Classification and structural connectivity of urban vegetation : A comparative study using different datasets

Abstract: Urban greenspace has an important role in supporting biodiversity and providing structural and functional connectivity between natural habitats. However, the mapping of vegetation in urban areas presents challenges, as urban vegetation is highly fragmented and heterogeneous. This study compared maps with respect to their strengths and weaknesses in providing ecologically relevant information in an urban area with the purpose to find how useful they are for local planning. The study took place in the urban part of Dún Laoghaire-Rathdown County, Ireland. The maps analysed were the Urban Atlas 2012 provided by Copernicus, the Prime 2 database provided by the Ordnance Survey Ireland, and two maps that were generated by this project from Sentinel 2 satellite imagery, one which showed vegetation based on calculated NDVI and one with four land cover classes calculated with a supervised classification. These maps were compared from three points of view: correspondence of vegetation classes, structural connectivity, and quality of vegetated areas. Analyses of structural connectivity were based on several landscape metrics, one of them was the degree of coherence (Cm) which reflects how connected the vegetation patches are in the landscape. Definition of quality was based on contextual indicators, such as the proximity to streams and areas with high conservation value, and vegetation present on a historical map from the 1830s. The results showed that the overlap of vegetation between the datasets varies between 27.0-89.1%. The different datasets definitions of vegetation affect how well they correspond in terms of where vegetation can be found. Resolution is also an important factor, as urban vegetation patches tend to be small and thus a coarse minimum mapping unit – as is the case for the Urban Atlas – masks important information on the configuration of vegetated areas in the urban area. The structural connectivity of vegetation differed little between Prime 2 (Cm = 7.95×10-2%) and the Urban Atlas (Cm = 5.87×10-2%). However, the distance between vegetated patches was on average shorter in Prime 2. This suggests that the Prime 2 dataset, because of its higher spatial resolution, contains more information on potential stepping stones for species to move around the landscape. The distribution of vegetated areas with higher contextual quality was mainly close to the boundaries of the urban area. Connectivity in a fragmented landscape like this urban study area plays a crucial role in maintaining populations of flora and fauna. It is therefore important to consider in the management of urban vegetation and in planning for development. This study offers a first insight in the structural connectivity of Dún Laoghaire-Rathdown County, which can be used to make more informed decisions that will sustain urban biodiversity.