Emily S Cassidy, Paul C West, James S Gerber and Jonathan A Foley 2013 Redefining agricultural yields: from tonnes to people nourished per hectare Environ. Res. Lett. 8 034015
http://iopscience.iop.org/1748-9326/8/3/034015/article

Worldwide demand for crops is increasing rapidly due to global population growth, increased biofuel production, and also changing dietary preferences. A Recent study estimates that by 2050 crop production will have to roughly double to meet demands. Meeting these demands for crops will be a substantial challenge, one that taxes the capability of our food system. But to increase food availability, we can also look at crops we are currently growing and how we are using them, to see if we can feed more people with current levels of production. Of particular interest are the crops that we use for animal feed and biofuels.  Currently, 36% of all produced calories are used for animal feed and we find that of those animal feed calories, only 12% end up contributing to human diets in the form of animal products. Now that we know not all of crop production is used for human food, this study re-examines agricultural productivity from its standard definition of yield (or tonnes per hectare) to the actual number of people that we are able to feed per hectare of land. We examine the production and allocation of 41 crops globally, to determine, what fraction of all produced calories are delivered to the food system? Or in other words, what fraction of the calories we produce actually become food that you and I can eat? 

Figure one in the main text shows a map of the proportion of the calories that actually become food. We can see that in areas in the green, a high proportion of the calories that they produce actually become food. Whereas the areas in the red a small proportion of the calories that they produce actually become food. One of the major findings of our paper is that only 59% of the calories we produce on croplands actually become food. In other words, we are losing 41% of the calories produced on croplands and most of this loss is due to inefficient feed to animal product conversion. Calorie delivery fractions differ a lot by country, though. In the U.S. for example, 67% of calories produced go to animals for feed. So if we look at all of the calories produced on U.S. croplands, only 34% of those calories actually become food.  When we look at the number of people we could feed on the calories we're currently producing, versus the number of people we are actually feeding, when you factor in crop allocation and livestock conversion, we find huge potential in countries such as the United States, China, and Brazil. When we look at the number of people we could feed on produced calories we find that on average we could feed about 10 people per hectare. But when we look at the number of people we are actually feeding given crop allocations and livestock conversions, we're only feeding about 6 people per hectare. Shifting crops from their current mix of uses to all food for human consumption could increase calorie availability by 70%, and that's enough calories to feed an additional 4 billion people. We also find that even small changes in diet like shifting away from beef to poultry and pork, or shifting from meat to an ovo-lacto vegetarian diet could increase feed conversion efficiency and provide enough additional calories for over 800 million people. 

Considering the number of people fed per hectare is important for the future of food security. Also recent cropland expansion into the tropics has come at tremendous ecological costs. In this study, we demonstrate that even small changes in diets or shifts in crop allocations, can have big benefits on agricultural productivity.