Start with the DietA precise answer is impossible because so many variable factors are at play, including the productivity of the agricultural land. But actually, the first step in answering this is to know the diet being considered (including any big holiday turkeys consumed). The current U.S. diet is shown nicely in the graph below from Visual Economics. To summarize, the average American consumes about 2000 lbs of food per year, which works out to about 5.5 lbs and 2700 calories per day–or nearly your entire body weight in food per month. Divide those daily 2700 calories by 5.5 lbs and you get 490 calories per pound of food, on average. There are differences in the quality of various parts of the diet that are important to appreciate, including caloric density. Fruits and vegetables are abundant in the diet by weight and give us the flavors, fiber, vitamins and minerals we crave, but only typically provide 50-150 calories per pound. By contrast, a single slice of my favorite bread (pictured below) has 110 calories and only weighs a tenth of a pound. Oils and fats are about three to four times more dense, calorie-wise, than bread. Meat tends to have slightly fewer calories per pound than high starch foods. For example, boneless lamb chops without the edge fat are around 976 calories per pound, according to the USDA’s Food-A-Pedia, which I could peruse for hours. Low-fat milk, which is mostly water, still has about 200 calories per pound (about a pint).
Converting to Area
If we take the average U.S. diet as our starting point, we can convert each component of this diet into the area needed to produce it by using average U.S. harvest yields. For example, the USDA reported recently that the average corn harvest was 147 bushels per acre, or about 8250 lbs. It takes a true professional to sort out how much of this corn gets into the human food supply, since corn is normally eaten in highly processed and modified forms. The vast majority of corn is roughly split between ethanol factories and animal feed, with perhaps 10% or less used for food directly (e.g., polenta) and via food processing (e.g., gummy bears).
This sort of complexity is why I must rely on others to make the diet to area conversion. The most recent studies I am aware of were done for the state of New York by a team of Cornell scientists led by Christian Peters. Here’s a link to one of the published papers, but a more accessible review is also available and highly recommend for those who hunger for more information.
Below I have posted a key summary graphic from the paper. Along the Y-axis is land area in hectares needed to feed one person for a year, which is dependent upon the model diets shown along the X-axis. Each model diet is labeled by two dietary factors: meat and fat.
Adding Meat Feeds More People
The finding that gained headlines from this study some years ago had to do with the fact that adding some meat and dairy to the diet, while increasing land area, actually fed more people. This is because much land is not suited to annual crops but can be sown in pasture (most of the “perennial crops” shown in the bar graph are pasture). Cutting the average meat consumption roughly in half, which would de-emphasize hogs and poultry in the diet as these rely on grains, actually feeds more people than a vegetarian diet.
And the Answer Is…
Since the area of production needed is most sensitive to meat and fat consumption, we can see which of the model diets in the Cornell study is closest to the typical American diet to estimate the per capita area given current habits. To gauge the average, look at the middle of the chart above the 190 grams of meat per day and you’ll see that this converts to about 0.45 hectares, which is just a bit over one acre.
It is fair to ask if New York is representative of the U.S. in terms of agricultural potential. I actually think it is pretty “average” having a mix of both good and poor soils, mountains and plains and a climate that is neither the most benign nor most extreme. Certainly California and Iowa are not average so we shouldn’t be extrapolating from those best cases.
It would be nice, and possibly critical, to have this sort of research done more extensively. To that end, the Cornell group has a grant to develop a Local Foodshed Mapping Tool. It is being created for New York but the methods should be applicable anywhere.
Those who are savvy about how food is produced will have many follow-up questions to this direction of thought. For example, crop yields are no longer a simple function of Nature’s endowment of soil, the blessings of good weather, and irrepressible seed germination. Nearly all farmers rely on a steady stream of outside inputs in forms such as ammonia-nitrate and super phosphate. These derive from concentrated below ground sources of energy and raw materials deposited over geologic time. As I’ve explored before on this blog, food supply is over-correlated for my comfort zone with oil supply. Over the past few years I’ve also written about techniques for de-linking food production from massive external inputs. But that is a long discussion that has no easy answer either.
I’ll just add that addressing the outside inputs conundrum makes one consider the role that well-managed grazing systems have in an agriculture that can sequester carbon, clean water, and build soil fertility more endemically. And for those who claim we don’t have the land area to do this, take a look at the acres of corn sown each year (about 100 million acres) and how much of that is used for direct human consumption (about 10 million acres) in the U.S. Looking at the numbers clearly shows we have a problem of too much artificially created demand. Why not put pasture on 90 million acres of cropland and let the ruminants eat their evolved diet?
Most people are not looking forward to a 10 minute lecture when they ask me a supposedly simple question. So while there are many variables and lots of imprecision when answering “How much land is needed to feed a person,” for today’s American diet, with today’s agricultural system, I’ll stick with about one acre.