Over the last 50 years, national, international and private stewardship and
conservation organizations have spent billions of dollars collecting marin
e ecosystem information. This remains divided among many custodians, scatte
red among thousands of published sources and, from a global perspective, is
fragmentary in nature. It is argued that new resource management questions
regarding coastal ecosystem health can be addressed through the recovery a
nd 'data mining' of this previously collected and often discarded informati
on. A retrospective marine epidemiological approach was developed to demons
trate that marine morbidity, mortality, and disease information is recovera
ble by keyword searching of academic journals and through the retrieval of
publicly available digital and print-media information. Observational recor
ds compiled from disturbances occurring within the Northwestern Atlantic, G
ulf of Mexico, and Caribbean Sea confirm that anomalous marine morbidity an
d mortality events have increased in number and frequency during the last 3
0 years. A global approach is summarized for systematically reconstructing
spatial and temporal disturbance indicator time series using data mining an
d data reduction techniques. (C) 2000 Elsevier Science Ltd. All rights rese
rved.