Between 2001 and 2010, New York City Transit sent more than 2,500 retired subway cars—stripped to their steel shells—from barges into the Atlantic to form artificial reefs. At Delaware’s Redbird Reef, 714 cars added structure to a largely featureless sandy bottom, and surveys reported roughly 400 times more marine food per square foot within the reef than on the surrounding sand.

New York City Transit gave thousands of retired subway cars an unusual final assignment between 2001 and 2010. Crews removed the running gear, interiors and other components, loaded the empty steel bodies onto barges, and deployed them at permitted artificial reef sites along the Atlantic coast.

The broad count and dates come from the New York Transit Museum, which records more than 2,500 cars placed off New Jersey, Delaware, Virginia, South Carolina and Georgia from August 2001 through April 2010. Delaware’s Redbird Reef became the best-known destination.

Its most repeated result needs careful wording. Delaware reports that organisms living on artificial reef surfaces can provide up to 400 times more food for fish per square foot than bare ocean bottom. That is a comparison of food density on hard reef material and neighboring sand, not a claim that the regional ocean contains 400 times as many fish.

The program was larger than Redbird Reef

The first cars were Redbirds, carbon-steel subway cars named for the red paint applied late in their working lives. They were aging out as newer trains entered service. Reefing offered a disposal route that converted the strongest part of each car into habitat, though it still required extensive preparation and transport.

New York did not lower intact trains into the sea. Contractors removed wheels, axles, doors, windows, seats and loose material, cleaned the remaining bodies, and left open steel shells. Barges carried batches to permitted sites selected by participating coastal states.

At Redbird Reef, Delaware’s reef guide maps 714 subway cars across a site about 16 miles east of Indian River Inlet. The reef also contains vessels, military vehicles, rock and other material. Any ecological change there reflects a constructed habitat assembled over time, not 714 identical objects operating in isolation.

That boundary matters when interpreting the famous number.

Why a steel shell changes a sandy seabed

Much of the mid-Atlantic continental shelf consists of sand or mud, with relatively little exposed natural rock. Animals such as mussels, barnacles, oysters, sponges and hydroids need a stable surface on which to attach. Larvae can drift through suitable water yet fail to establish if the bottom keeps shifting beneath them.

A hollow subway car supplies walls, openings, edges and cavities. Once attached animals establish themselves, they create food and shelter for other invertebrates and fish. Black sea bass and tautog use structured habitat, while the car’s door and window openings create passages through the reef.

The steel does not generate life. It adds a missing kind of surface.

ScienceBlog recently examined the colonization process at Redbird Reef. The wider program raises a second question: how much of the observed abundance represents new biological production, and how much is existing life concentrating around structure?

What the 400-fold comparison actually measures

The Delaware Department of Natural Resources and Environmental Control says the invertebrate community on artificial reef surfaces provides up to 400 times more food for fish than the bare ocean floor. Older accounts attribute the comparison to monitoring by the state reef program.

The contrast is understandable. A square foot of sand may contain burrowing animals, but it provides little stable attachment area above the sediment. A square foot of encrusted steel can hold layers of mussels and other organisms, with more animals living between and upon them.

The figure is not a census of all marine life at the site. It does not mean the whole reef, the surrounding water or Delaware’s fish stocks multiplied by 400. It does not isolate subway cars from boats, rock and military vehicles nearby. It compares food-bearing surface with adjacent natural bottom, and “up to” marks the upper end reported by the state.

Keeping the denominator visible makes the result more useful.

The cars also altered sediment and currents

Reef structures do more than provide attachment points. They redirect water, disturb sediment and settle under their own weight. A peer-reviewed study led by Nicole Raineault in Continental Shelf Research used repeated sonar surveys from 2008 through 2011 to track seafloor changes at Redbird Reef.

The researchers found scour moats from about 1 to 30 meters across, sediment coarsening around objects, and more than a meter of settling in some locations. Orientation and clustering changed the footprint of that scour. Storms and the geology beneath the sand mattered too.

Subway cars appeared to reach relative equilibrium with the seabed after six or seven years, while larger objects and clusters took longer. The finding does not judge the reef a success or failure. It shows why placement design cannot be separated from local waves, currents and sediment.

Not every cleaned object makes a good reef

Artificial reefing is not permission to discard material at sea. Preparation standards are meant to remove fuel, oil, loose debris and other pollutants before deployment. The EPA’s national preparation guidance for vessels also addresses asbestos-containing material and the need to prevent pieces from breaking free after sinking.

Structural durability can be counterintuitive. The older carbon-steel Redbirds retained useful three-dimensional form, while some later stainless-steel subway cars deployed elsewhere deteriorated much faster than expected. Construction details, joints and surface geometry can matter as much as the name of an alloy.

A structure that collapses early loses height and cavities. It may also create mobile debris. Those outcomes explain why modern reef programs assess stability, expected lifespan, contaminant removal and compatibility with a specific seabed rather than treating any large piece of scrap as habitat.

A 2020 meta-analysis by NOAA-affiliated researchers synthesized 39 studies and found that artificial reefs could support fish density, biomass, richness and diversity comparable to natural reefs overall. The results varied with geography and material. The authors’ practical conclusion was that artificial reefs are not interchangeable tools.

More food does not settle the production question

Artificial reefs can support attached invertebrates and provide shelter. They can also attract fish that would otherwise be spread across a wider area. Those processes may occur together, but they have different consequences for fisheries.

If new habitat improves survival, growth or recruitment, it can add biological production. If it mainly gathers existing fish, catches may become easier without a matching increase in the population. Fishing regulations can reduce that risk, but a busy reef is not by itself proof of a larger fish stock.

A 2025 review in Fish and Fisheries found only 12 studies that empirically quantified secondary production at artificial reefs, and only three used a control site. The authors called for before-and-after research with matched controls, along with measures of survival, recruitment, movement and food-web effects.

Redbird Reef clearly shows that hard structure can support a dense community where the natural bottom is mostly sand. Resolving what that means for fish populations requires longer comparisons that follow the reef, nearby control sites, harvest pressure and structural decay through the same years.