The eBird Conundrum

Editor's note: While it can be argued that eBird is already the most successful citizen science project ever, some birders remain unaware of this project, while others have yet to contribute sightings because they have concerns about the accuracy of the data being collected. It's also important to note that many birders have no inclination to keep track of the species or numbers of individual birds that they encounter in the field. To them, counting birds, keeping notes, and spending the time it takes to enter their sightings is beyond their interest level and/or viewed as  "work." This article takes a look at the history of amateur participation in avian data collection and specifically discusses some of the challenges that the eBird staff faces as they strive to involve more birders in this project. In publishing this piece, we at BirdFellow hope to initiate a dialogue between committed eBirders and those who have yet to embrace this project. After reading this article, we invite you to share your thoughts, concerns, and eBirding experiences by posting a comment.

A Historical Perspective

In the beginning, data generated by the modern birding/birdwatching activities of amateur hobbyists was at best suspect and at worst completely ignored by the professional ornithologists. Not so long ago, it would have been unheard of to publish any bird record that was not backed up by a specimen in a museum tray.

Then, during the middle third of the 20th century Ludlow Griscom and others, including a young Roger Tory Peterson, began demonstrating that most birds one might encounter could be identified by sight or sound without being collected. In 1934, Peterson published his first Field Guide to Birds, revolutionizing the methodology we (non-professional observers) use to separate one species of bird from another.  Peterson's guide focused on the significance of unique characteristics we call field marks and how they can be used to identify birds visually. Ongoing refinement of this methodology has resulted in the publication of numerous field guides, identification guides to certain groups of species (gulls, warblers, shorebirds, etc.), and highly focused articles that discuss the finer points of identifying birds that are very similar in overall size, shape, and appearance.

The Citizen Science Epiphany and the Launching of eBird

Over time, professional ornithologists came to realize that there was a massive pool of potentially useful data being generated by amateur birders. Until fairly recently, the most popular and well-known citizen science project relating to birds was the National Audubon Society's Christmas Bird Count. In an effort to collect even more detailed data about wintering birds, the Cornell Laboratory of Ornithology started Project Feeder Watch in 1988. Cornell enlisted  legions of North American feeder watchers who volunteered to record and report on birds that they observed at their feeding stations. Project Feeder Watch was focused on monitoring year-to-year as well as long-term changes in the wintering populations of backyard birds. As Internet use became nearly universal, the Cornell Lab and the National Audubon Society partnered to design the most ambitious citizen-science project to date, jointly launching eBird in 2002.

The aim of eBird is to collect year-round observations from birders all over North America and eventually the world. It is believed that data collected by amateur birders will prove extremely valuable as scientists endeavor to better understand the population trends and distributional changes of the world's avifauna. Prior to eBird, much of this data either went unrecorded or took up residence in the notebooks or other personal databases that may never be incorporated into a usable database. Since eBird is an easily-accessed online database and is open to anyone willing submit their sightings, it has become a highly popular repository where birders of every skill level can freely contribute observations. The eBird project now collects more than a one million individual bird observations per month.

Barriers to Full Adoption of eBird

By making eBird available to any and all comers, the Cornell Lab and National Audubon have opened themselves to criticism about the accuracy of the data being collected via eBird. Ironically, the same argument used decades ago by professional ornithologists to discredit the contributions of amateurs are now being used by some amateurs to discredit eBird, a database designed by professional ornithologists. Some veteran observers have publicly questioned the value of a database that by default includes some faulty data points (misidentified birds). I admit to having been among those with such concerns.

Perhaps part of the problem is that eBird circumnavigates the informal/formal filter systems that many of us have come to embrace. Over the last half-century or so, amateur birders have developed a network of gatekeepers to filter out incorrect and questionable reports. In addition to local field notes editors and the regional editors for North American Birds (NAB), states, provinces, and nations have created records committees whose sole purpose is to determine the validity of reports of birds that are rare or unusual in a particular region. I am currently a regional editor for NAB and also a member of the Oregon Bird Records Committee, thus I've had a role in these gatekeeping activities. Birders unfamiliar with these processes are often put off by the notion of some faceless individual or small group of individuals passing judgment on their reports. Sadly, these methods have produced an undesired schism of sorts that divides those who make such decisions and those whose reports are being examined.

To solve this dilemma, the eBird team has developed a self-policing filtration system that allows the observer to decide if they want to "confirm" their report of an unusual species or unexpectedly high number of individuals. Surely some less-experienced observers opt to delete reports that are red-flagged, while others will steadfastly stand by their initial report even when it is pointed out that what they are reporting is unusual. Beyond the self-policing controls in the system, eBird coordinators have recruited local experts to further refine filters and communicate directly with the observer base. In the end, local reviewers are still able to set aside suspect reports so that they are not incorporated into the database.

Since eBird relies on local volunteers for their review process, there is significant region-to-region variation in how the review process is executed. In areas where there are lots of active eBird contributors and a strong local review process, the quality of data is generally very high. Conversely, in areas with either low eBird adoption or a low number of contributing observers, there is often no local review person and the local filters are not fine tuned. The eBird staff is engaged in a continuing effort to make the review and filtration processes more uniform across all regions, states, provinces, and countries.

These imperfections and other perceived shortcomings in the eBird methodologies have led some very talented, experienced observers to avoid contributing their sightings to the eBird database. Initially, I was a reluctant contributor to eBird for some of these same reasons. Since then I've taken a serious look around and I see that there are currently no alternatives on the horizon. Like it or not, eBird is here to stay.

Why I Am an eBirder

Since I believe in the ultimate mission of eBird, my hope is to play a role in the ongoing refinement of eBird. I recognize that to have any influence I must first be willing to submit my own sightings and encourage others to do the same. One can only be part of the solution if they are actively engaged in the process. Remaining on the sidelines and criticizing a process that you are not engaged in is unlikely to result in the positive changes you hope to see. I have yet to hear any birder argue against the value of the type of database that is being created via the eBird project.

Generally speaking, the most active birders are usually the most experienced as well. It is probably safe to assume that the most experienced observers will see more species, more accurately identify the species that they encounter, and take a more serious approach to the data that they submit. Further, experienced observers are likely to recognize filter settings that need adjustment and offer substantive comments that will help the eBird staff make the appropriate alterations. In theory, the shear volume of sightings contributed by this core group of observers is likely to outweigh the occasional faulty data points contributed by birders who only get out into the field a few times a year. 

At the very least, I would encourage those who remain skeptical to enter a few of their checklists into the system in order to gain a full understanding of how it works. Don't just enter simple checklists, but keep close track numbers and the age and sex of the birds you see and then enter a more detailed report. Engaging in this sort of observation methodology will not only increase your birding skills, but I have found satisfaction in knowing that my observations are going into a system where they have value, even if that system is still a work in progress. 

The Imperative

Few birders would argue that the time to start channeling our collective observations into a single database is now. There is always risk and some element of resistance to change. Were professional scientists prudent in ignoring colloquial and informal data collected by the earliest amateur birders? Is it prudent now to dismiss or attempt to discredit current efforts to develop a collective database where our observations can be put to scientific use?  Is eBird so hopelessly flawed that we should abandon the incredible momentum that has been established and try to start anew? In my opinion, the answer to all three of these questions is no. Can eBird be improved in terms of its ability to reduce the number of faulty data points that make their way into the database? I think those at the core of this project would be the first to answer, "certainly." Regardless of how data is collected or who collects it, in the end there will be imperfections. This begs the ultimate question. Should we wait for someone to come along with what appears to be a more perfected methodology, or should we embrace an already popular methodology with known imperfections and endeavor to make it as perfect as we can?