Experimental indications of gardeners’ anecdotes that snails interfere with invasive slugs

Mesocosm experiment

We carried out a three-factorial mesocosm experiment between March and May 2016. Factors included were snail/slug presence (3 levels: only H. pomatia, only A. vulgaris, both H. pomatia and A. vulgaris present), watering regime (2 levels: addition of 150 ml tap water every day, addition of 450 ml every third day) and earthworm presence (2 levels: addition of Lumbricus terrestris, no earthworms). Every combination of factors was replicated 5 times, resulting in totally 60 mesocosms in a Latin row-column design. A. vulgaris individuals were collected in gardens in Vienna in March 2016, H. pomatia individuals were obtained from a commercial breeder in Vienna (Gugumuck–Wiener Schneckenmanufaktur; https://gugumuck.com/). Since A. vulgaris usually lives for one year and dies after egg-laying in autumn, the slugs were still adolescent at the time of the experiment. H. pomatia, lives for several years and speciemens used in our experiment were adult.

The mesocosms (35.5 l volume, diameter 35.5 cm, depth 36 cm) were located in a greenhouse of the University of Natural Resources and Life Sciences Vienna in Groß Enzersdorf and were equally filled with substrate which consisted of a mixture of peat-free commercial plant substrate “Guter Grund” and topsoil from an arable field of the university’s research farm. Mesocosms were planted with 3 lettuce seedlings (Lactuca sativa var. capitate; mean weight 17.45 g) each in a consistent pattern on 23^rd of March; seedlings were obtained from a local gardening shop (Gartenbauer Auer, Vienna, Austria). For the mesocosms with earthworm presence as a factor, 2 individuals of adult L. terrestris purchased in a local fisherman shop (Anglertreff Thomas Lux, Vienna, Austria) were added after recording their initial weight (4.82 ± 1.16 g per mesocosm, mean ± SD). During a 27-day adaptation phase for lettuce seedlings and earthworms, all mesocosms were watered daily with 150 ml tap water each. After this establishment period, a stable isotope solution (¹⁵N) was applied once onto lettuce leaves using a fine brush following the method by Putz et al. (2011). Marking lettuce leaves with stable isotopes made it possible to assess herbivory on a species level, even in those mesocosms, that contained both H. pomatia and A. vulgaris. The isotopes are taken up by the gastropods together with the lettuce and accumulate in their bodies. The isotope concentrations in the gastropod bodies then gives an indication of how much lettuce was consumed by each specimen during the experiment. We used a 97 atom% ¹³C, 2 atom% ¹⁵N urea solution, which was produced by dissolving 100 mg 99 atom% ¹³C urea and 2 mg 98 atom% ¹⁵N urea (Sigma Aldrich, Vienna, Austria) in 50 ml distilled water. Furthermore, we added 12.5 µl wetting agent (Neo-Wett, Kwizda, Vienna, Austria) to ensure good contact of the labelling solution with the leaf surface. The urea solution was carefully applied with the fine brush on both sides of each leaf to avoid contamination of the soil within the mesocosms. Afterwards the different watering regimes were maintained (150 ml resp. 450 ml of water for the respective mesocosms). The water was applied using a long neck to avoid water spilling over the leaves and washing away the urea solution.

Subsequently, A. vulgaris and H. pomatia individuals were added to the mesocosms 28 days after planting (20^th of April); A. vulgaris-only mesocosms contained 5 individuals (1.03 g ± 0.16), H. pomatia-only mesocosms contained 4 individuals (20.06 g ± 1.86), and mesocosms with both species contained 3 A. vulgaris (0.85 g ± 0.17) and 2 H. pomatia individuals (19.05 g ± 1.92) each. Before introduction and after a 24-hour starvation period, all slugs and snails were weighed individually.

The experiment was conducted for 14 days. During this period, mesocosms were surveyed every day. In mesocosms where lettuce plants were consumed completely, single, similar-sized lettuce leaves were provided as food in order to prevent starvation of gastropods. These leaves where labelled with urea solution as well. A table with detailed information on the added leaves can be found in the Supplemental Material (Table S1). Soil temperature, soil humidity and soil electric conductivity were recorded during the experiment using a handheld TDR-system (TRIME -PICO 64/32, HD2-hand held device; IMKO Micromodultechnik, Ettlingen, Deutschland) on four dates (April 23^rd, 24^th, 25^th; May 2^nd 2016). For the analyses, TDR measurements per mesocosm were averaged.

After the experiment was terminated (May 3^rd, 2016), we estimated lettuce herbivory (%) visually, and remaining lettuce plants were harvested and dried at 60 °C for 24 h and weighed. Earthworm individuals were collected, counted and weighed. Arion and Helix individuals were collected, counted, weighed and subsequently frozen at −18 °C. All samples (i.e. all slug and snail individuals and a random selection of salad leaves from four mesocosms) were freeze-dried for 72 h and analysed for their stable isotope signature at the Federal Institute of Education and Research Francisco Josephinum, Wieselburg, Austria. The isotopic composition of C (δ¹³C) and N (δ¹⁵N) was measured using a Delta V Isotope Ratio Mass Spectrometer, connected via a ConFlo IV interface with an elemental analyser (Thermo Fisher Scientific, Waltham, MA, USA). Between 0.95 and 1 mg of the sample material was introduced into tin capsules for C- and N -isotope analysis. Stable isotope values were expressed applying the conventional δ-notation in parts per thousand (‰), relative to C and N reference materials, as the Vienna Pee Dee Belemnite (VPDB) and on atmospheric N₂ (AIR), respectively, as follows:

$$\delta \mathrm{X}‰=((\mathrm{R}\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}-\mathrm{R}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{r}\mathrm{d})/\mathrm{R}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{r}\mathrm{d})\ast 1,000$$

where X is ¹³C or ¹⁵N and R is the ratio of ¹³C/¹²C and ¹⁵N/¹⁴N in each case. Replicate measurements of certified as well as internal laboratory standards show that the measurement errors for both carbon and nitrogen isotope analyses were < ±0.2‰. Sample measurements were carried out at least in duplicates.

To account for the weight difference between A. vulgaris and H. pomatia individuals upon introduction, we calculated the percentage of herbivory per mesocosm by the total initial weight of all gastropods in each mesocosm and used this as a response variable.

Food choice experiment

We set up a food choice experiment to test whether Helix consumes Arion eggs. A total of 200 eggs were collected in private gardens in the Austrian provinces of Vorarlberg and Lower Austria. However, eggs from Lower Austria were only used in three repetitions; the other 37 repetitions with eggs as food choice contained eggs collected from Vorarlberg. Helix individuals were again obtained by the commercial breeder Gugumuck–Wiener Schneckenmanufaktur. In the experiment we either offered lettuce leaves (2 g of Lactuca sativa var. capitate), 5 Arion eggs or both leaves and eggs to one Helix individual (mean weight 22.14 g ± 5.28) and repeated each setting 20 times, resulting in 60 samples in total.

For acclimation and starvation, we kept H. pomatia individuals for 24 h on wet paper towels in a transparent plastic container (30 × 20 × 10 cm; L × W × H). After 24 h of starving the snails were weighed, the paper towels were removed, and we offered the different food choices. No soil was added to the container. The containers were kept in a climate chamber at 22 °C with a 12 h day/night rhythm. Additionally, 30 ml of water were added to each box. After 24 h, snails and remaining food were weighed and counted.

Statistical analyses

For the statistical analyses we used R (Version 3.5.1) and R Studio (Version 1.1.456) (R Core Team, 2018). We calculated a correlation matrix and the variance inflation factor to ensure explanatory variables were not highly correlated. We used Shapiro Wilk tests and additionally quantile-comparison-plots to check for normal distribution of data and residuals respectively. Homogeneity of variances was checked using the Fligner–Killeen test.

We ran a linear model including soil humidity, soil electric conductivity, soil temperature, water regime and earthworm presence as explanatory variables, added an interaction with the factor species for each variable and subsequently ran a Tukey post-hoc test using the glht-function of the R-package “multcomp” (Version 1.4-8). For analysing differences in intraspecific gastropod weight, we used a Kruskal Wallis test.

For weight difference between Arion and Helix-individuals, we ran ANOVAs after checking for normal distributions of weight differences and homogeneity of variances. To account for the different sizes of the two species, we estimated the percentage of leaves eaten in each mesocosm and divided this percentage by the initial gastropod body mass per mesocosm. Since this percentage of herbivory per gastropod body mass was not normally distributed, we conducted a generalized linear model with a Poisson distribution since the calculated percentages were very low and checked for a constant variance of the standardised residuals.

For the results of the stable isotope analysis, we performed an ANOVA with a Tukey post-hoc test for ¹⁵N and 13C to determine differences between the different levels of interaction. For ¹³C we had to log-transform the data first. Although ¹³C was still not normally distributed, the Fligner–Killeen test showed homogeneity of variances; therefore, we decided to proceed with an ANOVA. We used the same explanatory variables as in the previous analysis. To account for potential species-specific differences in the uptake of ¹³C and ¹⁵N, respectively, we used the mean ¹³C/¹²C and ¹⁵N/¹⁴N ratios of the salad-samples as a baseline. In a next step we calculated the difference between the isotope ratios of our salad samples and the ratios of the gastropods. In a final step we divided the ¹⁵N differences through the ¹³C differences for the gastropod samples for each mesocosm. This procedure allowed us to get one value for each gastropod sample that represents the isotope uptake for both ¹³C and ¹⁵N combined, which will be called “overall isotope uptake”, which is similar to what Aberle et al. (2005) did in their investigation.

Mesocosm experiment

Helix individuals weighed more than Arion individuals upon introduction regardless of the different treatment levels (mean weight per individual ± SD per mesocosm across all treatments: Arion 0.938 g ± 0.188; Helix 19.556 g ± 1.934). As a result, Helix performed significantly more overall herbivory compared to Arion (p = 0.001), but not compared to mesocosms where both species were present (p = 0.573). No effect of watering regime could be detected (p = 0.665).

Considering herbivory per bodymass, Arion-only significantly surpassed Helix-only (p < 0.001) and the Arion and Helix mixed mesocosms (p < 0.01) in the performance of lettuce herbivory, irrespective of the watering regime applied (Fig. 1). However, the percentage of herbivory per body mass was significantly different in mesocosms containing both Arion and Helix compared to Arion-only mesocosms when watered every three days (p = 0.022).

When analysing environmental factors in the mesocosm experiment (earthworm presence, soil temperature, soil electrical conductivity and soil humidity) in interaction with the respective species we could not detect significant differences. Therefore, the following graphs show results without explicitly presenting these environmental factors as well. However, soil humidity was significantly influenced by watering regime (p = 0.022), resulting in a higher soil humidity in mesocosms with daily watering.

Helix specimens in average lost weight during the course of the experiment while Arion specimens gained weight (mean weight difference per individual ± SD: Helix -0.44 g ± 0.95, Arion +0.65 g ± 0.33). Weight gain in Arion was higher (p = 0.001) and weight loss in Helix (p = 0.010) was lower under daily irrigation than under irrigation every three days.

Overall, no interaction between Arion and Helix in regard to weight difference could be detected. Individuals of both species showed no significant difference in intraspecific weight at the end of the experiment no matter if they were combined with the other species or not (Arion: p = 0.82; Helix: p = 0.91). However, initial weight in both species was significantly lower (Arion: p = 0.002; Helix: p = 0.04) at the beginning of the experiment between mesocosms with both gastropods present and mesocosms with either Arion or Helix present.

The mean isotope ratios for ¹⁵N and ¹³C for Arion, Helix and the salad samples are presented in Table 1.

The analysis of the ¹⁵N stable isotope concentrations revealed a significant influence of watering regime (p = 0.04), the percentage of leaves eaten in a mesocosm (p = 0.010) and of an interaction between soil temperature and the presence of both gastropod species in a mesocosm (p = 0.010; see Table S2). In contrast, the analyses of the ¹³C-ratios showed significant influences of species (p < 0.01), weight difference of the molluscs between the beginning and the end of the experiment (p < 0.01) and soil electric conductivity (p = 0.011; see Table S3). After calculating the overall isotope uptake to get a more complete picture about isotope uptake by gastropods in general, we conducted an ANOVA and found a significant difference in gastropod species (p < 0.01), but no influence of co-occurrence or watering regime (Fig. 2).

Additionally, we found a significant influence of mollusc weight between the beginning and the end of the experiment and the percentage of leaves eaten in a mesocosm (Table 2).

Food choice experiment

We found very little egg predation of Arion eggs by Helix. Helix had been offered five eggs each in 60 repetitions, but only a total of three eggs were eaten in three different containers. In all three cases eggs had only been eaten when lettuce was offered to Helix as well. The lettuce was eaten completely in all 40 repetitions containing lettuce. No significant influence of food choice options on Helix weight differences between the beginning and the end of the experiment could be detected. We conducted no further statistical analysis due to insufficient data.