麻豆传媒

There are tens of thousands of species of orchid, making it the second most species-rich plant family on Earth. Their exceptional diversity is driven in part by tightly evolved relationships with pollinating insects and root-boosting fungi. Each species has its own favourite collaborators, a feature that has driven orchids to success, but that also makes them especially vulnerable to extinction when these relationships are perturbed.

So, when the delicate, lemon-scented mignonette leek orchid, unseen since the 1930s, turned up again in 2020, there was cause for celebration. Recent rediscoveries like these 鈥� from the 聽to the 聽hummingbird 鈥� are seemingly rare moments of joy amid the rapid loss of many plants and animals around the world. Globally, the number of species deemed to be at high risk of extinction has increased , and plant species in particular are being lost as much as 500 times faster than would be expected under normal evolution.

Yet, evolutionary biologist at Stockholm University, Sweden, doesn鈥檛 see rediscoveries as an unmitigated joy. As she told a recent summit at the Royal Botanic Gardens at Kew, London, the rediscovery of species also constitutes a failure. Extinct species are found again far more often than you might think; at the same time, problems with how and why we declare species extinct are also leading us to underestimate the scale of extinction worldwide.

麻豆传媒 spoke with Humphreys about why we need to stop thinking in terms of a species being either extinct or not, and instead take a more nuanced approach to a philosophical problem that has vexed thinkers for thousands of years: how do you prove an absence?

Penny Sarchet: I was surprised to hear you tell the earlier this year that every rediscovery is a sort of failure, because it means we鈥�ve got something wrong.

Aelys Humphreys: People react strongly to that message. I don鈥檛 want to give the impression that rediscoveries are a bad thing per se: they mean a species is not extinct, so conservation efforts can be renewed. Fundamentally, it鈥檚 a good thing.

But the flip side is that an expert decided that a species was extinct and then it turned out that it wasn鈥檛. That can also lead to the erroneous withdrawal of conservation resources from that species. People have called this the Romeo error.

Is that named after Shakespeare鈥檚 Romeo?

Yes. He thought that Juliet was dead, but she鈥檇 just taken a sleeping potion, and then he killed himself. That鈥檚 what I find interesting about rediscoveries: their potentially grave implications.

So, how high are rediscovery rates?

We found it is about 35 per cent for plants and 36 per cent for mammals.

Do you have any favourite rediscovery stories?

One is Psoralea cataracta, a South African plant in the pea family, which is very distinctive and unusual as it has very long peduncles 鈥� the stem that leads to the flower. This species had only ever been collected once, in 1804. No one had recorded it since then, and it was declared extinct in 2008, but 12 years later it was found by a junior botanist. It鈥檚 fantastic that something can be so poorly known, for so long, but is actually still out there.

Why is it so hard to say for sure that something isn鈥檛 there?

It鈥檚 very hard to prove absence. This philosophical question has troubled us since ancient times across a lot of different fields. In medicine, for example, just because no side effects have been recorded in a new treatment, it doesn鈥檛 necessarily mean that there are none. And it鈥檚 the same with law: just because there鈥檚 no evidence that you鈥檙e innocent, does that mean that you鈥檙e not innocent? Absence of evidence is not evidence of absence.

How do conservationists typically determine if a species is extinct?

It requires repeated surveys in an area where the species is known to have occurred 鈥� and ideally in similar habitats where it might have moved to. These surveys must be carried out by experts who know what they鈥檙e looking for, over a number of years, to reduce the chance that they weren鈥檛 just unlucky once or twice in not finding it.

When you consider all those factors, it鈥檚 somewhat amazing that we ever declare anything extinct.

Yes. Especially as species that are declared extinct may have been quite rare for a long time before that, and many of them are in remote areas that are difficult to access. It鈥檚 usually done by experts who feel they have the authority to declare something extinct because they鈥檝e been returning to the same area for a very long time and they feel confident enough to say that if the species was here, I would have found it.

But there is a reluctance to declare species extinct out of the fear of getting it wrong, with the implications of resources being removed and nobody continuing to look for it. So, even though many extinct species are later rediscovered, we are underestimating extinction, because there are plenty of species that people think might be extinct but are not taking the final step to say so.

On top of that, it鈥檚 been estimated that 3 in 4 plants that haven鈥檛 yet been described scientifically are already . That suggests a very high rate of so-called 鈥榙ark extinction鈥� 鈥� species that go extinct without us even knowing about them.

Absolutely. That鈥檚 a staggering statistic. It鈥檚 also been estimated that around 2000 new plant species are described scientifically every year. So, new species description is happening constantly, and if 75 per cent of them at the time of scientific description are already endangered or have a high extinction risk, that鈥檚 really striking.

In another study, botanists quantified how long new-to-science species that have been collected lie around in herbaria [plant specimen libraries] before they are actually described as species. They found an average time of 24 years.

Does that mean there鈥檚 a realistic chance that some of those are extinct by the time we realise that they鈥檙e new species?

Exactly. By the time it鈥檚 been described, it is already extinct.

To tackle these issues, you鈥檙e a proponent of introducing a more probabilistic, less binary, way of viewing extinction.

The way that species have traditionally been declared extinct is within a categorical system: 鈥渆xtinct鈥� or 鈥渘ot extinct鈥�. But in most [other] scientific practices, we have some kind of estimate of how sure we are.

So, this alternative way of assessing extinction gives a probability that a species is extinct, ranging between 0 and 1. Then we can assign some kind of cutoff at which we鈥檙e comfortable saying it鈥檚 extinct. For example, if there鈥檚 a probability of 0.9 that a species is extinct, we can be quite confident. But if we have a probability of 0.6, we may say that, based on the information that we currently have, we鈥檙e not sure. There is a gradient between definitely not extinct, definitely extinct and everything in between.

The last-ditch plan to save coral reefs from utter destruction

Bleaching has devastated reefs around the world, raising fears of an irreversible shift. Yet new interventions have revealed that corals can be remarkably resilient if we can give them enough help to recover

Could this approach help solve the rediscovery problem?

It offers a way out of this paradox where we want things to be rediscovered, but we ideally wouldn鈥檛 have declared them extinct in the first place. Using this approach, if there are rediscoveries, we can say well, actually, it only had a 75 per cent chance of being extinct. It avoids us sending out a very certain message that then turns out to be wrong 鈥� which can give the impression that scientists are exaggerating the extinction crisis.

And if you have a more quantitative scale for extinction likelihood, you can estimate this mathematically, and someone else with new evidence can go in and re-evaluate it. You鈥檝e got this model and you add the new evidence and you run the model again. For example, if you find that there was another sighting of a species five years ago, you can ask: how does that influence the probability that the species is extinct? The model will tell you. You don鈥檛 have to rely on individual experts taking the decision that a species is extinct.

What other kinds of information are fed into these models?

Some models take into account unsuccessful searches. Some models use information on how much of the possible habitat has been searched. There are also things like detectability. How easy is this species to find and identify? Are there other things in its habitat that look similar? Is it a big tree and it鈥檚 always there, or is it a herbaceous plant that might not be detectable if you go there in midwinter?

Free newsletter

Sign up to The Earth Edition

Sign up to newsletter

You鈥檝e found that when expert assessors use models, extinct species are much less likely to be rediscovered.

I did a with ecologist at the University of Queensland, Australia, in 2023, of species that have been listed as extinct. We uncovered 155 species that had had their extinction status evaluated by a probabilistic extinction model. Of these, only 2.5 per cent had been rediscovered.

That鈥檚 really promising. Although, these probabilistic models tend to be used on recently extinct species, which have had less time to be rediscovered. The best approach is to combine expert opinion with these models. If an expert says it鈥檚 extinct, and the model says it too, then we鈥檝e done all we can.

How will a more probabilistic view of extinction change our view of the biodiversity crisis as a whole?

It will allow us to scale up extinction estimates across a much bigger proportion of all species globally. The manual assessment of extinction takes a lot of effort for a single species, but if we use models, we can also use past surveys, herbarium information and natural history collections to get a first estimate of how many species are likely extinct. This will also flag the species that haven鈥檛 recently been searched for, so that they can be added to a most-wanted list and we can look for them.

Ultimately, we want a more accurate estimate of the number of species that are already extinct and species that are becoming extinct at the moment.

Can we also use technology to make searching for lost or most-wanted species easier?

I really like the stories of flying drones in Hawaii, where there have been many interesting rediscoveries. Hawaii has got this unfortunate name of the 鈥減lant extinction capital of the world鈥�. It鈥檚 partly because lots of unique species are there that don鈥檛 occur anywhere else. If something only occurs on a steep cliff on a single island and there鈥檚 a landslide, that鈥檚 it, they鈥檙e gone. Also, a lot of people live there, there鈥檚 a lot of human activity, and invasive species too. But by flying drones, a number of species have been rediscovered. Modern technology can help us find things that would be very difficult and time-consuming for manual discovery.

What would you like people to better understand about extinction?

I have worked mainly on global scales but there is another aspect, and that is local extinction. Species that are becoming extinct in Britain in local woodlands, for example, might still be present in other parts of Europe, so they are not listed as globally extinct. Species are becoming extinct everywhere, really close to home, in people鈥檚 back gardens.