Response of Eucalyptus-Dominated Savanna to Frequent Fires: Lessons From Munmarlary, 1973–1996

assess a replicated fire plot experiment undertaken between 1973 and 1996 in two Eucalyptus-dominated savanna vegetation formations (open forest, woodland), at Munmarlary, in monsoonal northern Australia. Four treatments, each with three replicates, were imposed on each vegetation type: annual early dry-season burning; annual late dry-season burning; biennial early dry-season burning; and unburned controls. Treatments were imposed faithfully, with noted exceptions, on 1-ha plots. Fire intensities were typically low (1000 kW/m) to moderate (1000-2500 kW/m), varied significantly between treatments, and generally were greater in woodland. In both woodland and open forest, pH was significantly lower and NO3-N was significantly higher in unburned plots. Organic C was not significantly greater. in unburned treatments. Effects of fire regime on other soil chemical properties differed between open forest and woodland sites. Among the grasses, invariant frequent burning led to the dominance of a small number of annual species, notably regionally dominant Sorghum. In the absence of burning, annuals declined generally, whereas some perennials increased while most decreased. These responses usually were apparent within the first five years of the experiment. At the relatively, small spatial scale of the grass sampling regime, there was high turnover of both annual and perennial grasses. Under low- to moderate-intensity, frequent burning regimes, woody vegetation dominated by mature eucalypts is structurally stable. In the absence of burning for at least five years, there was release of the non-eucalypt, woody component into the midstory; this occurred more rapidly in open forest. Accession of rain forest species occurred on some woodland plots, especially the unburned treatment. In contrast, eucalypts were not released significantly from the understory. Rather, as suggested by other studies, recruitment of eucalypts into the canopy appears to involve significantly reduced root competition through death of dominant eucalypts. Although the Munmarlary experiment provides invaluable quantitative data for exploring relationships between fire regimes and the responses of north Australian savanna systems, it has been less successful in meeting the complex information requirements of regional fire managers. Replicated experimental fire plot designs, no matter how elegant and rigorously implemented, may substantially fail the test of management relevance, given the fundamental requirement for savanna biodiversity managers to experience the integrated effects of fire regimes that vary idiosyncratically over multiple time and spatial scales. We suggest that such information requirements are better met through modest, targeted adaptive management studies, involving collaborative partnerships between managers and researchers.