Tracking the movement of PIT-tagged terrestrial slugs (Arion vulgaris) in forest and garden habitats using mobile antennas

The use of different electronic transmitters has proven to be an important tool for studying the movement of individual free-ranging animals (Cooke et al., 2004; Thorstad et al., 2013). For example, large animals can be tagged with GPS, acoustic or radio transmitters (equipped with batteries) and tracked over large distances (Thorstad et al., 2013). Passive integrated transponders (PIT), on the other hand, are small (typically 7–32 mm), relatively cheap and do not contain internal batteries. Instead, an antenna creates an electromagnetic field that induces electricity in the tag, which then emits a unique signal that can be read by the same antenna, allowing tracking of individual animals. This technology has been used extensively to identify recaptured tagged animals, but also to track animal movements with stationary antennas placed at strategic locations (Gibbons & Andrews, 2004). Spatially confined animals or animals performing relatively short movements can also be tracked actively using mobile, portable antennas (Nzau Matondo et al., 2019; Watz et al., 2019). The movements and behaviours of a vast number of fishes, crustaceans, birds, reptiles and mammals have been studied using PIT telemetry (Gibbons & Andrews, 2004). For research on terrestrial slugs, PIT tags have been used only in a few studies. For example, Grimm & Paill (2001) captured, PIT tagged, released and recaptured Spanish slugs, Arion vulgaris Moquin-Tandon 1855 (sometimes, but wrongly, referred to as A. lusitanicus in the literature (Quinteiro et al., 2005; Zemanova et al., 2016)), to estimate their daily activity and home range, and Ryser et al. (2011) estimated two Arion slug survival rates in the field using traps and PIT tags. No study, however, has to date used mobile antennas to track slugs actively. Such studies have the potential to provide high-resolution information on slug movement, do not have to rely on trap placement and are independent of time of day (slugs are typically caught in traps when using them as daytime refuges). The Spanish slug (Arion vulgaris), which is believed to be native to the forested habitats of southwestern Europe, has spread to and become established in anthropogenic landscapes in many countries in the northern Europe during the last 50 years (Proschwitz, 1997; Kozłowski, 2007; Zemanova et al., 2016). It is considered to be highly invasive and a serious pest, both in agriculture and private gardens (Frank, 1998; Kozłowski, 2000). The species has an annual lifecycle, with egg-laying season normally commencing in late June, peaking in August or September and continuing throughout the autumn depending on temperature (Kozłowski, 2000). A typical Spanish slug lays between 200 and 400 eggs in batches underground in the soil or under vegetation or various objects (Kozłowski, 2000). During autumn, after egg laying, adults die (Kozłowski, 2000; Slotsbo et al., 2013). Eggs are not freeze tolerant, and in cold climates, only juveniles survive winter (Slotsbo et al., 2013). Although slugs are typically introduced via anthropogenic vectors, such as transported garden plants or soil (Proschwitz, 1997; Kozłowski, 2007), dispersal movements may expand local distribution and advance the establishment of the species (Grimm & Paill, 2001; Knop et al., 2013; Tamburello & Côté, 2015). Local habitat conditions, such as food supply and shelter, likely influence slug movement (Grimm & Paill, 2001). Throughout its lifecycle, Spanish slugs perform several local migrations. For example, adults need to move to suitable habitat for egg laying. After hatching, juveniles typically overwinter in plant litter or within the soil; they tend to migrate in relation to these habitats in response to freezing and warming events during winter (Kozłowski 2000; Slotsbo et al., 2011b; Slotsbo et al., 2013). Like all slug species, A. vulgaris is sensitive to dehydration and it has been reported that individual slugs home to humid resting places in response to dry weather, often showing a diurnal movement pattern where they hide during dry and warm days and move out to feeding areas during nights (Grimm et al., 2000; Slotsbo et al., 2011a). In the present study, we investigated the feasibility of actively tracking PIT-tagged Spanish slugs and compared their movement patterns in two different habitats. We recorded the positions of the slugs in autumn (during the egg-laying season) in a typical private garden and in a nearby deciduous forest during the day and at night. The study sites were located c. 200 m apart in the suburbs of Karlstad, Sweden, with the garden site being located at 59.362428°N, 13.454957°E and the forest site at 59.361104°N, 13.455959°E. The area of the garden was mainly covered by a lawn, an area with tall grass (mainly Agrostis capillaris, Convallaria majalis, Festuca rubra, Trifolium pratense and Vicia cracca), young trees (Betula pendula, Populus tremula, Quercus robur and Sorbus aucuparia) and ferns (Dryopteris filix-mas), some fruit trees (Prunus sp.) and a hedge (Syringa josikaea). The garden lawn was not mowed during the study, and total canopy cover was estimated as being c. 10%. At the forest site, the dominant trees were Acer platanoides, Betula pendula, Picea abies, Populus tremula and Prunus padus (total canopy cover = 60%); the ground was covered mainly by Fragaria vesca, Hylocomium splendens, Oxalis acetosella, Phleum pratense and Vaccinium myrtillus. All large slugs found at the two sites during the study period were visually identified as A. vulgaris. Tagged slugs recaptured alive at the end of the study were examined anatomically; the morphology of the lower genitalia confirmed that all sampled individuals had been correctly identified as A. vulgaris. During the night between 17 and 18 September 2018, we collected 64 A. vulgaris at the two study sites, 54 from the garden and 10 from the forest site. Slugs were sedated for 5–15 min using MS-222 (tricaine methanesulfonate) dissolved in water at a concentration of 300 mg l−1. We surgically implanted 12-mm PIT tags (Oregon RFID, Oregon, USA) by inserting the tag through a small incision made in the middle of the foot, following Grimm (1996). Three slugs did not survive the tagging procedure. On 19 September 2018, we released the 61 slugs in the same habitat from which they had been collected. These consisted of 51 slugs in the garden and 10 slugs in the forest. The slugs were released from buckets kept at a single point in each habitat. In total, three slugs in the garden and one slug in the forest did not leave the release bucket and were removed from the study. Hence, a total of 57 slugs were used in the study from the start (mean body mass ± SD = 6.9 ± 2.2 g). We tracked the slugs during five surveys between 21 September and 11 October 2018. Three surveys consisted of single-day tracking events, and two surveys consisted of three events in a consecutive day–night–day design. Daytime tracking was carried out between 09:00 and 10:30 and nighttime tracking between 21:00 and 22:00 local time. Each tracking event lasted for 1 h, during which we systematically searched the garden (800 m2 including immediate surroundings) and forest (1,200 m2) sites for PIT-tagged slugs, starting at the release point and searching outwards in a spiral pattern, using a handheld portable PIT antenna and a backpack reader (Oregon RFID, Oregon, USA). The antenna has a circular coil (50 cm diameter) with a ‘blind spot’ at its centre, which in some cases (depending on tag orientation) permits users to locate tags with precision down to 10 cm (Linnansaari et al., 2007; Watz et al., 2016). Occasionally the blind spot method could not be used (e.g. when slugs were deep underground or concealed under objects), resulting in a precision of c. 0.5 m. When a tagged slug was detected, we recorded its location. In the garden, the positions of individual slugs were estimated to the nearest 1-m orthogonal coordinate and recorded on a map. In the forest, we used circular coordinates; positions were assigned to a radius and an angle (i.e. measured relative to the north, using a compass), with the point of release as origin. Furthermore, we noted if an electronically detected slug was seen. Temperature was measured with automatic loggers every 30 min (HOBO Pendant, Onset Computer, Bourne, USA). Before every daytime tracking, the air humidity was measured (Velleman DEM500, Gavere, Belgium). Some slugs were never detected during the course of the study; many slugs presumably died as a natural consequence of cold autumn nights. We used an individual’s tracked movements to decide whether or not it was alive at the time of detection. This was done on the basis of two criteria: (1) slugs were assumed to be alive up to the time of its last detection and (2) if a slug was confirmed dead, we assumed that it was alive as long as it moved (inferred from changes in the detected position). Only position changes of more than 1 m were counted as movements. After release and initial dispersal, we used the position at first detection as the starting point for each individual slug. From the second detection onwards, we used changes in position and time between detections to calculate the distance moved per day. If a slug was detected more than once during a day–night–day survey, we only used the earliest detection when calculating its movements. For individuals that were detected and considered alive (according to the criteria specified above) in at least 4 of 11 tracking events, home range was calculated as the minimum convex polygonal area covering the detection positions (Grimm & Paill, 2001; Edworthy et al., 2012). To examine if the slugs in our study showed any homing behaviour (Grimm & Paill, 2001), we used positions of slugs that were detected at all three tracking events during a day–night–day survey and had moved ≥2 m between the first daytime and the nighttime tracking. Individuals that returned to their initial position (within 1 m) during the second daytime tracking were considered to show homing behaviour. Data on distance moved per day displayed a positive skew and were square root transformed to achieve normality. We tested for differences in movement between the two habitats using a linear mixed model, with individual slugs as subjects and surveys as a repeated measures variable. As an additional check, the linear mixed model-based analysis was also carried out using only slugs that were recaptured alive at the end of the study. To test if habitat and time of day affected the probability of visually observing an electronically detected slug, we used a generalized linear mixed model with a binomial response variable (0 = only electronically detected, 1 = visually observed and electronically detected). In both these analyses, the covariance matrix structures for the repeated measures within subjects were specified as compound symmetry. Home range areas were log10(x + 1) transformed to achieve normality, and differences in home range between habitats were examined using a Student’s t-test. All analyses were carried out using SPSS 25 (IBM, Armonk, USA). During the study, humidity ranged from 50.3 to 89.1% and temperature from 4.0 to 23.2°C; no major differences were observed between the forest and the garden. We recorded movement data from 38 slugs (i.e. individuals that were considered alive and detected at least twice after release), with a total of 86 movement data points being recorded. Slug movement ranged between 0 and 3.5 m per day with the median daily movement 0.7 m. Slugs in the forest (nslugs = 7, nmovements = 18) moved longer distances (median = 1.55, mean ± SE = 1.38 ± 0.20 m per day) than in the garden (nslugs = 31, nmovements = 68, median = 0.59, mean ± SE = 0.81 ± 0.09 m per day). This was reflected in a linear mixed model using square root transformed data (F1, 23.1 = 10.08, P = 0.004), which showed a significant effect of habitat (Fig. 1). At the end of the study, we recaptured nine living slugs. A similar analysis with just these nine slugs (Fig. 1) had only 27 movement data points and thus poor statistical power; although not statistically significant, the result from this analysis agreed with the results from the first analysis (F1, 4.0 = 5.54, P = 0.078). The difference in movement between the forest and garden habitats might reflect the landscape properties of the two habitats. We note, however, that the number of slugs was low and unequally distributed between the habitats. Moreover, only one site per habitat was used, leaving further exploration of habitat effects on slug movement to future research. Mean daily distance moved by individual Arion vulgaris (n = 38) in forest and garden habitats, as estimated by PIT telemetry. Empty circles represent slugs that were recaptured alive at the end of the study, and filled black circles represent slugs that were considered alive according to their detected movement. For slugs with more than one data point, error bars indicate ±1 SE. Mean home range for Arion vulgaris (n = 27) in forest and garden habitats estimated by PIT telemetry. Error bars indicate ±1 SE. During the study, 43 slugs were detected at least once, resulting in 182 electronic detections of living slugs (in most cases inferred from changes in position). In 21 of these detections (12%), we also visually observed the living slug. In the forest, a larger proportion of visually observed slugs were found at night rather than during the day (67 vs 6%). Contrastingly, in the garden, the corresponding nighttime and daytime value did not differ greatly (5 vs 10%). This interaction effect between daytime and nighttime and between forest and garden was revealed by a generalized linear mixed model with binomial response (forest vs garden: F1,178 = 4.13, P = 0.044; day vs night: F1,178 = 3.45, P = 0.065; forest vs garden × day vs night interaction: F = 8.669, df = 1;178, P = 0.004). These results underline the difficulty of using visual observations to estimate the abundance of slugs present in an area. Surface counting of slugs is generally considered problematic as it increases the risk of missing small individuals (Frank, 1998). The efficiency of hand searching, a form of visual surveying, has also been observed to be affected by habitat, weather and temperature (Johnston et al., 2017). Our results also suggest that the efficiency of visual observations may be highly time- and habitat‐specific, as demonstrated by the higher ratio of visible to detected slugs in the forest at night. This is possibly related to the difference in movement patterns. We detected 27 slugs (6 in the forest, 21 in the garden) on at least 4 of the 11 tracking events. Slugs had home range areas that ranged up to 111 m2 (forest: median = 20.0, mean ± SE = 41.2 ± 18.3 m2; garden: median = 2.5, mean ± SE = 11.3 ± 6.4 m2; Fig. 2). A Student’s t-test using log10(x + 1)-transformed data revealed that home ranges were significantly larger in the forest than in the garden (t25 = 2.47, P = 0.021). These home ranges are comparable to those previously reported for A. vulgaris. Using a trap-based tag-recapture approach, Grimm & Paill (2001) estimated an average home range of 45 m2 during the summer months. Eight slugs were detected during all three tracking events, during at least one of the two day–night–day surveys and moved ≥2 m between the first daytime and first nighttime detection. Two of these slugs were detected in all three tracking events during both day–night–day surveys, resulting in ten possible cases of observed homing. In only one of these cases had the slug returned to its initial position in the second daytime tracking. In the other nine cases, the slugs did not show a homing behaviour. Instead, they remained in the area to which they had travelled during the night, or they continued to move further away from their initial position. Approximately 35% of the released slugs could not be traced by end of the study. In our analysis, they were considered to be alive until their last detected movement. Possibly, they emigrated out of the tracked area, were predated, sought shelter deep underground, or died and became undetectable (i.e. the tag was buried under soil, litter and/or vegetation and was thus beyond the detectable range of the antenna). Arion vulgaris has been reported to be capable of moving 7 m within a 2-h period (Kozłowski, 2007). This means that some individuals may move a relatively long distance and, in the context of our study, suggests that some slugs may have moved a substantial distance beyond the study sites. In the tag-recapture study by Grimm & Paill (2001) mentioned above, a third of the tagged slugs were never recaptured and were considered to be cases of long distance dispersal; this observation agrees with our study. A relatively high proportion (46%) of the tagged slugs ceased moving and were considered or confirmed dead. Indeed, the study was conducted in autumn after the peak of the egg-laying season; most A. vulgaris die shortly after finishing egg laying (Kozłowski, 2000; Slotsbo et al., 2013). Hence, natural mortality is likely to be the main explanation for the relatively high frequency of ceased activity (mortality) for the tagged slugs. Substantial negative effects of tagging are deemed unlikely (six tagged slugs were held in a plastic container for over a week without showing any mortality), but should be explored further in the future. Mobile tracking of large terrestrial PIT-tagged slugs using portable and stationary antennas may prove an important tool to investigate dispersal, movements between feeding and sheltering habitats and potential homing behaviour. The method could also potentially be used to evaluate the efficiency of natural and engineered pest slug barriers as used in agriculture, horticulture and gardening. We thank N. Carlsson and S. Larsson for providing assistance in the field and are grateful to the inhabitants of Björkås for allowing us to track slugs in their gardens. Two anonymous reviewers made many useful suggestions. This research was funded by the Royal Swedish Academy of Agriculture (GFS2017-0181). The data that support the findings of this study are available from the corresponding author on request. The authors declare that there is no conflict of interests.