Land cover change (LCC) has significant effects on the global ecosystem diversity and function. This topic has received increasing attention due, in part, to its relationship with climate change, and the availability of remotely-sensed imagery that is used to monitor LCC. However, studies analysing the factors that drive LCC at large spatial scales and over long temporal scales are uncommon. This study aimed to identify the factors driving long-term (44 years, 1972-2016) national level LCC in Zambia. Two analyses were conducted, with the first considering factors that led to any LCC. The second scenario identified factors associated with changes from forests to other land covers, and the reversion to forests from non-forested covers. Candidate factors considered in both analyses include accessibility, proximity, topography, climate, conservation and socioeconomics. A classification tree (CT) approach was used to relate the explanatory candidate factors to LCC. The results showed that the CT models predicted LCC with accuracies ranging from 71 to 85%. The first analysis showed that the major factors determining LCC were percentage of area under agriculture, distance to water bodies, change in crop yield, mean temperature and elevation. Meanwhile, the second analysis showed that primary, secondary and plantation forest cover losses were mainly influenced by human population density, crop yield per hectare and mean crop yield, respectively. Protection status was the most important factor for forest reversion and recovery, while a variety of factors including distance to the railway, elevation and total precipitation also influenced forest reversion and recovery. The findings from this study provide insights into the factors that influence LCC and are important for developing effective policies to reduce the negative impacts of LCC and to promote forest reversion and recovery through effective management of protected areas. While this study focused on factors associated with historical LCC, the findings will also help to predict and understand future LCC scenarios.
Keywords: Anthropocene; Climate change; Decision tree; LULC; Redd+; Remote sensing.
Copyright © 2019 Elsevier B.V. All rights reserved.