© Joanna M. Reuter 2019

This exploratory analysis accompanies an article in The Chat, newsletter of the Columbia Audubon Society.


Broad-winged hawks (BWHA) migrate south through Missouri in late September and early October, sometimes in large flocks numbering in the hundreds or thousands. Witnessing large numbers is a very exciting experience. Here, I take a look at the existing eBird data to better understand the timing and magnitude of migration.

This is exploratory analysis using the R programming language. Due to the volunteer nature of this project, the following text is not polished, explanation is limited, and some of the axes have labels that are illegible (cringe). Familiarity with eBird is assumed. Please contact me with questions (or ideas/requests for analysis):

This is a collection of plots based on eBird data, as follows:

I imported the data into R using the auk package.

Six-county region of interest

The plots in the first sections of this document consider the six-county area that is served by Columbia Audubon Society: Audrain, Boone, Cooper, Howard, Monroe, and Randolph Counties in Missouri.

Histogram of observation count

Considering all eBird observations of Broad-winged Hawks: Most observations are of just one bird. High count is 6,000. The first plot uses a standard scale; use of a log scale in the second plot helps to see the pattern a little better.

Distribution of BWHA observations within and between years

The plots below show the number of lists reporting BWHAs. Note that the total number of lists has been more or less increasing over the years; these plots do not take that into account, but plots later in this document will.

Broad-winged Hawks typically arrive in April. Most are passing through, but pairs sometimes nest in the area, and there are occasional summer observations. Fall migration peaks in late September. The number of lists reporting BWHAs is greater in the spring than the fall, but as we’ll soon see, there are also more lists in the spring than the fall. Frequency plots are presented further below.

Plots show only the part of the year for which BWHAs have been reported (April through October).

Caveat: Throughout this document, I’m using the term “week” loosely to mean either a 7 or 8 day period that is roughly a quarter of a month. This makes plots much easier to understand than using week numbers (when the heck is week 37 again?). This may not be the perfect way to handle the yearly time series, but given the oddities of our calendar, I don’t think it is the worst. In the plots, the week is identified by its start day in month-day format.

Lists and their characteristics

The number of species observations is better understood in the context of how many lists there are in a given time period. Before looking at the frequency plots, here’s a quick look at the number and types of lists over time.

eBird asks users a question when submitting a list: “Are you submitting a complete checklist of the birds you were able to identify?” The answer to this question determines whether a checklist is considered “complete” with all species reported.

eBird used to allow incidental lists to be marked as “complete”, but this is no longer the case. (Therefore, if birding was not your primary purpose, you are not allowed to mark a list as complete.)