Integrated population modelling of dependent data structures

The modelling of wild animal populations is of utmost importance in today's climate of global change. There is considerable threat to the survival of native species and it is necessary to determine why these threats are occurring and what can be done to prevent the loss of species forever. The mathematical modelling of animal populations facilitates the estimation of important demographic parameters and can confirm their relationship with spatial, environmental and individual covariates.

Simple models were satisfactory for simple data sets. However, the development of sophisticated statistical models is severely lacking given the wealth of detailed individual level data being collected on a huge range of animal populations. This fellowship will achieve the ultimate goal of developing an individual level model which accounts for fundamental correlations between data sets.

It is often the case that multiple data sets are compiled from a single population under study. Until recently analyses on the different types of data were analysed in a piecemeal approach, extracting the parameters of interest from each data analysis. However the theory of integrated population modelling demonstrated the benefits of modelling multiple types of data within one coherent framework. The theory of integrated population modelling relies on assumptions of independence of the component data sets. This assumption is violated if the same individuals contribute to more than one data set. Incorrectly fitting integrated population models to dependent data sets can result in biased estimates of model parameters.

The research proposed within this fellowship will provide a new individual level model which will include all available information and will correctly account for the dependence of the different data types. The new model will incorporate imperfect detection of individuals and offer an approach to estimate likely parentage using just life history data. Developments will also be offered to account for incomplete overlap between individuals contributing to demographic and population count data.

The new methodology will be derived in order to provide an all-purpose model and as such the potential applications are considerable. Within this fellowship the new models will be fitted to two long-running case studies: Isle of Rum red deer and Alpine ibex in the Gran Paradiso National Park, Italy. These case studies have been selected to allow the robustness of the new modelling approaches to be assessed for populations with varying degrees of overlap between component data sets and will facilitate the answering of important biological objectives.
Key statistical aspects of model discrimination and goodness-of-fit assessment will be addressed and software promoting the use of the new procedures will be released.

Statistical developments within the ecological field greatly impact the methodology used by quantitative biologists and have wider application to other areas of applied statistics. The dissemination of statistical results to a statistical, biological and public audience will be of great importance. Outside of the academic research community there are individuals who interpret model results who would benefit from a more thorough understanding of how these models are chosen for a particular data set. Without the basic understanding of modelling assumptions and model assessment, it is impossible to interpret correctly parameter estimates.

Non-academic bodies often employ statisticians for model fitting and these individuals will gain from the publication of results in peer-reviewed scientific journals. I have extensive experience in working in conjunction with data collectors with little or no statistical background. Engaging with these individuals encourages the collection of extensive data sets which are invaluable for our purposes. Without these sorts of interactions, data collection would not advance as profitably as it does at the moment.

This work can impact significantly on industry. Many companies seek to mitigate environmental impact, for example paper and timber industries and agriculture, basing their decisions on ecological evidence from models coupled with economic considerations. These studies, by improving and refining modelling techniques will make a significant impact on these important corporate decisions. During this fellowship I will engage with this sector by direct discussion and consultation. This will further develop my existing communication and presentation skills to statisticians and non-specialists which will be of increasing importance as my career progresses.

Critical to the dissemination of this work will be the translation of the new models into appropriate software for ecologists, decision makers and fellow statisticians. I will be writing an R package and will make it freely available to provide people with the tools to fit the new complex models to multiple data types simultaneously and will also provide short courses to teach people the power of the available software and will write and regularly update a website to promote and provide a resource for interested parties.

The sectors identified will all be made aware of the development of the specialised software through use of subject-specific mailing lists, website promotion and through engagement with relevant societies, such as the Royal Statistical Society, International Biometric Society and British Ecological Society. I will run a workshop, Integrated population modelling of dependent data structures, in conjunction with the bi-annual International Statistical Ecology Conference. The state-of-the-art video-conferencing facilities at the University of Kent will provide a means to link with my research collaborators and also for the development of new working relationships following meetings at conferences and workshops.

I will also host a series of guest evening lectures on the topic of the future direction of statistical ecology, to promote discussion and generate interaction with interested members of the public and local community.