While gorging may help you carry yourself through a few weeks of food scarcity, it won’t protect you against a year of starvation. One long-term solution is to make reciprocal agreements with neighboring groups about sharing food when one group’s area has enough food and another group’s area is suffering from a food shortage. Local food availability fluctuates with time in any area. But two areas located a sufficient distance apart are likely to have fluctuations in food availability that are out of phase. That opens the door for your group to reach a mutually advantageous agreement with another group, such that they allow you onto their land or send you food when they have enough food but you don’t, and your group returns the favor when it’s the other group that’s short of food.
For example, in the area of the Kalahari Desert occupied by the !Kung San, rainfall in a given month varies by up to a factor of 10 between different sites. The result, in Richard Lee’s words, is that “the desert may be blooming in one area and a few hours’ walk away, the land may still be parched.” As one example, Lee compared monthly rainfall at five sites in the Ghanzi district for 12 months from July 1966 to June 1967. The total rainfall for the year varied by less than a factor of 2 between sites, but rainfall in a given month varied among sites from no rainfall at all to 10 inches. The site of Cume had the highest annual rainfall but was nevertheless the driest of the five sites in May 1967 and the second driest in November 1966 and February 1967. Conversely, Kalkfontein had the lowest annual rainfall, but it was the second-wettest site in March 1967 and again in May 1967. Hence for any site, a group confined to that site would be certain to experience droughts and food shortages at certain times, but could usually find some other group whose site was wet and flourishing—provided that the two groups had agreed to help each other in times of need. In fact, such generalized reciprocity is essential to the !Kung’s ability to survive in their locally unpredictable desert environment.
Reciprocity (punctuated occasionally by hostility) is widespread among traditional societies. Trobriand Island villages distribute food between villages to even out local food shortages. Among the Iñupiat of northern Alaska, individual families in times of local famine moved to live with relatives or partners in another district. The most important fruits consumed by South America’s Yanomamo Indians come from groves of peach palm trees and plantain trees, both of which (especially the former) produce harvests more abundant than a local group can consume by itself. The fruits spoil after ripening and cannot be stored, so they have to be eaten while ripe. When a local group finds itself with a surplus, it invites neighbors to come for a feast, in the expectation that those neighbors will reciprocate when they in turn produce a food surplus.
Scatter your land
The other common long-term solution to the unpredictable risk of a local food shortage is to scatter your land-holdings. I encountered this phenomenon in New Guinea when, while out bird-watching one day, I stumbled across a New Guinea friend’s garden clearing in the middle of forest a mile northeast of his village, and several miles from his other gardens scattered to the south and west of his village. What on earth did he have in mind, I asked myself, when he chose that isolated location for his new garden? It seemed so inefficient to commit himself to a waste of travel time, and the garden’s remoteness made it hard to protect from marauding pigs and thieves. But New Guineans are smart and experienced gardeners. If you see them doing something that you initially don’t understand, there usually turns out to be a reason. What was his motive?
Other Western scholars and development experts have been equally puzzled by other cases of field scattering elsewhere in the world. The example most often discussed involves medieval English peasants, who tilled dozens of tiny scattered plots. To modern economic historians, that was “obviously” a bad idea because of the resulting wasted travel and transport time and inevitable unplowed strips between plots. A similar modern case of field scattering by Andean peasant farmers near Lake Titicaca, studied by Carol Goland, provoked development experts to write in exasperation, “The peasants’ cumulative agricultural efficiency is so appalling…that our amazement is how these people even survive at all…. Because inheritance and marriage traditions continually fragment and scatter a peasant’s fields over numerous villages, the average peasant spends three-quarters of his day walking between fields that sometimes measure less than a few square feet.” The experts proposed land-swapping among farmers in order to consolidate their holdings.
But Goland’s quantitative study in the Peruvian Andes showed that there really is method to such apparent madness. In the Cuyo Cuyo district, the peasant farmers whom Goland studied grow potatoes and other crops in scattered fields: on the average 17 fields, up to a maximum of 26 fields, per farmer, each field with an average size of only 50 by 50 feet. Because the farmers occasionally rent or buy fields, it would be perfectly possible for them in that way to consolidate their holdings, but they don’t. Why not?
A clue noticed by Goland was the variation in crop yield from field to field, and from year to year. Only a small part of that variation is predictable from the environmental factors of field elevation, slope, and exposure, and from work-related factors under the peasants’ control (such as their effort in fertilizing and weeding the field, seed density, and planting date). Most of that variation is instead unpredictable, uncontrollable, and somehow related to the local amount and timing of rain for that year, frosts, crop diseases, pests, and theft by people. In any given year there are big differences between yields of different fields, but a peasant can’t predict which particular field is going to produce well in any particular year.
What a Cuyo Cuyo peasant family has to do at all costs is to avoid ending up at the end of any year with a low harvest that would leave the family starving. In the Cuyo Cuyo area, farmers can’t produce enough storable food surpluses in a good year to carry them through a subsequent bad year. Hence it is not the peasant’s goal to produce the highest possible time-averaged crop yield, averaged over many years. If your time-averaged yield is marvelously high as a result of the combination of nine great years and one year of crop failure, you will still starve to death in that one year of crop failure before you can look back to congratulate yourself on your great time-averaged yield. Instead, the peasant’s aim is to make sure to produce a yield above the starvation level in every single year, even though the time-averaged yield may not be highest. That’s why field scattering may make sense. If you have just one big field, no matter how good it is on the average, you will starve when the inevitable occasional year arrives in which your one field has a low yield. But if you have many different fields, varying independently of each other, then in any given year some of your fields will produce well even when your other fields are producing poorly.
To test this hypothesis, Goland measured the yields of all the fields of 20 families—488 individual fields in all—in each of two successive years. She then calculated what each family’s total crop yield, pooled over all their fields, would have been if, while still cultivating the same total field area, they had concentrated all their fields at one of their actual locations, or if instead they had scattered their fields at 2, 3, 4, etc. up to 14 different ones of the actual locations. It turned out that, the more numerous were the scattered locations, the lower was the calculated time-averaged yield, but also the lower was the risk of ever dropping below the starvation yield level. For instance, a family that Goland labeled family Q, which consisted of a middle-aged husband and wife and a 15-year-old daughter, was estimated to need 1.35 tons of potatoes per acre of land per year in order to avoid starvation. For that family, planting at just a single location would have meant a high risk (37%!) of starving in any given year. It would have been no consolation to family Q, as they sat starving to death in a bad year such as arrives about once in every three years, to reflect that that choice of a single location gave them the highest time-averaged yield of 3.4 tons per acre, more than double the starvation level. Combinations of up to six locations also exposed them to the risk of occasional starvation. Only if they planted seven or more locations did their risk of starvation drop to zero. Granted, their average yield for seven or more locations had dropped to 1.9 tons per acre, but it never dropped below 1.5 tons per acre, so they never starved.