COS 126-7
Linking metrics of landscape pattern to hydrological process in a lotic wetland

Thursday, August 13, 2015: 3:40 PM
338, Baltimore Convention Center
Jing Yuan, School of Natural Resources and Environment, University of Florida, gainesville, FL
Matthew J. Cohen, School of Forest Resources and Conservation, University of Florida, Gainesville, FL
David Kaplan, Department of Environmental Engineering Sciences, University of Florida, Gainesville, FL
Subodh Acharya, Dept. of Environmental Engineering Sciences, University of Florida, gainesville, FL
Laurel L. Larsen, Geography, UC Berkeley, Berkeley, CA
Martha K. Nungesser, Everglades Division, South Florida Water Management District, West Palm Beach, FL
Background/Question/Methods

Strong reciprocal interactions exist between landscape patterns and ecological processes.  In wetlands, and particularly flowing systems, hydrology is the dominant abiotic driver of ecological processes and both controls, and is controlled, by vegetation presence and patterning.  We focus on binary patterning in the Everglades ridge-slough landscape, where longitudinally connected flow, principally in sloughs, is integral to landscape function.  Patterning controls discharge competence (i.e., the ability to route water) in this low-gradient peatland, with important feedbacks on hydroperiod and thus peat accretion and patch transitions.  Our goal was to quantitatively predict pattern effects on hydrologic connectivity and thus hydroperiod, a core objective of ecosystem restoration.  We evaluated three pattern metrics that vary in their hydrologic specificity.  The first, Landscape Discharge Competence (LDC), considers elongation and patch-type density that capture geostatistical landscape features, but does not explicitly account for hydrologic connectivity flowpaths.  A second, the Directional Connectivity Index (DCI), extracts both flow path and direction from a rasterized landscape based on graph theory.  The third, least flow cost (LFC), is based on a global spatial distance algorithm strongly analogous to landscape water routing, where ridges have higher flow cost than sloughs because of their elevation and vegetation structure.  Metrics were evaluated in comparison to hydroperiod estimated using a numerically intensive hydrologic model (SWIFT2D) for synthetic landscapes (4 anisotropy levels, 7 ridge density levels). 

Results/Conclusions

Both LFC and DCI were excellent predictors of hydroperiod (r2 = 0.92 and 0.88, respectively); LDC was less successful (r2 = 0.69), likely because landscape geostatistical attributes and flow connectivity are correlated but distinct properties. Later, fitted relationships between metrics and hydroperiod for synthetic landscapes were extrapolated to both contemporary and historical maps (1940 to 2004) to explore hydroperiod trends in space and time. Both LFC and DCI were useful for diagnosing how the modern landscape has, in some areas, reorganized in response to modified hydrology.  Metric simplicity and performance indicates potential to provide hydrologically explicit, computationally simple, and spatially independent predictions of landscape hydrology, and thus effectively measure of restoration performance.