Skógræktarritið - 15.05.2001, Qupperneq 180

died during the second instar.
Larvae that survived were signifi-
cantly heavier at the beginning of
the instar (0.75 ± 0.11 mg) than
those that died (0.68 ± 0 13;
F, |7?= 11.5, P = < 0.0009001).
Siirvival rates varied between
trees in such a way that differ-
ences were statistically signifi-
cant for the interaction between
proteins and gallotannins levels
(Table 2) with survival being
higher in trees with low levels of
both (96 %) than in those with
high levels of both (70 %).
An increment of protein concen-
tration increased mortality
(Fig. 2a), particularly when gal-
lotannin levels were low (Table
2). The increment of gallotannins
also increased mortality (Fig. 2b),
although differences were only
marginally significant when pro-
tein concentration was low
(Table 2), and non significant
when concentration of proteins
was high (Table 2).
As regards growth, the mean
increment of body mass during
the second instar was 1.79 mg
(± 0.37), and the mean larval
mass at the end of the instar was
2.54 mg (± 0.41). I did not find
significant differences between
growth of larvae feeding on
different trees, and none of the
pre-planned contrasts was
statistically significant
(P > 0.5).
100
LOW MEDIUM HIGH
GALLOTANNINS LEVEL
Fig. 2. Epirrita autumnata survivorship
observed in the greenhouse experi-
ment depending on a) proteins concen-
tration, b) gallotannins concentration.
Discussion
Life-history traits have been sug-
gested to modulate the selective
behavior of Lepidopterans (Tam-
maru and Haukioja 1996). Simple
non-selective oviposition behavior
is usually associated to
polyphagous species with non-
feeding adults and a low flight
capability of the females. Poly-
phagy decreases the risks of non-
selectivity but still there might be
a conflict between mother selec-
tion and offspring performance
Table 2. Results of the Generalized Linear Model fitted to test for differences in
survival of second instar E. autumnata larvae between trees differing in conœntra-
tion of either proteins, gallotannins, or both.
Contrast Compared trees Wald’s Ý P
Between gallotannin levels when concentration of proteins is high 7 vi. 9 0.11 0.74
Between gallotannin levels when concentration of proteins is low 21,27 vj. 4,11 5.36 0.02
Between protein concentrations when gallotannins are medium 11 vs. 29 4.15 0.042
Between protein concentrations when gallotannins are low 7 vs. 21,27 9.90 0.0017
Low proteins low gallotannins vs. high proteins high eallotannins 21,27 vs. 9 12.37 0.0004
(e.g., Nylin and )anz 1996). Larval
dispersal may contribute to allevi-
ate this conflict, and in fact bal-
looning has been linked to flight-
less (Roff 1990) and hence to the
same group of Lepidopteran
species described above. Epirríta
autumnata has been classified
among capital breeders even
when adult females can eat and fly
because they do not apparently '
do it (Tammaru and Haukioja
1996, Ruohomaki etal. 2000),
ovipositing females do not select
between host and non-host
species nor between birch trees
differing in leaf quality (Tammaru
et al. 1995). However, under labo-
ratory conditions larval perfor-
mance is affected by the individ-
ual host-tree in which they feed
(e.g., Kause et al. 1999, Lempa et
al. 2000) suggesting that individ-
ual trees differ in their quality as a
host. It rested to know whether
larvae were more prone to dis-
perse from trees where their per-
formance was worse and here 1
checked it by using the same trees
than Lempa et al. (2000). Results
from the ballooning experiment
suggested that neonate larvae
have the capability to move from
the plant in which they hatch but
this behavior is only used when
there is no food available (e.g.,
when they hatch in a non-host
plant), but not for selecting host
quality at intraspecific level.
Similar results have been found by
Harrison (1995) in Orgyia vetusta,
larvae only dispersed from dead
bushes but did not moved away
from live respond bushes differing
in their to the level of defoliation
level of alive bushes. Thus, risks
associated to this type of uncon-
trolled dispersal may preclude lar-
vae to escape from any suitable
food plant and in natural condi-
tions rates of dispersal from indi-
vidual plants would be mostly
determined by wind and micro-
habitat location (e.g., Ghent 1999,
178
SKÓGRÆKTARRITIÐ 2001 l.tbl