Developing a nested, multiscale framework for modelling animal movement processes from long-term tracking data


Models of animal movements allow us to predict change in the distribution or migratory patterns of wildlife, particularly in the presence of climate change, other anthropogenic disturbance, or introduction to a novel environment. However, modelling movement for animals with high levels of cognition is not simple because they make decisions at multiple temporal and spatial scales. A primary broad-scale movement decision will constrain the options available for finer-scale (shorter-term) choices. The processes manifest at one scale may differ from, or even oppose, those at another scale. For example, in a ‘dry season’, an animal may attempt to establish its home range near a river, but on a daily time scale it may try to avoid predators by visiting the river-edge only briefly. Dependencies between these scales mean that it is difficult to disentangle the drivers of decision-making, and a complex mathematical framework is required in order to adequately represent the ‘movement system’. This framework is currently lacking, and consequently, we are unable to predict the movement of animals in the wild over the extended timescales that multi-scale processes can be expected to occur.

This project aims to develop statistical models of the hierarchy of different spatial and temporal scales over which movement decisions are made. Specifically, we will:
1. Formulate the fundamental rules required to model movement and decision-making at multiple scales
2. Develop statistical methods for parameterising these behavioural rules from field data
3. Use these fully fitted models to predict movement across scales
4. Show, using case studies, how these methods generalise naturally to distinctly different species and natural histories


• Create a mechanistic model of multiple processes operating at nested scales, and represent different scales using different levels in the model, which are linked through stochastic elements representing spatial behaviour at each scale.
• Extract the basic behavioural rules that relate to each scale through the original model and test the ability of these models to predict movement across scales over specific landscapes.
• Test the generality of this approach by applying these methods to two different data sets.
1) Seabird Populations on the Isle of May
Seabird populations will choose a home range in which to raise their chicks, centering their movement for the period of time in which they are providing parental care. Within the breeding season, other factors which impact the spatial location of the birds’ foraging trips are nested within the restricted range imposed by this large-scale decision. However, this restriction does not apply throughout the year, and so movement processes vary seasonally. The Isle of May Seabird observatory has been used as a base of observation for the foraging and migratory movement of several species, including Atlantic puffins, razorbills and European shags.
2) Elk Reintroduction in Ontario, Canada
Reintroduction projects generate valuable opportunities to understand how animals explore novel environments, and construct ‘memory maps’ of the landscape that enable establishment of a stabilised spatial distribution. The processes operating on a reintroduced group are, at first, very localised but as individuals spread spatially through the environment, processes operating at larger scales (such as memory) begin to impact their movement. By predicting the movement pattern of a population prior to its establishment of a stable distribution within a novel environment, the likely outcome of proposed reintroduction projects can be evaluated. This substantial data set is idea as all 120 individuals released into the region were tracked for 4 years over 40,000 km2.
This is a strongly interdisciplinary project, that will explore the use of groundbreakingly efficient methods of analysis for flexible and realistically complex spatial and spatio-temporal modelling. Outcome should have impact both within the statistical and the ecological research community and will result in influential publications in both areas. The project will, for the first time, develop statistical models for animal movement at different scales through integrated spatio-temporal modelling. More broadly, this is relevant to research in areas such as city planning, the behaviour of human populations within nested social networks and any area where the links between the hierarchies of micro and macro behaviours need to be investigated from raw and partial data.

Project Timeline

Year 1

Statistical Training (APTS and SMSTC); model development.

Year 2

Application of model to proposed systems; data exploration.

Year 3

Data analysis; conference attendance; thesis write up.

Year 3.5

Data analysis; conference attendance; thesis write up.

& Skills

For further statistical training, the student will attend APTS statistical training courses to obtain the skills necessary to undertake the required statistical analyses. More generally, this project will generate unique training opportunities at the intersection of mathematics and ecology. The student will be introduced to the diverse networks of the supervisors, each of whom are recognised experts in their respective fields. These networks will enable the appointed student to access a range of different data sets relating to animal movement, and critically, connect with the biological expertise required in understanding the different movement systems. CEH offers training opportunities on a range of topics from data collection methods to statistical modelling, allowing the student to benefit from expertise outside of the University of Glasgow. The student will integrate with an internationally conspicuous concentration of research activity in movement ecology led by Glasgow researchers that is reflected by the Special Interest Group in Spatial Ecology – stimulated through a number of new faculty appointments in this area over recent years in both Dept of Mathematics and Statistics and the Institute of Biodiversity, Animal Health and Comparative Medicine.

References & further reading

Avgar, T., Deardon, R., and Fryxell, J.M. (2013). An empirically parameterized individual based model of animal movement, perception, and memory. Ecol. Model. 251, 158-172.
Compton, B.W., Rhymer, J.M., and McCollough, M. (2002). Habitat selection by wood turtles (Clemmys insculpta): an application of paired logistic regression. Ecology 83, 833-843.
DeCesare, N.J., Hebblewhite, M., Schmiegelow, F., Hervieux, D., Mcdermid, G.J., Neufeld, L., Bradley, M., Whittington, J., Smith, K.G., Morgantini, L.E., Wheatley, M., and Musiani, M. (2012). Transcending scale dependence in identifying habitat with resource selection functions. Ecol. Appl. 22(4), 1068-1083.
Illian, J.B., Sarbye, S.H., Rue, H. (2012). A toolbox for fitting complex spatial point process models using integrated nested Laplace approximation (INLA). Ann. Appl. Stat. 6(4), 1499- 1530.
Oliveira-Santos, L.G.R., Forester, J.D., Piovezan, U., Tomas, W.M., and Fernandez, F.A.S. (2016). Incorporating animal spatial memory in step selection functions. J. Anim. Ecol. 85(2), 516- 524.
Paton, R.S., and Matthiopoulos, J. (2015). Defining the scale of habitat availability for models of habitat selection. Ecology 97(5), 1113-1122.
Riotte-Lambert, L., and Matthiopoulos, J. (2018). Communal and efficient movement routines can develop spontaneously through public information use. Behav. Ecol. 30(2), 408- 416.

Further Information

Applications: to apply for this PhD please use the url:

Apply Now