The key point in this paper is this: whereas it has been shown annualism arose in the Aplocheilids twice (in Nothobranchius and Fundulopanchax) there is evidence it arose independantly in Callopanchax as well.

"It would be equally parsimonious to assume three independent origins in the three annual fish genera (i. e., Callopanchax, Nothobranchius and Fundulopanchax DELTRAN option), or two independent origins, once in Callopanchax and once in the base of the clade Nothobranchius + Fundulopanchax + Aphyosemion, with a subsequent loss in Aphyosemion (ACCTRAN option, Fig. 7B). Among these two alternative hypotheses, the cladogram topology derived from the analysis of osteological features parsimoniously supports the hypothesis of a triple annualism origin, once in each annual fish genus (Fig. 7A)."

Note that since the determination that Raddaella is in Aphyosemion and not Fundulopanchax then the "subsequent loss in Aphyosemion" is no longer true.