Abstract
Small changes in environmental conditions can unexpectedly tip an ecosystem from one community type to another, and these often irreversible shifts have been observed in semi-arid grasslands, freshwater lakes and ponds, coral reefs, and kelp forests. A commonly accepted explanation is that these ecosystems contain multiple stable points, but experimental tests confirming multiple stable states have proven elusive. Here we present a novel approach and show that mussel beds and rockweed stands are multiple stable states on intertidal shores in the Gulf of Maine, USA. Using broad-scale observational data and long-term data from experimental clearings, we show that the removal of rockweed by winter ice scour can tip persistent rockweed stands to mussel beds. The observational data were analyzed with Anderson's discriminant analysis of principal coordinates, which provided an objective function to separate mussel beds from rockweed stands. The function was then applied to 55 experimental plots, which had been established in rockweed stands in 1996. Based on 2005 data, all uncleared controls and all but one of the small clearings were classified as rockweed stands; 37% of the large clearings were classified as mussel beds. Our results address the establishment of mussels versus rockweeds and complement rather than refute the current paradigm that mussel beds and rockweed stands, once established, are maintained by site-specific differences in strong consumer control.