Hamilton and Konar: Implications of substrate complexity and kelp variability for Alaskan nearshore fish communities 191 



Surveys of kelp and fish 



For surveys of kelp, randomly placed 0.25-m2 quadrats 

 {n = 10) were examined per 120-m- transect. All under- 

 story kelps in each quadrat were counted and identified 

 to species. Because all understory kelp species were 

 structurally similar (in size and overall shape), they 

 were grouped as "annual" iCostaria costata, Cymathere 

 triplicata, and Laminaria saccharina,) or "perennial" 

 {Agarum clathratum, L. bongardiana, and L. yezoensis) 

 understory for statistical analyses. All data collected in 

 May 2002 were omitted from the analyses involving kelp 

 because understory data for that month were incomplete. 

 Analyses were conducted on the average understory 

 kelp densities per transect to enable comparison with 

 the relatively sparse densities of the canopy kelp and 

 fish communities. Because Nereocystis was relatively 

 rare, all individuals were counted within each 120-m- 

 band transect. 



For fish surveys, all fishes observed within each tran- 

 sect and within one meter of the bottom (30 mx4 mx 

 1 m=120 m-^) were enumerated and identified to species 

 whenever possible. Because few fishes were observed, 

 the three most abundant families (Hexagrammidae, 

 Scorpaenidae. and Gadidae) were analyzed by family 

 group. All other fishes were rarely observed and were 

 grouped as "other fishes" for the analyses. 



Statistical analyses 



Statistical analyses were performed by using multivari- 

 ate approaches and linear models with STATISTICA vers. 

 6 (StatSoft, Tulsa, OKI. Cluster analyses were used to 

 examine site variability in the kelp and fish communities 

 and how this variability relates to structural complex- 

 ity. Averages of all data were calculated by site across 

 month and year for the ordination analyses. Kelp and 

 fish densities were considered by species with the physi- 

 cal variables of rugosity, substrate size, and verticality. 

 The Bray and Curtis dissimilarity coefficient (Bray and 

 Curtis, 1957) was used and the Euclidean distance was 

 calculated for physical variables, kelp, and fish. Water 

 temperature did not vary among sites within months 

 and was not used in our analysis. A one-way ANOVA 

 was used for temporal variation of water temperature. 

 Partial correlation analysis (with Pearson's correlation 

 coefficient, r) was performed between kelp groups (based 

 on average annual understory, perennial understory, 

 and canopy kelp densities [no./120 m'-]) and physical 

 habitat data (rugosity, substrate size, and verticality: 

 average values per 120 m-; water temperature: °C per 

 month) while controlling for potentially intercorrelating 

 variables. These results were considered significant at 

 a<0.05. Because of the low number offish observed, fish 

 counts were converted to presence or absence data and 

 logistic regression was applied. Independent variables 

 were the four log-transformed physical variables (rugos- 

 ity, substrate size, and verticality: logjg [average values 

 per 120 m-]; temperature: logjQ [value per month],) and 

 the three log-transformed kelp groups (annual under- 



story kelp, perennial understory kelp, and canopy kelp: 

 logjg [number of kelps per 120 m^]). Analyses were 

 conducted separately for each fish family that composed 

 at least 20% of total abundance (Hexagrammidae, Scor- 

 paenidae, and Gadidae). 



Results 



Physical habitat variables 



Substrate in the study sites varied from complex (rocky 

 outcrops, large boulders, and bedrock) to homogeneous 

 (small cobble and sand). The ten sites were partitioned 

 by clustering techniques into two general structural 

 complexity groups based on dissimilarities among the 

 three measured structural characteristics (rugosity, 

 substrate size, and verticality; Fig. 2A). Water tempera- 

 ture varied significantly among months (Fjg 45g=1983.2, 

 P<0.001). Temperature ranged from 1.8°C in winter to 

 11.0°C in summer and was the only physical variable 

 that did not vary among sites. Water temperatures 

 also differed significantly between years (Fj 45g=1028.6, 

 P<0.001), and were higher in 2003 than 2002. 



Surveys of kelp and fish 



A comparison of cluster dendrograms revealed patterns 

 of spatial variation among the kelp and fish groups that 

 mirrored the substrate trends. When all biological data 

 (understory and canopy kelp densities and fish pres- 

 ence) were averaged across months and years, five sites 

 grouped with higher counts of kelp and fish exhibited 

 the greatest structural complexity (Fig. 2, B [kelp] and 

 C [fish]). Similarly, three of the structurally homogenous 

 sites were grouped consistently with lower values for 

 both kelp and fish. Two sites (LJA and OJA) showed 

 inconsistencies in these groupings. Little Jakolof Bay 

 (with a lower complexity designation) was in the higher 

 macroalgal count group but in the lower fish abundance 

 group. Outside Jakolof Bay (with a higher complexity 

 designation) was in the lower density groups for both 

 kelp and fish. 



Kelp communities were variable in species composi- 

 tion and density over space and time and understory 

 kelp communities were considerably denser than the 

 canopy kelp. Understory kelps were present every month 

 and perennial kelp dominated in all months except late 

 October 2002 (Fig. 3). The annual understory kelp C. 

 costata contributed at most 2% to the annual kelp rela- 

 tive abundance in any month, whereas L. saccharina 

 composed at least 75%. Annual understory kelps were 

 found in greatest densities during periods with warmer 

 water temperatures; perennial kelps, however, were not 

 significantly correlated with temperature (Table 1). Pe- 

 rennial understory kelps were found on all transects; an 

 overall equal contribution was made by A. clathratum 

 and the perennial Laminaria species (L. bongardiana 

 and L. yezoensis. densities lumped together). Both an- 

 nual and perennial understory kelps were found in 



