Modeling multiscale structures in ecological data via Spatial Eigenfunction Analysis
Mattia Pancerasa
DEIB PhD student
DEIB - Seminar Room "Alessandra Alario" (building 21, 4th floor)
May 31st, 2017
5.00 pm
Contacts:
Renato Casagrandi
Research Line:
Planning and management of environmental systems
DEIB PhD student
DEIB - Seminar Room "Alessandra Alario" (building 21, 4th floor)
May 31st, 2017
5.00 pm
Contacts:
Renato Casagrandi
Research Line:
Planning and management of environmental systems
Sommario
One of the most studied problem in Ecology is how the abundance of species is affected by environmental processes, whether biotic or abiotic. Ecological data present strong spatial structures, measured in terms of correlation at different distance scales, which may depend on environmental variables themselves or population dynamics. Such structures must be incorporated into statistical analysis using appropriate tools.
Spatial Eigenfunction Analysis (SEA) is a family of methods for multiscale analysis of univariate or multivariate response data. It combines early approaches developed by geographers for analyzing binary spatial connection matrices with more recent advances that account for distances among localities. SEA can map spatial correlation within ecological data by providing a set of orthogonal variables, which can be used in regression models and variance partitioning.
This seminar summarizes concepts and methodologies taught in the one-week course ‘Recent advances in Analysis of Multivariate Ecological Data’, held last October by Prof. P. Legendre and Prof. D. Borcard (Département de Sciences Biologiques, Université de Montréal) at Abdus Salam International Centre for Theoretical Physics in Trieste.
Spatial Eigenfunction Analysis (SEA) is a family of methods for multiscale analysis of univariate or multivariate response data. It combines early approaches developed by geographers for analyzing binary spatial connection matrices with more recent advances that account for distances among localities. SEA can map spatial correlation within ecological data by providing a set of orthogonal variables, which can be used in regression models and variance partitioning.
This seminar summarizes concepts and methodologies taught in the one-week course ‘Recent advances in Analysis of Multivariate Ecological Data’, held last October by Prof. P. Legendre and Prof. D. Borcard (Département de Sciences Biologiques, Université de Montréal) at Abdus Salam International Centre for Theoretical Physics in Trieste.