The large body of data and knowledge generated in a few experimental highly monitored sites in the South American páramos provide a starting point to generalize findings to a larger area in which water resources are used and managed. To prove the usefulness of pooling data from monitored sites to make predictions in ungauged basins, we used data generated from the iMHEA network of paired, collocated catchments with contrasting land-use types, and tested how such a monitoring design can improve the detectability of land-use change signals and thus the prediction of land-use impacts in ungauged basins. The regionalization exercise related a set of hydrological indices to physical and climatic descriptors using multilinear regressions (Figure 5). The collocation of catchments is intended to increase the contrast of land-use and land-cover variables between pairs and, at the same time, minimize climatic and physical differences. This was tested by treating each catchment as ungauged while keeping its pair with similar descriptors but contrasting land use in the derivation of regional models. The regression results showed that regionalization using paired catchments enhances the detectability of land-use change impacts improving model performance and predictive capacity for 66% of the 50 indices tested (Ochoa-Tocachi et al., 2016b). For the runoff ratio, baseflow index, and slope of the flow duration curve, the mean absolute error reduced on average by 53% and the variance of the residuals by 79%. In contrast to previous research that found it difficult to isolate land-use signals in regionalization (e.g., Visessri & McIntyre, 2016) the collocation of catchments increased the contrast of land-use and land-cover variables between pairs, at the same time it minimized climatic and physical differences. In consequence, the robust regionalization results are attributed to the paired catchment network setup that covers diverse physiographic characteristics, contrasting land-use types, and degrees of conservation/alteration (Ochoa-Tocachi et al., 2016b). This demonstrates that such a design of monitoring network is a useful strategy to optimize data collection, provide commonly available geographical information, understand the major controls of hydrological response, and provide robust predictions in ungauged basins in data-scarce regions such as the tropical Andes, with potential application elsewhere. Figure 5