How Many Species Are There on Earth and in the Ocean?

How many species share this planet with us? It sounds like a simple census question, but it intersects where curiosity meets urgency. We're losing biodiversity fast, yet we still don't know the size of the living library we're trying to save. For decades, estimates have swung wildly because we've never sampled the biosphere in a way that lets us scale up cleanly. Part of the problem is methodological. If you try to count everything directly, you run headfirst into incompleteness; most taxa are poorly described, so any global total built from observed species is shaky. If you rely on habitat-area rules of thumb, you assume that patterns seen in one group or region transfer neatly to others, which often fails. When you ask experts, you get experience and intuition, but also subjectivity and echo chambers. Opinions vary by orders of magnitude across groups. The result is a debate with more heat than light. A few years ago, Camilo Mora and colleagues offered a different route. Instead of chasing species directly, they suggested looking higher up the tree. Phyla, classes, orders, families, genera — those categories accumulate through time in a smoother, more gradual way than species do, and they're much closer to being complete. If you can describe the shape of those higher-taxon curves, Mora argued, you can infer the unseen bottom of the pyramid: total species. It's a clever flip. Use what we know well to anchor what we don't. To make that work across the tree of life, they pulled together a massive backbone of classifications for about one point two million currently valid species. The core came from the Catalogue of Life and the World Register of Marine Species; the year each taxon was named came from the Global Names Index. For bacteria and archaea, they leaned on the List of Prokaryotic Names with Standing in Nomenclature. To avoid recent lags in data entry that artificially flatten curves, they truncated the analysis at two thousand six. Then comes the modeling, and it's more disciplined than hand waving. For each rank from phylum down to genus, they built a time series of how many higher taxa had been named by each year since one thousand seven hundred fifty-eight — that's the Linnaean starting gun — with anything pre-one thousand seven hundred fifty-eight absorbed as a one-time jump at the beginning. They fit six kinds of asymptotic growth curves to those trajectories, from the familiar Michaelis–Menten shape to modified Weibull and negative exponential forms. To keep the early years authentic, they only accepted cutoff points that had at least a decade of data and where five of the six models actually converged. Rather than pick a favorite curve, they used small-sample Akaike weights to average across them, landing on a consensus asymptote and an uncertainty band for each rank. All that gives you an estimate for how many phyla, how many classes, and how many orders there probably are — not just how many we've named. The bridge to species is a simple regression: relate those asymptotic counts to the numerical rank and project down one more step to the species level. They validated that bridge on eighteen groups where we think we know the totals pretty well. The plotted points hugged the one-to-one line. In other words, when we can check it, the method doesn't go off the rails. There are places where the logic breaks down, and they acknowledge this. Prokaryotes don't show clean asymptotes in their higher-rank naming curves, and the very idea of a "species" in bacteria and archaea isn't the same as in animals or plants. For those groups, Mora's team used the raw higher-taxon counts as conservative lower bounds rather than pretending there is an asymptote where none was evident. Treat the prokaryote numbers as a floor, not a ceiling. A big worry in any name-based analysis is synonymy — the same species named twice under different labels. So Mora and colleagues didn't guess; they asked the people doing the naming. They contacted four thousand seven hundred seventy-one taxonomists, reached about two thousand nine hundred thirty-eight of them, and received five hundred forty-eight responses. The pattern they report is intuitive and important: synonymy is a headache at the species level, around eighteen percent on average in the survey, and it dwindles as you go up the hierarchy, to about one percent at the phylum level. That matters because the method leans on higher ranks precisely to sidestep the churn at the species boundary. Could the slowdown in higher-taxon discovery simply reflect waning effort? The data argue no. In their time series, the rate of new higher-taxon descriptions falls off, while the rate of new species descriptions stays relatively steady. That divergence is what you would expect if the big classifications — the families and orders — are nearing completeness, even as we keep finding species to file into them. A different check asks, what if our catalogs are missing whole chunks of higher taxonomy? The sensitivity analyses show the method tends to underestimate diversity when higher-taxon inventories are incomplete, with the bias most pronounced when genera are missing. They highlight two groups in particular — Chromista and Fungi in the ocean — where genus-level data were thin, so predictions there deserve extra caution. So, what does the tally look like when you run this framework across the eukaryotes — the animals, plants, fungi, and protists? The headline number is striking: about eight point seven four million eukaryotic species on Earth, with an uncertainty of plus or minus one point three million. Of those, roughly two point two one million live in the oceans, with a margin around zero point one eight million. And here's the gut punch: by their accounting, about eighty-six percent of Earth's species and ninety-one percent of ocean species have yet to be described. We've named a lot — on the order of one point two four million are catalogued in central databases — but we've barely dented the total. Breaking that down gives you a sense of scale by kingdom. Animals dominate the eukaryote ledger, at an estimated seven point seven seven million species. Fungi come in around six hundred eleven thousand. Plants — land plants plus their allies as defined here — are about two hundred ninety-eight thousand. Protozoa, the single-celled eukaryotes lumped under that historical label, are roughly thirty-six thousand four hundred. Chromista — another catch-all for certain algae and protists — lands near twenty-seven thousand five hundred, though the uncertainty is large because the genus-level data were patchy. Each of those numbers carries its own caveats about classification, but together they paint a consistent picture: the bulk of eukaryotic diversity is still out there, unnamed. Prokaryotes sit outside this main estimate for good reasons. Using their conservative bounds, there are on the order of ten thousand named bacterial and archaeal species globally, but Mora's group stresses that this is not a statement about microbial diversity per se. It's a mirror held up to our nomenclature. The species concept in microbes is different, the rate of description hasn't settled into clean asymptotes, and the databases don't support the same kind of projection. In the ocean, their lower-bound numbers are even smaller. Take them as a reminder of limits, not as a verdict on the microbial world. One more number set puts the challenge in human terms. At the current pace — about six thousand two hundred eukaryote species described per year — finishing the catalog would take on the order of one thousand two hundred years. To do it faster with today's productivity would require something like three hundred three thousand taxonomists. If you price a species description at a rough average of forty-eight thousand five hundred US dollars, the global bill to finish the job reaches about three hundred sixty-four billion dollars. Big numbers, yes. But they scale with the scale of the unknown. None of this is a blank check for the method. It stands or falls with the stability and completeness of higher taxonomy. If phyla and orders get reshuffled wholesale, the anchors move. Protozoa and Chromista, as used here, are historical groups rather than clean phylogenetic classifications, which can bend comparability across kingdoms. The approach will be most wrong where our higher-taxon inventories are thinnest — exactly the places we should be focusing effort. What I like about Mora's framework is that it's falsifiable. It makes explicit claims: higher-taxon curves are nearing their asymptotes; their asymptotes predict species totals; those predictions align with well-known groups; the method is relatively insensitive to the tangle of species-level synonyms; and it underestimates most when higher-taxon data are missing. As Chen and colleagues showed in a different context, that mix — clear structure plus honest uncertainty — is what lets a field advance. So where does this leave us? With a defensible bottom line and a to-do list. Roughly eight point seven million eukaryotic species on Earth, about two point two million in the sea, and the vast majority still unnamed. Prokaryotes remain outside the reach of this particular lens. The method isn't perfect, but it's transparent about its assumptions, and it shifts the conversation from guesswork to patterns we can test. The payoff isn't just a prettier number in a headline. It's a recalibrated baseline for conservation and exploration. If most species are unknown, then "no news" from a habitat is not good news — it's a sign we haven't looked hard enough. The path forward is practical: keep building and cleaning the global catalogues, invest in taxonomy where higher-taxon inventories are weakest, and tighten the links between naming and data systems so accumulation curves tell us something real. If we want to know what we're losing — and what we can still save — we need the map. This gets us closer.