Initial data came from the massive Ethnographic Atlas by Dr. George P. Murdock, and was analyzed further by Cordain and his collaborators. Cordain is a professor at Colorado State University, and a longtime proponent of paleolithic diets for health. He has written extensively about the detrimental effects of grains and other modern foods. Here's his website.
The researchers broke food down into three categories: hunted animal foods, fished animal foods and gathered foods. "Gathered foods" are primarily plants, but include some animal foods as well:
Although in the present analysis we assumed that gathering would only include plant foods, Murdock indicated that gathering activities could also include the collection of small land fauna (insects, invertebrates, small mammals, amphibians, and reptiles); therefore, the compiled data may overestimate the relative contribution of gathered plant foods in the average hunter-gatherer diet.There are a number of striking things about the data once you sum them up. First of all, diet composition varied widely. Many groups were almost totally carnivorous, with 46 getting over 85% of their calories from hunted foods. However, not a single group out of 229 was vegetarian or vegan. No group got less than 15% of their calories from hunted foods, and only 2 of 229 groups ate 76-85% of their calories from gathered foods (don't forget, "gathered foods" also includes small animals). On average, the hunter-gatherer groups analyzed got about 70% of their calories from hunted foods. This makes the case that meat-heavy omnivory is our preferred ecological niche. However, it also shows that we can thrive on a plant-rich diet containing modest amounts of quality animal foods.
The paper also discusses the nature of the plant foods hunter-gatherers ate. Although they ate a wide variety of plants occasionally, more typically they relied on a small number of staple foods with a high energy density. There's a table in the paper that lists the most commonly eaten plant foods. "Vegetables" are notably underrepresented. The most commonly eaten plant foods are fruit, underground storage organs (tubers, roots, corms, bulbs), nuts and other seeds. Leaves and other low-calorie plant parts were used much less frequently.
The paper also gets into the macronutrient composition of hunter-gatherer diets. I think his methods may have a tendency to de-emphasize fat consumption. He writes that
...the most plausible... percentages of total energy from the macronutrients would be 19-35% for protein, 22-40% for carbohydrate, and 28-58% for fat.He derives these numbers from projections based on the average composition of plant foods, and the whole-body composition of representative animal foods (includes organs, marrow, blood etc., which they typically ate). However, as he notes in his paper, humans can't tolerate more than about 35% of calories from protein before developing "rabbit starvation". Therefore, carnivorous groups must have been getting at least 65% of their calories from fat, and probably more in many cases.
Second, he generalizes using data from animals with "typical" body composition, whereas hunter-gatherers often actively sought the fattest animals they could find. For example, he cites squirrels as having a body fat percentage of 2%, but anyone who has seen a squirrel in the fall knows that they can be much fatter than that. Older, larger ungulates (deer, elk, etc.) were preferentially sought out by many groups due to their high body fat percentage. It's an obvious strategy because more fat = more calories. Natives on the North American Pacific coast rendered fat from fish, seals, bears and whales, using it liberally in their food. Here's an excerpt from The Northwest Coast by James Swan, who spent three years living among the natives of the Washington coast in the 1850s:
About a month after my return from the treaty, a whale was washed ashore on the beach between Toke's Point and Gray's Harbor and all the Indians about the Bay went to get their share... The Indians were camped near by out of the reach of the tide, and were all very busy on my arrival securing the blubber either to carry home to their lodges or boiling it out on the spot, provided they happened to have bladders or barrels to put the oil in. Those who were trying out [rendering] the blubber cut it into strips about two inches wide, one and a half inches thick, and a foot long. These strips were then thrown into a kettle of boiling water, and as the grease tried out it was skimmed off with clam shells and thrown into a tub to cool and settle. It was then carefully skimmed off again and put into the barrels or bladders for use. After the strips of blubber have been boiled, they are hung up in the smoke to dry and are then eaten. I have tried this sort of food but must confess that, like crow meat, "I didn't hanker arter it".Despite the bias against fat consumption, I was very impressed by the paper overall. I think it presents a good, simple model for eating well: eat whole foods that are similar to those that hunter-gatherers would have eaten, including at least 20% of calories from high-quality animal sources. Organs are mandatory, vegetables may not be. Sorry, Grandma.