• Arion;
  • Arnica montana;
  • Deroceras;
  • feeding deterrents;
  • food selection


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

1  In order to test the hypothesis that the sensitivity of field populations of the rare perennial Arnica montana to slug attack is due to its high palatability, we compared Arnica leaves and seedlings with 20 other plant species in greenhouse food choice experiments.

2  When slugs were given a choice of mature plant leaf discs, Arnica montana was one of the food plants most preferred by all three slug species tested. Arion lusitanicus, Arionsubfuscus and Deroceras agreste differed only slightly in their preference for particular plant species. Greater differences in acceptability were observed when parameters other than area consumed were evaluated. Arnica seedlings displayed similarly high susceptibility to slug grazing.

3  Morphological factors influence mollusc feeding behaviour, with epidermal cell wall thickness and hairiness related to acceptability. Removing the hairs by shaving the leaves increased the acceptability of one species, Centaurea jacea.

4  Leaves of undamaged Arnica montana plants were preferred to leaves of plants that had been predamaged by partial leaf removal, suggesting that there is an inducible component of chemical defence.

5  We discuss our results in comparison with other food choice experiments and attempt to extrapolate them to mollusc/plant interactions under field conditions.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Slugs and snails form an important part of the herbivore fauna of different vegetation types. Although more or less polyphagous, they show distinct food preferences. If a rare plant species shows exceedingly high acceptability to slugs it may become excluded from sites where the presence of other, more abundant food plants supports high mollusc activity. Mollusc herbivory could therefore be a major factor for limiting plant geographical ranges and, in particular, the restriction of montane plant species to higher altitudes may result from higher herbivory pressure in lowlands. We tested this hypothesis for the rare grassland perennial Arnica montana L. (Compositae) in the Harz mountains, Lower Saxony, Germany (Bruelheide & Scheidel 1999), where transplanted Arnica individuals were shown to suffer increasing slug damage with decreasing altitude. Here we present the results of mollusc food choice experiments under greenhouse conditions.

Food selection has been investigated extensively for both slugs (Gain 1891; Pallant 1972; Jennings & Barkham 1975; Dirzo 1980; Lawrey 1983; Rathke 1985; Briner & Frank 1998; Wardle et al. 1998) and snails (Grime et al. 1968, 1970, 1996; Speiser & Rowell-Rahier 1993). However, most studies have involved plant species of high abundance or agricultural importance, and mollusc herbivory of threatened plant species has rarely been investigated. We investigated whether slugs graze on the leaves of a highly endangered species, Arnica montana, in preference to other species from the same montane grassland. Species of the Compositae are generally of high mollusc acceptability, e.g. Leucanthemum vulgare (Cates & Orians 1975; Briner & Frank 1998) and Taraxacum officinale (Dirzo 1980; Cook et al. 1996; Wardle et al. 1998), but although Arnica montana is known to be harmful to cattle, which avoid this plant (Hegi 1987, p. 707), it has not been tested on any of its natural herbivores.

The results of earlier food choice experiments are as diverse as the mollusc species used. For example, the extensively tested Plantago lanceolata was not eaten by Deroceras caruanae (Dirzo 1980), Cepaea nemoralis (Grime et al. 1968) and Arion lusitanicus (Briner & Frank 1998), only slightly by D. reticulatum (Wardle et al. 1998) but greatly by Helix aspersa (Grime et al. 1996). We therefore offered our food plants to three slug species that are normally present in montane meadows to determine whether they show different preference patterns.

Molluscs are known to prefer seedlings to mature plants (Byers & Bierlein 1982). Hanley et al. (1995a) found similar attack rates for 7-day-old seedlings of three plant species, but species-specific changes of palatability with seedling age, possibly due to changing contents of chemical deterrents (Horrill & Richards 1986; Glen et al. 1990) or altered morphological properties (Hanley et al. 1995a). Hulme (1994) highlighted the importance of seedling size as the primary factor determining the risk of mollusc herbivory. For rare or endangered species, seedling survival can be assumed to be of pre-eminent importance for maintaining population sizes and we therefore assessed the extent of slug grazing on Arnica seedlings as well as on leaf discs from mature plants.

It has been suggested that leaf palatability to herbivores has three major components (Grime et al. 1996): chemical defence (as a result of distasteful or toxic substances), physical defence (hardness of plant tissue or presence of epidermal hairs) and substances that promote acceptability (by offering a pleasant taste or lucrative nutrient composition), although neither the effectiveness of any single factor nor the interactions between them is yet understood (Pennings et al. 1998). We examined the potential morphological defence mechanisms of Arnica montana. The effectiveness of epidermal hairs as a barrier against mollusc consumption is controversial. Neither Grime et al. (1968) nor Dirzo (1980) found much evidence to suggest that hairs reduce palatability of leaves, and Cook et al. (1996) stated that the hairiness of leaves does not deter molluscs from eating them. In contrast, Jennings & Barkham (1975) found that several species of Arionidae preferred glabrous, soft-textured plant species, and Westerbergh & Nyberg (1995) detected a preference of hairless ecotypes of Silene dioica in two Arion species. We compared feeding of slugs on normal leaves with those from which the hairs were removed by shaving.

Other physical factors may be important. For example, Dirzo (1980) found a high acceptability of plants with soft epidermis and Grime et al. (1968) suggested that a hard external surface may be an effective barrier against consumption by the snail Cepaea nemoralis. We therefore assessed the thickness of epidermal cell wall of the food plants as a measure of their ‘hardness’.

We also tested the potential for chemical defence. Arnica montana is known to produce Helenalin (Frohne & Jensen 1992, p. 201), a sesquiterpene lactone similar to compounds found to deter snails from consuming plants of the tribe Senecioneae (Compositae) (Speiser et al. 1992; Hägele et al. 1996). Production of such defence chemicals may be constitutive or inducible, and several investigators have found that changes in chemical (Van Dam & Vrieling 1994) or morphological (Pullin & Gilbert 1989) defence mechanisms follow defoliation. We tested whether Arnica montana showed such inducibility by comparing feeding on discs from undamaged plants with those taken from plants that had been damaged in the field 4 weeks previously.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Experiment 1: consumption of leaf discs

Arnica montana and 20 other plant species (Table 1) were offered to three different slug species. Arion lusitanicus (Mabille), A. subfuscus (Draparnaud), both Arionidae, and Deroceras agreste (L.) (Limacidae) occur in central European meadows and were collected at locations between Göttingen and the Harz mountains in Lower Saxony, Germany. The first experiment was conducted in June 1997 with A. lusitanicus from a garden at about 165 m a.s.l.; the second in July 1997 with A. subfuscus collected at an edge of a forest at about 320 m a.s.l. The third experiment was carried out in September 1997 using D. agreste found on a small slope at 160 m a.s.l. The immature specimens of A. lusitanicus, A. subfuscus and D. agreste that were used had a mean weight of 2.52 g, 0.89 g and 0.18 g, respectively. The experiments were conducted in the greenhouse at the New Botanical Garden in Göttingen, Germany. The slugs were kept under shaded conditions at maximum relative humidity with natural day length and a daily temperature range between 15 °C and 25 °C.

Table 1.  Plant species used. Nomenclature after Ehrendorfer (1973)
Plant speciesAbbreviationFamilyExperiment 1Experiment 2Experiment 3aExperiment 3bExperiment 4
Arnica montana L.AmAsteraceae+++++
Alchemilla monticola OpizAvRosaceae++
Betonica officinalis L.BoLamiaceae+++
Calendula officinalis L.CoAsteraceae+++
Centaurea jacea L.CjAsteraceae+++
Geranium sylvaticum L.GsGeraniaceae+++
Hieracium laevigatum Willd.HlAsteraceae+++
Hypericum maculatum L.HmHypericaceae++
Knautia arvensis (L.) Coult.KaDipsacaceae+++
Leucanthemum vulgare Lamk.LvAsteraceae++++
Plantago lanceolata L.PlPlantaginaceae+++
Phyteuma nigrum F.W.SchmidtPnCampanulaceae++
Poa chaixii Vill.PcPoaceae++
Polygonum bistorta L.PbPolygonaceae++
Rumex acetosa L.RaPolygonaceae+++
Silene dioica (L.) Clairv.SdCaryophyllaceae++++
Solidago virgaurea L.SvAsteraceae+++
Succisa pratensis MoenchSpDipsacaceae+++
Taraxacum officinale Web.ToAsteraceae++
Trifolium pratense L.TpFabaceae+++
Veronica chamaedrys L.VcScrophulariaceae+++

Most of the montane plant species listed in Table 1 were harvested in a meadow near Braunlage in the Harz mountains at about 600 m a.s.l. and the remainder near Bad Lauterberg on the southern border of the Harz mountains (320 m a.s.l.). The annual Calendula officinalis, which is not native to meadows, was used because it has chemical components with medicinal effects comparable to those of Arnica montana (Düll & Kutzelnigg 1992, p. 81); it was collected in a garden at 165 m a.s.l. Fully expanded, non-senescent leaves were collected and stored in polyethylene bags at 5 °C for up to 5 days.

Discs of 20 mm (for feeding to A. lusitanicus), 18 mm (A. subfuscus) or 15 mm (D. agreste) diameter were removed avoiding the mid-rib of the basal half of the leaf. Complete leaves of approximately the same area were used for the small-leaved species Veronica chamaedrys and Hypericum maculatum and 4-cm portions of the 4–5-mm broad leaves of the grass Poa chaixii. In preliminary experiments the acceptability of leaves with intact edges did not differ from cut discs.

Slugs were kept individually in 12-cm diameter clay pots covered with a perforated plastic film (Arionidae) or in 9-cm Petri dishes (D. agreste). The slugs were not starved prior to the experiment. Spare slugs were kept in pots containing a mixture of the plant species used.

Each slug was offered a combination of leaf discs of three different plant species for a period of approximately 24 h. The 21 plant species were combined in such a way that every species was paired with every other species exactly once, resulting in 70 combinations that were randomly allocated to the slugs.

The procedure was repeated eight times on consecutive days, using the same slugs with a different set of 70 species combinations, to give a series of replicates in time.

On the following day the leaf area consumed was measured by placing the discs on graph paper and assessing the area of the holes. Where leaf discs had expanded due to altered tissue tension following excision, the size of the holes was reduced proportionately. When a slug did not eat anything this combination was fed once again using a different slug the next day.

Dry weight was calculated from leaf area by regression, which was derived after drying cut leaf discs of defined area at 105 °C for about 20 h and weighing them. The leaf area eaten and corresponding dry weight was calculated for each species and its contribution to the total for that day was calculated. Within each tested combination we ranked the species according to area eaten, and calculated a mean rank from all combinations that involved that species.

Additionally, all discs that showed any signs of feeding were counted for each plant species.

Experiment 2: consumption of seedlings

To test whether selective consumption of leaf discs of mature plants varies from preferential grazing on seedlings, six species were selected (Table 1) and seedling mixtures subjected to slug grazing. Seeds were collected in July 1997 from a native habitat near Braunlage (see experiment 1) and were sown into garden soil at the beginning of August. Some days after germination three seedlings of each species were selected at random and transplanted with nearly equal spacing into 10-cm plastic pots filled with garden soil up to 2 cm below the upper edge of the pots. At this time all plants had two cotyledons and one (Solidago, Hieracium) or two (the others) primary foliage leaves. On 27 August, two individuals of D. agreste were released in each pot, which was then covered by a perforated plastic film and kept as in experiment 1. Before the test the slugs had been fed on plants from their native habitat (Aegopodium podagraria, Convolvulus sepium) rather than any of the test species.

From 28 August to 2 September, the leaf area eaten was estimated daily for each species without distinguishing between cotyledons and primary foliage leaves, and assigned to one of five categories: undamaged, up to 25%, up to 50%, up to 75% and up to 100%. A leaf no longer found or detached but not eaten was regarded as completely consumed. Plants where only the hypocotyl remained uneaten or where the hypocotyl had been cut through were evaluated as killed.

To test whether the decreasing consumption by slugs observed during the course of the experiment had been caused by the lack of acceptable food or other factors, such as the slugs becoming less healthy, the remains of the Hieracium plants were removed and two new plants of this species added per pot on 1 September.

Eight simultaneous replicates were assessed.

Experiment 3: physical defence mechanisms

(a) Epidermal cell wall thickness

‘Hardness’, which influences acceptability, is at least partly determined by the thickness of cell walls. Cross-sections through three leaves of each of the species used in experiment 1 were examined under the microscope and the mean thickness of the outer wall of the upper epidermal cells determined from values measured at five randomly chosen positions on each leaf.

(b) Epidermal hairs

Most of the plant species included in experiment 1 have epidermal hairs on their leaves. Type and density were determined microscopically.

Twelve such species were selected (Table 1) to determine the effect of epidermal hairs on acceptability. In September 1997, two individuals of D. agreste, kept in Petri dishes as in experiment 1, were offered one normal disc and one disc from the same species where the hairs on the upper side had been removed using a razor blade. We used the small Deroceras sp. because it was expected to be influenced by hairiness to a greater extent than the larger Arionidae. The eaten leaf area was assessed in the same manner as in experiment 1 and the mean percentage of the total eaten leaf area represented by each treatment was calculated for the 14 replicates used for each plant species.

Experiment 4: chemical defence mechanisms

The chemical composition of a leaf of an undamaged plant may differ from that of a plant individual that has previously been damaged and this may influence acceptability. To simulate herbivory, one leaf of every third, fourth and fifth leaf pair was removed from each of 20 randomly chosen Arnica shoots at Braunlage on 21 May. The leaves of the sixth pair were collected from these plants on 16 June, together with those from other undamaged shoots. No morphological differences between them could be found microscopically. Individuals of A. lusitanicus, kept as in experiment 1, were offered four discs (each 25 mm in diameter), two from damaged and two from undamaged plants for about 45 h. The consumed leaf area was assessed as in experiment 1 in 20 replicates.

Statistical analysis

Statistics were performed using SAS 6.02 (SAS Institute 1987). Tests of departure from Gaussian distribution refer to Shapiro–Wilk (proc univariate). anova calculations involved single comparisons after Scheffé (proc glm). Tests of sum of ranks were performed according to Kruskal–Wallis (proc npar1way) after manual calculation of single comparisons after Schaich & Hamerle (Bortz et al. 1990). The sign test was performed after Bortz et al. (1990, pp. 256–257).


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Experiment 1

Each slug species tested displayed highly selective feeding behaviour (Table 2). Arion lusitanicus ate 29% of the available leaf area, corresponding to 4.8 mg dry weight of leaf material per g slug live weight per day, compared with 13% (5.8 mg g–1 day–1) for A. subfuscus and 14% (20.3 mg g–1 day–1) for D. agreste.

Table 2.  Consumption of the 21 plant species by three slug species (experiment 1). For species codes see Table 1. Values represent means of the parameters concerned. Consumption does not differ significantly (after Scheffé; P < 0.05) for plant species with the same letter
Arion lusitanicusArion subfuscusDeroceras agresteArion lusitanicusArion subfuscusDeroceras agreste
Percentage of total consumed leaf area (%)Percentage of total consumed dry weight (%)
Mean rank in three-way combinationNumber of leaf discs gnawed (out of 10)

Figure 1 shows the mean proportion contributed by each plant species to the total area of leaf consumed. The order of acceptability assessed on this basis differed only slightly between the slug species (Table 2), with Hieracium laevigatum, Phyteuma nigrum and Taraxacum officinale being the most palatable for all slug species, Arnica consistently ranking fourth and Alchemilla monticola, Poa chaixii, Geranium sylvaticum and Hypericum maculatum among the five least palatable. Our results showed that using various consumption parameters (leaf area, dry weight, mean rank in three-way combination, number of discs gnawed) leads to different rank orders of preference and numbers of significant differences (Table 2). Using the non-parametric Kruskal–Wallis test instead of the anova removed most of the significant differences between food plants other than between the extremes.


Figure 1. Mean proportion of the total leaf area consumed represented by each of 21 plant species in food choice experiments with three slug species and mature leaf discs (experiment 1). For species codes see Table 1. Error bars indicate standard deviation.

Download figure to PowerPoint

Experiment 2

Slugs normally remained dormant in the soil during the day but at night they created holes in both cotyledons and true leaves, excised leaves or ate the seedlings completely. No plant was felled. As with leaf discs, the slugs clearly showed selective behaviour when feeding on seedlings (Fig. 2). They particularly favoured Hieracium laevigatum, consuming all plants of this species within 2 days, while all specimens of Arnica montana and Leucanthemum vulgare had vanished after 3 days and all Silene dioica plants, except for one which remained nearly undamaged up to the last day, had been consumed after 4 days. One individual of Solidago virgaurea remained entirely undamaged. Rumex acetosa suffered the least damage, with eight of 24 plants remaining intact and none eaten completely.


Figure 2. Mean leaf area consumed by Deroceras agreste, calculated as percentage of the initial total plant leaf area (experiment 2). For species codes see Table 1. Consumption of plant species with the same letter does not differ significantly after Scheffé (P < 0.05).

Download figure to PowerPoint

Of the 16 replaced Hieracium plants, 15 had been consumed after 1 day. This was a higher proportion than the number of the primary plants attacked in the first day, possibly due to a slight starvation of the slugs after a shortage of acceptable food in the preceding 2 days, or to a lack of alternative food source. It indicates clearly that the slugs were still capable of exhibiting a high level of feeding activity.

Comparing these results with those for Deroceras in experiment 1, the rank order Hl > Am > Sd > Lv > Sv > Ra at day 1 (Fig. 2) agreed with that obtained by feeding with leaf discs, except for the preference of Solidago to Rumex and of Silene to Leucanthemum, which were not significantly different in either test.

Experiment 3a

There was considerable variation in all parameters possibly influencing acceptability, i.e. fresh and dry weight per unit area, leaf thickness and epidermal cell wall thickness (Table 3). The best relationship to acceptability of the 21 tested plant species was found for epidermal cell wall thickness, which was partially correlated with acceptability for all slug species, especially the smaller ones A. subfuscus and D. agreste (Fig. 3). The slugs obviously preferred leaves with thin epidermal cell walls, with values for the four most preferred plant species (see experiment 1) being below 3 µm.

Table 3.  Leaf fresh and dry weight (measured in experiment 1), thickness of leaf and of upper epidermal cell wall and hairiness (experiment 3) of 21 plant species. Weight values are means ± SD, n = 20. Values of leaf thickness, cell wall thickness and hairiness are means from five randomly chosen positions of three leaves. us = upper side, ls = lower side of leaf. Species are arranged according to their acceptability to Arion subfuscus in experiment 1
Plant speciesFresh weight per area (mg mm–2)Dry weight per area (mg mm–2)Leaf thickness (mm)Thickness of upper epidermal cell wall (µm)Hairs per area (number mm–2)
Phyteuma nigrum0.163 ± 0.0200.0287 ± 0.00220.1101.53
Taraxacum officinale0.362 ± 0.0240.0248 ± 0.00490.1122.81
Hieracium laevigatum0.440 ± 0.0740.0452 ± 0.00470.2112.04  < 1 (us + ls)
Arnica montana0.329 ± 0.0280.0472 ± 0.01000.3212.047.0 (us), < 1 (ls)
Trifolium pratense0.190 ± 0.0490.0385 ± 0.00680.1472.041.8 (us + ls)
Silene dioica0.338 ± 0.0220.0456 ± 0.00690.2281.283.5 (us), 8.8 (ls)
Centaurea jacea0.217 ± 0.0670.0389 ± 0.00680.2143.578.8 (us), 12.3 (ls)
Solidago virgaurea0.158 ± 0.0190.0531 ± 0.00330.1756.63
Rumex acetosa0.456 ± 0.0550.0428 ± 0.00760.1977.05
Leucanthemum vulgare0.314 ± 0.0170.0436 ± 0.00570.3197.65
Knautia arvensis0.258 ± 0.0440.0373 ± 0.00460.1291.2812.3 (us), 1.8 (ls)
Succisa pratensis0.296 ± 0.0190.0468 ± 0.00470.2174.08  < 1 (us + ls)
Calendula officinalis0.286 ± 0.1300.0338 ± 0.00380.2191.537.0 (us), 5.3 (ls)
Betonica officinalis0.265 ± 0.0550.0707 ± 0.01470.1496.128.8 (us + ls)
Polygonum bistorta0.174 ± 0.0150.0417 ± 0.00450.1382.5517.5 (ls)
Veronica chamaedrys0.352 ± 0.0230.0334 ± 0.00300.1116.125.3 (us + ls)
Poa chaixii0.268 ± 0.0410.0328 ± 0.00220.1114.59
Geranium sylvaticum0.189 ± 0.0210.0413 ± 0.00240.1461.798.7 (us + ls)
Plantago lanceolata0.258 ± 0.0240.0420 ± 0.00290.2124.55  < 1 (us + ls)
Alchemilla monticola0.319 ± 0.0530.0440 ± 0.00310.1611.288.8 (us), 36.7 (ls)
Hypericum maculatum0.385 ± 0.0330.0385 ± 0.00370.1793.06

Figure 3. Mean proportion of the total consumed leaf area of 21 plant species (from experiment 1) compared with the thickness of the outer cell wall of the upper epidermis (experiment 3a). Trend lines with equations show linear regression for each slug species.

Download figure to PowerPoint

Experiment 3b

Most of the plants tested have only simple trichomes consisting of a single cell or a cell chain on both the upper and lower side of leaf (Table 3); only Arnica and Calendula also have glandular hairs. The effect of shaving on D. agreste feeding differed between plant species (Fig. 4); while shaved leaves of Arnica montana and some other species were preferred, only Centaurea jacea indicated a significant difference between the treatments using the Kruskal–Wallis test and the sign test. Surprisingly, for some other species the normal (hairy) discs were slightly preferred.


Figure 4. Mean proportion of total leaf area consumed represented by normal (hairy) and shaved leaf discs (experiment 3b). If nothing was eaten, both hairy and shaved forms got a proportion of 0%. For species codes see Table 1. Error bars indicate standard deviation. The asterisk (*) indicates significant differences between consumption of the treatments after the anova and the Kruskal–Wallis test (P < 0.01) and the sign test (in 10 of the 14 comparisons slugs preferred shaved leaf discs).

Download figure to PowerPoint

Experiment 4

Arion lusitanicus preferred leaf discs of undamaged Arnica plants. The amount of leaf area eaten differed significantly between the two treatments after the sign test: 16 out of 20 slugs ate larger quantities of the undamaged plants. On average, undamaged plants contributed 63% of the total leaf area consumed by each slug.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Rank order differences between slug species which differ significantly in size and habitat demands were remarkably low. Jennings & Barkham (1975) and Rathke (1985) also found a great similarity in the feeding responses to plants when the slug D. reticulatum was compared with several Arionidae species. Extreme differences in palatability of particular plant species can be deduced by comparison of several studies (see compilations in Dirzo 1980 and Briner & Frank 1998) but these are probably due to varying feeding conditions and point to the need to establish a standard method.

As a consequence of this ‘context specificity of preference’ (Peterson & Renaud 1989), we must agree with Rathke (1985) who suggested that ‘fine differences in values should not be overinterpreted’. As with our results (Table 2), Richardson & Whittaker (1982) found altered rank orders when consumption was measured as dry weight rather than leaf area eaten. In addition, these authors highlighted the crucial influence of which plant is selected as a reference material to be offered with each other food plant. Choosing a semi-acceptable plant as reference provides more significant differences between food species than choosing highly or poorly acceptable ones. We analysed our results in this way using species of high (Hieracium), medium (Succisa) or low (Poa) acceptability as ‘reference plant’. For each combination where a particular species occurred together with the reference plant (and any third species) in the eight replicates, we determined whether more of the test or reference species was consumed (+/– results, respectively). We too got different rank orders and different numbers of significant differences between species depending on the reference chosen. In agreement with Richardson & Whittaker (1982), we found the most differences when the semi-acceptable Succisa was used as the reference species (Table 4).

Table 4.  Rank orders of the 21 plant species in relation to a ‘reference species’. The values show the numbers of combinations of the eight where both were fed together (experiment 1), when a greater (+) or lesser (–) area of the tested species was consumed than of the ‘reference species’. Combinations where the two species were consumed in equal amounts were excluded. For species codes see Table 1
Reference species Hieracium laevigatum+/–Succisa pratensis  +/–Poa chaixii  +/–
To3/2Lv8/0Hl, To, Am, Lv, Sd, Co, Sp8/0
(Hl)0/0Am7/0Cj, Ka, Ra, Bo7/0
Sv2/5Hl, To7/1Pn, Tp, Vc, Hm6/0
Pn1/4Sd, Ka5/2Sv5/0
Am Lv, Co, Cj, Ka1/5 1/6Pn Ra5/3 4/2Pl Pb4/0 3/0
Sd, Sp1/7Cj4/3Gs1/0
Hm0/7Co,4/4Av (Pc)0/0
Tp, Vc, Ra, Bo, Pl, Pb, Gs, Av, Pc0/8(Sp) Bo Sv Tp, Vc Pl, Pb Av Hm, Gs, Pc0/0 3/5 2/5 2/6 1/7 0/7 0/8  

If each food plant is offered together with reference material the ratio of eaten test species to eaten reference species can be used as an ‘acceptability index’ or ‘palatability index’ (Grime et al. 1968; Dirzo 1980; Richardson & Whittaker 1982). These can be compared statistically by an anova because values of food plant consumption are then independent of each other. However, studies of this kind provide highly artificial results, which are not directly comparable to field conditions. It is more biologically meaningful to offer a range of different test species or ecotypes together, but then many species will remain statistically undifferentiated.

As a consequence, we used combinations of three of our 21 test species at random, in an attempt to approximate field conditions. Rollo (1988) observed the slug D. reticulatum to take two to four meals per night. Speiser & Rowell-Rahier (1993) found an average of one change of food species per day when investigating faeces of the snail Arianta arbustorum, and our preliminary experiments showed that two changes are not unusual.

Unfortunately, the statistical analysis of such experiments is very difficult. As the consumption of one species can be affected by one or both of the other two species in a given trial, the quantities consumed may be interdependent. Therefore, parametric analyses, such as anova, are not valid, although they have been used by several authors, e.g. Westerbergh & Nyberg (1995). Using non-parametric tests, such as the Kruskal–Wallis test, the number of significant differences in single comparisons between plant species is often reduced dramatically, and in this case only the highest ranks remain significantly different from the lowest.

Interestingly, the rank order of preference of D. agreste for seedlings differed only slightly from that obtained with leaf discs. In contrast, Wardle et al. (1998), assessing palatability of 20 herbaceous dicotyledons to D. reticulatum, found no correlation between seedling and leaf disc palatability. We used seedlings that were more than 3 weeks old when subjected to slug grazing and it is possible that younger seedlings would not differ in palatability between the species as Fenner (1987) and Hanley et al. (1995a) found seedling palatability to vary with age. Our results, unlike those of Hulme (1994), do not indicate any correlation between seedling size and mollusc behaviour. Two of the most preferred seedlings, Arnica and Leucanthemum, had the largest and smallest cotyledons, respectively, of the plant species tested.

The main factors influencing mollusc food choice therefore appear to be the same in seedlings and mature leaf discs. There are significant morphological differences between seedling cotyledons and mature foliage leaves, and we therefore suggest that seedlings and older leaves have similar chemical properties and that these are of primary importance for slug food selection. Nevertheless, physical factors may also influence food choice, although they are difficult to detect by comparing different plant species (Wardle et al. 1998). Westerbergh & Nyberg (1995) suggested that it is ‘impossible to evaluate the importance of hairiness on grazing unless the study includes distinct glabrous and hairy morphotypes of a single species’.

We used shaving to change a morphological trait of leaves of a single species. Cytoplasm may leak from the cut surface and this may increase the plant's palatability and attract molluscs, although similar effects are expected when cutting leaf discs and this does not appear to influence feeding. In addition, the method used cannot avoid injuring neighbouring epidermal cells and their walls, which may alter the hardness of the plant material, the glandular hairs might have a chemical effect after shaving, and sharp-edged cell wall residues may have a defensive effect. This approach therefore needs further analysis and improvement.

Similarly, the effect of hardness (experiment 3a) is almost certainly masked by several other plant traits. In Jennings & Barkham's (1975) experiments a behavioural preference for soft textured leaves was least visible in A. ater, the largest slug species studied. The largest of our tested species, A. lusitanicus, also showed the least pronounced preference for leaves with thin epidermal walls.

Although the results suggest that mechanical factors may act as feeding deterrents, the limited correlations indicate the importance of other, mainly chemical, properties.

No differences could be detected in the physical properties of Arnica leaves from predamaged plants that might explain their lower acceptability in comparison to undamaged plants. Chemical defence mechanisms are therefore highly probable, although we did not investigate their basis.

There are, however, also hints of chemical mechanisms that attract slugs to Arnica rather than deter them, perhaps by the same components. For each slug species Arnica ranks lower when number of leaves attacked is measured than when the percentage of eaten leaf area is considered. Once having started to gnaw on Arnica, the slug seems to be compelled to continue eating and, as a result, fewer partly consumed leaf discs remain, possibly because some chemical component causes a kind of ‘addiction’ to Arnica. As with humans eating potato crisps, bags, once opened, rarely remain half empty.

It was a primary intention of this study to extrapolate from the results of greenhouse experiments to slug–plant interactions under field conditions. Field observations of Arnica montana in meadows in the Harz mountains suggest that the risk of herbivory by slugs to whole plants growing in natural surroundings is particularly high adjacent to forest edges and in rainy years (Scheidel & Bruelheide 1999). In field experiments we have been able to show a dramatic increase in damage after releasing slugs into native Arnica populations and that damage by native slugs to transplanted Arnica decreases with increasing elevation in the Harz mountains (Bruelheide & Scheidel 1999). The results indicate that Arnica montana, the species with whose conservation we are concerned, is among the most palatable of those studied, both as leaf discs and seedlings.

In the case of Arnica montana, the results of food choice experiments therefore appear to match the field observations and field experiments. However, the results for some other species illustrate the pit-falls when laboratory results are extrapolated to field conditions. For example, although Hieracium laevigatum displayed the highest acceptability of all in food choice experiments, it was almost never eaten by slugs in the field and suffered almost no damage there. This may be because in the field attached leaves are not on contact with the ground and so are rarely encountered by slugs, which are further discouraged from climbing by the hairy stem. In contrast, in food choice experiments leaves are presented to the slugs in an easily accessible form.

Plants are most accessible at the seedling stage, when slug damage to plant individuals is also most effective. In our experiment, Deroceras individuals were able to kill several seedlings per day. This may be sufficient to prevent seedling establishment of preferred plant species completely, given the slug abundances of 50–60 individuals of Deroceras sp. per m2 reported from grassland (South 1965; Tischler 1980).

Indeed, Hanley et al. (1995b) were able to influence plant community dynamics in artificially created grassland gaps by molluscicide application. These authors suggested the existence of ‘a trade-off between fast growth rate and unacceptability to herbivores in determining competitive ability’ for their test species. Arnica montana, however, faces a particularly disadvantageous position: not only are its seedlings slow growing but they are also highly palatable.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank M. Runge, who provided all the facilities of his department, the staff of the New Botanical Garden for tolerating the ‘highly dangerous slug brute’ in their greenhouses, Anne Theenhaus for her valuable advice on food choice experiments, and D.F. Whybrew for polishing the English translation. We are grateful to L. Haddon, M. Fenner, and two anonymous referees for their very constructive criticism on an earlier version of the paper.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Bortz, J., Lienert, G.A., Boehnke, K. (1990) Verteilungsfreie Methoden in der Biostatistik. Springer, Berlin, Germany.
  • Briner, T. & Frank, T. (1998) The palatability of 78 wildflower strip plants to the slug Arion lusitanicus. Annals of Applied Biology, 133, 123 133.
  • Bruelheide, H. & Scheidel, U. (1999) Slug herbivory as a limiting factor for the geographical range of Arnica montana. Journal of Ecology, 87, 839 848.
  • Byers, R.A. & Bierlein, D.L. (1982) Feeding preference of three slug species in the laboratory. Melscheimer Entomological Series, 32, 5 11.
  • Cates, R.G. & Orians, G.H. (1975) Successional status and the palatability of plants to generalized herbivores. Ecology, 56, 410 418.
  • Cook, R.T., Bailey, S.E.R., McCrohan, C.R. (1996) Slug preferences for winter wheat cultivars and common agricultural weeds. Journal of Applied Ecology, 33, 866 872.
  • Dirzo, R. (1980) Experimental studies on slug–plant interactions. I. The acceptability of 30 plant species to the slug Agriolimax caruanae. Journal of Ecology, 68, 981 998.
  • Düll, R. & Kutzelnigg, H. (1992) Botanisch-ökologisches Exkursionstaschenbuch, 4th edn. Quelle & Meyer, Wiesbaden, Germany.
  • Ehrendorfer, F. (1973) Liste der Gefäßpflanzen Mitteleuropas, 2nd edn. Fischer, Stuttgart, Germany.
  • Fenner, M. (1987) Seedlings. New Phytologist, 106 (supplement), 35 47.
  • Frohne, D. & Jensen, U. (1992) Systematik des Pflanzenreichs, 3rd edn. Springer, Stuttgart, New York.
  • Gain, W.A. (1891) Notes on the food of some British molluscs. Journal of Conchology, 6, 349 360.
  • Glen, D.M., Jones, H., Fieldsend, J.K. (1990) Damage to oilseed rape seedlings by the field slug Deroceras reticulatum in relation to glucosinolate concentration. Annals of Applied Biology, 117, 197 207.
  • Grime, J.P., Blythe, G.M., Thornton, J.D. (1970) Food selection by the snail Cepaea nemoralis L. Animal Populations in Relation to Their Food Resources (ed. A.Watson). British Ecological Society Symposium 10. Blackwell Scientific Publications, Oxford, UK.
  • Grime, J.P., Cornelissen, J.H.C., Thompson, K., Hodgson, J.G. (1996) Evidence of a causal connection between anti-herbivore defence and the decomposition rate of leaves. Oikos, 77, 489 494.
  • Grime, J.P., MacPherson-Stewart, S.F., Dearman, R.S. (1968) An investigation of leaf palatability using the snail Cepaea nemoralis L. Journal of Ecology, 56, 405 420.
  • Hägele, B., Harmatha, J., Pavlik, M., Rowell-Rahier, M. (1996) Sesquiterpenes from the Senecioneae and their effect on food choice of the specialised leaf beetles Oreina cacaliae, Oreina speciosissima and the generalist snail Arianta arbustorum. Entomologia Experimentalis et Applicata, 80, 169 172.
  • Hanley, M.E., Fenner, M., Edwards, P.J. (1995a) The effect of seedling age on the likelihood of herbivory by the slug Deroceras reticulatum. Functional Ecology, 9, 754 759.
  • Hanley, M.E., Fenner, M., Edwards, P.J. (1995b) An experimental field study of the effects of mollusc grazing on seedling recruitment and survival in grassland. Journal of Ecology, 83, 621 627.
  • Hegi, G. (1987) Illustrierte Flora von Mitteleuropa, Vol. VI/4: Compositae II, 3rd edn. München, Hamburg, Germany.
  • Horrill, J.C. & Richards, A.J. (1986) Differential grazing by the mollusc Arion hortensis Fér. on cyanogenic and acyanogenic seedlings of the white clover, Trifolium repens L. Heredity, 56, 277 281.
  • Hulme, P.E. (1994) Seedling herbivory in grassland: relative impact of vertebrate and invertebrate herbivores. Journal of Ecology, 82, 873 880.
  • Jennings, T.J. & Barkham, J.P. (1975) Food of slugs in mixed deciduous woodland. Oikos, 26, 211 221.
  • Lawrey, J.D. (1983) Lichen herbivore preference: a test of two hypotheses. American Journal of Botany, 70, 1188 1194.
  • Pallant, D. (1972) The food of the grey field slug, Agriolimax reticulatus (Müller), on grassland. Journal of Animal Ecology, 41, 761 769.
  • Pennings, S.C., Carefoot, T.H., Siska, E.L., Chase, M.E., Page, T.A. (1998) Feeding preferences of a generalist salt-marsh crab: relative importance of multiple plant traits. Ecology, 79, 1968 1979.
  • Peterson, C.H. & Renaud, P.E. (1989) Analysis of feeding preference experiments. Oecologia, 80, 82 86.
  • Pullin, A.S. & Gilbert, J.E. (1989) The stinging nettle, Urtica dioica, increases trichome density after herbivore and mechanical damage. Oikos, 54, 275 280.
  • Rathke, B. (1985) Slugs as generalist herbivores: tests of three hypotheses on plant choices. Ecology, 66, 828 836.
  • Richardson, B. & Whittaker, J.B. (1982) The effect of varying the reference material on ranking of acceptability indices of plant species to a polyphagous herbivore, Agriolimax reticulatus. Oikos, 39, 237 240.
  • Rollo, C.D. (1988) The feeding of terrestrial slugs in relation to food characteristics, starvation, maturation and life history. Malacologia, 28, 29 39.
  • SAS Institute (1987) SAS/STAT Guide for Personal Computers, version 6. SAS Institute, Cary, NC.
  • Scheidel, U. & Bruelheide, H. (1999) The influence of slug herbivory on the large- and small-scale distribution patterns of Arnica montana. Verhandlungen der Gesellschaft für Ökologie, in press.
  • South, A. (1965) Biology and ecology of Agriolimax reticulatus (Müll.) and other slugs: spatial distribution. Journal of Animal Ecology, 34, 403 417.
  • Speiser, B., Harmatha, J., Rowell-Rahier, M. (1992) Effects of pyrrolizidine alkaloids and sesquiterpenes on snail feeding. Oecologia, 92, 257 265.
  • Speiser, B. & Rowell-Rahier, M. (1993) Does the land snail Arianta arbustorum prefer sequentially mixed over pure diets? Functional Ecology, 7, 403 410.
  • Tischler, W. (1980) Biologie der Kulturlandschaft. Fischer, Stuttgart, New York.
  • Van Dam, N.M. & Vrieling, K. (1994) Genetic variation in constitutive and inducible pyrrolizidine alkaloid levels in Cynoglossum officinale L. Oecologia, 99, 374 378.
  • Wardle, D.A., Barker, G.M., Bonner, K.I., Nicholson, K.S. (1998) Can comparative approaches based on plant ecophysiological traits predict the nature of biotic interactions and individual plant species effects in ecosystems? Journal of Ecology, 86, 405 420.
  • Westerbergh, A. & Nyberg, A.-B. (1995) Selective grazing of hairless Silene diocia plants by land gastropods. Oikos, 73, 289 298.

Received 9 July 1998revision accepted 18 March 1999