A predictive approach to assess urban biodiversity and plan for future development scenarios

Protecting and enhancing biodiversity in urbanized areas is recognized as an important priority. To achieve this through urban planning, there must be empirically derived tools to predict biodiversity at the appropriate spatial scales and resolutions given various options in urban designs to compare the expected biodiversity outcomes and make optimal decisions.
We demonstrate how this can be done by developing models that predict the expected species densities or ‘alpha diversity’ in urban landscapes for four animal groups: birds, butterflies, odonates and amphibians, based on assemblage data from spatiotemporally replicated surveys conducted in the tropical city of Singapore. We demonstrate two use cases for these predictive models: citywide assessment and future scenario planning.
For citywide assessment, sub-city ‘towns’ (equivalent to districts or suburbs elsewhere) were compared and benchmarked relative to all other towns, based on the average species densities as indicators of hab..., , # A predictive approach to assess urban biodiversity and plan for future development scenarios

Dataset DOI: [10.5061/dryad.2fqz6131p](10.5061/dryad.2fqz6131p)

## Description of the data and file structure

## Sampling locations

Four animal groups were surveyed at six towns (as officially delimited) in the equatorial city-state of Singapore located in Southeast Asia (1.3° N, 103.8° E). For each town, surveys were conducted every two months across a one-year duration.

Point locations for surveys were randomly sampled across two land-cover strata—natural vegetation and urban cover. Before each survey year, ‘natural vegetation cover’ (forest patches) within the respective study towns was preliminarily delineated in Google Earth; large water bodies were then excluded from the remaining area to form ‘urban cover’ (which includes cultivated vegetation). The two land-cover strata were used to randomly generate sampling points at a density higher than the target of 1 point per 50 ha, to prov...,