National Centre for Statistical Ecology - beyond 2010

Lead Research Organisation: University of St Andrews
Department Name: Mathematics and Statistics

Abstract

The National Centre for Statistical Ecology (NCSE) has established itself as an international centre of excellence, with a reputation for producing ground-breaking research. It was founded in October 2005, supported by a five-year multi-disciplinary grant from EPSRC, with nodes at the Universities of Cambridge, Kent and St Andrews. We now seek to build on this success by bringing together those active in developing statistical ecology methodology in the UK, to form a world-leading coordinated team. In this way, we can bring the best expertise to bear to solve key problems facing ecologists and wildlife resource managers, and maintain the UK's record of internationally-leading research in the field. We will build on the experience gained within NCSE for working effectively on joint projects within a geographically distributed Centre. This will involve regular meetings supplemented by video- and tele-conferences as well as video-seminars. Annual workshops will bring all members of the Centre together for research and learning. The planned research will span six broad themes, covering biodiversity monitoring, spatial and spatio-temporal models for ecological communities, stochastic models for population dynamics, movement models, and overarching new statistical methods and diagnostics. Current perspectives regarding loss of biodiversity are largely driven by results from the monitoring of specific sites, which typically are not representative. Further, it is usually assumed that individuals of all species are equally detectable, which is far from the case in most surveys. The research of one of the themes will develop indices for use in quantifying regional trends. In another theme, methods will be developed to describe the non-linear dynamics that are a feature of forest insects and zooplankton populations. Only recently have statistical methods been devised which take account of features such as animal movement and the interactions between different species; this is important for instance in determining the development of coral reefs. New methods developed by the grant investigators are allowing for greater complexity in stochastic models, which properly account for randomness. This is especially important when populations reduce in size. Common to the themes of the proposal is increasing realism in modelling, combined with the ability to match these developments with appropriate analytical, computational and inferential tools.The Centre will be run by an Executive Committee, chaired by the Director, which will meet regularly (usually by videoconference). The committee will select the Director annually from its members. An International Advisory Panel will provide the wider perspective. The new Centre will have post-doctoral research assistants at the Universities of Bath, Kent, Sheffield and St Andrews. In addition there will be annual appointments of cohorts of research students throughout the Centre, through which we will seek to maintain research activity in all six research themes. The Centre will link 8 university departments and 5 external agencies. We will endeavour to recruit and train the next generation of researchers in statistical ecology, foster the development of user-friendly computer software to ensure that the methods developed are readily available to the community, train scientists from the user community through a wide range of workshops, and provide a forum for our researchers to interact with the international community through the series of International Statistical Ecology Conferences, instigated in 2008 and organised by NCSE. The research of the Centre is timely and vitally important. It will ensure that national and international decisions regarding pressing contemporary issues, such as the effects of anthropogenic changes on the environment, as well as those of climate change and alternative energy generation, are made using the best possible science.

Planned Impact

We anticipate that the impact of our research will be realised in a number of ways: 1. Continuing publications in key statistics and ecology journals, ensuring impact within the user community, including consultant statisticians, research ecologists, natural resource managers, conservation managers, ecological modellers and impact assessment consultants. 2. Continuing development of computer software, to ensure that the methods developed are readily available to the user community. 3. Accessibility to the user community will be enhanced by publishing further books on the methods we develop. 4. The Research Unit for Wildlife Population Assessment (http://www.ruwpa.mcs.st-and.ac.uk/) is a unit within NCSE that carries out contract work, specialising in helping other organisations to implement the methods we develop. This provides us with a fund-generating way to make our new methodologies available to the user community. 5. We will continue to offer training workshops to the user community, and will expand the range of workshops offered. 6. We will continue to communicate work to a wider audience. Activities since NCSE was created in 2005 include development of a whale survey game for teenagers with EPSRC Partnerships for Public Understanding funding and a prize of a whale sightings cruise; three articles in the RSS Significance magazine's Darwin Anniversary Special Issue in 2009; delivery of an annual Mathematics Masterclass to top mathematics pupils in schools from Tayside. 7. We will continue to advise government agencies, international commissions and NGOs on a range of issues involving wildlife assessment, such as the impact of climate change and of large-scale windfarms on populations and communities. 8. We shall use the new methods we develop to analyse a range of valuable historical data sets. By this means we will be able to extract the vital information that they contain, and in particular be able to investigate possible time trends that may be the result of anthropogenic changes. Thus our work will inform the debate on such changes, and on the need for effective action to ameliorate the effects of change.

Publications

10 25 50
publication icon
Besbeas P (2014) Goodness-of-fit of integrated population models using calibrated simulation in Methods in Ecology and Evolution

publication icon
Blackwell P (2016) Exact Bayesian inference for animal movement in continuous time in Methods in Ecology and Evolution

publication icon
Borchers DL (2015) A Unifying Model for Capture-Recapture and Distance Sampling Surveys of Wildlife Populations. in Journal of the American Statistical Association

publication icon
Buckland S (2015) Model-Based Distance Sampling in Journal of Agricultural, Biological, and Environmental Statistics

publication icon
Buckland S (2017) Measuring temporal trends in biodiversity in AStA Advances in Statistical Analysis

 
Description Improved methods for assessing trends in abundance of animal populations. This allows organisations to assess the effects of management regimes, harvesting, habitat changes, climate change, etc.
Methods to quantify turnover in biodiversity were developed. These allow for example changes in communities due to climate change to be quantified.
Exploitation Route Our methods for spatially-explicit capture-recapture methods, for biodiversity monitoring, for line transect sampling, for movement modelling, for generalized additive modelling are already in wide use. We expect to extend this to other areas of research activity.
Sectors Aerospace, Defence and Marine,Agriculture, Food and Drink,Education,Energy,Environment,Healthcare

URL http://ncse.org.uk/
 
Description Spatially-explicit capture-recapture methods are now in wide use to estimate abundance of many species. Our methods are used by Defra, CEH, SNH, Natural England and a range of other organisations to quantify biodiversity trends, and how they vary spatially. CEH and Butterfly Conservation use our methods to quantify trends. Additive models are widely used in statistics; developments part-funded by this grant are now implemented in R. Our movement models were developed in response to improved technology for monitoring animal movement, and are being used in a number of studies.
First Year Of Impact 2012
Sector Aerospace, Defence and Marine,Education,Energy,Environment
Impact Types Cultural,Societal,Economic,Policy & public services

 
Title admbsecr 
Description Analysis of spatially explicit capture-recapture (SECR) data in AD Model Builder and R 
Type Of Technology Software 
Year Produced 2014 
Open Source License? Yes  
Impact Traditional capture-recapture approaches to estimating animal abundance or density ignore an obvious spatial component of capture probability; organisms close to traps are more likely to be captured than those that are far away. Explicitly accounting for an individual's location provides additional information from which to infer animal density. Spatially explicit capture-recapture (SECR) methods have been developed for this purpose. An advantage of these over traditional capture-recapture methodology is that they allow for animal density estimation using passive detectors (e.g., cameras or microphones) over a single sampling occasion. AD Model Builder (ADMB) is a statisical software package most widely used for nonlinear modelling, and is well suited to the implementation of maximum likelihood SECR methods. Although growing in popularity since becoming freely available, open-source software in 2008, ADMB is used by a minority of statisticians and ecologists, who, in general, are far more comfortable with the popular programming language and software environment R. The aim of admbsecr is to enable a user is able to fit SECR models that incorporate additional spatial information, using ADMB to fit the model and return the results to the R session. 
URL https://github.com/b-steve/admbsecr
 
Title mgcv: Mixed GAM Computation Vehicle with GCV/AIC/REML smoothness estimation 
Description Routines for GAMs and other generalized ridge regression with multiple smoothing parameter selection by GCV, REML or UBRE/AIC. Also GAMMs. Includes a gam() function. Major upgrade to models well beyond univariate exponential family. 
Type Of Technology Software 
Year Produced 2014 
Open Source License? Yes  
Impact This software package is a new release of one of a handful of 'recommended' packages supplied with the base distribution of R. Previous versions have been widely used in a wide range of applications, particularly in ecology and natural resource management, as well as medicine, epidemiology and economics. For example the energy company EDF use the methods for electricity load forecasting. In addition the package is currently used by 102 other software packages for R. For example the underlying fitting methods are sufficiently general that they can be efficiently leveraged for functional data analysis, as in the 'refund' package. 
URL http://cran.r-project.org/web/packages/mgcv/
 
Title mvnfast: Fast multivariate normal methods 
Description The package provides computationally efficient tools related to the multivariate normal distribution. The main functionalities are: simulating multivariate normal random vectors, evaluating multivariate normal densities and Mahalanobis distances. These tools are very efficient thanks to the use of C++ code and of the OpenMP API. 
Type Of Technology Software 
Year Produced 2014 
Open Source License? Yes  
Impact Too early to say 
URL http://cran.r-project.org/web/packages/mvnfast/
 
Title scam: Shape constrained additive models 
Description Routines for generalized additive modelling under shape constraints on the component functions of the linear predictor. Models can contain multiple shape constrained (univariate and/or bivariate) and unconstrained terms. The routines of mgcv(gam) package are used for setting up the model matrix, printing and plotting the results. Penalized likelihood maximization based on Newton-Raphson method is used to fit a model with multiple smoothing parameter selection by GCV or UBRE/AIC. 
Type Of Technology Software 
Year Produced 2013 
Open Source License? Yes  
Impact Too early to say 
URL http://cran.r-project.org/web/packages/scam/index.html
 
Title synlik: Synthetic Likelihood methods for intractable likelihoods 
Description Framework to perform synthetic likelihood inference for models where the likelihood function is unavailable or intractable. 
Type Of Technology Software 
Year Produced 2013 
Open Source License? Yes  
Impact Too early to say 
URL http://cran.r-project.org/web/packages/synlik/index.html