At the beginning of the nineteenth century, the German explorer and naturalist Alexander von Humboldt arrived in South America. Most of it still belonged to the stagnant Spanish Empire. Humboldt and his companion, a Frenchman named Aime Bonpland, would spend five years travelling through a region that had dropped out of European consciousness since the forced departure of the Jesuits decades earlier. Having paid their respects in coastal cities, they quickly made their way to the vast interior. Before Humboldt, no European or North American had realized the incredible diversity of animal and plant life in the tropics. In the area that would soon become Venezuela, Humboldt found forests that were still pristine, broken occasionally by settlements whose Spanish-speaking inhabitants were few. Even here, though, colonization had left a bitter legacy. Many of the indigenous peoples had been weakened and diminished by epidemics. Often the survivors lived at small mission stations scattered along the rivers that drained and shaped the jungle.

Humboldt rarely had a chance to speak his native language in South America. For the most part he relied on Spanish and French. During his years away from Berlin, he is said to have discovered the electric eel, made the first accurate drawings of Inca ruins, crossed the Andes four times and discovered the cold current off the shores of Peru that still bears his name. There was nothing, it seems, he didn't want to know. Curious and knowledgeable about a host of scientific disciplines, Humboldt was also a linguist. His great series of books Travels to the Equinoctial Regions of America includes much first-hand news about indigenous languages. Amid the upper reaches of the Orinoco, the sheer abundance of small languages surprised and troubled him. For without a lingua franca, "a traveller, however great may be his talent for languages, can never hope to learn enough to make himself understood along the navigable rivers."

We live in a time of extinctions.

The dodo and the passenger pigeon have more and more company every month: since our talking species began to roam the earth, thousands of species of animals and plants have died out, and many thousands more are at risk. Previous episodes of mass extinction likely occurred after a collision with some errant heavenly body. Now the extinctions are our own responsibility.

Hunting, the oldest profession, still endangers some animals the last wild gorillas and chimpanzees are turning into "bush meat" daily. But a much larger number of creatures find themselves in danger as an indirect result of other human behaviours: the ravaging of forests and coral reefs, the deployment of chemical pesticides and pollutants, the technologies of modern farming. Species vanish by accident and oversight; often their demise is casual. Most of the victims are neither large nor photogenic. One of the smallest birds to disappear was a flightless songbird called the Stephen Island wren, which died out in 1894 on its only home, a green rock off the New Zealand coast. Stephen Island had no human inhabitants except for a lighthouse keeper. But in 1894 the lighthouse keeper kept a cat named Tibbles, and Tibbles succeeded in exterminating the entire species.

On a visit to the Canadian Rockies a couple of years ago, I recall dangling a hand off a walkway beside a natural pool in Banff National Park, eighteen inches above a bed of brown-green algae. A dragonfly skimmed the surface. The algae rippled in the wind. "Pond scum," you might say. Except that this particular scum contained a few diminutive snails: Physella johnsoni, the Banff Springs snail, which appears to survive in just five springs near the townsite of Banff. The snails occur only where foul-smelling, lukewarm water spills out of Sulphur Mountains springs and into ponds. With their tiny black eyes and their coiled, globe-like shells, the snails survive by feeding on algae. Even the largest of them are no bigger than a small fingernail. I could have scooped a few up in the palm of my hand.

If you were to dip your hands into the turquoise water while wearing mosquito repellent or sunscreen lotion, you would threaten a threatened species even further. Over the millennia, the snails have adapted to a delicate environment containing large amounts of dissolved gypsum, little oxygen and no artificial chemicals. Human swimming in the hot ponds  illegal now but tempting  has destroyed some mats of algae, and with them the snails' eggs. "Are snails just as important as grizzly bears?" asks a Parks Canada leaflet. "You bet they are!... Healthy populations of Banff Springs snails indicate the integrity of their unique hot-spring ecosystems. Its all just a matter of scale." The snails used to exist at nine springs in the area; from four, they have disappeared. Many other species of hot-spring snails once flourished in Europe and North America. Most of them are now extinct.

To conserve Banff Springs snails, you dont have to worry about psychology. Give them a chance, and they'll breed. The only necessary task is to safeguard their habitat. If you degrade the habitat, whether or not you've even heard of Physella johnsoni, you'll end up by killing off the species. Anything we might call development would mean degradation to the snails. All they want is for their few pools of warm, sulphurous water to be left untouched. In the heart of Canada's most popular national park, that's easier said than done.

Still, the idea that we have a moral duty to preserve this snail - so small, so impossible to cuddle, so lacking in what you might call animal magnetism - is uncontroversial. To most Canadians, it seems only normal that their government should go to some lengths to grant the snail a future. Consider this a sign of the intellectual success the conservation movement has enjoyed over the past century. Many populations of animals continue to suffer a grievous decline. But the public-relations battle on their behalf was won decades ago.

The more that human actions show up the fragility of the natural world, the greater lip-service we pay to the fight against extinction. Words like "ecosystem" and "biodiversity" now trip off the tongues of children (the concept of linguistic diversity, by contrast, is still unfamiliar to many adults). From their earliest days in pre-school and nursery school, children grow up learning about endangered species. I remember a colouring book that my older daughter used at the age of two or three - the sea turtles looked cute enough but the pictures told of poachers stealing their eggs, bulldozers destroying their nesting sites and plastic litter choking their digestive systems. It's a heavy weight to lay on a two-year-old head.

Only forty-seven Indians were left at the village of Atures when Humboldt and Bonpland arrived there, and in the absence of its Jesuit founders, the Catholic mission was in "the most deplorable state." San Juan Nepomuceno de los Atures, to give the place its full title, had been built in 1748, taking its last name from the Indian people of the region. But less than sixty years later, the Atures had disappeared. Humboldt found that the families living in the wretched settlement spoke languages called Guahibo and Maco. According to a Guahibo tradition, the Atures were being pursued by yet another people called the Caribs when they took refuge on an island in the Orinoco. There they gradually died out. If so, it's very likely that disease, not warfare, killed them. All that survived of the Atures, Humboldt believed, were their tombs in a mountain cave high above the great river: "the place of sepulchre of a whole nation destroyed."

He climbed there one evening, stopping to admire the view over savannahs stretching far to the west. Looking down on the Orinoco, he could see the cluster of forested islands where the Atures were rumoured to have fled. A great rock jutted out of the granite mountain. In a hollow of the rock, Humboldt counted nearly six hundred skeletons, varnished with "odoriferous resins" and decorated with red paint. Each skeleton reposed in a palm basket. On the ground nearby were earthenware vases containing further bones. Humboldt chose several skulls and three complete skeletons, one of them a child's, and as darkness fell he loaded them onto his mules. He intended to ship the bones to Europe for scientific study.

It was night by the time he left the cave, and the starlight above the mountain was matched by a resplendent display of fireflies on the slopes below. Humboldt noticed the wild begonias and the sweet-scented vanilla growing near the entrance to the tombs. When he arrived back in Atures, the local people became aware of a different smell: a resinous odour, coming from the backs of the tired-out mules. They knew what it meant, and they made sure Humboldt understood their fury. But if they were furious, they were also powerless.

Rare species of animals that range over a wide territory - sea turtles, tigers, Siberian cranes and so on - are among the hardest to conserve. Many of these creatures are also slow-growing and slow-breeding. But when any population shrinks to just a few dozen individuals, drastic measures are needed to stave off extinction. You may have to eliminate human traffic altogether. Australia made that choice when it turned a flat, dry, grassy woodland in central Queensland into a national park without public access. The woodland - once part of Epping Forest, a big cattle station - is the only remaining home of a species called the northern hairy-nosed wombat. Park wardens have fenced in three hundred hectares, keeping all livestock out and encouraging the native grasses to regenerate. The wombats are imprisoned in their own small wilderness. Human intervention led, indirectly, to their death everywhere else: the clearing of forests, competition from sheep and cattle, infestations of foreign grasses. Now, human intervention - habitat management, genetic research, supplementary feeding and so on - is the wombats' only chance. At last count their numbers were down to sixty-five.

Languages need a larger habitat than hairy-nosed wombats, not to mention hot-springs snails. But for people who want to help languages endure, an essential challenge is similar: to find a way of giving them a chance to reproduce. If mothers and fathers and grandparents want to speak a language to their children, that language will survive. But if mothers and fathers and grandparents are constantly told that their language is pathetic, that it has absolutely no value, that it's old-fashioned and useless, that it's a source of scorn, that it must not be spoken in classrooms or courts or offices or shops, that it's inadequate for the purposes of modern technology, that by using it they will always remain powerless and backward, that they would be mad to look for the language on any kind of electronic screen - then not many mothers and fathers and grandparents are likely to keep up the struggle.

None of this means that any human environment has to stay unchanged. Change has always occurred; stasis is death. In an oral culture, word-for-word repetition is less faithful to tradition than creative adaptation. You can't fence a language in; you can't deny it's speakers access to the outside world. What you can do, perhaps, is make their transition to the raging future a little easier to undertake. Some kinds of change - vaccination programs, for instance - can rescue a language by the simple expedient of forestalling fatal epidemics among its only speakers.

But some forms of human sickness don't respond to Western medicine. The most damaging changes of all can be inadvertent.

A generation ago, disaster was inadvertent when it fell upon Ladakh, a Himalayan valley at the far northwestern edge of India. Its people, who are Buddhist and ethnically Tibetan, speak a language called Ladakhi, a distant relative of Boro. Until the mid-1970s, the great snow peaks of the Himalayas sealed their communities off from the world. But when India built a highway to connect the valley to the rest of the country, the people of Ladakh found they were "underdeveloped." They had no TV sets, no refrigerators, no recorded music. They lived in mud homes. Economists arrived and told them they were poor. For the first time, they began to feel imprisoned in their own small wilderness. Within fifteen years of the highway's arrival, Ladakh's self-sufficient local economy had been undermined by cash and credit, and it's ancient forms of knowledge were being eroded by government schools in which Ladakhi children learned that their culture and language are inadequate for the modern world. But the highway could not be unbuilt.

"Young children I had never seen before used to run up to me and press apricots into my hands," wrote the British anthropologist Helena Norberg-Hodge in her book Ancient Futures: Learning From Ladakh. "Now little figures, looking shabbily Dickensian in threadbare Western clothing, greet foreigners with empty outstretched hands. The films they see and the tourists they meet make their lives seem primitive." No longer do the Ladakhi people stand at the centre of their own lives; instead, they are conscious of existing on a periphery, far from the "real life" taking place in New Delhi or New York. Young travellers from rich countries flock to Ladakh, hoping to find an unspoilt refuge from the materialism of their own society. Instead they pass on the very values they were hoping to escape. They don't speak Ladakhi. They don't drink butter tea. Searching for the wisdom of the ages and the locals, they encourage the locals to speak English and buy cola. Defenders of highway-building and globalization can rightly say that the Ladakhis are richer than before. But it would be hard to claim they're better off.

If Humboldt had cared only about ornithology, he would still have found much in Venezuela to keep him busy. The nation contains 43 species of birds that are found nowhere else, some of them with memorable English names like the "handsome fruiteater" and the "guttulated foliage-gleaner." A secretive, insect-eating, ground-dwelling bird called the Tachira antpitta was last seen in Venezuela in 1956, and no-one can be sure if it's extinct - rare birds, like rare languages, sometimes survive unexpectedly long.

But the explorer had much else on his mind. Among other things, he wanted to trace the relationship of the vanished Atures to the surviving Indian peoples of the region. In the Orinoco's "labyrinth of petty nations," this proved to be a hard task. Without history texts to help, Humboldt decided that the best way to work out the labyrinthine interrelationships would be by "the analogy of tongues. These are the only monuments that have reached us from the early ages of the world; the only monuments which, not being fixed to the soil, are at once movable and lasting, and have traversed time and space." From the village of Atures, Humboldt moved along the Orinoco to a mission settlement by the name of San Jose de Maypures. There, two other indigenous languages were spoken.

He arrived by night, struck by the solitude of a village where nothing could be heard except the calls of nocturnal birds and the clamour of a distant waterfall. Even in the middle of the night, however, the mosquitoes were ravenous. In the morning, scratching his wounds, he walked around Maypures; it didnt take long. The little church, built of palm trunks, stood on a plain below the waterfall. Seven or eight huts surrounded the church, and in some of them the women were making pottery. Humboldt admired these huts for their "appearance of order and neatness, rarely met with in the houses of the missionaries." Nearby, plantains and cassavas were being cultivated. The introduced goats had been devoured by jaguars, but descendants of the Jesuits' black and white pigs had somehow managed to survive.

Nor were they the only domestic species in Maypures.

The greatest concentrations of linguistic diversity lie in the tropics. Languages tend to flourish where the web of biological diversity is also at its most intricate: in tropical rainforests, above all. More languages are indigenous to Venezuela than to all of central and eastern Canada - a country eleven times its size. Where food is abundant, territories dont need to be large. Where food is scarce, territories expand.

The wet, hilly, verdant island that is divided between the nation of Papua New Guinea and the Indonesian-controlled region of Irian Jaya has given birth to about 1,100 living languages. One in every six languages spoken on the planet comes from this island. (These are languages, remember, not just dialects.) In New Guinea each valley, each mountain, each tributary, each bay seems to have a language of its own. Five times more languages exist on the island than in the whole continent of Europe. New Guinea is also a hotbed of biological diversity on a scale almost inconceivable in colder realms. Late in 1996, Conservation International sponsored a team of scientists to undertake a rapid assessment of biodiversity in the Lakekamu Basin, a lowland rainforest in Papua New Guinea. Some of the sites they visited were already being logged by transnational companies. Working in an intact forest just a single square kilometre in size, the scientists counted more than 250 species of ants. A month of work in the Lakekamu Basin was enough for them to discover twenty-three species of previously unknown insects, eleven new species of frogs, seven new species of reptiles and three new species of fish.

Unknown to Western science, that is. Often the indigenous people of a region have been familiar with a "new" species since time immemorial. One of those species, the Arnhem Land long-necked turtle, lives in the far north of Australia. I first heard of the turtle at the visitor centre in Nitmiluk National Park. A few minutes' walk away, orange-red cliffs overshadowed a river whose relentless waters have pushed a deep gorge through the Arnhem Land plateau. Until the year 2000, the turtle that lives in rockholes among those cliffs had neither an English nor a scientific name - a name, that is, in the dead language of our taxonomy. The animal's small size and flattened head, along with its habit of feeding on water lilies as well as fish, distinguish it from Chelodina rugosa, the northern long-necked turtle. Yet one of the Aboriginal peoples of the plateau - the Gagudju - have always known that Burrungandji, the "new" species, is not the same as Almangiyi, the "old" one. It's just that nobody bothered to ask them. The Gagudju language is now endangered: young and middle-aged people no longer speak it. In tribute to Gagudju, the Australian scientists who formally described the species have called it Chelodina burrungandjii. If and when Gagudju dies, what other environmental knowledge will die with it?

Language centres in many countries are fighting against the clock, so as to record the words of elderly speakers before time swallows them up. I think of Ngaapa Wangka Wangkajunga, a wordbook in an Australian desert language that is, like Gagudju, at risk of extinction. An old woman called Dolly Snell enriched the book with dozens of Wangkajunga words for plants and bush tucker - essential knowledge in an arid environment. Perhaps her knowledge can be transferred across languages; perhaps not. Neither she nor the linguist who compiled the book had a precise English word for karlijita ("an edible grass seed"), nartutaka ("a small plum-like fruit from bushes that grow on hills"), ngalyilka ("a small edible mushroom") or purti-purti ("a plant similar to bush tobacco. Eating this plant will make a hungry person feel full"). The language reflects its value in the environment: Wangkajunga may have as many words for lizard as Wall Street English does for financial instruments. In the future, perhaps, zoologists will find that the language had reason to distinguish lungkurta ("a blue-tongue lizard") from lungminka ("a blue-tongue lizard") and ngintul ("a blue-tongue lizard"). They may well be able to clone a whole duneful of lungkurtas on demand. But by then, other kinds of meaning that went with the Wangkajunga knowledge system will no longer be available. A threatened language can never be cloned.

In an appeal for action, not just rhetoric, biologists have begun to warn of the terrible consequences that await us if the loss of biodiversity continues apace. "Humanity's food supply comes from a dangerously narrow sliver of biodiversity," Edward O. Wilson has written. "Throughout history, people have cultivated or gathered seven thousand plant species for food. Today only twenty species provide ninety per cent of the world's food and three - maize, wheat and rice - supply more than half." If variety was once a guarantee of safety, our current reliance on so few food species makes us vulnerable to plagues and climate change alike. We will never know what cures for disease are being lost with the wreckage of tropical forests. Many of our most common and effective medicines derive from wild plants; yet Western scientists have studied only a tiny proportion of plant species for their medicinal value. When the forests are erased, local peoples scatter or die. So do their languages.

Threaten people with a double blow - your children starve, you succumb to cancer - and they might support tough action to save the environment. That's what some biologists now believe. All other appeals having failed to make a significant difference, perhaps naked self-interest will succeed in altering our destructive behaviour. But linguists can't make such an appeal. You can argue with a cabinet minister or a chief executive officer about the intrinsic usefulness of an old-growth forest, pointing to the long-term economic benefits of tourism and sustainable harvesting - but it's tough to argue that sustaining an ancient language is good for anybody's gross domestic product. A language is not a commodity. It's also hard to make an argument on purely moral grounds that a threatened language has the right to be protected. Individuals have rights, companies and governments insist they have rights, animals may or may not have rights - but languages?

Almost in desperation, linguists have begun to adopt the vocabulary and metaphors of biologists. They too are speaking about the resilience that comes with diversity. They too are asking for niches of equilibrium to remain undisturbed. They too are warning of the dangers inherent in what Wilson calls an "impoverished and homogenized world," one in which a few dominant lifeforms have overrun and erased the diversity that used to sustain us.

But Wilson means rats and starlings. Linguists mean the tongue you're reading now.

In the village of Maypures Humboldt was delighted to see, not only black and white pigs, but also "tame macaws around the huts of the Indians, and flying to the fields like our pigeons.... These macaws, whose plumage glows with vivid tints of purple, blue and yellow, are a great ornament to the Indian farmyards." He had seen the species above the banks of other tropical rivers, and had even tasted its cooked flesh: "black and somewhat tough." Ever punctilious in his descriptions, Humboldt called it "the largest and most majestic species of parrot with naked cheeks that we found on our travels." It was, almost certainly, the blue-and-yellow macaw, a species that still survives in good numbers over wide stretches of South America.

Other macaws have been less fortunate. Of the eighteen recognized species, all of them native to Latin America, two are extinct and a third, Spix's macaw, perches on the brink. The last wild member of the species lived in eastern Brazil and disappeared in October 2000, leaving a few dozen captive specimens divided among a Qatari sheikh, a Philippine industrialist and a Swiss entrepreneur. For them, the species is not just a commodity but a trophy. The International Union for the Conservation of Nature describes the status of seven other types of macaw as vulnerable, endangered or critical. Only eight species appear sure to be with us for a good while to come.

Humboldt was shown a talking parrot in Maypures. It too was a trophy of sorts. He didn't specify whether it happened to be a blue-and-yellow macaw or a member of a different species. It was an old bird, a feathered survivor, and the local people insisted "they did not understand what it said." When Humboldt asked why, he was told that the parrot "spoke the language of the Atures" - the language of the people whose sweet-smelling bones his mules had carried off the granite mountain. The Atures language had died out among humans. It was last heard coming from a bird's beak.