Analyzing effective protection for roe deer (Capreolus capreolus) habitat in Iranian Zagros forests at two scales

Background/Question/Methods

Bozin and Marakhil forest was designated protected area in 1999 to restore habitat for roe deer (Capreolus capreolus). It is located in Kermanshah province, Iran, and covers 23724 ha of semi-arid Zagros forests. Research in Europe has identified the importance of landscape structural characteristics, such as patch edge and contiguity, as important determinants of home-range and population sizes of roe deer. We analyzed landscape changes from 2001 to 2009 inside and outside of this protected area at two scales (Landsat: 30m grain and ~ 6750000 ha extent; and IKONOS/GeoEye: 1m grain and ~16500ha extent). Radiometric corrections and orthorectification were followed by the classification of images to five cover classes using Random Forest algorithm. All available structure metrics were calculated at patch and class level for forests using a 8 cell neighborhood rule.  A principal component analysis was applied on the structure metrics at patch level in order to avoid redundancy and select orthogonal metrics for comparison. The selected metrics were used to compare forest structure at general Zagros forests as well as protected and unprotected areas at the two time steps. The areas of the unprotected area located in Iraq were clipped out in order to eliminate the socioeconomic effects and the differences in national management policies.

Results/Conclusions

At the Landsat scale the selected orthogonal metrics with the highest loadings on the first two principal components (explaining 88% and 86% of the variation in 2001 and 2009, respectively) were patch area, shape index, and contiguity index. Class level metrics showed a 14.5 % forest loss and a 5.4 unit decrease in edge density for the general Zagros forest from 2001 to 2009. While there was no significant change in amount of forest in the protected area or the immediate area, there was an increase in patch connectivity inside of the protected area (p < .01). However, no significance difference was found for the value of the selected patch level metrics for general Zagros structure at the two time steps. These analyses at the first scale suggest that either 1) the park is functioning to protect important roe deer habitat characteristics (and the protection has increased the spatial connectedness of habitats), or 2) the increased habitat connectedness is due to the higher initial forest cover in the protected area. This requires analyses at the fine scale.