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 habitat value for
each of the four animal groups. For future scenario planning, four
development scenarios were compared, and the compatibility of vector-type
planning layers with the models was tested. An open-source R package,
biodivercity, was developed that would facilitate the use of the same
workflow elsewhere: to build, apply and validate predictive models
elsewhere given similar available empirical data. Synthesis and
applications: The models developed can also be examined to generate
recommendations for further actions that can improve biodiversity across
different spatial scales. These techniques can be incorporated into
current planning practices to achieve a more quantitative and
performance-based approach to enhancing biodiversity at fine spatial
scales in human-dominated landscapes.