Egg-trading in simultaneous hermaphrodites

Studying seabasses (Serranidae) may reduce measurements that help determine whether Axelrod's payoff matrix is satisfied. In these shallow water fishes, simultaneous hermaphroditism is a common feature (Fischer 1988, Conner 1992). They are known to be obligate outbreeders with external fertilisation and planktonic eggs. Such a system almost certainly prevents kin selection (Fischer 1988). During mating, each individual divides its clutch of eggs into parcels and, subsequent to courtship displays, alternates with its mate in offering parcels of eggs for fertilisation.
Owing to anisogamy, producing eggs is probably more costly than producing sperm. Thus the 'temptation to defect' i.e. fertilising eggs without giving any to be fertilised, is presumably high. Why do serranids still cooperate? Fischer (1988) suggested the fishes were using TFT and the system would apply to an IPD. However, there are several contradictions to this hypothesis: Fischer himself already pointed out (1988) that by dividing the clutches into a large number of parcels, the benefit b (i.e. T) of fertilising a parcel of eggs is small. He also remarked that defection consists of the absence of an act. i.e. failure to offer eggs, rather than a specific behaviour. In his experiments (Fischer 1988) he referred to retaliation as waiting "significantly longer to release a batch of eggs than [...] if the partner did reciprocate". This is an obvious violation of the TFT rule. He concluded that the fish were playing a nicer version of TFT e.g. TF2T or GTFT. Yet, considering the evolution of the mating system, Connor (1992) proposed that an alternative explanation for cooperation in seabasses is more likely. It was suggested that simultaneous hermaphroditism evolved originally at a low density of conspecifics where this reproductive system has an advantage (Fischer 1988, Conner 1992). At that time the clutches were not parcelled. Changing ecological conditions increased the abundance of mates and parcelling appeared initially because it extended male mating success (Fischer 1988): a parcelling individual in a population of non-parcellers would therefore obtain more eggs from additional matings. Once parcelling became common, the seabasses found themselves in an IPD (Fischer 1988), possibly susceptible for a cheating strategy such as FR or Rover. Further decreasing each investment may ultimately have led to R>T (Connor 1992) and thus the system escaped the IPD. This still remains to be tested.
If Conner's model applies, both individuals will cooperate until the last move, until there is only one parcel left. Not. offering this parcel will favour the withholding of the parcel on the penultimate move and so on. In Serranid seabasses, short spawning periods prevent cheating on the last move, as eggs must be spawned on the same day they are produced or they will become inviable. Since the chance of finding a new mate late in the spawning period is low, it is better to get its eggs fertilised than not to mate at all (Fischer 1988, Conner 1992). Moreover, Lima (1989) reported that cooperation may be stable in a finite IPD whose end point is known to all players.