192 
Fishery Bulletin 111(2) 
receivers (VR2, VR2W; VEMCO) deployed throughout 
the study area (Fig. 1A). Most receivers were deployed 
by 8 June 2007 (receivers numbered 1-27), but 4 re- 
ceivers were deployed on 26 June (receivers numbered 
28-30) and 16 July (receiver numbered 31) to provide 
additional coverage. Receivers were attached to an an- 
chored line fitted with a buoy and positioned near the 
bottom of the water column (<1 m from the bottom of 
the ocean floor) with the hydrophone oriented down- 
ward. Range tests conducted throughout the study area 
indicated an approximate detection range of 350 m. 
We placed as many receivers as possible -700 m 
apart in the upper and lower channels to be able to 
monitor fish movements on the scale of 100s of me- 
ters (Fig. IB). Currents and boat traffic limited the 
placement of receivers in certain locations in the up- 
per channels and prevented the use of a directional 
gate (Heupel et al., 2006) at Wachapreague Inlet. In 
these cases, receivers were placed in the next suitable 
location. The tidal flat was too shallow for extensive 
receiver coverage; instead, we used receivers to moni- 
tor fish as they entered and exited the tidal flat. Most 
receivers were retrieved on 31 January 2008, but 12 
receivers (receivers numbered 5-7, 16-18, and 22-27; 
Fig. 1A) were left in the system to detect fish return- 
ing to the Wachapreague system the following year. To 
prevent receiver loss, we began retrieval of the receiv- 
ers that were farthest from the inlet in mid-April. All 
receivers were retrieved by 29 July 2008. 
A separate acoustic telemetry study conducted by re- 
searchers to examine movements of Cownose Ray ( Rh - 
inoptera bonasus) in the Wachapreague system over- 
lapped with the timing of our Summer Flounder study. 
Receivers from the Cownose Ray study were placed 
mostly in small channels far from Wachapreague Inlet 
in an area not covered by our receivers. Receivers for 
that study were deployed on 26 June 2007 (receiver 
labeled S3) and 26 July 2007 (receivers labeled S5-S9, 
S11-S12; Fig. 1A) and retrieved on 17 November 2007. 
The receivers in the Cownose Ray study were spaced 
too far apart to meet the specific objectives of our study 
and detections from these receivers were not used in 
our analyses. We did note, however, the extent to which 
Summer Flounder were detected in these small back 
channels and henceforth refer to these receivers as 
supplemental receivers. 
Migratory behaviors 
Data were examined over weekly intervals to examine 
patterns of seasonal migration. We considered a fish 
to have dispersed on the last day it was detected at or 
near Wachapreague Inlet (receivers 17-22, 31; Fig 1A). 
Likewise, we considered a fish to have returned when 
it was first redetected at Wachapreague Inlet or with- 
in the lagoon system. Weekly probabilities of disper- 
sal and return were calculated with the Kaplan-Meier 
estimator, a nonparametric approach that requires no 
assumptions about the underlying hazard function 
and accommodates censored fish (Pollock et al., 1989; 
Bennetts et al., 2001). Fish were censored from (i.e., 
not included in) this analysis if they were no longer 
detected but did not depart from the system through 
Wachapreague Inlet; the fate of such fish could not be 
conclusively determined. Censored fish may have re- 
sided in the system undetected, been removed by fish- 
ermen or predators, or have left the system through 
another route. 
We used a piecewise linear regression model to 
identify when dispersal rates changed (i.e., the change- 
point), and we fitted the model to the data with non- 
linear least-squares estimation (the NLIN procedure in 
SAS, vers. 9.2, SAS Institute, Inc., Cary, NC; e.g., Ryan 
et al., 2007). The time before dispersal rates changed 
was considered the residency period, a time during 
which most fish were found within the lagoon system. 
The time after dispersal rates changed was considered 
the emigration period, during which most fish were 
observed finally to have dispersed. We classified fish 
according to observed migratory behaviors: those fish 
that dispersed early (during the residency period) and 
those fish that dispersed late (during the emigration 
period). An odds ratio (Agresti, 2007) was used to test 
the association between the timing of dispersal (i.e., 
residency period vs. emigration period) and the like- 
lihood that a fish would return to the Wachapreague 
system the following year. 
Some fish were detected at or near the inlet (receiv- 
ers 17-22, 31) but were subsequently undetected for 
14 or more consecutive days before redetection. These 
fish were classified as temporary emigrants because 
they were presumed to have exited and re-entered the 
lagoon system. Such behaviors were consistent with 
activity reported in a previous study (Sackett et al., 
2007). Tagged fish, including temporary emigrants, 
were considered residents until final dispersal out of 
the inlet, and residence time was defined as the total 
number of days from the start of our study (8 June 
2007) until the last detection at or near the inlet be- 
fore final dispersal. The residence time of uncensored 
fish was used to calculate a mean residence time for 
Summer Flounder in the Wachapreague system. The 
mean residence time reported throughout this article 
is, therefore, an estimate of least (minimum) residence 
time because we do not know how long tagged fish 
were present in the lagoon system before the start of 
our study. Mean residence time and other mean values 
are reported as mean ±1 standard error. 
The effects of mean monthly temperature and mean 
monthly photoperiod on the percentage of fish that fi- 
nally dispersed in a given month (log-transformed to 
improve homogeneity of variance) were examined with 
a multiple linear regression (general linear model 
[GLM] procedure in SAS). We also examined the effect 
of fish size on the probability of final dispersal before 
and after dispersal rates changed with logistic regres- 
sion (LOGISTIC procedure in SAS). Goodness-of-fit sta- 
