Rainfall, native vegetation and agricultural land uses influencing water quality in subtropical agricultural catchments

Link To Share This Poster: https://cdmcd.co/q9YLGZ

Eduarda Romanini, Ecology, Bioscience Institute at the University of São Paulo, São Paulo, Brazil, Ricardo Hideo Taniwaki, Center for Engineering, Modeling and Applied Social Sciences, Federal University of ABC, Santo André, Brazil, Carla Cristina Cassiano, Faculty of Forestry Engineering of the Federal University of Mato Grosso, Cuiabá, Brazil, Silvio F. B. Ferraz, Forestry Sciences, University of São Paulo, Piracicaba, Brazil and Leandro R. Tambosi, Ecolgy, Bioscience institute of the University of São Paulo, São Paulo, Brazil

Background/Question/Methods
In natural conditions, soil properties, topography, atmospheric depositions, geology, climate, and catchment hydrology can affect water quality. Together with these natural influences, water quality varies with the land conversion of natural vegetation to anthropic areas. This conversion, coupled with the amount of rainfall, may result in in-stream water quality degradation.
In this study, we aimed to understand the effect of the amount of natural and agricultural vegetation (represented by the percentage of native vegetation and values of enhanced vegetation index), together with the quantity of rainfall (accumulated rainfall and number of dry days before data sampling) on water quality parameters: dissolved oxygen (DO), electrical conductivity (EC), suspender total solid (STS), total dissolved nitrogen (TDN) and total phosphorus (TP). We represented the water quality parameters by their measured values (represented with the parameters acronyms) and their values centred in relation to the average values of each catchment (here and after symbolized with "dev" after parameters acronyms).
The study area is located at the Corumbataí river basin, in the east-central portion of the state of São Paulo, Brazil. Data analysis were performed using linear mixed models and the Akaike Information Criterion (AIC) was used to select the best models. After model selection, we deeply analysed only the simplest model, according to the parsimony principle.

Results/Conclusions

The null model was selected when analysing STS. Models for EC and TDNdev were poorly adjusted. Variables related to the quantity of rainfall were selected to ECdev, TDN, STSdev, TP and DO models. The variables that represent the quantity of vegetation (natural and agricultural) influence DOdev and TP. Furthermore, interactions between the quantity of vegetation and rainfall variables affected TPdev and TDN.
Therefore, we demonstrated that rainfall is important for almost every water quality parameter studied. Thus, proper landscape management (e.g. building sediment traps, preserving riparian zones, and avoiding large areas with bare soil) is necessary to minimize possible negative effects of rainfall and vegetation cover, especially agricultural land uses, in water quality, once these effects may not be completely compensated by small amounts of native vegetation. Moreover, our results suggest that more frequent and intense storm and longer drought events (as expected in global climate change scenarios for tropical regions) (IPCC, 2013) may greatly affect water quality. Our findings can be useful to better understand future changes in-stream water quality and guide decision-makers to guarantee the provision of hydrological ecosystem services in scenarios of global climate changes.