Using high-resolution multi-spectral imagery to estimate habitat complexity in open-canopy forests: can we predict ant community patterns?

The structure and composition of arthropod assemblages are strongly associated with habitat complexity. Accurate, time efficient estimates of habitat complexity may provide insights for biodiversity management in natural systems. We obtained high-resolution (0.7 m pixel) multi-spectral aerial imagery of National Parks 20 km north and 20 km south of Sydney, Australia. We explored both the Normalised Difference Vegetation Index (NDVI) and the standard deviation of reflectance in the near-infrared spectrum (${\rm stdevR}_{{\rm NIR}}$) as indicators of low and high habitat complexity in sandstone forests north of Sydney. We then tested described predictions of ant community patterns (based on a previous study) using sites selected from high-resolution multi-spectral imagery in sandstone forests south of Sydney. Ground-scored habitat complexity was positively correlated with NDVIs and, to a lesser extent, ${\rm stdevR}_{{\rm NIR}}$ values in sandstone forests north of Sydney. As predicted, ant species richness was negatively correlated with NDVIs in forests to the south of Sydney. Also, ant species composition was different in areas with contrasting NDVIs. The ant species driving composition differences responded to habitat complexity in a similar way in forests to the north, and south, of Sydney. The strong association we detected between NDVIs and habitat complexity, most likely reflects the relatively exposed nature of the vegetative layers in the forests we sampled. Remote sensing, integrated with quantitative research testing predictive faunal responses to vegetation structure and biomass at landscape scales, may provide efficient means of estimating biodiversity for management in particular habitats.