Doyle et al.: Neustonic ichthyoplankton in the western Gulf of Alaska 



235 



abundance in each sector by the total number of sta- 

 tions sampled within that sector. 



Cooccurrences of larval fish taxa in the neuston 

 net samples were determined by using recurrent 

 group analysis (Fager, 1957). This analysis identi- 

 fies groups of taxa that occur together relatively fre- 

 quently and considers only joint occurrences and not 

 abundance. The procedure involves two steps: the 

 calculation of indices of affinity for each pair of taxa 

 and the formation of groups of taxa based on a cho- 

 sen minimum index value. The equation for the af- 

 finity index is 



N, l 



where/ = the affinity index (range 0-1); 



Nj = the number of joint occurrences; 



N a = the number of occurrences of taxon a, 



the less common taxon; 

 Nb = the number of occurrences of taxon b, 

 the more common taxon. 



species. The "flexible sorting" strategy was used and 

 a recommended value of -0.25 was chosen as the clus- 

 tering intensity coefficient (Lance and Williams, 

 1967). 



To aid in identification of groups from the dendro- 

 grams, the original data sets (species abundance x 

 stations) were rearranged into two-way tables accord- 

 ing to the order that species and stations appeared 

 in the dendrograms. In this manner, it was possible 

 to see how a group of stations was characterized by 

 the occurrence or definitive range of abundance of a 

 particular species or group of species. After the final 

 species and station groups were chosen, the two-way 

 tables were reduced by calculating the mean abun- 

 dance of each species, within the different species 

 groups, for each station group. The station groups 

 were then plotted on maps of the sampling area to 

 aid in the identification of geographically distinct 

 groups of fish larvae. 



Results 



For this study, minimum index values for the for- 

 mation of recurrent groups were set at 0.3 and 0.4, 

 respectively, for two separate runs of the analysis. 

 One or both of these values have been chosen previ- 

 ously by other workers who applied this method to 

 ichthyoplankton data (Kendall and Dunn, 1985; 

 Moseretal., 1987). 



Numerical classification was used to investigate 

 multispecies spatial patterns among the fish larvae 

 in the neuston. It involves grouping similar entities 

 based on numerical data such as, in this instance, 

 species abundance at a range of stations (Clifford and 

 Stephenson, 1975). An agglomerative, hierarchical 

 technique was chosen. Normal and inverse classifi- 

 cations were carried out on the data sets (i.e. both 

 the species and stations were classified into groups). 

 Only the dominant larval fish taxa occurring in >4% 

 of the samples were included in this analysis, as 

 scarce taxa did not contribute significantly to spa- 

 tial patterns overall. The numerical classification was 

 performed on each individual data set from the seven 

 neuston cruises, as well as on data combined for all 

 the cruises (i.e. mean abundance of larval fish spe- 

 cies in each of the previously chosen geographical 

 sectors). The data were log-transformed prior to 

 analysis. 



The first step in the classification procedure com- 

 prised the calculation of correlation coefficients for 

 each pair of species or stations in a data set. The 

 Bray-Curtis dissimilarity measure was used. An 

 agglomerative, hierarchical sorting strategy pro- 

 duced dendrograms depicting clusters of stations and 



Taxonomic composition and density 



A total of 24,327 fish eggs were collected in the neus- 

 ton samples. Eggs of 12 species representing five 

 families were identified from the samples (Table 2). 

 The numerically dominant taxa included the gadid 

 Theragra chalcogramma and several pleuronectids, 

 mainly Errex zachirus, Hippoglossoides elassodon, 

 Microstomas pacificus, and other unidentified 

 Pleuronectidae. Theragra chalcogramma was the 

 only taxon whose eggs occurred in greater than 10% 

 of all the samples. Although the density of Clupea 

 pallasi eggs was relatively high, this taxon occurred 

 in less than 1% of the samples. It is likely that the 

 presence of these demersal eggs in the neuston was 

 due to clumps of eggs breaking off the substrate and 

 floating to the surface. The low diversity and gener- 

 ally low density offish eggs in the neuston (relative 

 to the diversity and density of fish larvae) was in- 

 dicative of the scarcity of species that spawn pelagic 

 eggs in this region. 



In total, 41,157 specimens of larvae or early juve- 

 niles were caught in the neuston. The taxonomic di- 

 versity and overall density were higher than for the 

 eggs (Table 3). Thirty-five species were identified 

 representing a total of 18 families. Apart from T. 

 chalcogramma and Anoplopoma fimbria, which 

 spawn pelagic eggs close to the bottom, the numeri- 

 cally dominant taxa among the larvae were demer- 

 sal spawners. 



Among the dominant larvae, the families Hexa- 

 grammidae and Cottidae were best represented. The 



