Knowledge-based framework for delineation and classification of ephemeral plant communities in riverine landscapes to support EC Habitat Directive assessment

Abstract Riverine landscapes are shaped by the spatio-temporal dynamics of the water regime. Water level transitions induce a shift in plant species composition from aquatic to ephemeral vegetation communities in riparian habitats. Hence, the occurrence of these ephemerals is strongly related to the hydrological connectivity and therefore used as indicator for the assessment of riparian habitat types. The delineation and assessment of such habitat types is time-consuming due to the indifferent occurrence of the plants. Therefore, in this study a knowledge-based framework is presented to provide readily usable polygons to support subsequent field surveys on species level. Different hierarchical scales range from hydrological connectivity classes to watercourses and to the micro-morphological classification of riparian habitats. The object-based image analysis approach was used to extract information from terrain and groundwater models, aerial images, and thematic data. The study site is located in the Danube floodplains east of Vienna Natura 2000 site. The micro-morphological classification of the watercourses resulted in the delineation of the classes Waterbodies , Riparian Habitats and the remaining Transition Zones . Watercourses with high flow velocity or with low hydrological connectivity show a small portion of potentially suitable riparian habitats for ephemeral vegetation communities. The framework with focus on terrain models delineating the shape of the riparian habitats performed well with an overall accuracy of 90% (kappa = 0.74). The thresholds in the framework were set fixed or calculated automatically to facilitate an application by spatial ecologist due to the combination of remote sensing techniques and GIS functionalities. The knowledge-based framework can be adapted to provide a harmonised and standardised dataset for any riverine study area.