THE PLANT STRESS HYPOTHESIS AND VARIABLE

RESPONSES BY BLUE GRAMA GRASS

(Bouteloua gracilis) TO WATER,

MINERAL NITROGEN, AND

INSECT HERBIVORY

ANTHONY JOERN1,* and SIMON MOLE2,3

1Division of Biology, Kansas State University, Manhattan, KS 66506, USA2School of Biological Sciences, University of Nebraska-Lincoln, Lincoln,

NE 68588-0118, USA

(Received January 3, 2005; revised May 9, 2005; accepted May 17, 2005)

AbstractActing 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 sourceYsink relationships and foliarchemistry, 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 C4 grass

Bouteloua gracilis. Water, nitrogen fertilizer, and herbivory from the grass-

feeding 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 carbohy-

drate (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

0098-0331/05/0900-2069/0 # 2005 Springer Science + Business Media, Inc.

2069

Journal of Chemical Ecology, Vol. 31, No. 9, September 2005 (#2005)

DOI: 10.1007/s10886-005-6078-3

* To whom correspondence should be addressed. E-mail: ajoern@ksu.edu3 Current address: Boulder, CO, USA.

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 WordsVChewing insects, environmental stress hypothesis, functional-convergence-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 sourceYsink relationships (Mooney and Chiariello, 1984; Bazzazand Grace, 1997). The plant stress hypothesis states that environmental stresses

on plants decrease plant resistance to insect herbivory by altering whole-plant

sourceYsink resource allocation schedules and foliar chemistry, thus changingfood 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

(sourceYsink relationships, translocation products), and partitioning of resourcesto 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)Voften 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 deter-

mined 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

2070 JOERN AND MOLE

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 C4 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

2071PLANT STRESS HYPOTHESIS

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 mj2 (unpublished data). C3and C4 grass species typical of eastern tallgrass prairie and western shortgrass

steppe grasslands intermingle at the site. Dominant plants in this sand dune land-

scape form loose but recognizable vegetation associations along the existing

topographic gradient (Barnes, 1985). The grass canopy is intermingled with ex-

tensive bare ground, largely because of extensive disturbance from pocket gophers.

Long-term annual mean precipitation (1951Y1980) recorded 15 km fromArapaho Prairie at Arthur County, NE, averaged 47.1 cm (SD = 8.98 cm) from

FIG. 1. Precipitation patterns at Arapaho Prairie. (a) Annual rainfall with mean and 95%

confidence intervals, 1987Y2000. (b) Seasonal pattern of precipitation illustrated bycumulative amount by date for the 3 years of the study.

2072 JOERN AND MOLE

US Weather Bureau records; the recent 14-year record from Arapaho Prairie

(1987Y2000) averaged 37.3 cm (SD = 11.4 cm). The amount and timing ofprecipitation at Arapaho Prairie varies greatly among years (Figure 1). Below-

average 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 80Y85% sand with low nutrient concen-trations (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 C4 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 20Y30% of the relative cover of valleys andmidslope 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 experi-

ments were run concurrently, one addressing effects of water, N fertilizer, and

grasshopper herbivory on plant response, and the other investigating grasshop-

per performance in response to water and N-fertilizer treatments on plants.

Rectangular cages (basal area 0.5 m2, 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 Bturf^ in earlyJune, corresponding to the initiation of growth. Cages housing treatment com-

binations 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 re-

sponses to stress. Biomass accumulation and foliar chemical responses (% total

2073PLANT STRESS HYPOTHESIS

nitrogen, TN; and % total nonstructural carbohydrates, TNC) by B. gracilis to

multiple stresses was studied using a 3 2 2 full-factorial treatmentcombination (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 combi-

nations. 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 2full-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 2Y3 d from censuses of individualsremaining 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).

2074 JOERN AND MOLE

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 mj2

wkj1 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 (NH4NO3). Levels included 0, 3, and 6 g N mj2 of N fertilizer

(N0, N3, and N6 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. 2Y3 g dry weight (d.w.)] was collected, immediately flash-frozen in liquidnitrogen 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 H2SO4, a CuSeO4 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 stan-

dard without the hydrolysis of sucrose. Sucrose averaged about 0.4Y0.5% d.w.

2075PLANT STRESS HYPOTHESIS

of plant material compared with 17Y22% d.w. plant material for TNC asmeasured 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 50Y100% greater in an average rainfall

FIG. 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 mj2), and grasshopper

herbivory (GH0, GH+)] for each year of the study.

2076 JOERN AND MOLE

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 expe-

rienced varying degrees of overall stress. Both water (1989: F1,56 = 17.4,

P < 0.001; 1990: F1,56 = 6.6, P = 0.013; 1991: F1,56 = 11.3, P < 0.001) and N

fertilizer additions (1989: F2,56 = 4.2, P < 0.021; 1990: F2,56 = 11.2, P < 0.001;

1991: F2,56 = 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, F1,56 = 34.8, P < 0.001; 32% in 1991,

F1,56 = 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 1990Y1991 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 (3Y10%), and only significantly differentin 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

10Y20% higher level of total foliar-N compared with W+ treatments. Theweakest 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

2077PLANT STRESS HYPOTHESIS

responses between the 2 years (Figure 5). In very dry 1989, higher foliar TN

levels were seen in W0 only for the N0 treatment. No differences were seen

between W0 and W+ for the N3 and N6 fertilization treatment levels. In a year

of average rainfall (1990), there was no difference in foliar TN between water

treatments at N0, but significant and about equal increases in total N for N3 and

N6 treatments in interaction with water availability.

Grasshopper herbivory affected foliar TN levels significantly as a main

effect only in 1991. However, grasshopper feeding interacted with other treat-

ments 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 N3, a TN level that

was reached with N6 with no grasshoppers. In 1990, grasshopper herbivory

interacting with water availability led to higher TN level that was reached in the

FIG. 3. Responses (mean, SE) in (a) % total N and (b) % TNC to main treatments (water

addition, grasshopper herbivory, and N fertilizer) for each year.

2078 JOERN AND MOLE

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 5Y10%, with1989 exhibiting the highest foliar TNC levels. Differences in responses among

all treatment combinations showed little variation in 1989 and 1991 compared

FIG. 4. Percentage of total variance in foliar nutrient responses explained by experi-

mental treatments in each year of study. (a) % Total foliar nitrogen (TN) and (b) % total

nonstructural carbohydrates (TNC). Letters refer to main effects (N, nitrogen fertili-

zation; 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.

2079PLANT STRESS HYPOTHESIS

t1.1

TA

BL

E1

.A

NO

VA

OF

FO

LIA

RN

UT

RIE

NT

RE

SP

ON

SE

SO

FB

.g

raci

lis

LE

AF

TIS

SU

EA

TT

HE

EN

DO

FT

HE

SE

AS

ON

IN

RE

SP

ON

SE

TO

MU

LT

IPL

ES

TR

ES

SE

S

So

urc

e

All

yea

rs1

98

91

99

01

99

1t1.2

df

FP

df

FP

FP

FP

t1.3

Fo

lia

rT

N(G

N/G

lea

fd

ryw

eig

ht)

t1.4

Mo

del

R2

57

.36

8.7

61

.06

1.5

t1.5

Yea

r2

14

2.8

0.0

01

YY

YY

YY

Yt1.6

Blo

ck5

4.2

0.0

01

57

.40

.001

5.4

0.0

01

2.2

0.0

7t1.7

N-F

erti

lize

r2

10

.90

.001

26

.80

.002

4.9

0.0

18

.30

.001

t1.8

Wat

er1

17

.10

.001

11

4.9

0.0

01

18

.50

.00

11

4.0

0.0

01

t1.9

Gra

ssh

opp

er1

15

.70

.001

10

.01

0.9

1.3

0.2

65

.90

.02

t1.10

N-F

erti

lize

r

Wat

er2

0.6

0.4

52

4.5

0.0

26

.20

.00

43

.10

.05

t1.11

N-F

erti

lize

r

Gra

ssh

opp

er2

0.7

0.5

62

5.4

0.0

07

2.6

0.0

90

.20

.82

t1.12

Wat

er

Gra

ssh

opp

er1

3.4

50

.03

13

.80

.05

4.9

0.0

30

.90

.34

t1.13

N-F

erti

lize

r

Wat

er

Gra

ssh

op

per

21

.97

0.1

02

1.7

0.1

82

.90

.06

5.0

0.0

1t1.14

Err

or

19

75

5t1.15

t1.16

TN

Cs

(gT

NC

/gle

af

dry

wei

gh

t)t1.17

Mo

del

R2

39

.83

7.5

91

.66

6.0

t1.18

Yea

r2

25

.80

.001

YY

YY

YY

Yt1.19

Blo

ck5

0.6

0.9

50

.80

.63

.10

.02

2.9

0.0

2t1.20

N-F

erti

lize

r2

9.8

0.0

01

27

.30

.002

7.9

0.0

01

20

.80

.001

t1.21

Wat

er1

2.9

0.0

91

4.8

0.0

30

.02

0.8

96

.80

.0t1.22

Gra

ssh

opp

er1

29

.40

.001

11

.00

.34

42

.00

.00

10

.02

0.8

9t1.23

N-F

erti

lize

r

Wat

er2

0.8

0.5

20

.10

.92

1.3

0.0

01

2.2

0.1

3t1.24

N-F

erti

lize

r

Gra

ssh

opp

er2

1.0

0.3

52

2.6

0.0

92

1.5

0.0

01

4.0

0.0

2t1.25

Wat

er

Gra

ssh

opp

er1

0.0

60

.81

0.1

0.8

11

.60

.00

11

6.3

0.0

01

t1.26

N-F

erti

lize

r

Wat

er

Gra

ssh

op

per

20

.94

0.3

92

2.2

0.1

21

8.0

0.0

01

7.5

0.0

01

t1.27

Err

or

19

75

5t1.28

2080 JOERN AND MOLE

with 1990. Total variance in TNC levels among treatments was 1.5Y5 timesgreater 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 sig-

nificant as main effects, with no significant statistical interactions. When com-

pared 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: r2 = 0.014; 1990: r2 = 0.001;

1991: r2 = 0.038; P > 0.05 for all years). However, when years were analyzed

separately, interesting differences were observed.

In general, TNC declined 4Y6.5% with increased N fertilizer in all years,although no significant differences were observed between the 3 g and 6 g

FIG. 5. Responses of significant interactions among treatments for % total foliar N (TN)

for each year of study.

2081PLANT STRESS HYPOTHESIS

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 3Y4%. Generally, grasshopper herbivory was afactor 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 N6fertilizer 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

FIG. 6. Responses of significant interactions among treatments for % total nonstructural

carbohydrate (TNC) for each year of study.

2082 JOERN AND MOLE

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

FIG. 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 trans-

formed as natural log of number alive at each census period.

2083PLANT STRESS HYPOTHESIS

individuals remaining in cages of P. nebrascensis (Figure 7a) was significant

(Wilks l = 0.10, P < 0.001). However, although observed trends in survivalmay be suggestive, no significant effect of water and N fertilizer treatments

were detected. The significant difference in the repeated-measures ANOVA re-

flected 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, Wilks l = 0.137, F6,21 = 22.06, P /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 150 /GrayImageDepth -1 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputCondition () /PDFXRegistryName (http://www.color.org?) /PDFXTrapped /False

/SyntheticBoldness 1.000000 /Description >>> setdistillerparams> setpagedevice