Host-parasite dynamics involves coevolutionary arms races, commonly leading to host specialization. General understanding of evolutionary trajectories of specialization in brood parasites is compromised by restricted focus on bird and insect lineages. We studied host utilization and host specificity in a natural population of the cuckoo catfish (Synodontis multipunctatus) which is an obligate parasite of parental care of mouthbrooding cichlids in Lake Tanganyika. On a sample of 779 host broods from 20 cichlid species, we detected four host species (with prevalence of parasitism of 2-18%). Phylogenetic analysis based on genomic (ddRAD sequencing) and mitochondrial (Dloop) data from cuckoo catfish embryos showed an absence of host-specific lineages, despite former indications of two morphological forms of the cuckoo catfish. This was corroborated by analyses of genetic structure and co-ancestry matrix. All host species were from the tribe Tropheini, maternal mouthbrooders that spawn over a substrate (rather than in open water). Parasitized host individuals carried smaller clutches (as cuckoo catfish prey on cichlid eggs), but did not differ in their body size or habitat use from non-parasitized conspecifics. We conclude that the cuckoo catfish is an intermediate generalist, selecting a subset of available cichlid species as hosts but not forming host-specific lineages. Brood parasitism in the cuckoo catfish arose in a lineage which lacks any parental care and we discuss costs and benefits of host specialization in this species and brood parasites in general.

Sampling was conducted by SCUBA diving and snorkelling. During each dive, we set a benthic screen net (20 m long and 1.5 m high, mesh size 10 mm) along the bottom at a particular depth and searched for mouthbrooding cichlid fish in close vicinity. Mouthbrooding fish are easily recognized by their extended buccal cavity. Immediately after each dive, the collected females were identified to species level, their body length was measured to the nearest mm, and the broods were collected by gently washing them out of the maternal buccal cavity using the water jet of a Pasteur pipette. The offspring of each brood was photographed, counted, and the age was estimated based on offspring developmental stage. Parasitic eggs and embryos were clearly visible during brood inspections (smaller size, characteristic colour and shape). When parasitic offspring were detected, they were counted, measured to the nearest 1 mm, euthanised and stored in 96% ethanol.

For each brood, we recorded host species, host size, depth of host collection, host brood size, developmental stage of the brood, occurrence of parasitism and, when the brood was parasitized, the number and size of parasitic offspring. Host brood developmental stage was estimated visually based on embryo and yolk sac size. We considered this estimate reliable until the age of 3 weeks. To mitigate uncertainty around our brood age estimates, we then converted brood age into a new variable, termed brood stage. Brood stage comprised five categories; eggs (age 1-5 days), early embryo (6-10 days old), mid embryo (11-15 days old), late embryo (16-20 days old), and juvenile (21 days and older).