Oecologia

, Volume 139, Issue 3, pp 383–391

Genotype-specific response of a lycaenid herbivore to elevated carbon dioxide and phosphorus availability in calcareous grassland

Authors

    • Department of Integrative Biology, Section of Conservation BiologyUniversity of Basel
  • Andreas Erhardt
    • Department of Integrative Biology, Section of Conservation BiologyUniversity of Basel
  • Jürg Stöcklin
    • Department of Integrative Biology, Institute of BotanyUniversity of Basel
Plant Animal Interactions

DOI: 10.1007/s00442-004-1516-4

Cite this article as:
Goverde, M., Erhardt, A. & Stöcklin, J. Oecologia (2004) 139: 383. doi:10.1007/s00442-004-1516-4

Abstract

Effects of elevated CO2 and P availability on plant growth of the legume Lotus corniculatus and consequences for the butterfly larvae of Polyommatus icarus feeding on L. corniculatus were investigated in screen-aided CO2 control chambers under natural conditions on a calcareous grassland in the Swiss Jura mountains. Elevated CO2 conditions and P fertilisation increased the biomass production of L. corniculatus plants and affected the plant chemical composition. CO2 enrichment increased the C/N ratio and sugar concentration and decreased the N and P concentrations. C- and N-based allelochemicals (cyanoglycosides, total polyphenols and condensed tannins) were only marginally affected by CO2 enrichment. P fertilisation increased the specific leaf area and concentrations of water, N, sugar and P, while the C/N ratio and the concentration of total polyphenols decreased. Furthermore, P availability marginally enhanced the effect of elevated CO2 on the total dry mass and sugar concentration while the opposite occurred for the total polyphenol concentration. The changes in food-plant chemistry as a result of P fertilisation positively affected larval mass gain and accelerated the development time of P. icarus. Only a marginal negative effect on larval mass gain was found for CO2 enrichment. However, we found genotype-specific responses in the development time of P. icarus to elevated CO2 conditions. Larvae originating from different mothers developed better either under elevated CO2 or under ambient CO2 but some did not react to CO2 elevation. As far as we know this is the first finding of a genotype-specific response of an insect herbivore to elevated CO2 which suggests genetic shifts in insect life history traits in response to elevated CO2.

Keywords

Elevated carbon dioxide Lotus corniculatusNutrient availabilityPlant-insect interaction Polyommatus icarus

Introduction

The current anthropogenic increase in atmospheric CO2 is likely to affect plant-insect interactions (e.g. Watt et al. 1995; Coviella and Trumble 1999). Changes in leaf chemistry under elevated CO2, especially reduced leaf protein concentration which is most limiting for phytophagous insect development (Schoonhoven et al. 1998), can lead to decreased insect herbivore growth, increased larval development time, or compensatory feeding (e.g. Fajer et al. 1989; Ayres 1993; Lincoln et al. 1993; Bezemer and Jones 1998).

The impact of CO2 enrichment on herbivores has been investigated in about 60 studies (see e.g. Coviella and Trumble 1999; Goverde et al. 2002b). Most of these studies have been conducted under standardised experimental conditions with a relatively high power to detect even small effects of experimental treatments. However, the extrapolation of greenhouse and growth-chamber data to natural conditions is difficult and of limited value since the conditions of these experiments may neglect processes of ecological relevance and lead to the detection of effects not occurring in the field (Peters 1993; Körner 2000). For example, the effect of elevated CO2 on the development of a satyrid butterfly found in a greenhouse study (Goverde and Erhardt 2003) could not be found when using a screen-aided CO2 control (SACC) system under natural conditions on a calcareous grassland (Goverde et al. 2002b). The lack of agreement between these two experiments might be a result of natural climatic conditions, food-plant choice and long-term effects in the field experiment which mask the effect of elevated CO2 found in the greenhouse (Körner 1995; Goverde et al. 2002b).

The present study investigates effects of elevated CO2 on biomass production and leaf chemistry of the legume Lotus corniculatus L. (Fabaceae), and consequences for the butterfly larvae of Polyommatus icarus Rottemburg (Lepidoptera, Lycaenidae) feeding on L. corniculatus in a SACC system under field conditions [for further details on the SACC see Leadley et al. (1999)]. The same plant-insect system has been investigated in a growth-chamber and greenhouse study (Goverde et al. 1999; Bazin et al. 2002) and, therefore, comparisons of results can be made. Under elevated CO2 conditions L. corniculatus plants grown either in growth-chambers or in the greenhouse, increased concentrations of starch, condensed tannins and total polyphenols, and decreased cyanoglycoside concentrations (Goverde et al. 1999; Bazin et al. 2002). In both experiments, P. icarus developed faster under elevated CO2 than under ambient CO2 conditions. Since L. corniculatus plants were grown in a nutrient-rich soil, they were not N limited and, hence CO2 enrichment had no effect on the protein concentration of L. corniculatus. The faster development of the larvae on plants grown under elevated CO2 conditions was probably a result of the higher starch concentration of these plants compared to plants grown under ambient CO2 conditions (Goverde et al. 1999).

As shown by Stöcklin and Körner (1999) the biomass increase of legumes as a result of elevated CO2 conditions is much larger when plants are fertilised with P. This indicates that legumes’ responsiveness is largely controlled by P availability. Therefore, in the present study, we added a P-fertiliser treatment to investigate its direct and interactive effect with CO2 enrichment on the growth of L. corniculatus and its consequence for insect performance. Positive effects of fertilisation on insect growth were found in many other studies (e.g. Kinney et al. 1997; Kerslake et al. 1998; Hättenschwiler and Schafellner 1999). Larvae of the satyrid butterfly Coenonympha pamphilus reared on fertilised plants showed a lower development time and higher weights than larvae feeding on unfertilised plants (Goverde and Erhardt 2003). Furthermore, adult butterflies showed lower preferences for amino acid-rich nectar mimics when reared on fertilised plants, which indicates that their fertilised larval food plants were more nutritious than unfertilised control plants (Mevi-Schütz et al. 2003). However, in these studies interactive effects of elevated CO2 and fertilisation were found neither for larval development nor for adult nectar preference. Increased effects of elevated CO2 as a result of fertilisation were found for plant characters such as dry mass, leaf N and starch concentrations (e.g. Conroy et al. 1992; Entry et al. 1998; Goverde et al. 2002a), but so far not for insect growth (e.g. Williams et al. 1997; Hättenschwiler and Schafellner 1999; Goverde and Erhardt 2003 but see Kerslake et al. 1998).

Finally, the effect of elevated CO2 on insect performance has been shown to be species-specific or to depend on the insects’ feeding guild (Bezemer and Jones 1998). Genotype-specific responses to elevated CO2 have been shown for plants (e.g. Leadley and Stöcklin 1996; Goverde et al. 1999; Mansfield et al. 1999; Roumet et al. 1999; Lindroth et al. 2001). However, so far few studies have investigated the effect of elevated CO2 on different genotypes of herbivores. Larvae of P. icarus originating from different mothers show differences in larval, pupal and adult mass, in development time and in conversion efficiency (Goverde et al. 1999; Bazin et al. 2002). Thus the question arises if there are interactive effects of larval genotype and elevated CO2 conditions, which could result in an evolutionary advantage for a specific genotype within a population.

In the present study, we tested the following hypotheses:
  1. 1.

    Elevated CO2 conditions increase starch and C-based allelochemical concentrations of L. corniculatus plants.

     
  2. 2.

    Growth of L. corniculatus plants increases with P fertilisation, and the latter amplifies the effect of CO2 enrichment for plant growth.

     
  3. 3.

    The changes in leaf chemistry of L. corniculatus plants as a result of elevated CO2 and fertilisation positively affect the performance of P. icarus.

     
  4. 4.

    Larvae of P. icarus originating from different mothers differ in their responses to elevated CO2.

     

Materials and methods

Study site and CO2 enrichment

The present study was carried out in a semi-natural, undisturbed calcareous grassland near the village of Nenzlingen (47°27′N, 7°34′E) in the Swiss Jura mountains. The site is a moderately steep (ca. 20°), south-west-facing slope at an altitude of 520 m a.s.l.. The Rendzina soil consists of a 10- to 15-cm silty clay loam top horizon underlain by calcareous debris (Leadley et al. 1999). Originally covered by beech forest, these grasslands have been grazed by cattle for many centuries, leading to the characteristic Teucrio-Mesobrometum vegetation (Schläpfer et al. 1998). Snow usually covers the area for <1 month per year.

Beginning with the onset of the growing season in late March 1994, elevated CO2 was applied using SACC (for details see Leadley et al. 1997, 1999). Twenty-four experimental plots were arranged in a randomised block design. Each block contained six hexagonal SACC chambers reaching a height of 50 cm and enclosing 1.27 m2. In three plots per block, the atmospheric CO2 concentration was raised to 600 μl l-1, whereas the other three plots were exposed to present-day CO2 concentrations (368 μl l-1). In the outer part of the 1.27-m2 area of each plot several tubes (diameter 9.5 cm, depth approximately 30 cm) were dug in. For the present study L. corniculatus plants were planted in six tubes per plot (three tubes were situated in the upper left part of the plot and three in the lower right part).

Plant and insect material

Lotus corniculatus L. (Fabaceae) seeds were obtained from a local distributor (Wildstaudengärtnerei Willi-Durrer, Eschenbach), and were germinated in compost and transplanted after 12 days into small pots. On 12 March 1998, 27 days after sowing, two L. corniculatus plants per pot (10 cm diameter) were planted. The soil in these pots originated from the study site at Nenzlingen described above. To ensure nodulation of all plants, pots were watered with a solution containing rhizobial bacteria. Nodules of roots of L. corniculatus plants collected at Nenzlingen were crushed with a mortar and pestle, and left for 24 h in 100 mmol mannitol in tapwater (J. Müller, personal communication). This procedure was repeated 1 week later to ensure nodulation. Plants were grown in a greenhouse at the Botanical Institute Basel for 6 weeks. On 2 April 1998, L. corniculatus plants were randomly assigned to the tubes described above and planted in the field. To ensure growth, plants were watered on 25 April, and 4 8 11 and 13 May.

Polyommatus icarus Rottemburg (Lepidoptera, Lycaenidae) females were captured at different sites in the Swiss Jura mountains. Butterflies were kept in cages in the greenhouse to oviposit [for further detail see Goverde et al. (1999)]. Eggs from six different females were obtained and these were kept, separated by female, on moist filter paper in petri dishes covered by a nylon mesh. After hatching, batches of eight to ten larvae per female were placed in petri dishes containing shoots of L. corniculatus. Since the mortality of young larvae is high, P. icarus caterpillars were reared in the greenhouse until reaching their third larval instar (I3 ).

Experimental design

The six tubes per plot were randomly assigned to two groups. One group acted as control while the other group received granulated P fertiliser on 3 April and on 4 May. The total amount of P fertiliser per tube was 15 kg P ha-1.

On 8 June, 67 days after planting, shoots and leaves of all L. corniculatus plants were counted. Additionally, the cages for larval rearing were installed in the field. Each tube was covered by a fine nylon mesh of 30 cm height which was attached to a stick placed in the centre of the tube. Between 10 and 13 June freshly emerged I3 P. icarus larvae were weighed and released into the cages. Four of the six cages (two control and two with P fertiliser) per plot received one I3 caterpillar originating from one specific female. The two remaining tubes were used as control without herbivory. Per female genotype, two ambient and two elevated CO2 plots were used, resulting in four plots per female genotype and a total of six female genotypes.

Twelve days after they had been released in the cages, larvae were collected, weighed and returned to the original plants from which they were obtained. Between 7 and 9 July all larvae had reached their pupal stage. Pupae were collected and weighed, and singly placed in one of 18 compartments of a plastic box (21 cm×12 cm) with a nylon webbing floor. Boxes were placed in a greenhouse 5 cm above a pan with water to maintain high humidity. Boxes were daily checked for emerging adults. After they emptied their gut, emerged butterflies were weighed.

After removing the pupae, plant characteristics were measured. First, the number of leaves and shoots of the remaining L. corniculatus plants were counted. Then, to analyse cyanide concentration, six to seven leaves from each plant were removed, frozen in liquid nitrogen and stored at −80°C until analysis. To determine the specific leaf area (cm2 g-1 dry mass), a subsample of leaves was harvested and their leaf area was measured using a planimeter (LI-3100 area meter; LI-COR, Lincoln, Neb.). Finally, all remaining plants (above-ground and 20 cm of the root system) were harvested, weighed and dried at 60°C for 48 h. Leaf water concentration was measured as the proportional difference between fresh and dry leaf mass. Total shoot and root dry mass were determined. N and C concentrations were determined from dried leaves using a CHN analyser (model 932; LECO Instruments, St. Joseph, Mich.), which uses a combustion procedure. Non-structural carbohydrates (NSC) were determined using a series of enzymatic digestions (Wong 1990; Körner and Miglietta 1994). With this method, soluble-neutral sugar and NSC concentrations are measured whereas the starch concentration was calculated as the difference between NSC and the soluble-neutral sugar concentration. The P concentration was determined by using the molybdate blue ascorbic acid method (Watanabe and Olsen 1965). Cyanide samples were ground with liquid N using a mortar and pestle, and the HCN released was fixed in 0.5 ml of 1 N NaOH in a 37°C oven for 12 h. The HCN concentration was estimated by the method of Lambert et al. (1975) and expressed in equivalent mg CN g-1 dry mass. A standard absorption curve at 725 nm was established with commercial KCN (Prolabo, Briare, France). For total polyphenol and condensed tannin analysis, approximately 70-mg dried leaf samples were extracted in 10 ml of 70% acetone overnight (15 h) at 4°C in the dark. After filtration, the condensed tannin concentration [estimated at 550 nm and expressed by the extinction coefficient (E550)=150 in units of mg g-1; condensed tannins measured as anthocyanidin equivalents] was assessed using the acid butanol assay (Porter et al. 1986). For total polyphenols, the Prussian blue assay (Price and Butler 1977) was used with a standard curve of absorption at 725 nm established with commercial tannic acid (catechin; Fluka Chemie, Switzerland).

Statistical analysis

The data were analysed using the skeleton ANOVA presented in Table 1 which corresponds to a hierarchical split-plot design. For plant characters, first the variation of the factor block was fitted. Then, the main effect of elevated CO2 was fitted and tested against the variation among plots. All remaining factors and interactions were then tested against the residuals (Table 1). For insect characters the factor block could not be tested since the factor female genotype was not orthogonal to the factor block. The main factors CO2 and female genotype were tested against the variation in plots. The factor fertilisation and all other interactions were then tested against the residuals (Table 1). To eliminate the variation of differences between males and females, the covariate sex was fitted for both strata (plot-stratum and plots-by-units stratum; Table 1). Two statistical models were used for analysing larval mass after 12 days of growth in the field. First, larval mass was analysed using the model described above. Second, larval mass was analysed using the larval mass at the beginning of the experiment in the field facility as covariate. Finally, to meet the assumptions for ANOVA, some data were transformed. Square root transformation was applied for shoot/root ratio and total polyphenol concentration, log-transformation was used for sugar concentration and weight of third instar larvae, and inverse sine transformation was applied for leaf water concentration (Zar 1999). All statistical analyses were performed using Genstat 5 release 3.2 (Payne 1993). Values are given as mean ± 1 SEM.
Table 1

Skeleton ANOVA for measured plant characters and insect characters. Split-plot ANOVA, sum of squares type I, n =144 for plant characters; n =96 for weight of third instar larvae; n =81 for weight of larvae after 12 days; n =68 for pupal weight, adult weight and development time

Plant characters

Insect characters

Source of variation

df

Mean square (MS)

Variance-ratio (F)

Source of variation

df

Mean square

Variance-ratio (F)

Block (B)

3

MSB

MSB/MSP

CO2

1

MSCO2

MSCO2/MSP

CO2

1

MSCO2

MSCO2/MSP

Female genotype (FG)

5

MSFG

MSFG/MSP

Plot (P)

19

MSP

MSP/MSR

CO2×FG

5

MSCO2×FG

MSCO2×FG/MSP

Fertilisation (F)

1

MSF

MSF/MSR

Covariate sex (S)

1

MSS

MSS/MSP

Herbivory (H)

1

MSH

MSH/MSR

Plot (P)

11

MSP

MSP/MSR

CO2×F

1

MSCO2×F

MSCO2×F/MSR

Fertilisation (F)

1

MSF

MSF/MSR

CO2×H

1

MSCO2×H

MSCO2×H/MSR

CO2×F

1

MSCO2×F

MSCO2×F/MSR

F×H

1

MSF×H

MSF×H/MSR

FG×F

5

MSFG×F

MSFG×F/MSR

CO2×F×H (all)

1

MSall

MSall/MSR

CO2×FG×F (all)

5

MSall

MSall/MSR

Error (residuals; R)

114

MSR

S

1

MSS

MSS/MSR

Error (residuals; R)

59 a

MSR

aModel df, i.e. since there were some larvae or pupae lost, the effective error df is <59

Results

Effects of elevated CO2 and P fertilisation on plant characters

The effects of elevated CO2 and P fertilisation on plant parameters measured are given in Table 2. Elevated CO2 increased the total dry mass, C/N ratio and sugar concentration of L. corniculatus, and decreased the N and P concentrations (Table 2). Marginal effects were found for all allelochemicals measured. The cyanide concentration marginally decreased (F1,19 =3.13, P =0.09) while the total polyphenol and condensed tannin concentrations of L. corniculatus plants marginally increased under elevated CO2 conditions (F1,19 =3.19, P =0.09 and F1,19=3.13, P =0.09, respectively).
Table 2

Summary of plant parameters (means±SEM) under ambient and elevated CO2 conditions without (Control) and with P fertiliser. The proportional effect of elevated CO2 and P fertilisation (P) is indicated. Significant CO2 or P main effects are shown (for interactive and marginal effects see the text). n No. of samples per treatment, SLA specific leaf area, d.m. dry mass

n

Ambient CO2

Elevated CO2

CO2 effect

P effect

Control

P fertiliser

Control

P fertiliser

Control

P fertiliser

Ambient CO2

Elevated CO2

Number of shoots (before herbivory)

12

20.06±0.56

23.28±0.98

20.25±0.53

24.00±0.99

1.0%

3.1%

16.1%

18.5%

F1,114 =17.64,P <0.001

No. of leaves (before herbivory)

12

129.50±4.11

163.31±8.39

131.22±5.91

161.61±6.17

1.3%

−1.0%

26.1%

23.2%

F1,114 =27.57,P <0.001

No. of shoots (after herbivory)

12

25.19±0.96

28.78±1.27

26.06±0.86

29.44±1.66

3.4%

2.3%

14.2%

13.0%

F1,114 =8.40,P =0.005

No. of leaves (after herbivory)

12

159.83±8.82

225.28±13.42

160.89±5.79

220.53±14.02

0.7%

−2.1%

40.9%

37.1%

F1,114 =40.87,P <0.001

Total dry mass (g)

12

2.99±0.20

3.96±0.16

3.63±0.21

5.40±0.38

21.5%

36.3%

F1,19 =11.48,P =0.003

32.6%

48.8%

F1,114 =46.06,P <0.001

Leaf water (%)

12

67.5±0.6

69.1±0.8

66.6±0.7

68.2±0.8

−1.4%

−1.2%

2.4%

2.5%

F1,114 =11.92,P <0.001

SLA (cm2g−1d.m.)

4

167.94±7.93

180.19±11.54

147.07±17.73

187.80±20.76

−12.4%

4.2%

7.3%

27.7%

F1,9 =9.84,P =0.012

N (%)

12

2.82±0.08

3.13±0.06

2.54±0.05

2.78±0.05

−10.0%

−11.2%

F1,19 =22.98,P <0.001

10.9%

9.4%

F1,114 =32.95,P <0.001

C/N ratio

12

14.99±0.44

13.72±0.30

16.74±0.34

15.44±0.31

11.7%

12.5%

F1,19 =21.04,P <0.001

−8.5%

−7.8%

F1,114 =24.34,P <0.001

Sugar (%)

12

3.74±0.20

4.16±0.36

4.33±0.38

5.47±0.23

15.7%

31.5%

F1,19 =6.90,P =0.017

11.1%

26.3%

F1,114 =9.59,P =0.002

Starch (%)

12

9.74±1.05

8.64±1.09

10.89±1.11

10.46±1.42

11.9%

21.1%

−11.2%

−4.0%

P (%)

12

4.49±0.31

6.59±0.23

3.54±0.22

5.34±0.21

−21.1%

−19.1%

F1,19 =15.50,P <0.001

46.9%

50.7%

F1,114 =102.41,P <0.001

Cyanide (mg g-1)

12

0.97±0.08

1.03±0.10

0.87±0.09

0.85±0.05

−10.6%

−17.7%

6.2%

−2.2%

Total polyphenols (mg g-1)

12

2.98±0.12

2.93±0.13

3.48±0.15

3.04±0.17

16.8%

3.8%

−1.9%

−12.7%

F1,114 =4.78,P =0.031

Condensed tannins (mg g-1)

12

96.31±4.59

97.11±6.39

119.17±7.01

101.36±7.49

23.7%

4.4%

0.8%

−14.9%

Most plant parameters were positively affected by P fertilisation (Table 2). The strongest effects of P fertilisation were found for the number of shoots and leaves, for total dry mass and for P concentration of L. corniculatus plants (Table 2). A negative effect of P fertilisation was only found for the C/N ratio and the total polyphenol concentration (Table 2).

Additionally, one significant and some marginally significant interactive effects of elevated CO2 and P fertilisation were found. The effect of elevated CO2 on root dry mass was stronger when plants were fertilised (F1,114 =4.80, P =0.031). For total dry mass as well as for sugar concentration the effect of elevated CO2 tended to be more pronounced in fertilised plants (Table 2; F1,114=3.88, P =0.051 and F1,114=2.91, P =0.091, respectively). The opposite effect appeared for concentrations of total polyphenols and condensed tannins, where P fertilisation tended to decrease the effect of elevated CO2 (Table 2, F1,114=3.39, P =0.068 and F1,114=3.72, P =0.056, respectively).

Effects of herbivory on plant characters

The feeding of P. icarus caterpillars negatively affected the number of L. corniculatus leaves (F1,114 =5.03, P =0.027). Plants that experienced herbivory had 11.2% fewer leaves than control plants without caterpillars. The number of shoots and the total dry mass, however, were unaffected (F1,114 =0.67, P =0.42 and F1,114=0.10, P =0.76, respectively). Feeding of P. icarus caterpillars increased the concentration of N by 4.2% (F1,114 =5.31, P =0.023). Corresponding to this, the C/N ratio was significantly reduced by herbivory (F1,114 =5.95, P =0.016). The strongest effect of herbivory was found for starch concentration, where plants with herbivore damage had 26.2% less starch than control plants without caterpillars (F1,114 =26.43, P <0.001). Additionally, this effect depended on P fertilisation (F1,114 =5.16, P =0.025). The effect of herbivory on starch concentration was stronger in unfertilised (34.0%) than in fertilised (16.3%) L. corniculatus plants. Finally, herbivory marginally increased condensed tannin concentrations by 10.4% in plants with herbivore damage compared to control plants (F1,114 =3.66, P =0.058).

Effects on insect development and insect genotype-specific responses

Larval mass after 12 days of growth in the field was marginally reduced under elevated CO2 conditions (Fig. 1, Table 3). Larval mass was increased by P fertilisation (Fig. 1), and additionally this increase depended on the female genotype (fertilisation-by-female genotype interaction; Table 3). Some larvae originating from a specific mother reacted more strongly to P fertilisation than others. However, when using larval mass at the beginning of the experiment as a covariate only the main effect of fertilisation remains significant (F1,44 =23.57, P <0.001). For pupal and adult fresh mass no significant main effect of elevated CO2 was found (Table 3). However, a three-way interaction of CO2, female genotype and fertilisation appeared, i.e. the effect of elevated CO2 on pupal and adult mass depended on the origin of the larvae as well as on P fertilisation. Strong differences in pupal and adult fresh mass among larvae from different females were found (Table 3; female genotype effect). Additionally, as for larval mass, female genotype affected the outcome of the P fertilisation effect on pupal and adult fresh mass, while only a marginal main effect of P fertilisation on pupal fresh mass was found (Table 3). Finally, the effect of elevated CO2 on development time depended on female genotype, i.e. some larvae originating from a specific mother developed better under elevated CO2, some under ambient CO2 and some did not react to CO2 elevation (Fig. 2). As for larval mass, development time was affected by fertilisation (Table 3). Larvae feeding on fertilised L. corniculatus plants developed within 32.4±0.3 days, while they took 34.0±0.6 days to fully develop on unfertilised plants.
Fig. 1

Mass of Polyommatus icarus larvae after feeding for 12 days on Lotus corniculatus plants exposed to ambient (A) and elevated (E) CO2 conditions, without or with P fertilisation. Means+SEM, n =48

Table 3

ANOVA for larval, pupal and adult fresh mass and development time of Polyommatus icarus reared onLotus corniculatus plants grown under ambient or elevated CO2 conditions with or without P fertilisation

Source of variation

Larval mass after 12 days

Pupal fresh mass

Adult fresh mass

Development time

df

MS

F

P

df

MS

F

P

df

MS

F

P

df

MS

F

P

CO2

1

3,270.1

4.55

0.056

1

190.7

2.33

0.155

1

25.5

1.65

0.225

1

3.7

1.31

0.277

Female genotype (FG)

5

70.5

0.10

0.990

5

847.6

10.36

<0.001

5

192.0

12.42

<0.001

5

2.6

0.91

0.507

CO2×FG

5

1,099.8

1.53

0.258

5

43.1

0.53

0.752

5

32.7

2.12

0.139

5

16.2

5.71

0.008

Covariate sex (S)

1

8,794.2

12.25

0.005

1

517.3

6.32

0.029

1

62.3

4.03

0.070

1

20.7

7.32

0.020

Plot (P)

11

718.1

2.34

0.035

11

81.8

1.52

0.176

11

15.5

0.91

0.546

11

2.8

0.39

0.950

Fertilisation (F)

1

7,028.0

22.88

<0.001

1

207.3

3.84

0.059

1

21.2

1.24

0.274

1

50.6

7.02

0.013

CO2×F

1

391.2

1.27

0.269

1

32.8

0.61

0.441

1

53.6

3.14

0.086

1

0.5

0.07

0.792

FG×F

5

792.6

2.58

0.049

5

372.2

6.90

<0.001

5

69.2

4.05

0.006

5

10.8

1.49

0.220

CO2×FG×F

5

291.4

0.95

0.466

5

205.3

3.80

0.008

5

51.8

3.03

0.024

5

8.6

1.20

0.333

S

1

1,278.1

4.16

0.051

1

442.0

8.19

0.007

1

1.9

0.11

0.738

1

0.8

0.11

0.738

Error (residual)

27

307.2

1.11

31

54.0

31

17.1

31

7.2

Fig. 2

Development time from third larval instar to adult eclosion of P. icarus under ambient (open bars) and elevated (black bars) CO2 conditions. Means+SEM for the six different female genotypes (FG) are presented, n =8 in each case. For other abbreviations, see Fig. 1

Discussion

The novelty of the present study is to investigate the interactive effects of maternal butterfly-specific, i.e. genotype-specific, responses to elevated CO2 and P fertilisation under natural conditions in the field, thus larvae from a specific mother might react differently to elevated CO2 and/or P fertilisation than other larvae. Many studies have demonstrated that the effect of elevated CO2 on plants can influence insect herbivores. However, most studies investigating this effect were conducted under controlled conditions in greenhouse or growth-chamber experiments and, so far, genotype-specific responses of an insect herbivore have not been considered.

Although there are some studies investigating the effect of elevated CO2 on insect herbivores under more natural conditions (e.g. Bezemer et al. 1998; Jones et al. 1998; Stiling et al. 2003) there is still a lack of studies which connect findings from greenhouse studies with findings from experiments conducted under more natural conditions (but see Bezemer et al. 1999; Goverde et al. 2002b). The plant-insect system used here had already been investigated in a growth-chamber and a greenhouse experiment (Goverde et al. 1999; Bazin et al. 2002). In these studies elevated CO2 conditions caused L. corniculatus to increase its biomass, to accumulate starch and C-based allelochemical and to reduce N-based allelochemicals. However, the N concentration did not change. Larvae of the butterfly P. icarus feeding on plants grown under elevated CO2 conditions showed higher rates of consumption, conversion efficiency and consequently accelerated growth.

In the present study, an increase in atmospheric CO2 affected L. corniculatus plant growth and foliar chemistry. L. corniculatus plants grown under elevated CO2 conditions showed a decrease in N concentration. A lower N concentration under elevated CO2 conditions has been found in many plants (Cotrufo et al. 1998), but was not expected in the present study due to earlier findings from a greenhouse experiment (Goverde et al. 1999). The difference might be an effect of soil type. In the greenhouse study a nutrient-rich soil (compost) was used while in the present study plants were grown in their original soil in a nutrient-poor calcareous grassland. Results, thus, confirm that the choice of soil type and/or the environmental conditions of the experiments with elevated CO2 can affect the outcome. It is, therefore, difficult to extrapolate results gained from greenhouse experiments to natural conditions (see also Körner 2000). The surplus C in the elevated CO2 treatment was mainly stored as sugar and not as starch as predicted. However, since sugar and starch are both easily digested carbohydrates for insect herbivores, the surplus of carbohydrates might positively affect insect growth (Chapman 1998). Finally, the predicted increase in C-based allelochemicals under elevated CO2 conditions was only marginally significant.

Amplification of CO2 effects by P fertilisation

N-fixing plants obtain a large amount of their N requirement from the atmosphere and, therefore, N-fixing plants might rather be P-limited than N-limited and interactive responses of P availability and elevated atmospheric CO2 are likely (Stöcklin et al. 1998; Stöcklin and Körner 1999). The predicted interactive effect of P fertilisation and CO2 enrichment on the growth of L. corniculatus was significant for root dry mass and some marginal effects (for details see Results). As suggested by Stöcklin and Körner (1999), nutrient availability has an over-riding effect on CO2 responsiveness, and species-specific responses in biomass to elevated CO2 and fertilisation will affect community structure and biodiversity by suppressing the less responsive species. Thus, N-fixing plants may play a very important role in a future CO2-enriched environment (Cotrufo et al. 1998). Furthermore, as predicted in our second hypothesis, the growth of L. corniculatus plants was enhanced by P fertilisation. Fertilised plants had more shoots, leaves and a higher dry mass. Also, most other plant characteristics were increased (Table 2). Probably a combination of all these changes did affect insect growth positively. P. icarus larvae feeding on fertilised plants had higher weights and developed faster. However, the difference in larval weight disappeared in the adult stage, indicating compensatory feeding (Bezemer and Jones 1998).

The effect of herbivory on L. corniculatus

The feeding of P. icarus larvae on L. corniculatus plants had several, probably connected effects. Plants with herbivore damage had less leaves but their dry mass was the same. This indicates compensatory growth after herbivory which has already been found in this plant species (Bazin et al. 2002). Since young or fresh leaves have higher protein concentrations (e.g. Gleadow and Woodrow 2000; Alonso and Herrera 2002; Riipi et al. 2002), compensatory growth might also explain the higher N concentration in L. corniculatus leaves with herbivore damage. The lower starch concentration in L. corniculatus plants with herbivore damage indicates the preference for digestible C by P. icarus larvae. Herbivores feeding on N-fixing legumes might rather be limited by the accessibility of digestible carbohydrates than by protein (Goverde et al. 1999), and plants with higher starch concentrations are more digestible for herbivores (Slansky and Rodriguez 1987; Lincoln et al. 1993). Finally, the marginal increase in condensed tannin concentrations in L. corniculatus plants with herbivore damage may indicate a potential of the plant to show an induced response (Schoonhoven et al. 1998).

Effects on insect performance

Especially the higher C/N ratio of plants grown under elevated CO2 conditions reduces the tissue quality for herbivorous insects. This reduction in food-plant quality is likely to reduce herbivore fitness (Bezemer and Jones 1998; Stacey and Fellowes 2002). We predicted a positive effect of elevated CO2 on insect performance, since such an effect had been found in greenhouse studies for the same insect-plant system (Goverde et al. 1999; Bazin et al. 2002). However, the present study does not support this prediction. Elevated CO2 had no effect on pupal and adult mass, but a marginally negative effect on larval mass after 12 days—which, however, disappears when using larval mass at the beginning as a covariate. These results differ clearly from previous ones from a growth-chamber study where mass gain of P. icarus larvae was increased under elevated CO2 conditions (Goverde et al. 1999). As mentioned above, the choice of a low-fertility soil and the environmental conditions did affect the outcome of the experiment.

Larval genotype matters

In the present study we found insect genotype-specific responses to elevated CO2 and to P fertilisation. The effect of elevated CO2 on the development time of P. icarus depended on the origin of the larvae. Larvae of two mothers prolonged their development time while larvae of three mothers did not react, and the larvae from one mother even accelerated their development time under elevated CO2 conditions. A similar effect was found in the growth chamber and greenhouse study for the same plant-insect system (Goverde et al. 1999; Bazin et al. 2002). CO2 concentration can thus be considered as a potential selective factor for some genotypes of this butterfly species if rapid development is advantageous, i.e. a change in selective pressure might affect the genetic diversity of single species. Thus, we find not only species-specific but also genotype-specific responses to elevated CO2 in insect herbivores (Bezemer and Jones 1998). In addition, larvae from a specific mother reacted more strongly to P fertilisation-induced changes in larval food quality than other larvae (significant interaction between P fertilisation and female genotype). As in the response of development time to elevated CO2 this finding suggests genetic variation in insect life history traits in response to P fertilisation. Finally, the effect of elevated CO2 on pupal and adult mass depended not only on the origin of larvae but also on P fertilisation (three-way interaction). A further three-way interaction among CO2 enrichment, insect genotype and plant genotype does appear for several insect life history traits of P. icarus feeding on different L. corniculatus genotypes under ambient and elevated CO2 (Goverde et al. 1999). These results show the complexity of ecological systems and the challenge to investigate real effects of the ongoing increase in atmospheric CO2 on plant-insect dynamics.

Conclusion

In conclusion, we showed that CO2 enrichment in SACC chambers affected the growth and chemical composition of L. corniculatus plants. These changes in food-plant quality partially affected the development of the butterfly P. icarus. Remarkably, the interactive effect of CO2 enrichment and origin of larvae showed that there is genetic variability in the insect population leading them to react differently to changes in the quality of larval food-plants caused by increased atmospheric CO2 conditions and P fertilisation. Thus, the effect of environmental changes on insect-plant interactions can depend on genotype-specific responses in both plants and insects. So far, genotype-specific responses in insect-plant interactions have been poorly investigated. However, both are of high relevance for evolutionary processes, i.e. a change in selective pressure will affect the genetic composition within populations of a species and hence the genetic diversity of single species. Our study suggests that rising levels of CO2 might be such a selective factor, affecting both plant and herbivore populations and their interaction. Therefore, genotype-specific responses must be considered because this will affect the outcome of elevated CO2 for plant-herbivore interactions.

Acknowledgements

We are especially grateful to Kathrin Schweizer, Daniel Bretscher, Monika Wohlfender, Olivier Bignucolo, Susanna Pelaez-Riedl for technical support, to Bernhard Schmid and Christian Körner for providing us with the facility in Nenzlingen. This research was supported by grants from the Swiss Priority Program Environment of the Swiss National Science Foundation to A. Erhardt (no. 5001–044622/1) and to C. Körner (no. 5001–035214).

Copyright information

© Springer-Verlag 2004