Landscape-scale Remote Sensing and Classification of Lentic Habitats in a Tropical City

Ponds and lakes may be common in urban landscapes and frequently have high biodiversity and conservation value. The importance of landscape-scale conservation of lentic habitat networks has recently been recognised, yet little research has been conducted at this scale. Approaches to inventorying lentic habitats at the landscape-scale are needed to support science and conservation. This paper uses remote sensing to inventory lentic habitats across Greater Kuala Lumpur (GKL), Malaysia, characterising their distribution, abundance and type. Remote sensing images were employed to automatically identify and map the distribution of lentic habitats, capturing 1,013 individual ponds and lakes, which represent 74.3 % of the likely total. Partitioning Around Medoids (PAM), a multivariate hierarchical clustering method, was used to develop a typology using physical characteristics extracted from the imagery, yielding six distinct types. The diversity of habitat types was greater in peri-urban and suburban areas (Shannon’s H = 1.68) than in the urban core (Shannon’s H = 1.48), suggesting that urban structure influences the spatial pattern of lentic habitat diversity. Physicochemical and vegetation field data were collected from a sample of each type (n = 60). Comparisons of group membership in the typology produced from remote sensing and field data indicated demonstrable heterogeneity in local conditions within each type, and a lack of consistency between local and broader scale characteristics. Our work shows how inventories produced from remote sensing can provide insights into landscape-scale biodiversity drivers, and contribute to larger-scale conservation of lentic habitat networks. This approach can complement local studies to improve understanding of factors operating at different scales.