450 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 123. No. 3. September 2011 
percent divergence were based on uncorrected-p 
and Kimura 2-parameter. The average distance 
within the endemic lowland clade between 
subspecies of O. megalot is including Mimizuku 
gurneyi ranged from 3.6 to 5.8% (Table 3). The 
pairwise sequence divergences within the lowland 
non-Philippine endemics clade ( O. bakkamoena , 
O. lempiji, and O. lettia) were much lower at 
0.03-1.2%, and were comparable to the diver¬ 
gence distance values among the montane clade 
taxa, which ranged from 3.8 to 6.5%. 
Our preliminary divergence dating attempts 
using Beast Version 1.4.8 (Drummond and 
Rambaut 2007) suggested the First invasion of 
the archipelago started in Mindanao by M. gurneyi 
which branched off from the Indian Scops Owl 
group from the mainland (4.6 mya, 95% HPD = 
2.6-5.8). Dispersal to the Greater Panay Island 
block may have occurred after emergence of the 
landmass (2.4 mya, 95% HPD = 1.7-3.3), 
followed soon after by dispersal back to Mind¬ 
anao and to Luzon (2.2 mya, 95% HPD = 1.4- 
3.1). The montane invasion of the islands started 
at about the same time (2.5 mya, 95% HPD = 
1.2-3.7). 
Morphometric Variation. —Specimens of mega- 
lotis were significantly larger than everetti, which 
were significantly larger than nigrorum (Table 2). 
All 16 tests were significant except tail length was 
not significantly different between everetti and 
nigrorum (Fig. 3). Measurements of male mega- 
lotis were significantly greater than everetti for 
wing (F = 29.28. P < 0.0001), tail (F = 8.08, P 
< 0.01). culmen (F = 8.21, P < 0.01), and tarsus 
(F - 48.47, P < 0.0001). Female megalotis were 
also significantly larger than everetti for wing (F 
= 26.80. P < 0.0001), tail (F = 13.12, P < 
0.002), culmen (F = 46.78. P < 0.0001). and 
tarsus (F = 24.68. P < 0.0001). Measurements of 
male everetti were significantly larger than 
nigrorum for wing (Z = 2.82, P < 0.001), culmen 
(Z - 1.91. P < 0.02). and tarsus (Z = 1.62. P < 
0.05). but not tail (Z = 1.55, P < 0.07). The same 
pattern held lor female everetti. which were 
significantly greater than nigrorum for wing 
(Z = 2.04. P < 0.02), culmen (Z = 2.16, P < 
0.01), and tarsus (Z = 1.59. P < 0.05). but not tail 
(Z = 0.91. P < 0.18). 
Contrary to the typical reversed sexual size 
dimorphism (RSD) found in owls, only one of the 
16 measurements showed a larger character for 
males than for females. Male nigrorum have 
longer tarsus length than females (Table 2), but 
this should be interpreted with caution due to the 
limited sample size. 
Plumage Variation. —We focused on diagnostic 
differences among the three taxa. The taxon 
megalotis has two characters the other two lacked, 
scapular stripes and tarsal feathers extending onto 
the upper part of the foot. A reddish-colored head 
characterizes nigrorum in contrast to megalotis 
and everetti. 
The darkest taxon is everetti. which has dark 
brown underparts and is nearly black on the nape. 
Ln contrast, nigrorum is the lightest taxon with 
white-striated undeiparts and reddish nape. Inter¬ 
mediate is megalotis with ashy-brown underparts 
and brownish nape. 
A red morph is also found in megalotis. Two of 
21 (10%) specimens examined were red morph 
(one each from Luzon and neighboring Poliiio. 
both females |DMNH 2816. 14477]), and three 
(14%) appeared to be intermediates (from Luzon. 
2 females [DMNH 52949, 52951] and 1 male 
(CMNH 38249]). 
DISCUSSION 
We found that scops owls colonized the 
Philippines in at least two independent events with 
subsequent diversification occurring independently 
in both montane and lowland clades. This pattern is 
similar to that observed among endemic mammals, 
although time scales were different (Heaney 1986. 
1991. Steppan et al. 2003). 
There arc three prediction models that explain 
patterns of colonization in islands with montane 
and lowland biomes, including: (1) a montane 
clade nested w ithin a lowland clade on the same 
island is predicted when a single colonization 
event of the lowland biome is followed by a 
cladogenetic colonization by a new population of 
the montane biome. (2) a lowland clade nested 
within a montane clade w'ithin an island is 
predicted when a single highland colonization 
event is followed by lowland colonization by a 
new population, and (3) a separate montane and 
lowland clade is predicted during multiple parallel 
colonization events. During the latter, the lowland 
taxa between adjacent islands should form a 
monophyletie group while the montane taxa form 
a parallel clade. The first and second models 
suggest a cladogenetic speciation event within an 
island. The third model suggests independent 
parallel altitudinal colonization events. Cases 
supporting all three hypotheses have been report¬ 
ed in New World avian taxa (Rice et al. 1999). 
