Enquist & Leimar (1993). The evolution of cooperation in mobile organisms

From the abstract: 
Game theory models based on reciprocity don't account for the switching of partners afforded by mobility.
Their theory: mobility seriously restricts the evolution of cooperation, since an efficient free rider could move rapidly through a population of cooperators, searching out victims to exploit. Thus, behavioral adaptations, such as suspiciousness toward strangers and gossiping may exist to counteract the effects of mobility.
The possible importance of search time is illustrated by comparative data on the relationship between nest sharing and size of nest aggregations in sphecid wasps.

The model:
"Apart from a consideration of interactions within a coalition, one must also take into account properties of the social environment, the most important being the time required to find a new coalition partner."
Search time depends on (a) the density of individuals and (b) the speed with which an individual can move in the environment. To form a coalition, an individual must also find someone who is not already fully engaged and is willing to be a partner.
Results in an effective search time S per coalition. Once formed, a coalition can be terminated by either partner, otherwise it lasts an expected time T. Cooperating costs c and results in the benefit b.

Conditional cooperation (C) strategy starts off cooperating and leaves whenever a partner defects.
Free rider (FR) strategy always defects.

With expected lifespan L, the total utility for C in the majority is:
U(C,C) = (b - c) [T/(S + T)] L
and the total utility for a small minority of FR is:
U(FR,C) = b [1/(S + 1)] L
The results indicate that an increase in both coalition time and search time favors cooperation. 


Small populations
The above model assumes large population with random encounters. However, individuals may live in small groups or populations with restricted migration.
Here, assume that individuals remember defections and refuse to offer cooperation to defecting individuals, the number number of victims to a free rider becomes limited in smaller groups. Additionally, a decrease in group size favors cooperation when accompanied by a decrease in density.

Defenses against free riders: 
Suspiciousness. Cooperators are suspicious (i.e., defect) for a fixed time before offering cooperation. This has negative consequences for both cooperators and free riders, since an association may be terminated before cooperation is offered.  Suspiciousness will be maintained in a heterogeneous population if there is a high enough chance that a new partner will defect.
Gossiping. Information that an individual is unreliable and should not be interacted with spreads through a group. Gossip is exchanged when individuals meet.

Empirical evidence:
Females of solitary wasps (sphecidae) build nests, and provision cells with food for offspring. Sometimes females share nests. Sometimes there is a higher density (aggregation) of nests in an area, which is likely to make search time shorter. As predicted, nest sharing is virtually absent among highly aggregated species, but may be high among species with little to no aggregation.