Predicting the century-long post-fire responses of reptiles

• Published in: Global ecology and biogeography Vol. 21; no. 11; pp. 1062 - 1073
• Main Authors: Nimmo, Dale. G., Kelly, Luke T., Spence-Bailey, Lisa M., Watson, Simon J., Haslem, Angie, White, John G., Clarke, Mike F., Bennett, Andrew F.
• Format: Journal Article
• Language: English
• Published: Oxford, UK Blackwell Publishing Ltd 01.11.2012; Blackwell Publishing; Blackwell; Wiley Subscription Services, Inc
• Subjects: Amphibia. Reptilia; Animal and plant ecology; Animal, plant and microbial ecology; Australia; Biogeography; Biological and medical sciences; chronosequence; Ecological modeling; Fauna; fire; Fundamental and applied biological sciences. Psychology; General aspects; generalized additive mixed models; habitat accommodation model; Landscapes; mallee; Mallees; model transferability; Modeling; Predictive modeling; Reptiles; Species; Synecology; Vegetation; Vegetation structure; Vertebrates: general zoology, morphology, phylogeny, systematics, cytogenetics, geographical distribution; Australia; Oceania; generalized additive mixed models; reptiles; model transferability; Biogeography; Ecology; Additive model; Vertebrata; Chronosequence; habitat accommodation model; Fires; fire; Habitat; Reptilia; Australia; mallee
• ISSN: 1466-822X; 1466-8238
• DOI: 10.1111/j.1466-8238.2011.00747.x

Aim: We examined the century-long post-fire responses of reptiles to (1) determine the time-scales over which fauna — fire relationships occur, (2) assess the capacity of a conceptual model to predict faunal response to fire, and (3) investigate the degree to which models of fauna — fire relationships can predict species occurrence and are transferable across space. Location: A 104,000 km 2 area in the semi-arid Murray Mallee region of southeastern Australia. Methods: We surveyed reptiles at 280 sites across a century-long post-fire chronosequence. We developed generalized additive mixed models (GAMMs) of the relationship between time since fire and the occurrence of 17 species in two subregions, and compared modelled responses with predictions derived from the conceptual model. The predictive capacity of GAMMs was then assessed (1) within the subregion the model was developed and (2) when transferred into a novel subregion. Results: Eleven species displayed a significant relationship with time-since-fire, with changes in species probability of occurrence continuing up to 100 years post-fire. Predictions of the timing of species post-fire peak in occurrence were accurate for 9 of 13 species models for which a significant fire response was detected, but little success was achieved in predicting the shape of a species' response. GAMMs predicted species occurrence more accurately when applied within the subregion in which they were developed than when transferred into a novel subregion, primarily due to some species responding to fire more strongly in one part of their geographic range. Main conclusions: Fire influences the occurrence of reptiles in semi-arid ecosystems over century-long time frames. Habitat-use conceptual models have value in predicting the peak occurrence of species following fire, particularly for species with distributions strongly shaped by fire. Species relationships with fire can differ across their geographic range, probably associated with variation in climatic influences on post-fire succession and the consequent provision of habitat resources.