Last spring my daughter counted seven bees on one lavender plant while the rest of the garden stayed quiet. A neighbor leaned over the fence and said the GMO corn across the road was to blame. I knew enough to be suspicious of an answer that tidy. What I did not know yet was how the science had changed while the argument stayed frozen in place.

I am not an entomologist. I am a mom who reads primary studies because food labels stopped answering my questions. This is what I found after a few months of chasing the research.

The lab question: Bt protein and bee larvae

The first genetically modified crops I remember hearing about were corn and cotton that make proteins from Bacillus thuringiensis, a soil bacterium. Farmers and gardeners have sprayed Bt for decades, but in these crops the plant produces the protein in its tissues. The worry was straightforward: if bees eat Bt pollen, does it hurt them?

The early studies tested exactly that. Researchers fed adult honey bees and larvae high doses of Bt proteins, often far above what a bee would encounter in a field. A 2008 meta-analysis in PLoS ONE pooled 25 studies and found no effects on honey bee survival at field-realistic concentrations. Some larval studies showed reduced survival only at doses hundreds to thousands of times higher than pollen contains.

Researchers tested the direct-toxicity question carefully, and the honey bee data came back mostly reassuring. Wild bees received less attention in those early years, and that gap matters. Bumblebees, squash bees, and solitary bees have different life cycles, different foraging habits, and different exposure routes than a managed honey bee colony.

The field shift: why milkweed disappeared

Around the same time Bt crops arrived, seed companies introduced corn and soy that tolerate glyphosate. That second trait changed what a farm field looks like. A farmer could spray glyphosate over the crop during the growing season, killing weeds without killing the crop. That convenience came with an ecological side effect: milkweed, once common in Midwestern corn and soybean fields, lost its safe haven.

A 2013 study in Insect Conservation and Diversity estimated a 64 percent decline in milkweed in Midwest agricultural fields between 1999 and 2010. The authors tied that decline to the rapid adoption of glyphosate-tolerant crops. Milkweed is the only host plant for monarch butterfly caterpillars. When milkweed left the row crops, monarch breeding habitat shrank with it.

The GMO connection here is indirect. The crop's inserted gene did not poison the monarch. The farming system built around the trait removed a plant that monarchs need.

The monarch numbers: a landscape story

Eastern monarch winter colonies in central Mexico shrank from about 18 hectares of occupied forest in the winter of 1996-1997 to under one hectare in 2023-2024, according to surveys by the World Wildlife Fund and Mexican wildlife authorities. That decline has several causes. Logging in the oyamel fir forests, severe weather, and roadside mowing all take a share. The breeding habitat loss tied to row-crop weed control remains one of the clearest documented drivers.

This is where the street-level argument about GMOs and bees and butterflies falls apart. Blaming the gene misses the field. The more useful question is whether a farming system leaves room for the flowers, weeds, and edge habitat that pollinators move through. A single trait can matter less than the way that trait changes how crops are sprayed, how fields are kept clean, and how much wild vegetation survives from fence line to fence line.

The pesticide that is not a GMO

Some readers will ask about neonicotinoids, because bee decline discussions often land there. Neonicotinoids are a class of insecticides, not a genetic trait. They are applied as seed coatings on both genetically modified and conventional corn, soy, and other crops. The European Union restricted outdoor use of three neonicotinoids in 2018 after the European Food Safety Authority concluded they posed a high risk to bees.

The GMO label does not tell you whether a seed was treated with a neonicotinoid. A non-GMO corn field can carry the same insecticide coating as a GMO corn field. That overlap explains a lot of confusion. People often credit or blame GMO technology for pesticide choices that travel alongside it or instead of it.

Where the research is headed

The next generation of crop breeding is already moving past the two traits we have debated for twenty years. Researchers are testing gene-edited crops with traits like disease resistance, drought tolerance, and altered flowering time. Some plant ecologists are asking whether crops can be bred to support pollinators more directly, for example by flowering longer or producing more accessible nectar.

That is speculative, and I treat it that way. No grocery store shelf carries those crops yet. The more immediate shift is in risk assessment. Regulators used to ask whether the new protein harmed bees in a lab dish. They increasingly ask what the crop does to the whole farm landscape: flowering weeds, hedgerows, water sources, and the timing of pesticide sprays. The unit of study moved from the gene to the field.

What I ask at the farmer's market now

I stopped treating the GMO label as a pollinator safety score. It is not one. A genetically engineered crop can be grown with abundant hedgerows and integrated pest management. A non-GMO crop can be drenched in insecticides that harm bees. The label tells you less about pollinator health than the farmer's practices do.

When I buy produce, I ask or look for answers to three questions:

• Does this farm rotate crops and keep flowering strips or hedgerows?
• Does it use insecticide sprays during bloom?
• Does the seed source, GMO or not, use neonicotinoid coatings?

These questions translate better to the bees on my lavender than any aisle label.

In my own yard, I planted a small patch of common milkweed last fall. This July, I counted eleven bees on one coneflower. Not a controlled study. Just a reminder that the answers I can act on live one garden row away. The gene is one variable. The field is the system. And the bee on my coneflower does not care which word is on the seed bag.