Modelling habitat suitability of whales in the Southern Ocean
Ahmed El-Gabbas, Ilse Van Opzeeland, Elke Burkhardt, Olaf Boebel
2019
Poster
Abstract
Detailed information on cetacean distribution is crucial to identify large-scale conservation actions and management decisions. Understanding the ecological drivers behind their spatial patterns in the Southern Ocean is complicated by whales’ mobility and the logistic restrictions in collecting data in polar environments. Species distribution models have become essential tools in ecology and conservation. They relate information on species occurrence with environmental predictors thought to influence its habitat use, to predict its potential distribution and explain environmental drivers of the observed patterns.
In this study, we compiled opportunistic presenceonly data for seven whale species in the Southern Ocean from multiple sources. A quality-controlled data set was then used to model species distributions using Maxent software (under the point process modelling framework). Environmental predictors were prepared from multiple in-situ and remotely-sensed sources, based on our experience of the study area and species ecology. We estimated the best combinations of Maxent’s parameters & evaluated model performance on a species-specific spatial block cross-validation to maintain spatial independence between training and testing data.
For each species, block size and their spatial allocation into crossvalidation folds was objectively determined according to how much spatial-autocorrelation exists at occurrences. For each of species, we 1) predicted circumpolar potential distribution, 2) determined the most important variables, and 3) showed the relationship between habitat suitability and environmental variables. We believe that our results would be of great importance to explain the habitat preference of species in the Southern Ocean, for the first time for the majority of studied species. However, we argue that these models can only represent a hypothetical, mean state (which actually never becomes manifest) of the potential distribution of the species in space, and hence another set of dynamic models are required to consider the high dynamic environment in the Southern Ocean and the migratory nature of whales.
Publication
World Marine Mammals Conference (WMMC), Barcelona, Spain. December 9-12, 2019
Detailed information on cetacean distribution is crucial to identify large-scale conservation actions and management decisions. Understanding the ecological drivers behind their spatial patterns in the Southern Ocean is complicated by whales’ mobility and the logistic restrictions in collecting data in polar environments. Species distribution models have become essential tools in ecology and conservation. They relate information on species occurrence with environmental predictors thought to influence its habitat use, to predict its potential distribution and explain environmental drivers of the observed patterns. In this study, we compiled opportunistic presence-only data for seven whale species in the Southern Ocean from multiple sources. A quality-controlled data set was then used to model species distributions using Maxent software (under the point process modelling framework). Environmental predictors were prepared from multiple in-situ and remotely-sensed sources, based on our experience of the study area and species ecology. We estimated the best combinations of Maxent’s parameters & evaluated model performance on a species-specific spatial block cross-validation to maintain spatial independence between training and testing data. For each species, block size and their spatial allocation into cross-validation folds was objectively determined according to how much spatial-autocorrelation exists at occurrences. For each of species, we 1) predicted circumpolar potential distribution, 2) determined the most important variables, and 3) showed the relationship between habitat suitability and environmental variables. We believe that our results would be of great importance to explain the habitat preference of species in the Southern Ocean, for the first time for the majority of studied species. However, we argue that these models can only represent a hypothetical, mean state (which actually never becomes manifest) of the potential distribution of the species in space, and hence another set of dynamic models are required to consider the high dynamic environment in the Southern Ocean and the migratory nature of whales.