Advances in the analysis of animal movement
Wednesday 9th November 2011
Queries: email: firstname.lastname@example.org
|Document downloads for IBS members.|
Join us now.
|10.00 - 10.30||Registration and coffee|
|10.30 - 11.15||A non-exhaustive overview of animal movement models. Roland Langrock (University of St Andrews)|
While the development of statistical methodology for animal movement data may have lagged behind the technological advancements in animal tracking, it is nevertheless also true that the last decade has seen substantial progress in terms of our ability to incorporate ecological realism in animal movement models. This tutorial gives an overview of some of the most popular approaches to modelling animal movement, including a discussion of their respective advantages and disadvantages. Topics that will be covered include random walks (correlated, biased & mixtures), Lévy walks, stochastic differential equations (with a focus on Ornstein-Uhlenbeck processes) and state-space models. Such models will be the focus of subsequent talks during this meeting.
|11.15 - 11.45||A tale of two case studiesRuth King (University of St Andrews)|
We will consider two separate case studies of animal telemetry data and discuss different model fitting algorithms for discrete time animal movement models (as discussed by Roland Langrock in the previous tutorial). The two case studies that we consider are (1) an individual grey seal track; and (2) multiple bison tracks. Model fitting in general is non-trivial due to the complexities of the data, which may include observation/measurement error, unequally spaced observation (in time), missing observations and multiple individual paths. As we shall see these case studies are very different in terms of the form of the data and models fitted to the data. Consequently, we will consider different model fitting techniques for each of the case studies:
1) Case study 1 – grey seals – Bayesian data augmentation approach (observation error, unequally spaced observations, centres of attraction, missing data);
2) Case study 2 – bison – classical hidden Markov model approach (multiple individuals, multi-state model).
In each case we will discuss the advantages and disadvantages of the different model fitting approaches (including areas of current/future research).
This work is joint work with Juan Morales (Universidad Nacional del Comahue, Argentina), Brett McClintock (National Marine Mammal Laboratory, USA), Roland Langrock, Len Thomas, Jason Matthiopoulos and Bernie McConnell (University of St Andrews)
|11.45 - 12.15||The use of GPS tracking data to infer foraging behaviour in seabirds Adam Butler (BioSS)|
Adam Butler (BioSS) and Ellie Owen (RSPB)
The populations of many seabird species are in decline. In order to conserve these species it is important to understand the areas in which birds forage (feed), and the environmental conditions that are associated with foraging. Direct data on the foraging behavior of seabirds are extremely difficult to collect, but data from electronic tags - which monitor the location of individual birds at regular intervals, and thereby provide information on direction, speed and (sometimes) depth - are now becoming widely available. Hidden Markov and state space models provide a natural framework for using data on these observed variables to draw inferences about the unobserved behavioral state. Tracking data for a relatively large number of birds have been collected during 2010 and 2011 at part of the FAME project (Future of the Atlantic Marine Environment), and this talk will focus upon using these data to draw inferences about the spatial distribution of foraging behavior.
|12.15 - 13.30||Lunch and AGM|
Sandwich lunch and Annual General Meeting of the British and Irish Region of the International Biometric Society
|13.30 - 14.00||Modelling animal movement in continuous time. Paul Blackwell (University of Sheffield)|
Observations of animal locations generally take place at discrete times, often equally spaced, but the underlying movement is inherently a continuous-time phenomenon. Modelling in continuous time avoids imposing an artificial time-scale on real patterns of movement behaviour, and gives a natural way of combining or comparing data that are on different time-scales, whether by accident or design.
In this talk I will describe a rich class of models in which, at any instant, an animal can be in one of a number of different states. These states can represent different behaviours, e.g. encamped and exploratory, as in Morales et al (2004), and/or different movement patterns within the same kind of behaviour, e.g. attraction to different feeding patches. Switching between states can be allowed to depend on time and on spatial covariates in a very flexible way.
Both simulation and statistical inference can be carried out for these models without any approximation or discretisation error, despite the complex feedback between location and behaviour. I will outline a Bayesian statistical approach that allows learning from the data not only about parameters but also about the number of different behavioural states present.
These ideas will be illustrated using data on wild boar movement, from GPS collars (courtesy of FERA) - and perhaps other examples, if time permits.
|14.00 - 14.30||Modelling movement in the third dimension: Diving in seals Jason Matthiopouos (St Andrews)|
|14.30 - 15.00||Testing for biological Levy flights: theory, empirical patterns and potential process|
David Sims (Marine Biological Association)
|15.00 - 15.30||Title to be announced Robin Freeman (Microsoft)|
|15.30 - 16.00||Tea|
|16.00 - 16.45||Understanding and Predicting Mammalian Space-Use: Recent Advances, Future Directions Paul Moorcroft|
Paul Moorcroft (Harvard University )
The coming of age of GPS telemetry, in conjunction with recent theoretical innovations for formulating quantitative descriptions of how different ecological forces and behavioral mechanisms shape patterns of animal space-use, has led to renewed interest and insight into animal home range patterns. This renaissance is likely to continue as result of on-going synergies between these empirical and theoretical advances. In this article I review key developments that have occurred over the past decade that are furthering our understanding of the ecology of animal home ranges. I then outline what I perceive as important future directions for furthering our ability to understand and predict mammalian home range patterns. Interesting directions for future research include: (1) improved insights into the environmental and social context of animal movement decisions and resulting patterns of space use; (2) quantifying the role of memory in animal movement decisions; (3) examining the relevance of these advances in our understanding of animal movement behavior and space-use to questions concerning the demography and abundance of animal populations.
|16.45 - 17.15||General Discussion|
Existing members can login below to view all site content. Lost password?
Other visitors might be interested to learn more about the benefits of membership.
|14 Nov 14||Mendelian Randomisation: Past Success and Future Challenges|
|17 Sep 14||Statistics for big data|
|06 Jul 14 - 11 Jul 14||International Biometric Conference 2014, Florence|
|04 Jun 14||Clinical trials: past, present and glorious future?|
|14 Nov 13||Statistical software interoperability|