We studied the relationship between variation in age and shoot charact
eristics of the host plant Salix exigua Nuttall (coyote or sandbar wil
low) and the attack and survival of Euura sp. (an unnamed leaf-midrib
galling sawfly). Variation in shoot characteristics resulted from redu
ced growth as willow ramets aged. Mean shoot length per ramet and mean
longest leaf length per shoot decreased by 95% and 50% respectively b
etween 1- and 9-year-old willow ramets. All measured shoot characteris
tics-shoot length, longest leaf length, number of leaves per shoot, an
d mean internode length-were significantly negatively correlated with
ramet age (r(2) ranged from -0.23 to -0.41). Correlations between shoo
t characteristics were highly positive, indicating that plants also gr
ew in a strongly integrated fashion (r(2) ranged from 0.59 to 0.85). F
our hypotheses were examined to explain sawfly attack patterns. The ho
st-plant hypothesis was supported in explaining enhanced larval sawfly
survival through reduced plant resistance. As willow ramets aged, the
probability of Euura sp. attack decreased over 10-fold, from 0.315 on
1-year-old ramets to 0.024 on 2- to 9-year-old ramets. As shoot lengt
h increased, the probability of sawfly attack increased over 100-fold,
from 0.007 on shoots <100 mm, to 0.800 on shoots in the 1001-1100 mm
shoot length class. These attack patterns occurred even though 1-year-
old ramets and shoots >500 mm each represented less than 2% of the tot
al shoots available for oviposition. Host plant induced mortality of t
he egg/early instar stage decreased by 50% on longer leaves and was th
e most important factor determining survival differences between vigor
ous and non-vigorous hosts. Sawfly attack was not determined by the re
source distribution hypothesis. Although shoots <200 mm contained 82%,
Of the total leaves available, they contained only 43% of the galls i
nitiated. The attack pattern also was not explained by the gall volume
hypothesis. Although gall volume increased on longer shoots, there wa
s no significant variation in mid or late instar mortality over shoot
length, as would be expected if food resources within smaller galls we
re limited. The natural enemy attack hypothesis could not explain the
pattern of oviposition since predation was greater on longer shoots an
d leaves. In addition, larval survival was related to oviposition beha
vior. Due to a 69% reduction in late instar death and an 83% reduction
in parasitism, survival of progeny in galls initiated close to the pe
tiole base was 2.8 times greater than in galls initiated near the leaf
tip. A 75% reduction in gall volume over this range of gall positions
may account for the observed increases in late instar mortality and p
arasitism.