The global state of biodiversity and extinction

How fast is biodiversity being lost? The question sounds simple but answering it requires careful thought about what is being measured. Extinction counts the complete loss of species, which is final but slow to document. Population declines measure changes in abundance, which happen faster and matter more for ecosystems but are harder to aggregate. Extinction risk assessments estimate how likely species are to be lost in the future. This lesson examines the three main global yardsticks, the Living Planet Index, the IUCN Red List and estimates of extinction rates, together with the major global assessments that synthesize them, and then looks at the status of species in Canada and the United States.

The IPBES global assessment

In 2019 the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES), an independent body with more than 140 member states that plays a role for biodiversity similar to that of the Intergovernmental Panel on Climate Change for climate, published its Global Assessment Report on Biodiversity and Ecosystem Services. Prepared by about 145 expert authors drawing on some 15,000 sources, including Indigenous and local knowledge, it remains the most comprehensive synthesis of the state of nature. Its headline findings were that about one million animal and plant species, out of an estimated eight million, are threatened with extinction, many within decades; that human actions have significantly altered about 75 per cent of the land surface and 66 per cent of the ocean area; that more than 85 per cent of wetland area present in 1700 had been lost by 2000; and that the global rate of species extinction is already at least tens to hundreds of times higher than the average over the past ten million years and is accelerating.

The assessment ranked the direct drivers of change. On land, changes in land and sea use, mostly the expansion of agriculture, had the largest impact; in the ocean, direct exploitation, mainly fishing, came first. Climate change, pollution and invasive alien species followed. These direct drivers are in turn driven by indirect drivers: population growth, consumption, trade, technology and governance. The report concluded that business as usual would not meet global goals, but that transformative change across economic, social, political and technological factors could still reverse declines.

The Living Planet Index

The Living Planet Index (LPI), maintained by the Zoological Society of London and published every two years by WWF in the Living Planet Report, tracks the average change in the size of monitored populations of vertebrates: mammals, birds, reptiles, amphibians and fishes. The 2026 report, released in October 2026, found an average decline of about 73 per cent in monitored populations between 1970 and 2022, roughly 2.5 per cent a year on average. Freshwater populations showed the steepest decline, about 85 per cent, followed by terrestrial populations at about 69 per cent and marine populations at about 59 per cent. Among regions, Latin America and the Caribbean showed the largest average decline, about 95 per cent, and Africa about 80 per cent, while North America showed about 28 per cent and Europe and Central Asia about 23 per cent.

The LPI is often misreported as meaning that 73 per cent of all wild animals have disappeared. It does not mean that. The index is the geometric mean of population trends: it averages the relative change in each monitored population, so a population that falls from 100 to 10 animals and one that falls from 1,000,000 to 100,000 count equally. It is therefore an index of the average rate of change, not a measure of total animal numbers lost. The 2026 report also noted that about half of the monitored populations were stable or increasing; the average decline is driven by populations that have fallen steeply. And the index is biased toward well-studied regions and groups, since it can include only populations with repeated counts. Smaller regional declines in North America and Europe do not mean those ecosystems are healthy; much of their wildlife had already been lost before 1970, the baseline year.

import math
# How the Living Planet Index aggregates trends: the geometric mean of relative change.
# Five invented populations counted in 1970 and 2022.
pops = {"large fish stock": (2_000_000, 1_500_000), "frog A": (400, 20),
        "songbird": (12_000, 6_000), "river turtle": (900, 90), "seal colony": (3_000, 4_500)}
ratios = [end / start for start, end in pops.values()]
for (name, (s, e)), r in zip(pops.items(), ratios):
    print(f"{name:17s} {s:>10,} -> {e:>10,}  ratio {r:5.2f}")
geo = math.exp(sum(math.log(r) for r in ratios) / len(ratios))
total_start = sum(s for s, e in pops.values()); total_end = sum(e for s, e in pops.values())
print(f"index (geometric mean of ratios): {geo:.3f}, an average decline of {1 - geo:.0%}")
print(f"change in total individuals:      {total_end / total_start:.3f}, a decline of {1 - total_end / total_start:.0%}")

large fish stock   2,000,000 ->  1,500,000  ratio  0.75
frog A                   400 ->         20  ratio  0.05
songbird              12,000 ->      6,000  ratio  0.50
river turtle             900 ->         90  ratio  0.10
seal colony            3,000 ->      4,500  ratio  1.50
index (geometric mean of ratios): 0.309, an average decline of 69%
change in total individuals:      0.749, a decline of 25%

In this invented example the index shows an average decline of about 70 per cent, because three of the five populations fell steeply, yet the total number of animals fell by only about a quarter, because the largest population, a fish stock, declined modestly. Neither number is wrong; they answer different questions. The LPI asks how populations are faring on average, which is a sensitive warning signal. Total abundance asks how much wildlife is left, which matters for ecosystem function. Both are needed.

The IUCN Red List

The IUCN Red List of Threatened Species, maintained by the International Union for Conservation of Nature since 1964, is the world's most widely used system for assessing extinction risk. Since 1994 it has used quantitative criteria, revised in 2001, that can be applied to any species except microorganisms. The July 2026 update listed 175,909 assessed species, of which 49,505, about 28 per cent, were classified as threatened with extinction. Assessments are far from complete: only a small fraction of described species have been assessed, and coverage is much better for vertebrates and some plant groups than for insects and fungi. Nonetheless the Red List shows that about 41 per cent of amphibians, about a quarter of mammals, about 44 per cent of reef-building corals, more than a third of sharks and rays, and about one in eight birds are threatened.

The Red List places each assessed species in one of nine categories. Extinct (EX) means there is no reasonable doubt that the last individual has died. Extinct in the Wild (EW) means the species survives only in cultivation, captivity or as a naturalized population outside its past range. Critically Endangered (CR), Endangered (EN) and Vulnerable (VU) are the three threatened categories, indicating extremely high, very high and high risk of extinction in the wild. Near Threatened (NT) species are close to qualifying, Least Concern (LC) species are widespread and abundant, Data Deficient (DD) species lack enough information for an assessment, and Not Evaluated (NE) species have not been assessed.

The five criteria

A species qualifies for a threatened category if it meets any one of five criteria, labelled A to E. The thresholds for each criterion differ among the three threatened categories, and an assessment uses the highest category for which any criterion is met.

For criterion A1, which applies when the causes of decline are understood, reversible and have ceased, the thresholds are higher: 90, 70 and 50 per cent. Criterion A uses generation length because a decline of 30 per cent over ten years is far more serious for a long-lived tortoise, whose generation time may be several decades, than for a mouse; using three generations scales the time window to the species' biology, up to a maximum of 100 years. Criterion B targets species with small ranges that are also fragmented or declining; extent of occurrence is the area of the smallest convex polygon containing all known sites, while area of occupancy is the area actually occupied, measured on a grid of 2 kilometre by 2 kilometre cells. Criterion E is used rarely, because it requires a population model good enough to estimate the probability of extinction, a topic for lesson 6.

import math
# Criterion A2: estimate the decline over three generations from survey counts.
# Invented counts of mature individuals of a long-lived turtle; generation length 25 years.
years = [1990, 1996, 2001, 2007, 2012, 2018, 2024]
counts = [3100, 2850, 2400, 2200, 1900, 1650, 1500]
gen = 25
window = max(10, 3 * gen)                       # 75 years
# least-squares fit of log(count) on year gives an exponential rate of change
n = len(years); xm = sum(years) / n; ym = sum(math.log(c) for c in counts) / n
slope = sum((x - xm) * (math.log(c) - ym) for x, c in zip(years, counts)) / sum((x - xm) ** 2 for x in years)
print(f"estimated rate of change: {slope * 100:.2f}% per year (exponential)")
decline = 1 - math.exp(slope * window)
print(f"projected decline over {window} years (3 generations): {decline:.0%}")
cat = "CR" if decline >= 0.8 else "EN" if decline >= 0.5 else "VU" if decline >= 0.3 else "not threatened under A"
print("criterion A2 category:", cat)

estimated rate of change: -2.23% per year (exponential)
projected decline over 75 years (3 generations): 81%
criterion A2 category: CR

The surveys cover 34 years, less than half of the 75-year assessment window, so the decline must be projected using an assumed model, here an exponential decline at a constant rate. The IUCN guidelines allow such projections, including combinations of past and future periods (criterion A4), but the choice of model matters: if the decline were linear rather than exponential, the projection would differ. Here the projected decline of about 81 per cent lies just above the Critically Endangered threshold of 80 per cent, so a slightly different model or a few more years of data could move the species to Endangered; assessors document such assumptions and often report a range of plausible categories, here EN to CR. Note also that a decline of about half over 34 years would look serious on any time scale, but it is the three-generation window that places this species in a high category.

Extinction rates

How does the present rate of extinction compare with the past? Paleontologists estimate that the average lifespan of a species in the fossil record is about one to ten million years, which implies a background extinction rate of roughly 0.1 to 1 extinction per million species per year, often written E/MSY. With about 8 million species, a background rate of 1 E/MSY would mean about eight extinctions a year worldwide. Since 1500, more than 900 species have been documented as extinct by the IUCN, and the true number is certainly far higher because most species are unknown and extinctions are hard to confirm. For well-studied groups such as birds and mammals, studies such as those of Stuart Pimm and colleagues in 2014 and Gerardo Ceballos and colleagues in 2015 estimated current extinction rates at roughly 100 to 1,000 times the background rate.

Earth has experienced five mass extinctions, events in which roughly three quarters or more of species disappeared within a geologically short time, most recently at the end of the Cretaceous 66 million years ago. Many scientists argue that a sixth mass extinction is under way or imminent. Others caution that current documented losses, while alarming, are still far smaller than the past mass extinctions, and that the term should be used carefully; but they agree that the rate of loss is far above background and that if current trends continue, losses would reach mass-extinction levels within centuries. The distinction matters less than the trajectory, and the trajectory can still be changed.

A further complication is extinction debt, a term introduced by David Tilman and colleagues in 1994. When habitat is destroyed, some species do not disappear immediately; they persist for decades or centuries in populations that are no longer viable. Their eventual extinction is a debt that has already been incurred. Extinction debt means that today's extinctions reflect habitat loss that happened long ago, and that the full consequences of recent losses have not yet appeared. It also offers an opportunity: species in debt still exist and can sometimes be saved by restoring habitat in time.

North American losses

North America has its own record of extinctions since European colonization. The great auk, a flightless seabird of the North Atlantic, was hunted to extinction by 1844; the last known birds were killed off Iceland, but large colonies had existed on Funk Island off Newfoundland. The Labrador duck disappeared by about 1878. The passenger pigeon, once probably the most abundant bird in North America with a population in the billions, was reduced by market hunting and forest clearing; the last individual, named Martha, died in the Cincinnati Zoo in 1914. The Carolina parakeet, the only parrot native to the eastern United States, followed in 1918. Freshwater species have fared especially badly: dozens of species of freshwater mussels, snails and fishes in the southeastern United States have disappeared as rivers were dammed and polluted.

In Canada, COSEWIC has assessed several species as extinct, including the great auk, Labrador duck, passenger pigeon, sea mink, deepwater cisco of the Great Lakes, blue walleye, Banff longnose dace and eelgrass limpet, and many more as extirpated, meaning they no longer exist in the wild in Canada but survive elsewhere. Examples of extirpated species include the greater prairie-chicken, the paddlefish and, at times, the black-footed ferret and swift fox, both later reintroduced. Canada's periodic Wild Species reports on the general status of species, prepared jointly by federal, provincial and territorial governments, have found that roughly one in five assessed species is at some level of risk nationally, while most species in the country's vast northern regions remain secure.

Signs of hope

Global status assessments are not uniformly grim. The Red List includes many species whose status has improved because of conservation action, and a 2024 study in Science by Penny Langhammer and colleagues, a meta-analysis of 186 studies, found that conservation interventions improved the state of biodiversity or slowed its decline in about two thirds of cases, compared with no action. Examples include the recovery of the bald eagle and peregrine falcon after DDT was banned, the return of humpback whales after commercial whaling ended, and the improvement of the numbat in Australia from Endangered to Near Threatened in the 2026 Red List update after programs to control introduced cats and foxes. The message of the global assessments is not that conservation fails, but that it has been applied at too small a scale relative to the pressures.