The design does the work here. Species that come back fastest after fire tend to be short-lived, while eucalypts and banksia return slowly and then persist, so a single post-fire measurement can score a temporary pulse of growth as repair [9]. Unburned reference sites received the same three above-average rain years as the burned ones [3], which is why growth appearing in both cannot be counted as recovery.
Rainforest burned at all because of what came before: three years in which cool-season rainfall across southeastern Australia ran at half the normal amount, dry enough to carry fire into wetlands and rainforest that usually resist it [1][2]. The largest of the seven fires, Gospers Mountain, covered seven times the area of Singapore from one ignition point, the biggest fire from a single ignition in Australia's recorded history [5].
The satellite record has a gap: it can show that a burned rainforest site is not behaving like an unburned one, but it cannot show whether coachwood or sassafras seedlings are standing under what remains [12]. The sample is not broken out either: 760 sites spread across a million hectares of World Heritage Area [4][6] averages one site per roughly 1,300 hectares [19], and that total is divided again among seven plant communities and a severity range running from low to extreme [18][6].
Five years on, thick scrub had replaced charred undergrowth in many places, and much of the burned area carried denser understories than unburned country [10]. That is the number that reads as success from a helicopter, and it leaves the real bottleneck unmeasured: whether the shrub and canopy layers of a working community are back. Grassy woodland is the sharpest case, dry open country of the sort normally expected to cope, and its trajectory shows no recovery at all [13]. The authors suggest the heavy rain itself stopped grasses and tussocks adapted to dry conditions from re-establishing [14], which is a hypothesis fitted to a pattern rather than a tested cause.
The drought scenario comes from a model rather than direct observation. Under continued drought stress every community did worse, with wet sclerophyll forest and freshwater wetlands falling five times further than the rest [16], because those species need heavy rainfall to re-establish and wetlands often need to be flooded outright before species return [17]. So the one clear success is conditional: wet sclerophyll forest recovered inside five years with three wet years to spend [11][3], and the model implies it would not have managed that in a dry stretch [16].
My view, with its condition attached. For the rainforest and the grassy woodland, five years of clock plus three years of unusually favourable rain amounts to a fair trial of passive recovery, and neither community returned anything at any severity [12][13]. The condition is that Gondwanan canopy trees may work on a schedule no five-year series can resolve, and in a record this short, slow and stalled look much the same [12].
