An important goal of almost every birder is to identify birds correctly. To that end, one of the joys of birding is that there is always more to learn. As we study and gain experience in the field, we become more attuned to look for the best marks that lead to proper identifications.

Some field marks are diagnostic, leading us quickly to the correct identification of a species. A duck with a green head and yellow bill, for example, is not to be confused with anything but a Mallard. A hawk with a reddish tail and dark patagial patches is almost certainly an adult Red-tailed Hawk. Learning these diagnostic marks (or combinations of marks) leads to a definitive identification to species.

With the growing popularity of eBird, birders are now increasingly aware of opportunities to learn more about differences within species—in particular, subspecies. (See Fig. 1: What Is a Subspecies Anyway?) Some subspecies are widely known by the birding community, often because they differ so much from one another that, at some time in the past, they were considered to be separate species. Examples include subspecies of the Dark-eyed Junco and Yellow-rumped Warbler.

eBird has also increased our awareness of subtler variation within some species, focusing on differences that can be noted in the field when viewing conditions are adequate. It is important first to understand the difference between subspecies, as defined by scientific ornithology, and what eBird calls “field-identifiable forms” or “eBird groups.” Some species have many subspecies that are so subtly different in appearance that it would be impossible to identify them reliably without having a series of specimens in hand. Thus, recognizing them while birding is essentially impossible. In this article, we consider “field-identifiable form” and “eBird group” to be synonymous terms.

Here, we offer advice to help birders decide when to identify birds below the species rank to the level of field-identifiable form or eBird group. We first explain why the topic is interesting and provide examples. We then address some challenges of reporting birds to levels below species, paying particular attention to the interesting dynamic of using probability (location) vs. field marks to identify birds. Next, we offer a recommended decision-making process to be used when identifying and reporting field-identifiable forms to eBird. Finally, we summarize information on how one learns the identification criteria for eBird groups.

Besides the challenge and fun of looking at birds very closely, we still have much to learn about differences in geographic distributions, seasonal movements, and even identification criteria useful in the field. By gathering observations in a central database like eBird, the collective effort of thousands of birders accumulates information, including rich media such as photos, videos, and audio recordings, helpful for learning more about subspecies and forms.

The external differences we see in the colors and shapes of birds may be present because some subspecies are on different evolutionary tracks than others within their species. In birds, geographic isolation of populations, such as during glacial periods, often leads to accumulation of differences as each population experiences different selective pressures where it lives. In other words, if sufficient differences persist long enough, some subspecies become different enough from each other to warrant consideration as full species. Typically, this process is very slow, requiring tens to hundreds of thousands, even millions, of years. If differences we see among subspecies provide a signal that different forms may have some degree of evolutionary separation, it is super interesting, so why not gather information about it? In addition, current laws protecting birds extend to populations with evidence of evolutionary distinctiveness, which includes subspecies (Haig et al. 2006). By learning more about such differences, we also help collect data that may enhance bird conservation.

Altogether, we have much to learn about the geographic distributions, seasonal occurrence patterns, and identification criteria of many subspecies. It truly is a realm where eBirders can make substantial contributions when their observations are carefully gathered.

The opportunities to learn more and to share knowledge of subspecific information has prompted eBird to provide opportunities to track such information. Two examples can help clarify the reasoning for including field-identifiable forms as options in eBird.

In White-crowned Sparrows, pugetensis is a migratory subspecies found in the humid zones of the Pacific Northwest and northern California, whereas nuttalli is a resident subspecies occurring in central California. Both have pale lores and yellow bills and constitute the “Yellow-billed” group in eBird. In the field, they are normally not reliably separated by their field marks, but they are obviously different from the orange-billed (Gambel’s) and pink-billed, darker-lored subspecies (oriantha and nominate leucophrys) occurring elsewhere in North America. The degree to which the migration routes and the breeding and wintering areas of the different subspecies overlap is incompletely known but, because of the distinctive differences, careful observations of birders can improve our knowledge.

Likewise, Horned Larks have an impressive variety of plumages across their vast geographic range. The range-restricted strigata of western Oregon and Washington is distinctively brighter in color than the widespread subspecies of the arid Interior West, merrilli. Yet, strigata’s appearance is quite similar to subspecies found to the south in California, so eBird currently limits identifications to the “western rufous group” instead of allowing identification to the subspecies level. In part, this limitation is imposed because we still do not yet know the extent of migration and occurrence patterns for all the subspecies contained in the western rufous group nor do we fully understand the range of plumage variation that could influence our identifications. Thus, field-identifiable forms have been created to combine subtly different subspecies into groupings that can be reliably identified by birders.

Notwithstanding the extensive literature on species concepts, which can be characterized by rancorous debate and continuing research, we consider it to be reasonably accurate to suggest that recognition of bird species focuses on evidence of reproductive separation. That separation leads to reduction or elimination of gene flow. In contrast, subspecies differences are based on “morphological diagnosability” (Patten 2015). That is, subspecies are not species because we know or strongly suspect they can interbreed with each other and produce viable offspring. Therefore, they are not necessarily reproductively isolated. We are generally comfortable assuming they could interbreed if given the chance, so we focus more on differences in appearance.

It turns out that there really is no standard definition of subspecies. This is not surprising given that there is also no universally accepted definition of species! A starting point is the idea of diagnosability, defined as our accuracy in reliably differentiating between subspecies based on their appearance. Biologists have developed a “75% rule,” sometimes called “Amadon’s Rule,” for applying this standard of diagnosability (Amadon 1949, see Patten and Unitt 2002). This idea recognizes subspecies if 75% or more of the individuals in one group fall outside the range of variation of 99% or more of individuals in other subspecies. For example, if we examined 10 juncos we presumed to be Oregon Juncos and all 10 (100%) had brown flanks and a blackish head while none of, say, 100 Slate-colored Juncos had the combination of brown flanks and blackish head, we could conclude that the two groups are diagnosably different as subspecies. On the other hand, if we looked at Brewer’s Blackbirds and found that 70% of birds in eastern Oregon had glossy green head feathers while 20% of those in western Oregon also had glossy green head feathers, we would not conclude, based on that field mark alone, that the two are different subspecies.

The authority on taxonomy of North American birds, the American Ornithological Society, has not officially revised our subspecies lists since 1957, largely because of limited resources to do so for the large number of species occurring north of Colombia. To provide birders an opportunity to contribute via our field observations, eBird has combined similarly appearing subspecies into forms that are different enough to be visible in the field under good viewing conditions. The future will probably bring a merging of ideas on defining subspecies based on a combination of data from appearances and genetics (Pruett et al. 2008, Patten and Remsen 2017).

Finally, today’s subspecies may prove to be tomorrow’s species. In the past few decades, numerous eBird subspecies groups have “graduated” to species status, including Black and Common scoters, Xantus’s and Guadalupe murrelets, Woodhouse’s and California scrub-jays, Winter and Pacific wrens, and Sagebrush and Bell’s sparrows. No doubt more will follow.

The identification of birds to field-identifiable forms presents multiple challenges worthy of discussion. Our goal here is to focus on identification of birds reported to what eBird calls “field-identifiable forms” or “eBird groups.” The first step is to learn where to find information on what options currently exist in eBird. The Cornell Lab of Ornithology’s website Birds of the World (birdsoftheworld.org) gives full information on eBird groups, including their ranges and component subspecies, but this resource is currently available only by subscription. The Birds of the World taxonomy explorer (birdsoftheworld.org/bow/species), however, provides another way to see the subspecies groups. Within eBird, it is easy to see which options exist for each species by navigating to the Explore pages, then choosing Species Maps, and entering the species name in the search box. A list of all the available options appears.

A list of options without identification criteria, however, can lead to mistakes. We have noticed that many birders make faulty assumptions or generate higher rates of misidentifications when reporting subspecies. Yet, attempting to make subspecific IDs offers us a chance to learn something new, to renew our interest in familiar species, and to increase our appreciation for the remarkable diversity within species. Below, we discuss some potential challenges and work through an example to illustrate best practices.

The differences in field marks of field-identifiable forms can be subtle. To support identifications, exceptional views, extensive observer experience, and perhaps even photos or audio recordings are required. By studying one’s local birds in detail, it is possible to become familiar with the typical range of variation so that when something different appears, it becomes more apparent. Also, one needs to be aware of the circumstances for each species and the array of subspecific options from which it makes logical sense to choose. For example, the field-identifiable forms of Green-winged Teal can only be identified in the field for adult, breeding-plumaged males, not for females or juveniles. Most songbirds cannot be reliably identified below species level as juveniles. Best practices suggest that we should attach subspecific labels only in the proper species-specific circumstances.

Excellent views or diagnostic photos or recordings (for instance, with crossbill types) are normally required to identify birds to eBird group. So, when we find large groups of a species that may contain multiple field-identifiable forms, such as flocks of White-crowned Sparrows across the American West, we should identify to eBird group and report only the number of individuals actually viewed sufficiently. All others should be reported only to species level. For example, an eBird checklist in Idaho during fall migration might report five White-crowned Sparrows (Gambel’s) for which the key field marks were all clearly viewed and 50 White-crowned Sparrows that got away too quickly for detailed looks.

An especially difficult conundrum involves the perspective of each birder on how probability plays into bird identification. For example, Downy Woodpecker has three identifiable forms in eBird: Eastern, Rocky Mountain, and Pacific. We currently think Downy Woodpeckers are essentially residents, engaging in little or no migration. For this reason, a Downy Woodpecker in Illinois is almost certainly an Eastern Downy Woodpecker. So why not report all Downy Woodpeckers in Illinois as Easterns? The probability of being wrong is very low. On the other hand, if they are all Easterns, what’s to be gained by reporting them as a field-identifiable form? We already know they are almost certainly all Easterns. This is a case of “no harm no foul;” either way is unlikely to mislead our understanding of Eastern Downy Woodpecker distribution.

But what happens in other parts of their range, especially out West? Or in the western portions of the Eastern Downy range where they might mix with Rocky Mountain forms? In those places, our assumption that all Downy Woodpeckers are of the one resident form may be incorrect. We can’t know how often we might be wrong without documented evidence. Automatically assuming the most likely form can dilute our perception of how prevalent that form is. That is, if we report all birds as the local form by default, we “prove” our idea that they are all the local form, regardless of whether we actually have evidence.

Making identifications based on assumptions of geography and probability raises a thorny issue at the heart of establishing best practices in bird identification. Most birders use probability all the time. Take a simple example: hummingbirds. In eastern North America in June, if a hummingbird flies by, nearly all of us would feel comfortable calling it a Ruby-throated. We are almost certainly correct. Yet, technically speaking, identificationof hummingbirds, especially females and immatures, requires far more than a brief look at a flying bird. We could be wrong calling this a Ruby-throated, but it is very unlikely. If we are in west Texas, however, and a hummingbird flies by, even the best observers in the world will be challenged to pin the identification to a Ruby-throated vs. a Black-chinned; both options are far more probable than zero (so it is best to report as hummingbird sp.). Knowing the context of each situation really matters, and should influence the decision.

The context helps us understand the amount of information needed for a correct identification. For a June hummingbird or a Downy Woodpecker in Illinois, there’s not much chance of being wrong, even with incomplete details. August in west Texas begs us to get a closer look at the hummingbirds, just to get to a correct species-level identification. Likewise, context and probability can guide us toward likely subspecific identifications, but proper IDs must be made by use of field marks.

And that’s the tricky bit. It’s asking a lot of birders to know the context for all the combinations of subspecies and forms, the geographical locations, and the probabilities that each may occur on particular dates of the year. Consider this: If each species averages just two field-identifiable forms, that doubles the amount of information needed to nail the subspecific ID. Many species have more than two forms, so the amount of necessary information becomes large quickly. In states with hundreds of species, it requires a lot of knowledge just to decide which species might be encountered. For these reasons, we do not recommend that birders attempt to classify all species to subspecies group.

Just as you would with rare birds, be conservative with subspecific identifications until you have enough experience for reliable identifications. The eBird filters established by regional reviewers can guide which options are expected in your region, but the information used to construct the filters may be incomplete. Your careful observations can help improve the knowledge base upon which strong filters are built.

Remember:

1. Probability suggests options, but field marks establish criteria for correct identifications. Do not assume an identification based on location unless you know definitively that there are no other options (and those cases are rare). Use the flow chart in Fig. 1 to guide your decision.

2. Experience helps. Get to know your local birds well, and study plumages of the entire species to improve your understanding of variability across the whole geographic range.

3. The more specific the level of identification, the more information you need. Identification to field-identifiable forms or eBird groups requires more information (better views, as well as photos or recordings) than identification of most species. Correctly identifying birds (just) to species level is fine.

An important limitation to our ability to accurately identify birds is imposed by the set of resources we use. Most field guides, the primary resources birders consult, have incomplete information about subspecies. The goal of field guides is to help us identify species correctly. When subspecies are portrayed, it is normally to show readers a sample of the range of variation in appearances of the species. The intent was not to help us see all the subspecies because some species have many subspecies that vary in only subtle ways. Field guides, as most of them are currently composed, are not meant to help us identify field-identifiable forms or eBird groups. Some guides now include range maps for a select set of distinctive subspecies, which can guide your decisions. But remember that identifications should be based on field marks, not range maps.

If you become interested in identifying field-identifiable forms or eBird groups, we recommend consulting the aforementioned Birds of the World (birdsoftheworld.org), an online resource, through the Cornell Laboratory of Ornithology. Within each species account, the section on Systematics contains relevant details on subspecies and eBird groups. Many accounts also include illustrations of plumage differences across recognizable subspecies. This resource is continually being improved and is probably your best option today. The Clements (2019) checklist includes exhaustive lists of recognized subspecies and their general geographic ranges, as well as details on which subspecies fall into which eBird groups. However, it does not include identification criteria.

The addition of field-identifiable forms to eBird as “eBird groups” has opened the door for birders to contribute new information about subspecies. But bird identification below the level of species can be challenging. Birders who think carefully and logically, objectively weigh information from sightings, and consider the context of each situation can achieve reliable subspecific identifications. We recommend a cautious approach when identifying birds to subspecies groups. Birders can help advance our collective knowledge faster by focusing on field marks, avoiding assumptions, and documenting identifications with diagnostic photos or recordings. In the process, by taking a closer look at familiar birds to better understand and appreciate their variation, birders may find their birding re-energized.

Acknowledgments
We thank the many birders who contribute their observations to eBird. Their patterns of reporting subspecies prompted us to explore this topic more. The eBird team invests immense energy to design and run eBird in thoughtful ways, including decisions about inclusion of field-identifiable forms. WDR was supported by the College of Agricultural Sciences and the Bob and Phyllis Mace Professorship. Comments from T. Hallman, M. Hunter, M. J. Iliff, R. Namitz, D. Trochlell, and B. Tweit improved our thinking and the paper. We thank Tara Kate Designs for the decision tree illustration.

Works Consulted

Amadon, D. 1949. The seventy-five per cent rule for subspecies. The Condor. 51: 250–258.

Clements, J. F., T. S. Schulenberg, M. J. Iliff, S. M. Billerman, T. A. Fredericks, B. L. Sullivan, and C. L. Wood. 2021. The eBird/Clements Checklist of Birds of the World: v2019. Downloaded from https://www.birds.cornell.edu/clementschecklist/download/. Ithaca, New York.

Haig, S. M., E. A. Beever, S. M. Chambers, H. M. Draheim, B. D. Dugger, S. Dunham, E. Elliott-Smith, J. B. Fontaine, D. C. Kesler, B. J. Knaus, I. F. Lopes, P. Loschl, T. D. Mullins, and L. M. Sheffield. 2006. Taxonomic considerations in listing subspecies under the U.S. Endangered Species Act. Conservation Biology. 20 (6): 1584–1594.

Patten, M. A. 2015. Subspecies and the philosophy of science. The Auk. 132 (2): 481–485.

Patten, M. A., and J. V. Remsen, Jr. 2017. Complementary roles of phenotype and genotype in subspecies delimitation. Journal of Heredity 108 (4): 462–464.

Patten, M. A., and P. Unitt. 2002. Diagnosability versus mean differences of Sage Sparrow subspecies. The Auk 119 (1): 26–35.

Pruett, C. L., P. Arcese, Y. L. Chan, A. G. Wilson, M. A. Patten, L. F. Keller, and K. Winker. 2008. Concordant and discordant signals between genetic data and described subspecies of Pacific Coast Song Sparrows. The Condor. 110 (2): 359–364.