Species description years mislead estimates of Earth’s unknown biodiversity

How many species share our planet? The question sounds simple, but it remains one of the deepest unsolved problems in biology. For centuries, scientists have tried to close the Linnean shortfall — the gap between the species we have named and the species that actually exist — by plotting the years in which species were first described.

The year a species was described, the authors argue, often tells us almost nothing about when it was actually accepted as a species, and that mismatch systematically underestimates how much unknown biodiversity remains on Earth. The problem hinges on a technical distinction that most people never think about: the difference between an authority year and an acceptance year. Under the zoological naming rules governed by the International Commission on Zoological Nomenclature, the publication year of a name never changes, even when the taxonomic status of the animal does.

A population that spent a century listed as a subspecies, or buried as a synonym of another species, keeps the old description year when it is finally elevated to full species rank. Species accumulation curves built on description years therefore measure the age of scientific names, not the timing of their general acceptance as species — and the two can be separated by decades. The scale of this mismatch is striking.

In birds, only 32 of the 1,221 species added to the accepted list over an 11-year period represented genuinely new discoveries. The remaining 1,189 were previously known subspecies and populations whose taxonomic status changed. The authors illustrate the issue with the Californian climbing salamanders of the Aneides flavipunctatus complex.

Aneides flavipunctatus was described in 1870. Aneides iecanus was described in 1883 but treated as a synonym from 1925 until 2019. Aneides niger was described as a subspecies in 1948 but only recognized as a distinct species in 2019.

Aneides klamathensis was only accepted as a species in 2019, yet its correct authority year is 1948. A discovery curve would place all of these salamanders deep in the nineteenth and twentieth centuries, when in fact three of the four were only accepted within the last decade. To test whether this mismatch biases diversity estimates, the researchers compared two taxonomic snapshots taken roughly a decade apart for birds, amphibians, and mammals.

For birds, they used two versions of the BirdLife International checklist, from 2012 and 2022. For mammals and amphibians, they used the 2010 and 2023 versions of the IUCN Red List. These groups were chosen deliberately: they are among the best-studied organisms on Earth, with the vast majority of species already described, so methods based on discovery patterns should work best for them.

If the method were reliable, estimates based on the older taxonomy should closely match those based on the newer one. Instead, the researchers consistently found lower estimates of total diversity when relying on the older snapshot. The results were remarkably consistent.

For birds, the analysis of the 10,054 species in the old taxonomy suggested that a further 493 species could eventually be described. The new taxonomy contained 11,178 species — 1,124 more than the old one. At the order level, 15 of the 17 orders analyzed showed lower estimated diversity under the old taxonomy, a pattern that would be expected by chance only rarely (p = 0.0024 based on a binomial test).

The median increase in estimated total diversity across those 15 orders was 10 percent, with a range of 3 to 25 percent. For birds, mammals, and amphibians as whole classes, estimated diversity increased by 7, 10, and 11 percent respectively when comparing the old and new taxonomies. The researchers also ran a series of simulations to confirm that the bias behaves as expected.

Starting from the amphibian dataset, which showed the highest number of estimated undescribed species, they artificially increased the gap between authority year and acceptance year by subtracting random amounts from the empirical distribution of discovery intervals. The more biased the input data became, the fewer undescribed species the model predicted. In the empirical amphibian dataset, 13 percent of species remain undescribed based on the new taxonomy, but under the highest simulated biases, this dropped to 0.1 percent.

The drop was highly predictable, with consistently fewer undescribed species estimated as the magnitude of the bias increased. This confirms that the underestimation is not a quirk of the particular datasets but a direct consequence of the mismatch between description year and acceptance year. The geographic analysis revealed an even more troubling pattern.

The researchers looked at the authority years of all species added between the old and new taxonomies, reasoning that if authority year were informative of when species are considered full species, nearly all of these should have authority dates more recent than about 2005. Instead, they found that species added with centroids in Europe had a mean authority year of 1902, while those from Africa, Tropical Asia, North America, Oceania, and South America had mean authority years around 1951 to 1954. For mammals, authority years of the species added to Europe were significantly older than anywhere else.

For birds, nearly all pairwise comparisons were significant, with Europe having the oldest authority years followed by North America. This means the bias is most severe in the well-studied regions — precisely where previous analyses based on discovery curves have predicted the fewest undescribed species. This geographic pattern calls into question analyses that use species discovery years to infer spatial patterns in undescribed species richness, such as those presented by Moura and Jetz (2021) or Ondo et al. (2024).

These analyses predict a higher number of undescribed species in regions where the year of discovery is closer to the year of acceptance, and a lower number in regions with a larger difference between the two. The new study shows that these analyses will mainly underestimate the number of undescribed species in the regions where they predict the fewest — the well-studied temperate zones — while the tropical regions where they predict the most may be somewhat more accurate. The overarching conclusions of those studies, such as the claim that tropical Asia likely harbors many undescribed species, remain plausible, the authors note.

But the discovery curves provide little additional support for such conclusions beyond what expert biological knowledge already suggests. The authors are careful to note that the problem is not that all models are useless. As George Box famously said, all models are wrong, but some are useful.

The models examined here generally only work on very well-studied groups and may not converge to finite values unless the majority of species in a group is already known. In cases where approximate conclusions are already known a priori, the built-in bias of accumulation curves may not matter. A recent imprecise estimate of the number of birds that went extinct due to humans but did not leave any fossils, for example, used accumulation curves as a lower bound of the number of unknown extinct species — and the fact that the method is known to be an underestimate arguably makes it more suitable for this limited use.

Similarly, a study stressing that the discovery rate of fungi is still accelerating stands even though its use of authority year means it underestimates the size of the acceleration. The problem only extends to studies whose conclusions rest on a face-value reading of the results. The fix, at least in principle, is straightforward.

The problem should largely disappear if authority year in all analyses is replaced by the year a species was raised to species status, when it was formerly treated as a subspecies or synonym. The researchers therefore advocate that taxonomic databases and monographs should include additional information on the taxonomic history of each taxon, including when it was raised to species level — a call that many other researchers have also made. Until such data are available, the authors hope that less problematic methods for inferring undescribed diversity will be used instead of analyses of patterns in authority year.

The broader lesson for biodiversity science is sobering: a good statistical fit to the data does not necessarily imply that a model is informative. This study provides a real-world example of how a seemingly plausible model can yield misleading insights when its underlying assumptions are violated — a caution that extends far beyond the counting of species.

Subject of Research: Limitations of using species description years to estimate undescribed global species diversity

Article Title: Global patterns in species description year cannot reliably estimate undescribed diversity

Article References: Faurby, S., & Farooq, H. (2026). Global patterns in species description year cannot reliably estimate undescribed diversity. Frontiers of Biogeography, 19, Article e171197. https://doi.org/10.21425/fob.19.171197

Image Credits: AI Generated

DOI: 10.21425/fob.19.171197

Keywords: biodiversity, species description, taxonomic inflation, Linnean shortfall, species accumulation curves, authority year, undescribed species, biogeography, conservation, taxonomy, birds, amphibians