Oecologia (2003) 136:402–411
DOI 10.1007/s00442-003-1278-4
PLANT ANIMAL INTERACTIONS
Steven S. Lower · Sheril Kirshenbaum ·
Colin M. Orians
Preference and performance of a willow-feeding leaf beetle:
soil nutrient and flooding effects on host quality
Received: 13 September 2002 / Accepted: 2 April 2003 / Published online: 20 May 2003
Springer-Verlag 2003
Abstract The distribution and abundance of herbivores
on plants growing under different environmental conditions may depend upon preference and/or performance.
Soil nutrients and water availability are key determinants
of herbivore distribution, as both influence plant growth
and tissue quality. However, the effects of water on plant
quality may depend upon the availability of nutrients and
vice versa. Surprisingly few studies have examined the
interactions between the two. We investigated the effects
of soil nutrient and water availability on (1) the growth
and chemistry of the silky willow (Salix sericea Marshall), and (2) the preference and performance of the
imported willow leaf beetle (Plagiodera versicolora
Laichartig). We conducted two common garden experiments using a similar 22 fully factorial design with two
levels of soil nutrients (low, high) and two levels of water
availability (field capacity, flooded). In the first experiment (larval performance), larval development time and
pupal weight were not influenced by nutrient or water
availability to the plant. This occurred despite the fact that
plants in the high nutrient treatments had higher protein
concentration and lower foliar concentrations of the
phenolic glycoside 20 -cinnamoylsalicortin. In the second
experiment (adult preference), we caged four plants (one
from each treatment) and released beetles into cages. We
found that plant growth and leaf protein depended upon
the interaction between nutrient and water availability.
Plant growth was greatest in the high nutrient-field
capacity treatment and leaf protein was greatest in the
high nutrient-flooded treatment. In contrast, adults settled
and oviposited preferentially on the high nutrient treatment under flooded conditions, but we found no evidence
of interactions between nutrients and water on preference.
S. S. Lower · S. Kirshenbaum · C. M. Orians ())
Department of Biology,
Tufts University, Medford, MA 02155, USA
e-mail: [email protected]
Present address:
S. S. Lower, Department of Nematology,
University of California (Davis),
One Shields Avenue, Davis, CA 95616, USA
Thus, at least under flooded conditions nutrients affect
adult preference. We also found that foliar protein was
correlated positively with adult oviposition preference
and per capita egg production. Our results, then, suggest
that soil nutrients can influence adult preference, and that
adults choose high-quality hosts (high protein) that
promote egg production.
Keywords Resource availability · Nutrient water
interactions · Plagiodera versicolora · Herbivore
preference and performance
Introduction
The distribution and abundance of insect herbivores can
depend on herbivore preference and performance. Where
herbivores feed and oviposit, and how well they grow and
develop on their hosts may be influenced by several
components of plant tissue quality, including nitrogen,
water and secondary chemical content (Mattson 1980;
White 1984; Hanks and Denno 1993; Matsuki and
Maclean 1994; Kolehmainen et al. 1995). Herbivores,
however, do not always preferentially feed and oviposit
on hosts that maximize larval performance (Roininen and
Tahvanainen 1989; Denno et al. 1990; Vrieling and de
Boer 1999; reviewed by Mayew 2001). For example,
Roininen and Tahvanainen (1989) found that larvae of the
polyphagous sawfly Nematus pavidus did not perform
better on hosts preferred for oviposition by adults. There
are at least four possible explanations for the lack of
correlation. First, herbivores may be behaviorally constrained to low-quality hosts due to historical associations, i.e., phylogenetic constraints (Thompson 1998).
Second, herbivores may degrade the quality of their hosts
(Valladares and Lawton 1991). As a result, an initially
high-quality, preferred host may become an inferior one.
Third, predation pressure may force herbivores to utilize
suboptimal hosts (Denno et al. 1990; Hacker and Bertness
1995). Finally, adult preferences may be based on their
own performance rather than the performance of their
403
larvae. An important question that has received little
attention is how resource availability influences herbivore
preference and performance. Because the plant traits that
determine host preference may not be the same ones that
determine larval performance, and because resource
availability may have different effects on different traits,
resource availability may differentially influence preference and performance. In this study, we examine the
effects of resource availability (nutrients and water) on
plant traits, on host selection, and on the performance of
adult and larval leaf beetles. The availability of resources
(e.g., nutrients, water, light) is a key determinant of plant
quality and herbivore preference and performance (Tisdale and Wagner 1991; Mensah and Madden 1992; Jauset
et al. 1998; Koricheva et al. 1998; Mutikainen et al.
2000). Moreover, these factors often interact to alter plant
suitability to herbivores (Larsson et al. 1986; Coleman
and Jones 1988; Tisdale and Wagner 1991). For example,
Larsson et al. (1986) showed that light by nutrient
interactions were important determinants of tissue quality
and herbivore performance. Despite the fact that multiple
environmental factors operate to influence plant quality to
herbivorous insects, surprisingly few studies have addressed how interacting abiotic factors influence both
herbivore preference and performance.
Nutrient and water availability are known to affect
plant traits, e.g., tissue nutrients and secondary chemicals
(Gershenzon 1984; Bryant et al. 1987), that are potential
feeding and oviposition cues (Pimbert and Srivastava
1991; Bernays and Chapman 1994). Phenolic glycosides
can influence host choice in Plagiodera versicolora
(Tahvanainen et al. 1985), however there is little evidence
for correlations with leaf nitrogen (see Wait et al. 1998).
Hemming and Lindroth (1999) demonstrated that nutrient
addition increases leaf nitrogen and decreases condensed
tannin concentration in quaking aspen (Populus tremuloides). And English-Loeb (1997) demonstrated that dry
soil conditions increase phenolic compounds in tomato
and also decrease the growth of leaf-feeding Spodoptera
exigua larvae. In one of the few studies to investigate how
nutrient and water availability interact to influence
herbivores, Young and Hall (1986) showed that soil
nutrients and water stress interact to influence pupal mass
of the elm-feeding leaf beetle Xanthogaleruca luteola.
Specifically, they found that leaf beetles growing on
tissue from fertilized, well-watered seedlings had disproportionately high pupal weight compared with high
nutrient-low water, low nutrient-high water and low
nutrient-low water treatments.
Although flooding is known to decrease the concentration of nitrogen and protein in leaves (Kozlowski and
Pallardy 1984), little is known about its effects on plant
phenolics and herbivore growth and host selection.
Nutrient transport into the root generally is disrupted by
anoxic conditions created by flooding (Marschner 1986).
As a result, we would predict that plant quality does not
respond as readily to nutrient addition under flooded
conditions as under well-drained conditions. In a previous
study, we found that nutrient addition increased the
nitrogen concentration of leaves of Salix sericea Marshall
in both flooded and field capacity conditions and
increased the growth of Plagiodera versicolora Laichartig
larvae in a laboratory feeding assay (Lower 2002; Lower
and Orians 2003). However, leaf nitrogen concentration
was less responsive to nutrient addition in the flooded
than well-drained treatment. In addition, larval growth
was positively correlated with leaf N concentration.
Interestingly, in the field we have observed high densities
of beetles on flooded sites compared with nearby drier
sites. This casual observation has led us to investigate the
possibility that soil water may influence the suitability of
plants to herbivores.
In this study we investigate how the combined effects
of soil nutrient and water availability influence various
growth and chemical plant traits, and subsequent preference and performance of the imported willow leaf beetle
P. versicolora. Specifically, we quantified adult abundance and oviposition, adult performance (no. eggs/adult),
and larval growth and development on the silky willow,
Salix sericea. We addressed three major questions: (1)
How do nutrient by water interactions alter plant traits
important to herbivores? Our factorial design allowed us
to test for nutrient by water interactions. (2) Do adult
beetles settle and oviposit on the same soil nutrient and
water treatments on which their larvae perform best? (3)
Are beetle preference and performance correlated with
particular plant traits and, if so, are the correlations
consistent with the hypothesis that adults should prefer
hosts that are high-quality for the larvae? We expected
beetles to grow and develop better on plants given
additional nutrients under flooded and field capacity
conditions, because herbivorous insects generally prefer
foliage with high nitrogen concentration (Bach 1990;
Barros and Zucoloto 1999). We also expected beetles to
prefer high to low nutrient treatments since insect growth
is generally nitrogen limited (Scriber and Slansky 1981).
In previous work we found that nutrient addition
increased growth rates of P. versicolora (Lower 2002;
Lower and Orians 2003). However, we expected flooding
to limit the effect of increased nutrient addition and
therefore predicted that the difference in adult preference
between high and low nutrient treatments should be
smaller under flooded conditions. Finally, we expected
leaf nitrogen/protein to be positively correlated with both
performance and preference.
Materials and methods
Study system
The silky willow, Salix sericea Marshall (Salicaceae) is a shrubby
wetland species that has an extensive range in the northeastern USA
and eastern Canada (Argus 1986). It is most often found growing in
freshwater swamps and along riverbeds and streams, but it can also
occasionally be found on relatively drier upland sites (personal
observation). Under favorable growing conditions, a mature plant
can reach 3 m in height and consist of up to 20 or more individual
stems. The mode of reproduction and well-characterized secondary
chemistry of S. sericea make it an ideal study system for
404
investigating how plant chemical traits influence herbivore preference and performance. S. sericea is a dioecious species that
reproduces sexually by means of plumose seeds and asexually by
suckers. For this study we took advantage of the ability of S. sericea
to reproduce asexually by making cuttings to generate replicate
clones of a single genotype. S. sericea produces the phenolic
glycosides salicortin and 20 -cinnamoylsalicortin in concentrations
as high as 10% dry leaf mass (Nichols-Orians et al. 1993). Phenolic
glycosides have been shown to affect feeding patterns and growth
of insect herbivores (Tahvanainen et al. 1985; Lindroth 1989).
The imported willow leaf beetle, Plagiodera versicolora, was
introduced to the United States from Europe at the beginning of the
twentieth century and is a specialist on species in the Salicaceae.
Larvae feed communally in groups of 10–25 for the first two instars
then separate to feed individually or in smaller groups for the
remainder of larval development. At the end of the third instar,
larvae attach themselves to the underside of a leaf and pupate.
Adults are between 4–7 mm in length, with the females 1/4–1/3
heavier than males. The feeding preference of adult P. versicolora
is influenced by foliar chemistry (Tahvanainen et al. 1985; Orians
et al. 1997; Crone and Jones 1999).
The experimental plants
In March 2000, we generated 18 cm-cuttings from a single clone
(S53, female) originally derived from a population growing near
Oneonta, N.Y. Use of a single clone allowed us to focus on effects
of resource availability by reducing genetic variation. In previous
work in our laboratory, this clone showed phytochemical responses
to nutrients and flooding that were similar to two other clones
(Lower 2002). One end of each cutting was scored lightly several
times with a paring knife, dipped in rooting hormone (Rootone) and
placed in a rootcone filled with potting mix consisting of peatmoss
and vermiculite (Whittmore). We grew plants on a greenhouse
bench under a 16:8 h light: dark cycle and fertilized them once per
week with Peters 20-20-20 plant food (N-P-K, 10 ml of 0.5 g/l
fertilizer). On 31 May, 200 plants were transplanted into 4-l pots
and transferred to an open field site at the Waltham Experimental
Field Station (University of Massachusetts) in Waltham, Mass. The
potting mixture consisted of loam, peat moss and fine vermiculite
(Whittmore) in a 3:2:1 ratio by volume. All plants were watered
once per day until the initiation of experimental treatments, which
began after 3 weeks of growth in the field.
Experimental treatments
On 20 June we created 4 treatments by growing potted willows at 2
levels of nutrient availability (high and low) and 2 levels of water
availability (field capacity and flooded). Plants in the high nutrient
treatment received 100 ml of fertilizer solution (1.25 g/l Peters 2020-20) every 3–4 days; plants in the low nutrient treatment did not
receive supplemental nutrients. We watered all plants 3 times daily
using an automatic drip-line irrigation system. The flooded
treatment was created by lining pots with heavy-duty plastic,
which prevented drainage from the pots. We watered plants in the
field capacity treatment by running drip lines to the saucers
underneath the pots. This allowed water to wick up into the soil to
reduce nutrient leaching. Clear plastic saucers with a hole in the
middle for the stem covered the tops of all pots to shield the soil
from rain. We applied the treatments for 4 weeks before the
performance and 5 weeks before preference experiments were
begun (see below).
We conducted two separate experiments to test the effects of the
nutrient and water treatments on adult preference and larval
performance. The first experiment was designed to simultaneously
test the effects of nutrients and water on plant chemistry and larval
performance. Plant chemistry was measured on one shoot and
larval performance was assessed on a second shoot. This allowed us
to correlate larval performance with specific plant traits. The
second experiment examined adult feeding and oviposition in cages
containing plants from all four treatments. Because we found a
significant correlation between protein concentration and larval
performance in experiment 1, we decided to measure protein
concentration in experiment 2. This allowed us to test whether
adults also chose plants containing high concentrations of protein.
In addition, the results of these two experiments were compared to
determine whether adults preferred the treatments on which larvae
performed best.
Experiment 1. Larval performance
Sixty plants, 15 from each of the 4 treatment groups, were
randomly assigned to 60 positions in a common garden. The design
was as follows: 2 nutrient 2 water 15 replicates =60 plants.
Plant traits
We measured plant traits on a single shoot per plant and larval
performance on a second shoot on the same plant. We measured
three plant growth traits: total above-ground dry weight, total leaf
dry weight and leaf expansion rate. We harvested above-ground
plant parts at the end of the experiment (13 August), dried at 60C
for 48 h and separately weighed leaves and stems. We measured
leaf expansion on a single leaf per plant by marking a 1- to 2-cmlong first-formed leaf (first identifiable leaf below the meristem)
and measuring its change in length over a 7-day period. At 7 days
the fastest growing leaves had just reached full expansion. Leaf
expansion rate (cm/day) was measured as the increase in length
(cm) per day from the leaf tip to the base of the lamina. In addition,
we measured five traits likely to influence leaf beetle feeding and
growth: leaf water content, leaf nitrogen concentration, leaf protein
concentration and the concentration of the phenolic glycosides
salicortin and 20 -cinnamoylsalicortin. Samples for these five traits
were collected 1 day before feeding assays (19 July).
Elemental nitrogen was determined on 15€0.5 mg ground leaf
material with an NC 2500 CE Elantech C/N Analyzer (Thermoquest Instruments). Percent protein concentration was estimated on
four first fully expanded leaves using the BioRad Colorimetric
Protein Assay. We extracted 3.0€0.2 mg ground dry leaf material in
1.5 ml 0.1 M NaOH for 2 h at 100C. The protein extracts were
combined with the BioRad reagent (Coomassie Brilliant Blue) in
96-well microtiter plates and their absorbances measured at 595 nm.
We used bovine serum albumin as a standard.
We measured concentrations of the phenolic glycosides
salicortin and 20 -cinnamoylsalicortin using high performance liquid
chromatography (HPLC) (Orians 1995). Approximately 10 mg of
leaf powder was extracted in 1.0 ml cold methyl alcohol (0C) for
13 min under sonication. We centrifuged and filtered (0.45 mm) the
extracts into crimp-top vials and stored them at –20C. Analyses
were performed on a Hewlett-Packard 1100 Series HPLC within
24 h of extraction. For the analyses we used a reverse-phase NovaPak C18 column (Waters) and a gradient system of distilled water
and methyl alcohol. We used 1, 3-dimethoxybenzene as our internal
standard.
Beetle performance
We conducted feeding assays by placing approximately ten firstinstar larvae on each of the 60 plants on 20 July. Larvae from a
cluster of eggs laid by a single female were transferred as an intact
group on a leaf, which was then glued with superglue to the adaxial
surface of a first fully expanded leaf. We covered the branch on
which we placed larvae with a nylon mesh bag (tulle, length =0.5 m,
diameter=15 cm, mesh size =11 mm) to exclude predators.
However, upon dissection, 12% of larvae were found to be
parasitized by a small parasitic wasp (Schizonotus sp.); the
treatments did not differ in parasitism rate. Parasitized larvae were
excluded from the statistical analyses of larval performance. Plants
were inspected daily for pupated larvae. We removed the pupae
405
experiment (August 6) and dried them at 60C for 48 h. Although
plants were harvested after the adults had fed (see below), total
consumption was minimal and unlikely to have altered treatment
differences in leaf dry weight (personal observation).
Protein concentration was estimated using the BioRad Colorimetric Protein Assay as in experiment 1.
from the plants and stored them in glass vials at 25C. The time of
development from 1st instar to pupation was recorded and pupal
weights obtained. We allowed the pupae to emerge as adults and
preserved them in alcohol for later dissection to determine sex. Due
to sexual size dimorphism, we analyzed pupal weight separately for
males and females. We calculated mean pupal weight and larval
development time per plant, including only individuals that
survived to adulthood.
Beetle abundance and oviposition
We started a laboratory colony of P. versicolora in early June from
approximately 250 adults and larvae collected from the Hinsdale
Flats Wildlife Preserve in Hinsdale, Mass. Beetles were maintained
in petri dishes on a 16:8 h light/dark cycle at 25C in a growth
chamber. Every second or third day, they were provided with fresh
leaves from Salix eriocephala, a shrubby species that does not
contain phenolic glycosides.
On 25 July we released 20 adult beetles from a petri dish placed
at the center of each cage; the petri dish was equidistant from the
four experimental plants. The 20 beetles in each cage consisted of
15 large and 5 small beetles (females are visibly larger than males).
This maintained a similar sex ratio among cages. Beetles had the
opportunity to mate before being placed in the cages.
We recorded the number of beetles and eggs on each treatment
every second day for 7 days. Abundance may have been influenced
by plant size, therefore we reported both the absolute number of
beetles and eggs on each plant as well as the number of beetles and
eggs on a per gram dry leaf basis. We did this since larger plants
had a higher likelihood of intercepting beetles. Although we did not
quantify leaf damage, it was obvious from visual inspection that
plants with more beetles had more damage. We also calculated the
number of eggs laid per beetle using the last beetle count on day 5
and the last egg count on day 7. We used counts from the final dates
Experiment 2. Adult preference
Experimental set-up
On a separate set of plants we tested adult leaf beetle preference
using a 4-choice design. The same treatments used in the
performance experiment were also used to test the effects of
nutrients and flooding on adult preference. On June 19, we arranged
the plants in groups of 4 within a common garden; plants were
chosen randomly from each treatment (high nutrients-flooded, high
nutrients-field capacity, low nutrient-flooded and low nutrientsfield capacity). The following fully factorial design was used: 2
nutrient 2 water 20 replicates =80 plants. Each group of 4 plants
was placed in a 1 m 1 m 1.3 m PVC and nylon mesh (tulle) cage
in order to reduce predation. The nylon mesh reduced photon flux
density (mmol m2 s1) by approximately 15%. Plants were
positioned to avoid contact with each other.
Plant traits
We measured total leaf weight to control for the effects of plant size
on beetle preference. We harvested the plants at the end of the
Table 1 ANOVAs of the effects of nutrient and water treatments
on plant traits and herbivore performance. Significant effects are in
bold and comparisons among treatments are shown for significant
effects. Degrees of freedom are given in order water, nutrients,
nutrient water interaction and error
Effect
df
Water
Plant traits
Leaf weight (g)
Total stem weight (g)
Leaf expansion (cm/day)
% Leaf nitrogen
% Leaf protein
Salicortin(mg/g)
20 -Cinnamoylsalicortin (mg/g)
Nutrients
Nutrients Water
F
P
F
P
F
P
90.14
34.11
0.09
0.60
3.60
1492
0.22
<0.001
<0.001
0.761
0.442
0.05
<0.001
0.644
420.27
197.18
9.97
390.02
168.75
0.03
9.18
<0.001
<0.001
0.003
<0.001
<0.001
0.865
0.004
14.85
3.26
0.16
19.74
14.85
12.37
25.38
<0.001
0.007
0.960
<0.001
0.007
<0.001
<0.001
Table 2 Plant characteristics
(means and standard errors) in
two water treatments (flooded
and field capacity) crossed by
two nutrient treatments (high
and low)
Comparison
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
56
56
56
56
56
56
56
HN>LN; FC>FLD
HN>LN; FC>FLD
HN>LN
HN>LN
HN>LN
FC>FLD
LN>HN
Water treatments
Field capacity
Flooded
Nutrient treatments
Plant traits
Leaf weight (g)
Total stem weight (g)
Leaf expansion (cm/day)
% Leaf nitrogen
% Leaf protein
Salicortin(mg/g)
20 -Cinnamoylsalicortin (mg/g)
Low
High
Low
2.96 (0.14)
6.51 (0.29)
469 (0.52)
1.94 (0.03)
7.56 (0,16)
77.31 (1.43)
16.72 (0.74)
11.99 (0.66)
22.70 (1.27)
655 (0.59)
2.68 (0.03)
9.36 (0.23)
69.49 (1.92)
11.98 (0.45)
2.13
5.68
4.24
1.77
7.40
62.35
13.49
High
(0.08)
(0.21)
(0.46)
(0.04)
(0.17)
(2.85)
(0.74)
5.66
11.54
6.83
2.93
10.74
68.70
14.66
(0.39)
(0.84)
(0.80)
(0.08)
(0.29)
(1.46)
(0.78)
406
as estimates, since the majority of eggs were laid between days 5
and 7.
Statistical analysis
We used a two-way analysis of variance (ANOVA) to test for
effects of nutrient and water treatments on plant growth and
chemistry, and adult preference and larval performance (Proc GLM
procedure in SAS, 2000), with nutrients and water treated as fixed
effects Square-root and natural log transformations were used
where data did not fit the assumption of homogeneity of variances
and percentage data were arcsine square-root transformed (Zar
1996). We employed Pearson product-moment correlations to test
correlations of plant traits with beetle preference and performance.
The step-up procedure was used to correct for multiple correlations
(Hochberg and Hommel 1988).
Results
Experiment 1. Larval performance
Plants
Leaf weight was highest in the high nutrient-field
capacity treatment (nutrient water interaction) (Tables
1, 2). In the field capacity treatment, nutrient addition led
to a 305% increase in leaf weight over a 10 week period,
whereas the flooded treatment had a smaller increase of
166%. Total plant above-ground weight showed a nearly
identical pattern. Leaf expansion rate increased only in
the high nutrient treatment, and there was no interaction
between treatments.
Percent leaf N was higher in the high nutrient
treatments than in the low nutrient treatments, but there
was no overall effect of water availability (Tables 1, 2). A
nutrient water interaction resulted from the fact that
flooding reduced leaf N concentration compared with
field capacity within the low nutrient treatment, but
increased it in the high nutrient treatment. Leaf protein
concentration also increased with nutrient addition and
showed a similar, less pronounced interaction (Tables 1,
2). However, there was a marginally significant increase
in protein concentration in the flooded treatment that was
not present for leaf N concentration. Leaf N and protein
were positively correlated (r=0.89, P<0.001).
Both water and nutrient treatment altered phenolic
glycoside concentration (Tables 1, 2). Salicortin concentration was higher in the field capacity treatments and
there was a nutrient water interaction. Nutrient addition
reduced salicortin concentration in the field capacity
treatment, but it increased in the flooded treatment. 20 Cinnamoylsalicortin showed a similar pattern, with fertilization resulting in a 39% reduction in the field capacity
treatment and a 9% increase in the flooded treatment.
There was no effect of water treatment on 20 -cinnamoylsalicortin concentration, but there was an overall reduction in the high nutrient treatments.
Fig. 1 Female pupal weight (A), male pupal weight (B) and
development time (male + female) (C) in experiment 1 (Performance) on low and high nutrient treatments within field capacity
and flooded soil conditions. ANOVA results are given above the
graphs and bars =1 SEM
Beetles
Neither development time nor pupal weight was influenced by nutrient and water treatments (Fig. 1). After
correction for multiple comparisons using the step-up
procedure, there were no significant correlations between
plant traits and larval performance: protein and pupal
407
Fig. 2 The mean total dry leaf weight (A) and leaf protein
concentration (B) in experiment 2 (Preference) on low and high
nutrient treatments within field capacity and flooded soil conditions. ANOVA results are given above the graphs and bars =1 SEM
weight (r=0.34, P=0.30, n=32), protein and development
time (r=0.33, P=0.14, n=48), salicortin and pupal weight
(r=0.20, P=0.26, n=32), and salicortin and development
time (r=0.20, P=0.46, n=48).
Experiment 2. Adult preference
Fig. 3 Number of beetles per plant (A) and number of beetles per
gram dry leaf weight (B) in experiment 2 (Preference) on low and
high nutrient treatments within field capacity and flooded soil
conditions. ANOVA results are given above the graphs and bars =1
SEM
Leaf protein concentration was elevated in the high
nutrient and in the flooded treatments (Fig. 2B). There
was an interaction resulting from the fact that the
difference between the high and low nutrient treatments
was 13% under flooded conditions but only 1% in field
capacity conditions.
Plant traits
Beetle abundance and oviposition
Leaf dry weight was greater at high than low nutrient
availability, and greater in field capacity than in flooded
soil conditions (Fig. 2A). There was also a nutrient water interaction, which was a result of the extreme
enhancement of growth from nutrient addition in the field
capacity treatment compared with the flooded treatment.
In the field capacity treatment, nutrient addition resulted
in a 319% increase in dry leaf weight, whereas only a
68% increase was found in the flooded treatment.
The number of beetles per plant was greater on the high
nutrient treatments than the low nutrient treatments by
approximately 300% (Fig. 3A). Since plant size may
affect the attractiveness of plants in each treatment, we
also present beetle abundance expressed as the number of
beetles per gram dry leaf weight. In this analysis, only the
high nutrient-flooded treatment had high beetle abundance (Fig. 3B). Nutrient addition increased the number
408
tion (r=0.45, P=0.003, n=80) and egg production (eggs/
adult) (r=0.26, P=0.046, n=80).
Discussion
Soil nutrient and water treatments did not influence larval
performance, but appeared to influence adult preference.
Adult beetles clearly preferred fertilized plants to nonfertilized plants under flooded soil conditions. Although
we did not find a correspondence between adult preference and larval performance for the different treatment
combinations, we did find that adults settled and
oviposited on hosts with high protein concentrations
(positive correlation between abundances and leaf protein) and that these were the same plants on which adult
egg production was highest (positive correlation between
no. eggs/beetle and leaf protein). We predicted that plant
quality to herbivores would be less responsive to nutrient
addition under flooded than field capacity conditions.
Neither larval performance, which was not affected by
treatments, nor adult preference, which differed at least as
much in flooded as field capacity conditions, supports this
hypothesis. Moreover, we found that leaf nitrogen and
protein increased more due to nutrient addition in the
flooded than in the field capacity treatment.
Larval performance
Fig. 4 Number of eggs per plant (A) and number of eggs per gram
dry leaf weight (B) in experiment 2 (Preference) on low and high
nutrient treatments within field capacity and flooded soil conditions. ANOVA results are given above the graphs and bars =1 SEM
of beetles by 144% in the flooded but only by 4% in the
field capacity treatment when abundance was measured
per gram dry leaf weight.
Similarly, the number of eggs per plant was greater on
the high nutrient treatments by approximately 430%
(Fig. 4A). As found with beetle number, the number of
eggs/gram leaf mass increased dramatically in the flooded
(210%) but only slightly in the field capacity (15%)
treatment.
In contrast to beetle and egg number, the number of
eggs per beetle only showed a marginally significant
increase in the high nutrient compared with the low
nutrient treatment (F=3.84, P=0.055). Water treatment
did not influence per capita egg production and there were
no nutrient water interactions.
We found that leaf protein was positively correlated
with beetle abundance (r=0.33, P=0.006, n=80), oviposi-
In previous work we have shown P. versicolora larvae
raised in the laboratory on foliage from fertilized plants
attained higher pupal weights than larvae raised on
unfertilized plants (Lower 2002; Lower and Orians 2003).
In contrast, Wait et al. (1998) found that nutrient addition
to Populus deltoides had no effect on Plagiodera
versicolora feeding rate, although it increased the feeding
rate of the leaf beetle Chrysomela scripta. Our present
results and Waite et al.’s results are consistent with the
notion that the growth rates of P. veriscolora larvae may
be less responsive to plant nutrients than those of other
insects. An alternative explanation, however, is that
larvae feeding on nutritionally inferior hosts may compensate by increasing their feeding rate. The fact that our
earlier work showed an effect on larval performance
suggests that other factors in the field, whether differences in environmental variables or beetle behavior, may
swamp out the effects of nutrients.
Adult preference
Our experiment was designed to test whether nutrient and
water have interactive effects on preference. Although the
combined effects of soil nutrients and water on insect
herbivores have received little attention (but see Young
and Hall 1986; Estiarte et al. 1994), nutrients and water
have been shown to individually influence herbivore
preference. The positive effect of nutrient addition in
409
flooded conditions that we found is consistent with
numerous other studies. For example, nutrient addition
has been shown to increase oviposition by cabbage
butterflies and whiteflies on their hosts (Letourneau and
Fox 1989; Jauset 1998). Soil fertility also appears to
influence herbivore abundances in field settings. In a field
experiment, Orians and Fritz (1996) found that nutrient
addition to potted willows increased the abundance of a
suite of arthropod herbivores on S. sericea, including
skeletonizing leaf-chewing insects such as P. versicolora.
Similarly, fertilized oak trees have been shown to support
higher densities of herbivorous arthropods from different
feeding guilds (Forkner and Hunter 2000).
The effects of flooding on feeding and oviposition
choices of herbivorous insects are poorly understood. In
one of the few studies of the effects of flooding on
herbivores, Watt (1986) found no effect of waterlogging
on oviposition preference of a pine-feeding moth. We
found that salicortin concentration was particularly low
under flooded low nutrient conditions in experiment 1
(Performance) and leaf protein and/or N were particularly
high in the high nutrient-flooded treatment in both
experiments 2. This suggests that flooding influences
plant traits related to the nutritional quality of host plants
to herbivores. Our results on the effects of flooding on
adult preference are not definitive, however one set of
results does suggest that high nutrient-flooded plants may
in fact be preferred to the other three treatments. Thus,
nutrients and water may interact to influence adult
preference. Whether nutrient by water interactions are
relevant in the field is still unclear. A question that
remains to be answered is whether field sites with high
beetle density have both high soil nutrient availability and
high water availability. Biotic factors such as predation
and other abiotic factors such as temperature may also
influence leaf beetle abundance in the field (for discussion see Sipura 1999).
We demonstrate that adults prefer fertilized to nonfertilized hosts in both field capacity and flooded
conditions on a per plant basis (Fig. 3A). However, plant
size may have confounded results when abundance was
measured on a per plant basis, since plants in the high
nutrient-field capacity treatment were 4–5 times larger
(leaf dry weight) than plants in the other treatments. As a
result, beetle interception by plants may have been higher
on these larger plants. We attempted to control for plant
size by expressing abundance as number of beetles or
eggs per gram dry leaf weight. When we measured
abundance in this way we found that only the high
nutrient-flooded treatment had elevated numbers of egg
and beetle, indicating that nutrients and water may
interact to influence adult abundance. An important
caveat is that expressing abundance on a per leaf weight
basis may favor smaller plants too heavily. Consequently,
determining whether this interaction is real, i.e., whether
nutrient addition has a stronger effect on abundance under
flooded than field capacity conditions, will require further
experiments in which plant size is controlled. Although it
is disputable whether nutrients and water interact to affect
abundance, what is unequivocal is that adult beetles prefer
fertilized to unfertilized plants under flooded conditions.
Regardless of how beetle and egg abundances were
measured, they were 2–3 times higher in the high
nutrient-flooded than in the low nutrient-flooded treatment.
Do adult beetles choose high-quality hosts?
Plagiodera versicolora preference is affected by soil
nutrients—adult beetles preferred the high nutrient treatment under flooded conditions. The high leaf protein
concentration in the high nutrient-flooded treatment may
explain preferential adult utilization. Adult performance,
measured as per capita egg production, was high on these
same hosts. However, these treatment differences in
preference were not accompanied by treatment differences in larval performance. Thus, adults choose plants
with high protein and on which they perform well (high
egg production), while their host choice appears unrelated
to larval performance.
Although we found positive correlations between leaf
protein and adult beetle abundance/oviposition, they only
explained a small percentage of the variation (10%–20%).
What might explain the relatively low correlation? In
experiment 2 (Preference), leaf protein concentration was
only 10% higher in the high nutrient-flooded than in the
high nutrient-field capacity treatment, but there was over
a 190% difference in beetle abundance and 225%
difference in oviposition preference between the same
two treatments. Perhaps herbivores respond to small
differences in leaf protein. Alternatively, other traits such
as soil water availability could contribute to differential
adult preference (sensu Sipura et al. 2002). This experiment does not allow us to rule out one or the other of
these alternatives, but the results do suggest that adult
beetles are able to distinguish between hosts growing in
different soil nutrient conditions based on tissue quality.
Other leaf traits have also been implicated in feeding
and oviposition preference, including leaf water content,
defensive compounds such as phenolic glycosides, and
amino acids (Behmer and Joern 1993; Bernays and
Chapman 1994; Kolehmainen et al. 1995; Justus and
Mitchell 1996). We found that salicortin concentration
was lowest in the low nutrient-flooded (62.35 mg/g) and
highest in the low nutrient-field capacity (77.31 mg/g)
treatment in experiment 1 (Performance), while preference for these same two treatments was nearly indistinguishable in experiment 2 (Preference). Thus, salicortin
itself does not appear to be a good predictor of adult
abundance. Nonetheless, salicortin may play a role in
determining host preference through its combined effect
with leaf N. Tahvanainen et al. (1985) demonstrated that
P. versicolora prefers host species with moderate to low
levels of phenolic glycosides. Perhaps beetles prefer highnutrient plants with intermediate concentrations of salicortin and high protein concentrations. Salicortin concentration is relatively low and N concentrations is high
410
in the high nutrient treatments, a combination of character
states that may make these treatments preferred over the
low nutrient treatments. Still, from our results it is not
clear whether the variability in salicortin concentration
among treatments was sufficient to influence adult
preference. We feel this deserves further study, although
we recognize that other traits such as plant volatiles also
may have affected adult preference (Bernays and Chapman 1994; Takabayashi et al. 1994).
Conclusions
Our results support the notion that adult beetles make
oviposition decisions that benefit themselves by choosing
hosts with high protein concentration. However, we did
not find evidence that adults preferred nutrient and water
treatments on which larvae performed best. In fact, larval
performance was not affected by treatments. Soil nutrient
and water content, then, may not be good predictors of
how well larvae will perform in the field. Adult beetle
preference, on the other hand, responded strongly to
nutrients in flooded soil conditions. This implies that (1)
nutrient availability may potentially be an important
factor influencing herbivore distribution in the field and
that (2) preference and performance of adult beetles may
play a larger role than larval performance in determining
distribution and abundance. This study focused on the
short-term responses of young plants, therefore whether
our results apply to mature plants in field, where
numerous factors can potentially influence insect populations, will require further investigation. Specifically we
suggest that future work examine whether soil nutrients
and water influence beetle abundance amid the myriad
other factors that operate on herbivore populations in the
field.
Acknowledgements We are grateful to Dr. Michael Reed and
Durwood Marshall for statistical advice. We thank Dr. Frances
Chew for assistance with cage design. The laboratory of Dr. Adrien
Finzi at Boston University provided facilities and technical support
for nitrogen analyses. We thank the Tufts University Department of
Biology for support and facilities. We are grateful to Lisa
Tewksbury (University of Rhode Island) and the Systemic Entomology Laboratory (Baltimore, Md.) for identification of Schizonotus. This work was supported by the EPA STAR Fellowship
Program, the Draupner Ring Foundation, the Howard Hughes
Biomedical Institute and the National Science Foundation
(deb9981568). Dr. George Ellmore, Dr. Sara Lewis, Dr. Greg
English-Loeb, Dr. Margret Van Vuuren, Megan Griffiths, Benjamin
Babst, Brian Brannigan and two anonymous reviewers gave
valuable suggestions on the manuscript.
References
Argus GW (1986) The genus Salix (Salicaceae) in the southeastern
United States. Syst Bot Monogr 9:1-70
Bach CE (1990) Plant successional stage and insect herbivory: flea
beetles on sand-dune willow. Ecology 71:598–609
Barros HC, Zucoloto FS (1999) Performance and host preference of
Ascia monuste (Lepidoptera, Pieridae). J Insect Physiol 45:7-14
Behmer ST, Joern A (1993) Diet choice by a grass-feeding
grasshopper based on the need for a limiting nutrient. Funct
Ecol 7:522–527
Bernays EA, Chapman RF (1994) Host-plant selection by phytophagous insects. Chapman and Hall, New York
Bryant JP, Clausen TP, Reichardt PB, McCarthy MC, Werner RA
(1987) Effect of nitrogen fertilization upon the secondary
chemistry and nutritional value of quaking aspen Populus
tremuloides Michx. leaves for the large aspen tortrix Choristoneura conflictana Walker. Oecologia 73:513–517
Coleman JS, Jones CG (1988) Plant stress and insect performance:
Cottonwood ozone and a leaf beetle. Oecologia 76:57–61
Crone E, Jones CG (1999) The dynamics of carbon-nutrient
balance: effects of cottonwood acclimation to short- and longterm shade on beetle feeding preferences. J Chem Ecol 25:635–
656
Denno RF, Larsson S, Olmstead KL (1990) Role of enemy-free
space and plant quality in host-plant selection by willow
beetles. Ecology 71:124–137
English-Loeb G, Stout MJ, Duffy SS (1997) Drought stress in
tomatoes: changes in plant chemistry and potential nonlinear
consequences for insect herbivores. Oikos 79:456–468
Estiarte M, Filella I, Serra J, Penuelas J (1994) Effects of nutrient
and water stress on leaf phenolic content of peppers and
susceptibility to generalist herbivore Helicoverpa armigera
(Hubner). Oecologia 99:387–391
Forkner RE, Hunter MD (2000) What goes up must come down?
Nutrient addition and predation pressure on oak herbivores.
Ecology 81:1588–1600
Gershenzon J (1984) Changes in the levels of plant secondary
metabolites under water and nutrient stress. Rec Adv Phytochem 18:273–320
Hacker SD, Bertness MD (1995) A herbivore paradox: Why salt
marsh aphids live on poor quality hosts. Am Nat 145:192–210
Hanks LM, Denno RF (1993) Natural enemies and plant water
relations influence the distribution of an armored scale insect.
Ecology 74:1081–1091
Hemming JDC, Lindroth RL (1999) Effects of light and nutrient
availability on aspen: growth, phytochemistry, and insect
performance. J Chem Ecol 25:1687–1714
Hochberg Y, Hommel G (1988) Step-up multiple testing procedures. Encycl Stat Sci Suppl vol 2
Jauset AM, Sarasua MJ, Avilla J, Albajes R (1998) The impact of
nitrogen fertilization of tomato on feeding site selection and
oviposition by Trialeurodes vaporariorum. Entomol Exp Appl
86:175–182
Justus KA, Mitchell BK (1996) Oviposition site selection by the
diamondback moth, Plutella xylostella (L.) (Lepidoptera:
Plutellidae). J Insect Behav 9:887–898
Kolehmainen J, Julkunen-Tiitto R, Roininen H, Tahvanainen J
(1995) Phenolic glucosides as feeding cues for willow-feeding
leaf beetles. Entomol Exp Appl 74:235–243
Koricheva J, Larsson S, Haukioja E (1998) Insect performance on
experimentally stressed woody plants: a meta-analysis. Annu
Rev Entomol 43:195–216
Kozlowski TT, Pallardy SG (1984) Effect of flooding on water,
carbohydrate and mineral relations. In: Kozloski TT (ed)
Flooding and plant growth. Academic Press, New York, pp
165–181
Larsson S, Wiren A, Lundgren L, Ericsson T (1986) Effects of light
and nutrient stress on leaf phenolic chemistry in Salix
dasyclados and susceptibility to Galerucella lineola Coleoptera. Oikos 47:205–210
Letourneau DK, Fox LR (1989) Effects of experimental design and
nitrogen on cabbage butterfly oviposition. Oecologia 80:211–
214
Lindroth RL (1989) Biochemical detoxication mechanism of
differential tiger swallowtail tolerance to phenolic glycosides.
Oecologia 81:219–224
Lower SS (2002) The effects of soil moisture and water on the
suitability of silky willow for the imported willow leaf beetle.
Dissertation, Tufts University
411
Lower SS, Orians CM (2003) Soil nutrients and water availability
interact to influence willow growth and chemistry but not leaf
beetle performance. Entomol Exp Appl 197:69–79
Marschner H (1986). Mineral nutrition of plants. Academic Press,
London
Matsuki M, MacLean SF Jr (1994) Effects of different leaf traits on
growth rates of insect herbivores on willows. Oecologia
100:141–152
Mattson WJ Jr (1980) Herbivory in relation to plant nitrogen
content. Annu Rev Ecol Syst 11:119–161
Mayew PJ (2001) Herbivore host choice and optimal bad motherhood. Trends Ecol Evol 16:165–167
Mensah RK, Madden JL (1992) Factors affecting Ctenarytaina
thysanura oviposition on Boronia megastigma terminal shoots.
Entomol Exp Appl 62:261–268
Mutikainen P, Walls M, Ovaska J, Keinanen M, Julkunen-Tiitto R,
Vapaavuori E (2000) Herbivore resistance in Betula pendula:
Effect of fertilization, defoliation, and plant genotype. Ecology
81:49–65
Nichols-Orians CM, Fritx RS, Clausen TP (1993) The gentic basis
for varition in the concentration of phenolic glycosides in Salix
sericea: clonal variation and sex based differences. Biochem
Syst Ecol 21:535–542
Orians CM (1995) Preserving leaves for chemical analyses: a
comparison of methods using three willow taxa. J Chem Ecol
21:1235–1243
Orians CM, Fritz RS (1996) Genetic and soil nutrient effects on the
abundance of herbivores on willow. Oecologia 105:388–396
Orians CM, Huang CH, Wild A, Dorfman KA, Zee P, Dao Minh
Tam T, Fritz RS (1997) Willow hybridization differentially
affects preference and performance of herbivorous beetles.
Entomol Exp Appl 83:285–294
Pimbert MP, Srivastava CP (1991) The influence of rainfall deficits
on the abundance of Helicoverpa armigera and Heliothis
armigera in Andhra Pradesh, India. Biol Agric Hortic 8:153–
176
Roininen H, Tahvanainen E (1989) Host selection and larval
performance of two willow-feeding sawflies. Ecology 70:129–
136
Scriber JM, Slansky F (1981) The nutritional ecology of immature
insects. Annu Rev Entomol 26:183–211
Sipura M, Tahvanainen J (2000) Shading enhances the quality of
willow leaves to leaf beetles—but does it matter? Oikos
91:550–558
Tahvanainen J, Julkunen Tiitto R, Kettunen J (1985) Phenolic
glycosides govern the food selection pattern of willow-feeding
leaf beetles. Oecologia 67:52–56
Takabayashi J, Dicke M, Maarten A (1994) Volatile herbivoreinduced terpenoids in plant-mite interactions: variation caused
by biotic and abiotic factors. J Chem Ecol 20:1329–1354
Thompson JN (1998) The evolution of diet breadth: monophagy
and polyphagy in swallowtail butterflies. J Evol Biol 11:563–
578
Tisdale RA, Wagner MR (1991) Host stress influences oviposition
preference and performance of a pine sawfly. Ecol Entomol
16:371–376
Valladares G, Lawton JH (1991) Host-plant selection in the holly
leaf-miner: does mother know best? J Anim Ecol 60:227–240
Vrieling K, de Boer Nico J (1999) Host-plant choice and larval
growth in the cinnabar moth: do pyrrolizidine alkaloids play a
role? Entomol Exp Appl 91:251–257
Wait DA, Jones CG, Coleman JS (1998) Effects of nitrogen
fertilization on leaf chemistry and beetle feeding are mediated
by leaf development. Oikos 82:502–514
Watt AD (1986) The performance of the pine beauty moth Panolis
flammea on water-stressed lodgepole pine Pinus contorta plants
a laboratory experiment. Oecologia 70:578–579
White TCR (1984) The abundance of invertebrate herbivores in
relation to the availability of nitrogen in stressed food plants.
Oecologia 63:90–105
Young CE, Hall RW (1986) Factors influencing suitability of elms
for elm leaf beetle Xanthogaleruca luteola (Coleoptera:
Chrysomelidae). Environ Entomol 15:843–849
Zar JH (1996). Biostatistical analysis. Simon and Schuster, New
Jersey