FISHERY BULLETIN: VOL. 86, NO. 3 



Ainley^ and Szczepaniak^ in central California. The 

 number of harbor porpoise seen per kilometer of 

 transect was taken as an index of relative density 

 along each isobath. A simple descriptive model was 

 then constructed to give relative density as a func- 

 tion of water depth. 



Fifteen depth strata were used in abundance 

 estimation: 0-10, 10-20, 20-30, . . . , and 140-150 m. 

 The surface area within the strata was calculated 

 from digitized bathymetric data. Kelp beds were 

 assumed to be unsuitable as harbor porpoise habitat; 

 hence, kelp bed area was subtracted from the total 

 area within the 0-10 m stratum. Kelp bed areas for 

 the entire west coast were taken from Crandall 

 (1915). More recent estimates for limited areas in 

 central California are in good agreement with these 

 previous values (G. Van Blaricom^"). 



For each of 15 depth strata, the abundance of har- 

 bor porpoise was estimated as the product of their 

 density along the survey line (the 18 m isobath), the 

 density in that depth strata relative to that along 

 the survey line, the surface area included within that 

 depth strata, and the inverse of the estimated frac- 

 tion of trackline animals that were seen. Since 

 survey effort and harbor porpoise density both 

 varied geographically, abundance estimates were 

 made for each of 8 geographic regions (Fig. 2). 

 Areas within the depth strata were estimated from 

 NOAA bathymetric data. The estimate of total abun- 

 dance along the coast, A/'y, is therefore given by 



15 



^- = ^ I «^ .?. <^'  ^^.'> 



(4) 



where Dj = density of individuals observed on the 



transect line in the jth geographic 



strata, 

 4 = ratio of density in depth strata k to 



that on transect line (see Figure 4), 

 Aji^ = area in geographic region j' and depth 



strata k, and 

 F = the estimated fraction of trackline 



animals seen by the usual team of 5 



observers. 



'LaBarr, M. S., and D. G. Ainley. 1985. Depth distribution of 

 harbor porpoise off central California: A report of cruises in April 

 and May-June 1985. Report to U.S. National Marine Fisheries 

 Service, Northwest and Alaska Fishery Center, 7600 Sand Point 

 Way N.E., Seattle, WA. Contract No. 41-USC252. 



'Szczepaniak, I. D. 1987. Abundance and distribution of har- 

 bor porpoise (Phocoena phocoena) in the Gulf of the Farallones 

 National Marine Sanctuary. Contract report prepared for 

 National Park Service, Point Reyes National Seashore, Point 

 Reyes, CA 94956. 



'"G. Van Blaricom, U.S. Fish and Wildlife Service, University 

 of California, Santa Cruz, CA 93106, pers. commun. August 1986. 



Equation (4) was applied independently to the dif- 

 ferent surveys and, within surveys, to different sea 

 state strata. When combining estimates from differ- 

 ent sea states or different cruises, abundance was 

 calculated as the mean of the densities in each of 

 the stratum, weighted by the length of the transect 

 line within that stratum. 



In estimating standard error for total abundance, 

 variances of products were calculated using the 

 Goodman (1960) product variance formula, and vari- 

 ances of ratios were estimated using a Taylor 

 approximation (Yates 1953, p. 198). Area was 

 assumed to be known without error. Statistical error 

 in the indices of abundance for the depth strata could 

 not be estimated given the paucity of available in- 

 formation. To account for uncertainty in the model 

 of depth distribution, three versions of the model 

 are proposed to span a range of possibilities. 



RESULTS 



On the four surveys, 852 groups of harbor por- 

 poise were sighted (an estimated 1,818 individuals). 

 A distance of 6,590 km was surveyed during 56 

 days. The number of sightings per kilometer sur- 

 veyed varied geographically and these geographic 

 patterns appeared to change appreciably between 

 cruises (Fig. 1). 



Sighting Distributions 



The number of sightings on the inshore and off- 

 shore sides of the vessels were approximately equiv- 

 alent (383 and 392, respectively). The cumulative 

 distributions of perpendicular sighting distances 

 were not significantly different for these two sides 

 (P = 0.06). Therefore, sighting distributions were 

 assumed to be symmetrically distributed about the 

 trackline, and the distributions of perpendicular 

 sighting distances from both sides of the vessel were 

 pooled for subsequent analyses. 



The distributions of perpendicular sighting dis- 

 tances for the first three surveys were significant- 

 ly different from one another (P < 0.01 for all). This 

 was probably the result of the modifications in 

 survey methods between these cruises. Surveys 3 

 and 4 used the same methods, and sighting distribu- 

 tions were not significantly different (P = 0.39). 

 Given that changes in methods result in differences 

 in sighting distributions, all surveys were treated 

 separately in subsequent analyses. 



Distributions of perpendicular distance were not 

 significantly different between individuals sighted 



422 



