Affordable Access

Spatial scale-dependent prediction in marine distribution modeling: A case study

Authors
Publication Date
Disciplines
  • Computer Science
  • Ecology
  • Geography

Abstract

Spatial scale, defined as the spatial resolution and extent of environmental data, is known to influence species distribution models. With most studies investigating spatial scale effects in the context of predicting impacts of global change in the terrestrial realm, comparative species distribution modeling based on fine-scale versus coarse-scale environmental data in the marine realm remains to be investigated. We present a regional case study for three benthic seaweed species with different well-defined distribution patterns in two adjacent but contrasting seas: the spatially and temporally more homogeneous Gulf of Oman and the highly variable Arabian Sea. Rather than starting from a single environmental dataset with subsequent up- or downscaling, a genuine sub-100m resolution environmental dataset was compiled based on 10 mosaiced Landsat scenes for winter and summer. This resulted in habitat layers pertaining to sea surface temperature, nutrient content, turbidity and substrate availability. The coarse-scale dataset (9km resolution) is based on the global environmental dataset Bio-ORACLE, cropped to the same 2000-km long coastline. Models for the three species were generated with the Maxent algorithm using both environmental datasets. The Landsat-based models performed equally well in terms of AUC compared to the Bio-ORACLE models. However, important differences in output maps could be noted, capturing the difference between coarse-scale macroecological modeling and fine-scale habitat modeling. Overall, it appears that while both coarse and fine-scale models are in good agreement for all species in the less variable Gulf of Oman, coarse-scale models suffered from overprediction for all three species in the more heterogeneous Arabian Sea. This suggests that the choice of modeling scale is important for applied marine modeling studies such as the guiding of field work.

There are no comments yet on this publication. Be the first to share your thoughts.