Gulf Crossings: How to do Bad Science

Recently, I was included (with a group of ex-grad school cohorts and colleagues) in an email thread begun by a prominent ornithologist at my alma mater. In it, he refers to (and includes a PDF of) a recent paper published in the journal BioScience entitled "Changes in Bird Abundance in Eastern North America: Urban Sprawl and Global Footprint". The authors and reviewers will remain nameless - you can look it up if you're curious. However, let it be known that I wouldn't let a dermatologist do brain surgery on me, if you know what I mean.... The major conclusions of the paper are:

Overall, breeding bird species numbers in North America have declined 19% since 1965. In addition:

Forest bird numbers are increasing.

Neotropical migrant bird numbers are increasing.

Birds that breed and winter in North America are decreasing.

Rock Doves and House Sparrows are declining significantly.

Hmm. There seems to be something counter-intuitive about this. Most of us have learned or studied the effects of forest fragmentation and the loss of forest habitat over the last 300 years in North America. Fragmentation of forest exposes nesting birds to increased nest parasitism by Cowbirds, and increased predation by crows, jays, domestic cats, and raccoons, all of whom favor open habitats. Evidence is mounting that neotropical migrant numbers are decreasing due to deforestation in the tropics, which is rampant and increasing. Studies are currently being done which may point to the loss of stopover habitat in the southern U.S. as another cause of the decrease in populations of a number of migrant species.

To reach the conclusions listed above, Breeding Bird Survey data was analyzed from 1965 to 2005. The entire basis for the analysis done was a table listing bird species from the Atlantic and Mississippi flyway. Each species was catagorized by where it bred, and where it spent the boreal winter. Fully resident in North America, migrate south to southern U.S., Mexico, northern South America and the Amazon basin, or migrate to southern South America. Unfortunately, the table contained a significant amount of data that was just plain wrong. Chestnut-backed Chickadee winters in Mexico? The bird is a N. American resident, and occurs nowhere near the Atlantic or Mississippi flyway. Canada Warbler a N. American resident? The species winters in north and central South America. Cape May Warbler winters in Amazonia? Nope - the West Indies - no records for South America. And so on. Probably more than 20% of the table is wrong. In addition, on BBS counts it is widely understood that common species will be recorded in large numbers, and this must be accounted for in any population trend study.

A classic case of using statistics on bad data and coming up with stupid conclusions. Obviously the authors did not know much about the ranges of N. American birds (not sure what happened with the reviewers). Unfortunately, the conclusions of this paper are now in the literature, and may influence conservation decisions. Expect to see a retraction or major revision to this paper - none of the ornithologists I know are going to let it pass.

Bill