Resource selection by female moths in a heterogeneous

Journal of Animal
Ecology 2007
76, 854–865
Resource selection by female moths in a heterogeneous
environment: what is a poor girl to do?
Blackwell Publishing Ltd
SOFIA GRIPENBERG*, ELLY MORRIËN†¶, AILEEN CUDMORE‡, JUHAPEKKA SALMINEN§ and TOMAS ROSLIN*
*Metapopulation Research Group, Department of Biological and Environmental Sciences, PO Box 65
(Viikinkaari 1), FI-00014 University of Helsinki, Finland; †Institute of Ecological Sciences, Faculty of Earth and Life
Sciences, De Boelelaan 1087, 1081 HV Amsterdam, Vrije Universiteit Amsterdam, the Netherlands; ‡Department of
Zoology, Ecology and Plant Science, University College Cork, Distillery Fields, North Mall, Cork, Ireland; and
§Laboratory of Organic Chemistry and Chemical Biology, Department of Chemistry, FI-20014 University of
Turku, Finland
Summary
1. According to the preference–performance hypothesis, female insects select resources
that maximize offspring performance. To achieve high fitness, leaf miner females should
then adjust their oviposition behaviour in response to leaf attributes signalling high host
quality.
2. Here we investigate resource selection in Tischeria ekebladella, a leaf-mining moth of
the pedunculate oak (Quercus robur), in relation to two alternative hypotheses: (1)
females select their resources with respect to their future quality for developing larvae;
or (2) temporal changes in resource quality prevent females from selecting the best larval
resources.
3. Specifically, we test whether females show the strongest selection at the levels at
which quality varies the most (shoots and leaves); whether they respond to specific leaf
attributes (leaf size, phenolic content and conspecific eggs); and whether female
preference is reflected in offspring performance.
4. Female choice of leaves was found to be non-random. Within trees, the females
preferred certain shoots, but when the shoots were on different trees the degree of discrimination was about four times larger than when they were on the same trees.
5. While females typically lay more eggs on large leaves, this is not a result of active
selection of large leaves, but rather a result of females moving at random and ovipositing
at regular intervals.
6. The females in our study did not adjust their oviposition behaviour in response to
leaf phenolic contents (as measured by the time of larval feeding). Neither did they
avoid leaves with conspecific eggs.
7. Female choice of oviposition sites did not match patterns of offspring performance:
there was no positive association between offspring survival and counts of eggs.
8. We propose that temporal variation in resource quality may prevent female moths
from evaluating resource quality reliably. To compensate for this, females may adopt a
risk-spreading strategy when selecting their resources.
Key-words: host-plant quality, oviposition, preference–performance relationship,
spatiotemporal variation, Tischeria ekebladella.
Journal of Animal Ecology (2007) 76, 854–865
doi: 10.1111/j.1365-2656.2007.01261.x
© 2007 The Authors.
Journal compilation
© 2007 British
Ecological Society
Correspondence: Sofia Gripenberg, Metapopulation Research Group, Department of Biological and Environmental Sciences,
PO Box 65 (Viikinkaari 1), FI-00014 University of Helsinki, Finland. Tel.: +358 9191 57756. Fax: +358 9191 57694. E-mail:
[email protected]
¶Present address: Department of Multitrophic Interactions, Netherlands Institute of Ecology, PO Box 40 (Boterhoeksestraat 48),
6666 ZG Heteren, the Netherlands.
855
Resource selection
by Tischeria
ekebladella
© 2007 The Authors.
Journal compilation
© 2007 British
Ecological Society,
Journal of Animal
Ecology, 76,
854–865
Introduction
Herbivorous insects are rarely uniformly – or even
randomly – distributed in space (Denno & McClure
1983; Stiling, Simberloff & Anderson 1987; Crawley
& Akhteruzzaman 1988; Faeth 1990; Mopper &
Simberloff 1995; Eber 2004). Neither is the performance
of insects even, but varies for example among different
resource units such as individual plants or parts of
individual plants (Denno & McClure 1983; Suomela &
Nilson 1994; Roslin et al. 2006), or in concert with
certain host-plant attributes such as leaf size (Whitham
1978; Faeth 1991), leaf phenolics (Zucker 1982; Lane
et al. 1985; Beninger et al. 2004; Lahtinen et al. 2006),
or the presence of conspecifics (Denno, McClure & Ott
1995; Roslin & Roland 2005).
When host plants differ in their suitability for herbivorous insects, evolutionary theory predicts a positive
relationship between female choice of oviposition site
and the performance of the offspring (Thompson &
Pellmyr 1991). Some studies have established clear links
between female preference and offspring performance
(Leather 1985; Damman & Feeny 1988; Craig, Itami &
Price 1989; Briese 1996; Craig & Ohgushi 2002). The
existence of such preference–performance relationships
is thought to be particularly likely for insects with sessile
larval stages, such as leaf miners and gallers, where the
offspring is often confined to developing on the same
leaf that the female has chosen (cf. Mopper 1996).
However, there are also several forces counteracting
any simple coupling between female preference and
offspring performance (Thompson 1988; Mayhew 1997).
For example, when the resource quality is not constant
in time, or when host plants do not provide reliable signals
of quality at the time of female choice, females may face
a difficult task in choosing the best sites for offspring
development (Riipi et al. 2004; Ruusila et al. 2005;
Gripenberg, Salminen & Roslin 2007). Also, when females
have to take several aspects of resource quality into
account (including the direct effects on their own fitness;
Scheirs, De Bruyn & Verhagen 2000; Scheirs & De Bruyn
2002), the adaptiveness of their choice is not always
evident to the observer, although it is adaptive for the
female.
In our earlier studies of the leaf-mining moth Tischeria
ekebladella (Bjerkander) on oak trees (Quercus robur
L.), we have observed patterns of qualitative variation
in resource quality with likely implications for female
resource selection and the coupling of preference and
performance.
First, individual oak trees form mosaics of highly
variable resource quality (Gripenberg & Roslin 2005;
Roslin et al. 2006; Gripenberg et al. 2007). When
variation in a large number of host-plant traits and
measures of larval performance is partitioned among
several hierarchical levels, the differences are generally
small between tree individuals, while there is abundant
variation among shoots and leaves within trees (Gripenberg & Roslin 2005; Roslin et al. 2006). This leads us
to hypothesize that ovipositing females should select
their resources on rather small spatial scales – at the
level of shoots and leaves rather than individual trees.
In this case, we would also observe high offspring
performance on the resource units preferred by females.
Second, our observations on temporal variation in
resource quality provide us with an alternative hypothesis. Since the quality of different resource units (trees,
branches, shoots) is only moderately predictable both
within and between years (Gripenberg et al. 2007),
females might benefit from adopting a risk-spreading
strategy (den Boer 1968; Hopper 1999). They would then
scatter their eggs a little here and a little there, maximizing the chance that some offspring survive. In this case,
observing a direct coupling between female preference
and offspring performance would provide critical evidence
against this hypothesis.
Here we investigate patterns of resource selection by
females of T. ekebladella, and explicitly test the two
hypotheses presented above. We start by investigating
female choice of oviposition sites in relation to patterns
of hierarchical variation in resource quality, and in
relation to given leaf attributes. We then examine the
relationship between female preference and offspring
performance.
More specifically, we ask whether females discriminate
between trees and between shoots, and whether the
level of discrimination is roughly similar for both hierarchical levels (as predicted by our previous observations
of little variation among trees and abundant variation
among shoots within trees; Gripenberg & Roslin 2005;
Roslin et al. 2006). Given our previous observations of
increasing intraspecific competition among larvae with
increasing densities (Roslin et al. 2006), we then ask
whether females avoid leaves with conspecific eggs. To
assess what traits affect female choice, we also ask
whether females actively select leaves of a certain size
(Whitham 1978; Tuomi, Niemelä & Mannila 1981;
Simberloff & Stiling 1987; Faeth 1991), and whether
leaves selected by females differ in their chemical
profiles compared with leaves that are not selected
(Zucker 1982; Haribal & Feeny 2003). Finally, we evaluate the adaptiveness of female choice by comparing
female preference and offspring performance at the
level of both trees and shoots.
Materials and methods
 
Tischeria ekebladella is a leaf-mining moth in the family
Tischeriidae. In Finland, it is exclusively associated
with the pedunculate oak, Quercus robur. The moths fly
in June and early July, when the females lay their eggs
on the upper side of fully extended oak leaves, typically
next to a leaf vein. After approximately 3–4 weeks, the
eggs hatch and the larvae start excavating distinct white
blotch mines under the upper epidermis of the leaves
(for illustrations see Roslin et al. 2006). The larvae cease
856
S. Gripenberg et al.
Table 1. Materials used in the study
Hierarchical levels
in study design
n
No. eggs
No. initial mines
No. full-grown larvae
Leaf area
Tree pair
Tree
Bag
Shoot
Leaf
5
10
84
168
911
Experimental
No. eggs
Leaf area
Leaf
Direct observations
Experimental
No. visits on a leaf
Time spent on a leaf
No. eggs laid on a leaf
Leaf area
Female
Shoot
Leaf
9
19
110
Phenolics
Experimental
Concentrations (mg g)–1
of 23 phenolic compounds
Tree
Branch
6
20
Conspecific avoidance
Experimental
Density of conspecific eggs
No. eggs
Leaf area
Shoot
Leaf
24
215
20 trees
Observational
No. mines
No. full-grown larvae
Tree
Branch
Shoot
Leaf
20
100
1967
12128
18 trees
Experimental
No. initial mines
No. full-grown larvae
Tree
Bag
Shoot
Leaf
18
36
288
1973
Material
Type
Variables measured
Tree pairs
Experimental
10 trees
3284
A material is defined as observational if patterns of egg densities and offspring survival reflect wild, unmanipulated individuals,
and as experimental if the moths were confined to given leaves by bags or cages.
feeding in the autumn, retire to circular hibernation
cocoons inside the mine, and overwinter inside the
abscised leaves. Pupation occurs in spring inside the leaf.
Due to the close association between T. ekebladella and
the host plant lasting nearly 11 months of the year, we
expect host tree quality to be a key determinant of lifetime reproductive success. As the larvae are confined to
develop on leaves chosen by the females, we also expect
an intimate association between female preference and
larval performance.
   
© 2007 The Authors.
Journal compilation
© 2007 British
Ecological Society,
Journal of Animal
Ecology, 76,
854–865
This study combines controlled experiments and observational studies. Most data sets were collected on the
island of Wattkast (60°11′ N, 21°37′ E) in the archipelago
of south-western Finland throughout 2004–06. The
data were analysed using generalized linear mixed models
(Littell et al. 1996). Unless stated otherwise, statistical
analyses were implemented in  for  9·1,
using the GLIMMIX macro. As our study builds on
several data sets (seven) and a wealth of statistical
models (14), we have striven for maximum clarity by
summarizing sampling designs in Table 1, and specifying model structures in Table 2.
Levels of host-plant choice
The hierarchical patterns of female choice were investigated at the level of trees and shoots.
Do females discriminate more strongly among trees or
shoots?
To compare the strength of female choice at different
hierarchical levels, we conducted an experiment where
female moths were allowed to oviposit on two shoots of
variable origin. In one treatment, females were offered
shoots from different trees, whereas in another treatment, both shoots were on the same tree (material ‘tree
pairs’; Table 1).
Before the onset of the experiment, five pairs of neighbouring trees were selected on the island of Wattkast.
Each tree pair consisted of two trees standing close enough
to each other to allow shoots to be tied together. To
examine female choice among shoots on different trees,
22–24 shoots on each of these tree pairs were tied
together (using plastic-covered metal wire), forming 11
or 12 shoot pairs. Each pair of shoots was enclosed in a
20 × 30-cm muslin bag. These bags are referred to as
the ‘between-tree treatment’. To assess choice among
shoots within trees, 11 or 12 pairs of shoots were formed
within one of the trees in each tree pair. These were
enclosed in bags (referred to as the ‘within-tree treatment’).
To prevent wild females from laying eggs on leaves,
shoots in both treatments were tied together and bagged
well before the flight period of the moths.
The moths used for the experiment were collected as
larvae from trees on the island of Wattkast and surrounding islands in autumn 2003. During the winter,
larvae were stored outdoors in small muslin bags inside
cages of wire netting. In spring, mined leaves were
857
Resource selection
by Tischeria
ekebladella
Table 2. Structure of the generalized linear mixed models used to analyse the data
Link
function
Model Material
Response
Fixed effects
Random effects
1
Tree pairs
No. eggs
Treatment
Leaf area
Pair (treatment)
Log
Bag (pair treatment)
Shoot (bag pair treatment)
2
Tree pairs
No. eggs
Leaf area
Pair
Tree (pair)
Bag (pair)
Log
3
Tree pairs
Leaf area
Pair
Tree (pair)
Bag (pair)
Identity
4
Direct observations
Leaf visited or not (0/1)
Leaf area
Female
Logit
5
Direct observations
Time spent on a leaf
Leaf area
Female
Identity
6
Direct observations
Oviposition rate:
Leaf area
No. eggs laid/time spent on leaf
Female
Log
7
Phenolics
Concentration*
Treatment
Compound
Tree
Branch (tree)
Identity
8
Conspecific avoidance Density of eggs
Density of
conspecific eggs
Bag
Log
9
Tree pairs
Egg hatching:
No. initial mines/no. eggs
Pair
Tree (pair)
Bag (pair)
Logit
10
Tree pairs
Larval survival:
No. full-grown larvae/
no. initial mines
Pair
Tree (pair)
Mine density
Bag (pair)
Logit
11
20 trees
No. mines
Tree
Branch (tree)
Shoot (branch tree)
Log
12
20 trees
Larval survival:
No. full-grown larvae/
no. initial mines
Tree
Branch (tree)
Shoot (branch tree)
Logit
13
18 trees
No. mines
Tree
Bag (tree)
Shoot (bag tree)
Log
14
18 trees
Larval survival:
No. full-grown larvae/
no. initial mines
Tree
Bag (tree)
Shoot (bag tree)
Logit
No. mines
No. mines
*Square root-transformed to achieve homoscedasticity and normality of residuals.
Materials refer to entries in Table 1.
© 2007 The Authors.
Journal compilation
© 2007 British
Ecological Society,
Journal of Animal
Ecology, 76,
854–865
transferred to cylindrical hatching cages. Upon emergence between 10 June and 15 July, individuals were
sexed and immediately transferred to the bags on the
experimental trees. We transferred one female and one
male moth to each bag. To avoid overcrowding of eggs
in the bags, we checked the leaves ≈ 4 days after transplantation. If there were eggs on the leaves, the individuals were removed. This resulted in a mean count of
25·3 eggs per bag (SD = 19·0). If there were no eggs, the
individuals were returned to the bag, or (only if the moths
were dead or damaged) replaced by fresh individuals.
Hence the distribution of eggs within a bag would always
reflect the choice of a single female. We filled all bags on
one tree pair before proceeding to the next tree pair.
This was done to ensure sufficient replication within
each of the tree pairs included in the study.
When the moths had been removed from the bags,
we counted the exact number of eggs laid on each leaf.
To account for a possible effect of leaf area on female
choice, we also measured the size of each leaf using a
leaf-area meter (LI-3000A, Li-Cor, Lincoln, NE, USA).
In some bags, all leaves on one of the shoots had dropped
before the end of the experiment, and we were not able
to measure their area. These bags were discarded from
further analyses. The number of replicates in Table 1
correspond to bags in which we were able to assess leaf
areas on both shoots.
Data were modelled using model 1 (Table 2). To assess
whether females prefer one shoot over the other, we
extracted variance estimates and associated standard
errors at the shoot level, then applied Wald’s z-test to
test explicitly whether shoot-to-shoot variation added
significantly to overall variation in leaf-specific egg
counts. To test whether the preference for one shoot
over the other was similar in both treatments, variance
components associated with the shoot level were estimated separately for each treatment. We used parametric
bootstrapping in - ver. 6·1 to derive 95% confidence
858
S. Gripenberg et al.
limits for these variance components. If the variance
component of shoot in one of the treatments fell outside the confidence limits of the variance component of
shoot in the other treatment, female discrimination
among shoots was considered to be significantly stronger
in one of the treatments.
To test whether females in the between-tree treatment
on each of the tree pairs consequently prefer shoots
from one of the trees above shoots on the other tree, we
used model 2 (Table 2). Here, a significant effect of tree
would suggest that females preferred shoots from one
of the trees. We assessed pair-specific differences between
trees using t-tests of appropriate contrasts in . Finally,
to assess whether female preference for certain trees
could be due to differences in leaf areas between trees,
we built model 3 (Table 2). Pair-specific differences
between trees in terms of leaf area were assessed using
t-tests, then compared with pair-specific differences
between trees in terms of female preference (model 2).
Attributes affecting female choice
To assess patterns of female choice in relation to leaf
attributes known to affect herbivores in other systems,
we investigated egg distribution in relation to leaf size,
leaf phenolics and the presence of conspecifics.
© 2007 The Authors.
Journal compilation
© 2007 British
Ecological Society,
Journal of Animal
Ecology, 76,
854–865
Do females prefer large leaves?
The relationship between leaf area and the number of
eggs laid on a leaf was investigated using several approaches.
To explore the relationships between leaf area and egg
numbers, and between leaf area and egg density (number
of eggs laid per cm2), we produced graphical plots based
on experimental data. To assess the generality of any
patterns found, we used two independent data sets. First,
we used the ‘tree pairs’ material (cf. above; Table 1). Second,
we used data generated from an experiment involving
transplantations of moths to each of 10 bags on each of
10 experimental trees (material ‘10 trees’; Table 1). For
details on this experiment, see Gripenberg et al. (2007).
To reveal the processes behind the patterns, we conducted an independent experiment where females were
offered a set of leaves of variable size, and their oviposition behaviour directly observed (material ‘direct observations’; Table 1). Females of T. ekebladella are nocturnal,
laying most of their eggs at dusk and in the early hours
of the night. Therefore all observations were conducted
at night. On each night of observation, a small twig
with two shoots of numbered oak leaves was placed
into a cage of plexiglass (45 × 30 × 30 cm). At ≈ 10 pm,
one fertilized female was released into the cage. During
the hours of female activity we followed her movements
across the leaves and recorded the time spent on each
leaf. Afterwards we counted the number of eggs laid
on each leaf, and measured leaf areas. Altogether we
followed nine ovipositing females for over 12 h.
Data collected through direct observations of ovipositing females were summarized by three models (Table 2).
First, to test whether females are more likely to visit
large leaves than small leaves, we used model 4. Second,
to analyse whether, once a leaf is visited, the time spent
on the leaf is dependent on leaf area, we built model 5.
Third, to test whether females adjust their oviposition
rate with respect to leaf size, we built model 6. If females
moved randomly over the leaves, we would expect females
to visit large leaves more often than small ones (model 4);
females to spend more time on large leaves (model 5);
and the oviposition rate to be independent of leaf size
(model 6). If, as an alternative, females actively selected
large leaves, we would expect an increasing oviposition
rate with increasing leaf size (model 6). Finally, if females
simply treated leaves as equal units regardless of size,
we would expect no relation between leaf size and any
aspect of female behaviour (models 4–6).
Do females select leaves with respect to phenolics?
To examine differences in phenolic contents between
mined and unmined leaves, we collected leaf samples
on six oak trees in Läyliäinen, southern Finland, on 3
September 2001 (material ‘phenolics’; Table 1). In this
case, branch tips had been enclosed in muslin bags in
late May 2001 and females transplanted between 15
June and 18 July (for specific procedures see Roslin
et al. 2006). Two bagged branch tips were randomly
selected on each of two oaks, three bagged branch tips
on the other four. From each branch, we collected three
leaf samples: within each bag, we picked six leaves with
leaf mines and six leaves free of mines, and outside
the bag we randomly selected an additional six leaves.
Guidelines for sampling storage, preparation and
chemical analyses followed those of Salminen et al. (2004),
with one exception: to exclude the physical effects of
leaf mining (when specific leaf tissues are consumed
and others remain), we cut out and removed the leaf
mines themselves. To assess differences in the concentrations of phenolic compounds between treatments,
we used model 7 (Table 2).
Do females avoid conspecific eggs?
The effect of conspecific eggs on leaf selection was
studied experimentally in 2005 (material ‘conspecific
avoidance’; Table 1). On a tree with several easily accessible branches, we haphazardly selected 24 shoots with
at least eight leaves. To exclude a subset of leaves from
female oviposition, we covered approximately half the
leaves on each shoot in small bags of muslin to prevent
female oviposition. The full shoot was then enclosed in
a larger bag, into which a fertilized female moth was
released. After 2 days the moth was removed, eggs were
counted, and the small leaf bags used to control female
access was removed. A new female was then introduced
into the bag and allowed to oviposit for approximately
two nights, after which we re-counted the number of eggs
on each leaf. This allowed us to establish the number of
eggs laid by the second female. Finally, leaf areas were
measured.
To test the response of the second female to eggs laid
by the first, we built model 8 (Table 2). We excluded
859
Resource selection
by Tischeria
ekebladella
unbagged leaves on which the first female had not laid
any eggs, as these might have been perceived by females
as completely unsuitable.
Female preference vs. offspring performance
The coupling between female choice and larval performance was studied using several data sets, depending on the spatial scale considered. In all cases we
measured female preference (number of eggs laid or
number of small mines) and offspring performance on
the same resource units.
© 2007 The Authors.
Journal compilation
© 2007 British
Ecological Society,
Journal of Animal
Ecology, 76,
854–865
Do larvae perform better on preferred trees?
To study preference–performance coupling at the level
of trees, we used two data sets. First, we used data from
the between-tree treatment in the ‘tree pair’ material
(Table 1; see ‘Levels of host-plant choice’ above). Some
4 weeks after eggs had been counted, we assessed egghatching rates by counting the number of small mines
on each leaf. At this stage the bags had to be removed
from the trees in order to prevent the build-up of aphid
colonies within the bags. At the end of summer, before
leaf fall, we counted the number of living larvae on each
leaf and defined larval survival rates as the number of
living larvae divided by the initial number of mines on
a leaf.
To assess egg-hatching rates and larval survival rates
on different trees, we used models 9 and 10 (Table 2).
Differences between trees within pairs (and the direction
of the differences) were assessed through pair-specific
contrasts (t-tests). To assess the relationship between
preference and performance, we then compared the
tree-specific differences in offspring performance to
differences between trees within pairs in terms of female
preference (cf. model 2; Table 2).
At the level of trees, we also assessed the coupling of
preference and performance using survey data on wild
individuals of T. ekebladella on 20 trees, collected in the
autumn of 2004 (material ‘20 trees’; Table 1). Here the
distribution of mines was assumed to reflect female
choice. The trees were visited from 9–15 September,
and we recorded the number of leaf mines per leaf on
each of 20 shoots on each of five branches within each
tree. We also noted whether the larva inside each mine
was dead or alive, enabling us to assess larval survival rates.
For details on this material, see Gripenberg et al. (2007).
To study the relationship between preference and
performance of wild individuals at the level of trees, we
first wanted to correct for potential effects of intraspecific competition. Therefore we built models 11 and 12
(Table 2) and extracted fitted values for each tree (best
linear unbiased predictions or BLUPs; Littell et al. 1996)
for the average number of mines per leaf and for larval
survival, respectively. As the effect of mine density was
non-significant in model 12, we did not include that
effect in calculations of the BLUPs.
The relationship between the tree-specific BLUPs of
mine numbers and larval survival was investigated
using Spearman’s rank correlation coefficient. A
positive correlation between the two variables would
indicate a coupling between preference and performance: that a tree being popular among females is also
good in terms of larval survival.
Do larvae perform better on preferred shoots?
To study preference–performance coupling at the shoot
level, we used data from an experiment conducted in
the summer of 2005 (material ‘18 trees’; Table 1; S.G.
and co-workers, unpublished data). Between 21 June
and 10 July 2005, moths were experimentally transplanted to 18 trees on the island of Wattkast. Within
each tree, one female and one male moth were introduced
in each of two muslin bags (dimensions 50 × 60 cm;
mean number of leaves per bag = 54·3, SD = 27·8). Some
5 weeks after transplantation, we counted the number
of initial mines on each leaf (assumed to reflect female
preference given a high hatching rate). The bags were then
resealed to exclude natural enemies. Between 14 and 23
September we counted the numbers of living larvae on
each leaf, enabling us to assess larval survival rates.
Models 13 and 14 were fitted to the data. To adjust
for density-dependent effects, we again obtained model
predictions (BLUPs) for the number of mines per leaf,
and larval survival for each of the 288 shoots in our
material. Expected shoot-level survival was calculated
for the mean number of mines per mined leaf (average
= 2·87) across the whole material. The relationship
between the BLUPs for mine numbers and larval
survival was investigated using Spearman’s rank correlation coefficient.
Results
   
    
When given the choice among two shoots, females in
the tree-pair experiment preferred one shoot over the
other. This was the case for both within-tree treatment
and between-tree treatments, with variance estimates
being more than twice as large as their standard errors
in both groups (z = 2·1, P = 0·02; z = 3·3, P < 0·001,
respectively). Female preference was still stronger when
shoots were located on different trees (Fig. 1): when
shoots were on the same tree, variation at shoot level
explained 6% of the variation in female choice, compared with 22% of the variation when the shoots were
on different trees. This difference was statistically significant, as the confidence limits do not overlap with
the estimate means (Fig. 1).
The difference in preference between shoots could
not be attributed to corresponding differences in leaf
area: while the effect of leaf area was statistically significant (estimate = 0·03, SE = 0·003, t = 11·18, P < 0·0001),
this seemed largely due to the extreme power of the test
(df = 742). When a model including the covariate leaf
area was compared with a model excluding leaf area,
860
S. Gripenberg et al.
Fig. 1. Differences in the strength of discrimination among
two shoots located on the same vs. different trees. Proportion
of total variation in numbers of eggs per leaf attributed to the
shoot level in ‘between-tree’ and ‘within-tree’ treatments,
respectively, of the tree-pair experiment. Error bars show 95%
confidence limits.
the proportion of total variation explained by the
shoot level decreased by only 5%, from 24 to 19%. This
suggests that the preference for one shoot over another
is largely explained by factors other than leaf area.
When females were given the choice among shoots
from two different trees, there was some consistency in
their choice at the tree level: within some of the tree
pairs, females in different bags commonly preferred
shoots from one of the trees over shoots from the other
tree (Fig. 2a). However, these differences were evident
in only two of the five tree pairs. The differences between
trees could not be explained by differences in the area
of full-grown leaves: despite some overall differences in
leaf size between trees within tree pairs (F5,376 = 3·35,
P = 0·006), there were no statistically significant differences between trees in the two tree pairs where the
females were consistent in their choice (pair 1, t = 0·01,
df = 376, P = 0·99; pair 2, t = –0·98, df = 376, P = 0·33).
      

Leaf size
© 2007 The Authors.
Journal compilation
© 2007 British
Ecological Society,
Journal of Animal
Ecology, 76,
854–865
In both the large experimental data sets investigated
(‘tree pairs’ and ‘10 trees’), we found a positive relationship between leaf area and egg numbers (Fig. 3a,c),
and a negative relationship between leaf area and egg
density cm–2 (Fig. 3b,d). A qualitatively similar pattern
of egg distribution was evident following our direct
observations of ovipositing females (Fig. 3e,f ). Still,
our observational studies of ovipositing females provided no evidence for active female selection of either
large or small leaves. Overall, the females used for
direct observations of leaf selection laid 156 eggs. They
visited 57 of the 112 available leaves, and oviposited on
44 of them. Females visited large leaves more often
than small leaves (coefficient for leaf area = 0·04,
SE = 0·02, F1,100 = 5·25, P = 0·02), but this was only due
to females avoiding the very smallest leaves (Fig. 4).
Fig. 2. Preference and performance in the tree-pair experiment.
Box plots show (a) number of eggs laid on leaves on each of the
two trees in each tree pair; (b) bag-specific egg-hatching rates;
(c) bag-specific larval survival rates. Lines within boxes show
the median; boxes represent 25th and 75th percentiles;
whiskers indicate 10th and 90th percentiles; dots indicate
outliers. Statistically significant differences between trees are
indicated by asterisks: *, P < 0·05, **, P < 0·01, ***, P < 0·001.
When leaves <10 cm2 were excluded, there was no
longer any association between leaf area and whether
or not a leaf was visited (coefficient for leaf area = 0·02,
SE = 0·02, F1,77 = 0·71, P = 0·40). Once having landed
on a leaf, the time spent on the leaf was not significantly
related to leaf area (coefficient for leaf area = 0·24,
SE = 0·15, F1,47 = 2·49, P = 0·12), although there was a
weak trend for females spending more time on large
leaves (Fig. 5a). Also, the oviposition rate (number of
eggs laid on the leaf divided by time spent on the leaf )
did not vary with leaf size (Fig. 5b; coefficient for leaf
area = –0·002, SE = 0·005, F1,34 = 0·18, P = 0·68).
861
Resource selection
by Tischeria
ekebladella
Fig. 3. Numbers and densities (eggs cm–2) of eggs in relation to leaf area (cm2). (a,b)
Based on data from the ‘10 trees’ material; (c,d) data from the ‘tree pairs’ material; (e,f )
egg numbers and densities resulting from direct observations of ovipositing females
(material ‘direct observations’).
Fig. 5. Effect of leaf area on (a) total time (min) spent on a
leaf by a female; (b) oviposition rate (number of observed
oviposition events on a leaf divided by total time female spent
on leaf ). Different symbols show data points from different
females.
average phenolic contents between treatments (F2,1109 =
6·48, P = 0·0016). This, however, was due to a general
effect of bagging, with reference leaves outside the bag
differing from both mined (t1109 = –2·27, P = 0·02) and
unmined (t1109 = –3·57, P = 0·0004) leaves, but with no
significant difference between mined and unmined
leaves within bags (t1109 = 0·33, P = 0·18).
Conspecific eggs
Fig. 4. Fraction of leaves in different size categories visited by
female moths in the ‘direct observations’ material.
Females did not avoid leaves with conspecific eggs, as
the density of prelaid eggs did not have any negative
effect on the density of eggs laid by subsequent females.
Rather, different females oviposited on the same leaves
more often than explained by chance alone (estimated
effect of density of prelaid eggs on a leaf = exp(0·61
× number of eggs), SE = 0·30, F1,136 = 4·29, P = 0·04).
Leaf phenolic content
© 2007 The Authors.
Journal compilation
© 2007 British
Ecological Society,
Journal of Animal
Ecology, 76,
854–865
Female choice of leaves appeared to be independent of
the concentrations of phenolic contents measured in
early September. Twenty-three phenolic compounds
were detected in the HPLC analyses: nine hydrolysable
tannins and 14 flavonoid glycosides. The identities and
biochemical relationships between the compounds are
discussed further in another paper (Salminen et al.
2004). There were statistically significant differences in
   

Patterns of egg hatching and larval survival rates in the
paired tree experiment showed no consistency with
female choice at the level of individual trees (Fig. 2).
Hence offspring on shoots of the more preferred tree
fared no better than offspring on shoots of the less preferred tree. Similarly, our material on the distribution
862
S. Gripenberg et al.
Fig. 6. Relationship between female preference and offspring
performance (larval survival). (a) Relationship between best
linear unbiased predictions (BLUPs) for preference and
performance of wild individuals at tree level (material ‘20
trees’); (b) the same relationship for shoot level (material ‘18
trees’).
and survival of wild individuals on 20 trees provided no
evidence for a coupling between female preference and
offspring performance at the level of individual trees:
there was no correlation between the tree-specific BLUPs
for larval density and larval survival (Fig. 6a; Spearman’s rank correlation coefficient = 0·24, P = 0·32).
At the level of individual shoots, we found a weak
negative relationship between female preference and
larval performance (Fig. 6b; Spearman’s rank correlation coefficient = –0·16, P = 0·006). While statistically
significant due to the high number of observations
(n = 288), we consider this effect to be too small to be of
any real biological significance (r2 = 0·03).
© 2007 The Authors.
Journal compilation
© 2007 British
Ecological Society,
Journal of Animal
Ecology, 76,
854–865
Discussion
This study suggests that female moths face a difficult
task in selecting the best possible resources for their
offspring in a spatially and temporally variable environment. Patterns of oviposition did not mirror spatial
patterns of variation in offspring performance. Neither
did females adjust their behaviour to any of the leaf
attributes studied. When female preference and offspring performance were compared across the identical
resource units (at tree and shoot levels), there was no
detectable coupling between preference and performance.
Nevertheless, despite being apparently non-adaptive,
female resource selection showed non-random patterns
at several spatial scales.
When offered two shoots, female moths clearly preferred leaves from one shoot above leaves on the other.
This was the case both when the shoots offered to females
were on different trees, and when they were from the same
tree. Female preference for certain shoots was expected:
in a study on hierarchical variation in a number of host
attributes and several measures of larval performance,
we have previously found plenty of variation at the shoot
level (Roslin et al. 2006). Nevertheless, in the same study,
we also found differences between individual trees to be
minimal. Our observation, that female preference for
one shoot over another in the tree-pair experiment was
more than three times greater when shoots were on
different trees than when they were on the same tree,
therefore appears inconsistent with the spatial distribution of variation in leaf quality.
We do not know the reasons for the mismatch between
the spatial scales of variation in female choice and hostplant quality. One possible explanation is temporal shifts
in patterns of hierarchical variation in leaf quality. Our
previous studies have shown that, at the beginning of
the season, there are marked differences among trees in
the concentrations of many phenolic compounds. As the
season proceeds, these differences become less pronounced
(cf. decreasing standard errors in Fig. 2 of Salminen
et al. 2004). If female choice and offspring performance
are affected by leaf phenolics, the contrasting hierarchical patterns of female preference and offspring
performance therefore could reflect the shift in spatial
patterns of leaf chemistry. While the ovipositing female
would experience large tree-to-tree differences, the differences between trees would be comparatively small
during the time of larval development. Interestingly, the
two tree pairs where females showed a consistent preference for one tree over the other were those to which
females had been transplanted earliest in the season (pairs
1 and 2; Fig. 2a). This supports the idea that tree-specific
differences might appear largest early in the season. At
an interspecific level, we might then expect among-tree
differences in insect distribution and performance to be
larger for species feeding early in the season than for
species with a later phenology, such as T. ekebladella.
This prediction will be tested in future studies.
While many studies on herbivorous insects have
reported clear effects on offspring performance of leaf
size (Whitham 1978; Faeth 1991), phenolic content
(Lane et al. 1985; Beninger et al. 2004; Lahtinen et al.
2006), and the presence of conspecifics (Connor &
Beck 1993; Roslin et al. 2006), the females in our study
did not seem to respond to any of these leaf attributes.
863
Resource selection
by Tischeria
ekebladella
© 2007 The Authors.
Journal compilation
© 2007 British
Ecological Society,
Journal of Animal
Ecology, 76,
854–865
The relations were either non-existent or highly complex,
as with respect to leaf area. Here, we found more eggs
in total on larger leaves, while large leaves had fewer
eggs per cm2 than small leaves. Based on these patterns
alone, one could then argue that females prefer large
leaves, or that they prefer small leaves. Yet the most
parsimonious interpretation seems to be no preference
at all. Our experimental observations of actual female
behaviour suggest a scenario where females visit leaves
irrespective of leaf size (except for the very smallest
leaves, which are not visited). Once the female has landed,
leaf area has no major effect on the time a female spends
on the leaf. As the female also continues to lay eggs at
a constant rate irrespective of leaf size, we conclude that
females actively choose neither small nor large leaves.
The processes revealed by direct observation also illustrate that great care needs to be taken before drawing
conclusions regarding causal mechanisms based on
patterns alone.
While females showed no response to variation in
leaf size, they also showed no response to variation in
the chemical composition of leaves. We detected no difference in leaf phenolic contents between leaves that
had been oviposited on and leaves that had not been
chosen by the females. This could be a result either of
oak leaf phenolics being ecologically unimportant
and thus not influencing female ovipositional behaviour,
or of females being incapable of predicting temporal
changes in oak phenolic content. Numerous studies
have linked phenolic chemistry to herbivore performance (Ayres et al. 1997; Salminen & Lempa 2002; Park
et al. 2004), and we have no reason to doubt that the
performance of T. ekebladella could be adversely affected
by phenolic chemistry. We still have good reason to
suspect that temporal changes in oak leaf content might
prevent females from making appropriate choices
(Salminen et al. 2004). In T. ekebladella, the early larval
stage (a time when the offspring is probably particularly
sensitive to noxious compounds) occurs about 1 month
after female oviposition. Hence chemical changes occurring after eggs are laid, but before larvae start feeding,
may prevent females from making optimal choices.
If leaf attributes did not have any detectable effect
on female choice, then neither did the density of T.
ekebladella. This lack of response to conspecific eggs
suggests that intraspecific competition does not act as a
very strong agent of selection in the evolution of female
choice in our study system. Despite a consistent effect
of intraspecific competition on several measures of
larval performance, the estimated effect size is rather
small compared with constitutive variation in host-plant
quality (Roslin et al. 2006). Moreover, the conspecific
densities experienced by wild individuals of T. ekebladella are typically low, and might not often reach
situations where intraspecific competition becomes
severe (Roslin et al. 2006). The fact that the females in
our experiment often seemed to oviposit on the same
leaves suggests that there might be some other, more
important cues dictating their oviposition behaviour.
If leaf size, phenolics and the presence of conspecifics
are not affecting female oviposition behaviour, what
cues – if any – could then be guiding female choice?
In addition to the leaf traits investigated in this study,
numerous chemical and physical properties of leaves,
such as their nutritional value and toughness, could
influence patterns of oviposition (Valladares & Lawton
1991; Derridj et al. 1996; Nieminen et al. 2003; Schoonhoven, Van Loon & Dicke 2005). We suspect that the
tendency for females in some pairs of the tree-pair
experiment to prefer shoots from one of the trees above
shoots from the other tree might be at least partly due
to differences in leaf toughness, stemming from treespecific differences in phenology. Whether a trait such
as leaf toughness could also explain differences in egg
distribution at smaller spatial scales (within trees) is
unclear. As the current study does not allow us to assess
the importance of some other factors shown to influence
oviposition behaviour in other systems, such as abiotic
factors (Moore, Myers & Eng 1988; Potter 1992) and
the presence of potential competitors (Wilson & Faeth
2001), the role of those factors remains unclear.
Regardless of the exact attributes determining
offspring performance, the most striking result of this
study was a complete lack of any coupling between
female preference and offspring performance. Hence
our study adds to a large body of studies reporting no
link between female preference and offspring performance. Despite the clear and seemingly straightforward
predictions of the preference–performance hypothesis,
the empirical evidence in support of it is mixed. Sometimes female insects make ovipositional choices that
neatly translate into high larval performance (Leather
1985; Damman & Feeny 1988; Ng 1988; Craig et al. 1989;
Craig & Ohgushi 2002), but female choice seems to
be non-adaptive surprisingly often, with larval survivorship being poor on the plants actually chosen
(Auerbach & Simberloff 1989; Courtney & Kibota 1990;
Valladares & Lawton 1991; Underwood 1994; Ferrier
& Price 2004; Digweed 2006). In our study, female
choice did not match the distribution of offspring
performance at any of the spatial scales studied. An
important question emerges: is female choice really
unrelated to resource quality, or did we focus on irrelevant
measures of offspring performance (cf. Leather 1994)?
In all our assessments of coupling between preference and performance, we examined the relationship
between egg numbers and offspring survival rates. In
addition to survival, herbivore fitness could be influenced by, for example, larval growth rates and size
(Reavey & Lawton 1991). In this context, we consider
our focus on survival well justified: In the ‘18 trees’
material, model-fitted survival at the shoot level ranged
from 7 to 96%. Given such huge variation, the exact
shoot that a female chooses for her eggs may determine
either nearly complete survival or almost certain death.
At the level of trees, our inference of preference–
performance coupling is complicated by the fact that
the larvae were exposed to natural enemies throughout
864
S. Gripenberg et al.
most of their development. Thus we cannot exclude
the possibility that survival might have been higher
on preferred trees in the absence of natural enemies.
In the material used for assessing preference–performance
coupling at the level of shoots, however, larvae were
sheltered from natural enemies throughout their
whole development, and we still did not detect any
coupling between female preference and offspring
performance.
To conclude, our earlier work on T. ekebladella suggests two alternative hypotheses for the relationship
between female oviposition behaviour and offspring
performance: (1) female behaviour is adaptive and reflects
fine-scale variation in leaf quality within individual
oak crowns; or (2) temporal inconsistency in resource
quality prevents females from making adaptive choices.
In this study we found tentative support for the latter
hypothesis, but a rigorous test of the hypothesis will
require further work. The observed mismatch between
spatial patterns of female oviposition and offspring
performance and the lack of preference–performance
coupling suggests that female choice is largely independent of larval food quality. Hence female moths
seem to counter the unpredictability of the larval food
by adopting a risk-spreading strategy in their resource
selection, thereby reducing the risk of losing all offspring at the same time (den Boer 1968; Hopper 1999).
From the perspective of the plant, spatiotemporal
heterogeneity can be seen as an adaptation against
herbivory (Whitham 1983). Trees are long-lived, and
if their quality was stable in space and time, insects
could probably adapt quickly to their properties
and get an advantage in the co-evolutionary arms race
(Thompson 1994). In contrast, if trees provide resources
that are heterogeneous through space and time, herbivores might not be capable of tracking the best quality
resources. Hence the results from this study make the
evolution of fine-tuned discriminatory responses in
T. ekebladella appear unlikely. Rather, these results
provide an example of the difficulties of specializing
in a dynamic environment (Cobb & Whitham 1998;
Cronin, Abrahamson & Craig 2001; Ruusila et al.
2005).
Acknowledgements
© 2007 The Authors.
Journal compilation
© 2007 British
Ecological Society,
Journal of Animal
Ecology, 76,
854–865
The tree-pair experiment was designed based on an
idea by Dieter Ebert. We thank Bob O’Hara for
statistical advice and Otso Ovaskainen for perceptive
comments on data interpretation. Magnus Lindström
and Anna-Leena Mäkelä assisted in a pilot study on
ovipositional behaviour of T. ekebladella conducted at
the Tvärminne Zoological Station. Katja Bonnevier,
Virpi Lintunen and Riikka Kaartinen helped in the
field and Maarit Karonen conducted the chemical
analyses. Financial support by the Academy of Finland (grants 213457 and 111704), the Jenny and Antti
Wihuri Foundation, and the Oskar Öflund Foundation is gratefully acknowledged.
References
Auerbach, M. & Simberloff, D. (1989) Oviposition site preference and larval mortality in a leaf-mining moth. Ecological
Entomology, 14, 131–140.
Ayres, M.P., Clausen, T.P., MacLean, S.F., Redman, A.M. &
Reichardt, P.B. (1997) Diversity of structure and antiherbivore activity in condensed tannins. Ecology, 78, 1696–
1712.
Beninger, C.W., Abou-Zaid, M.M., Kistner, A.L.E. et al. (2004)
A flavanone and two phenolic acids from Chrysanthemum
morifolium with phytotoxic and insect growth regulating
activity. Journal of Chemical Ecology, 30, 589–606.
den Boer, P.J. (1968) Spreading of risk and stabilization of
animal numbers. Acta Biotheoretica, 18, 165–194.
Briese, D.T. (1996) Oviposition choice by the Onopordum
capitulum weevil Larinus latus (Coleoptera: Curculionidae)
and its effect on the survival of immature stages. Oecologia,
105, 464 – 474.
Cobb, N.S. & Whitham, T.G. (1998) Prevention of deme formation by the pinion needle scale: problems of specializing
in a dynamic system. Genetic Structure and Local Adaptation
in Natural Insect Populations: Effects of Ecology, Life-History,
and Behavior (eds S.S. Mopper & S.Y. Strauss), pp. 37–
63. Chapman & Hall, New York.
Connor, E.F. & Beck, M.W. (1993) Density-related mortality
in Cameraria hamadryadella (Lepidoptera, Gracillariidae)
at epidemic and endemic densities. Oikos, 66, 515–525.
Courtney, S.P. & Kibota, T.T. (1990) Mother doesn’t know
best: selection of hosts by ovipositing insects. Insect–Plant
Interactions, Vol. 96 (ed. E.A. Bernays), pp. 161–188. CRC
Press, Boca Raton, FL, USA.
Craig, T.P. & Ohgushi, T. (2002) Preference and performance
are correlated in the spittlebug Aphrophora pectoralis on
four species of willow. Ecological Entomology, 27, 529–540.
Craig, T.P., Itami, J.K. & Price, P. (1989) A strong relationship
between oviposition preference and larval performance in a
shoot-galling sawfly. Ecology, 70, 1691–1699.
Crawley, M.J. & Akhteruzzaman, M. (1988) Individual variation in the phenology of oak trees and its consequences
for herbivorous insects. Functional Ecology, 2, 409–415.
Cronin, J.T., Abrahamson, W.G. & Craig, T.P. (2001) Temporal
variation in herbivore host-plant preference and performance: constraints on host-plant adaptation. Oikos, 93, 312–
320.
Damman, H. & Feeny, P. (1988) Mechanisms and consequences of selective oviposition by the zebra swallowtail
butterfly. Animal Behaviour, 36, 563–573.
Denno, R.F. & McClure, M.S. (1983) Variable Plants in
Natural and Managed Systems. Academic Press, New York.
Denno, R.F., McClure, M.S. & Ott, J.R. (1995) Interspecific
interactions in phytophagous insects – competition
reexamined and resurrected. Annual Review of Entomology, 40,
297–331.
Derridj, S., Wu, B.R., Stammitti, L., Garrec, J.P. & Derrien, A.
(1996) Chemicals on the leaf surface, information about the
plant available to insects. Entomologia Experimentalis et
Applicata, 80, 197–201.
Digweed, S.C. (2006) Oviposition preference and larval performance in the exotic birch-leafmining sawfly Profenusa
thomsoni. Entomologica Experimentalis et Applicata, 120,
41– 49.
Eber, S. (2004) Bottom-up density regulation in the holly
leaf-miner Phytomyza ilicis. Journal of Animal Ecology, 73,
948–958.
Faeth, S.H. (1990) Aggregation of a leafminer, Cameraria sp.
nov. (Davis): consequences and causes. Journal of Animal
Ecology, 59, 569–586.
Faeth, S.H. (1991) Effect of oak leaf size on abundance,
dispersion, and survival of the leafminer Cameraria sp.
865
Resource selection
by Tischeria
ekebladella
© 2007 The Authors.
Journal compilation
© 2007 British
Ecological Society,
Journal of Animal
Ecology, 76,
854–865
(Lepidoptera, Gracillariidae). Environmental Entomology,
20, 196–204.
Ferrier, S.M. & Price, P.W. (2004) Oviposition preference
and larval performance of a rare bud-galling sawfly
(Hymenoptera: Tenthredinidae) on willow in northern
Arizona. Environmental Entomology, 33, 700 –708.
Gripenberg, S. & Roslin, T. (2005) Host plants as islands:
resource quality and spatial setting as determinants of
insect distribution. Annales Zoologici Fennici, 42, 335–345.
Gripenberg, S., Salminen, J.-P. & Roslin, T. (2007) A tree in the
eyes of a moth – temporal variation in oak leaf quality and
leaf-miner performance. Oikos, 116, 592–600.
Haribal, M. & Feeny, P. (2003) Combined roles of contact
stimulant and deterrents in assessment of host-plant quality
by ovipositing zebra swallowtail butterflies. Journal of
Chemical Ecology, 29, 653–670.
Hopper, K.R. (1999) Risk-spreading and bet-hedging in
insect population biology. Annual Review of Entomology,
44, 535–560.
Lahtinen, M., Kapari, L., Haukioja, E. & Pihlaja, K. (2006)
Effects of increased content of leaf surface flavonoids on the
performance of mountain birch feeding sawflies vary for
early and late season species. Chemoecology, 16, 159–167.
Lane, G.A., Biggs, D.R., Sutherland, O.W.R., Williams, E.M.,
Maindonald, J.M. & Donnell, D.J. (1985) Isoflavonoid
feeding deterrents for Costelytra zealandica: structure–activity
relationships. Journal of Chemical Ecology, 11, 1713–1735.
Leather, S.R. (1985) Oviposition preferences in relation to
larval growth rates and survival in the pine beauty moth,
Panolis flammea. Ecological Entomology, 10, 213–217.
Leather, S.R. (1994) Life history traits of insect herbivores in
relation to host quality. Insect–Plant Interactions, Vol. 5
(ed. E.A. Bernays), pp. 175–207. CRC Press, Boca Raton,
FL, USA.
Littell, R.C., Milliken, G., Stroup, W.W. & Wolfinger, R.D.
(1996) SAS System for Mixed Models. SAS Institute, Cary,
NC, USA.
Mayhew, P.J. (1997) Adaptive patterns of host plant selection
by phytophagous insects. Oikos, 79, 417– 428.
Moore, L.V., Myers, J.H. & Eng, R. (1988) Western tent
caterpillars prefer the sunny side of the tree, but why? Oikos,
51, 321–326.
Mopper, S. (1996) Adaptive genetic structure in phytophagous insect populations. Trends in Ecology and Evolution,
11, 235–238.
Mopper, S. & Simberloff, D. (1995) Differential herbivory in
an oak population – the role of plant phenology and insect
performance. Ecology, 76, 1233–1241.
Ng, D. (1988) A novel level of interactions in plant–insect
systems. Nature, 334, 611–613.
Nieminen, M., Suomi, J., Van Nouhuys, S., Sauri, P. &
Riekkola, M.L. (2003) Effect of iridoid glycoside content on oviposition host plant choice and parasitism in a
specialist herbivore. Journal of Chemical Ecology, 29, 823–
844.
Park, C.G., Lee, K.C., Lee, D.W., Choo, H.Y. & Albert, P.J.
(2004) Effects of purified persimmon tannin and tannic
acid on survival and reproduction of bean bug, Riptortus
clavatus. Journal of Chemical Ecology, 30, 2269–2283.
Potter, D.A. (1992) Abundance and mortality of a specialist
leafminer in response to experimental shading and fertilization of American holly. Oecologia, 91, 14–22.
Reavey, D. & Lawton, J.H. (1991) Larval contribution to
fitness in leaf-eating insects. Reproductive Behaviour of
Insects (eds W.J. Bailey and J. Ridsdill-Smith), pp. 293–329.
Chapman & Hall, London.
Riipi, M., Haukioja, E., Lempa, K., Ossipov, V., Ossipova, S.
& Pihlaja, K. (2004) Ranking of individual mountain birch
trees in terms of leaf chemistry: seasonal and annual
variation. Chemoecology, 14, 31– 43.
Roslin, T. & Roland, J. (2005) Competitive effects of the forest
tent caterpillar on the gallers and leaf-miners of trembling
aspen. Ecoscience, 12, 172–182.
Roslin, T., Gripenberg, S., Salminen, J.-P. et al. (2006) Seeing
the trees for the leaves – oaks as mosaics for a host-specific
moth. Oikos, 113, 106–120.
Ruusila, V., Morin, J.-P., van Ooik, T., Saloniemi, I., Ossipov, V.
& Haukioja, E. (2005) A short-lived herbivore on a long-lived
host: tree resistance to herbivory depends on leaf age. Oikos,
108, 99–104.
Salminen, J.-P. & Lempa, K. (2002) Effects of hydrolysable
tannins on a herbivorous insect: fate of individual tannins
in insect digestive tract. Chemoecology, 12, 203–211.
Salminen, J.-P., Roslin, T., Vienola, M., Sinkkonen, J., Pihlaja,
K. & Pulkkinen, P. (2004) Seasonal variation in the content
of hydrolysable tannins, flavonoid glycosides and proanthocyanidins in oak leaves. Journal of Chemical Ecology,
30, 1693–1711.
Scheirs, J. & De Bruyn, L. (2002) Integrating optimal foraging
and optimal oviposition theory in plant–insect research.
Oikos, 96, 187–191.
Scheirs, J., De Bruyn, L. & Verhagen, R. (2000) Optimization
of adult performance determines host choice in a grass
miner. Proceedings of the Royal Society of London Series B
– Biological Sciences, 267, 2065–2069.
Schoonhoven, L.M., van Loon, J.J.A. & Dicke, M. (2005)
Insect–Plant Biology. Oxford University Press, Oxford, UK.
Simberloff, D. & Stiling, P. (1987) Larval dispersion and
survivorship in a leaf-mining moth. Ecology, 68, 1647–1657.
Stiling, P.D., Simberloff, D. & Anderson, L.C. (1987) Nonrandom distribution patterns of leaf miners on oak trees.
Oecologia, 74, 102–105.
Suomela, J. & Nilson, A. (1994) Within-tree and among-tree
variation in growth of Epirrita autumnata on mountain birch
leaves. Ecological Entomology, 19, 45–56.
Thompson, J.N. (1988) Evolutionary ecology of the relationship between oviposition preference and performance
of offspring in phytophagous insects. Entomologia Experimentalis et Applicata, 47, 3–14.
Thompson, J.N. (1994) The Coevolutionary Process. University
of Chicago Press, Chicago, IL, USA.
Thompson, J.N. & Pellmyr, O. (1991) Evolution of oviposition behavior and host preference in Lepidoptera. Annual
Review of Entomology, 36, 65–89.
Tuomi, J., Niemelä, P. & Mannila, R. (1981) Leaves as islands
– interactions of Scolioneura betuleti (Hymenoptera)
miners in birch leaves. Oikos, 37, 146–152.
Underwood, D.L.A. (1994) Intraspecific variability in host
plant quality and ovipositional preferences in Eucheira
socialis (Lepidoptera: Pieridae). Ecological Entomology,
19, 245–256.
Valladares, G. & Lawton, J.H. (1991) Host plant selection in
the holly leaf-miner: does mother know best? Journal of
Animal Ecology, 60, 227–240.
Whitham, T.G. (1978) Habitat selection by Pemphigus aphids
in response to resource limitation and competition. Ecology,
59, 1164–1176.
Whitham, T.G. (1983) Host manipulation of parasites: withinplant variation as a defense against rapidly evolving pests.
Variable Plants in Natural and Managed Systems (eds R.F.
Denno & M.S. McClure), pp. 15–41. Academic Press, New
York.
Wilson, D. & Faeth, S.H. (2001) Do fungal endophytes result
in selection for leafminer ovipositional preference? Ecology,
82, 1097–1111.
Zucker, W.V. (1982) How aphids choose leaves – the roles of
phenolics in host selection by a galling aphid. Ecology, 63,
972–981.
Received 9 January 2007; accepted 16 April 2007