Report Open Access

Use of species distribution modeling in the deep sea

Kenchington, E.; Callery, O.; Davidson, F.; Grehan, A.; Morato, T.; Appiott, J.; Davis, A.; Dunstan, P.; Du Preez, C.; Finney, J.; González-Irusta, J. M.; Howell, K.; Knudby, A.; Lacharité, M.; Lee, J; Murillo, F. J.; Beazley, L.; Roberts, J. M.; Roberts, M.; Rooper, C.; Rowden, A.; Rubidge, E.; Rowden, A.; Stanley, R.; Stirling, D.; Tanaka, K. R.; Vanhatalo, J.; Weigel, B.; Woolley, S; Yesson, C.

Use of Species Distribution Modeling in the Deep Sea.  Published in the Canadian Technical Report of
Fisheries and Aquatic Sciences 3296 (2019)

In the last two decades the use of species distribution modeling (SDM) for the study and management of marine species has increased dramatically. The availability of predictor variables on a global scale and the ease of use of SDM techniques have resulted in a proliferation of research on the topic of species distribution in the deep sea. Translation of research projects into management tools that can be used to make decisions in the face of changing climate and increasing exploitation of deep-sea resources has been less rapid but necessary. The goal of this workshop was to discuss methods and variables for modeling species distributions in deep-sea habitats and produce standards that can be used to judge SDMs that may be useful to meet management and conservation goals. During the workshop, approaches to modeling and environmental data were discussed and guidelines developed including the desire that 1) environmental variables should be chosen for ecological significance a priori; 2) the scale and accuracy of environmental data should be considered in choosing a modeling method; 3) when possible proxy variables such as depth should be avoided if causal variables are available; 4) models with statistically robust and rigorous outputs are preferred, but not always possible; and 5) model validation is important. Although general guidelines for SDMs were developed, in most cases management issues and objectives should be considered when designing a modeling project. In particular, the trade-off between model complexity and researcher’s ability to communicate input data, modeling method, results and uncertainty is an important consideration for the target audience.

157
95
views