Kagley et a!.: Residency, partial migration, and late egress of Oncorhynchus tshawytscha and O. kisutch 
547 
time that each receiver was active varied from a few 
months to years (median: 155 d) according to the pri¬ 
mary focus of the study for which they were deployed 
(detailed receiver location and status information are 
available from the Ocean Tracking Network website 
and Hydraphone Data Repository website). The num¬ 
ber of receivers deployed in each basin varied through 
time but was always more than would be feasible for 
any single study. 
Data analysis 
Fish last detected crossing uninterrupted receiver ar¬ 
rays in Admiralty Inlet, the Strait of Juan de Fuca, 
or Johnstone Strait, or detected on individual receiv¬ 
ers along the Washington State coast without exit¬ 
ing detections, were classified as transients because 
these detections indicated that they left Puget Sound. 
In contrast, fish last detected within Puget Sound 
(Whidbey Basin, Admiralty Inlet, Hood Canal, cen¬ 
tral or southern Puget Sound) and having no coast¬ 
al or Strait of Georgia detections were classified as 
residents. Based on acoustic detections, the minimum 
amount of time it took a tagged fish to leave central 
Puget Sound and reach the Strait of Juan de Fuca 
was 6 d. Consequently, we excluded from analysis 
any fish with final receiver detections within 6 days 
of tagging because the tracking duration was not 
sufficient to detect a departure. Given the distribu¬ 
tion of receivers (Fig. 1), our probability of detect¬ 
ing a fish exiting the study area for coastal waters 
was high, but detection of an individual remaining 
in Puget Sound depended upon the movements and 
location of that fish. A Welch’s Gtest was used to de¬ 
termine whether fork length differed between fish 
that were included and excluded from the analysis. A 
chi-square test was used to determine whether there 
was a difference in proportion of wild vs. hatchery 
fish for fish included and excluded from the analysis. 
A chisquare test was employed to determine wheth¬ 
er remaining a resident was independent of origin 
(hatchery or wild) or month of tagging. Welch’s t- 
tests were used to evaluate whether fork length or 
condition index (Anderson and Neumann, 1996) had 
any influence on whether or not fish remained a resi¬ 
dent within Puget Sound. Linear regression analysis 
allowed us to determine whether the total detection 
time for fish that remained in Puget Sound was in¬ 
fluenced by fork length of the fish. We also calculat¬ 
ed the total distance fish moved per number of days 
with obtained detections, and we used a Welch’s t- 
test to determine whether the total distance traveled 
per day was different for fish that remained as resi¬ 
dents and those that left Puget Sound. 
To investigate the use of different areas of Puget 
Sound by Chinook salmon that remained resident (ex¬ 
cluding transient fish), a “site-use rank sum metric” 
was calculated as a composite variable for each sta¬ 
tionary receiver. This variable was created by using 4 
metrics similar to those used by Rohde et al. (2013): 
1) number of unique fish detected at each receiver, 2) 
number of days that a receiver detected at least 1 fish, 
3) total amount of time spent at each receiver by all 
fish, divided by the number of individuals detected 
there (i.e., average time spent per detected fish), and 
4) total number of visits to each receiver, divided by 
the number of individuals detected there. These 4 met¬ 
rics were combined by summing the rank scores of each 
variable for each receiver. Low values of the site-use 
rank sum metric corresponded with low values of the 
4 variables and high values of the metric corresponded 
with high values of the 4 variables. The amount of time 
spent by each fish at each receiver was calculated as 
the time between the first and last detections. If a fish 
was not detected for more than 1 h or the fish visited 
another receiver the duration period was terminated. 
A minimum of 2 detections within 1 h were needed to 
create a duration period. 
To characterize areas where fish spent time, the 
maximum depth within the listening distance (a 520- 
m radius— an estimate of receiver range based on 
favorable environmental conditions and the tag out¬ 
put and settings) and the distance to shore were de¬ 
termined for each receiver (Smith et al., 2015). Re¬ 
ceivers were then categorized as onshore-shallow (<1 
km from shore and <95 m deep), onshore-deep (<1 
km from shore and >95 m deep), or offshore-deep (>1 
km from shore and >95 m deep). These categories 
were chosen because there was a clear break between 
the data at these cutoffs; no receivers were located 
offshore in shallow water (Fig. 2). 
To determine whether site use (measured as the 
site-use rank sum metric) differed among receiver 
types (onshore-shallow, onshore-deep, offshore-deep), 
an analysis of variance (ANOVA) was used. Univari¬ 
ate normality of the residuals was examined with a 
quantile-quantile plot, homogeneity of variances was 
examined by plotting standardized residuals against 
fitted values, and independence was examined by plot¬ 
ting the residuals by each factor of receiver type. A 
post-hoc Tukey’s honestly significant difference (HSD) 
multiple comparison test was used to determine which 
receiver types were significantly different. 
Diel patterns of receiver recordings of fish that re¬ 
mained as residents were examined by determining 
the number of discrete movements and presence of fish 
during each hour of the day. Following Chamberlin et 
al. (2011), we defined movements as discrete detec¬ 
tions between individual receivers. These movement 
data were summarized by receiver type (offshore-deep, 
onshore-deep, and onshore-shallow) and tested for cir¬ 
cular uniformity among hours by using Rayleigh tests 
with R package circular, vers. 0.4-7 (Agostinelli and 
Lund, 2013). 
Additionally, the movements of individually tagged 
Chinook salmon were compared with those of tagged 
coho salmon (Rohde et al., 2013) by categorizing re¬ 
ceivers as detecting 1) both Chinook and coho salmon, 
2) only Chinook salmon, or 3) only coho salmon. We 
examined whether the proportion of fish among these 
