In the Banda district of west-central Ghana, July is the hungry season. This year's sorghum, yams and millet are still young and green in the rain-fed fields, and for most farmers, last year's harvest is long gone.

People survive on cassava. They grind the roots and cook a polenta-like porridge called tuo zaafe and they stir the leaves into a soup. But there isn't enough to go around always, and the meal lacks protein. It's hard to know whether autumn will bring more food: Rains in Banda have been erratic lately and harvests sparse. The region has been in the midst of a 40-year drought.

It's easy to think that life has always been this way in Banda — a poor, mostly agricultural district, a 10-hour drive from Ghana's thriving capital, Accra. But according to Northwestern University archaeologist Amanda Logan, that could not be further from the truth.

Logan says the hungry-season gap likely didn't exist in the past. In fact, her research shows that before the mid-19th century, people here usually had enough to eat — even when rains failed.

In a paper published Wednesday in American Anthropologist, Logan reports that food security in Banda peaked about 500 years ago, smack in the middle of an epic drought. By contrast, a much milder dry spell is currently wreaking havoc on local diets.

Logan has spent the past eight years examining archaeological artifacts — dug up by her and archaeologists before her — spanning a 1,000-year timeline, searching for indirect clues to food shortage and abundance.

She began by looking at charred grains — typically left over from cooking, and occasionally from kitchen fires. These grains provide a window into past eating habits of residents.

Using more than 300 samples from 10 sites representing different time periods, Logan identified the grains and tracked changes in their relative quantities.

She wanted to know when people were eating foods that they loved, and when they switched to less-preferred ones, a key sign of food insecurity. (It's like switching from steak and potatoes to mac and cheese when you're broke.)

From the 11th through 15th centuries, she found, people mostly ate pearl millet, a grain historically loved by communities all over West Africa. Other artifacts, such as beads from Afghanistan and locally made iron bangles, show that during this period, merchants were plugged into trade networks, and local artisans were busy. That suggests there was enough food to feed a significant number of people who weren't farming. In other words, the people of Banda were thriving.

Then, in the middle of the 15th century, a two-century-long drought set in — sedimentary records from nearby Lake Bosumtwi tell the story.

"That drought, in terms of its severity and length, is like nothing we've seen in modern Africa," Logan says. "It's really intense."

But here's the mystery: The archaeological record during this period shows no signs of food stress — no big increase in wild plant remains, which people often eat to get through famines; no shift to less-preferable foods; no major declines in population. People kept eating millet. And a wide range of iron, copper, ceramic, ivory and cloth artifacts show that trade and craft production were still thriving.

It wasn't until the mid- to late 1800s, long after the drought ended, that Logan began to turn up evidence of food stress. Present-day residents of Banda still talk about ancestors around that time eating wild plants to survive, and the archaeological record backs them up: Four new types of wild plant seeds appear in dwellings from that period. A little later on, people went from eating millet to maize, a historically less-favored staple. They also began to eat cassava, also not a favorite among locals historically. Today, the hungry season has become a fact of life in Banda.

So what happened between the 15th and 21st centuries to explain these changes?
According to Logan, two key things: The slave trade siphoned off many young farmers and artisans, and Banda was incorporated into Britain's Gold Coast colony in the late 1800s. The British wanted to expand markets for their own industrial goods like iron and cloth, so they undercut local production of these items.

"Five hundred years ago, Banda was a producer as well as a consumer of highly sought-after stuff [like] gold, ivory, iron and copper," she says. "As you get to the colonial period, Banda stops being a producer of anything but agricultural and locally consumed goods" like pottery.

These changes weakened Banda's economy, and consequently, crippled residents' ability to survive drought and other disasters. The region remained reliant on agriculture even after Ghana became independent in 1957.

Today, over 70 percent of residents work in farming, fishing or forestry. Because they sell much of their harvest to earn cash, families often run short of food for themselves and have to buy more at the market. If crops fail or prices rise at the wrong time, they go hungry.

Back during the drought in the 1400s, Logan thinks people may have used income from craft production to buy food. Or that non-farm income spared them from selling food they grew, leaving them with enough to get through the year. They also may have shared food among themselves, so that the poor did not starve.

The geographer Michael Watts has shown that this latter strategy was common in northern Nigeria before that region became a British colony.

Scott MacEachern, a professor of anthropology at Bowdoin College and president of the Society of Africanist Archaeologists, says Logan has strong evidence documenting the long-term decline in Banda's food security. And her argument explaining that decline is convincing.

"It fits really well with the historical record," says MacEachern, who was not involved in the study. "We tend to think of colonization as a fairly dry process, as essentially changes in government. On the ground, they were fantastically disruptive processes to the patterns of everyday life. So it's entirely plausible that the decline in food security she talks about is associated with those processes."

Logan isn't the first to highlight the role of colonialism on food security in parts of the world. Geographers like Watts and economists like Amartya Sen have linked colonial policy to hunger for decades. But Logan is among the first to do so using archaeological evidence, says MacEachern.

The new study is important because it extends the story much further into the past, says Arizona State University archaeologist Michelle Hegmon.

Logan's findings, Hegmon says, parallel what economists and historians have already found — that food insecurity isn't caused simply by drought. "It's caused by economics and colonialism and the way people have to produce for market and things like that," she says.