Reduced precipitation treatments often are used in field experiments to explore the effects of drought on plant productivity and species composition. However, in seasonally snow-covered regions reduced precipitation also reduces snow cover, which can increase soil frost depth, decrease minimum soil temperatures and increase soil freeze-thaw cycles. Therefore, in addition to the effects of reduced precipitation on plants via drought, freezing damage to overwintering plants at or below the soil surface could further affect plant productivity and relative species abundances during the growing season. We examined the effects of both reduced rainfall (via rain-out shelters) and decreased snow cover (via snow removal) at 12 sites globally as an add-on to the International Drought Experiment, a coordinated distributed experiment. Snow was removed opportunistically from the beginning until the middle of winter to increase soil freezing, and snow removal stopped before the end of winter to minimize its effects on spring melt water. In the subsequent spring, half of the snow removal and ambient snow plots were covered with rain-out shelters designed to reduce rain throughput at each location by an amount equivalent to a one-in-one hundred year drought. Plant cover was estimated at the species level and aboveground biomass was harvested.
The snow removal effect on minimum soil temperature varied widely among sites; for example, for the sites that experienced very little accumulation, predictably, there was no significant effect of snow removal on soil temperature. However, across all sites, there was a significant correlation between the snow removal effect on minimum soil temperature and the effect on subsequent plant productivity. The rain-out shelters likewise produced widely variable soil moisture effects among sites (e.g. similar to snow removal, there was no rain-out shelter effect in the absence of rain, and the rain-out shelters did not produce stressful drought conditions for sites that experienced wet years). Only 3 of the sites exhibited significant rain-out shelter effects on plant productivity, and across sites, there was no significant correlation between the rain-out shelter effect on soil moisture and the effect on plant productivity. There was no significant interaction between snow removal and rain-out shelters for plant productivity, but the test of the interaction was inadequate, because the two factors rarely exhibited significant effects simultaneously for any given site. Overall, our results reveal that reduced snowfall, when it decreases minimum soil temperatures, can explain a substantial component of the total effect of reduced precipitation on plant productivity.