97th ESA Annual Meeting (August 5 -- 10, 2012)

PS 8-110 - Classification and mapping of wetland plant communities in the Burren National Park, West Ireland

Monday, August 6, 2012
Exhibit Hall, Oregon Convention Center
Daniel A. Sarr1, Micheline Sheehy-Skeffington2, Lorin Groshong1 and John Curtin2, (1)Klamath Network-National Park Service, Ashland, OR, (2)Botany & Plant Science, National University of Ireland, Galway, Galway, Ireland
Background/Question/Methods

The Burren National Park, west Ireland, encompasses extensive and globally significant wetlands. At the time of this study no formal wetland map or inventory existed to inform park management and interpretive programs.  After meeting with park staff, we designed a rapid wetland inventory that could be co-analyzed with existing imagery to develop a wetlands map characterizing the dominant wetland plant communities of the park. In spring 2008, we sampled wetland vegetation in all major wetland basins in the park. At each wetland, we sampled vegetation composition, soils, and hydrological data at 2 x 2 relevé samples along 32 centripetal transects from the edge to interior. We classified wetland vegetation data using hierarchical cluster analysis and optimized cluster resolution using Indicator Species Analysis in PC-Ord and the SIMPROF function in Primer-E.

We conducted a parallel image classification on mosaicked 1m resolution Irish Odnance Survey orthophotos provided by park staff.  Training pixels were chosen on the imagery for each possible wetland community type at each corresponding field site. Then we ran a supervised maximum likelihood classification in ERDAS Imagine to generate mapping units across the wetland landscape, cleaning pixel scatter with a 5x5 to 7x7 pixel scanning window using the Neighborhood tool in ERDAS GIS Analyst.

Results/Conclusions

The Indicator Species and the SIMPROF analyses guided our final selection of 10 community types from the vegetation data.  These 10 vegetation groups corresponded to 9 recognizable mapping units, which along with open water, formed 10 distinct mapping units. These groups generally formed a gradient from temporally flooded fen and shrubland community types at the upper edges of the wetlands to seasonally and semipermanently flooded turlough, fen, and reedswamp types deeper in the wetland basins. Floristic and environmental profiles were developed for each community type. Of the vegetated wetland area (ca. 433 ha), the most extensive community types included black bogrush (Schoenus nigricans) fen (34% ), sawsedge (Cladium mariscus) fen (9.7%), and and turlough floor meadow (8.6%). Least extensive types included raised sweet gale – crossleaf heaf (Myrica gale-Erica tetralix) bog (2%) and bulrush-giant reed (Schoenoplectus lacustris-Phragmites australis) reedwamp (1%). We estimate that the minimum recognizable mapping unit is approximately 0.01 ha (10m x 10m).

Overall, the color, texture, and geomorphic features in the imagery helped to refine and generalize the floristically-developed community types, illustrating that complementary biophysical factors contribute to meaningful wetland typologies for national park management and interpretation.