Landscape fragmentation and habitat structure across the national parks
Description
Landscape structure shapes the capacity of protected areas to sustain viable wildlife populations, yet patch-level landscape metrics have not been reported for the national parks of Punjab, Pakistan. We compiled a vector land-use/land-cover dataset of 537 habitat patches in eight habitat classes across all eight national parks (2,879.0 km²) and calculated FRAGSTATS-consistent metrics: number of patches, patch density (PD), mean patch size, patch-size coefficient of variation, largest patch index (LPI), and Shannon’s diversity (SHDI) and evenness indices. Bare rock/gravel (52.0%) and subtropical broad-leaved forest (16.5%) dominated network composition. PD varied almost six-fold, from 8.3 patches per 100 km² at Lal Suhanra to 48.7 at Kheri Murat, and LPI ranged from 23.2% (Murree–Kahuta–Kotli Sattian) to 89.6% (Salt Range). SHDI was highest at Murree–Kahuta–Kotli Sattian (1.30; eight classes) and lowest at Salt Range (0.36), where one class covered 90.7% of the park. Landscape dominance and compositional diversity were strongly inversely related (Spearman ρ = −0.93), whereas patch density was unrelated to diversity (ρ = 0.10) and was higher in smaller parks (ρ = −0.81). Kheri Murat and Chinji combined the highest patch densities with moderate dominance, making them the most internally fragmented reserves, while Tilla Jogian and Pabbi–Rasul showed bimodal structures in which one dominant patch coexisted with many small fragments. These contrasting configurations indicate that the parks require differentiated management: connectivity restoration at Kheri Murat and Chinji, protection of the heterogeneous mosaic at Murree–Kahuta–Kotli Sattian, and evaluation of habitat homogeneity at Salt Range.
Files
Steps to reproduce
Class- and landscape-level metrics were calculated following standard FRAGSTATS definitions (McGarigal, 2015; McGarigal & Marks, 1995), adapted to vector patch data. At the landscape (whole-park) level we calculated the number of patches (NP); patch density (PD = NP/total area × 100, patches per 100 km²); mean patch size (MPS = total area/NP, km²); the patch-size coefficient of variation (PSCV = standard deviation of patch area/mean patch area × 100, %), which quantifies size heterogeneity among patches; and the largest patch index (LPI = area of the largest patch/total area × 100, %), which quantifies the degree to which a landscape is dominated by a single contiguous patch. Because MPS is the reciprocal of PD at the landscape level (MPS = 100/PD), the two metrics were interpreted as a single fragmentation axis. At the class level, class area, patch count, mean patch size, LPI and the percentage of landscape occupied by each class (PLAND) were calculated for each habitat class within each park. Compositional diversity was quantified using Shannon’s diversity index (SHDI = −Σ pi ln pi, where pi is the proportion of park area occupied by class i) and Shannon’s evenness index (SHEI = SHDI/ln m, where m is the number of classes present in the park) (McGarigal & Marks, 1995; Shannon & Weaver, 1949). SHDI increases with both the number of classes and the evenness of their areas, whereas SHEI scales diversity from 0 to 1 independently of class number, allowing comparison among parks with different numbers of classes. Edge and shape-complexity metrics (edge density, landscape shape index and mean shape index) were not calculated because the perimeter attributes in the source geodatabase could not be validated against patch area. All reported metrics are therefore restricted to patch number, area, size distribution, dominance and compositional diversity, each computed directly from validated patch areas.