Parnel et al.: Effects of oceanic conditions on ichthyoplankton communities in the Columbia River plume 
163 
Table 1 
Total number (by year and month) of ichthyoplankton samples collected off the Columbia River, Oregon, during survey cruised 
conducted by the National Oceanic and Atmospheric Administration, National Marine Fisheries Service. 
1999 
2000 
2001 
2002 
2003 
2004 
Total 
January 
2 
0 
0 
0 
0 
0 
2 
February 
0 
0 
1 
0 
0 
0 
1 
March 
2 
2 
0 
0 
0 
0 
4 
April 
3 
1 
0 
2 
2 
2 
10 
May 
4 
3 
4 
1 
6 
6 
24 
June 
2 
2 
6 
3 
6 
4 
23 
July 
2 
2 
5 
2 
4 
4 
19 
August 
0 
0 
0 
0 
0 
0 
0 
September 
1 
0 
0 
0 
0 
0 
1 
October 
0 
0 
0 
0 
0 
0 
0 
November 
0 
0 
0 
0 
0 
0 
0 
December 
1 
0 
0 
0 
0 
0 
1 
Total 
17 
10 
16 
8 
18 
16 
85 
bottle was used to collect a water sample for chlorophyll 
analysis at the 3-m depth at each station. A conductiv- 
ity-temperature-depth (CTD) cast was also conducted to 
within 5 m of the bottom at each station. 
Before they were sorted, ichthyoplankton samples 
were rinsed in fresh water with a 332-pm mesh sieve. 
The samples were then poured into a large Pyrex dish 
on a light box. All fish larvae and eggs were removed 
from the samples. Identifications were made to the low- 
est possible taxa by using a dissecting microscope and 
taxonomic information, primarily from Matarese et 
al. (1989) and Moser (1996). Zooplankton volume was 
determined by allowing the sample to settle in a gradu- 
ated cylinder overnight. Zooplankton volumes and ich- 
thyoplankton densities were standardized by adjusting 
these numbers by the volume of water filtered by the 
net. 
Data analysis 
Species diversity was calculated with the Shannon- 
Weaver index of diversity, H' (Shannon and Weaver, 
1949; Krebs, 1989), where higher values denote greater 
species diversity. Species evenness was calculated with 
Simpson’s index of evenness. A, which ranges between 0 
and 1, and where a value of 1 indicates that all taxa have 
the same density within a sample (Krebs, 1989). 
To eliminate the effect of rare and uncommon taxa, 
data from each station were filtered to remove those 
taxa that occurred in less than 5% of the samples, 
thus leaving 12 egg taxa and 12 larval taxa for further 
analysis of the community structure. Agglomerative hi- 
erarchical two-way cluster analyses were conducted to 
identify taxa and sample assemblages. For this analy- 
sis, densities were averaged from both stations from a 
single collection date. Samples with no ichthyoplankton 
were excluded. Catch distributions were highly skewed 
and thus were log 10 (cafc/i/1000 m 3 +l) transformed to 
de-emphasize the few high catches. We constructed 
separate two-way clusters for eggs and larvae data with 
the Bray-Curtis distance measure and a flexible beta 
(/3=-0.25) clustering algorithm, using PC-ORD software 
(vers. 5, MJM Software Design, Gleneden Beach, OR). 
A nonparametric multiresponse permutation proce- 
dure (MRPP) (Kruskal, 1964; Mather, 1976) was used 
to test the hypothesis of no difference between two 
or more groups. Factors tested were station (buoy 1 
and 2), year (1999, 2000, 2001, 2002, 2003, 2004), and 
season (downwelling and upwelling), before and after 
the spring transition date as defined by Logerwell et 
al. (2003). The date of the spring transition is the day 
when the ocean conditions switch from downwelling to 
upwelling, as identified by analyzing Bakun upwell- 
ing indices and sea level conditions along the Pacific 
Northwest (Logerwell et al., 2003). Identification of the 
primary taxa associated with each grouping was done 
by using an indicator species analysis (ISA). The ISA 
measures the fidelity of a taxon within a particular 
group in relation to its abundance in all groups and 
assigns an indicator value for that taxon per group. A 
statistical significance was then calculated by a Monte 
Carlo resampling method with 1000 runs. These tests 
were also run with PC-ORD V5 software. Relationships 
between environmental conditions and the ichthyoplank- 
ton community were investigated by using the BVSTEP 
procedure in the PRIMER software package (vers. 5, 
PRIMER-E Ltd, Lutton Ivybridge, UK). This analysis is 
analogous to forward stepwise multiple regression and 
compares nonparametrically paired species and environ- 
mental similarity matrices. Species similarity matrices 
were calculated by using Bray-Curtis similarities on 
fourth-root transformed data. Environmental matrices 
