300 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124. No. 2. June 2012 
We collected arthropods in 2009 from 0930 to 
1130 hrs CST during mid-morning from mid-June 
through early July coinciding with peak breeding 
of Dickcissels. We visited three separate locations 
in each habitat type where Dickcissels were 
primarily foraging (burned buffers, disked buffers, 
control buffers, pasture fields, hay. milo, corn, 
soybean, riparian, roads). We look 50 sweep net 
samples per site (38.1 cm diameter net). We dried 
all arthropods at 60 C for 24 hrs (Southwood 
1978). Arthropods from each size category were 
weighed to the nearest ±0.0001 mg. We assigned 
a mean weight to apply to arthropods of the same 
taxa (Order) and size categories observed on 
video (Rogers et al. 1977). Less than 1.2% of the 
total items brought to nestlings were <5 mm. and 
we only identified arthropods >5 mm and placed 
them into one of the three size classes. Certain 
species of ground-dwelling arthropods may have 
been under-represented in sweep net samples 
(Doxon et al. 2011), but these were more likely 
to be small or fast moving insects not typically 
collected by Dickcissels (e.g., orthoplerans, lepi- 
dopterans). We estimated availability of arthropod 
species and mass of each arthropod si/e class used 
in biomass calculations. 
Foraging Observations.—We recorded forag¬ 
ing trips of Dickcissels from a 2.5-tn ladder 
positioned >30 m from the nest during 2-hr 
monitoring periods concurrent with 4-hr video¬ 
taping sessions. We recorded straight-line dis¬ 
tance traveled from the nest to where food was 
collected on georeferenced maps for each forag¬ 
ing trip. We grouped foraging distances into bands 
of 10-25. 26-50. 51-75, 76-100. 100-200. and 
200+ m for analysis. We recorded cloud cover as: 
0 (clear), 25 (1-25% cloudy), 50 (26-50% 
cloudy), 75 (50-75% or mostly cloudy), and 
100% (complete overcast). We recorded wind 
speed using a modification of the Beaufort wind 
scale: 0-1.6 (calm). 1.6-8.0 (light breeze, grass 
and leaves slightly moving). 8.0-14.5 (grass, 
leaves, and small twigs constantly moving), and 
16.1+ km/hr (small tree branches moving, ground 
debris blowing around). 
Statistical Analysis .—We Calculated provision¬ 
ing rate as number of visits per nest per hour 
divided by the number of nestlings (Sejberg ct al. 
2000, Britschgi et al. 2006). Total biomass (g) was 
the sum ot biomass brought to the nest by adults 
(both males and females) per hour divided by 
number oi nestlings (Sejberg el al. 2000). We 
identified nestling diet composition from video 
observations. Foraging distance was the distance 
(m) from the nest to the location where parents 
collected food for their nestlings. 
We used general linear mixed models i 
account for multiple nests in the same field 
(random effect) and repeated observation periods 
on individual nests to lest hypotheses about 
continuous response variables (provisioning rate, 
biomass, foraging distances) (Littell et al. 2006. 
SAS Institute Inc. 2007). Predictor variable' 
included nestling age, nest locations (buffer vs. 
non-buffer habitat), nestling number (foraging 
distance only), and male helping (male vs. no 
male). We included weather variables to test for 
effects of day-to-day variation in weather condi¬ 
tions on provisioning behavior before testing for 
effects of predictor variables. We dropped any 
weather variables that were not significant at t - 
0.10. We calculated provisioning rates and 
foraging distance in both 2008 and 2009 (biomass 
was calculated for 2009 only), and included year 
as a covariate in all analyses. We included 
observer presence (observers making foraging 
observations) as a covariate in nestling provision¬ 
ing and biomass analyses to account for the 
possibility that technicians observing foraging 
trips could have affected provisioning activities. 
We used selection ratios (Manly et al. 2002) 
for eight Orders delivered to nests to measure 
selection of prey types. Orthoptera. Lepidoptera. 
and Araneuc comprised 99% of prey items and we 
restricted subsequent analyses of prey taxa tc 
those three groups. We tested hypotheses about 
factors influencing prey type and prey size using 
multinomial generalized linear mixed models 
(PROC GLIMMIX; SAS Institute Inc. 2007) to 
account for multiple observations li.e.. each prey 
item provisioned) from the same nest. We used 
prey tava and prey size (small, medium, and large 
as response variables and buffer, observer pres¬ 
ence, nestling age, and nestling number as 
predictor variables. We used a separate multino¬ 
mial model to test if foraging distance was related 
to prey size, because foraging distance data only 
existed lor a subset of the video observations 
u hen an observer was present. We used general¬ 
ized linear mixed models to test effects of nestling 
age and nestling number on probability of male 
helping. We used ot = 0.10 for all tests. 
RESULTS 
We filmed 18 nests in 2008 and 25 nests in 
2009 for 282, 1 -hr observation periods (125 in 
