A generalized residual technique for analysing complex movement models using earth mover's distance

@article{Potts2014AGR,
  title={A generalized residual technique for analysing complex movement models using earth mover's distance},
  author={Jonathan R. Potts and Marie Auger-M'eth'e and Karl S. Mokross and Mark A. Lewis},
  journal={Methods in Ecology and Evolution},
  year={2014},
  volume={5}
}
Complex systems of moving and interacting objects are ubiquitous in the natural and social sciences. Predicting their behaviour often requires models that mimic these systems with sufficient accuracy, while accounting for their inherent stochasticity. Although tools exist to determine which of a set of candidate models is best relative to the others, there is currently no generic goodness‐of‐fit framework for testing how close the best model is to the real complex stochastic system. We propose… 

Figures from this paper

Making sense of ultrahigh‐resolution movement data: A new algorithm for inferring sites of interest
TLDR
The algorithm is given an efficient method for turning a long, high‐resolution movement path into a schematic representation of broadscale decisions, allowing a direct link to existing point‐to‐point analysis techniques such as optimal foraging theory.
Beyond resource selection: emergent spatio-temporal distributions from animal movements and stigmergent interactions
A principal concern of ecological research is to unveil the causes behind observed spatiotemporal distributions of species. A key tactic is to correlate observed locations with environmental
Integrated step selection analysis: bridging the gap between resource selection and animal movement
TLDR
iSSA relies on simultaneously estimating movement and resource selection parameters, thus allowing simple likelihood‐based inference of resource selection within a mechanistic movement model and demonstrates the utility of iSSA as a general, flexible and user‐friendly approach for both evaluating a variety of ecological hypotheses, and predicting future ecological patterns.
Incorporating periodic variability in hidden Markov models for animal movement
TLDR
The results suggest that incorporating additional structure in statistical models of movement can allow more accurate assessment of appropriate model complexity and reduce the selected number of movement states closer to a biologically interpretable level, although there is further room for improvement here.
Characterising menotactic behaviours in movement data using hidden Markov models
TLDR
This poster presents a probabilistic procedure to estimate the temperature and salinity of the Arctic Ocean using a £20,000 (US$25,000) net worth estimation method.
The Conditionally Autoregressive Hidden Markov Model (CarHMM): Inferring Behavioural States from Animal Tracking Data Exhibiting Conditional Autocorrelation
TLDR
The conditionally autoregressive hidden Markov model (CarHMM) is motivated and developed, a generalization of the HMM designed specifically to handle conditional autocorrelation, and guidelines for all stages of an analysis using either an HMM or CarHMM are provided.
Across atoms to crossing continents: application of similarity measures to biological location data
TLDR
Generally-used similarity measures are reviewed jointly with several measures less used in biological applications and those which can be useful in noise elimination or, conversely, are sensitive to spatiotemporal scales to highlight how similarity measures contrast regarding computational complexity.
The “edge effect” phenomenon: deriving population abundance patterns from individual animal movement decisions
TLDR
The distance over which animals make their decisions to move between habitats turns out to be a key factor in quantifying the magnitude of certain observed edge effects, deriving analytic expressions describing oft-observed population abundance patterns from a model of movement decisions near edges.
Solving the sample size problem for resource selection functions
TLDR
The results demonstrate that the most biologically relevant effects on the utilization distribution can often be estimated with much fewer than M=30 animals, and it is argued that these equations should be a mandatory component for all future RSF studies.
Solving the Sample Size Problem for Resource Selection Analysis
TLDR
It is argued that random sampling of background data violates the underlying mathematics of RSA, leading to incorrect values for necessary M and N and potentially incorrect RSA model outputs, and should be a mandatory component for all future RSA studies.
...
...

References

SHOWING 1-10 OF 70 REFERENCES
A unifying framework for quantifying the nature of animal interactions
TLDR
This paper focuses on combining the various mechanistic models of territorial interactions in the literature with step selection functions, by incorporating interactions into the step selection framework and demonstrating how to derive territorial patterns from the resulting models.
Inferring spatial memory and spatiotemporal scaling from GPS data: comparing red deer Cervus elaphus movements with simulation models.
TLDR
This study demonstrates how inference regarding memory effects and a hierarchical pattern of space use can be derived from analysis of GPS data.
Predicting local and non-local effects of resources on animal space use using a mechanistic step selection model
TLDR
This work proposes a mechanistic model where the non-local effect of resources naturally emerges from the local movement processes, by taking into account the relative utility of both the habitat where the animal currently resides and that of where it is moving.
EXTRACTING MORE OUT OF RELOCATION DATA: BUILDING MOVEMENT MODELS AS MIXTURES OF RANDOM WALKS
TLDR
A framework for fitting multiple random walks to animal move- ment paths consisting of ordered sets of step lengths and turning angles, which allows for identification of different movement states using several properties of observed paths and lead naturally to the formulation of movement models.
A SPATIALLY EXPLICIT HABITAT SELECTION MODEL INCORPORATING HOME RANGE BEHAVIOR
TLDR
This paper extends a previous model by formulating the probability of selecting a habitat as a function of its distance from the animal's current location and home range center, leading to more parsimonious models when applied to a koala radio-tracking data set from eastern Australia.
Mechanistic home range models capture spatial patterns and dynamics of coyote territories in Yellowstone
TLDR
This analysis indicates that the spatial arrangement of coyote territories in Yellowstone is determined by the spatial distribution of prey resources and an avoidance response to the presence of neighbouring packs, and shows how the fitted mechanistic home range model can be used to correctly predict observed shifts in the patterns of Coyote space-use in response to perturbation.
Step selection techniques uncover the environmental predictors of space use patterns in flocks of Amazonian birds
TLDR
It is shown that movement decisions are significantly influenced by canopy height and topography, but depletion and renewal of resources do not appear to affect movement significantly, and a mechanistic model is constructed from which predicted utilization distributions (home ranges) of flocks are derived.
Accounting for animal movement in estimation of resource selection functions: sampling and data analysis.
TLDR
A model for step selection functions (SSF) that is composed of a resource-independent movement kernel and a resource selection function (RSF) is proposed and it is suggested that distance always be included as a covariate in SSF analyses.
Differentiating between the L\'evy and the area-restricted search strategies
TLDR
A method consisting of likelihood functions, including a hidden Markov model, and associated statistical measures that assess the support for each strategy is presented, which can differentiate between the two search strategies over a range of parameter values.
ROBUST STATE-SPACE MODELING OF ANIMAL MOVEMENT DATA
TLDR
It is shown how known information regarding error distributions can be used to improve inference of the underlying process(es) and demonstrated that the frame- work provides a powerful and flexible method for fitting different behavioral models to tracking data.
...
...