758 
THE WILSON JOURNAL OF ORNITHOLOGY . Vol 123. No. 4. December 2011 
TABLE 2. Differences (two-independent samples 
test) between dense grass and nest areas, Yalu River Estuary, Chim 
Variables Mann-Whiuiey V 
Wilcoxon W z 
P (2-lailedl 
Cover (n, = 20, n 2 = 53) 
Height of total grasses («, = 32, n 2 = 53) 
Number of total grasses (n, = 34, n 2 = 53) 
37 
17.5 
109.5 
1.468 -6.111 
1.448.5 -7.54 
1.540.5 -6.886 
<0.001 
<0.001 
<0.001 
an entrance. The mean (± SD) entrance orienta¬ 
tion was 296.5 ± 64.1° ( n = 21, r = 0.535, P < 
0.01), significantly different from the southeast 
direction ot the wet monsoons in this region (% 2 = 
23, df = I, P < 0.001). The difference in live or 
lotal grasses between nests with and without 
entrances was significant (U = 157 and 165, W = 
388 and 396, Z = -3.256 and -3.11. P = 0.001 
and 0.002, respectively). 
Nests (n = 41) failed because of predation of 
eggs and or young. Small mammals were the 
major cause of nest failure, accounting for 88% (» 
- 36) oi all nest failures. No nests were lost to 
flooding. The Northern Harrier (Circus cyaneus) 
was a predominant predator for adult warblers, but 
not for eggs and nestlings. Common Cuckoo 
.Cuculus canons canorus) parasitism did not 
influence the breeding of Marsh Grassbirds 
Analyses of Variables.-No nests were found in 
dense Calamagrostis, where cover and number of 
grasses were significantly different (cover = 97 o 
±3.0%, density = 574.5 ± 90.6 in 0.25 nr „ = 
(Tahf B 00 ? fr ° m ‘ hOSe M the nest 
TaWe 2 . Nest height was significantly different 
W - 116. Z = -2.98. P = 0.003) between 
jmgation (2006 and 2007, and nomirrigat,™ 
Three models were constructed through rnulti- 
pk Itnear regress,on analyses between nes, heigh, 
and other independent variables, and cover, water 
epth, and dry reed number entered the models 
successrvely (Table 3). Model I was best (lowe 
AIC c value) with Y= -6.088 + 0.3I2X where Y 
is nest height. X is cover (45% < cover < 100%) 
and the constant may be considered to be deleted 
,0 the equation (t = -0.838, P = 0.406). Ten of 
I I variables in 2009 (after deep irrigation) varied 
(P < 0.05) from those in 2006 and 2007 (Table 4 
indicating deeper irrigation decreased grasses and 
increased reed productivity. 
Effects ofReedbed Management.—We searched 
18 dry grass patches (5 in 2006, 8 in 2007. and ? 
in 2009). which ranged in size from 4 nr tc 
20,820 nr. Mean (± SD) patch size (MPS) was 
2,269.5 ± 5.145.2 nr’, and mean (± SD) number 
of nests/100 nr’ was 3.75 ± 6.82. The largest dn 
grass patch attracted most nests (n = 26 in 2007). 
Females were often observed searching for nes! 
materials along dikes (7 individuals/200 m) and 
hillocks, which were dominated by reeds and 
Phacelurus latifolius when the ground was 
flooded. 
The ratio of live Calamagrostis to total live 
grass was 77.7% in 2006 and 2007. and the ratio 
ol live./. gracillimus to total live grass was 84.2% 
in 2009. The ratio of nests constructed with live/ 
gracillimus to total nests was 7.5% in 2006 and 
2007. but 84.6% in 2009 (r = 30.017, df=lP 
< 0.001). This suggests the grass communin' wfl' 
changed by deeper irrigation (>30 cm). I 
gracillimus and S. planiculmis replaced C 
rpigejos in 2009. and deeper irrigation supported 
fewer males and no nests (Table 5). 
DISCUSSION 
Our data show that deep irrigation influenced 
nest placement and success of Marsh Grassbirds . 
as Calamagrostis was replaced by aquatic grasses 
Cutting of reeds also influenced nesting of Marsh 
Grassbirds. The difference in nest height between 
irrigation (2006 and 2007) and non-irrigation 
(2009) was significantly different. We infer Marsh 
use. Yalu River Estuaiy. China!^ (depCndent vanable nest height: w, is the Akaike weight), to predict Marsh Grassbird 
