LAUSR

Habitat losses occur non-randomly within human-modified landscapes, resulting in high spatial heterogeneity of local habitat histories. Although local habitat history can modulate the existence of extinction debt (i.e., the number of populations predicted to become extinct) in a landscape, its role in detecting extinction debt has not been examined explicitly. We aimed to compare the detectability of extinction debt among populations of an endangered semi-natural grassland species, Echinops setifer (Compositae), in the grassland landscape of Mt. Aso, Japan. We classified populations into three groups that differed in local habitat history: stable (habitat loss ≤30% since the 1930s), moderately decreased (30% < loss ≤ 90%), and severely decreased (loss >90%). We then evaluated whether the effects of habitat areas during the 1930s and 2000s varied among groups to explain population size by GLMMs and estimated coefficient of explanatory variable by Bayesian MCMC methods. Within the groups, stable group showed significant positive relationships with both past and current habitat areas. The moderately decreased group only showed significant positive relationships with past habitat areas, indicating the existence of extinction debt in these populations. The severely decreased group only showed significant positive relationships with current habitat areas, indicating that they may have already paid their extinction debt because the rate of grassland loss exceeded the extinction threshold. Even within the same landscape, extinction debt varied in response to local habitat history. In spatially heterogeneous landscapes, evaluation of effects of local habitat history can elucidate the habitat-based extinction risks for plant populations.