At coastal sites, trees are exposed to marine aerosols that may cause foliar necrosis and shoot dieback, which can result in deformed crowns and contorted stems. A six-year study of leaf primordia in terminal buds of black cottonwood trees (Populus trichocarpa Torr. & Gray) on Heimaey Island off the south coast of Iceland was undertaken to elucidate the physiological events associated with salt-deposition-related bud failure. Leaf and bud lengths, dry mass, water content and chloride concentrations were monitored and related to four phenological stages: (1) bud set; (2) dormancy induction; (3) dormancy release; and (4) bud break. The trees set buds in July and shed their leaves by late September. Leaf primordia generally stopped growing by September 10 +/- 22 days and attained midwinter water content in late September. Leaf growth commenced in the terminal buds by March 2 +/- 16 days, but mean dates of bud swelling and bud break were April 29 +/- 19 and May 10 +/- 12 days. In summer and until November, chloride concentrations in leaf primordia were low, but increasing. Chloride concentrations remained stable from December to February, even though the dormant trees were exposed to large amounts of marine aerosols. In February and March, three events occurred more or less simultaneously: (1) leaf extension growth commenced; (2) chloride concentration surged in the leaf primordia; and (3) the leaf primordia began to hydrate. Following dormancy release, growth and hydration of leaf primordia were negatively related to chloride concentration in the leaf primordia, with inhibition of leaf growth, tissue hydration and chloride acquisition occurring at a chloride concentration threshold estimated at 7.3 mg Cl- g(-1) tissue water. Necrosis of leaf primordia was observed above 14 mg Cl- g(-1) tissue water. Growth and hydration of leaves at bud break in mid-May was explained by a three-parameter logistic model of chloride concentration in leaf primordia at the end of March. By mid-May, 90% of all buds remained non-necrotic, but only 56% the terminal buds had broken. Salt alone explained the observed growth suppression of leaf primordia in the buds and the resultant failure of terminal buds to break by mid-May.