Willow Hybridization Differentially Affects Preference and Performance Of

Entomologia Experimentalis et Applicata 83: 285±294, 1997. 285

c 1997 Kluwer Academic Publishers. Printed in Belgium.

Willow hybridization differentially affects preference and performance of

herbivorous beetles

; Colin M. Orians1; , Cynthia H. Huang1, Alexander Wild1, Katherine A. Dorfman1 , Pamela Zee2,MinhTamT.Dao2 & Robert S. Fritz2 1Department of Biology, Vassar College, Poughkeepsie, NY 12603, USA; 2Department of Biology, Williams

College, Williamstown, MA 01267, USA; Present address: Department of Biology, Tufts University, Medford,

MA 02155, USA; Mt Holyoke College, Department of Geology and Geography, South Hadley, MA 01075, USA

Accepted: February 6, 1997

Key words: Salix, hybridization, phenolic glycosides, herbivore preference, herbivore performance

Abstract

We examined the preferences and performances of ®ve beetle species (four chrysomelids and one scarab) on two species of willows (Salix sericea and S. eriocephala) and their interspeci®c hybrids. Beetle species differed markedly in their responses. In preference assays, two chrysomelid beetle species (Calligrapha multipunctata bigsbyana and Plagiodera versicolora) preferred hybrids, two chrysomelids (Chrysomela scripta and Ch. Knabi) preferred hybrids and S. sericea, and the scarab beetle (Popillia japonica) preferred S. eriocephala. Experiments with puri®ed salicortin indicated that salicortin concentration may contribute to these preferences. The relative performance (growth rate, pupal/adult weight and survivorship) of these beetles on the three willow taxa did not correspond with their feeding preferences. Three species exhibited intermediate performance on hybrid willows (the two Chrysomela spp. and P.japonica); the Chrysomela spp. performed best on S. sericea, while P.japonica performed best on S. eriocephala. One species performed equally well on all three taxa (C. multipunctata bigsbyana). The performance of Pl. versicolora was not tested. Our results support the general pattern that willow taxa with phenolic glycosides are more acceptable to specialist willow herbivores while those taxa without phenolic glycosides are more acceptable to generalist herbivores. We also show that to predict the relative susceptibility of hybrid and parental plants to herbivores, consideration must be given to the inheritance of traits affecting both preference and performance.

Introduction There is no reason to expect one single pattern to dominate. For one to dominate, we would have to Hybridization among plants is more common than pre- assume that traits that affect herbivore responses, pref- viously recognized (Arnold, 1992). This realization erence or performance, such as secondary chemistry has led to numerous studies focusing on the ecolo- (Tahvanainen et al., 1985; Lindroth et al., 1988; Roin- gical and evolutionary consequences of hybridization, inen & Tahvanainen, 1989; Schultz, 1989), foliar nutri- especially in the ®eld of plant-herbivore interactions ents (Mattson 1980; Bryant et al., 1987), leaf water (see Strauss, 1994, and references therein). Hybrids content (Horton, 1989), shoot morphology(Price et al., and parental species typically differ in resistance to 1987), plant phenology (Aide, 1988), and plant onto- herbivores. Contrary to Whitham's (1989) original geny (Bryant, 1981; Kearsley & Whitham, 1989) show hypothesis that hybrids are ecological and evolution- similar patterns of inheritance. We would also have ary sinks to pests (because they are more susceptible), to assume that herbivores respond similarly. Because patterns of hybrid resistance are diverse (Boecklen & the inheritance patterns of traits in hybrids are vari- Spellenberg, 1990; Fritz et al., 1994; Strauss, 1994). able (Rieseberg & Ellstrand, 1993) and herbivore species are known to differ in their responses to intra-

GR: 201003416, Pips nr. 135632 BIO2KAP

*135632* ento1530.tex; 17/06/1997; 11:56; v.7; p.1 286 and interspeci®c variation in plant quality (Matsuki & Materials and methods MacLean, 1994; Orians & Fritz, 1996; Orians & Floyd, 1997), no single pattern of hybrid susceptibility should The system. This willow system has three distinct be expected. advantages. First, Salix sericea and S. eriocephala Among willow species, much of the variation in are largely sympatric and hybridize naturally (Argus, susceptibility has been attributed to the presence or 1974, 1986). Second, there is a diverse herbivore com- absence of phenolic glycosides (Rowell-Rahier, 1984; munity that feeds on the three willow taxa. Finally, to Tahvanainen et al., 1985; Denno et al., 1990; Kelly & control for confounding environmental effects, these Curry, 1991; Rank, 1992; Kolehmainen et al., 1994). willows can easily be cloned and grown in a com- Phenolic glycosides stimulate feeding by some herb- mon garden (Paige & Capman, 1993; HanhimakiÈ et ivores but deter feeding by others (Matsuda & Mat- al., 1994; Christensen et al., 1995). Clearly this is an suo, 1985; Tahvanainen et al., 1985; Rank, 1992; excellent system in which to investigate the effects of Kolehmainen et al., 1994). Phenolic glycosides are hybridization on the responses of various herbivores. also known to affect the growth of herbivores (Denno et al., 1990; Kelly & Curry, 1991; Matsuki & MacLean, Plants. In 1992 we used leaves from ®eld plants at 1994). In our study system, Salix sericea Marshall and our study site near Milford, NY. This is an area that is S. eriocephala Michx. and their hybrids differ dramat- dominated by S. sericea but also contains S. erioceph- ically in their secondary chemistry (Orians & Fritz, ala and their hybrids. In 1993 and 1994 we used leaves 1995). S. eriocephala foliage contains high concentra- from cuttings grown in a common garden at Williams tions of condensed tannins and no phenolic glycosides, College in Williamstown, MA. In 1992 and 1993 we while S. sericea foliage contains high concentrations of used a single genotype of each taxon. In 1994 we used phenolic glycosides, especially salicortin, but low con- four genotypes of each willow taxon (with 5 replic- centrations of condensed tannins. The concentration ate clones per genotype). To ensure that taxon effects of salicortin in S. sericea often exceeds 10% dry leaf were not confounded by plant gender effects (willows weight (Nichols-Orians et al., 1992, 1993). Hybrids are dioecious), three of the four genotypes of each contain intermediate concentrations of both classes of taxon were female. However, beetle responses to the chemicals (Orians & Fritz, 1995). Given the differ- male genotype were not distinct from their responses ences in phenolic glycoside chemistry in this willow to the female genotypes (data not shown). This result system, we hypothesized that the responses of herbi- is consistent with previous results (R. Fritz, unpubl. vores to parental and hybrid willows would be variable data). The identity of the parental and hybrid willows and that chemistry would be an important determinant was con®rmed using RAPD analysis, and the hybrids of those responses. selected for this study contain intermediate concentra- We have been quantifying the abundance of numer- tions of phenolic glycosides and appear to be F-1 and ous insect herbivores on three willow taxa, S. sericea, F-2 type hybrids (Fritz et al., 1994; Orians & Fritz, S. eriocephala and their hybrids. From our censuses 1995; S. Brunsfeld, pers. comm.) of ®eld plants we have found a diversity of herbi- In the spring of 1993, we collected lea¯ess bud- vore responses (Fritz et al., 1994). We have not until sticks (cuttings) from plants at our study site in NY. now explored potential mechanisms behind the differ- 20 cm cuttings were dipped in rootone to stimulate ent responses. Here we report on the feeding prefer- root growth, and planted into Agway's Potting Soil ences and performance of ®ve beetle species: Calli- Mix. Cuttings were watered daily and fertilized once grapha multipunctata bigsbyana, Plagiodera versicol- with 20-20-20 Agway fertilizer. In May, they were ora, Chrysomela scripta and Chrysomela knabi,and potted into 8 liter pots containing loam, peatmoss and Popillia japonica to S. sericea, S. eriocephala,and vermiculite (3:1:1) and fertilized weekly with 140 ml of their interspeci®c hybrids. By painting salicortin onto 20-20-20 soluble fertilizer (3.2 g/l).The plants were cut leaves, we further determined whether salicortin may back in the late fall of 1993 and allowed to regenerate contribute to the variable responses. in the spring of 1994. In the spring of 1994 they were transplanted into 16 liter pots, fertilized with 15 ml of 10-10-10 insoluble fertilizer (Agway) and watered as needed. To avoid the use of pesticides, we removed herbivores by hand.

ento1530.tex; 17/06/1997; 11:56; v.7; p.2 287

Beetles. The responses of four chrysomelids, a wil- Beetles were collected from S. sericea at our New York low leaf beetle (Calligrapha multipunctata bigsby- ®eld site in July. The beetles were brought to Hartwick ana (Kirby) ± henceforth referred to as C. multi- College, in Oneonta NY and maintained on S. sericea punctata), the imported willow leaf beetle (Plagiodera leaves until the choice tests were performed. Choice versicolora (Laicharting)), and two Chrysomela wil- tests were terminated after 4±6 h and the remaining low leaf beetles (Chrysomela knabi Brown & Chryso- leaf area of each disc was measured. (Because using mela scripta Fabricius) were tested. In addition, one beetles reared on S. sericea may bias choice tests, 1993 scarab, the imported Japanese beetle (Popilliajaponica and 1994 assays were done on beetles maintained on S. Newman) was tested. bebbiana, another shrubby willow, prior to the choice tests.)

Preference assays. Beetle preference was determined In 1993, we compared the preferences of adult C.

= n = using multiple choice tests. Beetles were introduced multipunctata (n 20) and adult P.japonica ( 20) into a petri dish lined with moistened ®lter paper that for a second randomly selected genotype of each taxon. contained three leaf discs, one disc from each taxon. The plants were grown in a common garden. C. mul- Discs were of similar size (2 cm2 in 1992, 1.5 cm2 tipunctata were collected in Berkshire County, MA in 1993/1994). After a speci®ed duration, the relat- from Salix bebbiana and maintained on S. bebbiana ive amounts of each taxon consumed were compared. until assays were performed. P. japonica were collec- Leaves used for the assays were clipped at the peti- ted from Parthenocissus tricuspidata (Boston Ivy) on ole and transported back to the lab in a cooler. All the Williams College campus and tested immediately. assays were started within 30 min of leaf collection. After 6 (P. japonica)or24(C. multipunctata)hwe No individual beetle was tested more than once and determined the area consumed. This difference in dur- only the ®rst fully expanded leaves were used so as to ation was designed to ensure signi®cant consumption ensure that leaves of similar developmental ages were but ensure no single leaf disc would be completely compared. consumed.

The analysis of feeding preference experiments can In 1994, the multiple choice tests were done with =

be problematic due to lack of independence (Roa, 3rd instar C. multipunctata larvae (n 20), adult

= n =

1992). Therefore we determined which taxon was con- Pl. versicolora ( n 12), adult Ch. scripta ( 30) = sumed the most in each trial, and employed a G-test and adult Ch. knabi (n 31). Beetles were collected to determine if one willow taxon was preferred over from S. sericea in Bennington County, Vermont. As in others (Sokal & Rohlf, 1981). Our null hypothesis was 1993, we maintained all beetles on S. bebbiana prior to equal consumption of all three taxa. This G-test is a the experiments. Each choice test lasted roughly 16 h. statistically conservative test because we were not com- Because of their small size, four Pl. versicolora adults paring the absolute amount of each taxon consumed. were used in each of the 12 trials. All other species As another measure of preference, we censused were tested individually. During each choice test, we Japanese beetle damage to ®eld plants in 1992. A randomized which genotypes were compared. census of damage was not completed for other beetles because, unlike Japanese beetle damage, their damage Effects of salicortin on beetle preferences. We is rare and more dif®cult to distinguish. We censused determined the preferences of C. multipunctata, Japanese beetle damage on each of 20 S. erioceph- Ch. knabi and P. japonica for S. eriocephala leaf discs ala, S. sericea and hybrid genotypes. For 25 shoots painted with either a solution containing salicortin or a per plant we counted the number of shoots with more solvent-only control. (Recall that S. eriocephala does than 50% of their leaves damaged by Japanese beetles. not produce phenolic glycosides.) Salicortin-painted We employed a chi-squared test to determine if the and control leaf discs (taken from the same leaf) were amount of damage was equally distributed across the placed into a petri dish containing moistened ®lter three willow taxa (Sokal & Rohlf, 1981). paper. A single beetle was then introduced into the dish and after a speci®ed duration, the area consumed Speci®cs of multiple choice assays. In 1992 we com- from each disc was determined. To determine the con- pared the preferences of C. multipunctata 3rd instar lar- centration to apply, we ®rst measured the average dry

= n =

vae ( n 40) and adults ( 63) and of Ch. knabi 3rd weight of S. eriocephala leaf discs (2 cm ). We then

= n = instar larvae (n 47) and adults ( 30). We used calculated the concentration of salicortin required to leaves from a single genotype of each willow taxon. obtain 1%, 4%, 7%, and 10% salicortin by dry leaf

ento1530.tex; 17/06/1997; 11:56; v.7; p.3 288 weight. Hybrid and S. sericea foliage typically con- In 1993, we measured the performance of newly

tain 1±4% and 7±10% salicortin respectively (Orians hatched C. multipunctata larvae on 4 different gen- = & Fritz, 1995). Because acetone causes discoloration otypes of each willow taxon (n 6±8 lar- around the edges of leaf discs, we cut smaller leaf discs vae/genotype/taxon). Larvae were randomly assigned, (1.5 cm2) from the larger (2 cm2) leaf discs immedi- given fresh leaves every 2 days and kept in a growth ately after the acetone had evaporated. chamber (23 C and L16:D8). Pupae were weighed fol-

In 1993, we determined the effect of 10% salicortin lowing pupation. Due to heavy pupal mortality we were =

on the preferences of adult C. multipunctata (n 15) unable to measure adult mass. Within each taxon, we =

and adult P. japonica (n 20). The area consumed found no genotype effect on pupal weight (S. sericea:

F = : = : F = : ;

after six (P. japonica)or24(C. multipunctata)hwas 3;15 0 65, P 0 60; S. eriocephala: 3 21 1 44,

= : F = : = : determined. In 1994, we determined the effects of 1%, P 0 26; hybrids: 3;23 0 96, P 0 43), so the 4%, 7% and 10% salicortin on the host selection of effects of willow taxon on pupal weight were determ-

adult P. japonica, adult Ch. knabi, and adult C. mul- ined with a one-way analysis of variance. = tipunctata (n 20 for all species). Assays lasted 6, In 1994, performance assays were done with 8 and 24 h for P. japonica, Ch. knabi,andC. mul- Ch. scripta, C. multipunctata and P. japonica.Two tipunctata, respectively. As before, the differences in different assays were completed. First, we randomly

duration ensured no leaf disc would be completely con- assigned newly hatched larvae of Ch. scripta to a = sumed. Beetles that failed to eat were excluded from single genotype of each willow taxon (n 46/taxon). the statistical analyses, and each beetle species was We weighed the beetles when they reached the adult analyzed separately. stage. (Rearing conditions were identical as in 1993). For the 1993 data, we used binomial probability Ch. scripta are sexually dimorphic so the adults were analysis to determine if the beetles showed a signi®c- sexed to control for the effects of gender. A 2-way ana- ant preference (Dixon & Massey 1983). For the 1994 lysis of variance (with willow taxon as a ®xed effect data, we used a regression analysis to determine if and beetle gender as a random effect) was used to test the proportion of salicortin-treated leaf area consumed for differences in adult weight. depended upon the concentration of salicortin. Second, using all three beetle species, we compared the weight gain or loss of adult beetles over an eight

Performance assay. Performance was measured as day period when fed leaves of each willow taxon (1 =

either larval growth rate, pupal/adult mass or adult genotype per taxon). C. multipunctata (n 45) and =

weight gain (only Pl. versicolora was not tested). P. japonica (n 45) had been fed S. bebbiana prior = Although unknown for these species, pupal and adult to the experiments, while Ch. scripta adults ( n 58) weight are often correlated with adult fecundity. All had been reared as larvae on the same willow tax-

beetles were reared individually. In 1992 we measured on on which they were tested. Measuring changes in =

the growth of Ch. knabi (n 31) and C. multipunctata adult mass is unusual in performance assays. Because = (n 27) larvae. Adult beetles were collected from P. japonica only feeds on willows as adults, this assay S. sericea at our New York ®eld site and maintained allowed us to compare the ability of these three beetle on S. sericea leaves until eggs were laid. The resulting species to maintain body mass when feeding on these larvae were weighed and randomly assigned to each different willow taxa. Furthermore, signi®cant weight plant taxon and reared on whole leaves (at room tem- loss or gain could affect fecundity. perature in a laboratory at Hartwick College) for one (Ch. knabi)ortwo(C. multipunctata) weeks. These time differences re¯ect species speci®c differences in Results development time. Leaves were changed daily and larvae were weighed again at the end of the experiment. Preference. Despite differences in the methods used (We used the same plant genotypes as in the 1992 pref- each year, the responses within each beetle species erence tests.) Data on larvae that died were excluded. were consistent. Larvae and adult C. multipunctata

We calculated larval weight gain per day and tested showed non-random host selection (1992 larvae and

: < :

for the effects of willow taxon with a one-way analys- adults, G = 21 23 and 8.58 respectively, P 0 05;

: < : =

is of variance using StatView 4.02 (Abacus Concepts, 1993 adults, G = 19 43, P 0 001; 1994 larvae, G

= : 1992). 5:93, P 0 05; Figure 1A). They preferred hybrid leaves more than those from either of the two parental

ento1530.tex; 17/06/1997; 11:56; v.7; p.4 289

Figure 1. The proportion of multiple choice tests in which each willow taxon was the most consumed. (A) Calligrapha multipunctata bigsbyana

n = n n =

( n 20), (B) Plagiodera versicolora ( 12), (C) Chrysomela knabi (CK; 31) and C. scripta (CS; 20), (D) Popillia japonica = ( n 20). See Methods for further details of sample sizes. species, and did not prefer leaves from one parent over 1992. However, Ch. knabi larvae were also reared on

another (Figure 1A). Pl. versicolora showed the same S. sericea in 1992 but did not prefer S. sericea.

: < : pattern (G = 7 44, P 0 05) (Figure 1B). P. japonica preferred S. eriocephala over both the

= : <

Both Chrysomela species also exhibited prefer- other two taxa (1992, 19 6, P 0.001; 1993,

: = P :

ences (1992 Ch. knabi larvae and adults G = 19 3 G 16.0, 0 001) (Figure 1D). Although, the lack :

and 17.5 respectively, P < 0 001; 1994 Ch. knabi of P. japonica acceptance of S. sericea precludes stat- :

and Ch. scripta adults G = 6 21 and 15.5 respect- istical analysis, hybrids appeared to be slightly more : ively, P < 0 05) (Figure 1C). Both species avoided acceptable (Figure 1D). S. eriocephala while accepting S. sericea and hybrids. In 1992 but not in 1994, Ch. knabi adults appeared Effects of salicortin on beetle preferences. C. multi- to prefer S. sericea over hybrids. This may have been punctata, Ch. knabi and P. japonica differed in their due to the fact that they were reared on S. sericea in response to salicortin. In 1993, 13 of 14 C. multipunctata adults consumed more control than 10%

ento1530.tex; 17/06/1997; 11:56; v.7; p.5 290

reared larvae from egg hatch to the adult stage. All but one Ch. scripta larva died when feeding on S. eriocephala,andCh. scripta survivorship was intermediate on hybrids (0.02, 0.41 and 0.83 on S. eriocephala, hybrids and S. sericea, respectively). When comparing the whole life weight gain of Ch. scripta on hybrids

vs. S. sericea, we found a signi®cant effect of beetle

F = : < :

sex ( 1;71 44 98; P 0 001), but no effect of wil-

F = : : < < :

low taxon ( 1;1 108 72; 0 05 P 0 1), and no

F = : = :

interaction effect ( 1;71 0 11; P 0 74). When adult C. multipunctata, P. japonica and Ch. scripta were fed the three taxa in 1994 they differed in their ability to maintain weight (Table 1).

C. multipunctata gained 1±2 mg on all three taxa and

F = :

weight gain did not differ by taxon ( 2;39 0 36;

Figure 2. The effects of salicortin concentration on the area of leaf : consumed by Calligrapha multipunctata bigsbyana. Values above P = 0 70). Japanese beetles lost weight when feeding

0 indicate that more was consumed from salicortin-amended discs on all three taxa but weight loss was taxon specif-

= : < : than from control leaf discs. Values less than 0 indicate that control F ic ( 2;37 41 05, P 0 001). They lost most on discs were favored over salicortin-amended leaf discs. Five beetles S. sericea (21 mg) and hybrid (15 mg), and least on were tested at each concentration.

S. eriocephala (3.6 mg). For Ch. scripta there was no

F = :

effect of taxon (hybrids vs. S. sericea)( 1;1 129;

P < : = : F = : = :

salicortin-painted S. eriocephala leaves ( 0 001). P 0 50), beetle sex ( 1;54 3 69; P 0 06) or their

F = : = :

In 1994 we found that as the concentration of sali- interaction ( 1;54 1 45; P 0 23) on adult weight cortin dropped, C. multipunctata beetles were much gain.

= :

more likely to consume the painted leaves ( R 0 30,

: n = P = 0 01, 20) (Figure 2). Although insigni®cant, there is a trend suggesting that low concentrations may Discussion stimulate feeding.

Ch. knabi feeding was stimulated by the presence Preference. The beetle species included in this study

< :

of salicortin at all four concentrations ( P 0 05, displayed different preferences for the three willow = n 7). (Extensive mortality precluded higher sample taxa. C. multipunctata and Pl. versicolora both pre- sizes.) In 6 of the 7 trials more of the painted discs ferred hybrid willows. Pl. versicolora are also more were consumed, and there was no indication that high abundant on hybrid plants when both parental and concentrations stimulated feeding more than did low hybrid clones are grown in a common garden (Orians, concentrations. unpublished data). The Chrysomela species avoided

In 1993 and 1994,P.japonica were deterred by sali- S. eriocephala but found both S. sericea and hybrids : cortin (P < 0 05). In both years, 10 of 13 preferred the acceptable. In contrast, P. japonica found S. erioceph- control discs. In 1994, the concentration of salicortin ala acceptable but avoided both S. sericea and hybrids,

= : = :

did not affect acceptability (R 0 08, P 0 34, a response consistent with previous results (Orians & = n 13). Floyd, 1997). These same patterns were evident when we compared the absolute amount of each taxon con- Performance. The beetle species also differed in their sumed by each beetle species (data not shown). performance on the three taxa (Table 1). In 1992 the Although diet-induced feeding preferences can growth rate of C. multipunctata larvae did not differ occur (Richardson & Whittaker, 1982), C. multi- among the willow taxa over the course of 2 weeks. Nor punctata shows similar preferences whether they were was there a taxon effect on pupal weight in 1993. fed leaves of a phenolic glycoside (1992) or a con- The two Chrysomela species performed differently densed tannin (1993 and 1994) containing host prior to on the three taxa (Table 1). In 1992, the growth rate the tests. Although Ch. knabi adults appeared to prefer per day of Ch. knabi larvae was slowest on S. erio- S. sericea when fed S. sericea prior to the choice tests cephala and intermediate on hybrids. A similar pat- (Figure 1C), other experiments with both the Chryso- tern was observed for Ch. scripta in 1994 when we mela species indicate that diet-induced preferences are

ento1530.tex; 17/06/1997; 11:56; v.7; p.6 291

Table 1. The effects of willow taxon on beetle performance. 1992 data show the mean (s.e.) larval growth rate (mg/day) (Calligrapha multipunctata bigsbyana and Chrysomela knabi) on the three willow taxa. Full-life assays illustrate the effects of plant taxon on pupal (C. multipunctata) or adult (Chrysomela scripta) weight (mg) when larvae were reared on one of the three taxa. Adult weight gain (C. multipunctata, Ch. scripta,andPopillia japonica) shows the effects of plant taxon on adult weight change over 8 days. Signi®cance of differences were determined with an analysis of variance with willow taxon as a ®xed effect and beetles sex (Ch. scripta only) as a random effect. Different letters indicate that

treatments differ at P <0.05

Salix eriocephala Willow Taxon Salix sericea Eriocephala Hybrid Sericea

1992 ± growth rate

C. multipunctata 2.74 (0.37) a 2.93 (0.33) a 2.67 (0.36) a

= n = n = n 9 9 9

Ch. knabi 0.50 (0.15) a 1.68 (0.17) b 3.10 (0.17) c

= n = n = n 6 11 10

1993 ± Full-life assay

C. multipunctata 43.0 (1.17) a 44.0 (1.43) a 41.00 (1.17) a

= n = n = n 20 27 25

1994 ± Full-life assay Ch. scripta

females ± 18.41 (1.05) a 21.19 (0.93) a

= n = n 9 21

males ± 13.69 (0.47) b 15.99 (0.38) b

= n = n 25 21

1994 ± Adult weight gain

C. multipunctata 1.85 (0.01) a 1.18 (0.01) a 1.86 (0.01) a

= n = n = n 15 14 13

Ch. scripta :

females ± 1 19 (0.49) a 0.46 (0.63) a

= n = n 6 20

males ± 0.81 (0.55) a 0.92 (0.39) a

= n =

n 13 19

: : :

P. japonica 3 63 (1.14) a 14 98 (0.87) b 20 83 (1.95) c

= n = n = n 13 13 14

lacking (Orians, unpublished data). Rank (1992) also Importance of chemistry. Numerous studies have found no evidence of diet-induced feeding preferences shown that phenolic glycosides are key determinants in Chrysomela aeneicollis. of herbivory on willows (e.g. Rowell-Rahier, 1984; Fritz et al. (1994) illustrate ®ve possible responses Tahvanainen et al., 1985; Denno et al., 1990). Con- of herbivore towards hybrids: susceptibility (hybrids sequently, we hypothesized that phenolic glycosides are more acceptable), Additivity (hybrids show inter- would be an important determinant of host selection mediate acceptability), Dominance (acceptability of in this system. The preference of C. multipunctata and the hybrid resembles one of the parental species), Res- Pl. versicolora for hybrids could indicate that they are istance (hybrids are least acceptable), and No Differ- inhibited by high concentrations of phenolic glycosides ence (the taxa do not differ). Two of these patterns but stimulated to feed by low concentrations. Although were supported by our preference data: Susceptibility further work is required to determine if this is true, (C. multipunctata and Pl. versicolora) and Dominance there are two reasons we believe phenolic glycosides (the two Chrysomela species and P. japonica). may be important. First, only high concentrations of

ento1530.tex; 17/06/1997; 11:56; v.7; p.7 292 salicortin inhibited feeding by C. multipunctata (Fig- there was a growth and survivorship cost to consum- ure 2). Second, Tahvanainen et al. (1985) found that ing hybrid leaves. Larval growth of both Chrysomela Pl. versicolora prefer willow species with moderate to species was intermediate on hybrids, as was the surviv- low glycoside concentration and that high concentra- orship of Ch. scripta. In contrast to our results, Floate tions slowed growth but low concentrations appeared et al. (1993) suggested that the performance of Chryso- to elicit feeding. mela con¯uens was higher on poplar hybrids. They Both Ch. knabi and Ch. scripta convert phen- attributed this to differences in plant phenology. The olic glycosides for their own defense. Therefore their differences between our studies either re¯ects contrast- preference for S. sericea and hybrids was expected. ing resistance traits of the hybridizing species or dif- Rank (1992) also found that Chrysomela aeneicol- ferences among Chrysomela species in their response lis prefer willow species rich in phenolic glycosides. to those traits. We suspect the former, since Chryso- The fact that these Chrysomela spp. found S. sericea mela spp. are known to respond positively to phenolic and hybrids equally acceptable suggests that they may glycosides (Rank, 1992). respond to the presence of the glycosides, not to the P. japonica adults performed poorly on S. sericea concentration in each taxon. The fact that all concen- and hybrids. Overall, weight loss was highest on trations of salicortin stimulated feeding by Ch. knabi S. sericea, intermediate on hybrids and least on S. erio- lends support to this conclusion. cephala. In contrast, P. japonica preferred S. eriocephala Thus two different patterns described by Fritz et al. but avoided both S. sericea and hybrids. P. japon- (1994) were supported by our performance data:Addit- ica are generalists that are often found feeding on ivity (the two Chrysomela species and P. japonica), hosts containing condensed tannins. [Interestingly, andnodifference(C. multipunctata). Unlike the pref- condensed tannins do not appear to stimulate feed- erence results, we found no cases of Dominance or ing (Orians, unpublished data).] Their avoidance of Susceptibility. S. sericea and hybrids also is not surprising in light of the fact that few generalists consume willows that Conclusion. In the ®eld, higher herbivore densities produce phenolic glycosides (Rowell-Rahier, 1984). on hybrids could be because the herbivores prefer to Thus, P. japonica appears pre-adapted to feeding on S. feed on hybrids and/or because they perform better on eriocephala, which produces condensed tannins, but hybrids (Floate et al., 1993; HanhimakiÈ et al., 1994). not on S. sericea or the hybrids of S. eriocephala and When evaluating the relative susceptibility of hybrids S. sericea. to herbivores, few studies have looked at preference and performance simultaneously. As we found, pref- Performance. The performance of each beetle spe- erence and performance are not necessarily coupled cies was not always consistent with its preference. (Futuyma, 1991). This has important implications. C. multipunctata preferred hybrids but did not perform First, this means that the two sets of traits (those affect- better on them. In this study we found no difference in ing preference or performance) exhibit different pat- larval or adult growth across these three taxa. In anoth- terns of inheritance in hybrids. Second, year to year er experiment we actually found that short-term larval or site to site variation in the response of herbivores to growth was greatest on S. eriocephala, intermediate on hybrids may re¯ect changes in which factor,preference hybrids, and least on S. sericea (Orians, unpubl. data). or performance, determines abundance. We know there Here, we were unable to rear the beetles to the adult can be year to year variation in the response of some stage, so we could not control for sexual dimorph- herbivores to these hybrids (Fritz et al., 1996), and per- ism and this may have masked differences. Matsuki & haps this was due to differences in which mechanism MacLean (1994) found that phenolic glycosides inhib- determined abundance. it Calligrapha verrucosa growth, so we suspect that, We argue that to predict the relative resistance of as our short-term experiments suggested, that the per- hybrid and parental willows it is necessary to know the formance of Calligrapha multipunctata bigsbyanawill inheritance of putative resistance traits and the natural be lowest on S. sericea. Clearly, additional whole-life history of the attacking herbivores. Herbivores, such as studies are necessary. those studied here, are diverse and differing responses Both Chrysomela species avoided S. eriocephala to plant traits are expected. Detailed studies of oth- and performed very poorly on that taxon as well. er plant hybrid systems are likely to uncover similar Despite similar preference for S. sericea and hybrids, complexity.

ento1530.tex; 17/06/1997; 11:56; v.7; p.8 293

Acknowledgements theses, genetics, and variable responses in a diverse herbivore community. Oecologia 97: 106±117. We would like to thank Len and Ellie Sosnowski for Fritz, R. S., B. M. Roche, S. J. Brunsfeld & C. M. Orians, 1996. Interspeci®c and temporal variation in herbivore responses to letting us work on and collect from their property and hybrid willows. Oecologia, 108: 121±129. Hartwick College Biology Department for providing Futuyma, D. J., 1991. Evolution of host speci®city in herbivorous laboratory space. Doug Futuyma identi®ed several of insects: Genetic, Ecological and phylogenetic aspects. In: P. W. Price, T. M. Lewinsohn, G. W. Fernandes & W. W. Benson (eds), the beetle species. A special thanks to Rebecca Barnes Plant-Animal Interactions: Evolutionary Ecology in Tropical who assisted with data collection. We give thanks to and Temperate Regions. John Wiley & Sons, Inc., NY, pp. 431± Frank Messina and several anonymous reviewers for 454. comments on earlier versions of this manuscript. A. Hanhimaki,È S., J. Senn & E. Haukioja, 1994. Performance of insect herbivores on hybridizing trees: the case of the subarctic birches. Wild was supported by the New England Consortium Journal of Animal Ecology 63: 163±175. for Undergraduate Science Education. This research Horton, D. R., 1989. Performance of a willow-feeding beetle, was supported by a National Science Foundation grant Chrysomela knabi Brown, as affected by host species and diet- (#DEB 92007363) to R. Fritz, C. Orians and S. Brun- ary moisture. Canadian Entomologist 121: 777±780. Kearsley, M. J. C. & T. G. Whitham, 1989. Developmental changes sfeld, by the Vassar College general Research Fund, in resistance to herbivory: Implications for individuals and pop- and by the Bronfman Science Center at Williams Col- ulations. Ecology 70: 422±434. lege. Kelly, M. T. & J. P. Curry, 1991. The in¯uence of phenolic com- pounds on the suitability of three Salix species as hosts for the willow beetle Phratora vulgatissima. Entomologia Experi- mentalis et Applicata 61: 25±32. References Kolehmainen, J, H. Roininen, R. Julkunen-Tiitto & J. Tahvanain- en, 1994. Importance of phenolic glucosides in host selection Abacus Concepts, 1992. StatView 4.02. Abacus Concepts, Inc., of shoot galling saw¯y, Euura amerinae,onSalix pentandra. Berkeley, CA. Journal of Chemical Ecology 20: 2455±2466. Aide, T. M., 1988. Herbivory as a selective agent on the timing of Lindroth, R. L., J. M. Scriber & M. T. S. Hsia, 1988. Chemical eco- leaf production in a tropical understory community. Nature 336: logy of the tiger swallowtail: mediation of host use by phenolic 574±575. glycosides. Ecology 69: 814±822. Argus, G. W., 1974. An experimental study of hybridization and Matsuda, K. & H. Matsuo, 1985. A ¯avonoid, luteolin-7-glucoside, pollination in Salix (willows). Canadian Journal of Botany 52: as well as salicin and populin, stimulating the feeding of leaf 1613±1619. beetles attacking Salicaceous plants. Applied Entomology and Argus, G. W., 1986. The genus Salix (Salicaceae) in the southeastern Zoology 20: 305±313. United States. Systematic Botany Monographs 9: 1±170. Matsuki, M. & S. F. MacLean Jr, 1994. Effects of different leaf traits Arnold, M. L., 1992. Natural hybridization as an evolutionary pro- on growth rates of insect herbivores on willows. Oecologia 100: cess. Annual Review of Ecology and Systematics 23: 237±261. 141±152. Boecklen, W. J. & R. Spellenberg, 1990. Structure of herbivore Mattson, W. J., 1980. Herbivory in relation to plant nitrogen content. communities in two oak (Quercus spp.) hybrid zones. Oecologia Annual Review of Ecology and Systematics 11: 119±161. 85: 92±100. Nichols-Orians, C. M., T. P. Clausen, R. S. Fritz, P. B. Reichardt Bryant, J. P., 1981. Phytochemical deterrence of snowshoe hare & J. Wu, 1992. A new phenolic glycoside isolated from Salix browsing by adventitious shoots of four Alaskan trees. Science sericea Marshall. Phytochemistry 31: 2180±2181. 213: 889±890. Nichols-Orians, C. M., R. S. Fritz & T. P. Clausen, 1993. The genetic Bryant, J. P., T. P. Clausen, P. B. Reichardt, M. C. McCarthy & R. A. basis for variation in the concentration of phenolic glycosides Werner, 1987. Effect of nitrogen fertilization upon the second- in Salix sericea: Clonal variation and sex-based differences. ary chemistry and nutritional value of quaking aspen (Populus Biochemical Systematics and Ecology 21: 535±542. tremuloides Michx.) leaves for the large aspen tortrix (Choris- Orians, C. M. & T. Floyd, 1997. The susceptibility of parental and toneura con¯ictana (Walker)). Oecologia 73: 513±517. hybrid willows to plant enemies under contrasting soil nutrient Christensen, K. M., T. G. Whitham & P. Keim, 1995. Herbivory and conditions. Oecologia 109: 407±413. tree mortality across a pinyon pine hybrid zone. Oecologia 101: Orians, C. M. & R. S. Fritz, 1995. Secondary chemistry of hybrid and 29±36. parental willows: Phenolic glycosides and condensed tannins Denno, R. F., S. Larsson & K. L. Olmstead. 1990. Role of enemy- in Salix sericea, S. eriocephala and their hybrids. Journal of free space and plant quality in host-plant selection by willow Chemical Ecology 21: 1245±1253. beetles. Ecology 71: 124±137. Orians, C. M. & R. S. Fritz, 1996. Genetic and soil-nutrient effects Dixon, W. J. & F. J. Massey Jr., 1983. Introduction to Statistical on the abundance of herbivores on willow. Oecologia 105: 388± Analysis, 4th edition. McGraw Hill Inc., Sidney. 396. Floate, K. D., M. J. C. Kearsley & T. G. Whitham, 1993. Elevated Paige, K. N. & W. C. Capman, 1993. The effects of host-plant herbivory in plant hybrid zones: Chrysomela con¯uens, Populus genotype, hybridization and environment on gall aphid attack and phenological sinks. Ecology 74: 2056±2065. and survival in cottonwood: the importance of genetic studies Fritz, R. S., C. M. Nichols-Orians & S. J. Brunsfeld, 1994. Interspe- and the utility of RFLP's. Evolution 47: 36±45. ci®c hybridization of plants and resistance to herbivores: hypo- Price, P. W., H. Roininen & J. Tahvanainen, 1987. Why does the bud- galling saw¯y, Euura mucronata, attack long shoots? Oecologia 74: 1±6.

ento1530.tex; 17/06/1997; 11:56; v.7; p.9 294

Rank, N. E., 1992. Host plant preference based on salicylate chem- cialisation of some of their herbivorous insects. Oecologia 62: istry in a willow leaf beetle (Chrysomela aeneicollis). Oecologia 26±30. 90: 5±101. Schultz, J. C., 1989. Tannin-insect interactions. In: R. W. Hem- Richardson, B. & J. B. Whittaker, 1982. The effect of varying the ref- ingway & J. J. Karchesy (eds), Chemistry and Signi®cance of erence material on ranking of acceptability indices of plant spe- Condensed Tannins. Plenum Press, New York, pp. 417±433. cies to a polyphagous herbivore, Agriolimax reticulatus.Oikos Sokal, R. K. & F. J. Rohlf, 1981. Biometry: The Principles and 39: 237±240. Practice of Statistics in Biological Research, 2nd edition. WH Rieseberg, L. H. & N. C. Ellstrand, 1993. What can molecular Freeman and Company, NY. and morphological markers tell us about plant hybridization? Strauss, S. Y., 1994. Levels of herbivory and parasitism in host Critical Reviews in Plant Sciences 12: 213±241. hybrid zones. Trends in Ecology and Evolution 9: 209±214. Roa, R., 1992. Design and analysis of multiple-choice feeding- Tahvanainen, J., R. Julkunen-Tiitto & J. Kettunen, 1985. Phenolic preference experiments. Oecologia 89: 509±515. glycosides govern the food selection pattern of willow feeding Roininen, H. & J. Tahvanainen, 1989. Host selection and larval leaf beetles. Oecologia 67: 52±56. performance of two willow-feeding saw¯ies. Ecology 70: 129± Whitham, T. G., 1989. Plant hybrid zones as sinks for pests. Science 136. 244: 1490±1493. Rowell-Rahier, M., 1984. The presence or absence of phenolglycos- ides in Salix (Salicaceae) leaves and the level of dietary spe-

ento1530.tex; 17/06/1997; 11:56; v.7; p.10