______________________________________________ MARMAM mailing list [email protected] https://lists.uvic.ca/mailman/listinfo/marmam
