Ball et al. • NOCTURNAL PROVISIONING OF THRUSHES 
511 
TABLE ^ Candidate model set assessing whether pre-dawn to post-dusk hourly provisioning rates varied based on 
2% hrs of observations from 20 nests. 
Umodelk) 
Base + night provisioner X time period 
Base + night provisioner 
-680.17 
-688.17 
-690.21 
19 
12 
11 
1.401.10 
1.401.44 
1,403.35 
0.00 
0.34 
2.25 
1.00 
0.84 
0.32 
0.46 
0.39 
0.15 
1 B&e model includes nestling age -r nestling number * lime period. 
r Model likelihood given data x = expl - ' t AAIC,). 
A model that considered day length alone 
received overwhelming support for explaining 
variation in provisioning rate during the night 
period at Fort Simpson (Table 3). Provisioning 
rate increased sharply from zero when day length 
exceeded 19.3 hrs (Fig. 3). Only the lull model 
received more support (AAIC r = 0.97) hut the 
small evidence ratio (1.62) suggests little im¬ 
provement over the more parsimonious day length 
model. Temperature also explained some variu- 
lion in provisioning rate during the night period. 
The evidence ratio of the temperature-only model 
was 59 times that of the base brood demand model 
and the linear temperature coefficient did not 
include zero in its confidence interval (95% Cl: 
~0.23--0.02). Increasing temperature from 8 to 
18° C, which was the approximate range of night 
temperature at Fort Simpson, reduced the predict¬ 
ed night provisioning rale from 3.03 (95% Cl. 
1.50-6.12) to 0.85/hr (95% Cl: 0.42-1.73). 
Twenty-nine nests successfully fledged one or 
more young. We did not record provisioning 
during the night period at Chinchaga and were 
unable to include study area as a covanate in our 
reproductive success models. We assessed the 
effects of night provisioner status on reproductive 
performance for all nests combined and for Fort 
Simpson nests only. Results for Fort Simpson 
were similar to those for all nests. We found no 
evidence that nests provisioned during the night 
period had improved reproductive success. Night 
provisioning did not affect the probability of 
being successful (f. = 0*38, P = 0.54), the 
number of young fledged from successful nests 
( x * ( = 0.06, P = 0.81), or age at fledging (F L27 - 
f) 66. P = 0.42; Fig. 4). 
FIG. 2. Predicted hourly provisioning rule by Swain- 
wn's Thrush between dawn and dusk at Fort Simpson. NT 
'hat did (gray bars) and did not (hatched bars) provision 
'heir nestlings at night. Values arc marginal means based on 
'he full model that considered nestling age. nestling 
number, and the interaction between night provisioning 
status (yes/no) and time period (Tabic 2). Nestling age and 
nestling number are set to average values of 6.32 and 
3.47, respectively. 
discussion 
Our data clearly demonstrate the provisioning 
period for Swainson s Thrush is not restricted to 
the hours between dawn and dusk. Pairs extend 
\ provisioning well before sunrise and well after 
| sunset, particularly at higher latitudes where 
provisioning continues through the night. Intui¬ 
tively this result is not surprising given the 
tendency of many songbird species, including 
Swainson's Thrush, to migrate at night (Mack and 
Wang 2000, Newton 2008). Some nocturnal 
migrants possess a special brain function that 
enhances night vision outside of the migratory 
period (Mouritsen et al. 2005), which would 
enable parents to extend their provisioning period. 
However, migratory flights arc in open skies and 
likely require less visual acuity than navigating 
and foraging in a forest. This may explain why we 
did not record night provisioning until day length 
exceeded 19.3 hrs. This threshold approximates 
the day length at the latitude of our northern site 
