The identification of spatial scales is a fundamental topic in ecology, commonly addressed through multiscale analyses like the Moran's Eigenvector Maps (MEMs). This technique entails the derivation of scales through the definition of a model of spatial organization, i.e. of the spatial relationships among the elements of ecological systems. Defining the spatial organization model is highly nontrivial, but has fundamental implications for both the technical implementation of the analysis and, most importantly, the ecological understanding of the system. To address this issue, we developed the Unravelled Voting Algorithm (UVA), a novel paradigm and analytical framework based on MEMs, to investigate the spatial organization and scales of ecological systems. UVA revolves on three key points: i) consensus in moving from the univariate to the multivariate domain, ii) post-selection inference, iii) rank-based techniques. Its performances were evaluated, through simulations and real data, in terms of accuracy in identifying the spatial organization model and scales, computational costs and flexibility. UVA demonstrated an outstanding accuracy in recognizing both the spatial organization model and the spatial scales, and can be used with signal-to-noise ratios as low as ≈ 0.5. The current implementation for the R programming language can already attain ×50 faster computation time than related procedures, with far superior accuracy. The generation of novel non-spatial attributes, coding the relative preference of each element of the ecological system toward different spatial organization models, allows exploring the relative variations in spatial patterns. UVA defines a modular framework setting new standards in the investigation of spatial organization and scales of ecological systems. Its flexibility makes it adaptable to any analytical requirement, open to large improvements and future-proof. On top of its outstanding accuracy, it paves the way to the analysis of the variations in patterns and scales among the elements of ecological systems.
The Unravelled Voting Algorithm: a novel framework to investigate the spatial organization of ecological systems
Bellino A.
;Baldantoni D.
2022-01-01
Abstract
The identification of spatial scales is a fundamental topic in ecology, commonly addressed through multiscale analyses like the Moran's Eigenvector Maps (MEMs). This technique entails the derivation of scales through the definition of a model of spatial organization, i.e. of the spatial relationships among the elements of ecological systems. Defining the spatial organization model is highly nontrivial, but has fundamental implications for both the technical implementation of the analysis and, most importantly, the ecological understanding of the system. To address this issue, we developed the Unravelled Voting Algorithm (UVA), a novel paradigm and analytical framework based on MEMs, to investigate the spatial organization and scales of ecological systems. UVA revolves on three key points: i) consensus in moving from the univariate to the multivariate domain, ii) post-selection inference, iii) rank-based techniques. Its performances were evaluated, through simulations and real data, in terms of accuracy in identifying the spatial organization model and scales, computational costs and flexibility. UVA demonstrated an outstanding accuracy in recognizing both the spatial organization model and the spatial scales, and can be used with signal-to-noise ratios as low as ≈ 0.5. The current implementation for the R programming language can already attain ×50 faster computation time than related procedures, with far superior accuracy. The generation of novel non-spatial attributes, coding the relative preference of each element of the ecological system toward different spatial organization models, allows exploring the relative variations in spatial patterns. UVA defines a modular framework setting new standards in the investigation of spatial organization and scales of ecological systems. Its flexibility makes it adaptable to any analytical requirement, open to large improvements and future-proof. On top of its outstanding accuracy, it paves the way to the analysis of the variations in patterns and scales among the elements of ecological systems.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
