eBird Patterns: Eastern Phoebe Migration

Eastern Phoebe spends the winter primarily across the southeastern United States, ranging as far north as Virginia, southern Kentucky, and central Oklahoma and as far south as central Mexico. The extent of the winter range predicted here roughly matches that shown in many field guides (try comparing!). Interestingly, the winter range appears to contract at its periphery over the course of the winter, probably indicating winter mortality in northern climes, but which may also represent local movements in response to harsh weather conditions. Eastern Phoebe is a "half-hardy" species that may not always survive successfully at the northern limits of its wintering range, and the STEM model seems to take this into account based on your eBird observations. This pattern will likely appear in other half-hardy species, since in colder areas many birders know that it is much easier to find a Eastern Phoebe or Common Yellowthroat in December than it is in February.

Fig. 1. Animated map of Eastern Phoebe occurrence for the Lower 48 United States in 2008. This map uses the STEM analysis to predict occurrence, and was developed specifically for use with eBird data. White areas indicate a high likelihood of encountering Eastern Phoebe, corresponding with 20% or greater frequency of occurrence.

Northward migration begins in early March, and by mid-March phoebes are setting up territories in many areas where they do not winter. By early April migrants have arrived throughout most of the breeding range, but the predicted occurrence continues to increase as later arriving birds fill in the range throughout the month. Note how accurately STEM interprets phoebe arrival in the mid-Atlantic, with occurrence reaching ~10% to 15% on 15 March 2008. Check the arrival timing in your local area and let us know if you think these maps are predicting it well--for the areas we know well, this timing seems to be spot on (send comments to ebird@cornell.edu, or post them on our Facebook page).

Fig. 2. Eastern Phoebe occurrence for the Lower 48 United States on 15 March 2008.

The wave of phoebes surges northward through early May, but then the movement appears to cease. The predicted occurrence waxes and wanes throughout the summer (see summer discussion below), but only in September does the drop in occurrence signal the beginning of fall migration. The southward migration is clear since each successive week shows progressively lower predicted occurrence in the north until it finally fades to black, at which point the predicted occurrence is effectively zero. One interesting thing is that the more westerly populations seem to migrate south somewhat earlier in fall. Compare the first arrival in South Texas to the first arrival in Florida. This is supported by the eBird bar charts: arrival in southeast Florida begins in early October with a peak in early November; at comparable latitudes in southeast Texas Eastern Phoebes arrive in force in September.

Fig. 2. Eastern Phoebe occurrence for the Lower 48 United States on 20 September 2008. Note the earlier fall arrival in south Texas as compared to Florida.

Fig. 3. Eastern Phoebe frequency in southeast Texas (red) and southeast Florida (blue), showing the earlier fall arrival in Texas. This image is taken straight from eBird's "View and Explore" data, but modified in excel to show both regions together.

On the breeding season phoebe map, apparent fluctuations in its the predicted occurrence are likely due to changes in detectability (see Northern Cardinal discussion). For example, most phoebes have arrived on their breeding grounds by May 1st, and for the next month they remain highly detectable as they are singing and defending territories. By mid-June detectability drops, presumably because the birds are not as vocal or conspicuous as they were just one month ago. Note how they appear to "disappear" from Texas Hill Country; this is surely due to a relatively scarce population of phoebes seeming to "blink out" when they stop singing and become less obvious.

The ability to watch migration patterns unfold day-by-day is one of our primary goals at eBird, and something we will be very excited to provide through eBird's "View and Explore Data" once we can produce these types of models for all species. Remember that these animations are generated by a computer that is working only with your eBird data, and GIS information on climate and habitat. Model research and development is ongoing and this is a work in progress, but we are excited to be able to share these early results with all eBirders. Until then, your day-to-day observations and effort-based complete checklists ensure that the migration maps that we produce will be as accurate as possible.

As always, we welcome feedback via Facebook or through email to ebird@cornell.edu. And if you appreciate new features like this, and hope to see more of them, please remember that we do welcome donations at whatever level of support you can provide.

NOTE: There are a few differences between this STEM model and the Northern Cardinal one. First, this one uses both Stationary and Traveling Count data. Second, the increase in data volume means that the predictions can be made at finer spatial scales, so there is less of a problem of extrapolation into areas where phoebes do not occur. Third, the scale is different: the cardinal map peaks at 1.0 (i.e., 100% probability of occurrence) while the phoebe map peaks at 0.2 (i.e., 20%). Bird gurus might pick nits at the apparent edges of the range (where Eastern Phoebes are rare at best), but overall these maps are extremely accurate and match those found in most modern field guides!