Using multilevel models to identify drivers of landscape-genetic structure among management areas
RACHAEL Y. DUDANIEC,* 1 JONATHAN R. RHODES,* 1 JESSICA WORTHINGTON WILMER, MITCHELL LYONS,* KRISTEN E. LEE, CLIVE A. MCALPINE* and FRANK N. CARRICK
School of Geography, Planning and Environmental Management, The University of Queensland, Brisbane, QLD 4072, Australia
Natural Environments Program, Queensland Museum, PO Box 3300, South Brisbane, QLD 4101, Australia
Centre for Mined Land Rehabilitation, Sustainable Minerals Institute, The University of Queensland, Brisbane, QLD 4072, Australia
Landscape genetics offers a powerful approach to understanding species’ dispersal patterns. However, a central obstacle is to account for ecological processes operating at multiple spatial scales, while keeping research outcomes applicable to conservation management. We address this challenge by applying a novel multilevel regression approach to model landscape drivers of genetic structure at both the resolution of individuals and at a spatial resolution relevant to management (i.e. local government management areas: LGAs) for the koala (Phascolartos cinereus) in Australia. Our approach allows for the simultaneous incorporation of drivers of landscape-genetic relationships operating at multiple spatial resolutions. Using microsatellite data for 1106 koalas, we show that, at the individual resolution, foliage projective cover (FPC) facilitates high gene flow (i.e. low resistance) until it falls below approximately 30%. Out of six additional land-cover variables, only highways and freeways further explained genetic distance after accounting for the effect of FPC. At the LGA resolution, there was significant variation in isolation-by-resistance (IBR) relationships in terms of their slopes and intercepts. This was predominantly explained by the average resistance distance among LGAs, with a weaker effect of historical forest cover. Rates of recent landscape change did not further explain variation in IBR relationships among LGAs. By using a novel multilevel model, we disentangle the effect of landscape resistance on gene flow at the fine resolution (i.e. among individuals) from effects occurring at coarser resolutions (i.e. among LGAs). This has important implications for our ability to identify appropriate scale-dependent management actions.
Keywords: habitat fragmentation, landscape genetics, mammal dispersal, multilevel model, spatial scale, wildlife management
Received 22 November 2012; revision received 2 April 2013; accepted 5 April 2013