Mueter and Norcross: Spatial and temporal patterns in the demersal fish community off Alaska 



571 



Table 6 



Ki'sults from NMDS l multidimensional scaling) ordinations of species CPUEs in five dimensions (axes 1-5) by year, and GAM 

 (generalized additive models! results for regressions of axis scores on depth, alongshore distance, and Julian day. Variance indicates 

 proportion of overall variation accounted for by each axis. Stress indicates Kruskal's stress values for each ordination. Numbers 

 for the full model indicate pseudo-;- values for the best model fit ( 1 - deviance of the best fitting model/deviance of the null model). 

 Numbers following individual variables indicate pseudo-coefficient of partial determination, computed from a reduced model that 

 excluded the variable (= 1 - deviance of best model / deviance of reduced model). Only coefficients of determination exceeding 0.1 

 are shown. Coefficient of determination for Julian day did not exceed 0.1 in any year or for any axis. AD = alongshore distances. 



Year 



A.xis I 



Axis ; 



Axi.s 3 



Axi.s 4 



Axis 5 



Stress 



1>)S4 



1487 



I QMO 



1993 



1946 



Variance 

 I'ull model 

 Depth 

 AD 



Variance 

 Full model 

 Depth 

 AD 



Variance 

 Full model 

 Depth 

 AD 



Variance 

 Full model 

 Depth 

 AD 



Variance 

 Full model 

 Depth 



0,102 



O.IOO 



0.096 



0.098 



0.099 



Kodiak, and Yakutat areas ( Figs. 10 and 11). Another elas- 

 mobranch (Pacific sleeper shark, Somniosus pacificus), 

 two osmerids (capeHn, Mallotus villosus, and eulachon, 

 Thaleichthys pacificus), three flatfish species (Dover sole, 

 rex sole, and arrowtooth flounder), and one rockfish spe- 

 cies (Pacific Ocean perch, Sebastes alutus) increased sig- 

 nificantly in one or more areas and depth strata (Fig. 10). 

 The frequency of occurrence of capelin in the trawl survey 

 increased significantly, whereas CPUE-where-present did 

 not change or. in some cases, decreased significantly. Clos- 

 er examination revealed that the increase in frequency 

 of occurrence occurred primarily between 1993 and 1996 

 (Fig. 12). Capelin were caught in 14% of all hauls in 1996, 

 compared to only T^, in 1993. Changes in the CPUE- 

 where-present of three flatfish species differed among 

 species. Although Dover sole increased primarily in the 

 eastern GOA (Yakutat and Southeast), rex sole increased 

 most strongly in the Chirikof area (Fig. 10). In contrast, 

 arrowtooth flounder increased in all areas, but only in the 

 0-100 m depth stratum. Although all three flatfish species 

 increased significantly over time, estimates of their gulf- 

 wide average CPUEs were highest in 1990 and declined 

 from 1990 to 1996 (Table 2). Both the frequency of occur- 

 rence and CPUE-where-present of Pacific Ocean perch 



increased strongly in the Chirikof area, and frequency of 

 occurrence decreased in the Yakutat area. 



CPUE-where-present of the four species that were nega- 

 tively correlated with the time index did not decrease sig- 

 nificantly, and, in the case of rock sole (Lepidopsetta spp.), 

 even increased in some areas (Fig. 13). However, frequency 

 of occurrence decreased significantly for three of the spe- 

 cies, particularly for bigmouth sculpin (Hemitripterus 

 holini. Fig. 13). 



Discussion 



Research survey data provide standardized indices of rela- 

 tive abundance that often track abundance trends more 

 accurately than methods that use commercial catch data 

 (Pennington and Stromme, 1998). However, results need 

 to be interpreted with caution because the catchability 

 of species may change over time owing to changes in 

 survey gear, vessel type, spatial distribution, or the length 

 composition of a population. In our study we were par- 

 ticularly concerned that differences in gear type (Table 1) 

 might bias our results. We opted to include gear type 

 directly into the analysis as a "nuisance" variable, instead 



