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



569 



100 200 300 400 500 

 Depth (m) 



1000 2000 



Alongstiore distance (km) 



a 

 o 



0.6- 

 0.2- 

 -0.2- 

 -0.6- 



140 



180 220 260 



Julian day 



'84 '87 '90 '93 '96 

 Year 



0.4 

 0.0 

 -0.4 

 -0.8 



160 172 



Gear type 





Figure 6 



Trends in total CPUE by depth, alongshore distance, Julian day, year, 

 and gear type. For details see Figure 2 



o 0-100 m • 100-200 m 



X 200-300 m A 300-500 m 



-06 -0.4 -0.2 0.2 04 06 



Axis 1 



Tt D Shumagin o Cliirikof • Kodiak 



X Yakutat A Southeast 



A A A 



• .• 2'*>^''*^A 





Jo O A 



O 



•0 6 -0.4 -0 2 0.2 04 6 



Axis 1 



Figure 7 



Plots of the first two axes from an NMDS (mul- 

 tidimensional scaling) ordination of 48 strata 

 sampled during each of fivfe years. Distances 

 between two points (strata) in the ordination dia- 

 gram approximately reflect their dissimilarity in 

 terms of species composition. Symbols indicate 

 depth strata (A) and geographic area (B). 



sions). These findings suggest that species composition 

 within the summer remains relatively stable over the 

 14—16 week survey period. 



To examine effects of temperature on species composi- 

 tion we used only those hauls from the 1993 and 1996 

 data for which temperature measurements were avail- 

 able (n=726 and «=716 respectively, Table 1). Results 

 for other years are not reported because of the relatively 

 small number of temperature measurements. The avail- 

 able temperature data for earlier years were often local- 

 ized in certain strata, whereas large areas had few if any 

 measurements. The lack of contrast did not allow definite 

 conclusions with regards to the temperature effect. For 

 1993 and 1996, we repeated the GAM analysis for the five 

 axes of species composition with temperature included. 

 Temperature effects were apparently large. However, wide 

 confidence intervals and inconsistent patterns between 

 years suggest that the apparent effects were at least in 



part due to confounding of temperature with alongshore 

 distance and depth. Therefore we first adjusted for the 

 effects of depth and alongshore distance using the re- 

 gression models summarized in Table 6. Temperature 

 explained a small but significant portion of the remaining 

 residual variation in three of the ordination axes in both 

 1993 and 1996 (Pseudo-r^ values <0.1). 



To identify which species were most strongly related 

 to the major gradients we computed rank correlations 

 between each species and the three axes of species compo- 

 sition based on the NMDS ordination of strata averages. 

 We considered all species that had a positive or negative 

 rank correlation of at least 0.4 with one of the axes for 

 further examination, based on a visual inspection of scat- 

 terplots. Species associated with the first two axes were 

 clearly separated along gradients of depth and alongshore 

 distance (Figs. 8 and 9). The first axis was positively cor- 

 related with a number of deep-water species that were 



