Guns, Germs & Steel

900; and at 0.2 miles per year, for the llama from Peru north to j These differences could be even greater if corn was not domesticated i Mexico as late as 3500 b.c., as I assumed conservatively for these tions, and as some archaeologists now assume, but if it was instead * ticated considerably earlier, as most archaeologists used to assume (a many still do).
There were also great differences in the completeness with which! of crops and livestock spread, again implying stronger or weaker to their spreading. For instance, while most of Southwest Asia's fou crops and livestock did spread west to Europe and east to the Indus Va neither of the Andes' domestic mammals (the llama / alpaca and the | pig) ever reached Mesoamerica in pre-Columbian times. That astor failure cries out for explanation. After all, Mesoamerica did develop farming populations and complex societies, so there can be no doubt i Andean domestic animals (if they had been available) would have valuable for food, transport, and wool. Except for dogs, Mesoamerica ' utterly without indigenous mammals to fill those needs. Some South ican crops nevertheless did succeed in reaching Mesoamerica, such as i ioc, sweet potatoes, and peanuts. What selective barrier let those through but screened out llamas and guinea pigs?
A subtler expression of this geographically varying ease of spread i|f phenomenon termed preemptive domestication. Most of the wild species from which our crops were derived vary genetically from : area, because alternative mutations had become established among; wild ancestral populations of different areas. Similarly, the required to transform wild plants into crops can in principle be

SPACIOUSSKIES AND TILTED AXES • I 7 9
t by alternative new mutations or alternative courses of selection to • Id equivalent results. In this light, one can examine a crop widespread historic times and ask whether all of its varieties show the same wild tation or same transforming mutation. The purpose of this examination is to try to figure out whether the crop was developed in just one area or else independently in several areas.
If one carries out such a genetic analysis for major ancient New World rops many of them prove to include two or more of those alternative wild variants, or two or more of those alternative transforming mutations. This suggests that the crop was domesticated independently in at least two different areas, and that some varieties of the crop inherited the particular mutation of one area while other varieties of the same crop inherited the mutation of another area. On this basis, botanists conclude that lima beans (Phaseolus lunatus), common beans (Phaseolus vulgaris), and chili peppers of the Capsicum annuutn I chinense group were all domesticated on at least two separate occasions, once in Mesoamerica and once in South America; and that the squash Cucurbita pepo and the seed plant goosefoot were also domesticated independently at least twice, once in Mesoamerica and once in the eastern United States. In contrast, most ancient Southwest Asian crops exhibit just one of the alternative wild variants or alternative transforming mutations, suggesting that all modern varieties of that particular crop stem from only a single domestication.
What does it imply if the same crop has been repeatedly and independently domesticated in several different parts of its wild range, and not just once and in a single area? We have already seen that plant domestication involves the modification of wild plants so that they become more useful to humans by virtue of larger seeds, a less bitter taste, or other qualities. Hence if a productive crop is already available, incipient farmers will surely proceed to grow it rather than start all over again by gathering its not yet so useful wild relative and redomesticating it. Evidence for just a single domestication thus suggests that, once a wild plant had been domesticated, the crop spread quickly to other areas throughout the wild plant's range, preempting the need for other independent domestications o e same plant. However, when we find evidence that the same wild ancestor was domesticated independently in different areas, we infer that e crop spread too slowly to preempt its domestication elsewhere. The evi ence for predominantly single domestications in Southwest Asia, but requent multiple domestications in the Americas, might thus provide

I 8 O •GUNS,GERMS, AND STEEL
more subtle evidence that crops spread more easily out of Southwest Asia than in the Americas.
Rapid spread of a crop may preempt domestication not only of the same wild ancestral species somewhere else but also of related wild species. If you're already growing good peas, it's of course pointless to start from scratch to domesticate the same wild ancestral pea again, but it's also pointless to domesticate closely related wild pea species that for farmers are virtually equivalent to the already domesticated pea species. All of Southwest Asia's founder crops preempted domestication of any of their close relatives throughout the whole expanse of western Eurasia. In contrast, the New World presents many cases of equivalent and closely related, but nevertheless distinct, species having been domesticated in Meso-america and South America. For instance, 95 percent of the cotton grown in the world today belongs to the cotton species Gossypium hirsutum, which was domesticated in prehistoric times in Mesoamerica. However, prehistoric South American farmers instead grew the related cotton Gossypium barbadense. Evidently, Mesoamerican cotton had such difficulty reaching South America that it failed in the prehistoric era to preempt the domestication of a different cotton species there (and vice versa). Chili peppers, squashes, amaranths, and chenopods are other crops of which different but related species were domesticated in Mesoamerica and South America, since no species was able to spread fast enough to preempt the others.
We thus have many different phenomena converging on the same conclusion: that food production spread more readily out of Southwest Asia than in the Americas, and possibly also than in sub-Saharan Africa. Those phenomena include food production's complete failure to reach some ecologically suitable areas; the differences in its rate and selectivity of spread; and the differences in whether the earliest domesticated crops preempted redomestications of the same species or domestications of close relatives. What was it about the Americas and Africa that made the spread of food production more difficult there than in Eurasia?
To answer this question, let's begin by examining the rapid spread of food production out of Southwest Asia (the Fertile Crescent). Soon after food production arose there, somewhat before 8000 b.c., a centrifugal wave of it appeared in other parts of western Eurasia and North Africa

SPACIOUSSKIES AND TILTED AXES • I 8 I
farther and farther removed from the Fertile Crescent, to the west and east. On this page I have redrawn the striking map (Figure 10.2) assembled by the geneticist Daniel Zohary and botanist Maria Hopf, in which they illustrate how the wave had reached Greece and Cyprus and the Indian subcontinent by 6500 b.c., Egypt soon after 6000 b.c., central Europe by 5400 b.c., southern Spain by 5200 b.c., and Britain around 3500 b.c. In each of those areas, food production was initiated by some of the same suite of domestic plants and animals that launched it in the Fertile Crescent. In addition, the Fertile Crescent package penetrated Africa southward to Ethiopia at some still-uncertain date. However, Ethiopia also developed many indigenous crops, and we do not yet know whether it was these crops or the arriving Fertile Crescent crops that launched Ethiopian food production.
The spread of Fertile Crescent crops across western Eurasia