A species of small mongooses in Africa called meerkats share sentinel (guard) duties to warn other group members by repeating alarm calls if a predator is seen. This is an important job, because when meerkats are foraging, their heads are in the ground seeking prey, and they cannot see a predator coming. The question is, why do group members take turns acting as sentinels? Kin selection, that is, being able to save the lives of family members can be one hypothesis for this type of sentinel behavior. Family members share copies of a meerkat's genes. Kin selection is achieved by helping a meerkat's own offspring as well as non descendant kin, including sibling, nieces, nephews, aunts, and uncles. Therefore, if members of a certain group are closely related, a sentinel ensures that copies of its genes can be passed on to future generations by saving the majority of family members by alerting others, even at the expense of its own life. Assuming this hypothesis is true, we can predict that group members have close genetic ties. Otherwise, kin selection would not work. But this prediction does hold true. A dominant, breeding female is mother to 75 percent of all the litters in a group, and one dominant male fathers 75 percent of all the pups born. Even though a typical meerkat group includes a few immigrants, most subordinate adults are siblings or half siblings. Therefore, it is likely that subordinate adults share 25 or 50 percent of their genes. On account of most meerkat group members being family, it is possible that kin selection has favored sentinel behavior. Nonetheless, by itself, a close inherent relationship is not enough evidence to conclude that kin selection has played a role. Thus, we need further evidence, and must improve the prediction. Based on the same hypothesis, a more specific prediction is that each mongoose should increase the frequency of sentinel behavior when they are guarding family members. This new prediction needed testing, so the group was observed to determine which members stand guard and when. The immigrants without any kin relations to other group members acted as sentinels just as much as the individuals with many relatives nearby. Therefore, the result of this test does not support the kin-selection hypothesis. Another hypothesis that is often suggested to explain such cooperative behavior is that it results from reciprocal altruism~each individual takes turns standing guard to benefit the rest of the group, rather than itself. The reciprocal altruism theory can work only when those who cheat by avoiding guard duty can be identified and punished by the rest of the group. This hypothesis produces the prediction that there should be a regular rotation of sentry duty within the group and that the ones who neglect this duty should be chastised. However, this is not observed. In fact, the group members do take turns on sentry duty, but there is no predetermined order for this. In addition, when some members shorten their shift, other group members increase their contributions to compensate. The predictions and observations of the reciprocal-altruism hypothesis do not coincide with each other. Yet another hypothesis for the evolution of meerkat sentinel behavior is that it results from selfish antipredator behavior. This idea stems from the fact that the meerkat watching for predators increases its personal safety, and warning others does not harbor any disadvantage. So, when a meerkat has had enough to eat, it should watch for predators. The sentinel on duty can then return to foraging. This hypothesis produces a prediction that sentinel duty is not dangerous or risky in any way. This does seem to be true. Over the course of 2,000 hours of observation, no sentinels were attacked or killed by predators. They may actually be safer because they are the first to sense the predator. Moreover, they generally stand guard within 5 meters of a burrow, and are the first underground when a predator comes close. If a meerkat's personal safety is increased with serving as a sentinel, it would be possible to predict that an individual would spend a proportion of its time guarding, whether it was solitary or part of a group. As predicted, individual meerkats spend about the same time on guard duty as members of large groups. Groups with more members suffer less predation because there is a sentinel for a longer portion of foraging time than in small groups.