Diversidade de ácaros edáficos (acari: oribatida) e redução do esforço amostral no gradiente ecológico de 10 mil ha de floresta de terra-firme da Reserva Ducke, Manaus, AM, Brasil

Abstract

Sampling oribatid mites in large areas using conventional methods is expensive, timeconsuming, and this constrains their use on monitoring program. In 2001/2002, a system of 9 N-S and 9 E-W trails was installed in the reserve. Following the RAPELD method, plots were distributed over a 64 km2 grid of trails spaced 1 km apart, resulting in 72 plots with a minimum distance of 1 km between them. Inside this large grid, the Program for Planned Biodiversity Research (PPBio) delimited a smaller grid of 25 km2 with 30 plots, whose localization was defined by logistic and financial considerations. This size grid is the standard for most sites using large RAPELD grids. To estimate the taxonomical diversity of oribatid mites in a gradient of soil and vegetation in the reserve, one of the most intensively studied sites in Amazon, in 2001-2002, 360 sub-samples of litter and soil were obtained in 72 plots of 250 m of length, and mites were extracted using a Berlese-Tullgren apparatus. To the effort reduction in the field, we evaluated the reduction of the number of plots to 62, 52, 42, 32 e 30 in a random way and also taking groups of neighbor plots. To the effort reduction in the field and in the laboratory, the same procedure was done excluding the rare species from the data. Wet upland forests in central Amazonia are not homogeneous, but have a gradient of soil, topographys, water retention, vegetation structure, and light penetration to the understory. Because of it, we also evaluated if the localization of six alternative 25 km2 grids inside the 100 km2 of the reserve would affect conclusions about the abundance and diversity of the mites. A total of 161 taxa (adults) was registered, being 82 (51%) identified to known species. The survey increased from 54 to 106 the number of taxa not previously registered in the reserve. For the State of Amazonas and for the reserve, 24 and 105 were new registers respectively, increasing to 254 the number registered in the area. Galumna, Rostrozetes and Scheloribates were the most speciose. Fifty seven taxa (35%) were singletons or doubletons. Rostrozetes foveolatus, Pergalumna passimpuctata and Archegozetes longisetosus were the most abundant, being the first two also the most frequents. Although the Amazon region is one of the most investigated in Brazil, there is a relative deficiency in the taxonomic knowledge of oribatid mites, compounded by lack of descriptions of the same species of the most common. The dissimilarity matrices were generated for each 30 plots grid based on the Bray Curtis (abundance data) and Sørensen (presence absence data) indices. Mantel correlations were used for assessing relationships between pairs of matrices, assuming that values ≥ 0.7 indicate a satisfactory similarity of information from pairs of grids. The ecological pattern of the mites community could be detected in different levels of effort reduction. Using multiple multivariate regressions, the effect of the clay percentage in the composition of mites was detected using data based in every level of reduction. The number of palm trees was not correlated to the mite community. Using abundance data of the mite community, correlations values varied from 0.41 to 0.63, indicating differences in information depending on the position of the grid. Presence/absence data varied from 0.79 to 0.88 showing that species compositions were similar between grids. The localization of the 25 km2 grids greatly affects information on the abundance but not the composition of species between grids. The most frequent species occurred in the majority of habitats in Reserva Ducke. If the standard PPBio grid was placed in another location of the reserve, only the lesser abundant species of oribatid mites would not have been sampled. These smaller grids carried ~77% of the species captured in the 64 km2 grid and the utilization of data originating from these grids still capture the main ecological pattern of the community. Estimating sampling and processing time for the 30 plots grids material, the costs of the project are reduced to ~58%. For rapid surveys to detect ecological pattern, it is more efficient to use presence and absence data of the oribatid mites community and exclude rare species, reducing time in the identification of the material and providing faster access to the data and faster publication of the results.