524 



Fishery Bulletin 104(4) 



A 



3J N- 



Figure 1 



(A) High-resolution bathymetry of Tanner Bank and map of the southern Cali- 

 fornia coast. Black lines represent ROV transect track lines in 2002 and yellow 

 lines represent track lines in 2004. Pink dots on track lines represent white 

 abalone (Haliotis sorenseni) sightings. Depth strata are distinguished by color 

 (see legend). CA=California; SCI = San Clemente Island; TB=Tanner Bank; 

 CB = Cortes Bank. (B) High-resolution bathymetry of Cortes Bank and San 

 Clemente Island. Black lines represent ROV transect track lines. Pink dots 

 on track lines represent white abalone tHaliotis sorenseni) sightings. Depth 

 strata are distinguished by color (see legend). 



for multibeam habitat classification (Kvitek et al.''). A 

 rugosity measure was used to evaluate the "texture" of 

 the seafloor by quantifying the surface area to planar 

 area. Rugosity calculations were done with ArcView 3.2 

 (ESRI, Redlands, CA) by using the "Surface Area and 

 Ratios from Elevation Grid , vers 1.2." (Jenness, 2002). 



^ Kvitek, R. G., P. J. lampietro, and E. Summers- 

 Morris. 2003. Integrated spatial data model tools set for 

 the auto-classification and delineation of species-specific 

 habitat maps from high-resolution, digital hydrographic 

 data. Report prepared for NOAA National Ocean Service 

 (NOS), p 1-74. NOAA Coastal Services Center, 2234 South 

 Hobson Ave., Charleston, SC 29405. 



Rugosity measures ranged from 1 to 4, and higher values 

 indicated more texture (i.e., rocky terrain). TPI was cal- 

 culated in ArcGIS, vers. 8.3 by a multistep process. For 

 each cell the focal mean of the digital elevation model 

 (DEM) was calculated to compute an average depth 

 value used to obtain a measure of relative elevation. 

 Relative elevations were classified into six TPI classes 

 to allow for visual representation of elevations. Bottom 

 types were noted during ROV surveys and compared to 

 bottom types determined by sonar data, and in all cases 

 results were confirmed. 



Principal component analysis (PCA) was used to ex- 

 amine potential relationships between habitat type and 



