I began ethnographic research in India’s cotton belt at a crucial moment in the history of agricultural biotechnology. Genetically modified (GM) crops had appeared in the mid-1990s and were being adopted by industrial farmers in a few developed countries. But interest was rising in the possibility that GM crops could benefit smallholders in developing countries.

Hopes for a GM crop revolution were particularly high in India. By the late 1990s India’s cotton sector was famously troubled – plagued by pest infestations, rampant insecticide use, farmer debt and suicide. One of the new GM crops was Bt cotton, which produced its own bio-insecticide. 

When I began field research in India in Spring 2000, Monsanto and their Indian partner were still testing Bt seeds. The seeds were approved for release in 2002, and within a decade accounted for 90% of all cotton planting in India. India’s Bt cotton is by far the leading example of smallholder adoption of any GM crop. But the technology’s actual effects continue to be hotly debated.

Activists promptly declared the technology a disaster. But I kept coming back to the cotton belt every year, funded by 2 grants from the National Science Foundation, and I wasn't exactly seeing a pattern of disaster. But there was also a rash of short-term studies by applied economists that credited the seeds with phenomenal yield boosts and “sustainable” reductions in pesticide use, which led to claims that the seeds had been an instant “winner” and a “remarkable success”. As I wrote, these studies tended to give narrow and biased snapshots of a complicated and changing situation.

India’s cotton agroecosystem includes over a dozen major insect pests, including leaf-eating caterpillars, bollworms, and numerous sucking pests. Pest populations are affected by insecticide sprays, Bt toxins, fertilizer, water use, and weather. A long-term analysis of Bt cotton’s effects was badly needed, especially one that considered pest population dynamics alongside other changes in agricultural practices over time.

In conversations with KR Kranthi last year I found that he shared my interest in conducting just such a study. A renowned entomologist and former director of India’s Central Institute for Cotton Research, Kranthi is probably the leading expert on the agroecology of Indian cotton. Together we began a close examination of spatial and temporal variation in Indian cotton over the last 20 years, leading to "Long-term impacts of Bt cotton in India" in Nature Plants.

A major takeaway of our analysis was that cotton yields did rise sharply in 2003-2004 which was soon after Bt seeds were released, but this was not when the new seeds were adopted; the yield rise preceded Bt adoption by several years. Anyway insecticide spraying was actually climbing in 2003-2004 when yields were climbing — which would make no sense if Bt seeds were actually behind the increased yields. The rise in yields correlated primarily with increased fertililization and secondarily with a surge in new insecticides. The data were unequivocal: Bt traits had little to do with it.

Bt seeds were not widely adopted until 2005-2007; yields rose only slightly in those years. Insecticide spraying dropped – but only briefly. The additional fertilizing that had caused yield rises also led to surging sucking pest populations. And the dedicated cotton-feeding pink bollworm also developed Bt resistance and began a dramatic resurgence.

Activists who quickly proclaimed Bt to have been a disaster have been overstating their case. But Bt's contribution was modest and quite ephemeral; it has proved no match for Indian insects. Cotton yields have not improved in 13 years and with the surging populations of both sucking pests and pink bollworms, Indian cotton farmers today spend much more on insecticides than before Bt was introduced. A lot more. And cotton farming is more capital-intensive than ever.