Testing the 'two water worlds' hypothesis under variable preferential flow conditions

Widespread observations of ecohydrological separation are interpreted by suggesting that water flowing through highly conductive soil pores resists mixing with matrix storage over periods of days to months (i.e., two 'water worlds' exist). These interpretations imply that heterogeneous flow can produce ecohydrological separation in soils, yet little mechanistic evidence exists to explain this phenomenon. We quantified the separation between mobile water moving through preferential flow paths versus less mobile water remaining in the soil matrix after free-drainage to identify the amount of preferential flow necessary to maintain a two water world's scenario. Soil columns of varying macropore structure were subjected to simulated rainfall of increasing rainfall intensity (26 mm h(-1), 60 mm h(-1), and 110 mm h(-1)) whose stable isotope signatures oscillated around known baseline values. Prior to rainfall, soil matrix water delta H-2 nearly matched the known value used to initially wet the pore space whereas soil delta O-18 deviated from this value by up to 3.4 parts per thousand, suggesting that soils may strongly fractionate O-18. All treatments had up to 100% mixing between rain and matrix water under the lowest (26 mm h(-1)) and medium (60 mm h(-1)) rainfall intensities. The highest rainfall intensity (110 mm h(-1)), however, reduced mixing of rain and matrix water for all treatments and produced significantly different preferential flow estimates between columns with intact soil structure compared to columns with reduced soil structure. Further, artificially limiting exchange between preferential flow paths and matrix water reduced bypass flow under the most intense rainfall. We show that (1) precipitation offset metrics such as lc-excess and d-excess may yield questionable interpretations when used to identify ecohydrological separation, (2) distinct domain separation may require extreme rainfall intensities and (3) domain exchange is an important component of macropore flow.