You are on page 1of 7

Oecologia (1995) 104:72-78

9 Springer-Verlag 1995

J.A. A r n o n e III- J.G. Zaller 9 C. Ziegler 9 H. Zandt Ch. Kiirner

Leaf quality and insect herbivory in model tropical plant communities after long-term exposure to elevated atmospheric CO2

Received: 27 July 1994 / Accepted: 21 April 1995

Abstract Results from laboratory feeding experiments

have shown that elevated atmospheric carbon dioxide can affect interactions between plants and insect herbivores, primarily through changes in leaf nutritional quality occurring at elevated CO 2. Very few data are available on insect herbivory in plant communities where insects can choose among species and positions in the canopy in which to feed. Our objectives were to determine the extent to which CO2-induced changes in plant communities and leaf nutritional quality may affect herbivory at the level of the entire canopy. We introduced equivalent populations of fourth instar Spodoptera eridania, a lepidopteran generalist, to complex model ecosystems containing seven species of moist tropical plants maintained under low mineral nutrient supply. Larvae were allowed to feed freely for 14 days, by which time they had reached the seventh instar. Prior to larval introductions, plant communities had been continuously exposed to either 340 gl CO 2 1-1 or to 610 gl CO 2 1-1 for 1.5 years. No major shifts in leaf nutritional quality [concentrations of N, total non-structural carbohydrates (TNC), sugar, and starch; ratios of: C/N, TNC/N, sugar/N, starch/N; leaf toughness] were observed between CO 2 treatments for any of the species. Furthermore, no correlations were observed between these measures of leaf quality and leaf biomass consumption. Total leaf area and biomass of all plant communities were similar when caterpillars were introduced. However, leaf biomass of some species was slightly greater - and for other species slightly less (e.g. Cecropia peltata) - in communities exposed to elevated CO 2. Larvae showed the strongest preference for C. peltata leaves, the plant species that was least abundant in all communities, and fed relatively litJ.A. Arnone III ( ~ ) 9J.G. Zaller 9Ch. K6rner Department of Botany, University of Basel, Sch6nbeinstrasse 6 CH-4056 Basel, Switzerland Fax: +41 61 267 35 04 E-mail: arnone@urz.unibas.ch C. Ziegler 9H. Zandt Department of Zoology, University of Basel, Rheinsprung 9, CH-4051 Basel, Switzerland

tle on plants species which were more abundant. Thus, our results indicate that leaf tissue quality, as described by these parameters, is not necessarily affected by elevated CO 2 under relatively low nutrient conditions. Hence, the potential importance of CO2-induced shifts in leaf nutritional quality, as determinants of herbivory, may be overestimated for many plant communities growing on nutrient-poor sites if estimates are based on traditional laboratory feeding studies. Finally, slight shifts in the abundance of leaf tissue of various species occurring under elevated CO 2 will probably not significantly affect herbivory by generalist insects. However, generalist insect herbivores appear to become more dependent on less-preferred plant species in cases where elevated CO 2 results in reduced availability of leaves of a favoured plant species, and this greater dependency may eventually affect insect populations adversely.
Key words Carbon dioxide enrichment. Feeding preference 9Leaf quality 9Mesocosms 9 Spodoptera eridania

Introduction
Interactions between herbivorous insects and their host plants are expected to be altered significantly as atmospheric CO 2 concentrations continue to rise (cf. Fajer et al. 1989; Bazzaz 1990; Lincoln et al. 1993). Results from laboratory studies indicate that exposure to elevated CO 2 can result in reductions in leaf nutritional quality, and that these reductions can dramatically influence insect herbivory. Relative to leaves produced under ambient CO> leaves grown under elevated CO a have often been shown to have greater C:N ratios (mainly due to accumulation of non-structural carbohydrates; e.g. Cave et al. 1981; Wong 1990; Johnson and Lincoln 1991; K6rner and Arnone 1992; Wullschleger et al. 1992; K6rner and Miglietta 1994), higher concentrations of non-carbonbased allelochemicals (Lincoln and Couvet 1989; Johnson and Lincoln 1991; Lindroth et al. 1993), and increas-

OECOLOGIA 104 (1995) 9 Springer-Verlag es in leaf toughness (Lindroth et al. 1993). Responses of insect herbivores to CO2-induced reductions in leaf quality are highly variable but may include compensatory feeding (Lincoln et al. 1984; Fajer 1989; Fajer et al. 1989; Lincoln and Couvet 1989; Johnson and Lincoln 1990, 1991), slower or incomplete development of larvae (Fajer et al. 1991), decreased fecundity (Fajer et al. 1989) and increased mortality (Osbrink et al. 1987; Fajer 1989; Fajer et al. 1989). Laboratory feeding studies and growth chamber experiments have provided useful information about the responses of insect herbivores to particular leaf tissue or to individual plants grown under ambient and elevated CO 2. However, they are inherently inadequate for predicting how herbivory may be affected in complex plant communities where herbivores not only respond to variations in leaf nutritional quality, but where they can choose among species and positions in the canopy in which to feed. Thompson and Drake (1994), however, do show that reductions in leaf N concentration can occur in native plant communities after several growing seasons of treatment with elevated CO 2, and that these reductions were correlated with increases in insect herbivory. It is uncertain, however, to what extent species shifts observed in these communities under elevated CO 2 (Arp et al. 1993) might indirectly affect these insect herbivores. Long-term exposure of plant assemblages to high CO 2 has been shown to result in shifts in species dominance to varying degrees (e.g. Williams et al. 1986; Reekie and Bazzaz 1989; Nie et al. 1992; Arp et al. 1993; Owensby et al. 1993; Polley et al. 1994; Arnone and K6rner 1995). Such shifts could lead to increases or reductions in plant species favoured by native insect herbivores. Particularly in these cases, even slight shifts could substantially affect insect herbivory. Herbivores themselves may then in turn amplify any shifts in plant species' abundance which first result from exposure to elevated CO 2. More studies [furthering the strides made by Thompson and Drake (1994)] are urgently needed (Lincoln et al. 1993) in order to understand the ecological implications of the CO2-induced alterations in feeding behaviour observed in laboratory studies, and to judge the relative importance of tissue-level and community-level effects. We created an outbreak of insect herbivores by introducing equivalent larval populations of a lepidopteran generalist species to model communities of moist tropical plant species which had been exposed for 1.5 years to current ambient and elevated concentrations of atmospheric CO 2 (Arnone and K6rner 1995). Our objectives were to determine: (1) if long-term exposure to elevated CO 2 results in shifts in leaf nutritional quality at the entire canopy level, (2) how closely feeding is correlated to leaf nutritional quality, and (3) the extent to which the slight CO2-induced shifts in species abundance (Arnone and K6rner 1 9 9 5 ) m a y affect herbivory. Our results indicate that the potential impacts of elevated CO 2 on insect herbivory in plant communities cannot be reliably predicted simply based on changes in leaf tissue quality and from traditional laboratory feeding experiments.

73

Material and methods
Construction of artificial communities and growth conditions Model tropical plant ecosystems were constructed as described in Arnone and KOrner (1995) by installing equivalent communities of moist tropical plants in each of four large-scale (17 m3) clear plastic houses. Seven species of plants, each represented by 6-12 individuals in each community (77 plants), were installed over a ground area of 6.7 m2 in a common soil substrate. The seven species were the dicots Cecropia peltata (Cecropiaceae), Ficus benjamina (Moraceae) and Ficus pumila, and the monocots Ctenanthe lubbersiana (Marantaceae) Elettaria cardamomum (Zingiberaceae), Heliconia humilis (Musaceae) and Epipremnum pinnatum (Araceae). Two of the communities were maintained at daytime CO2 concentrations of 340 gl CO21-1 and two at 610 gl 1-1 for 530 days. All communities were maintained under humid tropical conditions, and nutrients were supplied at levels limiting to growth. Details of the growth conditions are described by Arnone and K6rner (1995). Larvae Larvae of the lepidopteran generalist Spodoptera eridania (Noctuidae) were raised from eggs on a nutrient-rich synthetic diet in petri dishes at 23~ until the fourth instar. We introduced 100 of these larvae to each community (15 larvae per m2 of ground) in the evening of day 515 of CO2 treatment. Larvae were placed randomly on leaves of all species in each community. Larvae were allowed to feed for a total of 14 days, by which time they had reached the seventh instar. Fourth instar larvae were used because we thought that their larger size would facilitate censusing in the dense plant communities, and because we anticipated that any CO2-induced differences in herbivory would be most easily detected during the later instars when insects normally consume greater amounts of leaf tissue. Results of a preliminary experiment showed that the rate of survival of first and second instar larvae placed on individual plants (in adjacent growth chambers) was highly variable.This indicated a high risk of creating very unequal starting larval populations in each of our communities and further convinced us to introduce slightly older larvae.The number of larvae feeding or sitting on leaves in each plant community was estimated from counts made every several days (and also at night) from outside each plastic house. Leaf quality Leaf quality was measured to help explain the variability observed in larval leaf consumption. Since larvae fed on leaves of all ages and at all positions in the canopy of each species, we lumped leaves from each individual together at harvest. Leaf subsamples were then taken from this pool, immediately dried at 70~ and subsequently ground to powder for chemical analyses. Leaf nonstructural carbohydrates [sugars and total non-structural carbohydrates (TNC)] were determined following the procedure used by Wong (1990) and described in K6rner and Miglietta (1994). In this procedure, sugars were measured spectrophotometrically after the hexokinase reaction on samples which were first heated in boiling water followed by reaction with invertase and isomerase. TNC was determined in the same way except that heated samples were first incubated with clarase. Starch concentration was calculated as the difference between TNC and sugar concentration. N concentrations of dried and powdered subsamples from leaves of each individual plant were determined with a CHN analyser (Model 932, LECO Instruments, St. Joseph Mich, USA). The following additional measures of leaf quality were also examined: ratios of C/N, TNC/N, sugar/N, starch/N. Specific leaf areas (SLA, cm2 g 1) were also determined for leaf subsamples taken from each plant to calculate leaf biomass consumed, as well as pre- and post-larval

74

O E C O L O G I A 104 (1995) 9 Springer-Verlag performed. A ranking of herbivore species preferences was then calculated using post hoc contrast analyses and the Scheff~ test.

leaf biomass. SLA was used as a measure of leaf toughness - lower SLAs being equivalent to tougher leaves.

Quantification of the leaf canopy and herbivory Since the actual amount of leaf tissue consumed in all communities was small, we examined each individual leaf at harvest for herbivore damage and measured the area eaten by the larvae. This was accomplished by placing the damaged leaves on top of overhead transparencies and, using the leaf as a stencil, filling in the missing parts created by the larvae with a black opaque felt-tipped pen. The marked transparencies were then run through a Licor LI3000 leaf area meter to measure the area consumed. Total pre-larval leaf area and the leaf area index (LAI) were then calculated by adding the area consumed to the total post-larval area measured with the LI-3000 at harvest. Leaf growth in all of the species was negligible during the 2-week observation period.

Results
Canopy pre-larval leaf quality and availability Leaf quality was surprisingly similar under both CO 2 treatments, but varied significantly among species for TNC and SLA (Table 1). Leaf N concentrations ranged from 16 to29 mg g-l; however, within a species, N concentrations were remarkably similar at both CO 2 levels. SLAs of monocot leaves, and those of Cecropia peltata, were also not significantly different between CO 2 treatments, however SLAs of leaves of the two Ficus species were lower at elevated CO 2. Furthermore, no significant changes were observed in the concentration of TNC at elevated CO 2 for any of the species. Age distribution of leaves within individual species were similar in all communities, based on observations at harvest. Leaf subsamples taken from each individual for chemical analyses adequately represented leaf age distributions. The monocot Elettaria cardamomum represented most (62-64%) of the total community leaf biomass in all communities when larvae were introduced, followed by Ficus benjamina, and the two monocots Heticonia humilis and Ctenanthe lubbersiana (Table 2). Cecropia peltata was one of the most dominant species in the canopy at the early part of the CO 2 treatment period (e.g. 17.3% of initial community leaf biomass; Arnone and K6rner 1995). However, by the time we introduced the larvae its leaves were the least abundant (0.4-0.9% of total leaf biomass) in all communities. LAI (Arnone and K6rner 1995) and leaf biomass at both the level of the individual species and the entire community were nearly identical under ambient and elevated CO 2 before larvae were introduced (Table 2). No statistically significant (P < 0.05) species-specific CO 2 effects were observed in leaf biomass; however, one group of species gained in leaf biomass under elevated
total non-structural carbohydrate concentration (TNC); and specific leaf area (SLA, as a measure of leaf toughness) (dwt dry weight) SLA (cm 2 g-l) Elev. CO 2 10.7• 6.2• 4.5• 20.1• 10.0• 14.9• Amb. C O 2 208• 235• 283• 267• 175• 128• a 262• b Elev. C O 2 205• 238• 280• 284• 164• 115• 210•

Calculation of herbivore preference We first calculated relative consumption of each species in each community. This was expressed as the amount of leaf biomass of a species consumed by larvae in a community relative to the total amount of leaf biomass of that species available at the beginning of the 14 day feeding period. Larvae were considered to have shown a preference for a particular plant species in a community if they consumed more leaf biomass of that species than would have been expected as a result of a simple chance encounter with leaves of that species (Crawley 1983).

Statistical analyses A nested statistical design and ANOVA as described by Arnone and K6rner (1995) were used to test the effects of CO 2, plant species, and the plant species x CO 2 interaction for all dependent variables. In addition, linear regressions of leaf biomass consumption on all measures of leaf quality were performed for each plant species and for each CO 2 level. Regressions were based on individual plant data. Fisher LSD post hoc contrast analyses were used to test for significant differences in leaf quality between CO s treatments at the species level. Data were log transformed to conform to parametric ANOVA principles of equal variance. Since neither the CO 2 effect nor the CO 2 x species interaction term was statistically significant for the relative consumption rates, values from both CO 2 treatments (i.e. n = 4 per species) were pooled and a one-way ANOVA with species as the independent variable was

Table 1 Leaf quality averaged over entire plant communities for each species at harvest after 530 days exposure to ambient (Arab.) and elevated (Elev.) CO 2 (mean_+SE; n=2): nitrogen concentration; Leaf [N] (mg g-l) Amb. CO 2 Monocots Elev. C O 2 18.5• 23.0• 23.6_+1.2 21.8• 17.7• I 22.4_+0.9

TNC (%, dwt) Arab. CO 2 8.8• 5.9• 5.3• 21.2• 9.1• 14.5• No data

Elettaria cardomomum Ctenanthe lubbersiana Heliconia humilis Epipremnumpinnatum
Dicots

20.3• 22.8• 23.8• 22.1• 19.4• 23.4• No data

Cecropia peltata Ficus benjamina Ficus pumila

a, b The only significant C O 2 effects occurred for SLA of the two Ficus species (a P<0.05; b P<0.15; t-tests). ANOVAs showed no significant CO 2 or COz x species effects

Table 2 Community pre-larval leaf biomass by species (see text for full names), and each species' contribution to plant community pre-larval leaf biomass, after 530 days of exposure to ambient (Arab) and elevated (Elev.) CO2; amount of community leaf biomass consumed by larvae during 14 days of feeding, and the percentage of total consumption Community leaf consumption Biomass (mg m -2 ground) Amb. CO 2 76 _ + 3 7 8.6_+7.1 78 - + 2 7 26 _+11 449 _+288 42 _+ 26 0 680 _+398 329-+161 100.0 163_+82 49_+40 0 62.8 _+5.6 5.9 _+0.4 0 33+10 10_+ 1 54_+23 20-+ 7 12.1 _+1.7 1.00-+0.46 13.9 _+4.1 4.3 - + 0 . 8 11.2_+2.5 4.0_+1.8 17.2_+1.5 6.6-+1.2 48.8_+1.1 12.1_+5.9 0 100.0 Elev. C O 2 Amb. C O 2 Elev. CO 2 % Total biomass consumption Relative consumption ~ % g Consumption g-1 available Arab. CO 2 Elev. CO 2

represented by each plant species (means_+SE; n=2); and the percentage of available (i.e. pre-larval) community leaf biomass which was consumed (i.e. plant species preference shown by Spondoptera eridania larvae)

Community preqarval leaf availability % Total biomass Amb. C O 2 62.4 +1.4 10.2 _+0.8 9.5 _+1.6 0.62+0.12 0.91-+0.59 15.5 _+4.2 0.93-+0.12 100.0 100.0 0.37-+0.01 20.4 _+1.1 1.6 _+0.6 64.4 _+0.6 5.1 _+2.6 7.6 +0.3 0.47-+0.04 Elev. CO 2

Larval preference ~ (rank) b (n=4 per species, both CO2 treatments lumped)

Biomass ( g m -2 ground)

Arab. C O 2

Elev. CO 2

Monocots Elettaria Ctenanthe Heliconia Epipremnum

122.5+6.1 19.9_+0.1 18.5_+1.9 1.2_+0.3

130.5 +5.6 10.2 _+4.9 15.4 _+1.1 0.94-+0.06

0.06_+0.03 0.04_+0.03 0.41-+0.11 2.6 _+1.6 25.4 _+1.6 0.18+0.15 0 0.36_+0.23

0.03_+0.01 3 0.13_+0.06 3 0.36+0.17 3 2.1 _+0.06 2 9 22.2 -+11.7 0.13-+0.11 0 0.17_0.08 1 3 3 o 9 9 >

Dicots Cecropia F. benjamina F. pumila ___7

1.7_+ 1.0 31.1_+10.5 1.8-+ 0.4

0.74_+0.02 41.3 _+3.6 3.3 _+1.4

Total

197 +14

202

Results of ANOVAs <0.01 0.42 <0.01 0.26 <0.01 0.41

P 0.02

P 0.09

P 0.75

P 0.04
<0.01 0.26

P 0.80
<0.01 0.98

df Factora P 6 species <0.01 21 resid.

4~

,g
tal

df 1 2 6 6 12

Factor CO 2 residual Species CO2xspecies residual

9

a Consumption of leaf biomass of each species relative to the amount available before larvae were introduced. Larval preference was calculated using relative consumption data b Post hoc contrast analyses of larval preference using the Scheff4 test

--M t~

76
o1:,

OECOLOGIA 104 (1995) 9 Springer-Verlag
0,4' --, A m b .

E

CO2 D[]
CO 2 [] []

(a)

A

A

(b)

(c)

Etev.

0.3

A
A [] A A A A A 9

8
0.2" E
0

[]

m[]_.q
r-i

O 9 O
A
o 9 9 .

0.1
~mu []
i 9

J

0.0

. . . .

t

. . . .

i

. . . .

0

10

20

30

0

10

2'0

0

160

2()0 3()0

Leaf [N] (mg N g-l)

TNC (% dwt.)

SLA (cm 2 g-l)

Fig. la-c Cecropiapeltata leaf biomass consumption by Spodoptera eridania larvae in 14 days of feeding as a function of: a leaf nitrogen concentration, b leaf total non-structural carbohydrate (TNC) concentration, and c specific leaf area (SLA; increasing SLA equivalent to decreasing leaf toughness) for plants growing in communities which had been exposed for 515 days to either ambient (340 gl 1-1) or elevated (610 gl 1-1) C O 2 prior to the introduction of larvae to the communities. Larvae fed upon all leaves of C. peltata in all communities, and tissue quality was measured for these leaves. Points represent individual plants from the four communities (open symbols ambient CO2, filled symbols elevated CO2, dwt dryweight)
C O 2 while another group lost (P < 0.01). Some margin-

ally significant (P < 0.15) species-level shifts in leaf biomass were observed under elevated CO 2 (e.g. Ctenanthe lubbersiana leaf biomass was 50% lower in communities treated with elevated CO2; Table 2). For Cecropia peltata, an average of 55% less leaf biomass was present in communities at elevated CO 2 (Table 2; Fig. 1) but these stand-level differences (not significant) were due to the presence of relatively high values in one of the two ambient CO 2 communities.

Herbivory Larvae actively fed throughout the entire 14 day experimental period on leaves of all ages and at all positions in the canopy (i.e. sun and shade leaves) for all species, apart from the ground creeper Ficus pumila where no consumption was seen (Table 2). We observed an apparent decline in feeding activity towards the end of this period as estimated from larval counts made every several days from outside each house. Night-time activity was greater, but at any given point during daylight hours between 20 and 50 larvae could be seen feeding in all four communities. Larvae were also frequently observed crawling along the soil surface, presumably in search of new plants or leaves. While moulting, larvae which were visible remained anchored on leaves (mainly dicots) or at leaf bases (mainly monocots). A thorough assessment of larval populations during the experiment turned out to be intractable because we found that feeding was badly disrupted by our movements within the dense canopy (larvae dropped from their leaves to the ground). We terminated the herbivory experiment (harvested the plant com-

munities) when the larvae in all communities which were visible had reached the seventh instar and could no longer be seen feeding. Consumption of leaf tissue was unrelated to any of our measures of leaf quality for any of the species, either within a CO 2 treatment or across both CO 2 treatments. This lack of correlation could clearly be shown for data from individual Cecropia peItata, where all leaves were heavily fed upon (Fig. 1). Larvae consumed between 0.17 and 0.36% of total biomass available to them in all communities (Table 2). As might be expected from the absence of CO2-induced differences in leaf quality, the amount of teaf biomass consumed by larvae in communities treated with ambient and elevated CO 2 did not differ significantly at either the level of the individual species or the entire community (Table 2). This was also the case even when just the three most-consumed species were considered (Cecropia peltata, Epipremnum pinnatum and H. humilis). Although not statistically significant, a trend (with respect to the means) toward decreased consumption of leaves of these species was apparent at elevated CO 2 (Table 2). Larvae showed clear and similar preferences for particular plant species in both ambient and elevated CO 2 communities (Table 2). Cecropia peltata, one of the least abundant species, was by far the most preferred, while Elettaria cardamomum, the most abundant species, was among the least preferred. Larvae also demonstrated a secondary preference for Epipremnum pinnatum. The overall reduction in the availability of leaves of Cecropia peltata under elevated CO 2, although not statistically significant, did result in a greater proportional feeding on less-preferred plant species (e.g. Ctenanthe lubbersiana and Epipremnum pinnatum; Table 2). However, we found no evidence that leaves of less-preferred species were consumed any differently under ambient and elevated CO 2 (Table 2).

Discussion

CO2-induced effects on herbivory via leaf quality? The absence of reductions in leaf N concentration (Table 1; Fig. 1), and lack of increases in leaf TNC levels (Table 1), under elevated CO 2 in our present study was surpris-

O E C O L O G I A 104 (1995) 9 Springer-Verlag

77

ing and is in marked contrast to most of the published data, including our own (K6rner and Arnone 1992). We attribute this mainly to the fact that: (1) communities were maintained under relatively low nutrient supply designed to mimic conditions prevailing in many native ecosystems (Arnone and K6rner 1995), and (2) we analysed pooled samples taken from the entire leaf canopy instead of selecting specific leaves. While nutrient availability certainly constrained biomass accumulation in all of the plant communities (Arnone and K6rner 1995), we found that: (1) leaf N concentrations observed were typical of those measured in native tropical plant communities (cf. Mattson 1980; Jordan 1985; Vitousek and Sanford 1986), (2) the amount of nutrients supplied was not excessively low (Arnone and K6rner 1995), and (3) percolation water contained negligible amounts of nutrients (J.A. Arnone et al., unpublished data). Nutrient limitation in our ecosystems may have also precluded dramatic increases in photosynthesis, and thus precluded large increases in leaf TNC and dilution of leaf N. In systems where nutrients are relatively plentiful, reductions in leaf N concentrations under elevated CO 2 still appear to be important determinants of herbivory at the community level (Thompson and Drake 1994). Leaf TNC and N data from field studies are rare. K6rner and Miglietta (1994) reported increases in TNC concentration and decreases in leaf N concentration in a large number of Mediterranean herbaceous species growing near natural CO 2 springs. However, they report only slight or no changes in these properties in leaves of the climax tree species Quercus ilex, and in several other woody species growing in the shade. Likewise, Cipollini et al. (1993) also found no changes in leaf TNC under elevated CO 2 in the woody understory plant Lindera benzoin. In our previous study (K6rner and Arnone 1992), much higher leaf TNC concentrations were observed at elevated CO 2 in species common to both the 1992 experiment and our current study. However, differences in leaf TNC concentrations between ambient and elevated CO 2 in monocot species in the 1992 study, as well as in understory leaves of all species, were also much smaller. In the cun'ent study, monocot species represented more than 80% of the leaf biomass in all communities (Table 2). Data from these earlier studies, as well as those from our present experiment, suggest that increases in leaf TNC and decreases in leaf N may not always occur under elevated CO 2, especially when viewed at the level of the entire canopy (i.e. CO2-induced differences in leaf TNC and N weighted by the actual amount of leaf biomass in the canopy showing a given shift in these measures). Thus, canopy-level shifts in leaf quality may actually be better predictors of the potential effects of elevated CO 2 on generalist herbivores feeding in dense canopies than shifts based on unweighted leaf values. Despite the lack of CO 2 effects on leaf quality, we still expected to find a reasonable correlation between herbivory and leaf N concentration (Mattson 1980). However, consumption was unrelated to leaf N even when calculated across both CO 2 treatments and even for

the most preferred species (Cecropia pe#ata; Fig. 1). The lack of correlations between consumption and all measures of leaf quality examined in our study may have been due to the absence of large variation in these measures within and between CO 2 treatments (e.g. Fig. 1). This may also hold true for many native plant communities where naturally low levels of soil nutrients (e.g. Chapin 1980) may greatly reduce the predicted adverse effects of elevated CO 2 on leaf quality and herbivory. While leaf N is considered to be the most important overall driver of herbivory (e.g. Mattson 1980), other physical and chemical properties of leaves have also been shown to be critical determinants of herbivory for some plant species (cf. Coleman 1986). For example, relatively low concentrations of defence compounds (e.g. polyphenols, tannins) may account at least in part for the overwhelming preference shown for Cecropia peltata leaves in all communities in our study, as indicated by the short leaf life spans of the pioneer species C. peltata (cf. Coley et al. 1985).

CQ-induced effects on herbivory via shifts in species' leaf abundance? We observed no statistically significant differences in either leaf abundance or leaf consumption under ambient and elevated CO 2 at either the individual species level or the entire canopy level. Thus, the marginally significant CO2-induced shifts in species' leaf biomass observed before larvae were introduced had no apparent effects on herbivory (Table 2). Given the absence of shifts in the leaf qualities measured, increased consumption of leaves via compensatory feeding behaviour (e.g. Fajer et al. 1989; Johnson and Lincoln 1991) of the larvae of these generalist insects was not expected and was not found to occur (Table 2). However, trends observed in mean leaf biomass consumption per unit ground area suggest that feeding may even decrease in plant communities grown under elevated CO 2. Spodoptera eridania larvae had to climb over 1.2 m of leafless stem to reach the foliage of Cecropia peltata and thus actively selected this species above all others in all communities (Table 2). Larvae also fed on this species in proportion to its leaf availability. However, because of the high variability in both the amount of C. peltata leaf biomass available and the amount of C. peltata leaf biomass consumed in communities under ambient CO 2 treatment, it is unclear to what extent the interaction of elevated CO 2 and generalist insect herbivores will affect the competitive fitness of favoured plant species. Thus, our results indicate that leaf nutritional quality may not necessarily be reduced at high CO 2 in ecosystems where the availability of nutrients is relatively low. Furthermore, herbivory was not related to key measures of leaf quality, even in the most preferred plant species. Hence the potential importance of these leaf qualities as determinants of herbivory at the plant-community level

78

OECOLOGIA 104 (1995) 9 Springer-Verlag

may be overestimated if based on traditional laboratory feeding studies. Finally, slight shifts in the abundance of leaf tissue of various species occurring under elevated CO 2 will probably not significantly affect herbivory by generalist insects. However, generalist insect herbivores appear to become more dependent on less-preferred plant species in cases where elevated CO 2 results in reduced availability of leaves of a favoured plant species (Table 2), and this greater dependency may eventually affect insect populations adversely.
Acknowledgements We thank: R StOcklin of CIBA-Geigy for providing us with S. eridania larvae and for his advice on maintaining them, B Sch~ippi for conducting the carbohydrate analyses, and H. Mfiller-Schfirer and two anonymous reviewers for valuable comments on a previous version of this manuscript.

References
Aruone JA III, KOrner Ch (1995) Soil and biomass carbon pools in model communities of tropical plants under elevated CO 2 Oecologia 104:61-71 Arp WJ, Drake BG, Pockman WT, Curtis PS, Whigham DF (1993) Interactions between C3 and C4 salt marsh plant species during four years of exposure to elevated atmospheric CO 2. Vegetatio 104/155:133-143 Bazzaz EA (1990) The response of natural ecosystems to the rising global CO 2 levels. Annu Rev Ecol Syst 21:167-196 Cave T, Tolley LC, Strain BR (1981) Effect of carbon dioxide enrichment on chlorophyll content, starch content and starch grain structure in TrifoIium subterraneum leaves. Physiol Plant 51:171-174 Chapin FS III (1980) The mineral nutrition of wild plants. Annu Rev Ecol Syst 11:233-260 Cipollini ML, Drake BG, Whigham DF (1993) Effects of elevated CO 2 on growth and carbon/nutrient balance in the deciduous woody shrub Lindera benzoin (L.) Blume (Lauraceae). Oecologia 96:339-346 Coleman JS (1986) Leaf development and leaf stress: increased susceptibility associated with sink-source transition. Tree Physiol 2:289-299 Coley PD, Bryant JR Chapin FS III (1985) Resource availability and plant antiherbivore defense. Science 230:895-899 Crawley MJ (1983) Herbivory: the dynamics of animal-plant interactions. Blackwell, Oxford Fajer ED (1989) The effects of enriched CO 2 on plant-insect herbivore interactions: growth responses of larvae of the specialist butterfly, Junonia coenia (Lepidoptera: Nymphalidae). Oecologia 81:514-520 Fajer ED, Bowers MD, Bazzaz FA (1989) The effects of enriched carbon dioxide atmospheres on plant-insect herbivore interactions. Science 243:1198-1200 Fajer ED, Bowers MD, Bazzaz FA (1991) The effects of enriched CO 2 atmospheres on the buckeye butterfly, Junonia coenia. Ecology 72:751-754 Johnson RH, Lincoln DE (1990) Sagebrush and grasshopper responses to atmospheric carbon dioxide concentration. Oecologia 84:103-110

Johnson RH, Lincoln DE (1991) Sagebrush carbon allocation patterns and grasshopper nutrition: the influence of CO 2 enrichment and soil mineral limitation. Oecologia 87:127-134 Jordan CF (1985) Nutrient cycling in tropical forest ecosystems. Wiley, New York K6rner Ch, Arnone JA III (1992) Responses to elevated carbon dioxide in artificial tropical ecosystems. Science 257:1672-1675 KOrner Ch, Miglietta F (1994) Long-term effects of naturally elevated CO 2 on Mediterranean grassland and forest trees. Oecologia 99:343-351 Lincoln DE, Couvet D (1989) The effect of carbon supply on allocation to allelochemicals and caterpillar consumption of peppermint. Oecologia 78:112-114 Lincoln DE, Sionit N, Strain BR (1984) Growth and feeding responses of Pseudoplusia includens (Lepidoptera: Noctuidae) to host plants grown in controlled carbon dioxide atmospheres. Environ Entomol 13:1527-1530 Lincoln DE, Eajer ED, Johnson RH (1993) Plant-insect herbivore interactions in elevated CO 2 environments. Trends Ecol Evol 8:64-68 Lindroth RL, Kinney KK, Platz CL (1993) Responses of deciduous trees to elevated atmospheric CO2: productivity, phytochemistry, and insect performance. Ecology 74:763-777 Mattson WJ Jr (1980) Herbivory in relation to plant nitrogen content. Annu Rev Ecol Syst 11:119-161 Nie D, Kirkham MB, Ballou LK, Lawlor DJ, Kanemasu ET (1992) Changes in prairie vegetation under elevated carbon dioxide levels and two soil moisture regimes. J Veg Sci 3:673-678 Osbrink WLA, Trumble JT, Wagner RE (1987) Host suitability of Phaseolus lunata for Trichoplusia ni (Lepidoptera: Noctuidae) in controlled carbon dioxide atmospheres. Environ Entomol 16:63%644 Owensby CE, Coyne PI, Ham JM, Auen LM, Knapp AK (1993) Biomass production in a tallgrass prairie ecosystem exposed to ambient and elevated CO 2. Ecol Appl 3:644-653 Polley HW, Johnson HB, Mayeux HS (1994) Increasing CO2: comparative response of the C 4 grass Schizachyrium and grassland invader Prosopis. Ecology 75:976-988 Reekie EG, Bazzaz EA (1989) Competition and patterns of resource use among seedlings of five tropical trees grown at ambient and elevated CO 2. Oecologia 79:212-222 Thompson GB, Drake BG (1994) Insects and fungi on a C3 sedge and a C4 grass exposed to elevated atmospheric CO 2 concentrations in open-top chambers in the field. Plant Cell Environ 17:1161-1167 Vitousek PM, Sanford RL (1986) Nutrient cycling in moist tropical forests. Annu Rev Ecol Syst 17:137-167 Williams WE, Garbutt K, Bazzaz FA, Vitousek PM (1986) The response of plants to elevated CO 2. IV. Two deciduous-forest tree communities. Oecologia 69:454-459 Wong S-C (1990) Elevated atmospheric partial pressure of CO 2 and plant growth. II. Non-structural carbohydrate content in cotton plants and its effect on growth parameters. Photosyn Res 23:171-180 Wullschleger SD, Norby R J, Hen&ix DL (1992) Carbon exchange rates, chlorophyll content, and carbohydrate status of 2 forest tree species exposed to carbon dioxide enrichment. Tree Physiol 10:21-31