FISHERY BULLETIN: VOL. 83, NO. 2 



137 m) at each of seven transects (Fig. 1) with an 

 otter trawl towed along a depth isobath at 1.1 m/s 

 for 10 min. A 7.3 m (headrope length) trawl was 

 used from 1971 to 1974, when it was replaced by a 

 7.6 m net; a 1.25 cm mesh cod end liner was used in 

 both nets. From 1971 through 1978, two 

 samples — one between April and June and one 

 between October and December — were collected 

 annually at each depth; occasionally, additional 

 trawls were made and these were included in the 

 analyses. Quarterly trawling began in 1979 and 

 has continued to the present. Sampling was dis- 

 continued at transects T2, T3, and T6 in 1977. 



Trawl catches were sorted by species and pro- 

 cessed on board ship. The standard length of each 

 individual was determined on a measuring board 

 (BSL = board standard length). External 

 abnormalities (e.g., fin erosion, parasites, tumors) 

 were recorded along with length. 



Trends in the incidence of fin erosion per 10-min 

 trawl were determined from linear regressions of 

 the proportion ip) offish with the disease (trans- 

 formed to arcsin \ p) versus time [numbered in 

 consecutive months from 1971 through 1982 (i.e., 1, 

 2, 3, ..., 143, 144)]. Collections where only one 

 individual of the species of interest was caught 

 were dropped from the analysis because the result- 

 ing transformed datum (either or 90) often had a 

 large effect on the residual sum of squares. Trends 

 in the total catch per 10-min trawl were deter- 

 mined from linear regressions of numbers 

 [transformed to logio (x -I- D] versus time. The null 

 hypothesis that the regression function (slope) 

 was equal to zero was tested with a ^-test. The null 

 hypothesis of equality of the regression functions 



was tested by analysis of covariance (ANCOVA). If 

 the null hypothesis was not accepted, significant 

 differences were detected with a Newman-Keuls 

 multiple range test (Zar 1974). Calico rockfish, 

 Sebastes dalli, rex sole, Glyptocephalus zachirus, 

 and Dover sole. Microstomas pacificus, accounted 

 for 96.6% of the fish with fin erosion (Table 1) so the 

 trend analysis was performed on each species in- 

 dividually. Less than 1% of the fin eroded fish came 

 from the 23 m stations; these stations were then 

 dropped from the analyses. 



The quarterly trawl data (1979-82) from tran- 

 sects T4 and T5 were examined for seasonal trends 

 in the total catch of Dover sole, the number with 

 fin erosion, and the proportion with fin erosion 

 with the following model: 



Y,=fiT^,S,) + e^ 



where Y = observed abundance (or proportion) in 

 period t,T^ = trend factor of time series in period 

 t, S^ = seasonal factor of time series in period t, 

 f = function relating observed abundance (or pro- 

 portion) to the trend and seasonal components, 

 and e^ = irregular factor of time series in period t 

 (Bowerman and O'Connell 1979). Multiplicative 

 and additive models were fitted to the quarterly 

 trawl data after transformation [logio (x -I- 1) and 

 arcsin \ p ]. Multiplicative models gave the best fit 

 (lowest residual sum of squares) so only those re- 

 sults are presented. After determining the quar- 

 terly trends, the original transformed data were 

 "deseasonalize'd" by dividing each value by the 

 corresponding seasonal factor. Trends in the de- 



X 



.'b'^ 



<^ 



% 



^o. 





Figure l. — Location of sampling transects on the Palos Verdes shelf. Three depths (23, 61, 

 and 137 m) were sampled at each transect. Joint Water Pollution Control Project outfalls are 

 located between transects T4 and T5. Net current flow is northwest. 



196 



