288 
Fishery Bulletin 11 6(3-4) 
Table 2 
Presumed mortality of kelt Atlantic salmon (Salmo salar ) released at an upriver and a downriver site 
on the Penobscot River, Maine, on 17 November 2015 and tracked with radio telemetry from time 
of release to mid-June 2016. Fish presumed to be initial mortalities are those that remained within 
1 km of their release site for the duration of the study. Fish presumed to be overwinter mortalities 
are those that remained at a fixed location in the river for several months, including the end of the 
study in mid-June 2016. Total presumed mortality includes both initial mortalities and overwinter 
mortalities. Mortality between the two release sites was compared using a chi-square (% 2 ) test. 
Mortality Upriver (28 individuals) Downriver (27 individuals) x 2 df P 
Initial 2(0.071) 3(0.111) 0.002 1 0.966 
Overwinter 14(0.500) 7(0.259) 2.432 1 0.119 
Total 16(0.571) 10(0.370) 1.496 1 0.221 
1997). Because spawning of all the fish in our study 
was artificially induced in a hatchery, many still had 
gametes in the abdominal cavity when we surgically 
implanted tags. Physiological mechanisms to motivate 
upstream migration before spawning have been noted 
in previous studies (Bendall et al., 2012), therefore it 
is possible that our kelts were behaving more like ripe 
adults than postspawning adults, given that gametes 
were still present. The majority (95.3%) of upstream 
movements in individuals from both release sites 
ceased within a month of release. Therefore, without 
further knowledge of the physiology of each fish, it is 
difficult to determine a trigger for upstream movement. 
Regardless of the mechanism behind upstream move¬ 
ment, the fact that kelts continue to move throughout 
the fall and early winter may be of concern to river 
managers. During this time of year, fish passage struc¬ 
tures (in both upstream and downstream directions) 
are typically closed to avoid damage from ice and de¬ 
bris (Simpson 6 ). Keeping these facilities operational in 
the winter may require substantial expenditures of ef¬ 
fort and capital, but may also provide an opportunity 
for better survival, particularly for kelts moving in a 
downstream direction. 
Because of the logistics of maintaining the radio re¬ 
ceiver array, we were unable to place receivers on the 
Stillwater Branch of the river. Because of this gap in 
coverage, we were unable to determine the predomi¬ 
nant path choice around Marsh Island (rkm 56-63) for 
fish outmigrating from the upriver release location. In 
the future, securing access to sites along this portion 
of the river could help determine whether path choice 
around the island influences mortality in the lower 
river. Previous research on Atlantic salmon smolts has 
shown that downstream passage at hydroelectric fa¬ 
cilities results in decreased survival farther downriver 
(Stich et al., 2015). Because path choice influences the 
number of dams passed (e.g. there is one dam on the 
east side of the island and three on the west, Fig. 1), 
6 Simpson, M. 2017. Personal commun. Maine Dep. Mar. 
Resour., 21 State House Station, Augusta, ME 04333-0021. 
it may be also be linked to survival. Just as smolts 
are at risk of being descaled, of movement delays, and 
depredation at dams, these sites may also pose risks to 
downstream migrating kelts. 
In both release groups, 8 tags were lost to the radio 
network, and we were unable to ascertain their area of 
disposition. The lack of transmissions from these tags 
could be due to a number of factors. First and foremost, 
the tags were all repurposed from an earlier study of 
upstream migrating Atlantic salmon, and therefore 
their batteries had already been somewhat depleted. 
Additionally, some tags may have been removed from 
the watershed either by poachers or wildlife (e.g. birds 
of prey). Finally, the most downriver receiver in the 
system covered a portion of the river that is inundated 
with salt water at high tide. Salt water can interfere 
with radio signals and likely reduced the probability of 
detection at that receiver during certain time periods. 
Survival of kelts released into the Penobscot River 
was relatively low, compared with survival in Euro¬ 
pean rivers, which may contribute to the observed de¬ 
clines in the number of multispawning individuals in 
the population (Maynard et al., 2017). Apparent initial 
mortality (fish that remained within 1 km of release 
for the duration of the study) was <15% in both re¬ 
lease groups. There was no significant difference in 
apparent overwinter mortality between the groups. 
Apparent survival to outmigration was at least 23.6% 
(13/55 individuals) , which is considerably lower than 
documented survival to outmigration in the River Imsa 
(65-85%; Jonsson et ah, 1990) in Norway and in the 
Burrishoole River in Ireland (40%; Piggins, 1990). Be¬ 
cause 2015-2016 was an unusually warm winter, an 
absence of ice cover on the river may have contributed 
to low survival because kelts are known to use ice for 
refuge from aerial predators (Cunjak et al., 1998) such 
as bald eagles (Willson and Halupka, 1995). Since the 
initiation of the Penobscot River Restoration Project, 
the proportion of bald eagle diets made up of marine- 
derived nutrients has been increasing (Call, 2015). Al¬ 
though survival to repeat spawning varies among river 
systems (2-25%, Jonsson et ah, 1991), multispawning 
