Modeling a Red deer population reintroduced in an area of Northern Apennines (N-Italy)

At the beginning of the past Century only two remnant populations of red deer (Cervus elaphus) were present in Italy : one in the Eastern Alps (Monastero and Venosta valleys) and the other in the Po Plain near the coast of Adriatic Sea (Mesola Forest). The present Alpine population was originated by immigration from Switzerland, Austria, and Slovenia (Central and Eastern Alps) and by reintroductions (Western Alps), whereas in the Apennines red deer was reintroduced since 1960-70. At present Apennine populations are fragmented and isolated even if an increasing trend of density and occupied range was recorded. We monitored from 2002 to 2012 a protected population of red deer reintroduced in late eighties in the hilly and mountainous areas of the Piacenza province (Northern Apennines), in order to detect changes of population size and range, and to formulate a model of habitat suitability for the prediction of future expansion of population. We carried out counts of roaring males in September of each year and collected systematic observations all year round to define population structure and reproductive success. We used species locations, to define yearly ranges and core areas by Kernel Analyses (KA) at 99% and 50%, curve fit regression to detect trend of the population and its range, and Ecological Niche Factor Analysis (ENFA) to model habitat suitability and potential range. For ENFA we subdivided the study area in 4-km2 cells and in each cell we measured the altimetry, exposure, slope, and land use variables; the cells where at least one observation was collected in the study period were considered as presence ones. Finally we carried out Population Viability Analyses (PVA) to predict population trend and to explore the harvesting possibility. Deer population showed a slow increasing trend (from 4 roaring males in 2002 to 70 in 2012). Only KA50% resulted in a low increasing trend (F=13.78; df=1; P=0.014; R2=0.73). ENFA showed a global marginality of 0.53 and a tolerance of 0.71. The main variables positively related to the marginality were broad-leaved and conifer forests, pastures, East aspect, altitudes from 1000 to 1800 m a.s.l., and slopes between 10° and 30°. The model provided a habitat suitability map that defined a potential red deer range of 580 km2 (22.4% of the Piacenza province). K-Fold cross validation and ROC curves showed a high predictive power of the model (Rho=0. 96; P<0.0001; AUC=0.99; P