Selection of reference lakes and adaptation of a fish multimetric index of biotic integrity to six amazon floodplain lakes

This study seeks to define a bioassessment tool for the floodplain lakes of the lower Solimões River (Amazonas, Brazil). We defined lakes in pristine condition as reference lakes and adapted a fish multimetric index of biotic integrity. The floodplain lakes analyzed were Baixio, Preto, Ananá, Araçá, Maracá and Poraqué, located 30-400 km from Manaus, the state capital. To select the reference lakes we adapted a rapid bioassessment protocol that analyzes satellite imagines from Google Earth. Landscape characteristics were investigated in two different screening phases: preliminary (presence/absence) and final (semi-quantitative). Each screening phase used two landscape scales: buffer-zone (a 6 km-diameter circle around each lake) and local (an area extending 500 m from the lake's shoreline). The landscape attributes selected in these screening phases allowed us to define an Environmental Gradient Index (EGI) that represented the effect of the human presence on the lakes. Lakes Araçá and Ananá, which had no or very few signs of anthropogenic disturbance, were classified in the high EGI category and selected as reference lakes. A floodplain-lake index of biotic integrity (FL-IBI) was developed from twenty candidate fish assemblage metrics, of which four were selected after range, sensitivity, responsiveness and redundancy tests for inclusion in the final index (total number of species, total number of individuals, total number of individuals with moderate-high vulnerability and percentage of carnivores). Metrics were scored continuously from 0 to 10. Final FL-IBI scores were calculated by adding the scores for each selected metric and dividing the result by the number of metrics. To facilitate comparison with other indices, this was weighted to range from 0 to 100. The FL-IBI proved to be able to distinguish well between reference and non-reference lakes. Index scores had wide seasonal and temporal variability, largely because of the major changes in habitat caused by the intensity, duration, frequency and occurrence of the Solimões flooding cycle. For the reference lakes, the lowest seasonal index variability (CV < 5%) was observed in the receding-water period, allowing us to consider this as the best index period for biomonitoring purposes. The greatest annual variability for the study period (2004-2007) was observed in 2005 (CV > 20%) and was associated with the severe drought occurred in that year. These results show that the fish assemblages in the lakes studied are highly resilient and perfectly adapted to the hydrological cycle of the Solimões River.