Contrasting topoclimate, long-term macroclimatic averages, and habitat variables for modelling ant biodiversity at landscape scales

• Published in: Insect conservation and diversity 2015-01, Vol.8 (1), p.43-53
• Main Authors: Gollan, John R., Ramp, Daniel, Ashcroft, Michael B.
• Format: Article
• Language: English
• Subjects: Climate; energy limitation hypothesis; Formicidae; generalised dissimilarity modelling; habitat complexity
• ISSN: 1752-458X; 1752-4598
• DOI: 10.1111/icad.12081

Spatial modelling is part of the solution for incorporating insects into conservation policy. Uptake, however, rests on identifying robust environmental predictors. Coarse‐grained climate models based on long‐term averages and similarly coarse environmental features may not be adequate, especially at regional scales where most planning is done. Here, we test whether topoclimatic variables, which are derived from local‐scale climate forcing factors, are more important for structuring ant assemblages. We quantified ant richness and species composition at 86 sites across a large (200 × 300 km) temperate region of southeast Australia, and tested the explanatory power of three groups of environmental variables: (i) topoclimatic variables, (ii) long‐term climatic averages modelled from global data, and (iii) habitat features, namely, habitat complexity, soil pH, and soil texture. Generalised Additive and Generalised Dissimilarity Models were used to test predictors. In univariate models, the topoclimatic estimator of maximum temperature (95maxT) explained the largest amount of variance in both richness and compositional turnover (20% and 24% of deviance respectively). The plot for richness indicated a positive but decelerating function of 95maxT. This was consistent for two of three habitat types. Habitat complexity was the most important predictor in cleared habitat (28%). While a topoclimatic variable was a strong predictor of ant biodiversity across the landscape, this was not a ‘magic bullet’. Other predictors such as complexity may be more applicable in certain habitat types. We concluded that tailored predictors are needed for landscapes with a mosaic of different land use.