236 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124. No. 2. June 2012 
v NorthCoast 
The^y*axis indicates /!!• s ^ ec ' e ^ compos ‘tion ,mo, 8 all 27 transects with symbols indicating the four ecoregit*. 
o™ tTi„ “ " ,ndiVidl,al tra " SeCIS or The ‘-wer the value, the closer two transects 
ui groups are in terms of species composition. 
the curve tends to flatten beyond this point 
(Fig. 2). 
SI lasted an average of 5.5 hrs with five 
transects lasting >6 hrs. due mostly to unexpected 
logistic delays. On average, 39% of the species 
were detected during the first hour of the survey, 
after which detection of new species continuously 
decreased: 64 to 97% of the species were detected 
during the first 4 hrs. The last 2 hrs allowed 
detection of >20% of the species for 9 of ?7 
transects. 
Species Composition.— The survey sites 1 
each region clustered together or in groups 
biological significance, the Llanos being 
clearest example (Fig. 3). The avifauna of Gu 
nan locations in northern Bolivar (Caicara. C 
and El Manteco) had higher affinity with 
Llanos, as savanna vegetation in north 
Guayana is similar to that «>r the southern Lla 
he remaining Guayanan locations, predomin 
Nwn “ afeaS ’ c,uslered lo 8 c ther. ' 
WCord.lleras locations clustered together 
rth ^ Lag0> Which dusicmd wi. 
subset of NorthCoast sites: Rosario de Perija , 
Quebrada Arriba, and Pifiango, which appears 
completely isolated from the rest of the transects. 
NorthCoast locations were the most diverse, 
forming three subgroups: Paria-Araya, Maracaibo 
Lake Basin (Rosario de Perija and Quebrada 
Arriba), and thorn scrub coastal ecosystems 
(Macanao, Paraguana, and Rio Tocuyo). 
Sixty-four of 71 potential (expected) avian 
families were recorded either by the surveys or 
in the checklists (Table 3). Most unrecorded 
families were monospecific (6 of 7). and their 
species were either nocturnal (Tytonidae. Stea- 
tornithidae, Nyctibiidae) or associated with 
habitats poorly covered by roadside surveys 
(Heliornithidae. Phoenicopteridae, Psophiidae, 
Cincilidae). There was no significant difference 
between observed and expected species composi¬ 
tion by family, i.e., the proportion that each 
family contributed to the total number of species 
(X 2 = 57.025. P = 0.153 based on 2,000 Monte 
Carlo simulations). We detected a mean of 6^% 
o( the expected number of species/family, al¬ 
though there was some variability, especially ter 
families with low numbers of species (Table 3). 
