Landscape fragmentation as an indicator of coastal landscape quality: an application along the Apulian coast (southern Italy),45. MININNI M., MINUNNO F., LERONNI V., TARANTINO C. MAIROTA P..(2007), Landscape fragmentation as an indicator of coastal landscape quality: an application along the Apulian coast (southern Italy),

This work has being carried out within the framework of the IMCA (Integrated Monitoring of Coastal Areas) Research Project, among the activities aimed at drawing coastal landscape quality maps through the use of indicators derived from satellite RS images. The overall research project, moving from the experience of the European Landscape Convention, tackles the landscape quality issue via a multi temporal and spatial scale approach. The present contribution focuses on fragmentation as this phenomenon, as well as the loss of heterogeneity, initiated by urban settlement processes of dislocation and diffusion, represents the main cause of the landscape ecological efficiency decrease, of the area decay and of the beginning of diseconomy in its management (Forman, 1995). In order to quantify fragmentation, at a given spatial scale (defined in terms of both grain and extent), a set of LPI (ED, LSI, ENN_AM, PLADJ, MESH, SHDI) was computed at the landscape level on a sample plot population, extracted via an unaligned random samplingprocedure from the whole southernmost part of the Apulian peninsula (Southern Italy) and for which intepretation of recent aerial photographs had already been performed within the framework of the IMCA research project (Miacola et al. 2006). The same protocol was applied to categorical maps of the same area, derived, both by past aerial photointerpretation and by (unsupervided and supervised) segmentation, from medium (Landsat TM) resolution satellite images of two time steps. Preliminary results are encouraging in many respects. The distribution analysis performed on the indexes computed on the different data sets show, for this particular landscape at the given scale, a significant trend towards a normal distribution model, thus contributing to the ongoing debate (Remmel and Csillag 2003) on the uncertainties about the possibility to statistically compare indexes computed in different times and places, deriving by the lack of knowledge about their distribution. Principal component analysis performend on the indexes obtained from the different data sets, yields the ordination of sample plots along a fragmentation gradient, that migth be used to construct framentation intensity maps at the subregional scale, as well as to interpreting the change processes and obtain intelligent maps based upon the integration of field (aerial-photo interpetation) and and RS data, thus achieving the twofold purpouse of performing a phenomenological study aimed both at tmodelling coastal landscape transformations and identifying new survey categories that may have the temporal dimension as a reading parameter (e.g.. speed of change). As far as the relations between the indexes computed on the different data sets are concerned, they allow for the assessment of the potentials for using unsupervised categorical maps for the description and monitoring of landscapes fragmentation, as well as for testing hypoteses concerning fragmentation scaling relations in both space and time (Wu, 2004; Jelinski and Wu 1996)