This paper is written to compare the results of theoretical investigations of sympatric speciation with the relevant experimental data. We understand sympatric speciation as a formation of species out of a population whose spatial structure is not important genetically. A necessary prerequisite for speriation is an action of disruptive selection on sufficiently polymorphic traits. The present analysis confirms the view that such a selection is ecologically realistic. The genetical part of speciation begins with a development of reproductive isolation between those individuals that are opposed in strme characters. It is shown that selection for reproductive isolation may be quite strong. Extinction of intermediate individuals, which completes speciation, proceeds under a wide range of ronditions, including those when the newly formed species differ in quantitative characters, though most of the genes arc likely to remain the same in both species. The whole process seems possible if differences i n srveral (up to 10) loci are sufficient to adapt the forming species to different niches and to establish reproductive isolation. It is shown that populations with bimodal distributions of some genetically determined quantitative characters can have a considerable life-time. Such distributions may he formed either as a transition stage of sympatric speciation or represent a stationary state under conditions close to those necessary to complete speciation. They are very important Tor experimental investigations. Sympatric speciation always follows the same principal course; it does not contradict the idea of a genome coadaptedness. The occurrence of sympatric speciation is different for different taxa depending rather on how frequently populations are subjected to the appropriate kind of selection than on their ability to obey it.

Among the Lake Tana large barbs of about 10 cm SL only representatives of the ' acute ' morphotype can be distinguished, in the size range 10-20 cm SL ' bigmouth big eye ' can be identified also. As for the rest, very few individuals can be confidently affiliated with a particular morphotype most of them looking like ' intermedius '. Even within the range of 20-30 cm SL some individuals are still difficult to identify. Principal component analysis of cranial characters revealed discrete groups of morphotypes. Differences in both external and cranial characters of the morphotypes result from divergence which is most pronounced when fish are 4-5 years old and 20-25 cm SL. This divergence cannot be related exclusively with differences in the growth rates of individuals representing different morphotypes. Differences in food composition between the morphotypes probably increase in parallel with their morphological divergence. Differences between the morphotypes in the lateral line (ll) and the gill rakers (Sb) counts were revealed using ANOVA. Comparison of the Lake Tana Barbus complex of forms with that previously known from Lake Lanao (Philippines) suggests that in both lakes the different forms arose sympatrically but that sympatric speciation in Lake Lanao has advanced further than in Lake Tana. 1996 The Fisheries Society of the British Isles

No abstract

During the rainy season (September-November 1994), in tributaries of Lake Tana (Ethiopia), the Gumara, Gelda and Negashu, barbs of different morphotypes were ready to spawn, running and spent. The most numerous were barbs similar to Barbus intermedius common in many rivers of the Ethiopian Highlands. They were also found spawning in the Gumara and Gelda Rivers in the dry season (February 1996(February , 1997. There were no evident differences in time and place of spawning of barbs representing different morphotypes. Only zurki, bigmouth mini-eye and bighead were not found in the rivers. Spatial and temporal segregation may contribute to reproductive isolation of morphotypes spawning in rivers, but if reproductive isolation exists the main role must be played by differences in mating behaviour. Morphotypes of the Lake Tana barbs have different migration patterns. Barbs similar to the riverine B. intermedius may spend their lives in rivers or in the lake, moving to and fro. Barbs of acute, bigmouth big-eye and, perhaps, other morphotypes migrate downstream as alevins or fry, stay in the lake for several years and migrate back into rivers for spawning after reaching maturity. 1999 The Fisheries Society of the British Isles

In the Lake Tana barbs the PCA of skull measurements reveals "ontogenetic channels" within which individual ontogenetic trajectories are located. Some individuals seem to acquire states of characters specific for a certain morphotype quite late when they are 20-25 cm SL and 3-5 years old. Morphologically different sympatric forms of large barbs are also found in the Didessa (Blue Nile basin), Genale (Wabi Shebeli basin), and Gibe (Omo basin) rivers. In all the three rivers barbs resembling certain morphotypes of the Lake Tana barbs occur. Most probably, allopatric bigmouthed forms originated independently from sympatric forms which have no conspicuous specific traits and that their similarity results from homoplasies.