A dynamic simulation model of land cover in the Dulce Creek Basin, Argentina

The Dulce Creek Basin (1000 km2) is situated in the Buenos Aires's Sate and flows into the Mar Chiquita lagoon, (MAB Biosphere Reserve – UNESCO). While land cover on the upper basin is dominated by field crops, the low basin is mostly occupied by livestock grazing systems. The increased use of inputs (agricultural intensification), and a high degree of riverside vegetation's degradation in the grazing zone, represent an increasing pollution risk for surface water bodies and groundwater. The aim of this work was to find suitable driving factors for a dynamic land cover simulation model (Dyna-CLUE) since1997 to 2005 using logit regression models. Land cover change (Crop, Pasture, Grassland, and Others) was simulated using eleven driving factors accounting for geographic, climatic and soil and infrastructure features. We assessed multicolinearity from the Pearson correlation matrix, and only uncorrelated variables were applied in the logistic regression. Waterlogging, altitude, geomorphology and soil productivity were the main driving factors for the predictions of land cover dynamics, as selected from logistic models. The land-cover simulation model was run according to the actual trends of agricultural expansion. The accuracy of predictive models was tested using receiver operating characteristic (ROC) curves and the good fit showed by the Dyna-CLUE model to data, support its suitability for simulation of land cover dynamics of the Dulce Creek Basin. The expected trajectory of land cover according to current trends warns about a progressive decrease of grasslands cover, and the associated loss of their ecosystem services. Further simulations upon alternative scenarios can help land use planning for production and conservation objectives.