Factors affecting land use and land cover change and fragmentation in selected protected areas in the Philippines

The establishment of the protected areas (PAs) is one of the most prominent biodiversity conservation policies in the Philippines. Some of the functions of the country's PA policy include reduction of deforestation, forest conservation, and decreasing of human pressure on natural resources. This study therefore aimed to assess the drivers of forest intactness and fragmentation in selected terrestrial PAs. Using the Management Effectiveness Tracking Tool (METT) scores, the study looked at the impact of drivers of land use and land cover (LULC) change, including economic development, population, road networks, cultivation, and PA management, on the percentage of forest cover and different fragmentation metrics in the PAs. The study conducted a Canonical Correlation Analysis (CCA) and Multiple Regression Analysis to identify the significant drivers of LULC change for each individual response variable. The CCA results show a relationship between the LULC and the driving factors, with the following as the most important predictors based on their canonical weights: cultivation within the PA (-0.775), the human development index (-0.543), and the regional gross domestic product (0.540). The results of the Multiple Regression Analysis support the results of the CCA. Cultivation inside the PA was the most important driving factor from the CCA and found to be the common significant variable for most of the dependent variables. Other significant factors were poverty index, population, and road density, which all could provide basis for the observed PA fragmentation. The R-squares of the best fit models range from 15% to 43%, and are sufficient in identifying drivers of change. The study also found that higher METT scores had a negative influence on the open forest and a positive effect on patch richness (SHDI). This means that as PA management improves, open forests tend to decrease while patch richness tends to increase. Furthermore, there is no data to support that MET scores have significant influence on the other response variables.