D-Track-A semi-automatic 3D video-tracking technique to analyse movements and routines of aquatic animals with application to captive dolphins.

Patrícia Rachinas-Lopes,Ricardo Ribeiro,Rui M Costa,Manuel E Dos Santos

doi:10.1371/journal.pone.0201614

Patrícia Rachinas-Lopes, Ricardo Ribeiro + Show 2 more

Open Access

https://doi.org/10.1371/journal.pone.0201614

Copy DOI

Abstract

Scoring and tracking animal movements manually is a time consuming and subjective process, susceptible to errors due to fatigue. Automated and semi-automated video-based tracking methods have been developed to overcome the errors and biases of manual analyses. In this manuscript we present D-Track, an open-source semi-automatic tracking system able to quantify the 3D trajectories of dolphins, non-invasively, in the water. This software produces a three-dimensional reconstruction of the pool and tracks the animal at different depths, using standard cameras. D-Track allows the determination of spatial preferences of the animals, their speed and its variations, and the identification of behavioural routines. We tested the system with two captive dolphins during different periods of the day. Both animals spent around 85% of the time at the surface of the Deep Area of their pool (5-meters depth). Both dolphins showed a stable average speed throughout 31 sessions, with slow speeds predominant (maximum 1.7 ms-1). Circular swimming was highly variable, with significant differences in the size and duration of the “circles”, between animals, within-animals and across sessions. The D-Track system is a novel tool to study the behaviour of aquatic animals, and it represents a convenient and inexpensive solution for laboratories and marine parks to monitor the preferences and routines of their animals.

Highlights

Ethologists and other behavioural scientists have traditionally scored animal movements in experimental settings by hand. This low cost, traditional technique [1,2,3], has the drawback of being a very time consuming process and susceptible to errors due to fatigue, drift and subjectivity [3,4]. To overcome these problems, automated video-based tracking methods have been developed over the years in controlled environments to explore and extract the detailed information that is contained in videos of moving animals (e.g., Kabra et al studies [5])
To overcome the obstacles to animal tracking in a large pool with varying depths, we developed D-Track, a software that tracks aquatic animals in their habitat, without disturbance
Bottlenose dolphins in captivity are held in habitats that are obviously much smaller and less complex than their natural environments, which inevitably leads to behavioural alterations

Summary

Introduction

Ethologists and other behavioural scientists have traditionally scored animal movements in experimental settings by hand. This low cost, traditional technique [1,2,3], has the drawback of being a very time consuming process and susceptible to errors due to fatigue, drift and subjectivity [3,4]. To overcome these problems, automated video-based tracking methods have been developed over the years in controlled environments to explore and extract the detailed information that is contained in videos of moving animals (e.g., Kabra et al studies [5]).

Methods

Results

Conclusion