Keywords
NVC, ordination, pseudoquadrats, vegetation, sub-community.
Start Date
26-6-2018 10:40 AM
End Date
26-6-2018 12:00 PM
Abstract
Many countries have developed phytosociological classifications of their vegetation to describe semi-natural and natural vegetation communities. Can methods be developed for any classification system to allocate newly surveyed quadrats, i.e. data gathered from new field surveys, into the most likely vegetation community? Algorithms or software already exist to allocate quadrats for some classifications, but these are not generalisable to any system. We test the robustness of several generalisable approaches to allocate quadrats to an existing phytosociological classification, using the British National Vegetation Classification (NVC) as a case study. Vegetation from 167 quadrats was classified using two-way indicator species analysis (TWINSPAN) and the resultant groups allocated to communities within the NVC using the NVC-specific 'MAVIS' software. Sets of artificial 'pseudoquadrats' for potential communities were computer-generated based on either the published NVC community descriptions or from the subset of species surveyed. Distance in ordination space of observed quadrats from pseudoquadrats was used to predict community type. The conventional NVC-specific MAVIS classification produced 11 sub-communities at the site, and this was assumed to be the most reliable descriptor of the vegetation communities. There was a close match between the pseudoquadrat-based community predictions and the MAVIS predictions, although pseudoquadrats based on the subset of species observed at the site appeared to be slightly more reliable. Our results demonstrate that the use of pseudoquadrats provides a flexible, generalisable means to allocate objectively vegetation quadrats into any extant classification system.
Generalisable Methods for Vegetation Classification Using Computer-Generated Pseudoquadrats
Many countries have developed phytosociological classifications of their vegetation to describe semi-natural and natural vegetation communities. Can methods be developed for any classification system to allocate newly surveyed quadrats, i.e. data gathered from new field surveys, into the most likely vegetation community? Algorithms or software already exist to allocate quadrats for some classifications, but these are not generalisable to any system. We test the robustness of several generalisable approaches to allocate quadrats to an existing phytosociological classification, using the British National Vegetation Classification (NVC) as a case study. Vegetation from 167 quadrats was classified using two-way indicator species analysis (TWINSPAN) and the resultant groups allocated to communities within the NVC using the NVC-specific 'MAVIS' software. Sets of artificial 'pseudoquadrats' for potential communities were computer-generated based on either the published NVC community descriptions or from the subset of species surveyed. Distance in ordination space of observed quadrats from pseudoquadrats was used to predict community type. The conventional NVC-specific MAVIS classification produced 11 sub-communities at the site, and this was assumed to be the most reliable descriptor of the vegetation communities. There was a close match between the pseudoquadrat-based community predictions and the MAVIS predictions, although pseudoquadrats based on the subset of species observed at the site appeared to be slightly more reliable. Our results demonstrate that the use of pseudoquadrats provides a flexible, generalisable means to allocate objectively vegetation quadrats into any extant classification system.
Stream and Session
I plan, if accepted, to submit an oral presentation.
Stream B: (Big) Data Solutions for Planning, Management, and Operation and Environmental Systems
Session B2: Hybrid modelling and innovative data analysis for integrated environmental decision support