Date of Graduation
Doctor of Philosophy in Environmental Dynamics (PhD)
Frank L. Farmer
Stephen K. Boss
Second Committee Member
Kimberly G. Smith
Third Committee Member
W. Fredrick Limp
This research critically examines extant data systems and their linkage of scientific research to policy and public education in east African highland forest conservation preserves. The research indicates that the current state of data and monitoring systems in the region imposes substantial limitations on the ability to manage these preserves. Major outcomes of those constraints manifest in unprecedented cross-boundary pressure such as encroachment and conversion of forest to agriculture, increased human-wildlife conflicts, and constrained relations among primary stakeholders. Current monitoring is biodiversity focused within preserves and fails to capture human dimensions in adjoining areas across preserve boundaries. This study proposes that expanding the scientific basis of data systems, monitoring and research to include human variables changes taking place across preserve boundaries allows better understanding, anticipation and response to environmental change.
An integrative conceptual framework and analytic model based on human ecological perspectives is developed. The empirical utility of the model that allows inclusion of human variables in preserve monitoring and policy decision making is demonstrated. Computer based geo-spatial analysis techniques are applied on remote-sensed imagery to quantify and determine patterns of land use/cover change over the three period 1975 to 2000. The mid-western area around Kipipiri and the north-western areas showed most activity during 1970s to 80s. During 1980s to 2000, the south west and north east areas showed most change.
Forest-Agriculture land conversion is modeled using spatial logistic regression techniques on data derived from remote imagery and human population census. Patterns of land conversion similar to those displayed in the land cover change analysis were achieved by applying the predictive model. The model achieves average 69.5% pixels predicted correctly for the three decadal periods, with ROC (Relative Operating Characteristic) of 0.75, 0.68 and 0.69 for 1975-1985, 1985-2000 and 1975 to year 2000 respectively. ROC values above 0.5 are indicative of some association between the predictor and dependent variables.
Linking remotely sensed ecological data, long term records from parks managers and demographic and socio-economic data from areas in and around biodiversity hotspots is not only timely but also offers novel insight and solutions to current management problems. The methodological tools and techniques are broadly applicable to other areas of conservation concern, and essential for addressing conservation objectives in the face of human development activities. They provide important insight for land use planning and development agendas, and enhance opportunities for interdisciplinary integration and institutional collaboration.
Njuguna, Peter K., "Integrating Human Variables in Cross-Boundary Monitoring of Aberdare Preserves in Kenya" (2012). Theses and Dissertations. 281.