One morning in March of last year, I set out from Gulfport, Mississippi, on a three-week mission aboard the U.S. Navy research vessel Atlantis. The 274-foot ship, painted a crisp white and blue, stood tall in the bright sunlight. On its decks were winches, cranes, seafloor-mapping sonar, a machine shop, and five laboratories. Stowed in an alcove astern was Alvin, the federal government’s only manned research submarine. “Research vessel Atlantis outbound,” A. D. Colburn, the ship’s captain, reported into the ship radio.

The water was calm and the bridge crew quiet as they steered us into open water. For the next fourteen hours, we would sail toward the site of BP’s Macondo well, where, in April 2010, a blowout caused the largest offshore-drilling oil spill in history. Once there, Atlantis’s crew would launch Alvin and guide it to the bottom of the ocean, reaching depths as great as 7,200 feet below the surface. Over the next twenty-two days they would send the submersible down seventeen times, to gather animal, plant, water, and sediment samples. Their goal was to determine how BP’s spill had affected the ocean’s ecosystem from the seabed up. I would get the chance to join them in the submarine as they went closer to the Macondo wellhead than anyone had gone since the blowout.

Illustration by John Ritter

Data gathered by the Atlantis would likely be used in the federal legal proceedings against BP, which began in December 2010. A few months after our mission, U.S. district judge Carl Barbier found the company guilty of gross negligence and willful misconduct. In January 2015, he ruled that the amount of oil the company was responsible for releasing into the Gulf totaled some 134 million gallons, a decision both sides have appealed. By the time this article went to press, Barbier had yet to make his third and final ruling, which will determine how much BP owes in penalties under the Clean Water Act. (If his judgment about the size of the spill is not overturned, the company will face a $13.7 billion fine.) Meanwhile, the Environmental Protection Agency, the National Oceanic and Atmospheric Administration, the Department of Agriculture, and the Department of the Interior are concluding an ecological-damages assessment to determine how much BP must pay to restore the Gulf Coast. The trial and the assessment are likely to result in the largest penalty ever leveled against an oil company.¹

Dr. Samantha Joye, a biogeochemist at the University of Georgia and the lead scientist on the mission, estimated that 30 million gallons of oil from the BP spill remain in the Gulf — the equivalent of nearly three Exxon Valdez spills — and that about half of this amount has settled on the ocean floor, where its ecological effects could be devastating. A BP spokesman told me that most of the spilled oil had been removed, consumed, or degraded, and that there had been “very limited impact from the oil spill on the seafloor.” But Joye’s research indicates that any damage the oil has done to creatures inhabiting the deep-sea waters — from tube worms to sperm whales — threatens the ecosystem, harming organisms that rely on those species. Among those at risk are phytoplankton, the sea vegetation that produces about half of the planet’s oxygen. “If you short-circuit the bottom, you threaten the entire cycle,” Joye told me. “Without a healthy ocean, we’ll all be dead.”

The Gulf of Mexico is one of the world’s most ecologically diverse bodies of water, home to more than 15,000 marine species. Its coastal areas contain half of the wetlands in the United States — some 5 million acres — which provide habitat for a variety of birds and fish. Mangrove forests line the Gulf shore, and its shallow waters are filled with sea grass. Before the oil spill, more than 60 percent of all oysters harvested in U.S. waters were caught in the Gulf. Today that share has dropped to about 40 percent. Many threatened and endangered species also live in the Gulf: sea turtles, Florida manatees, whooping cranes, and bald eagles. Dolphins are a frequent sight. Giant squid, jellyfish, and octopuses swimming through the Gulf’s waters pass some of America’s lushest and most imperiled coral reefs.

As Atlantis carried us farther from shore, the fishing boats dropped away and we gradually entered clear, open ocean. Mississippi and Florida are the only Gulf states that do not allow offshore drilling in their coastal waters, which extend several miles out from shore. But as soon as we entered federal waters, oil rigs and drilling platforms began to appear. We sailed through the rigs for twelve hours and, near eleven o’clock at night, arrived at our destination, Mississippi Canyon Block 252 (MC252). This was the roughly 5,760-acre lease area that contained the Macondo oil well.

Illustration by John Ritter

Much of the western and central Gulf of Mexico has been parceled into a patchwork of oil and natural-gas lease blocks. A web of underwater pipelines carries the fuel to shore. MC252 sits in an industrial corridor that is occupied by the world’s largest oil companies — including ExxonMobil, Chevron, Shell, Eni, Noble Energy, Hess, and BP. Some 17 percent of U.S. oil comes from the Gulf, nearly 80 percent of that from depths of 1,000 feet or more below the ocean’s surface. In 2008, BP leased MC252 from the Department of the Interior, paying $34 million for ten years. The company then leased the Deepwater Horizon, a semisubmersible oil rig, from Transocean, which also ran the rig’s daily operations. On February 15, 2010, the crew of the Deepwater Horizon started drilling at the Macondo site, some 5,000 feet below the ocean’s surface. Eventually they dug through more than two miles of rock and sediment, to a depth of 18,360 feet.

The Gulf of Mexico’s high concentrations of methane, along with other natural features, make it an especially dangerous place to drill. Natural gas, which is lighter than oil, can get into oil pipes and overwhelm a well’s pressure-control systems, leading to a blowout. At 9:45 on the night of April 20, 2010, as the crew worked to prepare the well, methane escaped from the Macondo and shot up the steel pipe that connected it to the Deepwater Horizon. Inside the rig, sparks from machinery ignited the gas, setting off a series of explosions. The two men who were working near the pipe, in the rig’s mudroom, were quickly incinerated. They were the first of eleven crew members to die. At 10:21 a.m. on April 22, the Deepwater Horizon collapsed into the ocean. Its wreckage remains strewn around the well to this day.

When the Macondo well blew, none of the oil companies operating in the deep waters of the Gulf were prepared, even though the largest among them — ExxonMobil, Chevron, Shell, and BP — had claimed to the Department of the Interior that they could handle a far worse deepwater blowout. So BP applied methods designed for smaller, shallow-water spills. In May, employing what were known as the “top kill” and “junk shot” approaches, BP dumped drilling mud, golf balls, and tire rubber onto the well. Nothing worked. In the eighty-seven days it took to secure the well with a temporary cap, more than a hundred million gallons of oil and half a million tons of natural gas — most of which was methane — escaped into the Gulf. Eventually, BP performed the one operation that, however risky and time-consuming, it knew how to do: it drilled another well. One hundred and fifty-two days after the blowout, BP’s relief well intersected the Macondo borehole, allowing the company to pump in mud and cement. This remains the only proven method for permanently sealing a blown-out well in deep water.

By May 2011, BP’s oil had sickened or killed more than 100,000 Gulf animals: 28,500 sea turtles, 82,000 birds, and more than 26,000 marine mammals, including several sperm whales. Too small or too numerous to count were the vast numbers of dead fish, crustaceans, insects, and plants that washed up onshore. Most of the other organisms initially killed by the spill died at sea and were never seen.

The creatures that inhabited the coldest and darkest realms of the Gulf were not spared, either. Until the nineteenth century, when a pioneering British naval expedition was able to collect living samples from the seafloor, there was no evidence that animals could survive in the ocean’s deepest waters. In 1977, more than a century later, researchers used Alvin to explore the Galápagos Rift, in the Pacific Ocean, at depths never before visited. To their surprise, they observed a broad diversity of life, including many previously unknown species. A 1984 Alvin dive revealed abundant populations thriving in the deepest parts of the Gulf, as well. In December 2010, Joye and her crew surveyed the sea life near the Macondo wellhead. The view from the submarine revealed a barren landscape. The spill had chased out or killed anything that had been living down there. “It’s so strange to see nothing along the seafloor, particularly at this depth and in this area,” Joye wrote in her 2010 Alvin dive report. “I saw nothing on the bottom that was living,”she told me.

When we arrived at MC252, Atlantis’s twenty-four-person science party, most of them women in their twenties, immediately got to work. One group began deploying equipment from the decks, including sampling tools so large they had to be lowered into the water by crane. A device called a MOCNESS — the Multiple Opening and Closing Net with Environmental Sampling System — floated alongside the ship like a green sea monster, gathering tiny creatures to analyze how oil and gas were being taken up into the food web. Joye and her team disappeared belowdecks to prepare for the next day’s dive.

Life aboard Atlantis revolved around Alvin. A twelve-person team of pilots, engineers, and technicians was constantly at work on the sub. Alvin was commissioned by the Woods Hole Oceanographic Institution (which continues to operate it, on behalf of the Navy) and built by General Mills in 1964, six years after the establishment of the U.S. space program. In 1966 Alvin located a lost hydrogen bomb in the Mediterranean Sea; in 1986 the sub surveyed the wreckage of the Titanic. For many of the scientists, a key payoff for their long hours and grueling working conditions (they rarely slept during their three weeks at sea) was the chance that they would get to take part in a dive.

Since Joye’s 2010 mission, Alvin had undergone a three-year overhaul that replaced 75 percent of its components and cost $41 million. Despite the work, the sub proved finicky. On a test dive a few weeks before we set sail, the oxygen scrubber, which removes carbon dioxide from the air, had failed, requiring Alvin’s three passengers to put on oxygen masks during their rapid return to the surface. Technicians later fashioned a simple umbrellalike device to temporarily fix the problem — just one of many contrivances that kept the sub running. (In a workroom one day I found a box filled with condoms. An Alvin technician explained that they were especially useful for protecting electrical wiring from exposure to seawater.)

Left and center: Alvin lifts soil samples from the seabed. The dark layer visible at the top of the sample is degraded Macondo oil. Photographs by the author. Right: Coral killed by Macondo oil. Photograph by Samantha Joye

Alvin is twenty-three feet long, round and squat with a bright white shell, a red hatch, and long metal arms reaching from its sides. The sub was named for Allyn Vine, a pioneering submersibles engineer, but members of its crew, following nautical tradition, refer to Alvin as “she,” or sometimes “the Ball.” They never call Alvin “it.” The language they use is that of a very demanding and complicated personal relationship. At a media event the day before we left port, Alvin had appeared on deck, slowly rolling out on yellow tracks. I caught myself and several other reporters turning our recorders and microphones toward her, as if we expected her to speak.

By five o’clock on the morning of April 1, the day I was scheduled to dive, the Alvin crew was already up and preparing the sub. My dive would be the second of the trip, and the first to approach the Macondo wellhead.

After it rolled out to the stern’s edge at seven o’clock, Alvin was tethered to a mechanical winch with a thick white rope that would lower it into the water. During my time on the Atlantis I had learned how to maneuver my way into an oversize red rubber survival suit (the “Gumby suit”) in case we had to abandon ship. I’d learned, too, how to use the emergency-breathing device, which looked like a cross between a 1940s gas mask and a canister vacuum cleaner, in the event that the sub’s scrubber failed again. I knew not to bring shoes on the dive, and to wear thick socks and warm clothing, because we’d be sitting still for eight hours. It gets very cold at the bottom of the ocean.

Bruce Strickrott, the Atlantis expedition leader, became an Alvin pilot in 1997, after retiring from the navy. He compared diving in the sub to space travel, and Alvin’s silver and black interior does indeed resemble a spacecraft. The top half is packed with computers, monitors, electrical cables, and dashboards covered with red and black buttons and silver switches. Most of the sub’s volume is given over to thrusters, banks of lead-acid batteries, and bundles of electrical wiring. Its three passengers share a sphere seven feet in diameter. Inside the sphere are five viewing portholes — three at the sub’s nose, in front of the pilot, and one on each side — and the walls are draped in black insulated fabric to keep out the cold. As Joye, Bob Waters — the pilot for my dive — and I entered the sub in socks, sweats, and wool hats, I felt as though we were settling in to a cozy camping tent. A more sinister feeling crept in, however, when Alvin’s ladder was drawn up and the hatch closed securely behind us.

After Strickrott gave us permission to depart, we gently descended underwater. Through my tiny circular window I saw the sunlight and clouds give way to bubbling blue liquid.

“Damn it!” Waters grumbled, after we’d dropped just a few feet below the surface. Out the front window, we watched as a pair of milk crates, which had been secured to the front of the sub to carry the tubes used to collect samples, sank into the deep. A pair of crew members in swim trunks, riding atop the sub in case of just this sort of situation, dove after the crates. They saved one, but the other got away.

Once the salvaged crate was reattached, Strickrott’s soothing voice came back over the radio. “Let’s do this again,” he said, and we descended once more.

Joye and I are both small, and we sat with our legs stretched out on the cushioned floor. Waters crouched on a low stool that faced the computers he used to pilot Alvin. Once we reached the bottom, we were to identify any visual changes from the December 2010 dive and to collect water and sediment samples for further study. The scientists from Joye’s research consortium wanted to know how much oil was on the ocean floor, whether microbes were consuming the oil that remained, and how the sea life was faring in the depths.

Our descent took two hours, at a pace that felt motionless. We could see no farther than a few feet from our windows, and there was little change in the scenery. Waters reported regularly to the Atlantis via an underwater telephone, and the digital display on the Fathometer marked off our continuing dive, but the only visible sign of our depth was the color of the water — powder blue became turquoise and then navy, before fading into total darkness.

At 1,223 feet, the gloom was suddenly illuminated by zooplankton, which appeared as sinuous black lines with glowing tops. Joye explained that these tiny animals hosted luminescent bacteria that lit up when the zooplankton were surprised or alarmed. When we reached 3,609 feet, two thirds of the way down, I asked Waters if he was excited.

“About what?” he responded. This was his 120th dive, give or take. He’d started working with Alvin in 1995, after building a laser-tag system for a Defense Department training program. Much of the rest of the Alvin crew was also composed of engineers, because, as Waters told me, “you’ve got to be able to fix it, not just fly it.”

At 5,272 feet, we hit bottom. Alvin’s LEDs came on. If there was oil down here, we couldn’t see it. Outside was an endless gray underwater desert: barren, flat, and stark. Through our tinted windows it looked like a vast moonscape.

Our dive brought us within two nautical miles of the wellhead. Any nearer and we would have risked getting caught in the wreckage of the Deepwater Horizon. We traveled about a mile and a half in five hours, tracing half a circle around the site, and stopped periodically to collect samples. By manipulating Alvin’s robotic arms and fingers, Waters was able to remove each sample tube from the crate, push it into the sediment, and delicately return it to its place.

On the previous day’s dive, Joye and Joseph Montoya, a biological oceanographer at Georgia Tech, had been disturbed by their observation of dead and damaged coral — healthy coral provides habitat for thousands of species; dead coral is home to nothing. Today, however, there was a bit of good news: we saw a handful of sea cucumbers, small white fish, red crabs, blue eels, and pink shrimp. Etched along the seafloor we noted little pencil-shaped lines, evidence of organisms called infauna, which burrow, wormlike, into the sediment. As we passed, sea creatures struck attack poses: eels hung vertically in the water; crabs extended their claws and hind legs in our direction. Joye exclaimed at the sight of a vampire squid, a rare cephalopod, which showed us its bright red body and small webbed tentacles as it sped past. Later we saw a giant isopod, which Joye described as a “swimming cockroach.” These gave her hope. But she told me that before the oil spill, we would have expected to see many more of these and other creatures: fish, urchins, sea fans, and perhaps even whales and sharks.

After a few hours we stopped to eat lunch. Outside the sub, it was 39 degrees Fahrenheit. The cold seeped in, causing condensation along the walls that soaked anything pressed against them.

Shortly after we resumed our journey, the flat seabed topography was broken by a set of mounds running in straight, parallel lines. They were too symmetrical not to be human-made. “Someone’s been messing around down here,” said Joye. We followed the lines for a while, then returned to our sampling.

At three in the afternoon, Waters announced the completion of our mission to the Atlantis, which gave us clearance to ascend. The sub had no toilet, so Joye, who was in need of one, was forced to use the dreaded pee bottle. I held up a blanket, Waters turned up the music — Adele — and Joye regaled us with the story of “the first guy to poop in the sub.” When we reached the surface, we spent twenty minutes bouncing in the waves until the Atlantis’s swimmers were able to hook us onto the winch. Once we were back on deck, Montoya and another researcher dumped two giant buckets of seawater over my head, initiating me into the exclusive club of Alvin divers.

Over the next ten days I spent dozens of hours in the ship’s labs, watching the researchers sort and analyze specimens. The sediment samples we’d gathered were, it turned out, virtually identical to the ones collected in 2010. The layer of oil residue deposited four years earlier was still there. “It looks the same no matter where you are,” Joye said. “And it hasn’t changed.”

Today a coating of degraded oil, as much as two inches thick, extends across nearly 3,000 square miles of ocean floor. It is expected to remain there forever. In the Atlantis’s computer lab, Andreas Teske, a microbial ecologist at the University of North Carolina at Chapel Hill, told me, “When another expedition comes here in a hundred or a thousand years, they will say, ‘Ah, okay. Here is the 2010 oil spill.’ ”

There are many reasons that oil remains on the seafloor. The cold, dark bottom of the ocean is a naturally preservative environment. In addition, Joye found that microbes that consumed some of the oil and methane in the first few months after the spill have largely stopped eating. What they’ve left behind — the parts they have not yet broken down — are among the most toxic components of the oil, including polycyclic aromatic hydrocarbons, which are known human carcinogens. The microbes have also been inhibited by Corexit, a toxic chemical dispersant that was supposed to keep the oil from drifting ashore. Nearly two million gallons of the dispersant were used in the aftermath of the spill. Joye’s research now suggests, however, that Corexit was not only environmentally harmful; it was also counterproductive.

What does it mean that a blanket of oil remnants will cover thousands of miles of ocean floor for the foreseeable future? “We don’t know exactly,” Joye told me. Nothing close to the size and duration of this disaster has ever been studied, and it will take years to fully understand the spill’s effects. But the data collected so far is alarming. Last June I talked to Dr. Paul Montagna, a marine ecologist at Texas A&M University who studies benthic organisms. He had found significant declines in a range of species that live on the Gulf seafloor. Within a nine-square-mile area around the Macondo wellhead, he measured a 50 percent loss in the biodiversity of tiny invertebrates called meiofauna and slightly larger species called macrofauna. These species are a critical food source for larger organisms. Meiofauna and macrofauna have suffered losses as far as ten miles from the well. A die-off at any link in the food web threatens the species that depend on it, but it can also affect those farther down. Phytoplankton, for instance, rely on seafloor macrofauna such as tube worms to help decompose organic matter and release nutrients back into the water.

The increase in sea life that we had observed on Alvin could signal the start of an ecological recovery, Montagna said. But those returning species were now also being exposed to the oil on the seafloor, which they would pass along to the creatures that ate them. Joe Montoya’s research on phytoplankton has uncovered clear evidence that oil and gas carbon are moving through the food web. Ultimately, these contaminants, in potentially harmful concentrations, could reach “things like big fish that people are commercially interested in,” Montoya told me.

One study has already demonstrated that the spill was followed by immediate declines in the larval production of tuna, blue marlin, mahi-mahi, and sailfish. Macondo oil has also been linked to life-threatening heart defects in embryonic and juvenile bluefin and yellowfin tuna, as well as in amberjack. Perhaps even more troubling has been the effect on dolphins, which are predators at the top of the food chain and therefore indicators of the ecosystem’s degradation. In 2011, dolphins were stillborn or died in infancy at rates six times the historical average. Last year, the number of dolphins found dead on the Louisiana coast was four times higher than the annual average before the spill.

The effects of BP’s disaster have now spread far from the Gulf. Traces of oil and Corexit, for instance, have been found in Minnesota, Iowa, and Illinois, in the eggs of white pelicans that were in the Gulf at the time of the spill. Nor are humans immune to the damage. In January 2013, BP agreed to a medical-benefits settlement that provides twenty-one years of health monitoring and potential monetary compensation — up to $60,000 per person — to Gulf Coast residents and cleanup workers who can demonstrate spill-related respiratory, gastrointestinal, eye, skin, and neurophysiological conditions. (Researchers have also found that crude-oil contamination can lead to cancer, birth defects, and developmental and neurological disorders such as dementia, though none of these are covered by the settlement.) Of the more than 200,000 people who were potentially eligible for remuneration, however, only 12,144 had filed claims by the end of 2014, a spokesperson for the court-appointed medical-benefits-claims administrator told me. Of those, a mere 1,304 have been approved for payment.²

In June of last year, a tar mat composed of degraded BP oil that weighed more than a thousand pounds was found on a beach near Fort Pickens, Florida. BP’s oil also remains lodged in shoreline marshes, where it is killing plants and intensifying coastal erosion. In March, another tar mat, this one weighing 29,000 pounds, was found buried in the marshes of East Grand Terre Island, Louisiana. The beaches of Bay Jimmy and Bay Batiste, also in Louisiana, are still so heavily oiled that they remain closed to shrimping, crabbing, fin fishing, and oystering. “Our catch is down by a hundred percent,” Byron Encalade, a Louisiana oysterman, told me in March. And although the Food and Drug Administration declared in fall 2010 that many Gulf fish products were safe for human consumption, one 2011 study found that people who ate a diet heavy in Gulf seafood (or who were medically vulnerable) could be at risk of developmental disorders and cancer. “When people say, ‘Oh, the oil spill is over,’ ” Joye told me, “they’re not realizing that the full impacts are on a very long timescale, of decades or more.”

Though there has been no drilling in MC252 since the Macondo was sealed, this won’t be the case for long. During my Alvin dive, Waters had steered the sub to follow the suspicious mounds we’d seen running along the seafloor. They looked like freshly laid receiver cables, which are used in seismic surveys to map underground reserves of oil and gas. We were all shocked: although surveying the lease area is not illegal, everyone had been under the impression that no further oil- or gas-related work had been undertaken in MC252 since the well was sealed. After my return to shore, I confirmed that oil and gas companies were again exploring the site and would soon begin drilling.

It is not surprising, of course, that oil companies remain interested in the site. After all, the Macondo well was estimated to contain as many as 1 billion barrels of oil, which could be worth $50 billion. I later confirmed with the Department of the Interior’s Bureau of Ocean Energy Management (BOEM) that WesternGeco, a subsidiary of Schlumberger, one of the world’s largest oil-services companies, had received a permit in October 2013 authorizing “geophysical exploration for mineral resources” in MC252. WesternGeco began its survey in January 2014. The cables we saw in April were most likely used to generate a visual representation of the area’s oil and natural-gas potential.

In May 2014, BOEM approved the division of MC252 into two lease areas. BP retains just 270 acres, composed primarily of the Macondo well site and the Deepwater Horizon wreckage. The rest of the area, 5,490 acres, was turned over as a new lease to LLOG Bluewater Holdings, a privately owned offshore-oil company that is already active in the Gulf. BP cannot develop its area for oil or gas, but LLOG is allowed to drill in its lease area. Last October, LLOG received approval for two exploratory wells in MC252. They will be less than a mile from the Macondo site. When I contacted LLOG in April, the vice president of deepwater projects said that the company had no immediate plans for surface operations in MC252. But he did not dispute that the company intends to recover oil from the lease block. In a March filing, LLOG outlined a plan to dig from an adjacent lease block to reach the oil in MC252. The company is expected to complete its first such well this month.

The oil industry would like us to believe that it has corrected the problems that led to the Deepwater Horizon disaster, but in June of last year, the U.S. Chemical Safety Board, an independent agency established by Congress to investigate industrial incidents, released a report indicating that another “catastrophic accident” remains possible. “People in the industry have to recognize that Macondo was not just a one-off,” said Cheryl MacKenzie, a lead investigator at the CSB. Virtually every investigation into the disaster has found that it resulted from industry-wide, not company-specific, failures. Yet only a handful of industry policies have changed. William K. Reilly, an EPA administrator under George H. W. Bush and the co-chair of a national commission on the BP spill, warns that even good regulation and oversight cannot prevent another disaster from happening. “Drilling in very deep water is a highly challenging affair that involves highly complex technologies, and they sometimes fail,” he told me. “One should not suffer the delusion that it can be done risk-free.”

Nonetheless, the number of rigs operating in the deep waters of the Gulf has continued to increase — from twenty-nine in 2011 to fifty-one in 2014. The amount of oil under the Gulf is estimated to be 5 billion barrels, worth about $250 billion at today’s prices. Oil production there is expected to increase by 30 percent this year — meaning that by the end of 2016, oil companies will be extracting as much as 80 million gallons of oil from the Gulf each day. To reach that oil, they have begun drilling at sites nearly twice as deep as Macondo. “We’re fully back in,” Richard Morrison, the regional president of BP’s Gulf of Mexico business, recently told a Houston energy newspaper.

Ten days after my dive, a thirty-eight-foot twin-diesel catamaran traveled seven hours from Cocodrie, Louisiana, to meet the Atlantis at sea. It brought us fresh honeydew melon, zucchini, eggplant, and parsley, as well as a crew member who was returning from the hospital after falling about twenty feet on his back during a maintenance operation on the Atlantis the previous week.

After the supplies were unloaded, I boarded the catamaran for the return trip to shore. As we glided toward Cocodrie, deep-sea oil rigs gave way to drilling platforms that proliferated so thickly that the captain had to weave around them. The contrast with Mississippi’s open water was stark: Louisiana’s coast is packed with rickety and abandoned oil platforms, many of which have been there for decades, the oil they sought long since gone.

Cocodrie is a tiny fishing community perched on the southern edge of Louisiana — just a few hundred homes stretching along the bayou. There are a few restaurants and one or two shops that cater to visiting sport fishers and the cannery workers who commute from farther up U.S. 56. After disembarking, I visited a grocery store that doubled as a knickknack shop. The store, owned by Cecil and Etta Lapeyrouse, was built by Cecil’s grandfather in 1914. A gas pump sat out front, and in back, hidden behind a garden of flowering white oleander and green cypress trees, a path opened onto a wooden deck drenched in sunlight, where boats could pull up to refuel. There were none in sight that day. The town’s economy had suffered after BP’s oil covered the water and shores near Cocodrie. Things were turning around, Cecil told me, but with the drop in business he wasn’t sure whether his store would survive.

Sitting on the deck behind the shop, I admired the beauty of the empty water, now unblemished by human activity. But I looked out to sea knowing, better than ever, what lay below the placid surface.