Abstract. — Aerial surveys de- 

 signed to detect trends in the abun- 

 dance of harbor porpoise Phocoena 

 phocoena were conducted each au- 

 tumn, 1986 through 1990. The num- 

 ber of porpoise seen per kilometer of 

 survey effort was used as an index 

 of abundance. Based on these sur- 

 veys, an analysis of covariance was 

 used to model porpoise abundance. 

 Year was treated as a covariate, and 

 factors which affected sighting rates 

 were included as categorical vari- 

 ables. No significant changes were 

 seen in the abundance of porpoise 

 over the five survey years. Monte 

 Carlo simulations were performed to 

 determine the power of the ANCO VA 

 to detect trends in abundance. We 

 conclude that the ability to detect 

 trends is poor if traditional levels of 

 statistical significance (o = 0.05) are 

 used. A larger a-error may be appro- 

 priate in the management context of 

 this species and increases the power 

 to detect trends. Additional survey 

 years similarly improve the power to 

 detect trends. Based on the results 

 of the simulations, we suggest that 

 power should be defined to include 

 only the detection of the correct 

 trend when two-tailed tests are 

 employed. 



Detecting Trends in Harbor Porpoise 

 Abundance from Aerial Surveys 

 Using Analysis of Covariance 



Karin A. Forney 



La Jolla Laboratory, Southwest Fisheries Science Center 



National Marine Fisheries Service, NOAA, P.O. Box 271, La Jolla, California 92038 



Doyle A. Hanan 



California Department of Fish and Game, c/o Southwest Fisheries Science Center 

 La Jolla Laboratory, National Marine Fisheries Service, NOAA 

 P.O. Box 271, La Jolla, California 92038 



Jay Barlow 



La Jolla Laboratory, Southwest Fisheries Science Center 



National Marine Fisheries Service, NOAA. P.O. Box 271, La Jolla, California 92038 



Manuscript accepted 25 March 1991. 

 Fishery Bulletin, U.S. 89:367-377 (1991). 



Harbor porpoise Phocoena phocoena 

 are caught incidentally during halibut 

 fishing with gillnets along the central 

 California coast (Diamond and Hanan 

 1986; Hanan et al. 1986, 1987; Bar- 

 low 1987; Barlow and Hanan 1990). 

 To assess the potential impact of this 

 fishery mortality, ship and aerial sur- 

 veys have been used to estimate the 

 abundance of harbor porpoise along 

 the coast of California, Oregon, and 

 Washington (Barlow 1988, Barlow et 

 al. 1988). These authors showed that 

 although aircraft can be used to sur- 

 vey a large area very quickly, abun- 

 dance estimates from aerial surveys 

 must be multiplied by a very large 

 and uncertain correction factor to 

 account for the majority of animals 

 that will be underwater at any given 

 instant. For this reason, ship surveys 

 were concluded to be preferable for 

 estimating absolute porpoise abun- 

 dance. 



The requirements are, however, 

 less stringent if the only goal is to 

 detect trends in the abundance of 

 porpoise over time, rather than de- 

 termining absolute abundance. The 

 ability of aircraft to cover great dis- 

 tances relatively quickly and inexpen- 

 sively makes them a logical platform 

 for such surveys. If the fraction of 



animals detected from the air does 

 not change over time, the correction 

 factor becomes irrelevant, and in- 

 dices of relative abundance can be 

 used in place of absolute abundance 

 measures. 



We describe a series of five aerial 

 surveys for harbor porpoise con- 

 ducted in central California during 

 autumn of 1986, 1987, 1988, 1989, 

 and 1990. These surveys were de- 

 signed specifically to detect changes 

 in porpoise abundance. We used twin- 

 engine aircraft to fly predetermined 

 transect lines which zigzagged up the 

 coast between Point Conception and 

 the mouth of the Russian River (Fig. 

 1). Line transect methods were used 

 with one observer on each side of the 

 aircraft and a belly-port observer. A 

 fourth person recorded information 

 pertaining to sightings of porpoises 

 and sighting conditions. Each year 

 within the survey period, the transect 

 lines were repeated 3-7 times, de- 

 pending on weather conditions. 



The number of porpoise seen per 

 kilometer of search effort was used 

 as a measure of relative abundance. 

 A stepwise analysis of covariance 

 procedure (ANCOVA) with year as 

 the covariate was used to identify the 

 best model describing porpoise seen 



367 



