• emergence;
  • mortality;
  • seedbank;
  • selectivity;
  • weed;
  • wet-dry tropical;
  • wetland

Abstract We studied the seedbank of floodplain vegetation in three major tropical river systems in northern Australia, which had been variously invaded by the tropical woody weed Mimosa pigra. The sites selected had not previously been treated with tebuthiuron, a herbicide which is widely used in northern Australia to control Mimosa. We collected soil seedbank samples from two floodplain vegetation types (Melaleuca swamp and sedgeland), and, within each type, from areas in which Mimosa was either present or absent. The effects of treatment with tebuthiuron at 15 kg ha−1, twice the usual recommended rate, was subsequently assessed in die laboratory on the soil-seedbank samples. Ordination of the species composition of seedlings which emerged from the soil seedbank samples showed no effect of (i) the vegetation community from which the samples were collected, (ii) the presence of adult Mimosa, or (iii) treatment with tebuthiuron. The effect of tebuthiuron on the emergence and mortality of seedlings from four functional groups (grasses, sedges, forbs and Mimosa) was also tested on the seed bank samples. Emergence was significantly decreased by tebuthiuron only for forbs from Melaleuca swamps. The mortality of Mimosa was significantly higher than that of the other functional groups, but there was some mortality of forb and grass seedlings. Sedges, however, were unaffected. The impact of tebuthiuron on Mimosa depended on soil clay content—in the soils with lowest clay content, tebuthiuron was the most effective in killing Mimosa seedlings. Mortality in forb and grass seedlings, in contrast, was not affected by soil clay content. Tebuthiuron was therefore selective against Mimosa seedlings. However, even at twice the recommended rate of application for killing adult Mimosa, under ideal conditions for distribution of the herbicide through the soil, 43% of Mimosa seedlings survived. Given the size of the Mimosa seedbank under field conditions (∼10 000 seeds/m2), tebuthiuron can therefore not be considered an effective herbicide against Mimosa seedlings.