The distribution and abundance of aquatic insects were studied in the longitudinal gradient of the watershed of Macaé River, a coastal Atlantic Forest river in South-eastern Brazil. Sampling stations were selected in the first, second, fourth, fifth, and sixth orders and sampled in April, July, and October 1995. This represented the end of the rainy season, the dry season, and the beginning of another rainy season, respectively. In each month four samples were collected using a Surber sampler from each of the following substrates: sand, litter deposited in pool areas, litter in riffle areas, and stones. A total of 46,431 specimens of aquatic insects belonging to ten orders were obtained. The data were analyzed by the multivariate methodologies of Correspondence Analysis (CA) and Cluster Analysis (UPGMA) using the similarity index of Morisita, for all three months. Both showed a significant faunal disrupture in the river, which can be divided in two sections: the upper one, from first to fourth orders, and the lower section, including fifth and sixth orders. The same results were obtained with presence-absence matrices, using Jaccard similarity index, showing that the changes are not only due to quantitative differences. A Mantel test was used to compare the assemblage composition temporally and no difference was detected between the three months. Moreover, a Canonical Correspondence Analysis (CCA) was applied to the data to check which of the 14 physical and chemical variables significantly explained macroinvertebrate community variation. The most significant variables were conductivity, CPOM, and pH for the upper stations (1st, 2nd and 4th orders), and alkalinity, FPOM, and HCO3 for the lower stations (5th and 6th orders).
aquatic insect; longitudinal gradient; tropical river; multivariate analysis
