Raphael and others 



Chapter 18 



Landscape-level Analysis of Habitat in Washington 



Table 4 Landscape pattern indices output by DISPLA Y and used in the basin level analysis 



Pattern index 



Possible values 1 



Description 



Landscape diversity 



Landscape dominance 



Landscape contagion 



Number of different types 



Proportion of each type 

 in landscape 



Number of patches of each 

 type in landscape 



Mean patch size by type 



Perimeter/area fractal 

 dimension 



Grid based fractal 

 dimension 



0-o 



0-o 



- oo 



WDFW 2 = 9 

 WDNR 3 = 6 



0-1 



- total 

 landscape area 



1 .0 - 2.0 



1 .0 - 2.0 



Measures proportion of landscape in different types; = lowest 

 diversity (only 1 type); larger value indicates more diverse 

 landscape 



Extent to which 1 or few types dominate the landscape; as 

 value approaches 0, all types are present in equal proportions; 

 max. value depends on number of types in landscape 



Extent to which landscape is aggregated or clumped; as value 

 approaches 0, many small patches exist; max. value depends 

 on number of types in landscape 



Number of types possible in landscape, also termed 

 "patch richness" 



Percent of total area 



Count of patches by type 



Sum of patch area by type divided by total area 



Index of patch edge complexity, contrasts log (patch 

 perim.) with log (patch area) 



Index of patch edge complexity, calculated using a 

 grid-cell counting method 



1 Values reported are theoretical limits, not actual ranges. 



2 WDFW = Washington Department of Fish and Wildlife 



3 WDNR = Washington Department of Natural Resources 



DISPLAY and also calculates additional landscape-level 

 and patch-level indices (table 5). We attempted to use 

 FRAGSTATS for the basin level analysis, but these landscapes 

 were too large to process using this program. We tabulated 

 indices of pattern for each of the 261 circular areas and 

 compared site-level attributes among survey-status attributes 

 (those sites where murrelets were not detected, were detected, 

 or classified as occupied). 



We computed additional site-level variables using the 

 GIS to add other environmentally related measures to the 

 multivariate comparison of site-level pattern and occupancy 

 status. Distance to closest coastline (meters) was calculated 

 for each murrelet survey location using the NEAR function 

 in ARC/INFO. This represents a straight-line distance between 

 a survey location and the closest body of salt water. 



We identified patch size and type for each survey location 

 by recording the contiguous patch on the overall landscape 



(whether or not that patch was outside of the 0.5-mi radius 

 circle) directly underneath each survey point. The definition 

 of patch used here differs significantly from the concept of a 

 stand typically used by foresters. In this case, a patch is 

 defined in terms of the GIS map as each unique set of 

 contiguous cells of the same cover class type. Some of these 

 patches can be quite large (up to 25,000 hectares) and should 

 not be considered equivalent to typically defined stands in 

 forest management. Rather, these are areas defined by pixels 

 sharing the same class value. 



We determined survey-site elevations by overlaying the 

 map of survey locations on a digital elevation model and 

 interpolating the elevation at each point using GIS operations. 

 United States Geological Survey 1:250,000 scale digital 

 elevation models were used to derive an elevation surface 

 for the state of Washington. These elevation models are a 

 regular (grid) sample of elevations and have a vertical accuracy 



USDA Forest Service Gen. Tech. Rep. PSW-152. 1995. 



181 



