Abstract
Abstract
Attributing hydrological change to mining is difficult when only model outputs and a basin-internal reference are available. We tested whether annual baseline outputs from the Soil and Water Assessment Tool Plus (SWAT+) for Mandena, southeastern Madagascar, showed a post-2008 shift relative to an internal reference. A retrospective before–after–control–impact (BACI)-style comparison used area-weighted values from 216 mining-zone and 308 reference-zone hydrological response unit intersections over 1993–2022. For nine water-balance outputs, annual mining-minus-reference differences were regressed on a post-2008 indicator using Newey–West standard errors and Holm correction. No threshold-associated shift was detected (all adjusted p = 1.000). Percolation had the largest negative estimate (− 5.46 mm yr⁻¹), but increased in both zones and varied widely among years. This result does not demonstrate the absence of mining effects or insufficient alteration of soil properties. In a separate sensitivity experiment, low-, intermediate- and high-magnitude parameter changes were imposed on vegetation, surface and soil terms in 158 selected mining-zone units. These changes redistributed simulated evapotranspiration, runoff, water yield and percolation, but do not establish that equivalent changes occurred historically. The study distinguishes temporal detection from model sensitivity and supports environmental monitoring that combines outputs at finer temporal resolution with discharge, piezometry and separate wetland observations.