Oecologia (2005) 143: 578–587 DOI 10.1007/s00442-005-0008-5
P L AN T A N IM A L I NT E R AC TI O NS
Kathleen L. Prudic Æ Jeffrey C. Oliver Æ M. Deane Bowers
Soil nutrient effects on oviposition preference, larval performance, and chemical defense of a specialist insect herbivore
Received: 1 September 2004 / Accepted: 11 January 2005 / Published online: 14 April 2005 Springer-Verlag 2005
Abstract This study examined the effects of increased leaf nitrogen in natural host-plants (Plantago spp.) on female oviposition preference, larval performance, and larval chemical defense of the butterfly Junonia coenia. Increased availability of soil nutrients caused the hostplant’s foliar nitrogen to increase and its chemical defense to decrease. Larval performance did not correlate with increases in foliar nitrogen. Larval growth rate and survival were equivalent across host-plant treatments. However, larvae raised on fertilized host-plants showed concomitant decreases in chemical defense as compared to larvae reared on unfertilized host-plants. Since most butterfly larvae cannot move long distances during their first few instars and are forced to feed upon the plant on which they hatched, J. coenia larval chemical defense is determined, in large part, by female oviposition choice. Female butterflies preferred host-plants with high nitrogen over host-plants with low nitrogen; however, this preference was also mediated by plant chemical defense. Female butterflies preferred more chemically defended host-plants when foliar nitrogen was equivalent between host-plants. J. coenia larvae experience intense predation in the field, especially when larvae are not chemically well defended. Any qualitative or quantitative variation in plant allelochemical defense has fitness consequences on these larvae. Thus, these results indicate that females may be making sub-optimal oviposition decisions under a nutrient-enriched regime, when predators are present. Given the recent increase in Communicated by Oswald Schmitz K. L. Prudic (&) Æ J. C. Oliver Æ M. D. Bowers Ecology and Evolutionary Biology, University of Colorado, 334 UCB, Boulder, CO 80309, USA E-mail:
[email protected] Present address: K. L. Prudic Ecology and Evolutionary Biology and the Bio5 Institute, University of Arizona, PO Box 210088, Tucson, AZ 85721, USA Present address: J. C. Oliver Interdisciplinary Program in Insect Science, University of Arizona, PO Box 210036, Tucson, AZ 85721, USA
fertilizer application and nitrogen deposition on the terrestrial landscape, these interactions between female preference, larval performance, and larval chemical defense may result in long-term changes in population dynamics and persistence of specialist insects. Keywords Nutrient enrichment Æ Lepidoptera Æ Herbivore performance Æ Female preference Æ Host-plant quality
Introduction Recent anthropogenic nutrient availability changes in terrestrial systems have had a significant effect on ecosystem dynamics (Jefferies and Maron 1997; Whittaker 2001; Matson et al. 2002; Vitousek et al. 2002). Nutrient enrichment from agricultural and atmospheric sources has the potential to alter plant–insect interactions via changes in plant growth and defense (Whittaker 2001; Coviella et al. 2002; He et al. 2002; Richardson et al. 2002). Soil nutrient augmentation has been shown to change plant community structure (Pauli et al. 2002; Richardson et al. 2002), as well as the corresponding plant–herbivore dynamics (Kinney et al. 1997; Kerslake et al. 1998; Whittaker 2001; Richardson et al. 2002), although the direction of these changes can be systemspecific. While changes in the plant allelochemistry due to nutrient enrichment have been demonstrated (e.g. Oyeyele and Zalucki 1990; Fajer et al. 1992; Hugentobler and Renwick 1995; Bezemer et al. 2000; Coviella et al. 2002), few studies have investigated the effect of nutrient enrichment on specialist insects’ chemical defenses (Rank et al. 1998). Host-plant nitrogen content is viewed as the ultimate limiting nutrient for most chewing insects (Mattson 1980). Augmenting soil nutrients often increases plant nitrogen concentration and reduces production of some allelochemicals, resulting in higher growth and consumption rates in generalist phytophagous insects (e.g.
579
Schafellner et al. 1996; Lindroth and Kinney 1998). Because host-plant nitrogen limits phytophagous insect growth and development, specialist phytophagous insects should also demonstrate increased larval performance on high nutrient plants. However, these presumed benefits may be offset by other changes in plant chemistry which may affect specialist phytophagous insects differently from generalist phytophagous insects. Specialist phytophagous insects commonly use plant allelochemicals for host-plant location and hostplant identification, and these allelochemicals may influence female oviposition choices (Chew 1979; Feeny et al. 1983; Honda 1986; Pereyra and Bowers 1988). Some specialist phytophagous insect species also sequester plant allelochemicals for protection against their own predators (Bowers 1990). Lower levels of plant chemical defense often correlate with lower levels of insect chemical defense (e.g. Pasteels et al. 1988; Malcolm 1995) with potential consequences on insect palatability (Brower et al. 1970; Camara 1997b). As a result, variation in plant allelochemicals has potential fitness consequences on specialist insects that rely on these compounds for host-plant location, oviposition cues, or chemical defense. The present study examines the effects of soil nutrient enrichment on the interactions between a hostplant and a specialist insect. We used the buckeye butterfly, Junonia coenia Hu¨bner (Nymphalidae) and two principal host-plants, narrow-leafed plantain, Plantago lanceolata L., and common plantain, Plantago major L. (Plantaginaceae) to evaluate how soil nutrient changes affect plant chemistry, larval performance, larval defensive chemistry, and female oviposition preference. Previous experiments demonstrated that one of the major groups of secondary metabolites in P. lanceolata, iridoid glycosides, decreases with an increase in soil nutrients (Fajer et al. 1992; Jarzomski et al. 2000). Also, J. coenia larvae reared on P. lanceolata, which has high levels of iridoid glycosides, are rejected more often by an invertebrate predator than larvae reared on P. major, which has low levels of iridoid glycosides (Theodoratus and Bowers 1999). High larval predation in the field (90%) and local adaptation to plant allelochemicals by J. coenia demonstrate the importance of larval chemical defense for larval survival (Camara 1997b, c) Most butterfly larvae cannot move long distances during their first few instars and feed upon the plant on which they hatched (Mayhew 1997); therefore, J. coenia larval chemical defense will be determined, at least in part, by female oviposition choices. In J. coenia butterflies, host-plant iridoid glycoside concentration influences female oviposition preference (Pereyra and Bowers 1988; Klockars et al. 1993). Females prefer to lay eggs on substrates containing relatively high levels of the iridoid glycoside catalpol than on those containing relatively low, or no catalpol (Pereyra and Bowers 1988). We designed a set of experiments to answer three questions about the potential effects of increased soil
nutrient availability on the dynamics of plant chemistry, larval performance, larval chemical defense, and female oviposition choice in a specialist phytophagous insect: (1) How does soil nutrient enrichment affect plant performance and chemistry? (2) How do host-plant chemical changes resulting from soil nutrient enrichment influence larval performance and defensive chemistry? and (3) How do host-plant chemistry changes in response to soil nutrient enrichment influence female oviposition choice? To address questions 1 and 2, we performed a greenhouse experiment using J. coenia and P. lanceolata. Since both insect and plant growth are generally nutrient limited, we expected P. lanceolata and J. coenia to grow and develop better under increased soil nutrient regimes. Also since iridoid glycosides are carbon based allelochemicals, we predicted plant defensive chemistry would decrease with increasing soil nutrient availability (e.g. Fajer et al. 1992; Jarzomski et al. 2000). Since J. coenia larvae sequester iridoid glycosides from their larval hostplants, we predicted the larval defensive chemistry would also decrease with increasing soil nutrient enrichment. To address question 3, we evaluated the relative roles of both host-plant nutrient availability and hostplant defensive chemistry for female oviposition preference. To investigate host-plant nutrient availability, we compared J. coenia oviposition behavior on fertilized P. lanceolata (high nitrogen, low defensive chemistry) to unfertilized P. lanceolata (low nitrogen, high defensive chemistry) in the greenhouse. To investigate the role of host-plant chemical defense on female oviposition choice, we compared female oviposition preference between unfertilized P. lanceolata (high defensive chemistry) to unfertilized P. major (low defensive chemistry) in the greenhouse. If predation is a strong selective agent against larval survival, then we expected females to preferentially oviposit on more chemically defended host-plants (P. lanceolata, unfertilized > P. lanceolata, fertilized = P. major, unfertilized). If nitrogen is a limiting resource for larval development, we expected females to preferentially oviposit on higher nutrient host-plants regardless of host-plant defensive chemistry (P. lanceolata, fertilized > P. lanceolata, unfertilized = P. major, unfertilized). By combining all experimental results, we were able to evaluate the relative importance of host-plant nitrogen uptake and chemical defense on female oviposition choice, larval performance, and larval chemical defense in a specialist phytophagous insect.
Methods Study system Plantago lanceolata and P. major are cosmopolitan, herbaceous weeds, which are annual or facultative perennials (Cavers et al. 1980; Kuiper and Bos 1992).
580
They were introduced in North America approximately 200 years ago, thriving in disturbed habitats such as gardens and agricultural areas (Thomas et al. 1987). Both species contain iridoid glycosides, a group of cyclopentanoid monoterpene-derived compounds found in about 50 plant families (El-Naggar and Beal 1980; Boros and Stermitz 1990). P. lanceolata primarily contains two iridoid glycosides, aucubin and catalpol (Duff et al. 1965; Bowers and Stamp 1992, 1993); although trace amounts of other iridoid glycosides (which are not sequestered by J. coenia larvae) have been identified (Willinger and Dobler 2001; Taskova et al. 2002). Aucubin is the biosynthetic precursor to catalpol (Damtoft et al. 1983), and it is less toxic to herbivores than catalpol (Bowers 1991). P. major contains aucubin but not catalpol, and has lower amounts of total iridoid glycosides than P. lanceolata (Massa and M.D. Bowers, unpublished data; this study). The insect herbivore used in our experiments, J. coenia Hu¨bner (Lepidoptera: Nymphalidae), consumes only plants containing iridoid glycosides (Bowers 1984), and P. lanceolata and P. major are commonly used hostplants (Shapiro 1974; Scott 1986). Iridoid glycosides serve as larval feeding and female oviposition stimulants for J. coenia (Bowers 1984; Pereyra and Bowers 1988, respectively). J. coenia larvae sequester iridoid glycosides from their host-plants (Bowers and Collinge 1992), and these compounds deter predation by invertebrate (Dyer and Bowers 1996) and vertebrate (Bowers and Farley 1990) predators. Plant treatments Our nutrient enrichment regime was based on three factors. First, since both the butterfly and the host-plant are commonly found in agriculture and other nutrient enriched areas, our fertilization treatments should fall within range applied by farmers. In Colorado, the average farmer applies 140 kg/ha/year with a range of 0– 340 kg/ha/year (USDA 2002); we applied the equivalent of 89 kg/ha/year. We also wanted our nutrient treatments to be ecologically relevant. Based on previous nutrient experiments, this regime produced P. lanceolata foliar nitrogen levels within the documented range of forbs under natural conditions (Joern and Behmer 1998). Finally, our treatment had to be potent enough to result in significant changes in both plant foliar nitrogen and iridoid glycosides. Plantago lanceolata and P. major seeds were collected from two sites on the University of Colorado, Boulder campus. Three-week-old (first true leaves) plants were transplanted into 4-l pots containing Fafard Nursery mix#2 (Amherst, MA, USA). Half of the P. lanceolata seedlings were transplanted into normal potting soil, and the other half were transplanted into normal potting soil with 300 mg of Osmocote 14N:14P:14K fertilizer: 150 mg mixed in the soil and 150 mg applied to the soil surface to more closely simulate incidental fertilization
caused by agricultural runoff. All P. major seedlings were transplanted into normal potting soil without additional fertilizer. Plants were grown in the greenhouse at an average daytime temperature of 24C with 60% RH and 10light:14dark. Plants were watered every third day. Herbivory treatments Three-week-old (first true leaves) P. lanceolata plants were transplanted into 4-l pots with Fafard Nursery mix#2 (Amherst, MA, USA) with the fertilization treatments described in ‘Plant treatments’. Plants were then randomly assigned to one of the four treatments: 1. Unfertilized, no herbivory: each plant (n=10) was watered to saturation every third day. Neither fertilizer enrichment nor herbivory occurred in this plant treatment. 2. Fertilized, no herbivory: each plant (n=10) was watered to saturation every third day. With regular watering, fertilizer nutrients were released at a constant rate. 3. Unfertilized, herbivory: each plant (n=25) was watered to saturation every third day. A single second instar J. coenia caterpillar was placed on an 8-weekold plant. 4. Fertilized, herbivory: each plant (n=25) was watered to saturation every third day. Fertilization regime was the same as described for the ‘‘Fertilized, no herbivory’’ treatment (no. 2 above). A single second instar J. coenia caterpillar was placed on an 8-weekold plant. Throughout the experiment, the plants were kept in a greenhouse randomly arranged on a single 2·4 m raised bench. The average daytime temperature was 24C with 60% RH and 10light:14dark. The experiments began at week 8 (average leaf height of 20 cm) and lasted until the caterpillars molted into the fifth instar (n=70). After the larvae reached fifth instar, the plant above ground biomass was harvested and prepared for chemical analyses. Plant performance and chemical analyses At the end of the experiments, biomass and iridoid glycosides were measured in plants from all experiments (n=123). Plants were oven dried at 50C, and total above ground dry biomass was measured as a plant performance proxy. Dried plants were weighed, ground with a Wiley Mill (#40 mesh), and stored in a glass jar at 40C until further analysis. Iridoid glycosides were extracted from each sample and analyzed by gas chromatography (Gardner and Stermitz 1988; Bowers 1991). Total foliar nitrogen for a subset of P. major and P. lanceolata (n=68) was quantified with a copper sulfate
581
catalyst procedure using a Lachat colorimetric autoanalyzer (Bowman et al. 1993). Insect performance and iridoid glycoside sequestration Survival, larval mass gain, rate of larval mass gain, larval final mass and number of days to the fifth instar were measured for all larvae (n=50) from the herbivory experiment. A second instar caterpillar, taken from a laboratory colony raised on P. lanceolata, was weighed and then placed on a single plant described previously in the plant herbivory treatment. All plants were covered with netting secured at the bottom with a large twist tie. Every second day, each larva was removed from the plant, weighed and returned to the plant. Larvae were collected after their fifth instar molt because their chemical defense steadily decreases from this ontological point through pupation (M.D. Bowers, unpublished data). Because of this limitation, the larvae were not allowed to pupate, and we could not use the measure of relative growth rate used in previous Lepidoptera performance experiments (Wiklund et al. 1991; Fischer and Fiedler 2000). To quantify larval iridoid glycosides, newly molted, fifth instar caterpillars were collected and starved for 12 h to ensure empty guts. Aucubin and catalpol were measured using methods similar to those used for plants, except that the caterpillars were lyophilized instead of oven dried (Gardner and Stermitz 1988; Bowers and Collinge 1992). To determine whether caterpillar iridoid glycoside content depended on host-plant content, we tested for a correlation between host-plant and caterpillar iridoid glycosides. Female oviposition preference Females were simultaneously offered two 10-week-old plants, one each from two of the following treatments: unfertilized P. lanceolata (high iridoid glycoside content, n=19), fertilized P. lanceolata (low iridoid glycoside content, n=20), and unfertilized P. major (low iridoid glycoside content, n=21). There were three experiments in all: fertilized P. lanceolata versus unfertilized P. lanceolata (n=9); fertilized P. lanceolata versus unfertilized P. major (n=11); and unfertilized P. lanceolata versus unfertilized P. major (n=10). Female and male J. coenia were taken from a laboratory colony, separated by sex, and placed into cylindrical net bags (40 cm H·30 cm D). The butterflies were provided with honey water as a food source, but no suitable oviposition substrate. After 48 h of isolation from the opposite sex, males and females were placed together in net bags, in a 2 male:1female ratio (three or six individuals total). Following 72 h of potential mating time, females were removed from the net bag and used in the oviposition experiment. For each choice test, individual plants from two of the three plant treatments
were selected randomly and each plant was used only once. Plants were placed 20 cm apart in a clear plastic container (80 cm W·35 cm D·35 cm H) in the greenhouse. A honey water source was placed in the middle of the container, equidistant from both plants. The container was covered with fine mesh and secured with twist-ties. A single female butterfly was then placed upon the honey water source in the container, and allowed to oviposit for 48 h. The container was checked after 24 h to refill the food supply and to water the plants if necessary. After 48 h of oviposition, the plants were removed from the container and the eggs were counted as individual leaves were removed from the plants. These leaves were then prepared for subsequent chemical analysis. Statistical analyses The percentage based data (aucubin, catalpol and total iridoid glycoside concentrations) were arcsine transformed to normalize the data prior to analysis. In order to assess the effect of plant treatment on oviposition, the proportion of eggs laid on each plant was transformed, and these transformed data were then used to calculate a difference between eggs laid on each treatment: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi r1 þ ð3=8Þ r2 þ ð3=8Þ Difference in eggs ¼ n þ ð3=4Þ n þ ð3=4Þ where r1 and r2 are the number of eggs laid on each of the treatments, respectively, and n is the total number of eggs laid (Judd and McClelland 1989). t-tests were performed to determine if females preferred one plant treatment over another. The effects of fertilization and herbivory on plant performance and plant chemical defense were analyzed using a two-way ANOVA. The effects of soil fertilizer on plant nitrogen concentration, insect performance, and insect defensive chemistry were analyzed using one-way ANOVAs. Larval survival was compared using a chi-square test. Relationships between plant defensive chemistry and insect defensive chemistry (percent total iridoid glycosides, percent aucubin, and percent catalpol) were tested for using regression analyses. All statistical analyses were conducted with SAS version 8.1 (SAS 2000).
Results How does soil nutrient enrichment influence plant performance and chemistry? Soil fertilization significantly increased the foliar nitrogen concentration (measured as percent dry mass) in P. lanceolata (Table 1; Fig. 1), as well as plant above ground biomass (Table 1). However, from the larval performance experiments, herbivory also significantly
582
Percent dry mass
3.5
Catalpol
2.5 2.0
Nitrogen
1.5 1.0 0.5
P. lanceolata
Unfertilized
Fertilized
P. major Unfertilized
Plant species and treatment Fig. 1 The effect of soil fertilization on chemical defenses (iridoid glycosides aucubin and catalpol) and foliar nitrogen of Plantago lanceolata and P. major measured in percent dry mass. Mean ± SE presented
iridoid glycosides
6.0
How do host-plant chemistry changes influence larval performance and defensive chemistry? There was no significant effect of fertilization on any measure of larval performance. Neither total larval mass gain, rate of larval mass gain, larval final mass, nor days to fifth instar (n=40) was affected by fertilization treatment (Table 1). Larval survival (n=50) did not differ between fertilized and unfertilized plants (fertilized=84%, unfertilized=76%, v2 =0.503, P=0.4783). Iridoid glycoside concentrations were significantly lower in larvae reared on fertilized P. lanceolata than those reared on unfertilized P. lanceolata (Table 1;
Aucubin
Iridoid glycosides
3.0
P. lanceolata
Larval percent
affected foliar dry mass (Table 1). On average, foliar dry mass in the herbivory treatment was 1.3 g greater than foliar dry mass of the plants in the non-herbivory treatment. The interaction between herbivory and fertilization was not significant (Table 1). For P. lanceolata, total iridoid glycoside concentration was negatively affected by fertilization (Table 1). On average, fertilized plants contained about half the total iridoid glycosides of unfertilized plants (Fig. 1). This effect was attributable to differences in aucubin, not catalpol (Table 1). Aucubin concentrations (percent dry mass) of non-fertilized plants were, on average, 1.7 times higher than fertilized plants (Fig. 1). P. major contained no catalpol, but did have significantly less aucubin than plants in either of the P. lanceolata treatments (Fig. 1; ANOVA n=123, F1,121=35.52, P N(4.8 g) H(2.1%) = N(1.9%) H(4.08%) = N(4.49%) H(0.28%) = N(0.25%)
All are ANOVA results with significant effects labeled in bold. Direction of effects and mean of treatment shown F fertilized Plantago lanceolata, U unfertilized P. lanceolata, H P. lanceolata with herbivory, N P. lanceolata without herbivory
583 Table 2 Correlation results between host-plant defensive chemistry and larval defensive chemistry Number of correlations
df
F
P
R2
Total foliar iridoid glycosides versus total larval iridoid glycosides Foliar aucubin versus larval aucubin Foliar catalpol versus larval catalpol
40 40 40
1, 38 1, 38 1, 38
26.03 23.40 25.13
0.001 0.001 0.001
0.441 0.414 0.432
Proportion of total eggs laid
Correlation
1.0 0.8 0.6 0.4 0.2
A
P. lanceolata P. major Unfertilized Unfertilized
B
P. lanceolata P. major Fertilized Unfertilized
C
P. lanceolata P. lanceolata Unfertilized Fertilized
Host-plant and treatment Fig. 3 The effect of host-plant species and soil fertilization on female buckeye (J. coenia) oviposition preference. Original data was arcsin transformed for statistical analyses; proportion data is shown for illustrative purposes only. Mean ± SE presented
Fig. 2). On average, the larvae raised on unfertilized plants contained 4.1 times more total dry mass iridoid glycosides as compared to larvae raised on fertilized plants (Fig. 2). This effect is attributable to differences in aucubin, not catalpol (Table 1). Larvae reared on unfertilized plants, on average, sequestered 3.9 times more aucubin than larvae reared on fertilized plants (Fig. 2). Host-plant iridoid glycoside concentration showed significant positive correlation with insect iridoid glycoside concentration (Table 2). This was the case for aucubin, catalpol and total iridoid glycosides (Table 2). How do host-plant chemistry changes influence female oviposition choice? These results are presented as a series of three pair-wise experiments. In all three tests, females exhibited a clear preference for plants in one treatment over those in another plant treatment (Fig. 3). In the unfertilized P. lanceolata versus unfertilized P. major trials, females preferred unfertilized P. lanceolata (t test, n=10, t1,8=30.25, P=0.0004), and only rarely laid eggs on unfertilized P. major. In the fertilized P. lanceolata versus unfertilized P. major trials, females preferred fertilized P. lanceolata (t test, n=11, t1,9=89.11, P