Context . Because they are dependent on water, drought can have a deleterious impact on aquatic-breeding amphibians. One such species, the threatened growling grass frog (Litoria raniformis) occurs in south-eastern Australia, a region that has recently emerged from a decade-long, severe drought.

Aims . We aimed to identify features of drought refugia that facilitate persistence of L. raniformis, so as to provide guidance to natural-resource managers attempting to conserve populations of this species during drought.

Methods . We conducted repeat surveys for L. raniformis at 90 water bodies at the end of the ‘millennium drought’. We recorded the following six environmental variables for each water body: origin (natural or not), type (lotic or lentic), proportion of aquatic vegetation cover, conductivity of water, riparian tree-canopy cover and distance to the nearest woodland. We used occupancy models to relate the presence of L. raniformis to these variables, while accounting for uncertain detection.

Key results . Water-body type (natural or artificial, lentic or lotic) had minimal influence on the probability of occupancy by L. raniformis. We found a strong negative relationship between occupancy and conductivity of water (a surrogate for salinity), and a positive relationship between occupancy and the proportion of aquatic vegetation. We found a negative relationship between detection and the extent of aquatic vegetation, and a mildly negative effect of canopy cover on occupancy.

Conclusions . Habitat characteristics are more important indicators of the quality of drought refugia for L. raniformis than is the type of water body per se. Consequently, we identified aquatic vegetation and salinity as important targets for management when planning the retention, creation or restoration of habitat to facilitate persistence of L. raniformis during drought.

Implications . Our results highlighted aquatic vegetation and water-quality parameters that are likely to facilitate the persistence of L. raniformis during drought. Assessing the effectiveness of our recommendations in an experimental framework would ensure that conservation management of this frog can be refined over time.