In the Great Lakes of the East African Rift Valley, cichlid fishes have diversified into hundreds of species with great variety of ecomorphological, secondary sexual and behavioural characteristics. A strong role for sexual selection in their speciation is indicated by the presence of many closely related ecologically similar sympatric species which differ in male secondary sexual characteristics. A review of previous research finds that reproductive isolation by direct mate choice may be a common isolating barrier among sympatric species. Observations of partial assortative mating among divergent geographic populations have lead to theories of intralacustrine allopatric speciation of habitat specialists by divergence of signal traits involved in conspecific recognition. Here I demonstrate that signal and preference divergence among populations is not limited to patchily distributed lacustrine endemics, but can also occur in the widely distributed riverine generalist lineage that is phylogenetically basal to the large lacustrine radiations, suggesting a role for divergence of secondary sexual traits in allopatry throughout adaptive radiation. This thesis also adds to the evidence for ecological divergence and peripatric speciation of lacustrine habitat specialists in the absence of significant colour differentiation. In simulated intruder choice tests, males tend to bias aggression towards males of their own species or populations, which may aid in the co-existence of allopatrically diverged populations under secondary sympatry and help to facilitate speciation, or even drive divergence of male traits that are involved in signalling during both courtship and territorial interactions. Results from preliminary investigations into individual variation in mate preference suggest that partial assortative mating among allopatric populations may be due to variation within, rather than among, individuals in their choice of mate. Within-individual variation may be unlikely to cause divergence, but has the potential to aid in colonisation of new areas by a relaxation of directional selection allowing for divergence of male traits by environmental or other social selection pressures. To draw any firm conclusions about the impact of variation in mate choice on speciation, much more data from a wider variety of lineages is needed. Mate choice experiments, carried out as part of a larger ongoing project into sympatric divergence of haplochromine cichlids in isolated crater lakes, found little consistency of female mate choice despite morphological and genetic differentiation of forms. This indicates that, in contrast to previously tested sympatric species of cichHds, at the beginnings of adaptive radiation variety may not be maintained solely through reproductive isolation by direct mate choice. Observations of interactions among animals may frequently reveal unexpected results with regards to what is and is not a species - for example, strong assortative mating among phenotypically similar allopatric populations and that assortative mating alone may not maintain diversity in sympatric populations. Behavioural experiments can complement more modern genetic and genomic techniques and therefore continue to be a valuable tool in the study of speciation and the mechanisms involved in divergence and the maintenance of diversity.

Aggression is ubiquitous in the animal kingdom and a major field of investigation within the area of behavioural ecology. The haplochromine cichlid fish have been used extensively in studies relating not only to the behavioural ecology of aggression but also with respect to how aggression may play a role in the spectacular diversity found within the African Great Lakes. In the first half of this thesis, I investigate the nature of aggression within the rock dwelling fish of Lake Malawi. Aggression biases within polymorphic populations are initially investigated with field work determining the consequences of rare morph advantage. This is followed by experiments determining which cues males and females may use; lastly a pair of recently diverged allopatric species is studied to determine what may happen should secondary contact occur. The second half of this thesis focuses on the behaviorual ecology of female aggression. Females exhibiting the ancestral condition of post brood care are compared to a species with the more derived condition of no post release care. Non-maternal aggression is investigated with comparisons of the type of behavior used by males and females. When and why females are aggressive is also investigated. The results of this thesis suggest that overall a common morph bias may exist in some populations and the consequences are manifested in a rare morph advantage. Colour is important in aggression biases but potentially not limited to the dorsal region. Females and males use different kinds of aggressive behaviors and brooding females are able to vary levels of aggression towards different types of threat.

Parental care has evolved across the animal kingdom to increase the probability of offspring surviving in an environment fraught with danger. While parental care is common among mammals and birds, it is relatively rare in fish with the vast majority of fish showing no form of parental care at all, whilst those that do, often just provide parental care to developing eggs pre-hatch. The provision of parental care in discus fish (Symphysodon spp.) is, therefore, interesting in that parents provide mucus to offspring as a source of nutrition during the first few weeks of care. In mammals this post-birth provision of parental care can lead to the development of the parent offspring conflict. It is, however, possible that this conflict is also present in discus fish. This thesis examines both the interesting parental care strategy of discus fish along with the potential for the parent offspring conflict to develop. To examine the dynamics of parental care in discus fish, a range of behavioural and mucus composition studies were carried out. The analysis of mucus revealed that similar to mammals, parents provided offspring with an initial high quantity of nutritional and non-nutritional factors including antibodies (IgM), essential ions and hormones. Behavioural studies also revealed that initially parents were highly diligent in providing care to offspring but that after two weeks of care, the behaviour of parents changed making it harder for offspring to obtain mucus. At this point a weaning period was initiated where offspring began spending less time with parents and more time foraging for external food sources. The initiation of this weaning period suggests the presence of the parent offspring conflict and indicates that a point is reached where the energetic demands of offspring are too great and that energy is better invested in to future offspring. Research into the bite size and feeding rate of fry suggest that during the weaning period fry could demand excessive amounts of mucus, which may be energetically unsustainable leading to the observed offspring avoiding behaviour of parents. As parental care behaviour is known to be intimately associated with mate choice, mate choice behaviour was also assessed in discus fish with the hypothesis that the ability to provide mucus would be selected for by prospective mates. While my dietary experiment did not influence mucus quality, the mate choice experiment did reveal the importance of hierarchies in discus fish, indicating that dominant individuals were significantly more likely to pair than subordinates. This is similar to that observed in closely related cichlids where the ability to be dominant and protect a territory was indicative of the ability to successfully raise offspring. In conclusion, the parental care behaviour of discus fish appears to share more similarities with that seen in mammals than that observed in fish. The implications of these findings indicate that parental care in discus fish could be a new model of parent offspring conflict hitherto unseen in fish which could ultimately help our understanding of the evolution of parental care in fish.