Biological water quality in tropical rivers during dry and rainy seasons: a model-based analysis

Abstract

Recent studies on water quality in tropical rivers indicate substantial differences between seasons. However, investigations on the needs and added value of season-specific models are lacking. Thus, this paper aims to determine the accuracy and relevance of season-specific and season-overarching models to predict biological water quality. Additionally, we investigated the variation of prediction accuracy using sub-datasets from different parts of the Cuenca River basin. This study was accomplished in the rivers that pass through the urban and suburban areas of the city of Cuenca, which is located in the southern Andes of Ecuador. The Andean Biotic Index (ABI) was used as an indicator of biological water quality. Subsequently, models were developed to predict the ABI, with physicochemical and morphological variables as predictors, which were collected in 43 sites during both the dry and the rainy seasons. The predictions were obtained using three kinds of generalized linear models (GLMs): Gaussian, Gamma and Inverse Gaussian. The season-specific models were more accurate than the season-overarching models. Similarly, the predictions of the biological water quality in sites sampled in the urban area were more accurate than the forecasts performed in reference sites. The major variables predicting the ABI during the dry season were five-day biological oxygen demand (BOD5), ammonium and orthophosphate, while dissolved oxygen (DO), oxygen saturation (OS), nitrate, total solids proved to be important during the rainy season. The results of this research emphasize the importance of developing season-specific models and the implementation of different key actions for river restoration during both the dry and rainy seasons. The accuracy and the replication of these models could be improved and checked with more data taken from new sampling events. The modelling approach developed in this study can be applied to similar basins in the tropics and reveals that environmental investments need to count on monitoring strategies and data and analyses of the biological water quality variation in dry and rainy seasons, within the context of sustainable development.