Abstract
Population connectivity maintains genetic diversity and underpins adaptive capacity and long-term persistence. When assessing connectivity, landscape genetic analyses have largely focused on static features such as topography and have rarely incorporated disturbance regimes like fire, which may also shape genetic connectivity. We assessed genetic diversity and structure for three declining mammal species in northern Australian savannas: the northern brown bandicoot (Isoodon macrourus), northern brushtail possum (Trichosurus vulpecula arnhemensis) and black-footed tree-rat (Mesembriomys gouldii melvillensis). Using genetic distance data, we optimised resistance surfaces to evaluate how landscape variables, including fire history, rainfall, vegetation, topography, watercourses, and feral species, influence gene flow. Analyses were conducted at fine (< 20 km between individuals) and broad (whole-island) scales across Bathurst (up to 60 km) and Melville (up to 120 km) Islands, which differ markedly in disturbance regimes. Genetic patterns and their drivers varied by species, island and scale. The influence of fire was most apparent on Bathurst Island, where feral predators and herbivores are less abundant and rainfall is high. Northern brown bandicoots showed weak broad-scale genetic structure on Bathurst Island, while high fire frequency reduced connectivity at fine scales. For northern brushtail possums, high rainfall reduced genetic connectivity at both broad and fine scales. On Melville Island, where multiple interacting threats of fire, feral predators and herbivores are more prevalent, topographic ruggedness promoted connectivity for northern brown bandicoots and northern brushtail possums at broad scales. At fine scales, geographic distance best explained the genetic patterns for these species. For black-footed tree-rats, which occur only on Melville Island, broad-scale connectivity was reduced in low fire frequency areas (e.g., plantation and mangrove), while at fine scales low rainfall and high ruggedness were associated with reduced gene flow. Our results show that fire can influence resistance to gene flow, but effects are species- and context-dependent. Effective conservation requires accounting for species-specific ecology and local disturbance regimes when evaluating connectivity. Fire management should be a priority where fire strongly structures gene flow, while landscapes with multiple interacting threats require broader strategies including feral species control and habitat protection. Our study underscores the value of incorporating disturbance regimes into landscape genetics to guide context-specific conservation.