In present paper, an outline is given of results of investigation concerning regionaland seasonal changes of some criterin of bottom fish communities in the continenta shelf of northern Nan--Hai. Fish collection was made by three research vessels duringMarch 1964 through Febraury 1965. Fifty-one sampling sites were assigned within 8arbitrarysub--areas of the shelf, which cover an area of 137,667 Km². Towing time oftrawlnet around each site was 2 hours, Every sample of fish was considered as a fishcommunity, of which only the data of March, June, September and December weretreated as representatives of spring, summer, fall and winter respectively. Species diversity is the principal criterion of fish community and there are severalmodels for computing the index of diversity. The modified Shannon--Weaver formulawas employed in the present paper. For eliminating the bias arosing from the discre-pancy of weight of individuals in samples, we used biomass units instead of numbers.Thus, the actual formula we used was H" = -∑ from (w_t/w )log₂(w_t/w)where w_t is the biomass of i-th species in sample, w the biomass of sample. The resultsof implementation of biomass units suggest that this modified computing method mightbe convenient for practical purpose, but the values obtained in terms of biomass tendedto be slightly higher then those in terms of numbers. In the range of the indexes of diversity overall sites within the studied periord,the lowest was 1. 41 occurred at a site of V sub-area in September and the highest was4. 40 at a site of IV sub-area in June. In comparing with the published values of H" ofcoastal bottom-trawled fishes, our figures may be ranged as the highest one. An isopleth diagram of H" was drawn, it was found that the distribution patternof isopleth families in different seasons was distinct each other, but no explicit relation-ship could be ruled out, as the data were still insufficient. The relations of H" withshelf-depth and temperature were also investigated and it seems some possible correla-tions could be found. In conjunction with species diversity, 3 other parameters were calculated:number of species contained in community (S); (2) maximum of species diversity(H_(max)); (3) evenness of Species (J). The Pattern of variations was apparent in theindexes calculated. The variation of H_(max) followed the pattern of S, while the evennessof species (J) shows close association with the trend of species diversity (H"). The component of dominant species is a second fundamental critirion of fishcommuncties and we proposed to formulate a concept of dominant species in con-nection with imformation theory, its interpretation being given in detail. Thus, anyspecies of fish comprised 20--60% of catches in one specified sample is defined as adominant species (D. S.); 24 D. S. were observed in 8 sub-areas and 4 months, andtheir frequencies of occurrence in each sub--area and month showed a wide range. Acode diagram is drawn to demonstrate the regional and seasonal distribution of frequen- cy of D. S., so that the transmission and replacemeat of D. S. between areas or seasonscan be found intuitively. But the situation is very complex, therefore the possiblerelationship between species diversity and dominant species still remains unknown, andthe problem of dynamic of dominant species population itself needs further consecutivestudy and data accumulation.