Chen etal • SEARCH BEHAVIOR OF MIGRATORY SONGBIRDS 
349 
We noted whether the bird was in a flock or 
foraging alone. A mixed-species flock is defined 
as a group of two or more species that move in 
concert and behave cohesively while foraging 
iHutto 1987, 1994). Large numbers of Nearctic- 
Ncotropic migrant landbirds appeared at the study 
sites from time to time during the study period. 
These migrant “fallouts” often coincided with 
occurrence of severe weather conditions, espe¬ 
cially thunderstorms (Lowery 1945, 1955: Gau- 
threaux 1971). High migrant density in the plot 
caused by migrant fallout was also recorded since 
the density of migrants may increase 10- to 100- 
fold compared to regular days (Chen 1996; W.C. 
Barrow, unpubl. data). We recorded whether vine 
tangles were in the area where the bird had been 
searching beginning in 1994. Vine tangles were 
vegetated areas comprised of intertwined vines 
(most often Vitis spp.) supported by trees, usually 
sugarberry. 
Statistical Analyses .—Morrison (1984) recom¬ 
mended that a sample of at least 30 individuals, or 
about 150 sequential observations, was needed for 
data analysis of attack behavior or foraging site. 
We recorded many more search movements than 
attack behaviors during a foraging sequence. We 
included species for which we had at least 20 
individuals or 200 sequential observations. We 
excluded search sequences with durations 
<10 sec. Thirty-one species were included in 
Ihe analysis (Table 1). 
Search rate was defined as the number of search 
movements per minute (Robinson and Holmes 
1982). and was computed by dividing the total 
number of search movements in a sequence by the 
•■equence duration in minutes. An average search 
rate was calculated from all sequences for each 
species. We did not distinguish between within- 
patch flight and between-patch flight in 1993, and 
^ed only data from 1994 and 1995 to calculate 
night distance. 
The relationships between proportion of hop 
and search rate, and between flight distance and 
search rate of the 31 migrants were formulated 
us 'ng a linear model. The two flycatchers were 
deleted in the former analysis since they primarily 
used hops to change direction instead of searching 
for prey. We further examined if hop distance was 
correlated to tarsus length and whether flight 
distance was correlated to wing length of 
migrants. Morphological data were derived from 
birds banded at the sites (Barrow et al. 2000; 
Appendix). Search movement, search rate and 
movement distance were correlated, and we 
analyzed these three variables together using a 
cluster analysis. The frequencies of different types 
of search movement used by a bird, search rate, 
and movement distance are all related, and we 
examined these three variables together in a 
cluster analysis. Cluster analysis with a complete 
linkage (SAS Institute 1999) was used to group 
the 31 species by search rate, flight frequency, and 
hopping distance. Hopping frequency and flight 
distance were highly correlated to search rate; 
therefore we used flight frequency and hopping 
distance in the cluster analysis. 
The influence of environmental and social 
factors, including site, plot, presence of vine 
tangles, flocking, and migrant density in a plot 
were examined on search movement of migrants. 
We used data from 1994 in this part of the 
analysis because we stayed at only one site in both 
1993 and 1995. We pooled all observations of 
birds in (lie same family, and found only warblers 
and vireos had sufficient samples for analysis. A 
Likelihood-ratio Chi-square test (SAS Institute 
1999 ) was U sed to evaluate the association 
between type of search movement and each 
variable, individually. Logistic models (Agresti 
1990, SAS Institute 1999) were used to further 
test whether these factors collectively affected the 
frequency of search movements by warblers and 
vireos. We used a generalized linear model 
(GLM) to examine if these factors had any effect 
on search rate and movement distance of warblers 
and vireos separately (SAS Institute 1999). A 
Type I error (a level) of 0.05 was chosen for all 
statistical tests. 
RESULTS 
Search Movements Among Families— The rel¬ 
ative frequencies of hops used by migrants 
differed among families: warblers 85.0%. vireos 
77.0%, tanagers 71.1%, and flycatchers 18.1% 
(Table 1). The relative frequency of flight was 
inversely related to frequency of hops tor all taxa; 
flight and hop combined composed a large 
proportion of search movements. Flycatchers used 
flight intensively, 82.6% for Acadian Rycatcher 
(Empidonax virescens) and 81.2% tor Eastern 
Wood-Pcwee (Contopus virens). Hooded War¬ 
blers ( Wilsonia citrina) had the highest relative 
frequency of flight (22.6%), more than twice that 
of other warblers. Buttering was rare (< 5.0%) 
for most species and flycatchers did not flutter at 
all (Table 1). 
