Melt pond formation atop Arctic sea ice is a primary control of shortwave energy balance in the Arctic Ocean. During late spring and summer, the ponds determine sea ice albedo and how much solar radiation is transmitted into the upper ocean through the sea ice. The initial formation of ponds requires that melt water be retained above sea level on the ice surface. Both theory and observations, however, show that first year sea ice is so highly porous prior to the formation of melt ponds that multiday retention of water above hydraulic equilibrium should not be possible. Here we present results of percolation experiments that identify and directly demonstrate a mechanism allowing melt pond formation. The infiltration of fresh water into the pore structure of sea ice is responsible for blocking percolation pathways with ice, sealing the ice against water percolation, and allowing water to pool above sea level. We demonstrate that this mechanism is dependent on fresh water availability, known to be predominantly from snowmelt, and ice temperature at melt onset. We argue that the blockage process has the potential to exert significant control over interannual variability in ice albedo. Finally, we suggest that incorporating the mechanism into models would enhance their physical realism. Full treatment would be complex. We provide a simple temperature threshold‐based scheme that may be used to incorporate percolation blockage behavior into existing model frameworks.

To understand the controls on distributions of ice algal communities in spring and the role of ice algae in under‐ice bloom development through possible seeding, we sampled the ice and water column in the Chukchi and Beaufort Seas over spring and summer. Field observations showed that high springtime concentrations of bottom ice algal communities were released from the ice into the water column by summer. Furthermore, during our spring sampling, bottom ice algal concentrations were highly variable. Declines in spring ice algal biomass and physiological state were correlated with ice melt, rather than light or nutrient availability. Nonparametric multivariate data analysis of the seasonal succession of phytoplankton and ice algal community composition illustrated that the loss of algae from the sea ice temporarily elevated water column chlorophyll a (Chl a) levels, as ice‐derived taxa dominated the phytoplankton biomass. Model simulations, constrained by field observations from this study, further suggested that seeding by ice algae was brief and alone could not account for the phytoplankton biomass concentrations exceeding 2 mg Chl a m−3 observed in our study. Ice algal sloughing from the sea ice to the water column contributes biomass to the phytoplankton community. However, this signal of ice‐derived taxa is brief and non‐ice derived taxa dominate phytoplankton blooms later in the spring and summer.