A system for representing bird taxonomy

By collision, we mean a case where two or more names would have the same code while applying the rules. For example, Blackburnian Warbler and Blackpoll Warbler would both be encoded as BLAWAR. In that case we devise substitute codes that still suggest the original names: in this case, BKBWAR for Blackburnian Warbler and BKPWAR for Blackpoll Warbler.

Therefore, code BLAWAR, and the other collision forms, are not allowed, even in cases where one of the names is extremely unlikely to occur. For example, American Goldfinch collides with an ancient and obsolete name for Common Goldeneye, “American Goldeneye.” We disallow the collision code AMEGOL and require the use of the substitute codes AMEGFI and AMEGEY. This is a necessity in the Christmas Bird Count database because so many old names occur in the early days of that census.

The author's long career working with banding data has convinced him that the four-letter codes devised by the US Fish & Wildlife Service's Bird Banding Lab are highly error-prone. Roughly 100 names are involved in collisions in this system!

Most banders work with a fairly small set of local species: a bander on the East Coast may encoded Carolina Wren as CAWR, not realizing that this code is a three-way collision with Cactus Wren and Canyon Wren. When the code CEWA shows up on a banding sheet, it is not always clear whether it means Cerulean Warbler or Cedar Waxwing. The BBL also generally does not use the collision form. In this case they supply substitute codes CERW and CEDW, so that the appearance of code CEWA on a banding sheet is always an error.

However, they violate this rule in a few cases where one of the colliding names is highly unlikely to occur on a banding sheet. For example, because Barnacle Goose is quite rare, code BAGO is allowed for Barrow's Goldeneye, with code BRNG used for Barnacle Goose. The author feels that this is risky, since it assumes that everyone working with bird records knows something about bird distribution.