Changes during regeneration after sand mining for heavy minerals were studied on an area that previously supported open-forest on the Holocene high dune system in Myall Lakes National Park. Sixteen study areas, on which topsoil was replaced 0.5–10 yr ago, provided information on changes to be expected on any one site over that time period. A number of environmental variables was used in a linear multiple regression analysis to determine which of them is important in accounting for the variance and patterns observed in the biomass of two rodent species colonizing the regenerating areas.
The first small-mammal colonist is the introduced house mouse (Mus musculus), an opportunistic species present on all sites. Its population density increases rapidly to a maximum at 3 yr after which it declines. Three environmental variables: percentage of bare sand, hardness of the soil in the first 30 cm, and a vegetation structure variable, account for 73% of the variance in M. musculus biomass.
The native New Holland mouse (Pseudomys novaehollandiae) first appears between 4 yr and 5 yr after topsoil replacement and peaks between 8 yr and 9 yr, after which it also declines. A highly significant multiple regression accounts for 69% of the variance in P. novaehollandiae biomass using five significant variables: the proportíon of heath plants present, two vegetation structure variables, the amount of dead plant cover, and the topsoil depth. The species replacement observed confirms the succession indicated by previous work and suggests competitive interaction between these species.
This study confirms the seral positions of rodent species in successions following both mining or fire. There is a stretched time axis for the mining succession following the more complete disruption of the substrate so regeneration more closely approximates primary rather than secondary succession. The non-linearity of the parameters makes it impossible to predict a recovery time and verifies our previous prediction that recovery estimates from heathland should not be extrapolated to forest.