Uncertainties in the geographic representation of landscape connectivity metrics for area-based conservation

Keywords: sensitivity analysis, uncertainty, landscape metrics
Abstract Type: Paper Abstract

Abstract

Uncertainty is ubiquitous in applications of Geographic Information Science (GIS) and is essential to accurate understanding of the subject being analyzed. Despite its importance, uncertainties in landscape metrics, which are used to measure land cover spatial patterns, have been rarely examined. With emerging global conservation targets of connectivity and a pressing need for countries to measure and report progress toward those targets, it is critical to identify and understand the multiple aspects of uncertainties when using landscape connectivity metrics. From existing literature and previous research, we identify three layers of uncertainties in the geographic representation, including uncertainties caused by scale and data aggregation, data collection and use, and sensitivities to changing spatial patterns. To address these issues, we review 15 most used landscape connectivity metrics and test their sensitivity to changing spatial patterns. We simulate possible spatial pattern changes using the protected area network of the State of California, US, and compare results to area-based conservation targets using statistical methods. Our results verify the general consistencies in metric performance and identify metric differences in their sensitivity to changing spatial patterns. Our findings provide insights into novel ways to explain and account for the uncertainties posed by such studies. Comparisons of metrics can improve understanding of uncertainties in landscape connectivity analyses to promote better interpretation and help provide guidance on metric selection to measure against area-based conservation targets.