| Current models of stream drying require extensive data and drying predictions
could be improved if we understood (1) the spatial scales of autocorrelation in stream
drying, (2) its drivers, and (3) the synchrony of wetting and drying. We measured
relative electrical conductivity, a proxy for the absence or presence of water, at 92
(2020) and 121 (2021) locations across the ~16.8-km2
Gibson Jack watershed (Idaho,
USA). We then calculated seasonal streamflow permanence at each location,
developed a predictive kriging model, and calculated the number of wet sites each day.
We found that (1) seasonal streamflow permanence is autocorrelated on scales of ~400
m, (2) topographic, lithologic, and pedologic variables were the top three drivers of
stream drying, and (3) drying largely occurred asynchronously whereas rewetting
occurred synchronously. These results suggest that the hierarchy of drivers of stream
drying may be dynamic and scale-dependent in both space and time. |