Randomness and aggregation: analysis of dispersion in an epiphytic chironomid community

1. Patterns of dispersion in a chironomid community on the submersed macrophyte Myriophyllum spicatum were analysed. 2. Random dispersion commonly occurred throughout the year, with an average of 40% of all species being observed with random spatial patterns. The frequency of occasions with random dispersion varied among chironomid species, ranging from 3.3% in Rheotanytarsus curtistylus to 56% in Thienemanniella majuscula. 3. Estimates of the negative binomial parameter k show that 26% of all cases demonstrate strong aggregation (0 < k < 1.0) while nearly half (47%) have quasi-random dispersion. Interspecific variation in k was not significant statistically when all the species were considered together, although a pairwise comparison between two abundant species Tvetenia calvescens and Rheotanytarsus curtistylus demonstrated a marginally significant difference. When different instars were compared, the percentage frequency of strong aggregation (0 < k < 1.0) declined from first instars (49%) to later instars (II-38%, III-24% and IV-27%). 4. Variance/mean and m*—m regressions (m* is Lloyd's mean crowding statistic and m is the sample mean) both fitted the data well, but there was little indication of significant interspecific variation in parameter values, particularly the slope of regression. 5. Dispersion patterns were examined along with the analysis of spatial overlap in this community. Forty-two per cent of species-pairs with reduced spatial overlap (spatially ‘segregated’ pairs) contained one or both species with random dispersion, while the corresponding value for spatially unsegregated pairs was 57%. This suggests that spatial segregation is not necessarily caused by strong, independent aggregation of both species. Comparing spatially segregated vs. unsegregated pairs, the former tend to have one species with a stronger tendency of aggregation than species of the latter. 6. Patterns of dispersion observed were considered in the light of ‘random patch formation’. Random patch formation emphasizes the stochasticity of patch-forming processes as well as the stochastically dynamic nature of resultant patches. Unlike terrestrial drosophilid assemblages, where strong aggregation is a predominant pattern, this chironomid community demonstrates widely varying degrees of dispersion with high occurrence of randomness, reflecting the stochasticity of dispersal and recolonization processes. It is suggested that, in terms of species coexistence, more emphasis should be placed on stochasticity rather than on aggregation in this type of community.