Archival geolocators have revolutionized #ornithology & movement #ecology but what happens when you only see data from individuals that survive? New paper out in
@AmOrnith Applications on the analysis & interpretation of tracking data from these tags https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/ornithapp/duab010/6232326?redirectedFrom=fulltext

Brief background on how this study came about. We spent several years deploying light-level geolocators on male Painted Buntings along the southeastern coast. Our goal was to better understand migratory connectivity and winter distribution of this declining population

However, in the first two years of the project, we confirmed at least 4 of our tagged birds were captured in Cuba as part of the illegal pet trade. Aside from the obvious conservation issue, we realized this might be a problem for interpreting the data

Basically, if birds that go to Cuba are less likely to return, we won't see them in our tracking data (at least not in proportion to how often they actually go there)! If this bias is extreme enough, we could conclude (incorrectly) that Cuba isn't an important wintering area

And this is just an inevitable bias of archival tags. Because you only see data from surviving individuals, you will systematically underestimate connectivity to regions with low survival (and overestimate regions with high survival)

As we show in the paper, we can lessen this bias by analyzing geolocator data with CMR data. Basically, survival estimates from CMR data tell us about both connectivity & regional survival - if lots of birds go to low survival region, average survival will be low (& vice versa)

One cool side effect of the model is that we also get estimates of regional survival for birds going to different wintering areas. For PABU, our data indicate that males wintering in Cuba have nearly 20% lower apparent survival than birds wintering in FL or Bahamas

If you use archival tags, this could be an important source of bias for interpreting your data! All of the code necessary to replicate analyses shown here is included in the supplementary files but feel free to reach out if you have any questions

And shout to my awesome co-authors, including @a_sharp6 & @full_o_fulmars for doing most of analysis, twitterless Scott Sillett ( @SMBC) & Vivivana Ruiz ( @CornellBirds) for helping to get this project going, & Paul Sykes for providing access to his long-term resight data