Parameter redundancy with applications in statistical ecology

摘要: This thesis is concerned with parameter redundancy in statistical ecology models. If it not possible to estimate all the parameters, a model termed redundant. Parameter commonly occurs when parameters are confounded so that could be reparameterised terms of smaller number parameters. In principle, use symbolic algebra determine whether or certain ecological can estimated using classical methods inference. We examine variety different models: We begin by exploring models based on marking animals and observing same at future time points. These observations either animal marked then recovered dead mark-recovery modelling, then recaptured alive capture-recapture modelling. also explore capture-recapture-recovery where both recoveries recaptures observed study. go occupancy which used obtain estimates probability presence, absence, for living species repeated detection surveys, these have advantage individuals required marked. A examined included addition season-dependent group-dependent species-dependent, along other models. We investigate deriving general results model's dependencies relaxed suited studies. analyse how change specific data sets sparse influence redundant procedures written Maple. theory vital correct valid inference made.