Description: The field portion of the mapping project consisted of two primary elements: vegetation classification and field-based mapping. Due to the relatively small size of the Reserve, the vegetation classification portion of the project was accomplished concurrently with field mapping, with sampling located within and referenced to the mapped polygons of vegetation and other features. We conducted an initial scouting trip by car and foot to quickly survey the range of variability in the Reserve. We then revised of the draft list of potential vegetation associations (Rocchio and Crawford 2009) and identified additional vegetation zones areas that were not currently represented in the draft list, or that could prove challenging to classify due to reasons such as access. When the classification was complete, we developed a physiognomic and floristic key for identifying plant associations and/or alliances in the field. Summary descriptions were written for new provisional plant associations and alliances, and descriptions for existing plant associations were revised to reflect how they are expressed at the Reserve. NPS staff tested the key and provided feedback prior to the accuracy assessment field work. NVC Hierarchy 30 provided the reference for all plant association and hierarchy names. WNHP personnel reviewed these names to ensure standardization. Nomenclature for plant associations and alliances not yet incorporated into Hierarchy 30 were borrowed from their source documents. While many of these associations have been recognized by NatureServe, only a few have been entered into their online database at the time of this project. We attributed every polygon we visited with association or map class label using ArcMap 10.2 (ESRI 2013). If it was uncertain whether the association call was appropriate for the full extent of a large polygon (this happened primarily in forested areas), an observation point was placed to denote the association call for only the sampled portion of the polygon. The polygon was later attributed when the full extent had been visited and confirmed, or if all surrounding polygons ended up with the same association. Some polygons were never visited, either because they lay in unobservable portions of private property or on steep seaside cliffs. In most cases, a justifiable association call could be made for these polygons based on imagery interpretation, the DEM, and the association designations given to nearby observable vegetation patches. Creating the vegetation map was an iterative process. Polygons with the same association calls, alternate association calls, and notes were merged to form single polygons. As the mapping proceeded, associations that were not found at the Reserve were deleted from the draft list of associations and new associations were added. Priority areas that needed return field visits were identified during the digitizing process. Highways and major byways were digitized as polygons, while minor roads and driveways were merged with nearby residential zones. Polygons were checked against updated 2013 NAIP imagery that became available during the course of the project. We updated polygons that had changed between the 2011 and 2013 imagery, usually due to new housing developments. To finalize the appearance of the polygon boundaries, we used a PAEK (Polynomial Approximation with Exponential Kernel) algorithm (Kraak et al. 1996) with a 10 m tolerance to smooth the rough edges of the initial polygons. During the vegetation classification of Ebey’s Landing National Historical Reserve, we inventoried 51 alliances or associations, of which 31% were forests and woodland communities, 57% were shrublands and herbaceous types, and 10 % were agricultural or developed types. The long and varied history of human land use at Ebey’s Landing National Historical Reserve has resulted in ruderal classification for nearly one third of the forest, woodland, shrubland, and herbaceous types. Ruderal types are dominated by novel combinations of native and non-native vegetation due to human disturbance. The Reserve is home to seventeen plant associations considered critically imperiled or imperiled in the state of Washington, or globally. These communities include dry forests or woodlands restricted to a limited range in the rain shadow zone of the Puget Sound basin, remnant and restoration prairie types, rare wetlands, a salt marsh type, and a fen. To check the final polygon attributes we followed a standard procedure: 100 random points were created within the Reserve boundary. The polygon in which each point fell (as well as neighboring polygons) was checked against annotations on the field map and checked that the polygon boundaries reflected the actual landscape. Each vegetation association was highlighted in ArcMap and observed at a Reserve-wide scale to ensure that the extent and specific locations of each association agreed with the field surveyor’s notes. Every tenth line on the field observation data sheets was checked against the final digitizing feature class to ensure that the described polygon had been placed in and merged with the appropriate vegetation association.Map accuracy assessment focused on the natural vegetation types. Overall map accuracy is 81.1%. Twenty-one of the 29 classes assessed met the threshold of 80% producer’s accuracy. Twenty of the 29 classes assessed met the threshold of 80% user’s accuracy. Lower accuracy resulted from a variety of challenges, including low sample sizes, similar dominant and diagnostic species leading to confusion among associations in different alliances, patchy understory shrub distributions, and canopy tree species overlooked during the mapping but detected during the assessment.
Copyright Text: North Coast Cascades Network (NCCN) of the National Park Service (NPS)