Better management of grasslands, and some of their inevitable, but undesirable, expansion in Latin America, Africa, and Asia (as nearly all new grazing land will be created through deforestation) will supply only a small fraction of future increases in meat and milk production and more than 90% of additional output of animal foods will have to come from growing more concentrate feeds, above all corn and soybeans, in direct competition with food crops. This trend has been one of the most obvious features of the 20th-century cropping. In 1900 just over 10% of the global grain harvest was consumed by animals; by 1950 the share surpassed 20% and now it is approaching 50%, with national shares ranging from nearly 70% in the USA to less than 5% in India.
But future feeding requirements are not irrevocably determined by a given demand for animal protein. As these proteins are all of the highest quality they are mutually substitutable and long-term dietary shifts can reduce the relative, or even absolute, consumption of one kind of animal food while greatly increasing the demand for another. For example, in 1950 chicken made up less than 10% of average US per capita meat consumption, but in 2000 its share was about one-third (USDA, 1950–2001). Consequently, a future combination of animal foodstuffs that is inherently more efficient to produce can lead to substantial feed savings in comparison to the prevail-ing consumption pattern.
Milk is by far the most efficient animal food. Highly productive animals need just 1–1.1 kg of concentrate feed per kg of milk. They convert more than 30% of the total metabolizable feed energy, and 30–40% of feed protein to milk protein. Eggs come second (current best practices need less than 3 kg of feed per kg of eggs), and chicken third. Large-scale broiler production requires fewer than 5 units of concentrate feed per unit of edible tissue and feed proteins are converted into meat proteins with efficiency averaging 20%, twice as high as in pigs. Pigs grown for their lean meat can turn more than 20% of metabolizable energy into edible tissues and they need about 7.5 units of concentrate per kg of meat; pigs are also the most efficient convertors of feed energy into edible lipids which give meat its satisfying palatability.
Conversion efficiencies for beef clearly indicate the extravagant cost of that meat; feedlot-fed animals need at least 20 kg of corn and soybeans for each kg of meat, and only 5% of all fed protein is converted into protein in beef. Obviously, only those ani- mals that do not require any concentrates, i.e., raised completely by grazing, or con-suming crop-processing residues such as brans and oilseed cakes, that is digesting biomass that cannot be used by nonruminant species, can be seen as efficient users of agricultural resources. Consequently, it is difficult to imagine a less-desirable change of dietary habits than the global expansion of hamburger empires.
Aquacultured herbivorous fishes are better convertors of feed than pork or chicken. Conversion ratios (kg of feed per kg of live weight) for semi-intensively bred carp in warm waters are 1.4–1.8, and for catfish 1.4–1.6. As a smaller share of fish total mass is wasted in comparison with mammalian or bird carcasses, herbivorous fish need fewer than 2.5 kg of concentrate feed per unit of edible weight, and their protein con-version efficiency is as good as chicken (Smil, 2000). Salmon are even better protein convertors, but these carnivores need fish oils and proteins and their feeding actually results in a net protein loss. Use of concentrate feed in aquaculture of herbivorous species is thus an excellent way of increasing global availability of animal protein and the FAO believes that this rapidly growing enterprise has excellent prospects.
