88 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 123, No. 1, March 2011 
in front of the same flowers without being 
displaced. 
No other aggressive behaviors were reported 
although other hummingbird species occurred in 
feeding areas used by H. re galls. No interactions 
were observed in lower elevation forests above 
Yankuam (below 1,700 m). Intense disputes were 
observed among Brown Violetears ( Colibri del- 
phinae ), Rufous-vented Whitetips ( Urosticte rufi- 
crissa ), Violet-fronted Brilliants ( Heliodoxa lead- 
beateri), and Blue-fronted Lancebill (D. johannae) 
at flowering forest trees and epiphytes, but no H. 
re galls was observed in those interactions. The 
only female found above Miazi was observed 
displacing an unidentified hummingbird and 
perching upright, with her neck stretched, possi¬ 
bly in territorial dispute. 
Vocalizations. —We heard and tape recorded 
several types of vocalization. A tame and curious 
male approached one observer (J. F. Freile) and 
remained perched motionless for 1 min. He 
uttered a sharp, high-pitched, fast chichup chup! 
before take off with the last note more emphatic. 
A female in the same area perched in a natural 
forest gap uttered a thin, high-pitched tzifp! We 
failed to record both vocalizations. 
A feeding male uttered an emphatic tchup or 
chip! every 3.5-6 sec (deposited at www.xenocanto. 
org; XC 45910; Fig. 2A 1 ), whereas other feeding 
calls were uttered at shorter intervals (2-2.5 sec), or 
in short two- to three-notes descending series tchup- 
tchup , tchUp-tchup-tchup with the first note higher 
(XC 45911; Fig. 2A 2 ). The duration of each note 
was -0.1 sec for the feeding call (Fig. 2A 2 ), while 
frequency ranged from 1.5-1.8 to 17-19 kHz (call 
2A 1 ), and 1.5 to 21 kHz (call 2A 2 ). 
Two displaying males produced a fast chattered 
series of very high tEEp or jeet notes lasting 2.5— 
4.5 sec (XC 45912; Fig. 2B). This vocalization 
contained 19 notes of 0.10—sec mean duration 
(0.07-0.15 sec) with frequencies ranging from l .7 
to 16.69 kHz. Two males were observed in 
stunted shrubland and heard in intense dispute, 
constantly vocalizing an endless, thin, high- 
pitched jumble jijijit’jijit’jijit’jijiji. .., notes uttered 
at a much faster rate than regular vocalizations. 
This dispute and chatter was only interrupted 
when a third male dashed towards them. Our 
recordings are uploaded to www.xentocanto.org 
^XC 45910-45915) as only a single recording of 
H. regalls is currently deposited in a public audio 
library (Macaulay Library 18046; N. Krabbe, 
pers. comm.). 
Display Flight. —Three males were observed 
performing display flights in ridgetop stunted 
shrubland, one above a nectar source ( Macro- 
carpaea harlingii) and two above a shrub edge 
and rocky outcrop. In display, one male ascended 
~5 m in vertical flight, described one oval, 
possibly two, at the highest point and descended 
describing a semicircle to the same perch; the 
second male followed the first’s display flight 
constantly vocalizing (Fig. 3A). A somewhat 
similar display was also observed with a male 
ascending 10-12 m and then descending in a 
semicircle, and diving out of sight (Fig. 3B). 
DISCUSSION 
Heliangelus regalis is generally regarded as 
locally fairly common (Seddon et al. 1996; 
Schulenberg et al. 2001, 2007; Dauphine et al. 
2008). Schulenberg et al. (2001) suggested the 
Cordillera Azul might represent the center of 
abundance of H. regalis , and that it might be less 
threatened than currently believed. However, with 
a global range of only 2,100 km 2 , its global 
population has been roughly estimated at 2,500- 
9,999 individuals (BirdLife International 2009). 
Our observations suggest the abundance of H. 
regalis can vary across different vegetation types. 
However, in accordance to previous reports 
(Fitzpatrick et al. 1979, Davis 1986, Seddon et 
al. 1996, Dauphine et al. 2008), stunted shrubland 
appears as its preferred habitat, at least judging 
from relative abundances at the three vegetation 
types surveyed in this study. Higher numbers in 
stunted shrubland and paramo-like vegetation 
suggest these habitats provided plentiful food 
resources during the study season. Low detection 
in dense lower montane and foothill forests might 
also account for lower numbers in these forests 
(Poulsen et al. 1997). We cannot Rile out seasonal 
movements along vegetation gradients as previ¬ 
ously suggested by Seddon et al. (1996). Habitat 
suitability appears to be higher for stunted shrub- 
land, but defining optimal habitats solely based on 
relative abundance has proven to be misleading 
(van Home 1983, Morris 1987). 
H. regalis in stunted shrubland likely fed on 
most available nectar sources. Flowering was 
limited to a few individual plants of a few species. 
Our data suggest that H. regalis was not strongly 
dependent on one food plant species as found in 
the type locality, where the species was reported 
as highly dependent on Brachvotum quinquenerve 
(Melastomataceae) (Fitzpatrick et al. 1979); it 
