Weitkamp et a!.: Seasonal abundance, size, and host selection of Lampetra ayresii and Entosphenus tridentatus 
217 
true. Therefore, we viewed densities as indicators of 
relative abundance, rather than of absolute abundance 
and did not quantitatively compare densities between 
studies or gear types. 
Because the CREDDP sampling sites were distrib- 
uted throughout the estuary, we used station-specific 
catch data to calculate the mean location (rkm) for 
each species and age class of lamprey. Differences be- 
tween groups were evaluated by using Kruskal-Wallis 
(KW) one-way analysis of variance (ANOVA) on ranks, 
followed by a Bonferroni multiple comparison test if 
significant differences were detected (Zar, 1984). 
Finally, we wanted to determine whether the size 
of western river lamprey differed by study or gear 
type, in part because such variation in size may in- 
dicate differences in life history or gear selectivity. 
However, because the length of western river lamprey 
increased over the sampling season, we used analysis 
of covariance (ANCOVA) for these comparisons, with 
ordinal date as a covariate (Sokal and Rohlf, 1995). 
We tested whether the mean size and change in size 
over time (i.e., slope) of western river lamprey var- 
ied between the CREDDP and EPS studies and be- 
tween gear types (purse seine versus trawl) for the 
CREDDP. This analysis produced adjusted means for 
each group (study or gear type), which were adjusted 
for the effect of the covariate (ordinal date) for that 
group by using linear regression procedures (Sokal 
and Rohlf, 1995). 
Prey selectivity by lampreys Our analysis of fishes with 
lamprey wounds was designed to determine 1) the 
frequency of wounded fish of each species overall, 2) 
whether there was a seasonal pattern to the presence 
of wounded fishes, and 3) whether lampreys appeared 
to be selecting fishes of a particular size. We also exam- 
ined the frequency with which lampreys were caught 
together with wounded prey as an additional measure 
of the seasonality of lamprey predation. 
We examined the frequency of lamprey wounds by 
comparing the abundance of a particular fish species 
with and without wounds at annual and seasonal time 
scales. For annual estimates, we first determined the 
number of cruises in which fish with lamprey wounds 
were observed and divided that number by the num- 
ber of cruises in which fish of that species were caught 
(with or without wounds) each year. We also calculated 
the percentage of fish with and without wounds caught 
during each cruise for each species and averaged the 
values over cruises in which at least one wounded fish 
was observed. This estimate was based on total catch- 
es of 21,484 surf smelt ( Hypomesus pretiosus), 19,816 
American shad ( Alosa sapidissima ), 7422 Pacific her- 
ring ( Clupea pallasii), 4380 subyearling Chinook salm- 
on, 2696 shiner perch ( Cymatogaster aggregate t), 1231 
steelhead (O. my kiss), 681 longfin smelt ( Spirinchus 
thaleichthys ), and 106 sockeye salmon (O. nerka). To 
explore seasonality in lamprey wounds on the most 
frequently wounded species, we calculated the number 
of fish of a given species with wounds divided by the 
total number of fish of that species caught within each 
2-week period of each year (1 April-15 October), and 
then we averaged the values across years. 
To determine whether there was size selectivity in 
host choice by lampreys, we compared the size of fish 
(of a particular species) with lamprey wounds with the 
size of fish (of the same species) without wounds from 
the same cruise using a Mann-Whitney test of medi- 
ans (MW; Zar, 1984). Low numbers of wounded fish of 
a particular species from a single cruise, however, re- 
sulted in low statistical power. Because the size of most 
fishes increased over the course of the summer, we also 
we used a nonparametric Friedman test (an ANOVA 
analog) to test for size differences between fish with 
and without wounds; cruise was treated as a block (So- 
kal and Rohlf, 1995). We used probabilities of a<0.05 as 
the significance level for both tests. We also examined 
the overlap between catches of lampreys and wounded 
fishes both by haul and by cruise. 
Results 
Lampreys in the Columbia River estuary 
During the 18-month-long CREDDP, 42 western river 
lamprey and 42 Pacific lamprey (35 juveniles and 7 
adults) were caught. During the EPS study, 44 west- 
ern river lamprey and 4 Pacific lamprey (2 juveniles 
and 2 adults) were caught over 10 years (Table 1). One 
lamprey ammocoete (of an unidentified species) was 
caught during the CREDDP. 
The frequency of occurrence (FO) of lampreys in the 
Columbia River estuary varied in both studies by spe- 
cies, age class, and year and in CREDDP, by gear type 
(Table 1). Within the EPS study, FO reflected the in- 
frequent catch of Pacific lamprey (mean FO: 2.1%) and 
the frequent and relatively consistent catch of western 
river lamprey (mean FO: 24.4%). Although no west- 
ern river lamprey were encountered during 2001 and 
2002, at least one western river lamprey was caught 
during nearly one-third of all cruises conducted dur- 
ing 2003-2012 (mean FO: 29.8%). During the CREDDP, 
catch of the different species and age classes varied by 
gear type and to a lesser extent by year (Table 1). The 
FO for Pacific lamprey juveniles was much higher in 
trawls (mean FO: 42.4%) than in purse seines (mean 
FO: 9.1%), higher for Pacific lamprey adults in purse 
seines (mean FO: 22.0%) than for adults in trawls 
(mean FO: 4.6%), and relatively high for western river 
lamprey in both purse seines (mean FO: 35.6%) and 
trawls (mean FO: 60.6%) (Table 1). 
The density of western river and Pacific lamprey 
juveniles and adults exhibited clear seasonal patterns 
that were consistent across 3 data sets: the data set 
from EPS (purse seine) study and 2 subsets of data 
from CREDDP (trawl and purse seine) study (Fig 2). 
Adult Pacific lamprey had the highest densities during 
January-March and lower densities during April-May, 
and no adults were caught during June-December. Ju- 
