Journal of Chemical Ecology, Vol. 31, No. 9, September 2005 (
#
2005)
DOI: 10.1007/s10886-005-6078-3
ANTHONY JOERN
1, * and SIMON MOLE 2,3
1
Division of Biology, Kansas State University, Manhattan, KS 66506, USA
2
School of Biological Sciences, University of Nebraska-Lincoln, Lincoln,
NE 68588-0118, USA
(Received January 3, 2005; revised May 9, 2005; accepted May 17, 2005)
Abstract —Acting simultaneously or sequentially, plants encounter multiple stresses from combined abiotic and biotic factors that result in decreased growth and internal reallocation of resources. The plant stress hypothesis predicts that environmental stresses on plants decrease plant resistance to insect herbivory by altering biochemical source Y sink relationships and foliar chemistry, leading to more palatable food. Such changes in the nutritional landscape for insects may facilitate insect population outbreaks during periods of moderate stress on host plants. We examined the plant stress hypothesis with field experiments in continental grassland (USA) using the C
4 grass
Bouteloua gracilis . Water, nitrogen fertilizer, and herbivory from the grassfeeding grasshopper Ageneotettix deorum were manipulated. Combined stresses from water and mineral-N in the soil decreased plant growth and altered foliar percent total N (TN) and percent total nonstructural carbohydrate (TNC) concentrations in an additive fashion. Grasshopper herbivory affected final biomass only in dry years; plants compensated for tissue loss when rainfall was abundant. Foliar TN and TNC concentrations were dynamic with respect to variable climatic conditions and treatment combinations, showing significant interactions. Grasshopper herbivory had its greatest impact on TN or TNC in dry years, interacting with other forms of stress.
Herbivory as a single factor had strong effects on TNC in years with normal precipitation, but not in a dry year. Performance (developmental rate and
* To whom correspondence should be addressed. E-mail: ajoern@ksu.edu
3
Current address: Boulder, CO, USA.
2069
0098-0331/05/0900-2069/0 # 2005 Springer Science + Business Media, Inc.

2070 J OERN AND M OLE survival) by the grasshoppers Phoetaliotes nebrascensis and A. deorum were not greatly affected by plant stress in a manner consistent with the plant stress hypothesis.
Key Words V Chewing insects, environmental stress hypothesis, functionalconvergence-to-plant-stress hypothesis, grasshopper, insect herbivory, total foliar nitrogen, total nonstructural carbohydrates.
INTRODUCTION
Dynamic biochemical, physiological, and morphological responses by plants to environmental conditions are integrated at organ and whole-plant levels through a variety of source Y sink relationships (Mooney and Chiariello, 1984; Bazzaz and Grace, 1997). The plant stress hypothesis states that environmental stresses on plants decrease plant resistance to insect herbivory by altering whole-plant source Y sink resource allocation schedules and foliar chemistry, thus changing food palatability (Rhoades, 1983; Mattson and Haack, 1987; Louda and
Collinge, 1992; White, 1993; Redak and Capinera, 1994; Koricheva et al.,
1998; Huberty and Denno, 2004). Plant resource acquisition (light, water, carbon, elemental nutrients), internal resource allocation among tissues
(source Y sink relationships, translocation products), and partitioning of resources to different plant functions (growth, maintenance, reproduction, repair, defense, senescence) ultimately prescribe the nature and distribution of nutritional constituents within plants to herbivores (Mooney and Gilman, 1982; Bazzaz et al., 1987; Chapin et al., 1987; Mooney et al., 1991; Aerts and Chapin,
2000) V often considered growth optimization processes (Mooney and Winner,
1991). Variation in water and soil nutrient availability coupled to herbivory may cause unpredictable levels of stress that alters plant metabolism in response to the action of one or all factors with consequences for plant growth (Trlica and
Cook, 1971; Bokhari, 1978; Mooney et al., 1991; Louda and Collinge, 1992).
The plant stress hypothesis was proposed as an environmentally determined explanation for outbreaks of insect herbivores operating through plant condition (Rhoades, 1983; Waring and Cobb, 1992; Watt, 1992; Koricheva et al., 1998), in which improved nutritional quality of host plants experiencing intermediate levels of stress resulted in increased demographic performance by herbivores. Rhoades (1983) extended the hypothesis to also include reduced production of chemical defenses under stress conditions in addition to elevated nutritional quality. Experimental tests of the plant stress hypothesis for forest insects provide little general support of the hypothesis (Rhoades, 1983; Waring and Cobb, 1992; Watt, 1992; Koricheva et al., 1998). Although some insect feeding guilds (e.g., boring and sucking feeders) responded as predicted in experimental tests in woody plants, other groups including chewing insects did

PLANT STRESS HYPOTHESIS 2071 not generally respond to plant stress as predicted (Waring and Cobb, 1992;
Watt, 1992; Koricheva et al., 1998; Huberty and Denno, 2004). However, about
67% of the examples are consistent with predictions (Waring and Cobb, 1992) in observational studies of trees along environmental stress gradients, although alternate explanations exist (Watt, 1992). Although this system may be prototypical for the action of the plant stress hypothesis, few tests with grasses exist (Waring and Cobb, 1992; Redak and Capinera, 1994).
We seek to clarify the nature of interactions among multiple stresses as they impact growth and variable leaf chemistry in blue grama grass, Bouteloua gracilis (H.B.K.) Lag. ex Griffiths, according to predictions of the plant stress hypothesis.
B. gracilis is a dominant C
4 grass species in western North
American (USA) grasslands. Two primary predictions of the plant stress hypothesis were examined in the short grass B. gracilis experiencing naturally occurring and variable abiotic conditions: (1) reduced water or soil nitrogen levels coupled to insect herbivory will negatively affect plant growth and increase the palatability of tissues to insect herbivores, (2) chewing insect herbivores will perform better on stressed host plants with higher concentrations of primary nutrients (protein and carbohydrate). In addition, we examined the relative contribution to responses of stresses when combined under field conditions. We examined direct effects and interactions among three common forms of stress to B. gracilis : water availability, plant nutrient availability, and grasshopper herbivory within natural levels in the field. Experiments repeated over 3 years included a wide range of weather conditions against which to gauge plant responses. We expected that the imposition of moderate water or nutrient stress should modify plant physiology in such a way that resistance to herbivores decreases, with a concomitant increase in availability of primary nutrients in leaves to herbivores. As food plant palatability increases following moderate stress to B. gracilis , performance by the grass-feeding grasshoppers
Ageneotettix deorum (Scudder) and Phoetaliotes nebrascensis Thomas should be enhanced as levels of primary nutrients in leaf tissues, especially protein and carbohydrates, increase.
B. gracilis does not produce allelochemicals that are expected to influence responses to primary nutrients by herbivores in this experiment (Mole and Joern, 1994), allowing us to restrict our attention to the nutritional component of the problem.
METHODS AND MATERIALS
Study System.
We conducted field experiments at Arapaho Prairie (Arthur
County, NE, USA), a protected research site in Nebraska sandhills grassland.
The site is characterized by upland sandhills grassland composed of large stabilized sand dunes with steep upper ridges that gradually slope into broad flat

2072 J OERN AND M OLE valleys. Most plants at Arapaho Prairie experience at least some water and nutrient stress in most years (Barnes, 1985; Mole et al., 1994).
Vegetation at Arapaho Prairie is an open-canopy mixed-prairie, modified by sandy substrate (Barnes, 1985). Grasses contribute 80% to total plant biomass, with long-term NAPP ranging between 75 and 250 g m j 2
(unpublished data). C
3 and C
4 grass species typical of eastern tallgrass prairie and western shortgrass steppe grasslands intermingle at the site. Dominant plants in this sand dune landscape form loose but recognizable vegetation associations along the existing topographic gradient (Barnes, 1985). The grass canopy is intermingled with extensive bare ground, largely because of extensive disturbance from pocket gophers.
Long-term annual mean precipitation (1951 Y 1980) recorded 15 km from
Arapaho Prairie at Arthur County, NE, averaged 47.1 cm (SD = 8.98 cm) from
F IG.
1. Precipitation patterns at Arapaho Prairie. (a) Annual rainfall with mean and 95% confidence intervals, 1987 Y 2000. (b) Seasonal pattern of precipitation illustrated by cumulative amount by date for the 3 years of the study.

PLANT STRESS HYPOTHESIS 2073
US Weather Bureau records; the recent 14-year record from Arapaho Prairie
(1987 Y 2000) averaged 37.3 cm (SD = 11.4 cm). The amount and timing of precipitation at Arapaho Prairie varies greatly among years (Figure 1). Belowaverage precipitation was observed in two of the three years of this study
(Figure 1a), with rainfall in 1990 equaling the average amount for the site.
Perhaps more importantly, the seasonal timing of rainfall over the growing season differs in important ways among years (Figure 1b). Both 1989 and 1991 received approximately the same amount of precipitation, but rain fell early in the season in 1991 compared with late-season rainfall in 1989. In 1990, rainfall occurred throughout the growing season, compared with 1989 and 1991, each of which experienced large periods without significant amounts of rain.
Arapaho Prairie soils contain 80 Y 85% sand with low nutrient concentrations (Barnes et al., 1984). Total nitrogen in soil in the top 10 cm ranges from
0.02 to 0.07% of total soil weight according to landscape position. Valleys exhibit the highest soil total N levels, but all landscape positions are generally low (Alward and Joern, 1993). Nitrate concentrations range from 0.04 to 15 ppm, and ammonium concentrations varied from 0.17 to 3.3 ppm. Light is seldom a major limitation to plant growth because of the open canopy and large proportion of sunny days at this site.
B. gracilis is an often dominant C
4 short-grass species throughout the shortgrass steppe of the Rocky Mountain foothills to the mixed-grass prairies of the central Great Plains of North America. In Nebraska sandhills grasslands, it is commonly found in fine-textured soils typical of dry valleys. At Arapaho
Prairie, B. gracilis comprises up to 20 Y 30% of the relative cover of valleys and midslope dunes but is nearly absent from dune ridges (Barnes et al., 1984).
B.
gracilis productivity is correlated with soil moisture, and biomass peaks in early
August although yearly variability exists.
B. gracilis is an important dietary component of graminivorous grasshopper species at this site, including A.
deorum and P. nebrascensis (Joern, 1985).
Experimental Design and Statistical Analyses.
Overall, two related experiments were run concurrently, one addressing effects of water, N fertilizer, and grasshopper herbivory on plant response, and the other investigating grasshopper performance in response to water and N-fertilizer treatments on plants.
Rectangular cages ( basal area 0.5 m
2
, 80 cm high) were constructed of 0.64-cm mesh and buried 10 cm after severing possible root connections to neighboring ramets. Cages were placed over natural stands of B. gracilis B turf ^ in early
June, corresponding to the initiation of growth. Cages housing treatment combinations of both experiments were intermingled randomly within each block, but experiments were analyzed separately.
Plant Responses.
We manipulated levels of water, nitrogen fertilizer, and grasshopper herbivory within natural levels to understand variation in plant responses to stress. Biomass accumulation and foliar chemical responses (% total

2074 J OERN AND M OLE nitrogen, TN; and % total nonstructural carbohydrates, TNC) by B. gracilis to multiple stresses was studied using a 3 2 2 full-factorial treatment combination (N fertilizer, water availability, and grasshopper herbivory, respectively) experiment in a randomized complete block design, nested within each of 3 years. Six sites (blocks) were arbitrarily selected in a range of natural habitats for B. gracilis along a gradient stretching from slope vegetation to valley vegetation. Sites were selected based on the criterion that a sufficient density of B. gracilis was available to set up a full set of treatment combinations. Treatment combinations were randomly assigned to predetermined patches of B. gracilis within each block.
Grasshopper Performance.
Grasshopper performance was evaluated in a field experiment executed in parallel with the plant stress experiment by using a similar experimental design and identical water and mineral-N fertilizer additions using cages as described above. Cages were intermingled randomly with those of the plant stress experiment. The experimental design was a 3 2 full-factorial treatment combination experiment (N fertilizer and water availability, respectively) arrayed in a randomized complete block design, nested within each of 3 years. Six blocks were used. A repeated-measures analysis of variance (ANOVA) was used to examine grasshopper survival.
Responses of two grasshopper species to plant stress were evaluated in different years (1989, P. nebrascensis ; 1990, A. deorum ), but specific responses between species cannot be compared directly because of overall differences in naturally occurring stress between years. Ten fourth instar nymphs were added to each cage in late June or early July to match natural phenological development of each species in the field. The number of survivors and the developmental stage of individuals were determined every 2 Y 3 d from censuses of individuals remaining in each cage.
Statistical Analyses.
Statistical analyses were performed using ANOVA, with treatments evaluated as fixed effects in the ANOVA. To normalize data, dependent variables expressed as percent of the total sample weight were transformed by applying arcsine(square root) to original data before statistical analyses. We present and discuss values in the nontransformed state. Treatment variables were treated categorically in analyses.
Manipulations of Plant Stress from Water, Mineral Nitrogen, and Grasshopper
Herbivory
(1) Water . Two water levels were used: W+, in which water was added weekly for the 10-wk duration of the experiment, and W0, where no additional water beyond ambient rainfall was added. We considered W0 to be more stressful than W+ as water stress is common in grasses (Heinisch, 1981; Barnes, 1985).

PLANT STRESS HYPOTHESIS 2075
In the first 2 wk of the experiment, all plots received water in addition to N fertilizer if scheduled for that cage. After this, W+ cages received 2 l m j 2 wk j 1 of supplemental water over the course of the experiment. No attempt was made to standardize the absolute level of plant water stress among years.
(2) N-Fertilizer . Soil-nitrogen levels were manipulated using ammonium nitrate (NH
4
NO
3
). Levels included 0, 3, and 6 g N m j 2 of N fertilizer
(N
0
, N
3
, and N
6 treatments, respectively). N fertilizer was applied in two half-strength additions over several days in early June in each year.
(3) Grasshopper Herbivory . Moderate densities of the B. gracilis -feeding grasshopper, A. deorum , were added to cages to assess foliar responses to insect herbivory. In the GH+ treatment, we added four adult grasshoppers to each cage in late June. This density corresponded to eight individuals per square meter, about double the long-term average of all grasshoppers at
Arapaho Prairie (A. Joern, unpublished data), but about half the economic threshold. Moreover, the densities used in the experiments are routinely observed in some vegetation patches in most years. No grasshoppers were added to cages in the GH0 treatment. Initiation of the grasshopper treatment corresponded to the phenological presence of the adult A. deorum in the field. Grasshoppers were replaced weekly to maintain relatively constant levels of herbivory.
Final Biomass Estimates and Chemical Analyses of Leaf Material.
Leaf samples of B. gracilis were collected at the end of the experiment (mid-August) and prepared for chemical analysis. Initially, a subsample of green leaf material
[ca. 2 Y 3 g dry weight (d.w.)] was collected, immediately flash-frozen in liquid nitrogen in the field, and then prepared for chemical analyses. Samples were lyophilized for 48 hr and stored under desiccant in a freezer. Dried leaf material was ground with a Wiley Mill (40-mesh sieve) before chemical analysis. After collecting leaf material for chemical analyses, remaining plant biomass in a cage was clipped, dried (80 C for 24 hr) and weighed.
Total Nitrogen.
Total nitrogen was analyzed by using modified micro-
Kjehldahl techniques (AOAC, 1984) with a standard digest on 100-mg samples of ground leaf material (2 ml H
2
SO
4
, a CuSeO
4
Kjeltab catalyst tablet). Total N was determined by measuring ammonia generated after adding 100 ml of 5 M
NaOH to the digest using a selective ion electrode (Orion). The ammonium probe was calibrated daily with an ammonium sulfate standard.
Total Nonstructural Carbohydrates.
Total nonstructural carbohydrates were extracted following the method of (Smith, 1981) except for the use of amylglucosidase (Sigma A-7255) as the enzyme preparation in the digest. These were analyzed by the titrimetric method of Smith (1981) with glucose as a standard without the hydrolysis of sucrose. Sucrose averaged about 0.4
Y 0.5% d.w.

2076 J OERN AND M OLE of plant material compared with 17 Y 22% d.w. plant material for TNC as measured and did not vary with TNC concentration (S. Mole, unpublished data).
RESULTS
Total Plant Biomass.
On average, total biomass in B. gracilis plots at the end of the season (Figure 2) was about 50 Y 100% greater in an average rainfall
F IG.
2. End of season B. gracilis biomass (mean, SE) according to stress treatment conditions [water (W0, W+), N-fertilization (0, 3 and 6 g N m j 2
), and grasshopper herbivory (GH0, GH+)] for each year of the study.

PLANT STRESS HYPOTHESIS 2077 year (1990) as in dry years (1989, 1991), which were similar.
B. gracilis biomass was significantly different among experimental treatments depending on the number of stresses applied, indicating that the plants in this study experienced varying degrees of overall stress. Both water (1989: F
1,56
= 17.4,
P < 0.001; 1990: F
1,56
= 6.6, fertilizer additions (1989: F
2,56
P = 0.013; 1991:
= 4.2, P
F
1,56
= 11.3,
< 0.021; 1990: F
2,56
P < 0.001) and N
= 11.2, P < 0.001;
1991: F
2,56
= 7.1, P < 0.001) resulted in increased biomass in all years as additive, direct effects; no statistical interactions were detected for water and N fertilizer in any year (Figure 2).
Feeding by grasshoppers reduced the final B. gracilis biomass in the dry years of 1989 and 1991 (67% in 1989, F
1,56
F
1,56
= 34.8, P < 0.001; 32% in 1991,
= 6.5, P = 0.012), but no effect from grasshopper feeding was detected in
1990, a year of normal rainfall. This indicates that complete compensation for foliage loss was observed in this year with normal rainfall. No statistical interactions among grasshopper herbivory, water availability, and N fertilizer treatments were observed in their combined effect on final B. gracilis biomass, but were additive instead. Although biomass estimates do not include the amounts consumed by grasshoppers, these should be similar between years as the grasshopper encounter rate was controlled.
Foliar Total Nitrogen.
Foliar TN differed significantly among treatments, year, and block (Figures 3a and 4a, Table 1). TN concentrations were highest for all treatments in 1989, the driest year, a year with almost no precipitation occurring early in the growing period (Figure 1b). TN at the end of the experiments in August 1989 averaged 1.73% total dry weight in all treatment combinations compared with 1.01% (1990) and 1.14% (1991) TN in subsequent years, representing a notable decrease in 1990 Y 1991 compared with 1989.
Foliar TN levels varied in response to both N fertilizer and water treatments in some fashion in all years (Figures 3a and 4a, Table 1), with water addition explaining the most variation in responses (Figure 5). Depending on the year, N fertilizer addition increased foliar TN levels from 5 to 21% dry mass compared with no fertilizer addition treatments. An average 13% increase in foliar TN over the 3-year period was observed. Differences in foliar TN between 3N vs 6N treatments were of smaller magnitude (3 Y 10%), and only significantly different in 1991.
Although the main effects of treatments were pronounced in all cases
(Figure 3), treatment interactions that were important and insightful to underlying processes were sometimes detected. W0 treatments resulted in a
10 Y 20% higher level of total foliar-N compared with W+ treatments. The weakest response to water (9.5%) was observed in the driest year (1989), possibly because extreme drought stress in that year was not proportionally offset by the water addition treatment compared to other years. A significant N fertilizer by water interaction existed in 1989 and 1990 but with different

2078 J OERN AND M OLE
F IG.
3. Responses (mean, SE) in (a) % total N and (b) % TNC to main treatments (water addition, grasshopper herbivory, and N fertilizer) for each year.
responses between the 2 years (Figure 5). In very dry 1989, higher foliar TN levels were seen in W0 only for the N
0 between W0 and W+ for the N
3 and N
6 treatment. No differences were seen fertilization treatment levels. In a year of average rainfall (1990), there was no difference in foliar TN between water treatments at N
0
, but significant and about equal increases in total N for N
3
N
6 treatments in interaction with water availability.
and
Grasshopper herbivory affected foliar TN levels significantly as a main effect only in 1991. However, grasshopper feeding interacted with other treatments to influence total foliar TN in all years (Figures 4a and 5). In 1989, there was an increase in TN up to the maximum level observed at N
3
, a TN level that was reached with N
6 with no grasshoppers. In 1990, grasshopper herbivory interacting with water availability led to higher TN level that was reached in the

PLANT STRESS HYPOTHESIS 2079
F IG.
4. Percentage of total variance in foliar nutrient responses explained by experimental treatments in each year of study. (a) % Total foliar nitrogen (TN) and (b) % total nonstructural carbohydrates (TNC). Letters refer to main effects ( N, nitrogen fertilization; W, water; G, grasshopper herbivory) and statistical interactions (N*W, N*G,
W*G) as indicated in the experimental design of Table 1. B is the block (site) effect.
Percentage of total variance in response was calculated as the variance associated with the treatment combination compared with the total variance of the experiment.
W0 treatment compared with the W+ treatment for which there was no significant difference between grasshopper treatments.
Foliar Total Nonstructural Carbohydrates.
Significant responses in foliar
TNC concentrations were also observed (Table 1, Figures 3b, 5b, and 6) in response to combined stresses. Among-year differences averaged 5 Y 10%, with
1989 exhibiting the highest foliar TNC levels. Differences in responses among all treatment combinations showed little variation in 1989 and 1991 compared

2080 J OERN AND M OLE

PLANT STRESS HYPOTHESIS 2081
F IG.
5. Responses of significant interactions among treatments for % total foliar N (TN) for each year of study.
with 1990. Total variance in TNC levels among treatments was 1.5
Y 5 times greater in the average rainfall year (1990) than in the other years. Over all 3 years, combined nitrogen fertilizer and grasshopper treatments for all levels were significant as main effects, with no significant statistical interactions. When compared against the N fertilizer treatments, W0/GH0 had the lowest TNC levels and
W+/GH+ had the highest levels on average, with each decreasing along the
N-fertilization axis. Levels of TN and TNC in leaves were uncorrelated in all years for all treatments combined (1989: r
2
= 0.014; 1990: r
2
= 0.001;
1991: r
2
= 0.038; P > 0.05 for all years). However, when years were analyzed separately, interesting differences were observed.
In general, TNC declined 4 Y 6.5% with increased N fertilizer in all years, although no significant differences were observed between the 3 g and 6 g

2082 J OERN AND M OLE
F IG.
6. Responses of significant interactions among treatments for % total nonstructural carbohydrate (TNC) for each year of study.
N fertilizer treatments. When water treatments were significant (1989 and
1991), TNC was greater in W+ compared with the W0 treatments, with differences on the order of about 3 Y 4%. Generally, grasshopper herbivory was a factor when interacting with either N fertilizer or water treatments (Table 1). In
1990, GH+ resulted in a large 23% increase in % foliar TNC, and important interactions with N fertilizer and water were detected.
The nature of interactions among sources of plant stress differed among years. Numerous interactions were observed in both 1990 and 1991 (Figure 6), average and below average rainfall years, respectively. In 1990, all two-way interactions and a three-way interaction were significant. % TNC in the N
6 fertilizer treatment increased in the W0 treatment, but the trend otherwise was for TNC to drop with increased N fertilizer. Grasshopper treatments interacted with both N fertilizer and water in both 1990 and 1991, but the TNC responses were different. In the very dry 1989, no interactions were detected, and all contributions to the variance in TNC content were additive. Inclusion of grasshoppers resulted in increased TNC in high-resource environments ( N or

PLANT STRESS HYPOTHESIS 2083 water) compared with the GH0 treatments. In 1991, the opposite response was observed where TNC levels under high-resource conditions were lower if grasshoppers were present.
Grasshopper Performance. P. nebrascensis . This species was studied in a very dry year with late season rainfall. No significant effect of treatment combinations was observed for developmental rate although there is a suggestion that W0/6N develops faster. Repeated-measures ANOVA of the number of
F IG.
7. Mean survival of two grasshoppers in response to plant stress treatments.
Experiments were performed in different years as described in the text. Data are transformed as natural log of number alive at each census period.

2084 J OERN AND M OLE individuals remaining in cages of P. nebrascensis (Figure 7a) was significant
(Wilk’s l = 0.10, P < 0.001). However, although observed trends in survival may be suggestive, no significant effect of water and N fertilizer treatments were detected. The significant difference in the repeated-measures ANOVA reflected the decrease in the number of survivors over time, not treatments.
Ageneotettix deorum.
This species was studied in a normal rainfall year.
No significant effect of water and N fertilizer treatments on developmental rate was detected.
A. deorum survival (Figure 7b) varied in response to experimental treatments (repeated-measures ANOVA, Wilk’s l = 0.137, F
6,21
= 22.06, P <
0.001). A significant N Fertilizer Water interaction was detected (repeatedmeasures ANOVA, F
2,6
= 6.3, P = 0.006). In W0 treatments, survivorship was greatest in treatments with no N fertilizer and decreased when N was added. In
W+ treatments, survival was highest on fertilized plots, at least for the first half of the trajectory when N0 and N6 N fertilizer treatments converged.
DISCUSSION
This study examines how multiple environmental stresses interact to affect growth and foliar chemistry in the grass B. gracilis , and whether plant responses to these stresses increase herbivore performance. Since White’s (1993) formulations of the plant stress hypothesis, much effort has been directed at understanding its overall importance and generality to understanding insect herbivore population responses to plant stress (Waring and Cobb, 1992; Koricheva et al.,
1998). Testing this hypothesis becomes a greater challenge when multiple stresses operate (Mooney et al., 1991). Soil nutrient stress varies according to substrate type and nutrient cycling characteristics of the site, and ecological processes that deplete nutrient availability such as uptake rates by plants or use by soil microbes (Aerts and Chapin, 2000). Water availability varies at multiple scales within and among years, where plants in natural environments are often water stressed, including B. gracilis studied here (Mole et al., 1994). For much vegetation, local light availability is influenced primarily by accumulated total biomass at the site, although intermittent cloud cover can be important. In this study, light was unlikely to limit photosynthesis because of the low stature of vegetation growing in an open habitat under conditions of sunny skies on most days. Tissue loss from herbivory further modifies physiological responses in plants and potentially interacts with other sources of stress in nonlinear ways
(Mattson and Haack, 1987) and is the basis of Jones and Coleman’s (1991)
B phytocentric model ^ of plant Y insect herbivore interactions.
In response to moderate stress to B. gracilis , plant growth decreased, concentrations of primary nutrients in leaf tissue increased, and the ability to compensate for tissue loss from herbivory was reduced. Here, foliar chemistry

PLANT STRESS HYPOTHESIS 2085 in response to stress conditions is highly dynamic and effects of combined abiotic stresses are additive, but sometimes appear idiosyncratic when combined with herbivory from grasshoppers. Results observed for B. gracilis are consistent with those of other studies. Multiple environmental stresses to plants regularly reduce plant growth compared to the maximum performance possible
(Mooney et al., 1991; Louda and Collinge, 1992), and nutrient and water stress or tissue loss often alter tissue palatability to herbivores (Mattson and Haack,
1987; Louda and Collinge, 1992; White, 1993; Redak and Capinera, 1994).
Understanding the integrated responses by plants to combined stresses from abiotic conditions and herbivory is limited by our ability to incorporate the consequences of multiple stresses into a predictive framework (Jones and
Coleman, 1991; Mooney et al., 1991; Bazzaz and Grace, 1997). Because plants function as integrated units, whole-plant growth responses reflect the underlying coordination and allocation among competing resource sinks (Mooney et al.,
1991; Bazzaz and Grace, 1997).
Consequences of Environmental Stresses to B. gracilis. B. gracilis biomass varied significantly with manipulation of water availability, nitrogen fertilizer, and grasshopper herbivory treatments in the field, showing that these factors contribute importantly to plant stress. Responses differed among years as weather conditions varied (hot, dry vs normal precipitation), and both water and nitrogen fertilizer manipulations affected plant growth in each year in an additive fashion. Moreover, plant biomass decreased relative to controls in response to grasshopper herbivory in the driest years (1989, 1991), but herbivory did not affect final biomass accumulation in a year with normal rainfall (1990). These results indicate that B. gracilis compensates for tissue losses from herbivory when provided sufficient water and nutrients to support photosynthesis and growth.
Foliar chemistry of B. gracilis is highly variable among years; variability in foliar chemical concentrations should increase in response to stress according to variability in plant stress. This is to be expected, as critical soil nutrients, light, and water required for plant growth routinely shift in time and space (Bazzaz and Grace, 1997) under natural conditions. Water stress regularly resulted in increased concentrations of foliar TN in all years and was the most important stress to B. gracilis . However, N fertilizer, grasshopper, and year effects contributed greatly to the expression of TN, showing a variety of outcomes among years.
Insect herbivory should affect plants in a manner similar to other environmental stresses in that it alters the capacity for photosynthesis by removing leaf material and changes source Y sink relationships to favor regrowth of leaves. Photosynthesis, growth, and foliar nutrients routinely vary in response to the timing and degree of herbivory (Redak and Capinera,
1994). Of greatest interest is the highly variable nature of responses of foliar

2086 J OERN AND M OLE chemistry to grasshopper herbivory, especially among years and the large number of interactions that were observed between other plant stresses and grasshopper herbivory.
The Plant Stress Hypothesis.
By influencing metabolic activity in general, environmental stresses often alter plant resistance to herbivory, especially because of changes in foliage quality to herbivores (Rhoades, 1983; Bazzaz et al., 1987; Mattson and Haack, 1987; Louda and Collinge, 1992). For example, reduced soil water availability often reduces resistance to herbivory because of increased nutrient concentrations in leaf tissue available to herbivores (McNeil and Southwood, 1978; White, 1993; Redak and Capinera,
1994), decreased or elevated concentrations of defensive compounds (Rhoades,
1983; Gershenzon, 1984; Redak and Capinera, 1994), or some trade-off between nutritional and defensive qualities in leaf tissues that make them more or less palatable to herbivores (Bazzaz et al., 1987). Consequently, increased nutritional quality combined with decreased defensive capability results in improved herbivore performance. Mature grasses contain few chemical defenses compared to other plant taxa (Mole and Joern, 1994), thus simplifying the problem. Primary nutrients are also typically found at much lower concentrations in grasses than are typically observed in forbs and wood plants, decreasing the plant’s value as food to herbivores (Bernays and Barbehenn,
1987). Given the expected low concentrations of limiting nutrients relative to consumer needs, small shifts in their availability may provide large fitness consequences to individuals feeding on them (Joern and Behmer, 1997).
B. gracilis responded to multiple stresses more or less as expected, but grasshopper performance was not consistent with predictions of the plant stress hypothesis. No changes in developmental rate were observed in either species, and survival in P. nebrascensis showed no significant differences. Survival in
A. deorum differed among stress treatments, but higher survival was observed in treatments with lower levels of foliar TN, contrary to survivorship patterns expected based on feeding studies with controlled diets (Joern and Behmer,
1997).
Reviews of the plant stress hypothesis on woody plants (Waring and Cobb,
1992; Koricheva et al., 1998; Huberty and Denno, 2004) indicate weak support for the notion at best. No significant concordance with expectations was observed in a meta-analysis of experimental studies (Koricheva et al., 1998;
Huberty and Denno, 2004), although comparative studies in the field were reasonably consistent with expectations (Waring and Cobb, 1992; Watt, 1992).
Fewer studies are available for nonwoody plants, especially for grasses, and results generally conflict with predictions of the plant stress hypothesis. Redak and Capinera (1994) showed that heavy defoliation of western wheat grass
( Pascopyrum smithii ) either by mechanical means in the laboratory or from herbivory by P. nebrascensis in the field altered foliar nutrients and palatability,

PLANT STRESS HYPOTHESIS 2087 but in the opposite direction required for support. Similarly, polyphagous leaf miners on grasses respond positively as expected to foliar nutritional quality, but fertilizer stress has a negative impact on population responses (Scheirs and
De Bruyn, 2004).
Without doubt, insect herbivores typically encounter heterogeneous nutritional landscapes while foraging, one largely resulting from the integrated responses of plants to variable environmental conditions (Jones and Coleman,
1991; Louda and Collinge, 1992). In turn, variable nutritional quality is expected to directly influence herbivore fitness and subsequent population fluctuations
(Rhoades, 1983; Jones and Coleman, 1991; White, 1993). However, responses by chewing insect herbivores to plant quality are highly variable, with some species responding positively and others negatively or not at all to specific responses (Joern and Behmer, 1997; Fischer and Fielder, 2000). This is especially evident for understanding the dynamic changes in foliar TN and TNC in response to combined stress, which in turn affect subsequent levels of insect herbivory.
Results from our study with grasshoppers are consistent with others that examine free-living, chewing insects. Reviews of the plant stress hypothesis indicate that under continuous stress, most insect herbivore guilds either are negatively impacted or do not respond to plant stress (Watt, 1992; Koricheva et al.,
1998). The best current model (Huberty and Denno, 2004) may be the B pulsed plant stress hypothesis ^ in which foliar nutrients accumulating from stress conditions only become available to herbivores, especially sap feeders, as plants recover from stress. Still, many other insect herbivore guilds are not explained.
Nutritional quality of host plants generally affects performance by grasshoppers, as documented by laboratory and field studies (Joern and Behmer, 1997;
Simpson and Raubenheimer, 2001). Why does the plant-stress hypothesis not explain responses by free-living chewing insects, such as grasshoppers, despite the appealing logic inherent in the formulation of plant stress hypothesis? A combination of multiple factors might explain. It may not be possible to directly relate performance to stress treatments because of intervening pathways that interact in unknown ways. (1) Grasshoppers perform best on a diet that is balanced between protein and carbohydrates; as nutritional ratios deviate from the target, performance drops as well. Stressed plants may have elevated foliar protein that is not always in balance with carbohydrates. (2) Grasshoppers can compensate for poor-quality food by altering diets (Simpson and Abisgold,
1985) or by modifying retention time in the gut (Yang and Joern, 1994), a response that could affect nutrient acquisition and insect performance in a nonlinear fashion. Lack of response to stress may actually be B hidden ^ because of such physiological adjustments. (3) As observed in this study, chemical responses in B. gracilis leaves to stress treatment combinations are highly variable, making it particularly difficult to predict performance. Response by grasshoppers may best fit predictions of Price’s (1991) B plant vigor hypothesis ^ ,

2088 J OERN AND M OLE which argues for elevated insect herbivore performance on tissues in actively photosynthesizing tissue; this may allow for a wider range of combinations of protein and carbohydrate. It is noteworthy that TNC and TN concentrations among leaf samples were uncorrelated (unpublished data), making it particularly difficult for grasshoppers to use simple phagostimulatory cues to select balanced diets. (4) Effects of stresses themselves to plants are multifactorial
(water, temperature, trampling, N addition, herbivory) such that the expected effects on grasshoppers are not predicted by simple plant stress models; Jones and Coleman (1991) provide a basis for a more comprehensive model.
Ultimately, some combination of each of the above and other factors influencing foraging by free-living grasshoppers will determine the relative contribution of plant stress to performance in combination with other factors.
It is important to resolve the ability to relate insect herbivore response to plant stress. Historically, forecasts of insect pest outbreaks including grasshoppers often assume that insect populations do better under hot/dry weather conditions because of better food quality (Rhoades, 1983; Mattson and Haack,
1987). This weather-induced link between plant stress and insect performance is often circumstantial at best (Huberty and Denno, 2004) and remains to be vetted carefully before it is recognized as a key mechanism underlying insect outbreaks. However, it is also true that many experimental studies, including those with grasshoppers, implicate elevated host plant quality as an important determinant of insect performance and population responses in the field.
Alternate explanations that include multiple factors (Belovsky and Joern, 1995) or incorporate more sophisticated views of nutritional contributions and environmental influences on observed population variability (Simpson and
Raubenheimer, 2001) must be developed to fill the void left by the inability of the plant stress hypothesis to explain natural patterns.
Acknowledgments V Logistical support from Cedar Point Biological Station is gratefully acknowledged. M. Thomas, M. Zeisset, C. Holtmeier, S. Behmer, Y. Yang, and L. Kang provided help in the field. Y. Chen, T. Minnick, M. Thomas, L. Snyder, and M. Zeisset helped analyze plant samples. B. Danner and K. Stoner provided comments on the manuscript. Research was supported by USDA/NRI and the National Science Foundation.
REFERENCES
A ERTS , R. and C HAPIN , F. S. I. 2000. The mineral nutrition of wild plants revisited: a reevaluation of processes and patterns.
Adv. Ecol. Res.
30:2 Y 67.
A LWARD , R. D. and J OERN , A. 1993. Plasticity in grass responses to herbivory.
Oecologia
95:358 Y 364.
AOAC. 1984. Official Methods of Analysis. Association of Official Agricultural Chemists,
Washington, DC.

PLANT STRESS HYPOTHESIS 2089
B ARNES , P. W. 1985. Adaptation to water stress in the big bluestem Y sand bluestem complex.
Ecology 66:1908 Y 1920.
B ARNES , P. W., H ARRISON , A. T., and H EINISCH , S. P. 1984. Vegetation patterns in relation to topography and edaphic variation in Nebraska Sandhills Prairie.
Prairie Nat.
16:145 Y 158.
B AZZAZ , F. A. and G RACE , J. (eds.). 1999. Plant Resource Allocation. Academic Press, San Diego,
CA.
B AZZAZ , F. A., C HIARIELLO , N. R., C OLEY , P. D., and P ITELKA , L. F. 1987. Allocating resources to reproduction and defense.
Bioscience 37:58 Y 67.
B ELOVSKY , G. E. and J OERN , A. 1995. Regulation of grassland grasshoppers: differing dominant mechanisms in time and space, pp. 359 Y 386, in N. Cappucino and P. W. Price (eds.). Novel
Approaches for the Study of Population Dynamics: Examples from Insect Herbivores.
Academic Press, New York.
B ERNAYS , E. A. and B ARBEHENN , R. 1987. Nutritional ecology of grass foliage-chewing insects, pp. 147 Y 175, in F. Slansky Jr. and J. G. Rodriguez (eds.). Nutritional Ecology of Insects. Wiley
Interscience, New York.
B OKHARI , U. G. 1978. Nutrient characteristics of blue grama herbage under the influence of added water and nitrogen.
J. Range Manag.
31:18 Y 22.
C HAPIN , F. S. I., B LOOM , A. J., F IELD , C. B., and W ARING , R. H. 1987. Plant responses to variable environments.
Annu. Rev. Ecolog. Syst.
16:363 Y 392.
F ISCHER , K. and F IELDER , K. 2000. Response of the copper butterfly Lycaena titrys to increased leaf nitrogen in natural food plants: Evidence against the nitrogen limitation hypothesis.
Oecologia
124:235 Y 241.
G ERSHENZON , J. 1984. Changes in the levels of plant secondary metabolites under water and nutrient stress.
Recent Adv. Phytochem.
18:273 Y 320.
H EINISCH , S. P. 1981. Water allocation and rooting morphology of two Bouteloua species in relation to their distributions in the Nebraska sandhills. M.S. thesis, University of Nebraska.
H UBERTY , A. and D ENNO , R. F. 2004. Plant water stress and its consequences for herbivorous insects: a new synthesis.
Ecology 85:1385 Y 1398.
J OERN , A. 1985. Grasshopper dietary (Orthoptera: Acrididae) from a Nebraska sandhills prairie.
Trans. Nebr. Acad. Sci.
8:21 Y 32.
J OERN , A. and B EHMER , S. T. 1997. Importance of dietary nitrogen and carbohydrates to survival, growth and reproduction in adult Ageneotettix deorum (Orthoptera: Acrididae).
Oecologia
112:201 Y 208.
J ONES , C. G. and C OLEMAN , J. S. 1991. Plant stress and insect herbivory: toward an integrated perspective, pp. 249 Y 280, in H. A. Mooney, W. E. Winner, and E. J. Pell (eds.). Response of
Plants to Multiple Stresses. Academic Press, San Diego.
K ORICHEVA , J., L ARSSON , S., and H AUKIOJA , E. 1998. Insect performance on experimentally stressed wood plants: a meta-analysis.
Annu. Rev. Entomol.
43:195 Y 216.
L OUDA , S. M. and C OLLINGE , S. K. 1992. Plant resistance to insect herbivores: a field test of the environmental stress hypothesis.
Ecology 73:153 Y 169.
M ATTSON , W. J. and H AACK , R. A. 1987. The role of drought in outbreaks of plant-eating insects.
Bioscience 37:110 Y 118.
M C N EIL , S. and S OUTHWOOD , T. R. E. 1978. The role of nitrogen in the development of insect Y plant relationships, pp. 77 Y 98, in J. B. Harborne (ed.). Biochemical Aspects of Plant and Animal
Coevolution. Academic Press, London.
M OLE , S. and J OERN , A. 1994. The feeding behavior of graminivorous grasshoppers in response to host Y plant extracts, alkaloids and tannins.
J. Chem. Ecol.
20:3097 Y 3109.
M OLE , S., J OERN , A., O ’ L EARY , M. H., and M ADHAVAN , S. 1994. Spatial and temporal variation in carbon isotope discrimination in prairie graminoids.
Oecologia 97:316 Y 321.
M OONEY , H. A. and C HIARIELLO , N. R. 1984. The study of plant function: the plant as a balanced

2090 J OERN AND M OLE system, pp. 305 Y 323, in R. Dirzo and J. Sarukhan (eds.). Perspectives on Plant Population
Ecology. Sinauer Associates, Sunderland, MA.
M OONEY , H. A. and G ILMAN , S. L. 1982. Constraints on leaf structure and function in reference to herbivory.
Bioscience 32:198 Y 206.
M OONEY , H. A. and W INNER , D. A. 1991. Partitioning response of plants to stress, pp. 499 Y 518, in H. A. Mooney, D. A. Winner, and E. J. Pell (eds.). Response of Plants to Multiple Stresses.
Academic Press, San Diego, CA.
M OONEY , H. A., W INNER , W. E., and P ELL , E. J. (eds.). 1991. Response of Plants to Multiple
Stresses. Academic Press, San Diego.
P RICE , P. W. 1991. The plant vigor hypothesis and herbivore attack.
Oikos 62:244 Y 251.
R EDAK , R. A. and C APINERA , J. L. 1994. Changes in western wheatgrass foliage following defoliation: consequences for a graminivorous grasshopper.
Oecologia 100:80 Y 88.
R HOADES , D. F. 1983. Herbivore population dynamics and plant chemistry, pp. 3 Y 53, in R. F. Denno and M. S. McClure (eds.). Variable Plants and Herbivores in Natural and Managed Systems.
Academic, New York.
S CHEIRS , J. and D E B RUYN , L. 2004. Excess nutrients results in plant stress and decreased grass miner performance.
Entomol. Exp. Appl.
113:109 Y 116.
S IMPSON , S. J. and A BISGOLD , J. D. 1985. Compensation by locusts for changes in dietary nutrients:
Behavioural mechanisms.
Physiol. Entomol.
10:443 Y 452.
S IMPSON , S. J. and R AUBENHEIMER , D. 2001. The geometric analysis of nutrient Y alleleochemical interactions: a case study using locusts.
Ecology 82:422 Y 439.
S MITH , D. 1981. Removing and analyzing total nonstructural carbohydrates from plant tissue.
Wisconsin Agricultural and Experimental Station. Report R2107. University of Wisconsin,
Madison, WI.
T RLICA , M. J. and C OOK , C. W. 1971. Defoliation effects on carbohydrate reserves of desert species.
J. Range Manage.
24:418 Y 425.
W ARING , G. L. and C OBB , N. S. 1992. The impact of plant stress on herbivore dynamics, pp. 167 Y 226, in E. A. Bernays (ed.). Insect Y Plant Interactions. CRC Press, Boca Raton, FL.
W ATT , A. D. 1992. The relevance of the stress hypothesis to insects feeding on tree foliage, pp. 73 Y 85, in S. R. Leather, A. D. Watt, N. J. Mills, and K. F. A. Walters (eds.). Individuals,
Populations and Patterns in Ecology. Intercept Ltd., Andover, Hampshire, UK.
W HITE , T. C. R. 1993. The Inadequate Environment: Nitrogen and the Abundance of Animals.
Springer-Verlag, Berlin.
Y ANG , Y. and J OERN , A. 1994. Influence of diet, developmental stage and temperature on food residence time.
Physiol. Zool.
67:598 Y 616.