The Nile tilapia, Oreochromis niloticus, escaped from a fish farm in Mazabuka, probably in 1987, and was found in September 1992 on a 75 km stretch of the river, mainly downstream from the site of release . In most areas, except near Mazabuka, O. niloticus was relatively uncommon and so far does not seem to have hybridized with the local tilapias .

In aclear-water Nigerian stream, Tilupiumuriue showed lunar periodicity in breeding activity over a 9-week study period. Most clutches were laid during the last quarter of the moon cycle. It is suggested that spawning before full moon enhances the effectiveness of the parental care by allowing a maximal amount of moonlight during the most critical phases in the development of the young and by providing a cue for the pairs to spawn at the same time.

In 152 isolated pairs of Tilap'. m a k e the parents ate 41 % of their broods before these reached the freeswimmin stage. The cannibalized broods were usually eaten at the egg-stage. Most brood-eating was done by t i e males, as concluded from direct observauons and from variation between males (but not between females) in the tendency of eat broods. Cannibalistic behaviour was more common when: 1. the relative weight of female (female weight to male weight) was small (< .4) 2. the relative wei ht was large (> .7) but not too large (> 35) 3. the fernale was fzge (> 70 g) U.S. Copyright Clearance Center Code Statement: 0175-8659/86/0202-0065 $02.50/0 J. Appl. Ichthyol. 2 (1986) 65-74 0 1986 Verlag Paul Parey, Hamburg und Berlin ISSN 0175-8659 1. la taille relative de la female (le poids de la femelle par celui du mile) itait petite (< .4) 2. la taille rilative Ctait grande (> .7) mais pas trop grande (> 3 5 ) 3. la femelle ttait grande (> 70 g) 4. le paire avait fray6 endant des saisons hors du vrai temps du frai 5. le nombre des o e u t des fretins itait rarement petit (< 100 oeufs).Quand il s'agissait des frais h nombre des oeufs normal ni le nornbre des oeufs ni la taille du frai ok des oeufs n'itaient en relation avec le cannibalisme. La probabiliti de cannibalisme n'itait pas alter& ar suite d'un premier frai malhereux; Plusieurs des changes de probabiliti de cannibalisme trouvis sont His au succes reproductif des miles. Eventuellement, ils refletent une variation sous-jacente naturelle adaptive, en partie modifiie par le milieu d'aquarium.

SynopsisParental care of Tilapia mariae was observed in nature (Ethiop River, Nigeria) and in aquaria with or without intruders present. In the field, 2530% of nests are guarded by one parent, normally the female. It is assumed that most missing males have deserted. Males who participate in brood care exhibit both close brood guarding and brood defence at a lower level than females, and hence seem to invest less than females. Broods were guarded under three distinct types: (1) female at the brood, male in surroundings, (2) parents take turns, or (3) parents stay together at the brood. Each pair used predominantly one type until the young swam freely, thereafter type 3. Females defended most in type 3, but male attack rate did not differ among the types. Type 3 seems related to increased risk of brood predation and type 2 to the female's foraging needs, being more common when she is small and the mates do not differ much in size. The unequal guarding times of type 1 indicate rather a low parental investment by the male (and thus risk of desertion) than specialization in roles on equal investment basis. Parental behaviour exhibited in aquaria differed in many ways from that in nature. The role types were indistinct and there were more signs of motivational conflict between the mates. Isolated pairs avoided joint guarding in the embryo period and while switching, female turns were much longer than male turns, unlike in nature. When intruders were added, males attacked them more than did females.

No abstract