Oecologia (2001) 129:87–97 DOI 10.1007/s004420100703
Robert S. Fritz · Cris G. Hochwender Debra A. Lewkiewicz · Sara Bothwell · Colin M. Orians
Seedling herbivory by slugs in a willow hybrid system: developmental changes in damage, chemical defense, and plant performance Received: 6 August 2000 / Accepted: 26 January 2001 / Published online: 15 May 2001 © Springer-Verlag 2001
Abstract We evaluated feeding preference and damage by the slug, Arion subfuscus, on seedlings of two willow species, Salix sericea and S. eriocephala, and their F1 interspecific hybrids. Trays of seedlings were placed in the field and excised leaves were presented to slugs in choice tests. Slugs preferred feeding on and caused the most damage to S. eriocephala seedlings. S. sericea seedlings were least preferred and least damaged. F1 hybrid seedlings were intermediate in preference and damage. Slug preference of and damage to these seedlings decreased over time, suggesting developmental changes in resistance. Seedlings were sampled for phenolic glycoside and tannin chemistry weekly to coincide with the field and laboratory experiments. Concentrations of phenolic glycosides and tannins increased linearly with seedling age, coincident with changes in slug preference and damage, indicating a developmental change in defense. Slug deterrence was not detected at low concentrations of salicortin when painted on leaves or discs, but both salicortin and condensed tannins deterred slug feeding at concentrations between 50 and 100 mg/g, levels found in adult willows. Seedling performance was related to damage inflicted by slugs. Due to lower levels of damage when exposed to slugs in the field, S. sericea plants had significantly greater biomass than S. eriocephala plants. Biomass of F1 hybrids was equal to S. sericea when damaged. However, undamaged S. eriocephala and F1 hybrid plants had the greatest biomass. Because F1 hybrid seedlings performed as well as the R.S. Fritz (✉) · C.G. Hochwender · D.A. Lewkiewicz S. Bothwell Department of Biology, Vassar College, Poughkeepsie, NY 12604, USA e-mail:
[email protected] Fax: +1-914-4377315 C.M. Orians Department of Biology, Tufts University, Medford, MA 02155, USA Present address: D.A. Lewkiewicz, Department of Biology, University of California, Riverside, CA 92521, USA
most fit parent in all cases, slugs could be an important selective factor favoring introgression of defensive traits between these willow species. Keywords Arion subfuscus · Fitness · Hybrid · Chemical defense · Salix
Introduction Detailing the relationships between hybridization, resistance to herbivores, and plant fitness will help to clarify the evolutionary implications of plant hybridization. Interspecific hybridization can alter plant resistance to herbivores, such that hybrids either have lower resistance, higher resistance, or intermediate resistance compared to that of parents (for reviews see Strauss 1994; Fritz 1999; Fritz et al. 1999). Often, hybrids do not show lower resistance than both parents (i.e., hybrid susceptibility), but where hybrid susceptibility does occur, it could lead to decreased hybrid fitness relative to parental taxa (e.g., Drake 1981; Morrow et al. 1994; Cummings et al. 1999), potentially resulting in restricted populations and distributions of hybrids. In those cases, hybridization might have few influential evolutionary implications beyond strengthening species boundaries. However, selection can favor certain genes in hybrids, even when overall hybrid fitness is low, and thus hybridization can still have an important evolutionary effect on lineages (Arnold and Hodges 1995; Arnold et al. 1999). When hybrid resistance is equal to or greater than that of the parental taxa, or when selection favors particular genes in hybrids, hybrid plants may have equivalent or higher fitness compared to parents (Arnold and Hodges 1995; Arnold 1997), leading to larger hybrid populations with more widespread distributions. Hybridization could have at least two important evolutionary consequences – the creation of and selection for recombinant hybrid progeny (F2 or backcrosses) that have novel combinations of defense traits (Rieseberg and Wendel 1993; Fritz et al. 1999; Orians, in press), and introgression of defensive
88
traits between species (Stutz and Thomas 1964; Manley and Fowler 1969; Wu et al. 1996). In order to understand the influence of hybridization on fitness via herbivore resistance, particular attention should be paid to early developmental stages of plants. Herbivores can have their greatest impact on plant fitness by consuming seedling or juvenile plants (Huntly 1991; Bryant and Julkunen-Tiitto 1995). Further, seedlings often have the highest mortality of any developmental stage (Harper 1977; Crawley 1983). In addition, the environmental factors that influence seedling survival may not be the same for mature plants, so seedlings must be evaluated to understand which factors are critical during plant establishment (Harper 1977). For example, molluscs have a propensity for feeding on seedlings, thereby making themselves an influential mortality agent of seedlings (Hatto and Harper 1969; Dirzo and Harper 1980; Rai and Tripathi 1985), but not of adult plants. In addition, slugs and snails are often highly selective in their choice of host plant species (Cates 1975; Cates and Orians 1975; Dirzo 1980; Whelan 1982; Rai and Tripathi 1985; Rathke 1985; Hanley et al. 1995; Fenner et al. 1999; Scheidel and Bruelheide 1999), so mollusc herbivory can act as an important selective force on the abundance and distribution of plant species (Bruelheide and Scheidel (1999). Differences in the level of defenses expressed by plant species may underlie such molluscan preferences. Some plant species exhibit high levels of defenses in newly emerged seedlings (e.g., Glen et al. 1990; Thakray et al. 1990; Morse et al. 1991), whereas defenses are not immediately expressed in some species (Horrill and Richards 1986). Changes in seedling palatability can affect risk of damage by molluscs. For example, slugs and snails are often highly selective among chemical and morphological phenotypes of plants (Jones 1962; Cates 1975; Westerbergh and Nyberg 1995), and as seedlings age, the likelihood of attack often decreases (Bryant and Julkunen-Tiitto 1995; Hanley et al. 1995; Fenner et al. 1999). Thus, slugs may act as a major selective force favoring greater seedling defense within a lineage, as well as altering interspecific plant abundance. We have been studying herbivory by arthropods on two willow species (Salix eriocephala and S. sericea) and their interspecific hybrids to determine the patterns of relative susceptibility of hybrid and parent plants. S. eriocephala is a relatively fast growing species that utilizes condensed tannins as a foliar defense, whereas S. sericea is a slower growing species that utilizes phenolic glycosides (salicortin and 2′-cinnamoylsalicortin) (Orians et al. 1997). The concentrations of both phenolic glycosides and condensed tannin in hybrids in the field and in F1 and F2 hybrids produced from crosses of pure parents are generally intermediate, suggesting additive inheritance of these traits (Fritz et al. 1994; Orians and Fritz 1995; Orians et al. 2001; C. M. Orians, unpublished data). Seedlings of both parental species and F1 hybrids are susceptible to slug herbivory in the field, but a developmental shift in the palatability of willows to slugs seems to occur before the end of the first year of growth.
Given the potentially important impact of slugs on willow seedlings at our site, we evaluated the interaction between slugs and willow seedlings in relation to chemical defense. Specifically, we addressed the following questions: (1) Does foliar feeding by slugs differ among parental and hybrid willow seedlings in choice tests, and do levels of consumption change over time? (2) Do parental and hybrid willow seedlings sustain different levels of damage by slugs in the field, and do those damage levels change over time? (3) Do chemical defenses of parental and hybrid willows change with seedling age (are there developmental changes in defense)? (4) Do putative chemical defenses deter slug herbivory? (5) Do differences in slug damage experienced by plants in the field alter the relative performance among taxa?
Materials and methods Study area and system This study was conducted in 1997 and 1998 at the Sosnowski site; a low-lying, willow swamp about 3 km from Milford, Otsego County, N.Y. Two willow species, Salix sericea and S. eriocephala, and their interspecific hybrids occur naturally at the site, along with two other willows, S. discolor and S. bebbiana. For this study we performed intraspecific and interspecific crosses using genetically pure S. sericea and S. eriocephala plants (Fritz et al. 1996) to produce three independent crosses of S. sericea, S. eriocephala, and F1 seeds. RAPD markers were used to determine parents’ purity (Hardig et al. 2000). Seeds were germinated in trays of potting soil (Metro Mix 360) during the second week of June 1997 and the second week of May 1998. Two- to 3-week-old seedlings were transplanted into potting soil into 72-celled trays. The trays of seedlings were maintained in the greenhouse until used for experiments with regular watering and fertilization with Peters all-purpose 20: 20: 20 fertilizer as needed. The slug Arion subfuscus (Draparnaud) (Mollusca: Arionidae) is the primary herbivore of willow seedlings at our study site (personal observation). A. subfuscus was introduced from Europe (Chichester and Getz 1969, 1973) and is one of several introduced slug species in North America that are dominant molluscan herbivores in many communities. This species is among the most common slugs in many habitats, with the exception of forests (Chichester and Getz 1969). Early records documented A. subfuscus in Massachusetts by the 1840s (Chichester and Getz 1969). Therefore, this species has been established in the northeastern United States for at least 150 years. It is unknown how abundant native slug populations were prior to the introduction of European slug species or how they might have interacted with plant species (D. Strayer, personal communication). A. subfuscus is very common from May through July at our site, and it is an important herbivore of willow seedlings. Slug damage to young willow seedlings is often severe (personal observation). We collected and maintained a population of slugs throughout the summer for use in two experiments. In 1997, slugs were housed together in a large fiber bait box. In 1998, slugs were housed individually in 470-ml plastic containers lined with moist filter paper or paper towels. Slug containers in both years were buried in the ground in a shady spot at our field site to protect slugs from heat stress. Slugs were provided ad libitum with Romaine lettuce, except during periods of food-deprivation before experiments. Feeding choice experiments (laboratory) We conducted feeding preference experiments using excised leaves of parental and hybrid plants to control for confounding
89 factors that could influence slug herbivory of seedlings in the field, such as changes in slug abundance over time and changing environmental conditions. This experiment was conducted weekly to coincide with the weekly field choice experiments (see below). Seedlings used in these feeding choice experiments were obtained from replicate trays of seedlings of the same crosses and ages as seedlings in the field choice experiments. Seedlings were maintained under identical conditions in the greenhouse. We placed single, excised leaves of the same age and of similar sizes from seedlings of the three taxa abaxial side up along three equally spaced radii in a petri dish lined with moist filter paper. We placed a slug that had been deprived of food for at least 24 h into each petri dish and left them overnight to feed. Experiments were established between 2000 and 2100 hours. The next morning between 0800 and 0900 hours, we scored leaves for slug feeding in 10% damage intervals (0=0% feeding damage; 10=100% feeding damage) by visually assessing the proportion of leaf area removed. We analyzed damage scores using Wilcoxon signed-rank tests to determine differences in damage scores among taxa and over time. Slugs used in these trials had been collected from the field near where willows grow; thus some of them may have had prior exposure to willow seedlings, including both parent species and hybrids. Since seedlings in the field had low chemical concentrations at the time slugs were collected, it is less likely that slugs used in the trials had learned aversions to seedling. However, we cannot discount the possibility that slugs may have had aversions or preferences for willow seedlings.
within 1 m of each field tray location. We systematically searched for A. subfuscus individuals within our plots at night (2100–2300 hours) using head lamps. We compared slug numbers among weeks using a Kruskal-Wallis test. Changes in chemical defense in seedlings Adult S. sericea plants have phenolic glycosides and adult S. eriocephala plants have condensed tannins that act as defenses (Orians and Fritz 1995; Orians et al. 1997), but seedling plants have little or no chemical defenses; therefore, developmental changes in the onset of these chemical defenses could affect resistance. To quantify changes in concentrations of defensive chemicals, weekly samples of seedlings of each taxon were collected for chemical analysis from extra trays of seedlings maintained in the greenhouse under identical conditions as those used in the field tray and feeding choice experiments. Seedlings were not collected for chemical samples at the youngest ages used in our experiments, since a large number of seedlings would have to be sampled to provide enough tissue for analysis. Seedlings were cut at the base with scissors, placed in coin envelopes, and returned to the laboratory, where they were vacuum dried. Some seedlings were kept for a time in a freezer before vacuum drying. Care was taken to prevent thawing. Analysis of salicortin, 2′-cinnamoylsalicortin, and tannin concentrations followed protocols in Orians and Fritz (1995). Data were analyzed by performing regressions between the week that samples were taken and concentration of the each foliar chemical.
Feeding choice experiments (field trays) This experiment was designed to determine slug herbivory patterns among seedlings of S. sericea, S. eriocephala, and their interspecific F1 hybrids. In 1998, we randomly planted seedlings of the three taxa, 24 S. sericea, 24 S. eriocephala, and 24 hybrid seedling for each tray, in 23 trays (72 cells per tray). Colored plastic toothpicks were used to mark the locations of each taxon of seedling. Trays remained in the greenhouse until they were put in the field. A new set of trays was placed in the field in designated slug-dense areas on a weekly basis. The first set of trays was placed into the field during the first week of July (week 5, measured as weeks since seed germination) and the last set was placed in the field on the second week in August (week 10). After 1 week in the field, we returned trays to the greenhouse and scored each seedling for slug damage. Damage scores, ranging from 0–5, were given for 20% damage increments (0=0% damage, 5=100% damage). We compared damage scores among taxa and weeks using two-way ANOVAs. Initial seedling leaf number was used as a covariate to correct for plant size differences. Analysis of ranked data did not alter results, so we present analyses using non-transformed data. To determine if damage scores for each taxon changed over time, we compared the differences in mean damage score among weeks for each taxon. Taxa were then compared by week. We performed contrasts among taxa within each week to test the fit of observed slug damage to the hypothesized patterns of hybrid resistance (Fritz et al. 1996). We first compared the damage scores of the F1 hybrids to the average of the two parents (midparent). If F1 hybrids were significantly different than the mid-parent value, we then compared hybrid damage scores to damage score of the nearest parent (S. eriocephala in all cases). When this difference was significant, the pattern was classified as partial dominance; if not significant it was classified as dominance. If the hybrid versus midparent difference was not significant, we then compared the damage scores for the two parents. When this difference was significant, we then classified hybrids as having an additive pattern. When this difference was not significant, we classified hybrids as having no difference in resistance compared to either parent. To monitor slug density near the locations of field trays at the time trays were placed in the field, we conducted weekly censuses. We marked 1×2 m rectangular plots of similar habitat located
Deterrence of defensive chemicals To estimate the concentrations of salicortin and condensed tannins necessary to deter slug herbivory, we conducted chemical painting experiments. In 1997, we painted leaves from young S. eriocephala seedlings that only produce very low concentrations of tannins (see Results) with various concentrations of salicortin or condensed tannin solution. Salicortin was isolated from bulk samples of S. sericea leaves, and condensed tannins were isolated from bulk samples of S. eriocephala leaves. Solutions were prepared using acetone as a solvent, and concentrations were determined based on leaf dry weight. Choice tests between control leaves with and without acetone suggest that slugs did not discriminate against acetone (unpublished data). We painted leaves that had an estimated average dry weight of 3.113 mg with 7 µl of various solutions of salicortin or tannin and allowed leaves to dry before introducing slugs. We placed slugs, deprived of food for at least 24 h, into petri dishes lined with moist filter paper with one control leaf painted with acetone, and one treatment leaf painted with salicortin or tannin solution. In the first trial, we prepared 20 petri dishes using a 75 mg/g salicortin solution and 20 petri dishes using a 50 mg/g tannin solution. In the second trial, 30 petri dishes were prepared using 100 mg/g salicortin and 30 petri dishes were prepared using 100 mg/g tannin solutions. Leaves were scored for damage in 20% increments. Data were analyzed using two-tailed paired t-tests. In 1998, to further examine the effects of salicortin, we used Romaine lettuce instead of S. eriocephala leaves to eliminate the possible synergistic effects of tannins or other defense compounds in the leaves. Lettuce is the standard reference food for slug preference tests. We painted 3.5 cm2 disks of lettuce (average dry weight: 9.42 mg) with 30 µl of various concentrations of salicortin in acetone to achieve the following concentrations, based on dry weight: 1 mg/g, 3 mg/g, 6 mg/g, 10 mg/g, 20 mg/g, 30 mg/g, 50 mg/g, and 100 mg/g. These concentrations were chosen to span a possible developmental range of concentrations that might occur in seedlings. Control disks were painted with 30 µl of acetone. We placed slugs that had been deprived of food for 15 h into petri dishes lined with moist filter paper with one control and one treatment disk. After 8 h, we assessed slug damage in 20% intervals. Due to constraints in slug numbers during any 1 week, three sepa-
90 rate trials involving various combinations of salicortin concentrations were spread out over a period of 3 weeks. For each concentration, we used paired t-tests to compare control and treatment disks. Since conditions between the three trials did not differ, we grouped the results of trials. Because the painting assays lasted 8 h in a humid environment, chemical breakdown is likely. Therefore, these results provide a relative measure of how deterrence changes with increasing chemical concentration, not an absolute measure of deterrence of a particular concentration. Performance of damaged and undamaged seedlings Performance of damaged seedlings was determined by measuring dry biomass at week 12. We collected the surviving seedlings from trays placed in the field în weeks 6, 7 and 8. Trays were returned to the greenhouse after exposure to slugs and permitted to grow under greenhouse conditions (watering and ambient light), but received no further fertilizer supplements. Performance of two trays of undamaged seedlings maintained in the greenhouse throughout the study was also determined by measuring dry biomass at week 12, with leaf number having been measured at week 11. Seedling biomass data were adjusted for the initial number of leaves on each seedling at the time they were placed in the field (or, in the case of the undamaged seedlings, 1 week prior to harvesting) by entering initial leaf number as a covariate in an analysis of variance. This measure of residual biomass (hereafter just biomass) is therefore a measure of seedling performance with initial size taken into account. We compared biomass among the taxa for undamaged plants and at each week for damaged plants using two-way ANOVA. The effect of week that damage occurred was not significant. Therefore, we combined weeks and compared differences among the taxa using t-tests of least square means (corrected by use of sequential Bonferroni adjustments) generated in PROC GLM (SAS 1990).
Table 1 Results from pairwise Wilcoxon signed-rank test showing significance values of differences between damage scores of taxa by week for the feeding preference experiments. Only one significant difference would have been expected by chance alone given 18 separate tests Week
E vs H
E vs S
H vs S
5 6 7 8 9 10
0.001 0.001 0.007 0.026 0.047 0.001
0.001 0.001 0.001 0.111 0.001 0.069
0.001 0.001 0.001 0.001 0.293 0.001
Results Feeding choice experiments (laboratory) The results of the feeding choice experiments showed significant differences among taxa and over time in percent leaf consumption by slugs (Fig.1). In 1998, from week 5 to 7, S. eriocephala had significantly higher damage scores than hybrids, and hybrids had significantly higher damage scores than S. sericea (Fig. 1, Table 1). A marked decline in feeding on S. eriocephala and hybrid leaves occurred between weeks 7 through 9 (Fig. 1). Mean percent damage of S. eriocephala dropped from 64% to 2.6% from week 7 to 9. During the same period, percent damage to F1 hybrids decreased from 46% to 5% and S. sericea dropped from 17% to 5%. In weeks 9 and 10, all three taxa had very low damage scores; the differences among them were fewer and seem biologically less relevant because of the low damage levels (Fig. 1, Table 1). In similar feeding choice experiments conducted in 1997, S. eriocephala seedlings had the highest damage scores of the three taxa, S. sericea seedlings had the lowest scores, and hybrids had intermediate scores (data not shown). Slugs continued to feed on lettuce provided for them in their storage containers (personal observation), supporting the hypothesis that decline in consumption of willow leaves is related to changes in palat-
Fig. 1 Plot of percent leaf damage scores (mean ± 1SE) for Salix sericea, S. eriocephala, and F1 hybrids from plants of different age (weeks) offered to the slug A. subfuscus in tripartite choice tests
ability, and not a general decrease in slug’s consumption rate at that time. Feeding choice experiments (field trays) The percent damage to willow seedlings placed in trays in the field for 1 week differed among taxa (Table 2, Fig.2). Significant differences in damage among tray locations (blocks) occurred for weeks 5–7 and 10. For weeks 6 and 7, significant interactions between taxon and tray location also existed (Table 2). In addition, initial leaf number had significant effects during weeks 8 and 9. Still, the overriding relationship between damage and taxon was not obscured by any other significant effect; S. sericea had lower percent damage than did S. eriocephala (Fig. 2). For weeks 5–7, slug damage on S. eriocephala was between 63% and 85%, while only between 17% and 24% on S. sericea.
91 Table 2 F-values of 2-way ANOVA comparing proportion damage, and F-values of contrasts between Salix eriocephala (E) vs S. sericea (S), E vs hybrids (H), and S vs H in field tray damage
trials generated from the two-way ANOVA. Hybrid patterns of resistance detected included partial dominance of hybrid susceptibility (PD), an additive pattern (A), and no difference (ND)
Week
Initial leaf number
Taxon
Location
Taxon by location
Hybrid vs midparent
5 6 7 8 9 10
0.30 0.43 0.50 20.62*** 14.46*** 0.35
287.90*** 357.81*** 90.45*** 42.17*** 10.42*** 2.25
42.98*** 8.22*** 12.41*** 0.00 0.05 5.26*
1.46 2.15* 13.99*** 1.05 0.07 1.28
18.68*** 34.77*** 13.78*** 1.63 0.00 1.14
Parents
82.65*** 20.79*** 3.41
Hybrid vs near parent
Hypothesis supported
63.61*** 61.53*** 10.67***
PD PD PD A A ND
*P
