240 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124. No. 2. June 20/2 
FIG. 4. Abundance distribution (number of records) 
from data of Survey I (SI). Dark gray points correspond to 
common species, while light gray correspond to 'uncom¬ 
mon or rare'. 
Halpert 1987. NOAA 2011). Only two species of 
ducks, for example, were recorded in the surveys 
and six in the checklists of the potential \2 
species. In addition, numbers were unusually low, 
especially in the Llanos, where whistling ducks 
(■Dendrocygna spp.) in an average year can reach 
the thousands (Dallmeier 1991, Vilella ct al. 
2010). The average proportion by family of 
expected species detected in the surveys increases 
from 65 to 73% il waterfowl, shorebirds. 
nocturnal families, and hummingbirds (Anatidac. 
Charadriidae. Scolopacidae, Tvionidae. Strigidae. 
Steatornithidae. Nyctibiidae, Caprimulgidae, and 
Trochilidae) are removed. 
This relative limitation of our survey methc 
for sampling nocturnal, colonial or rare species 
possibly outweighed by the prospect of rapid I 
producing systematic, large-scale abundance dat 
for common species that could be of great use fc 
monitoring programs. A high proportion r 
common species (27%) was found in the stud 
cells, similar or higher than the expected propoi 
tion (17%). This is an indication that our survey 
represent the actual abundance, at least fc 
common, more conspicuous species (Fig. 4). 
The largest number of individuals (5,560) am 
records (2,673) was recorded in the Llanos 
despite being third in rank of species richnes 
among the lour ecoregions (Table 4). Thest 
numbers would have probably been much large 
a more typical year when thousands o, 
waterbirds gather in the water bodies (Vilella el 
al. 2010). Not surprisingly, the Llanos region also 
had a high proportion of common species (53% 
compared with 27% for the entire country), 
although the highest value corresponded to the 
NorthCoast with 56% (Table 4). This higher 
proportion of common birds in the lowlands is 
one of the reasons why species composition of 
some transects from Guayana and NWCordilleras 
(such as Sur del Lago) tend to overlap partly with 
these two lowland ecoregions, producing the 
regional ’blending’ (Fig 3). 
Our estimated values, all between 785 and 82S 
species, were below' the potential species richness 
of 1,033 (Fig. 2). However, that number is 
probably unrealistic since it includes all species 
that have ever been recorded in the study area, bur 
not necessarily present during our surveys, as well 
as colonial and nocturnal species. An important 
pattern observed is that the accumulation curve 
tends to flatten beyond the sampled 27 transects 
with an increase of only 110 new species from 
593 species (57% of the potential 1.033 species) 
at 27 transects to 703 species (68% of (lie 
potential) at 81 transects. Thus, only 18% more 
now species would be reported with 200% more 
sampling effort (Fig. 2). Therefore, these 27 
transects, a number obtained from the NeoMaps 
sampling design, appear to be the breakpoint ol 
optimal cost-effectiveness. A similar species 
accumulation pattern was obtained by Vilella et 
al. (2010) through waterbirds roadside counts in 
Venezuela Llanos, where the curve tends to 
llattcn after the first 10 of 54 sampled transects. 
The field techniques we used were based 
mostly on the North American BBS (particularly 
SI), a set ol techniques that has been developed, 
tested, and improved over the past 45 years 
(Bystrak 1981. Robbins et al. 1986. O'Connoret 
al. 2000, Robbins 2000). Some neotropical 
habitats may host bird diversities four or five 
times higher than in temperate zones, whereas 
individual species densities may be up to 10 time> 
lower, and most of these species, common or 
scarce, have low detection rates (Terborgh et al. 
1990). I his is a major challenge when censusing 
tropical birds by this technique, as the high 
number of species at each count site may be 
difficult to cover, especially early in the morning 
Rodriguez et al. (2007) hinted, after the NeoMaps 
pilot test, that a second survey (S2 in our study), 
conducted u day after the main survey (SI ) might 
prove useful for at least two reasons. First, it 
