Establishing an Open-source Animal-tracking Analysis Platform for Archival Geolocators
For centuries, naturalists have wondered how small migratory birds move between breeding sites and wintering locations, but our ability to study songbird migration has been limited by the logistical challenge of tracking tiny birds that fly thousands of kilometers. In just the past few years, researchers have been experimenting with a new tracking device that can revolutionize our understanding of migration by determining the migration routes of individual song birds. This key innovation is called an electronic tracking device called a geolocation datalogger (or geologger), which has recently been miniaturized to a mass of 0.5 grams—small enough to deploy on birds weighing around 10 grams. Geologgers work by logging light levels at regular intervals such that one can determine the duration of the daylight period (day length) and the midpoint between sunrise and sunset (solar noon). With these two bits of information from each day the bird wears the tag, we can estimate latitude from day length and longitude from the time of solar noon.
Although this method is simple and intuitive, it does not allow for error estimation. Moreover, current analysis practices generally have low repeatability and are very sensitive to perturbations in the light-level data caused by shading particularly during twilight events (caused by environmental factors or bird behavior). More sophisticated analysis tools are available, but they are underused because they require special skills (i.e. familiarity with the R computing environment) and lack extensive documentation.
The goals of the Animal-Tracking Analysis Working Group are:
1) Increase accessibility of analysis tools by creating a user-friendly web interface that can perform simple on-line analysis of light level data and format the data for use in more sophisticated tools
2) Promote transparency and data-sharing among those using light-level geolocation data loggers via the web interface, through use of analytical methods in the public domain, and through use of the Movebank archive and other data repositories
3) Update and compare existing analytical tools and publish guidelines for their use.
Membership of the working group consists of programmers who have written movement analysis software, field biologists with real-world data to analyze, and administrators and data curators of ecological data archives. Our meetings will not only result in the best possible analysis software but will promote individual research projects and multi-project collaborations among the working group members.
Example of a migration track from a geologger mounted on a Painted Bunting. The blue crosses show locations generated from simple calculations based on day length and the time of solar noon. The red and yellow colors show the likely ranges for residency and movement based on an animal movement model within an analysis package called TripEstimation. The movement model generates a narrower migration path because it assumes there are limits to the extent that a bird can fly in a single day. Movement models can also account for land cover and habitat types to further improve estimations of migration paths.
UPDATE: October 2013
Publications and Other Products
Progress on Goal 1: We have established TAGS (the Totally Awesome Geolocator Service), which is a simple web interface that allows users to upload raw light level data, quickly edit the data where necessary, and get a first look at location estimates for the tracked animal. TAGS can output both coordinate data and a data format that will work with the more sophisticated analysis tools. TAGS output includes a complete record of edits made to the original data set. Thus, researchers can make this output available to ensure that analysis by other researchers can replicate the original results.
Progress on Goal 2: TAGS output is also formatted for upload into the Movebank archive, which will promote data sharing among scientists and the general public. We are currently working toward a tighter interface between TAGS and Movebank that will allow a more direct flow of data between the two services.
Progress on Goal 3: Our group seeks to improve methods for analyzing light-level geolocation data by incorporating animal movement models into a bayesian framework that generates location estimates. We have brought together biologists who use light-level data loggers as well as programmers and statisticians who have generated different analytical approaches to the analysis of light-level data. The meetings have spurred several new directions with regard to analysis techniques, and two new analysis packages for the R programming language are nearing finalization.
To help promote the use of these new analysis techniques we are drafting a manuscript that will analyze simulated light-level data from a theoretical tag assigned to a specific (known) movement path that represent the migration of a songbird. Our analyzes will help determine how well different analysis methods capture various features of the underlying movement path and will help readers decide what analysis method best suits their needs. The working title of the manuscript is "New tools for the analysis of light-level geolocation data to derive migration tracks: guidelines and comparisons based on real and simulated data".
TAGS Web tool: http://tags.animalmigration.org
More information about this project.
This work is supported by the National Center for Ecological Analysis and Synthesis, a Center funded by NSF (Grant #EF-0553768), the University of California, Santa Barbara, and the State of California.