Modeling Wetland Plant Metrics to Improve the Performance of Vegetation-Based Indices of Biological Integrity
Location
Eccles Conference Center Auditorium
Event Website
http://water.usu.edu
Start Date
3-31-2015 1:50 AM
End Date
3-31-2015 2:00 AM
Description
Wetland managers need accurate and precise indices to assess wetland biological condition. Modeling techniques have recently been used in stream ecosystems to improve the precision and sensitivity of indices of biological integrity (IBIs). We used Ohio EPA’s wetland vegetation data to (1) assess metric responsiveness to natural environmental gradients and (2) determine if we could incorporate data from three different types of wetlands (emergent, forested, and shrub dominated) into one index that was more precise and sensitive than Ohio EPA’s existing wetland vegetation IBI (VIBI). We used Random Forests to model how 21 VIBI metrics varied across 82 reference sites in response to naturally occurring environmental variables. Modeling explained 14-52% of reference site metric values. We then selected eight of these metrics for inclusion in the modeled VIBI (MVIBI). We measured index precision as the coefficient of variation (CV) of reference site scores. The MVIBI was nearly twice as precise as the VIBI (CVs = 0.13 and 0.21, respectively). We examined index sensitivity by comparing index values at 170 test sites against the 10th percentile of reference site values. The two versions of the VIBI scored 142 sites similarly but disagreed on 28 sites indicating that modeling altered site specific estimates of the reference condition and hence site-specific inferences. These methods could be used to develop wetland IBIs applicable on regional and national scales.
Modeling Wetland Plant Metrics to Improve the Performance of Vegetation-Based Indices of Biological Integrity
Eccles Conference Center Auditorium
Wetland managers need accurate and precise indices to assess wetland biological condition. Modeling techniques have recently been used in stream ecosystems to improve the precision and sensitivity of indices of biological integrity (IBIs). We used Ohio EPA’s wetland vegetation data to (1) assess metric responsiveness to natural environmental gradients and (2) determine if we could incorporate data from three different types of wetlands (emergent, forested, and shrub dominated) into one index that was more precise and sensitive than Ohio EPA’s existing wetland vegetation IBI (VIBI). We used Random Forests to model how 21 VIBI metrics varied across 82 reference sites in response to naturally occurring environmental variables. Modeling explained 14-52% of reference site metric values. We then selected eight of these metrics for inclusion in the modeled VIBI (MVIBI). We measured index precision as the coefficient of variation (CV) of reference site scores. The MVIBI was nearly twice as precise as the VIBI (CVs = 0.13 and 0.21, respectively). We examined index sensitivity by comparing index values at 170 test sites against the 10th percentile of reference site values. The two versions of the VIBI scored 142 sites similarly but disagreed on 28 sites indicating that modeling altered site specific estimates of the reference condition and hence site-specific inferences. These methods could be used to develop wetland IBIs applicable on regional and national scales.
https://digitalcommons.usu.edu/runoff/2015/2015Posters/4