Ha et al. • MARIANA CROW NEST SITE SELECTION 
237 
METHODS 
Study Area .—This study was conducted from 
1997 to 1999 on the island of Rota (14 15'N, 
145 21' E), part of the Mariana archipelago in the 
northwestern Pacific Ocean. The island has an 
area of 86 km 2 and a maximum elevation of 
491 m. It is primarily composed of limestone, but 
contains exposed areas of volcanic origin, and is 
classified as high volcanic with raised coral 
terraces (Falanruw et al. 1989). The ecosystem 
is described as rain forest on the upper terraces 
and grasslands and scrub on lower volcanic slopes 
and limestone terraces. The majority of Rota’s 
native forest is still intact, unlike the other 
inhabited islands in the archipelago that are 
mostly deforested (Falanruw et al. 1989). 
Breeding activities of Mariana Crows were 
studied at six sites of ~ 1 km 2 each (Morton et al. 
1999). These sites were selected because they 
were relatively large contiguous land parcels 
vegetated by forest habitat types found elsewhere 
on Rota (cf Falanruw et al. 1989). The relative 
amount of limestone forest on these six sites 
ranged from 58 to 96% (estimated from Falanruw 
et al. 1989). However, forests on these sites may 
be slightly more pristine than the typical forest on 
Rota. Secondary vegetation, consisting primarily 
of introduced tangantungan (Leucocena leucoce- 
phalus). comprised only 6% of the forested area 
of the study sites but 16% of forests on the island 
(Falanruw et al. 1989). Native limestone and atoll 
forests represent 80% of the forested area on the 
study sites and 72% on Rota overall. 
Characterization of Active Nest Sites .—Nests of 
Mariana Crows were located by systematic 
searches of the six study sites. Nests were visited 
approximately once per week after activity was 
detected, and were examined each time with a 
mirror on a pole to count the number of eggs or 
nestlings. Nests were considered to be active if 
confirmed to contain eggs, nestlings or fledglings 
at any time. Active nests that fledged at least one 
young were considered successful. 
Specific coordinates (UTM) were obtained with 
a Global Positioning System (GPS) w'ith accuracy 
recorded of riO in. The species of the tree was 
recorded (Raulerson and Rinehart 1991). and the 
diameter (to the nearest mm) of the tree trunk was 
measured using tree calipers at breast height 
(DBH). Nest height was defined as the distance 
from the bottom of the center of the nest cup to 
the ground immediately below the nest (nearest 
cm). Tree height was measured from the base of 
the tree to the top-most branch (recorded to the 
nearest 0.1 m). The hranch length of a nest was 
the distance measured from the nest cup to the 
base of the branch. Nest aspect was the compass 
bearing of the nest relative to the tree trunk, and 
was measured with the observer's back against the 
trunk of the tree to record the compass bearing 
(degrees) of a marker placed below the nest. 
Canopy cover was obtained by taking 12 readings 
with a densiometer at three locations (4 readings 
per location). The densiometer was placed on a 
tripod and canopy cover was estimated immedi¬ 
ately above the marker placed below the nest 
center. Other readings were taken 5 m north and 
south of the nest. 
Characterization of the Vegetation of Nest 
Plots versus Random Plots .—Vegetation surveys 
of nest plots were conducted around the nest tree 
The structure and composition of the vegetation 
was characterized for 55 nest plots and 60 
stratified random (12.6-m circular) plots (Fig. 1) 
within the same six study sites (Tabic 1). Random 
plots were located by systematically stratifying 
the study sites into 10 units (to capture spatial 
variation), and randomly drawing a plot from each 
of Ihe 10 subunits. Random plots were within nest 
study areas; thus, they were available to birds as 
nest sites, but were not selected by the birds in this 
study. The vegetation characteristics included the 
number and DBH (>24 mm) of stems by species, 
the number and DBH of snags, canopy cover, 
slope, elevation, and distance and direction to 
edge, road or building (if <300 m). Buildings 
>300 m from the plot center were not recorded. 
Slope was measured with a clinometer at a 
representative location uphill and downhill from 
the nest tree (the 2 readings were 180° apart). The 
straight-line distance and direction from the plot 
center to the nearest road, edge, and building were 
recorded. Roads were classified as graded or 
paved; we did not include 2-wheel tracks. Edge 
was defined as an abrupt change in habitat such as 
beach strand to open beach, limestone forest to 
road edge, or secondary forest to open field. At 
times, the nearest road was the nearest edge. Plot 
elevations were recorded from a USGS topo¬ 
graphic map. 
Data Analysis.—Three comparisons were per¬ 
formed: vegetative and spatial differences be¬ 
tween active versus random plots, differences 
between the six study sites, and differences 
between successful and unsuccessful nest sites. 
