Since humans began to worry about having put too much carbon in the atmosphere, we’ve considered tropical forests an important “carbon sink.” Their fast growth rate, dense vegetation, and rich soils sucked more carbon out of the atmosphere then they produced. In other words, tropical forests were a natural greenhouse-gas vacuum. Except now, just when the world most needs them to be, they’re not.

At some point, it turns out, deforestation, drought, and other forest-disturbing factors tipped the scales, making tropical forests a net producer of carbon rather than a sink, according to a new study published today (Sept. 28) in the journal Science. Each year, instead of absorbing carbon, these degraded forests are a source of more carbon (roughly 425 teragrams of carbon per year) than an entire year’s worth of US transportation emissions.

Scientists at Woods Hole Research Center and Boston University spent two and a half years trekking to tropical forests in 22 countries, measuring trees’ thickness and recording their growth rate, which is a big factor in how much carbon a forest is absorbing. They then paired their field data with laser remote-sensing data and 12 years of satellite data from NASA’s MODIS satellites. The researcher’s combined approach allowed them to figure out not just losses from dramatic deforestation, but also the harder-to-calculate losses from less obvious factors, like selective logging and small-scale farming.

Previous studies have looked at large-scale deforestation in the tropics as a source of carbon, and more recent papers have pointed towards the subtler forms of degradation as a likely underestimated source.

Despite the bleak news, Alessandro Baccini, a remote-sensing scientist at Woods Hole and coauthor of the study, says it ought to inspire some hope.“What this study is telling us is that the forest is actually a great opportunity. Forests in the tropics are actually a source of carbon—meaning they are emitting more than removing—but you can change that.”

Ending tropical deforestation, they found, would reduce annual emissions by a minimum of 862 teragrams of carbon, or about 8% of annual global emissions. Eliminating deforestation may be an improbable goal, but the numbers suggest even incremental changes in forestation policy could make a significant difference.

Some countries are now investing heavily in the future of geoengineering—large-scale technological solutions to suck carbon out of the atmosphere. Baccini thinks we ought to be planting trees. “We are not there yet” with even the most promising geoengineering approaches. “For the next 50 years, it would be more effective to just do it using a more natural system,” he says. “The most effective way to do it fast, and in a cheap way, is doing it with a forest.”