Byproduct Mutualism among Safe Selfish Sentinels

What if sentinel behavior isn't dangerous after all? Bednekoff (1997) argued that sentinels might be safe and selfish rather than unsafe and selfless. He reasoned that even ifthey increase their risk ofbeing the target ofpredators, an improved ability to detect and avoid predators might outweigh this risk. Even if sentinels expose themselves to minimal predation risk, they must pay an opportunity cost because they cannot forage and act as sentinel simultaneously. We might expect, therefore, that individuals will act as sentinels only when their energetic reserves are high. In addition, if a single sentinel provides an adequate early-warning defense, then even a well-fed individual will serve as a sentinel only when no one else is doing so. According to Bed-nekoff's model, sentinel behavior depends on both the prospective sentinel's nutritional state and the actions ofothers.

Combining these state-dependent and frequency-dependent aspects ofsen-tinel behavior, Bednekoff developed a dynamic game to explore whether a coordinated sentinel system could emerge from the decisions of selfish individuals. In this game, each individual chooses forager or sentinel based on its energetic state and the actions of others. Provided group members share alarm information, a single sentinel greatly reduces everyone's predation risk, and additional sentinels add little protection from surprise attacks (fig. 10.7). Thus, an individual receives a large safety benefit if it acts as a sentinel when all other group members are foraging. However, an individual may not be able to forgo the foraging opportunity if its energetic state is too low. When no other individuals are acting as sentinels, Bednekoffs model predicts that a focal individual will serve as a sentinel even when its reserves are relatively low (fig. 10.8). However, when another individual is already acting as a sentinel, our focal individual is relatively safe, and so it should act as a sentinel only if its energetic state is near the maximum. The net effect is a sentinel system that appears highly coordinated even though simple selfishness guides the actions of each player. Thus, elaborate coordination and altruism emerge as a by-product of simple self-interested behavior.

Two recent studies support this model. First, meerkats (Suricata suricatta), a small social mongoose of South Africa, showed increases in various measures of sentinel activity in response to supplemental feeding (Clutton-Brock et al. 1999). Second, individual Arabian babblers (Turdoides squamiceps), a highly nj 0.15-

-sentinel

forager

• fully shared alarm information O alarm information spread by sentinel □ no shared alarm information

Figure 10.7. Per capita risk of predation for sentinels and foragers as influenced by the number of individuals acting as sentinels. Predicted predation risks are shown forthree information-sharing scenarios: full alarm information sharing by any individual that detects a predator, alarm information sharing only by the sentinel, or no alarm information sharing. An individual may be killed if it fails to detect an approaching predator or if it receives no alarm from other group members. The predation risk faced by a particular individual depends on the actions of others (i.e., whether they are foraging or acting as sentinel and whether they share alarm information). By assumption, each sentinel is three times more likely than a foragerto be killed by an undetected predator, but is also fourtimes more likely to detect an approaching predator. This combination makes a sentinel relatively safe, even if foragers fail to share alarm information. (After Bednekoff 1997.)

-sentinel

forager

• fully shared alarm information O alarm information spread by sentinel □ no shared alarm information

Number of others acting as sentinels

Figure 10.7. Per capita risk of predation for sentinels and foragers as influenced by the number of individuals acting as sentinels. Predicted predation risks are shown forthree information-sharing scenarios: full alarm information sharing by any individual that detects a predator, alarm information sharing only by the sentinel, or no alarm information sharing. An individual may be killed if it fails to detect an approaching predator or if it receives no alarm from other group members. The predation risk faced by a particular individual depends on the actions of others (i.e., whether they are foraging or acting as sentinel and whether they share alarm information). By assumption, each sentinel is three times more likely than a foragerto be killed by an undetected predator, but is also fourtimes more likely to detect an approaching predator. This combination makes a sentinel relatively safe, even if foragers fail to share alarm information. (After Bednekoff 1997.)

Figure 10.8. Predicted frequency distribution of numbers of sentinels underthree information-sharing scenarios. White bars show expected numbers of sentinels (based on binomial distribution) assuming an individual's decision making is independent of the actions of other group members. (After Bednekoff 1997.)

Figure 10.8. Predicted frequency distribution of numbers of sentinels underthree information-sharing scenarios. White bars show expected numbers of sentinels (based on binomial distribution) assuming an individual's decision making is independent of the actions of other group members. (After Bednekoff 1997.)

Was this article helpful?

0 0

Post a comment