Jaguar movement behavior: using trajectories and association rule mining algorithms to unveil behavioral states and social interactions

Abstract

Animal movement data are widely collected with devices such as sensors and collars, increasing the ability of researchers to monitor animal movement and providing information about animal behavioral patterns. Animal behavior is used as a basis for understanding the relationship between animals and the environment and for guiding decision-making by researchers and public agencies about environmental preservation and conservation actions. Animal movement and behavior are widely studied with a focus on identifying behavioral patterns, such as, animal group formation, the distance between animals and their home range. However, we observed a lack of research proposing a unified solution that aggregates resources for analyses of individual animal behavior and of social interactions between animals. The primary scientific contribution of this work is to present a framework that uses trajectory analysis and association rule mining [Jaiswal and Agarwal, 2012] to provide statistical measures of correlation and dependence to determine the relationship level between animals, their social interactions, and their interactions with other environmental factors based on their individual behavior and movement data. We demonstrate the usefulness of the framework by applying it to movement data from jaguars in the Pantanal, Brazil. This allowed us to describe jaguar behavior, social interactions among jaguars and their behavior in different landscapes, thus providing a highly detailed investigation of jaguar movement decisions at the fine scale.

Fig 1. Framework steps.

The framework is composed of three steps: (A) identify the behaviors of individual animals; (B) identify the influence of neighboring animals on the animal occurrences; and (C) identify the correlation between animal behavior and environmental factors. The framework receives animal movement data and environmental factors as input. First, the animal movement data are analyzed to identify the individual behavior of the animals. Then the environmental factors that occurred close to the animals are identified. In the last step, the interactions between animals and between animals and other environmental factors are identified. Each step provides information about the animals and their interactions that are presented in tables, graphs, or maps.

Fig 2. Behavioral state matrix of the jaguar Picolé.

The matrix shows the time intervals that the jaguar Picolé spent foraging, resting or in transit over the study period. Picolé presented a higher frequency of the resting state in the first days of February, and on days 10, 11, 16 and 17, there was a higher frequency of foraging. This distribution identifies the individual behaviors of the animal in its daily activities, as well as the changes in animal behavior over the days and months of the year.

Fig 3. Home range maps to the jaguars Caiman, Dale and Fera.

The graph shows the use of space by these jaguars over the months of February and March 2015. The kernel function [38] was used to calculate the home range of the jaguars in space and time and to determine the minimum area in which each animal had a high probability of being located. The results indicated that Caiman and Dale [A] and Caiman and Fera [B] presented home range overlap in February 2015. In March 2015, Caiman distanced itself from Dale and approached Picolé [C]. The home range overlap provides information about the social interactions between jaguars.

Fig 4. Home range maps of the jaguars Dale and Fera.

The graphs show the use of space by these jaguars over the months of February, March, April, and May 2015. The kernel function [38] was also used to calculate the home range of the jaguars in space and time and to determine the minimum area in which each animal had a high probability of being located. Dale and Fera presented home range overlap in all analyzed months in 2015: February [A], March [B], April [C] and May [D]. The home range overlap indicates a high frequency of interaction between these two jaguars.