The bananas you find in the average U.S. grocery store are pretty much the same: They're the genetic variety known as Cavendish.

In the market in Mayagüez, Puerto Rico, though, you have choices.

Brian Irish, a scientist who has been working at the U.S. Department of Agriculture's Tropical Agriculture Research Station in Mayagüez, points out our options. There are Cavendish bananas, to be sure, but also red-skinned varieties, miniature ones and others that seem extra plump. Shop owners are also here, buying their bananas from farms on the island.

Irish buys a few of a popular variety called Manzano, or apple. "Manzanos have a little acid, a little puckering of your mouth, especially if they're not very ripe," he says.

I take a bite. It's truly fabulous.

Irish nods. "Tasty!" he says. "It's one of my favorites, too. I really like it."

But whether you're someplace that relies on just one kind of banana, or someplace like Puerto Rico with great banana diversity, this fruit is vulnerable everywhere.

There's a deadly fungus that attacks banana plants. In the past century, an earlier version of this fungus wiped out commercial plantings of a banana variety called Gros Michel that once dominated the global banana trade.

Now history may be repeating itself. A new version of the fungus, called Tropical Race 4, is killing off the Cavendish variety.

Tropical Race 4 has marched across China and Southeast Asia, laying waste to banana plantations. It's killing bananas in Australia, and cases have been reported in southern Africa.

So far, the fungus has not spread to Latin America, but it can travel on the smallest particle of soil, even on a pair of shoes. When Irish visited a banana plantation in Australia recently, he decided to leave his boots behind. "It's such a small sacrifice to make for such an important cause," he says.

Some people probably don't take such precautions, though. "I think it's very realistic that the fungus will simply continue to spread," says Gert Kema, an expert on tropical plant diseases at Wageningen University in the Netherlands.

Kema and other scientists are commencing the search for a banana that is immune to Tropical Race 4 and could replace Cavendish.

The scientific sprint to beat the fungus relies on genetic diversity right here in Puerto Rico at the USDA's tropical agriculture research station. It's just one of many banana collections around the world that might just hold the key to stopping the fungus's deadly reach.

It's a forest of banana diversity. Bunches of red fruit dangle from some of the tall stalks; big green fruit from others. Some plants barely have any fruit at all.

"Many of them are wild," says Irish, who's been in charge of this collection for the past 11 years. "They have very small fruit, with seeds in them."

He grabs one tiny yellow banana and splits it open. It's full of black seeds, hard as rocks. "And how many in this small fruit? Hundreds. It's impossible to eat the pulp around these seeds," he says.

It may be useless to eat, but if this plant could withstand Tropical Race 4, it would be priceless.

The only way to know is by exposing it to the disease. It would be crazy to do this in Puerto Rico, of course, where the fungus hasn't yet spread. But Kema is doing such experiments in greenhouses at Wageningen University, which is one of Europe's biggest centers of agricultural research.

He has tested about 200 different banana plants so far. "Less than 10 percent of what we've tested has been resistant to Tropical Race 4," he says.

But the plants that resist the disease, so far, are not candidates to replace Cavendish. Some are wild varieties, with fruit full of seeds. Others are plantains, starchier relatives of the dessert bananas that are so popular in North America and Europe.

So it's possible that this search will fail, and the fungus will slowly destroy large-scale banana production around the globe.

But there's an optimistic scenario, too.

Kema says plant breeders can take those few disease-resistant bananas and mate them with others that taste good, creating offspring that might contain the best traits of both.

Copyright 2016 NPR. To see more, visit http://www.npr.org/.