Scleractinian corals are traditionally described mainly on the basis of colony morphology, but remarkable variation in colony form, skeleton micro-morphology, reproductive biology, symbiosis and competitive behavior exists within and between species of a single genus. As a consequence of this variability in form as well as function, species boundaries remain equivocal in many genera. With the recent addition of molecular genetic data, it has become clear that coral species boundaries are not straightforward.

The main objective of this thesis was to examine species boundaries between six morphospecies in the coral genus Madracis that are common to in the Caribbean i.e. M. mirabilis, M. decactis, M. pharensis, M. senaria, M. formosa and M. carmabi. These morphospecies occur in sympatry, but with overlapping distributions, across all depths at a single reef site. Madracis morphospecies are hermaphrodites and display a brooding mode of reproduction. Because of the ecological importance of Madracis on Caribbean reefs it is essential to reliably identify its species in order to understand their evolution and their ecological interactions—with each other, with their symbionts and with other reef residents. In this thesis, morphological and genetic data were utilized to investigate taxonomic and evolutionary relationships among the six commonly recognized morphospecies, as well as their dinoflagellate endosymbiotic. The main conclusions are as follows.

A molecular phylogenetic analysis based on ribosomal DNA ITS1-5.8S-ITS2 sequence data reveal a mixed pattern of clearly resolved monophyletic clades suggesting valid species (M. mirabilis and M. senaria) as well as an unresolved paraphyletic complex with high intra- and inter specific ITS polymorphism. The genus Madracis appears to be in the process of phylogenetic fission under a reticulate model of speciation following the surface circulation vicariance model, which has not been observed in brooding corals before.

Univariate analysis of micro-morphological linear measurements of six continuous characters and two meristic characters of the corallite skeleton revealed significant differences between morphospecies. However, no single linear measurement alone can function as a diagnostic character to identify the six morphospecies. Multivariate canonical discriminant analysis could identify six morphospecies clusters with little overlap between clusters. It was not possible to classify every individual to the correct morphospecies cluster using the discriminant functions. However, our results indicate that the use of overall colony morphology and number of primary septa appear to be reliable characters in the identification of Madracis morphospecies in the field.

Symbiont diversity in the genus Madracis as assessed by ribosomal DNA ITS1-5.8S-ITS2 sequence analysis reveal three clade B Symbiodinium ITS types that are specific for this coral genus. One Symbiodinium type was specifically found in M. mirabilis. This evidence suggests host-specificity in M. mirabilis and is one of the few indications of co-evolution between a coral and its alga. This finding bolsters previous results suggesting reproductive isolation of M. mirabilis.

Morphospecies boundaries in the genus Madracis do not correspond to genetically distinct evolutionary units. Introgressive hybridization as a result of weak or absent reproductive barriers and the decoupling of morphology and genotype are hypothesized to explain the complex relationships in the genus Madracis.