Playing with Geolocators

Posted on Mar 6, 2013

One of the major difficulties when it comes to studying the ecology of long distance migrants is our poor understanding of what they get up to in the non-breeding season – where they overwinter, how they get there, what habitats they use, etc. In the case of Afro-Palearctic migrants, political instability makes ground work tricky – plus, Africa is a massive! Traditionally we relied on people coming across dead birds which had previously been ringed in Europe – however, recovery rates are extremely low & biased towards inhabited areas, and the fact that we’re relying on dead birds means we might be sampling only the poorer areas and missing the better quality ones. As far as I know, only one Roller has been recovered in sub-Saharan Africa – a Bulgrian bird found in Tanzania in December – but for some reason there’s no record of this in the EURING database…

Remote tracking devices are increasingly being used to follow the migration of individual birds. Satellite tags use GPS (or similar) to record accurate location information, which is then beamed back to base over the satellite network. Satellite tags are being used by the BTO to track cuckoos to the Congo and back, by SEO to track 4 Spanish Rollers – who are starting to depart from their wintering haunts in the Serengeti – and (I think) by Nathalie Gilbert, a fellow PhD student at UEA, to track overwintering white storks in Portugal. However, satellite tags cost several thousand £s a pop, and have only just become light enough for use on birds as small as Rollers (which are still an order of magnitude heavier than, say, a warbler) – the white stork trackers weigh as much as whole Roller!

An alternative to the expensive & heavy satellite tags is an archival geolocator. ‘Archival’ means that the data is stored on board – not transmitted back to base. Unfortunately this means that the bird must be recapture to download data. This shouldn’t be a problem for fairly site-faithful nestbox breeders like Rollers, but it does mean that the data is biased towards those birds which survive, so there’s no way of finding out where mortality is occurring.  Geolocators (also called GLS, I believe – too many acronyms!), then, record light intensity every 10 minutes or so. They last upwards of a year, weigh just a few grams and cost a few hundred pounds. Once (if!) the bird is recaptured, the light intensity curve can be used to determine the time of sunrise and sunset (the GLS also carries a clock). The absolute timing of these twilight events allows longitude to be estimated (because the sun rises earlier in the East) and daylength allows latitude to be estimated (because daylength increases or decreases towards the poles, depending on the time of year). Sounds simple!

Geolocator #9 with teflon leg loops.

Geolocator #9 with teflon leg loops – you can imagine it as a frontless pair of pants (/ g-string?), with the geolocator fixed to the back.

Last year,  Simon was nice enough to deploy ~40 geolocators on French and Latvian Rollers (see photos below). One of the main objectives of the coming field season is to retrieve as many of these loggers as possible, then analyse the data to work out the migration routes of  birds from each population.

The light stalk of a geolocator, protruding through the plumage

The light stalk of a geolocator, protruding through the plumage

Close up of the light stalk

Close up of the light stalk

Two geolocators were left over from last year’s deployment, which Simon took on his travels; he then gave me the task of tracing his winter movements. Interpreting the light curves is more complicated than I’d anticipated. First you need to clear up all the ‘false’ twilight events caused by shading. You then need define the angle of the sun below the horizon at which sunset and sunrise occurs, so the astronomical algorithm can estimate latitude and longitude. Ideally, you’d say something along the lines of ‘When the light falls below ‘2’ (units unknown..), the sun is at -6 degrees. Whereabouts in the world was the sun at -6 degrees below the horizon at the time when light fell below 2?’ However, shading from clouds, vegetation and topography messes this all up, so that when the light falls below 2, the sun might still be at +3degrees, meaning that the estimated sunset will be too early, and your estimate of location will be out… Things also go wrong when the birds (selfishly) move large distances, as the sunset will be earlier or later than expected, so the time of noon will be shifted. During the spring and autumn equinoxes (when migrants are doing interesting stuff), latitudinal variation in daylength is minimal (especially in the tropics), so it’s difficult to estimate latitude with any accuracy.

Light curve screen grab. Green = sunrise, red=sunset

Light curve screen grab. Green = sunrise, red=sunset

So, it’s an imperfect science, and accuracy is between 10s and a few 100s of km. Nevertheless, the map below shows that geolocator #09 did a pretty good job of finding Simon in Norfolk (/ the North Sea), Portugal and New Zealand (/ the Pacific Ocean). The green markers show the actual locations (I’ve slightly moved the Norfolk marker to protect Simon’s true identity!), and the yellow points mark the three-day average locations estimated by the geolocator. On the grand scheme of things (given that Rollers migrate several thousands of kilometers) the accuracy is good enough. You should also notice that latitude vries more than longitude – this is because longitude is only really messed up if shading changes dramatically between one twilight event and the next, whereas any form of shading messes up latitude estimates.


Much easier to see when viewed in a larger map

Please note, I’m fairly new to all this geolocator stuff, so all of the above might be completely incorrect…

Next week – stable isotopes, if you’re lucky!

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload CAPTCHA.