234 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124. No. 2. June 2012 
farmlands, and dry forest, up to 780 m asl. (4) The 
Guayana includes all Guayanan and Amazonian 
lowland and highland locations, south of the 
Orinoco River, up to 1,450 m asl. Most rainforests 
are in this region. 
Data Analysis.—We pooled data from SI and 
S2 (which we refer to as 'SI + S2') for most 
analyses but, for those that required numbers of 
individuals, we used data from SI only. Check¬ 
lists were used only occasionally for selected 
cases. 
Data analyses were performed with R statistical 
software with the packages ade4. vegan, lahdsv ., 
and cluster (CRAN.R. Project 2011). We com¬ 
puted national-level species richness estimates, 
using the Chaol estimator based on abundance 
(numbers of individuals) of detected species with 
data from SI. The Chaol estimator is defined as: 
c _ c , "I ("I - 0 
‘h’ ~ *0 + -t: - —rr , 
2(a 2 + 1) 
where S n is the number of observed species and a, 
and a 2 denote the number of species with one or 
two individuals, respectively. We used the Chaol 
estimator based on frequency of detections 
(number of sites at which a species was observed) 
with data from S1 and from SI + S2. The Chaol 
estimator is: 
S n — Sq + 
“i(«i ~ 1) 
2(7/2 + !) ’ 
where a, and a 2 are the number of species present 
m one and two records, respectively (Chao 1987). 
We computed another estimate of species 
richness assuming the relationship between SI 
and real species richness (S p ) is similar to the 
proportion of species detected in the first 3 min of 
at X j2ii _ 
jl-3| 
S2 «l 
where SI is the number of species detected in S 
b2 n _ 3] , s the number of species delected in S2 (; 
three 3-min periods), and S2,,, is the number 
spec.es detected in the first 3-min period. 
We used the moments-based uncondition 
estimator and its standard deviation for interpol 
ion o a species accumulation curve based on 5 
r S’ " aL 2004 >■ We funher fitted 
^cumu'r e " rurmul “ 10 e '"-upola. e tf 
accumulation curve and to estimate its asymptol 
(Tjorve 2003): 
S = R 
(K+e) 
where S is the expected number of species for a 
value of sampling effort e , R is the total estimated 
richness (asymptote), and K is the effort necessary 
to reach half the value of R. Asymptote species 
richness values were estimated for the curve of the 
observed number of species (SI + S2 data) and the 
curve of the values estimated by the Chaol 
estimator, as both estimates are influenced by the 
number of sampling locations. The resulting curve 
was further extrapolated to 100 transects to 
predict the expected number of species with 
increasing sampling effort. 
We used a cluster diagram of species compo¬ 
sition per transect with data from SI + S2 to 
examine regional groupings of species. We first 
calculated a matrix of pairwise similarity distanc¬ 
es among transects based on the Sarensen index, 
defined as: 
D = 2a/(2a + b), 
where a is the number of shared species between 
two transects, and h is the number of species 
present in only one transect. We then applied an 
agglomerative nesting (agues) algorithm to the 
resulting matrix (Kaufman and Rousseeuvv 1990). 
We further compared the number of species 
detected per family in SI + S2 and the checklists 
in all transects against the expected numbers. We 
applied a ■/' goodness of fit to test for the null 
hypothesis that observed values had the same 
distribution as those expected. Some families had 
only one or a lew species and we pooled families 
with less than three expected species to perform 
this test. / J -values were computed using a 
MonteCarlo simulation with 2,000 replicates. 
Means (e.g.. species numbers, percentages) are 
reported with their standard deviation (— SD*. 
while richness estimates (Chaol. Chaol. and 
accumulation curves) are reported with their 
standard error (± SE). 
RESULTS 
Species Richness.— About one-half to two- 
diirds ol all potential species were detected by 
the surveys (45-65%), representing between a 
third and one-half (34-49%) of all known 
Venezuelan birds (Table 2). SI and S2 were 
obtained through independent counts that totaled 
