There are currently four world bird lists referenced by different stakeholders including governments, academic journals, museums and citizen scientists. Consolidation of these lists is a conservation and research priority. In reconciling lists, care must be taken to ensure agreement in taxonomic concepts—the actual groups of individual organisms circumscribed by a given scientific epithet. Here, we compare species-level taxonomic concepts for raptors across the four lists, highlighting areas of disagreement. Of the 665 species-level raptor taxa observed at least once among the four lists, only 453 (68%) were consistent across all four lists. The Howard and Moore Checklist of the Birds of the World contains the fewest raptor species (528), whereas the International Ornithological Community World Bird List contains the most (580) and these two lists are in the most disagreement. Of the disagreements, 67% involved owls, and Indonesia was the country containing the most disagreed upon species (169). Finally, we calculated the amount of species-level agreement across lists for each avian order and found raptor orders spread throughout the rankings of agreement. Our results emphasize the need to reconcile the four world bird lists for all avian orders, highlight broad disagreements across lists and identify hotspots of disagreement for raptors, in particular.

1. Introduction

Taxonomy is foundational to ecology, evolutionary biology, conservation, and wildlife management [1–4]. In particular, species-level taxonomy frequently provides the basis of decisions for conservation prioritization of both geographical areas and organisms. For example, levels of species richness are often used as a measure to compare areas for conservation action and funding (e.g. [5–7]). Inconsistencies in taxonomy can, therefore, impact prioritization of areas and species for conservation (e.g. [3,8,9]). Species-level conservation is also affected by taxonomic decisions, with recognized species generally receiving more attention than lower-level taxa such as subspecies or populations (e.g. [10–12]; but see [13]). Splitting species into multiple species thus tends to increase protection, whereas lumping might decrease conservation attention afforded to that group [14].

The science of taxonomy can be particularly difficult because it requires assignment of continuously changing pools of individuals into discrete categories. That is, the designation of a group of organisms as one species or another represents an attempt by humans to organize continuously varying and changing populations into discrete evolutionary units for the purpose of addressing our own communication needs [15,16]. Criteria used to designate organisms as species are thus varied and inconsistently applied such that, given the same information, separate groups of experts might classify taxa differently [3,15,17]. As a result, the four commonly recognized world bird lists contain differing numbers of species-level taxa [18,19].

These bird lists have different applications and are all uniquely influential. For example, the international ornithological journal IBIS follows the International Ornithological Community (IOC) World Bird List (hereafter IOC; [20]), whereas several major museums around the world use the Howard and Moore Checklist of the Birds of the World (hereafter Howard and Moore; [21]). The eBird/Clements Checklist of Birds of the World (hereafter eBird/Clements; [22]) is curated and applied by the Cornell Lab of Ornithology and is implemented by all of their programmes including the citizen science programme eBird (eBird.org), the Macaulay Library, and the species life-history resource Birds of the World (birdsoftheworld.org). Finally, the Handbook of the Birds of the World and BirdLife International Digital Checklist of the Birds of the World (hereafter HBW & BirdLife International; [23]) is the taxonomy followed by BirdLife International when conducting Red List assessments for the International Union for the Conservation of Nature (IUCN). Only species recognized by HBW & BirdLife International are therefore included on the Red List. The HBW & BirdLife International taxonomy is also followed by international agreements including the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) and the Convention on Migratory Species.

Confusion over taxonomic revision or disagreement can hamper conservation (e.g. [18,24,25]), although the outcome is difficult to predict [14,26]. Differences between the four world bird lists were recently given as an example of ‘taxonomy anarchy’ and thereby detrimental to conservation [18]. However, these differences might be tractable, and informal conversations of reconciling the four lists have already begun [19,27]. Indeed, Collar [19, p. 481] stated that unity and agreement on global bird taxonomy are desirable goals and that ‘we should take or support all reasonable measures to reach them’. The difficulty in doing so may be significant and will require compromise regarding how species are fundamentally conceptualized.

The above four world bird lists differ not only in the scientific names that they contain, but among other differences, there are disagreements regarding the groups of individual organisms to which each scientific name refers. Put differently, they employ differing taxonomic concepts—the actual individual organisms that scientific names circumscribe [28]. Separate lists can refer to different individuals even when using the same scientific names. For example, all world bird lists contain the scientific name Tyto alba, yet each list assigns different regional groups to this name. This discrepancy results in wide disagreement between world bird lists regarding where T. alba occurs (figure 1). Therefore, by analysing taxonomic concepts instead of scientific names, one can more precisely compare and contrast the distinct groups of birds that each list considers to be a species.

Figure 1. Map of countries in which the four taxonomic concepts applied to the scientific name *T. alba* occur. The Venn diagram illustrates the colour symbology of the countries where lists consider birds circumscribed under the name *T. alba* to occur. eBird, the eBird/Clements Checklist of the Birds of the World; HBW, HBW & BirdLife International; IOC, the IOC World Bird List; HM, the Howard and Moore Complete Checklist of the Birds of the World. White areas in the Venn diagram are combinations that do not occur. For example, there are no countries where only IOC and HM consider *T. alba* to occur. To determine where a given list places *T. alba*, examine the colours contained within the circle ascribed to that list in the Venn diagram. Therefore, the countries where HM considers *T. alba* to occur are in blue and purple. By contrast, the countries where IOC considers to harbour *T. alba* are in purple only. Note that eBird and HM agree where *T. alba* occur at the country level and thus share a circle in the Venn diagram. However, eBird and HM disagree over which populations within India are *T. alba* or *T. deroepstorffi*. Thus, even though all lists agree that something named *T. alba* exists, they do not all agree on the populations to which that name refers. (Online version in colour.)

Here, we compare taxonomic concepts across world bird lists using raptors (orders Accipitriformes, Cathartiformes, Falconiformes, Strigiformes, and Cariamiformes; [29,30]) as an example. Raptors are important in ecosystem functioning [31,32], provide ecosystem services [33], and are generally more threatened with extinction than other birds [34]. Therefore, disagreements in raptor taxonomy might create certain conservation challenges. We examine the amount of agreement between the four lists and highlight where disagreements occur—per family, genus, and geographically per country. Hotspots of disagreement highlight where challenges await in reconciling the four world bird lists and where research into the taxonomy of raptors is most needed to help prioritize our conservation efforts. Finally, for context across Class Aves, we compare the amount of agreement on raptor species with that for all bird orders. This study should therefore serve as a template for comparing the four world bird lists—a crucial step in development of a single list.

2. Methods

We examined Avibase—The World Bird Database (https://avibase.bsc-eoc.org/)—to determine differences and similarities between the four world bird lists. Avibase is designed to organize avian taxonomic concepts published by global and regional checklists [28]. We analysed world bird lists at the species level and therefore examined all taxonomic concepts that are considered to be species by at least one list (hereafter, ‘concepts’). We determined which concepts were agreed upon (i.e. included at the species level with the same genus and species names; hereafter, ‘agreements’) across all lists and which were subject to disagreement (i.e. differing genus or species names across lists, or either missing from or listed as subspecies in one or more lists; hereafter, ‘disagreements’). We quantified the amount of pairwise agreement between each list as the percentage of taxonomic concepts upon which they agree.

To determine which groups of species were subject to the most disagreement, we calculated the number of disagreements per family and genus. Avibase also records the taxonomic concepts that occur in each country. We thus examined the spatial distribution of disagreements in taxonomic concepts across the globe by calculating the number of disagreements per country. Finally, we examined where the five raptor orders ranked among all 38 bird orders (number of orders in Class Aves according to Howard and Moore) regarding the number of agreements. Because Howard and Moore does not recognize Cathartiformes, for this Aves-wide comparison, we separated Cathartiformes from Accipitriformes and thus examine 39 Avian orders in total.

These checklists are revised on differing time scales, with some updated multiple times per year, and others only annually or even less frequently. Consequently, differences between lists fluctuate regularly, especially in response to new information. Here, we analyse the Avibase dataset as of 11 September 2019. We thus compare HBW & BirdLife International v3 (November 2018), IOC v9.2 (June 2019), eBird/Clements v2019 (August 2019) and Howard and Moore 4th edition volume 1 (2013).

3. Results

In total, there are 665 (electronic supplementary material, table S1) taxonomic concepts of raptors that are considered species by at least one world bird list. IOC contained the greatest number of these taxonomic concepts, followed by eBird/Clements, HBW & BirdLife International and Howard and Moore (figure 2). Agreement was highest between IOC and eBird/Clements (89% agreement) and lowest between IOC and Howard and Moore (75% agreement, figure 2). Overall, the four world bird lists agree on—i.e. include and assign the same generic and specific epithets to—453 (68%) of the taxonomic concepts. Across lists, 57 species-level scientific names are applied to multiple concepts (electronic supplementary material, table S2). In particular, all four lists disagree upon the concepts applying to the owls Glaucidium capense and T. alba.

Figure 2. Comparison of the species-level taxonomic concepts for raptors across the four world bird lists. Percentages indicate the number of concepts that are shared by a given pair of lists. The total number of concepts per list is given parenthetically.

Across bird lists, the most common cause for disagreement among raptors was omission of taxonomic concepts (131 concepts). There are also 81 concepts that do not have the same scientific name usage across lists (electronic supplementary material, table S3)—67 concepts were treated at different taxonomic levels among lists (species versus subspecies, e.g. Bubo magellanicus versus Bubo virginianus magellanicus) and 13 had differences solely owing to genus assignment among lists (e.g. Phodilus prigoginei versus Tyto prigoginei). Further, eight concepts were not only either listed under different genera or taxonomic levels, but were also missing from at least one list (e.g. Mascarenotus murivorus, an extinct species, was not included in Howard and Moore and is listed under the genus Otus by eBird/Clements). Only one had a different specific epithet owing to the acceptance of a different original description (Buteo burmanicus versus Buteo refectus), but referring to the same population. At taxonomic levels higher than species, neither Howard and Moore nor IOC include the Order Cathartiformes. All lists included the same eight families (figure 3), but there were seven genera that did not occur across all lists (electronic supplementary material, table S4).

Figure 3. Families of raptors and circles representing the number of species level taxonomic concepts (saturation) and the proportion of disagreed upon species concepts (size) for each family. Note that darker circles indicate more total concepts and larger circles indicate more disagreed upon concepts. Per family, the numbers of disagreed upon concepts are on top, and the bottom number is the total amount of concepts. For example, Strigidae is the darkest circle, so has the most total concepts, and is the third largest in size, so has a moderately high proportion of disagreements. (Online version in colour.)

Of the disagreements, 67% were owls, yet owls constituted only 47% of all the resulting species-level concepts (χ²₁ = 55.07, p < 0.001). The lists agree on 81% of non-owl concepts, yet only agree on 54% of owl concepts. Further, the three families that contained the highest proportions of contested concepts were Pandionidae (the osprey (Pandion haliaetus); 100% of three concepts), and the owl families Tytonidae (79% of 34 concepts) and Strigidae (42% of 242 concepts; figure 3; electronic supplementary material, table S5). Similarly, the genus that contained the most contested concepts was an owl genus (Otus; 52%, 35 of 68 species concepts), and seven of the top 10 genera containing the most disagreements were owl genera (electronic supplementary material, figure S1 and table S6).

The three countries that contain the most species concepts are Indonesia (169 concepts), India (117 concepts) and Uganda (122 concepts; electronic supplementary material, table S7), and the three countries with the most disagreements are Indonesia (74 concepts), India (51 concepts), and China (73 concepts, figure 4; electronic supplementary material, table S7). Indonesia also contains the most disagreed owl concepts (45 concepts), followed by Australia (25 concepts) and Mexico (25 concepts; electronic supplementary material, table S7). Among all 39 bird orders, raptors ranked as tied for first (Cariamiformes and Cathartiformes, 1.0), then 15th (Falconiformes, 0.90), 21st (Accipitriformes, 0.78) and 35th (Strigiformes, 0.57) in proportions of agreement (figure 5, electronic supplementary material, table S8).

Figure 4. Map of the number of disagreed upon species-level taxonomic concepts of raptors per country. (Online version in colour.)

Figure 5. Proportion of agreed upon species-level taxonomic concepts per avian order. We consider concepts to be agreed upon only if they occur at the species level across all four lists. Names of raptor orders are in orange and identified by asterisks. Saturation indicates the number of taxonomic concepts per order ranging from 1 (Leptosomiformes and Opisthocomiformes, lightest) to greater than 600 (Passeriformes, darkest). Here, we use the Howard and Moore taxonomy at the order level, except that we separate Cathartiformes from Accipitriformes, for the ease of comparison. (Online version in colour.)

4. Discussion

Our results demonstrate where the four world bird lists differ in treatments of species-level concepts for raptors. Garnett & Christidis [13] asserted that differences between lists stem from the use of contrasting species definitions. Indeed, application of different species definitions can lead to large disparities between lists [2,35]. Howard and Moore [21], HBW & BirdLife International [36], and eBird/Clements [22] follow the Biological Species Concept (BSC), emphasizing reproductive isolation [37]. The IOC World Bird List adheres to the Evolutionary Species Concept (F. Gill 2020, personal communication), which stresses evolutionary lineage and maintenance of specific identity [38–40], but in most cases results in the same species limits as the BSC. Many of the disagreements between lists are therefore among taxa that probably fall within the ‘grey zone’ (e.g. [41–44]) along the speciation continuum, in which lineages have diverged or hybridized just enough that either splitting or lumping could both be justified depending on the species definition. Cases such as these are not questions of whether a lineage has diverged, but the level of divergence required for species delineation based on the philosophy underpinning the decision [41], the amount of data available and types of divergence levels. Therefore, perhaps the most important aspect for reconciling the differences between the four lists is a system to help resolve disagreements resulting from the use of differing species definitions.

Even when lists agree on the definition of a species, they might apply the definition differently. The four lists vary in their use of non-peer-reviewed research; and, even when considering peer-reviewed publications, not all taxonomic changes are accepted by each list owing to a process of filtering by regional taxonomic experts. HBW & BirdLife International is the only list that applies the Tobias criteria [45] to delineate species-level taxa. These criteria involve weighting morphological and acoustical differences when compared with the nearest putative relative of a given taxon. Importantly, the Tobias criteria largely discount genetic inference for species delineation [45]. Consequently, even though several lists generally follow the BSC, differences probably arise from inconsistent application.

Collar [19] recognized that managers of these lists have different work schedules and publication frequencies, which certainly exacerbates taxonomic and nomenclatural differences. Howard and Moore contains the fewest species and is in most disagreement with the other lists, but, of the four lists, the most time has passed since it was last updated. The IOC World Bird List is updated twice per year, and both eBird/Clements and HBW & BirdLife International have been updated yearly. Volume 1 of Howard and Moore was published in 2013, and thus many of the disagreements might be an artefact of Howard and Moore not yet incorporating recent research. Even among lists updated annually, relatively small differences may go unnoted (e.g. moving a subspecies among taxonomic concepts) and lead to discrepancies that do not reflect differences in the interpretation of the evidence, but simply different stages of scientific advancement. Those types of discrepancies should be the easiest to eliminate when relying on Avibase taxonomic concept cataloguing. Significant work has been achieved along these lines, though it remains incomplete.

Salient patterns emerge from our results. Hotspots of disagreement appear to be owls occurring within southern Asia. Strigiformes contains the fifth-most disagreements of all orders across Aves, and is the most speciose of those top five. Owl families and genera further ranked among those containing the most disagreements. Buechley et al. [46] identified owls as the least-studied group of raptors and thus listed them as the highest priority for future research. Many of the owl concepts that are subjects of disagreement are small, nocturnal and confined to islands, which probably contributes to their lack of research attention. Our results suggest research into taxonomy of owls, particularly in the genera Otus and Tyto, should be a priority. Southern Asia has previously been highlighted as a priority for research and conservation because it contains high raptor species richness and many threatened and understudied species [34,46]. That Indonesia, India, and China contain the most disagreements further highlights the need for research in southern Asia. The abundance of islands within the Indonesian Archipelago probably contributes to this country containing the most disagreements. Allopatric isolation among islands is well recognized as an important contributor to speciation and increased endemism (e.g. [47–49]), but allopatry also exacerbates differing interpretations of species limits. Indonesia is the most raptor-diverse country on Earth [34]. This high diversity, coupled with a lack of research across southern Asia and the problems of interpretation inherent in allopatric populations, probably contributes to the many disagreed upon taxonomic concepts in Indonesia.

The focus of conservation efforts is often at the species level; however, subspecies represent considerable genetic and ecological diversity and thus often warrant conservation action [36,50]. For example, the Puerto Rican sharp-shinned hawk (Accipiter striatus venator) is critically endangered, yet currently considered a subspecies by all four world bird lists. Extinction of this subspecies would result in the loss of the only resident accipiter in Puerto Rico [51]. Despite situations where subspecies conservation is obviously warranted, there is some agreement that conservation of species generally supersedes that of subspecies (e.g. [10–12]; but see [13]). Indeed, although general patterns are hard to identify, there are several instances of changes in taxonomic rank affecting the conservation of a given taxon [14,26]. Species assignment is, therefore, influential for the allocation of finite conservation resources.

Several disputed concepts are ranked as subspecies on at least one list, perhaps lowering their conservation priority relative to undisputed species. For example, three threatened owl concepts are considered to be subspecies by all lists except HBW & BirdLife International. One such taxon, the Annobon scops owl (Otus feae, according to HBW & BirdLife International) is categorized by IUCN as critically endangered, with taxonomy justifiably listed as a research priority [52]. Such taxa with disputed species status are thus especially crucial research priorities. Importantly, because IUCN follows HBW & BirdLife International in performing Red List assessments, we can only assess the conservation status of taxa listed as species by HBW & BirdLife International. This limitation highlights the influence of one list over others in setting global conservation priorities.

Raptors represent roughly 550 of the world's 10 000 bird species (5.5%). Our analysis therefore only highlights a small fraction of the discrepancies to be addressed in reconciling the four world bird lists for all birds. Further, we examined species-level taxonomic concepts, whereas discrepancies among subspecies will also need to be addressed. Based on the proportion of species-level taxonomic concepts in agreement among all four world bird lists, the five raptor orders are spread throughout the rankings of avian orders (figure 5). Our work with raptors, therefore, provides a good example of the challenges that exist for reconciling the four world bird lists for the remaining orders.

Our results highlight the importance of clarity in the taxonomic concepts employed by checklists, researchers and databases. Indeed, eBird is perhaps the largest and most influential bird monitoring programme in existence (e.g. [53–55]), making integration with taxonomic concepts employed by the eBird/Clements list, especially pressing. If a group of researchers or birders follow a taxonomy that employs differing taxonomic concepts from the eBird/Clements list, data cannot be reliably exchanged with eBird, even if the scientific names are identical. This problem applies to any researchers employing not only differing taxonomies, but also conflicting taxonomic concepts. For example, eBird/Clements and IOC agree that the Augur buzzard (Buteo augur) and Archer's buzzard (B. archeri) are distinct species, yet HBW & BirdLife International and Howard and Moore consider Archer's buzzard to be a colour morph of the Augur buzzard, not a separate species. It is, therefore, ambiguous whether birds regarded as Augur buzzards by the HBW & BirdLife International taxonomy are Augur or Archer's buzzards according to eBird/Clements, seriously complicating transfer of data collected under these two different taxonomies.

Comparison of taxonomic concepts between lists using data compiled by Avibase will be useful in reconciling lists across all Aves. However, it is beyond the scope of this study to prescribe the process of reconciling the four world bird lists in detail. Many discrepancies will probably be reconciled simply by updating all lists to reflect current scientific knowledge. Instances of differing specific epithets resulting from acceptance of different original descriptions (e.g. Buteo burmanicus versus Buteo refectus) will need to be adjudicated by the International Commission on Zoological Nomenclature. Most of the world bird lists reference regional checklists such as those compiled by the North [56] and South [57] American Classification Committees. Future compilers of a single world bird list might also use these regional lists to lessen the potentially immense workload of managing a global bird list. Collar [19] speculated on a potential framework for compiling and maintaining a single world bird list. We agree with Collar's [19] assertion that a single world bird list will require input from ornithological societies and, importantly, a spirit of collaboration based on scientific rigor and goodwill.

Vigorous scientific debate over taxonomy is essential for conservation [17,58], and the four world bird lists highlighted here are examples of such discourse. However, in addition to transparency and rigour, lawmakers and conservationists require a degree of taxonomic consistency and clarity [1,18,59]. The current system, with limited agreement and uneven rates of revision across four competing authorities, creates unnecessary confusion. Reconciling these four lists and maintaining a resulting single list will be resource-intensive, requiring substantial effort and funding. The lack of funding for taxonomic research is often bemoaned, but this study highlights how fundamental this work is to conservation of the world's biota.

Data accessibility

Data are uploaded as the electronic supplementary material.

Authors' contributions

C.J.W.M.C., S.E.S., L.D., L.Ca., B.W.R., D.L.A. and J.J. conceived of the study. C.J.W.M.C. and D.L. outlined the first draft of the manuscript and all authors provided comments on the outline. C.J.W.M.C. wrote the first draft of the manuscript. B.W.R. created figure 3 and illustrations therein. C.J.W.M.C., L.D. and D.L. performed analysis. L.Cr., P.C.R., T.S.S. and M.J.I. ensured that descriptions of world bird lists and interpretations of their differences were appropriate. All authors critically refined the manuscript and gave final approval for publication.

Competing interests

We declare we have no competing interests.

Funding

We thank the M.J. Murdock Charitable Trust and donors to The Peregrine Fund for funding.

Acknowledgements

We appreciate F. Gill and D. Donsker for providing information regarding IOC, and S. Butchart for providing information regarding HBW & BirdLife International.