NAGIOS: RODERIC FUNCIONANDO

Spatio temporal modeling of species distribution

Repositori DSpace/Manakin

IMPORTANT: Aquest repositori està en una versió antiga des del 3/12/2023. La nova instal.lació está en https://roderic.uv.es/

Spatio temporal modeling of species distribution

dc.contributor.advisor	Blangiardo, Marta
dc.contributor.advisor	López Quílez, Antonio
dc.contributor.author	Rodríguez de Rivera Ortega, Oscar
dc.contributor.other	Departament d'Estadística i Investigació Operativa	es_ES
dc.date.accessioned	2019-01-09T13:48:18Z
dc.date.available	2019-01-10T05:45:06Z
dc.date.issued	2019	es_ES
dc.date.submitted	17-01-2019	es_ES
dc.identifier.uri	http://hdl.handle.net/10550/68427
dc.description.abstract	The aim of this thesis is study spatial distribution of different groups from different perspectives and to analyse the different approaches to this problem. We move away from the classical approach, commonly used by ecologists, to more complex solutions, already applied in several disciplines. We are focused in applying advanced modelling techniques in order to understand species distribution and species behaviour and the relationships between them and environmental factors and have used first the most common models applied in ecology to move then to more advanced and complex perspectives. From a general perspective and comparing the different models applied during the process, from MaxEnt to spatio-temporal models with INLA, we can affirm that the models that we have developed show better results that the already built. Also, it is difficult to compare between the different approaches, but the Bayesian approach shows more flexibility and also the inclusion of spatial field or the latent spatio-temporal process allows to include residuals as a proxy for unmeasured variables. Compared with additive models with thin plate splines, probably considered one of the greatest methods to analyse species distribution models working with presence-absence data, comparable to MaxEnt, CART and MARS, our results show a better fit and more flexibility in the design. As a natural process we have realised that the Bayesian approach could be a better solution or at least a different approach for consideration. The main advantage of the Bayesian model formulation is the computational ease in model fit and prediction compared to classical geostatistical methods. To do so, instead of MCMC we have used the novel integrated nested Laplace approximation approach through the Stochastic Partial Differential Equation (SPDE) approach. The SPDE approach can be easily implemented providing results in reasonable computing time (comparing with MCMC). We showed how SPDE is a useful tool in the analysis of species distribution. This modelling could be expanded to the spatio-temporal domain by incorporating an extra term for the temporal effect, using parametric or semiparametric constructions to reflect linear, nonlinear, autoregressive or more complex behaviours. We can conclude that spatial and spatio-temporal Bayesian models are a really interesting approach for the understanding of environmental dynamics, not only because of the possibility to develop and solve more complex problems but also for the easy understanding of the implementation processes.	en_US
dc.description.abstract	The aim of this thesis is study spatial distribution of different groups from different perspectives and to analyse the different approaches to this problem. We move away from the classical approach, commonly used by ecologists, to more complex solutions, already applied in several disciplines. We are focused in applying advanced modelling techniques in order to understand species distribution and species behaviour and the relationships between them and environmental factors and have used first the most common models applied in ecology to move then to more advanced and complex perspectives. From a general perspective and comparing the different models applied during the process, from MaxEnt to spatio-temporal models with INLA, we can affirm that the models that we have developed show better results that the already built. Also, it is difficult to compare between the different approaches, but the Bayesian approach shows more flexibility and also the inclusion of spatial field or the latent spatio-temporal process allows to include residuals as a proxy for unmeasured variables. Compared with additive models with thin plate splines, probably considered one of the greatest methods to analyse species distribution models working with presence-absence data, comparable to MaxEnt, CART and MARS, our results show a better fit and more flexibility in the design. As a natural process we have realised that the Bayesian approach could be a better solution or at least a different approach for consideration. The main advantage of the Bayesian model formulation is the computational ease in model fit and prediction compared to classical geostatistical methods. To do so, instead of MCMC we have used the novel integrated nested Laplace approximation approach through the Stochastic Partial Differential Equation (SPDE) approach. The SPDE approach can be easily implemented providing results in reasonable computing time (comparing with MCMC). We showed how SPDE is a useful tool in the analysis of species distribution. This modelling could be expanded to the spatio-temporal domain by incorporating an extra term for the temporal effect, using parametric or semiparametric constructions to reflect linear, nonlinear, autoregressive or more complex behaviours. We can conclude that spatial and spatio-temporal Bayesian models are a really interesting approach for the understanding of environmental dynamics, not only because of the possibility to develop and solve more complex problems but also for the easy understanding of the implementation processes.	es_ES
dc.format.extent	130 p.	es_ES
dc.language.iso	en	es_ES
dc.subject	ecology	es_ES
dc.subject	species distribution model	es_ES
dc.subject	bayesian hierarchical model	es_ES
dc.title	Spatio temporal modeling of species distribution	es_ES
dc.type	doctoral thesis	es_ES
dc.subject.unesco	UNESCO::MATEMÁTICAS::Estadística	es_ES
dc.subject.unesco	UNESCO::MATEMÁTICAS	es_ES
dc.embargo.terms	0 days	es_ES