Judge et al. • REOCCURRENCE OF ‘OMA'O ON HAWAI‘1 ISLAND 
679 
TABLE 1. Model parameters and model selection results for ‘Oma'o from the 2010 HAVO survey. Models were sorted 
hv differences in second-order Akaike’s Information Criterion after correcting for small sample si/e (AAlCc) between each 
candidate model and the model with the lowest AKY value. Models included: H-norm = hull normal and 11-rate = hazard- 
rate key detection functions with scries expansions Cos cosine, IT-poly ~ hermite polynomial, and S-poly = simple 
polynomial. Covariates were incorporated with the most parsimonious model to improve model precision. Covariates 
included the continuous variable Time-con = time of detection. Categorical variables included Cloud = cloud cover. Cover 
amount of vegetation cover. DectType = detection type. Height = height of vegetation. Rain = amount of rain, 
Observer = observer. Panel = transect category. Time-cat = time of detection. Tract - HAVO tract, and Wind and Gusts 
= Beaufort wind scale. The number of estimated parameters (Variable) and Akaike model weight (u7) are provided for 
each model. 
Model 1 
Variables 
AAlCc 
AICc 
W'i 
H-rate Key 
2 
0 
2310 
0.771 
H-rate Key Tract 
8 
4 
2314 
0.104 
H-rate Key Cloud 
3 
8 
2318 
0.014 
H-rate Key Rain 
3 
8 
2318 
0.014 
H-rate Key Cover 
3 
8 
2318 
0.014 
H-rate Key Height 
3 
8 
2318 
0.014 
H-rate Key Time-cat 
3 
8 
2318 
0.014 
H-rate Key Time-con 
3 
8 
2318 
0.014 
H-rate Key Observer 
3 
8 
2318 
0.014 
H-rate Key Panel 
3 
8 
2318 
0.014 
H-rate Key Wind 
4 
It) 
2320 
0.005 
H-rate Key DectType 
4 
10 
2320 
0.005 
H-rate Kev Gusts 
5 
12 
2322 
0.002 
H-norm Key 
1 
29 
2339 
0.000 
J Model failed lo converge or 
parameters were highly correlated for the H 
-norm 
Cos. H-norm H-poly. H-rate Cos. and H-ratc S-poly models. 
unsuitable habitat to reoccupy leeward wood¬ 
lands. Captive-bred releases of Puaiohi on Kaua'i 
are generally successful, but most birds disperse 
<5 km to release sites and pairs rarely expand 
into new territories (Tweed e( al. 2003, Snetsinger 
et al. 2005). 
Little is known regarding the biology of the 
scrub alpine ’Onia'o population. Incidental detec¬ 
tions of ‘Oma’o in scrub alpine habitat have 
occurred over a long period. 1990s to present, and 
from throughout the year (Fcb-Nov). ‘Oma'o 
were detected since our 2010 survey in the scrub 
alpine habitat above both the Honomalino and 
Papa tracts (Mel Johansen. The Nature Conser¬ 
vancy. pers. comm.). ‘Oma’o were readily 
observed during seabird surveys near lava tube 
entrances, overhangs, and cavities, predominately 
in pahoehoe flows. These flows have more 
vegetation than a‘a flows because soil and 
moisture can accumulate on the smooth pahoehoe 
surfaces and in wash areas (Eggler 1971). ‘Oma'o 
in alpine habitats likely subsist on berries of 
‘bhelo (Vacciniiwi relit uUiimn), kukaenene 
(Coprosnui emntleoides). and pfikiawe (LepteCo- 
phylla tamdameiae ), as evidenced from drop¬ 
pings with these seeds found beside raised rocks 
where the birds had been perching (SWJ and 
JMG, pers. obs.), There is also a significant 
arthropod component in the diet of scrub alpine 
’Oma'o populations (Wakelee 1996). ’Oma'o 
nesting in scrub alpine habitat would be vulner¬ 
able to feral cat and rat depredation, which is a 
concern for this unusual solitaire population. 
Hawaiian forests are patchily distributed and a 
metapopulation framework can appropriately 
describe avian processes (Flaspohler et al. 2010). 
Bird populations in the Hawaiian Islands are 
disrupted by volcanic eruptions, habitat degrada¬ 
tion. and disease (Pratt et al. 2009, Flaspohler et al. 
2010). Habitat connectivity and dispersal rale of 
each population affect the success of a species 
(Gilpin and Hanski 1991). ‘Oma'o possess greater 
resource plasticity and have retained connectivity 
among wet mesic forest habitats, unlike Puaiohi 
that have distinct nesting and foraging require¬ 
ments (Wakelee 1996, Snetsinger et al. 2005). 
'Oma’o from the scrub alpine population would 
have to undergo substantial physiological and 
behavioral shifts to disperse to wet mesic forests. 
The processes needed for ’Oma'o to disperse 
between alpine and woodland habitats, however, 
would not be as restrictive as birds could forage in 
the mosaic ol dry woodlands and large lava Hows. 
Polymorphic microsatellite loci (Eggert et al. 
