Keywords
species distribution modelling, gis, open source, integration
Start Date
1-7-2012 12:00 AM
Abstract
Species Distribution Models (SDMs) are correlative models that use environmental and/or geographical information to explain patterns of species occurrences. Those models are being used in various fields including climate change, invasive species research, evolutionary biology and epidemiology. Thanks to the availability of increasing computational resources, new methods continue to be developed. However, software packages that include the SDM algorithms usually focus on one or few methods, and have different degrees of integration with other geographical and statistical software. Specifically, SDM implementations are often standalone programs developed by university laboratories, either as extensions to statistical software. In few cases they are written as extensions for the most common proprietary Geographic Information System (GIS) products, despite the strong geographical component present in the data. On the other hand, open source GIS software has loose connection with SDM implementations, and usually requires more effort to build a complete and well connected software stack. This paper investigate on the possible causes of the present separation of open source GIS and Species Distribution Modelling and on the benefits of a closer integration, and lists a selection of candidates for future joint development. A further step would be the adoption of open source principles in the implementation process of SDM. This will enable a peer-review mechanism on the computational code, that will strongly reduce the risk of attaining biased results due to inaccurate implementations.
Species distribution modelling and open source GIS: why are they still so loosely connected?
Species Distribution Models (SDMs) are correlative models that use environmental and/or geographical information to explain patterns of species occurrences. Those models are being used in various fields including climate change, invasive species research, evolutionary biology and epidemiology. Thanks to the availability of increasing computational resources, new methods continue to be developed. However, software packages that include the SDM algorithms usually focus on one or few methods, and have different degrees of integration with other geographical and statistical software. Specifically, SDM implementations are often standalone programs developed by university laboratories, either as extensions to statistical software. In few cases they are written as extensions for the most common proprietary Geographic Information System (GIS) products, despite the strong geographical component present in the data. On the other hand, open source GIS software has loose connection with SDM implementations, and usually requires more effort to build a complete and well connected software stack. This paper investigate on the possible causes of the present separation of open source GIS and Species Distribution Modelling and on the benefits of a closer integration, and lists a selection of candidates for future joint development. A further step would be the adoption of open source principles in the implementation process of SDM. This will enable a peer-review mechanism on the computational code, that will strongly reduce the risk of attaining biased results due to inaccurate implementations.
