A Species With Only Three Digits Left, Yet Scientists Can't Pin Down an Exact Number

August 31, 2026 11 min read A Species With Only Three Digits Left, Yet Scientists Can't Pin Down an Exact Number Mangrove Conservation Foundation (MCF) Dedicated to protecting wetlands and their biodiversity, practicing a socially participatory nature conservation model. Same name on all platforms, welcome to comment and exchange ideas. 11 min read Editor's Note: The Mangrove Conservation Foundation (MCF) is China's first publicly fundraising environmental foundation initiated by the public. It is dedicated to protecting wetlands and their biodiversity, practicing a socially participatory nature conservation model. Paibian has collaborated with MCF multiple times previously, including otter merchandise charity sales, World Migratory Bird Day promotional videos, the Mangrove Ecological Park "Red Person Market" offline event, and bird collision prevention New Year window decal design charity events. We hope this article helps everyone learn more about biodiversity conservation. Have you ever heard of a critically endangered bird called the spoon-billed sandpiper? It's okay if you haven't. Simply put, there are only three digits left of this bird worldwide. If you suddenly search online, you'll likely see a variety of numbers: Only about 443 spoon-billed sandpipers left globally; The IUCN Red List records 490 mature individuals; Winter surveys recorded only 330-340 mature individuals; Possibly fewer than 250. Many people are baffled at this point: Exactly how many spoon-billed sandpipers are there? Whose data is more authoritative? The International Union for Conservation of Nature (IUCN) Red List, the East Asian-Australasian Flyway Partnership (EAAFP) website, and even national scientific research institutions all give different figures. It's not just you folks. With so many different numbers popping up for the same bird, even we occasionally get tangled up when doing project communications. Let's tell a story below; after reading it, you might understand the tricks of the trade. Counting Books in a Flea Market. Imagine you are at a flea market, and your task is to figure out exactly how many copies of an extremely niche book are preserved in the market. The most "brute-force" method is to search stall by stall and row by row. You search three times, counting 278, 229, and 285 copies respectively. Thus, you've unlocked the first method—the Direct Count Method. In spoon-billed sandpiper research, it also has a specific name: the Scan Survey Method 1. The main idea is to count multiple times and take the highest number. But you also know full well: those pressed at the bottom of boxes, covered by other books, some copies you simply can't see... naturally, you haven't counted them all. Therefore, your hard-earned direct count can only yield a fuzzy upper limit: at most 285 copies these few times. Since searching book by book is like looking for a needle in a haystack, your clever brain starts working: why not first figure out the total number of old books in the market (this is easier, perhaps the market management has tallied the total stalls and average books per stall), then randomly sample a few stalls to see what proportion this niche book accounts for, and multiply it out. For example, if the market has 10,000 old books, and sampling shows this niche book makes up about 3%, then the total hidden copies are roughly 300. This is the Proportional Estimation Method 2. The benefit of this method is that you can estimate even if you can't count them all, but the downside is also obvious: the proportion comes from random sampling; in reality, it could be 3% or 7%, so the uncertainty is high, and the result can only serve as a reference. Moreover, both the direct count method and the proportional estimation method share a fatal flaw: you don't know how many you missed. So you decide to change your thinking, shifting from "estimating a rough figure" to "calculating a value." To achieve this, you need to create a known ratio. You go to the market again, this time bringing a stack of red bookmarks. You place a bookmark in every niche book you find, totaling 20 bookmarks. A few days later, you search again and find 120 copies, 5 of which still have the bookmark. Using the representativeness of the sample, you deploy this proportional structure, pay attention, remember this: Then: Estimated this way, the total number of niche books is about 120×20÷5=480. You don't need to care how many other books are in the market; by setting a ratio, you can calculate the number of niche books. Brilliant, right! This is the Mark-Recapture Method 3, specifically called the Lincoln-Petersen Estimator, and it is also the core method for spoon-billed sandpiper population estimation. Swap bookmarks for leg flags, and the flea market for mudflats, and the principle is exactly the same. From the proportional estimation method to the mark-recapture method, the accuracy of your data has greatly improved, but you still feel uncertain. So, you steel your heart and make a major decision: continue going to the market to insert bookmarks. But this time, add the element of time—ten years. A minor upgrade is that the new bookmarks are numbered, with time and other information, making the data more detailed. Congratulations, you've evolved to the Jolly-Seber Method, which, in addition to estimating the total population, can also estimate the Survival Rate and Recruitment Rate. These correspond respectively to "whether the previously marked books are still there this time" and "how many unmarked books were found this time." Actually, an even more powerful method is the Pradel Model, which can directly estimate the rise and fall of the entire collection, but we won't expand on that today. Because you face a new problem: the market is still too small, this niche book is already rare, and there are only so many bookmarks distributed. The data volume simply cannot support the two advanced methods, Jolly-Seber and Pradel. How can you draw more reliable conclusions from a limited amount of data? After thinking it over, you pull out another trick—Statistical Modeling. Different models can provide different kinds of help; since you're already using modeling,