The barbs of Lake Tana, Ethiopia : morphological diversity and its implications for taxonomy, trophic resource partitioning, and fisheries

The rediscovery of a unique species flock of cyprinid fish, its taxonomy and its feeding-biology are described. Fourteen species of barbs <em>(Barbus</em> spp, <em></em> Cyprinidae, Teleostei) were found in highland (1800 m) Lake Tana, in northwestern Ethiopia. Lake Tana is an isolated fresh-water system, because its only outflowing river, the Blue Nile, of which it is the source, drops over 40 m high waterfalls only 35 km from the lake. The carplike barbs in the lake are unique and show a wide variety of discrete, and consistent types, widely differing in 1) size and shape, 2) feeding behaviour, and 3) spatial distribution. Together with 4) the early morphological divergence of juvenile barbs, 5) the reproductive segregation of the adults, and 6) part of their genetic characters, this diversity was decisive in our description of fourteen endemic species, seven of them new to science. Lake Tana was formed by volcanic blocking of the Blue Nile and subsequent flooding of the Lake Tana basin. The <em>Barbus</em> species probably evolved within the lake itself from one common ancestor, very similar to the present-day, riverine <em>B. intermedius,</em> making it a unique species flock. The driving force of their evolution most likely involved radiation into the new lacustrine trophic niches that became available when Lake Tana filled up. The presence of this group of very closely related fishes provides a unique opportunity to study the evolution of adaptation, and the relation of morphology and ecology: it forms a 'natural laboratory'. By using an ecomorphological approach, based on properties of fish foods, and functional morphological studies of the relations between fish parameters and their significance in dealing with food properties, predictions on the diets and food partitioning of the barbs have been made. These predictions were based on a large set of characters (33), which made the method robust and allowed for accurate resolving power. Testing of the predictions with field data revealed that individual diets, but especially food partitioning can be well predicted. This new method provides insights into the dynamic trophic interactions among species without the need for an extensive ecological sampling programme, and could be instrumental in predicting shifts in fish fauna composition, due to environmental impact, such as overfishing or the introduction of new species. The understanding and prediction of such shifts will help in developing a strategy towards sustainable fisheries and the protection of biodiversity.