discussion / Biologging  / 6 September 2018

Mataki-LITE: lessons learned from petrel tracking (Part I: GPS loggers)

Dear colleagues,

I recently used Mataki-LITE GPS tags (Debug Innovations, Cambridge, UK) and I thought it would be helpful to share the set-up and configuration I used, as well as lessons learned. This is a lot of text so I'll publish it as replies. 

Deployment: From April 10 to 25, I partnered with Grupo Jaragua, a Dominican conservation NGO, to track the foraging movements of Black-capped Petrel (Diablotin, Pterodroma hasitata) through a conservation grant by the Neotropical Bird Club. I powered the Mataki-LITE with a 150mAh lithium-ion battery (TinyCircuits, Akron, USA; 3.8g) and waterproofed them with light-weight heatshrink tubing (F4(Z), Heatshrink.com, Ogden, USA). The total mass of the waterproofed devices were ~ 8g. I attached them at the base of the tail with 2-part epoxy and 3 strips of TESA cloth tape, to the underside of the 4 central rectrices. I finally secured the whole with a 0.2mm zip-tie. We deployed LITE loggers on 9 petrels. The handling process lasted less than 15 minutes and all birds were returned to their burrows.  

Logging schedule: To avoid draining the battery underground, I programmed the devices to turn on 24h after deployment, when the birds were back at sea. Based on previous results, we expected foraging trips to last 10-15 days thus GPS loggers were set to record GPS locations every 30 minutes in normal power mode to allow for a battery life of ca. 21 days. Upon return to the nest, loggers were set to upload tracking data to base-stations placed within 200m of clusters of burrows (See base-stations details in this other Wildlabs post).

Logging script: Mataki-LITE loggers are coded in BASIC language and I adapted a sample script provided by Debug Innovations. The script is attached (Appendix A). Briefly, I wrote the logging script to optimize power consumption by the GPS logger in two ways. First, I created two geofences (of respective 25 km and 150 km radius from the colony) to control the frequency of UHF communication “heartbeats” (which let the base-station know that a tag is near): a heartbeat was sent every 2 minutes when inside the 25 km geofence (based on the coordinates of the last recorded GPS location), every 10 minutes when inside the 150 km geofence, and no heartbeat was sent when more than 150 km away from the colony (Table B.1). Then, when the battery power decreased below a threshold of 3.5 V, a simple “low voltage” loop reduced the GPS logging schedule from every 30 minutes to every 180 minutes, and the frequency of UHF heartbeats was scheduled to every 10 minutes disregarding of the distance to the colony. Stationary tests were performed before deployment to debug the script and estimate battery life.

 

appendix_a_logging_script.pdf appendix_b_base-station.pdf


Results:  Of the 9 petrels equipped with LITE loggers, only 3 reported to a base-station during the deployment period. The 3 birds for which data were recovered were tracked for a single provisioning trip lasting 8 to 11 days.

Although they had been scheduled to do so, the GPS loggers did not record location data continuously (48 locations per day): as a result, an average of 12.5 locations per day were collected (Bird1: 15.9 locations per day, range 4-26; Bird2: 12.0 locations per day, range 6-21; Bird3: 12.4 locations per day, range 1-27). When locations were recorded, they were of good quality (HDOP < 9) and no locations were filtered out because of poor spatial precision or unrealistic speed.

Discussion:

- Data recording rate:

1) Attachment location: It is likely that our attachment location on the underside of the tail had a negative impact on the recording rate. To avoid rectrix breakage and undue bouncing when flapping, we decided to center the loggers’ weight as close to the base of the tail as possible: loggers placed on top of the tail would have covered the uropygial gland therefore we chose to attach them on the underside, between the undertail coverts and the rectrices, with the GPS antenna pointing up.

2) GPS fix time: The location of GPS satellites at the time of recording might have prevented successful logging. Indeed, while GPS devices used by the general public have an ample supply of power that allows them to always be in communication with the network’s satellites (“hot start”), low-power wildlife tracking devices only activate GPS communication at the time of logging (“cold start”). Thus the amount of time required for a successful GPS logging (which is limited a priori to optimize power consumption) is affected by the location of GPS satellites at the moment of the cold start. The 4-minutes limit on GPS fix might have been to short.

- Data recovery rate: Several possibilities exist for the lack of transmission:

1) Waterproofing failure: tests were performed to ensure that the waterproofing technique used in this study was resistant enough to sustain repeated immersions.

2) Death of tracked bird: The chicks of all tracked petrels were observed alive and in good condition in two visits in June and July, suggesting that tracked birds did not die during the tracking period.

3) Early loss of logger: Though the type of tape attachment I used in this study (which I reinforced with epoxy and a zip-tie) has been successful with other seabird, it is possible that rectrices broke off. I chose this technique instead of a dorsal attachment to keep the loggers from rubbing against the roof of rocky burrows, and to avoid the possibility of entanglement in hanging roots.

4) Early drain of logger battery:  Remote-download GPS loggers such as Mataki-LITE have two main drains of power: 1) activating the GPS module to listen to communications from the network’s satellites, and 2) using the UHF radio to communicate with the base-station and, later, upload tracking data. I had written the logging script to optimize battery use through a geofence and a low-voltage mode. Based on recovered tracking data, it appears that none of the functioning loggers had entered the low-voltage mode during the ca 10 days at-sea though two of them had battery levels below the low-voltage threshold at the time of uploading data to the base-station (Table B.2). Since logger battery levels are only recorded when they send a heartbeat to the base station, it is unclear if these low battery levels had been reached earlier during the tracking period without triggering the low-voltage mode, had triggered the low-voltage mode but GPS data were not recorded, or resulted from a power drain due to radio communications.

5) Not enough time within range of base-station: It is possible that the loggers that did not transmit tracking data were not able to contact their respective base-stations despite having sufficient battery voltages. Radar surveys and camera-trapping at breeding colonies showed that Black-capped Petrels may reach speeds up to 70km/h when arriving at colonies and enter their burrows instantly. Upon leaving the nest, petrels stay <5 minutes outside the burrow. Pre-deployment trials showed that loggers placed at a burrow’s entrance were able to communicate with base-stations but loggers placed inside a burrow’s tunnel were not. The 2-minute heartbeat activated inside the 25-km geofence to optimize power should have been sufficient to upload tracking data but loggers that had failed to trigger the 25-km geofence (or that had already entered the low-voltage mode) were less likely to be within range of the base-station (~200 m) for long enough to send a successful heartbeat.

Recommendations:

Mataki-LITE tags have a steep learning curve but they very adaptabile and cheap (for remote-download tags) so I hope these lessons-learned will prove useful for everybody. 

- Future deployments may benefit from an attachment above the tail, ensuring that the uropygial gland is not covered; to avoid undue stress to the rectrices, the center of gravity of the logger should be as close to the rump as possible. Try dorsal deployments on birds nesting in large sod burrows.

- Although it is possible to extend the time period allocated to successfully log a GPS location, this would drastically increase power consumption. Therefore, try to improve attachment location before extending the GPS “ON time”.

- Simultaneous logging of GPS locations and battery levels, and GPS logging frequency may have to be decreased to optimize power consumption.

- Wider geofence radius and a higher heartbeat frequency in low-power mode.

Feel free to start the conversation and share your own scripts, deployment techniques and fixes!

Thanks,

Yvan