5. Conclusions
In summary, we classified the natural hydrological regimes of NPRS located in the western Mediterranean region using daily flow data sets from 69 unaltered gauging stations. Using hydrological series of at least 11 years of duration, a total of 315 hydrological indices were calculated using different thresholds to define no-flow conditions (i.e. 0, 1, 2, and 5 l/s). We did not find notable statistical differences between the different thresholds and after redundancy analysis. A neural network model was calibrated with the 5 hydrological metrics that better describe the attributes of flow magnitude, duration, frequency, and timing of the zero-flow periods, and rate of change between flowing and non-flowing conditions. The 5 hydrological metrics describe annual attributes as number of days or percentage of months per year without flow mean annual flows, coefficient of variation of Julian date of the annual start of zero flow and annual rise rate. The analysis concluded that three different categories of NPRS can be defined, namely: ephemeral, intermittent, and temporal rivers. Ephemeral rivers are dry for most of the year and would have a low flow during the wet phase with water in their riverbed. Intermittent rivers have a dry phase not as long as ephemeral rivers, and, as a rule, more water runs through their bed. Temporal rivers are more briefly interrupted by a period of drought. Additionally, our classification revealed ranges that are not closed and different from the national legislation classifications of Spain and Italy. Ephemeral rivers could be dry for more than 181 days or 7 months, intermittent rivers between 100 and 187 days or between 4 and 7 months, and temporal rivers less than 46 days and 2 months. Therefore, our results could serve for the conservation and management of these rivers, establishing a necessary typology in future studies of ecological status of WFD, with well-defined groups based on statistical analysis. The lack of studies addressing the hydrological diversity of these freshwater ecosystems makes our study a basis for future efforts aimed at improving management strategies of this type of river courses. Additionally, future analyses guided to a better characterization of biological communities associated with different types of NPRS will allow a better understanding of the ecological implications of temporality on the biota and will advance in better ecological status classification and management of these ecosystems. Finally, due to the relative low number of NPRS analysed, the resulting classification should be considered as a first approximation and should be tested and calibrated with other gauging stations not included in the study.