We studied the patterns and mechanisms of regeneration of a 400 ha windthrow in an old-growth beech-hemlock forest caused by a tornado on 31 May, 1985. Starting in 1986, and over a period of six growing seasons, we recorded percent cover and density of woody stems, and monitored seedling demography of nearly 5000 seedlings in the windthrow and adjacent forest.
Plant community response to the disturbance was dramatic: by August of 1986, species richness, tree seedling density and total percent cover were significantly greater in the windthrow than in the adjacent forest. Shade-intolerant herbs (e.g. Erechtites hieracifolia) and shrubs (e.g. Rubus allegheniensis) established and rapidly increased in abundance during the first three years, but began declining by the fifth year of the study. Tree seedlings established in decreasing amounts through the six years of the study, and the young tree canopy was dominated in 1991 by seedlings and sprouts that established prior to 1987.
Fagus grandifolia, a shade-tolerant species that established via advanced regeneration, was dominant the first three years, but was surpassed in the fifth year by Betula alleghaniensis, a species of intermediate tolerance that established from seed germination just before or shortly after the disturbance. Tsuga canadensis seedling densities were initially high, but deer browsing prevented substantial growth and a drought in 1988 caused heavy mortality of browsed seedlings. Regeneration thus differed from the predictions of the gap and Hubbard Brook models of forest regeneration (which predict dominance by shade- intolerant species), and the severity model (which predicts dominance by shade-tolerant species). The differences point out important influences of availability of propagules and the impact of herbivory; and the need for more attention to models that incorporate multiple contingencies.