Abstract
The impact of negative liquid pressure on the freezing of xylem sap is a unique scientific problem. Equilibrium thermodynamics would suggest that, due to the density anomaly of ice, sap under tension should freeze at higher temperatures than under positive pressure. Yet, the opposite is observed in tree branches: freezing temperatures decrease as the pressure becomes more negative. Using a cold stage array, we measured the freezing points of sap samples extracted from Pinus sylvestris branches dehydrated to varying negative pressures and compared them with the freezing points of the sap-filled sample branches with similar negative pressures. We find that the freezing onset of extracted sap is c. 10 degrees lower than inside the branches, and uncorrelated with the water potential that was present before extraction. Taken together, these results strongly suggest that supercooling under tension is a purely physical phenomenon and that nucleation is initiated on the surface of the xylem tissue. By drawing an analogy with aerosol science, we propose that pores in the conduit walls may form either an ice embryo or a gas nanobubble, with the competition between the two determining the macroscopic freezing point. A phenomenological model based on the above mechanism reproduces the observed branch freezing onset temperatures within experimental uncertainty.