Fitting Probability Distributions to Animal Movement Trajectories: Using Artificial Neural Networks to Link Distance, Resources, and Memory

Abstract

Animal movement paths are often thought of as a confluence of behavioral processes and landscape patterns. Yet it has proven difficult to develop frameworks for analyzing animal movement that can test these interactions. Here we describe a novel method for fitting movement models to data that can incorporate diverse aspects of landscapes and behavior. Using data from five elk (Cervus canadensis) reintroduced to central Ontario, we employed artificial neural networks to estimate movement probability kernels as functions of three landscape-behavioral processes. These consisted of measures of the animals' response to the physical spatial structure of the landscape, the spatial variability in resources, and memory of previously visited locations. The results support the view that animal movement results from interactions among elements of landscape structure and behavior, motivating context-dependent movement probabilities, rather than from successive realizations of static distributions, as some traditional models of movement and resource selection assume. Flexible, nonlinear models may thus prove useful in understanding the mechanisms controlling animal movement patterns.

Mapping behavioral landscapes for animal movement: a finite mixture modeling approach

“…This approach allows for nonlinear models, but the forms of nonlinearity are restricted to exponential and logistic functions, and the movement response is based on a single probability density. Recently, animal movement models that combine movement, resource selection, and home range of an animal have been developed (e.g., Dalziel et al 2008, Johnson et al 2008, Forester et al 2009). Most of these models incorporate covariates in a probability density function of bivariate locations rather than turn angles and move length, and it is not always straightforward to assess the effect of landscape features.…”

“…Tracey et al (2011) developed a semiparametric regression approach using neural networks to relate the movement model parameters of probability distributions to covariates associated with landscape features without having to discretize space. While fully connected feedforward neural networks have proven useful in modeling animal movement (Dalziel et al 2008, Tracey et al 2011, neural network parameters are not directly interpretable in terms of animal behavior, so inference can be made only about the relationship between movement and landscape features in terms of the responses produced by the network. Other network-based approaches to modeling movement behavior are possible.…”

A framework for generating and analyzing movement paths on ecological landscapes

“…We have not dealt in any detail with the problem of movement mode identification that is the key to the deconstruction component other than stressing that the frequency with which data are collected limits our ability to identify CAMs and their underlying FMEs (12). To undertake such an analysis is not a trivial problem: it requires computationally complex methods that can only be successfully applied to high-resolution data, but a start has been made (33)(34)(35).…”

“…Furthermore, stochastic differential equation methods (42) can be used to construct vector fields from data on the contemporaneous movement of many individuals, and then thin plate splines can be used to fit potential fields to these vector fields to identify regions on the landscape that are either repealing or attracting the individuals at a particular time of day (2). Recently, techniques new to the field of movement ecology, such as wavelet analysis (33) and artificial neural networks (34), are being applied to obtain insights into the effects that the internal and environmental states of a system have on movement paths. Beyond these, as the resolution of movement and landscape data improves dramatically over the next decade, we should expect to see the application of state-space estimation methods (32) that can take advantage of formulations such as ours, because our formulation permits the inclusion of detailed landscape information.…”

Feasibility of Using Proximity Tags to Locate Female Lion Panthera leo Kills

“…In an early paper, Siniff and Jessen (1969) proposed a home-range simulation model in which individuals biased their movements toward locations that they had previously visited. More recently, Tan et al (2001Tan et al ( , 2002, building on earlier work by Sapozhnikov (1994Sapozhnikov ( , 1998 and Dalziel et al (2008), have analyzed the behavior of so-called ''self-attracting'' random walks in which individuals display an increased probability of moving toward previously visited locations. Their analyses showed that movement models of this kind result in individuals developing quasi-stable home ranges: over short timescales, the movements of an individual are largely confined to some characteristic area (i.e., a home range), whereas on longer timescales the center of the individual's home range drifts randomly around the landscape.…”