REMNANT SEED BANKS AND VEGETATION AS PREDICTORS OF RESTORED MARSH VEGETATION

Selection of sites for wetland restoration requires analysis of probab le success at recreating diverse natural vegetation. We studied the se ed banks and the remnant vegetation at restoration sites to determine their degree of similarity to the vegetation that developed following restoration of wetland hydrology. The study sites had been used for fo rage crops and pasture following long-term drainage. The seed bank was a very poor predictor of plant species abundances following restorati on. Similarity between remnant vegetation and restored vegetation was consistently higher than similarity values using seed banks at both re stored and natural wetlands. Our results suggest that seed-bank analys is is an inefficient technique for predicting restored vegetation in s ites with prolonged disturbance, and that analysis of remnant vegetati on on the sites is probably more useful. However, results would probab ly differ at sites with tile drainage that have limited remnant vegeta tion or created wetlands with bare soil where seed germination would p laya greater role in revegetation. In addition, seed-bank studies are important to determine if aggressive invasive species are present at p otential restoration sites.