From the referenced paper:
The frigid concentration or freezing of seawater is an important natural phenomenon in the polar regions and results in the precipitation of a different sequence of salts - and thus produces brines of different composition - to that formed during isothermal evaporation under temperate conditions (about 20-25 degrees C). Seawater freezing, however, has been studied less ... extensively than evaporation and somewhat greater uncertainty exists over the exact nature of the compositional pathway followed. Most investigators have shown that the precipitation of mirabilite (Na2SO4 - 10 H2O) or gypsum (CaSO4 - 2 H2O), which both occur at the same seawater concentration factor (SWCF), is the critical difference between frigid and evaporative concentration, respectively, a consequence of the very different temperature dependence of the solubilities of these salts, as well as the effect of sodium chloride on these properties. This difference can be considered to represent a temperature-dependent chemical divide in the closed-basin concentration of seawater because it determines significantly the major ion composition of the brine and the salt mineral assemblage precipitated on further evolution of the system. Recently new insights into seawater freezing have been achieved through improvements in existing chemical equilibrium models. Along with the results of some associated experimental work, this has provided evidence for the formation of gypsum during freezing, contradicting the accepted Ringer-Nelson-Thompson model of frigid concentration firmly established in the 1950's and through subsequent studies, but validating an alternative model proposed by Gitterman two decades later.