Response behaviour of native lizards and invading wall lizard to interspecific scent: implications for invasion success

The human-assisted movement of species beyond their native range facilitates novel interactions between invaders and native species that can determine whether an introduced species becomes invasive and the nature of any consequences for native communities. Avoiding costly interactions through recognition and avoidance can be compromised by the naivety of native species to novel invaders and vice versa. We tested this hypothesis using the common wall lizard, Podarcis muralis, and the native lizard species with which it may now interact in Britain (common lizard, Zootoca vivipara, sand lizard, Lacerta agilis) and on Vancouver Island (northern alligator lizard, Elgaria coerulea) by exploring species' responses (tongue flicks, avoidance behaviour) to heterospecific scent cues in controlled experiments. The tongue flick response of P. muralis depended on the different species’ scent, with significantly more tongue flicks directed to E. coerulea scent than the other species and the control. This recognition did not result in any other behavioural response in P. muralis (i.e. attraction, aggression, avoidance). Lacerta agilis showed a strong recognition response to P. muralis scent, with more tongue flicks occurring close to the treatment stimuli than the control and aggressive behaviour directed towards the scent source. Conversely, Z. vivipara spent less time near P. muralis scent cues than the control but its tongue flick rate was higher towards this scent in this reduced time, consistent with an avoidance response. There was no evidence of E. coerulea recognition of P. muralis scent in terms of tongue flicks or time spent near the stimuli, although the native species did show a preference for P. muralis-scented refuges. Our results suggest a variable response of native species to the scent of P. muralis, from an avoidance response by Z. vivipara that mirrors patterns of exclusion observed in the field to direct aggression observed in L. agilis and an ambiguous reaction from E. coerulea which may reflect a diminished response to a cue with a low associated cost. These results have significant implications for the invasive success and potential impacts of introduced P. muralis populations on native lizards.

Competition within and among species arises through overlap in utilization of limited resources, and plays an important role in determining species' distributions and abundance (Case & Gilpin, 1974;Schoener, 1983). The outcomes of conflict arising from this competition are often asymmetrical, commonly driven by factors such as contestant body size, residency and prior experience (Chen & Hsu, 2016;Chock, Shier, & Grether, 2018;Schoener, 1974). In time, contests (both direct behavioural interference and exploitative competition for limited resources) can lead to niche segregation, character displacement and exclusion of inferior competitors from optimal habitat (Heltai, Saly, Kovacs, & Kiss, 2015;Losos, 2000;Peiman & Robinson, 2010). The human-assisted movement of species beyond their native range can force novel competitive (and predation/prey) interactions between invaders and native species. Furthermore, behavioural traits indicative of successful invaders (i.e. high levels of boldness and aggression) can give a competitive advantage in such encounters (Chapple, Simmonds, & Wong, 2012;Damas-Moreira, Riley, Harris, & Whiting, 2019;Downes & Bauwens, 2002), with the potential to cause niche shifts and drastic declines or local extinctions of native species (Brzezinski, Chibowski, Gornia, Gorecki, & Zalewski, 2018;Cadi & Joly, 2003;Doody et al., 2009;Dorrestein, Todd, Westcott, Martin, & Welbergen, 2019;Hernandez-Brito, Carrete, Ibanez, Juste, & Tella, 2018). Understanding how species may interact when faced with novel competitors is therefore an important part of assessing the invasion potential of non-native species introductions and their impacts on native communities.
Conflict between species is a hierarchical process, beginning with contact and ending with physical interactions. However, mediating these physical interactions is a variety of behavioural decisions that can influence the severity of the outcome for one or both organisms (Langkilde, Lance, & Shine, 2005). Individuals might detect but choose not to interact with one another, such as if one individual perceives the other to be dominant (Brazill-Boast, 2013). Individuals might also engage in ritualized display behaviours that reduce the need for physical altercation by giving further information about the likely outcome (Baeckens, Driessens, Huyghe, Vanhooydonck, & Van Damme, 2018;Edwards & Lailvaux, 2012;Reichert & Gerhardt, 2014). When native species encounter novel non-native species, this system of recognition might be compromised by the lack of evolutionary history between the two taxa. Without such recognition, naïve/native species' responses may be suboptimal during encounters that leave them particularly vulnerable to pressures from introduced species (prey naïvety hypothesis; Sih et al., 2010;Ehlman, Trimmer, & Sih, 2019). Avoidance of costly encounters therefore requires accurate recognition of potential threats through sensory discrimination, followed by an appropriately gauged response that weights the relative costs of the threat. As stated by the threat sensitivity hypothesis, individuals should respond more strongly to cues associated with higher risks/costs (e.g. trade-off between avoidance of threat and reduced foraging time), but less strongly to cues with lower associated threat (Amo, Lopez, & Martin, 2005, 2007Cisterne, Vanderduys, Pike, & Schwarzkopf, 2014;Payne, Tillberg, & Suarez, 2004).
Chemosensory cues are an important source of information on which to base judgement of likely costs of encounters and a suitable response. They can reliably allow forewarning of the immediate or recent presence of predators, and in certain circumstances they may be the only cues available (Kats & Dill, 1998). Indeed, the majority of examples of behavioural response (or lack thereof) of natives to chemical cues of novel species come from predatoreprey systems (Cisterne et al., 2014;Hoffmann, McGarrity, & Johnson, 2018;Stanbury & Briskie, 2015). For example, the foraging behaviour of two native Australian lizards was compromised when individuals were exposed to scents of both native and invasive mammalian predators (Webster et al., 2018), suggesting prey naïvety is not the rule in native/non-native systems. The reverse situation, of invasive species' response to cues from novel predators, has also received attention, as in the avoidance response of the Asian house gecko, Hemidactylus frenatus, to some native predatory snake cues (Cornelis, Nordberg, & Schwarzkopf, 2019). There is less known, however, about behavioural responses to novel scent cues outside of predatoreprey systems. It is reasonable to suspect that the presence of an introduced competitor species might have an effect similar to that of a novel predator, and therefore native species might learn to avoid cues from invasive species if these cues were previously associated with an encounter that incurred a cost (Ferrari, Crane, Brown, & Chivers, 2015). Examples come from Spanish terrapins', Mauremys leprosa, avoidance of water pools when chemical stimuli of the invasive red-eared slider, Trichemys scripta, is detected (Polo-Cavia, Lopez, & Martin, 2009), and honey bee, Apis mellifera, avoidance of flowers occupied by invasive Argentine ants, Linepithema humile (Sidhu & Rankin, 2016). Responses may, however, be more ambiguous, such as the preference of both the endemic Barbados leaf-toed gecko, Phyllodactylus pulcher, and an invasive house gecko, Hemidactylus mabouia, for refuges conditioned with the scent of the other species (Williams, Pernetta, & Horrocks, 2016).
With a view to exploring the possible indirect competitive interaction between P. muralis and native lizards, our objective in this study was to examine experimentally the scent recognition and behavioural response of non-native P. muralis individuals to scent cues of native lizard species within its introduced range in Britain (L. agilis, Z. vivipara) and on Vancouver Island (E. coerulea), and vice versa. Based on the naïvety and threat sensitivity hypotheses, we predicted differences in behavioural responses dependent on the associated cost incurred through interspecific interactions and/or the degree to which chemical cues used for interspecific communication might be similar among phylogenetically related species. In the context of our model system, we therefore predicted the taxonomic distance separating P. muralis and native E. coerulea, combined with the recent shift into sympatry (i.e. P. muralis was introduced in 1967, but recent rapid range expansion has increased potential for contact with E. coerulea), would lead to no scent recognition, whereas, despite being a relatively recent introduction to Britain (Foster, 2015), the closer phylogenetic relatedness and substantial sympatry in continental Europe between P. muralis and the lacertids L. agilis and Z. vivipara (Sillero et al., 2014) would produce differences in behavioural response to scent cues.

Animal Collection, Husbandry and Welfare
All wild-caught animals (see below for species' sampling locations and methods) were sexed based on the presence of hemipenes in males (Schulte, 2008), health screened and checked for external parasites before being taken into captivity.
All lizards were transported to the study facilities in plastic vivaria (20 Â 12 cm and 16 cm high) with natural substrate and refuge objects obtained at the capture site. Lizard species were housed separately in large plastic vivaria (70 Â 30 and 50 cm high), with water supplied ad libitum and provision for basking, a thermal gradient (18e28 C) and shelter. Live food was offered every other day in the form of third-instar crickets, mealworms and waxworms. Light and heat were provided by incandescent (40 W) bulbs placed above each vivarium to provide a 14:10 h light:dark cycle; vivaria were moved outside during the day if weather conditions allowed. All lizards were marked dorsally with an identifying number in nontoxic marker and were given a 5-day acclimatization period to allow habituation to the general disturbances and handling prior to the experimental trials beginning.

Podarcis muralis
All P. muralis were caught by hand or noosing. Twenty-one adult males (snoutevent length >45 mm) with origins in the Emilian Apennines, Italy (Deichsel & Schweiger, 2004) were collected from the introduced population around the Prospect lake area (48.30 N, -123.25 E) and Fairfield district of Victoria, BC (48.24 N,.20 E). Nineteen adult male P. muralis of Venetian origin (Michaelides et al., 2015) were collected from an introduced population at West Worthing, Sussex, U.K. (50.48 N, 0.22 W). All P. muralis were euthanized as per approved ethical guidelines (see Ethical Note) following behavioural testing and retained for future research.

Zootoca vivipara
All Z. vivipara (11 females, seven males) were caught as part of active mitigation translocations at two sites: High Wycombe, Buckinghamshire, U.K. (51.61 N,e0.71 E) and West Malling, Kent, U.K. (51.28 N, 0.32 E) between 1 and 5 August 2018. Podarcis muralis is absent at both sites. Individuals were caught by hand under artificial refuges. These lizards were retained in captivity for 10 days for inclusion in the scent recognition experiments and subsequently released to the respective translocation receptor sites.

Lacerta agilis argus
Owing to the conservation status of L. agilis in Britain and necessary restriction on use of wild-caught animals, we used its most closely related subspecies L. a. argus (Andres, Franke, Bleidorn, Bernhard, & Schlegel, 2014). Five juveniles (three females, two males, born in August 2017) were acquired from captive stock in March 2018. These individuals were reared as a group in captivity and had reached adult size by August 2018. They were retained in a private collection after this study.

Elgaria coerulea
Elgaria coerulea (10 females, seven males) were collected, either by hand under natural refuges or by noose, from Kingzett Lake quarry (48.67 N,.34 E) on Vancouver Island in early July 2018. These lizards were retained in captivity for 10 days for inclusion in the scent recognition experiments after which they were returned to the point of capture. Podarcis muralis was absent at both sites in 2018.

Scent Recognition Experimental Procedure
The methodology for experimental trials of scent recognition was adapted from several chemosensory studies involving Podarcis spp. (Barbosa et al., 2005;Bertram, 2004;Font et al., 2012) and from pilot trials conducted in June 2017. Experimental trials were conducted on Vancouver Island, BC, between 10 and 20 July 2018, and in England between 15 and 21 August 2018. Trials were conducted between 0900 and 1700 h to coincide with the lizards' period of daily activity. The experimental enclosure was a clear plastic storage container (70 Â 30 cm and 15 cm high) with opaque back and sides. Two textured, nonabsorbent, washable liners were used as floor coverings which were alternated between trials. The centre line of the enclosure was marked on each liner to delineate treatment halves for observation and analysis. Two small refuges (10 Â 10 cm and 2 cm high) with a single entrance (3 cm long, 1 cm high) were created using slate and plastic building blocks and were placed against the side wall of each end of the enclosure (Fig. 1). A 60 W spot bulb was suspended directly overhead the experimental enclosure casting uniform heat (18 C) and light throughout.
Treatment was randomly assigned to each half of the enclosure before each trial using a random number generator. Four swabs were placed in each half of the arena: one in each corner of the arena, one at the entrance to, and one on top of, each refuge (Fig. 1). For the control treatment, swabs were dipped in deionized water. We did not use a pungency control because in many previous studies, including those specifically dealing with P. muralis and E. coerulea, it has already been well established that these lizards have highly developed olfaction and can discriminate the scent of congeners, predators and prey from biologically irrelevant scents (Cooper, 1990;Cooper & Perez-Mellado, 2002;Gabirot, Castilla, Lopez, & Martin, 2010). Scent treatment was obtained according to established protocol from similar studies by first dipping swabs in deionized water and then gently rubbing the swab over the body of the scent donor making sure to swab femoral pores and cloacal regions, since these are the body areas most frequently and intensely investigated by tongue flicking during social encounters (Cooper & Perez-Mellado, 2002;Lopez, Martin, & Cuadrado, 2002). Scent donors were always males randomly selected from the relevant test population.
Test subjects were introduced to the experimental enclosure underneath a transparent container (15 Â 10 cm and 10 cm high) placed on the central line of the enclosure. Once the lizard showed relaxed movement behaviour the container was gently removed, Figure 1. Diagram of the enclosure used for controlled experiments of scent recognition between Podarcis muralis, Zootoca vivipara, Lacerta a. argus and Elgaria coerulea. Scent/control swabs are depicted in the four corners of the enclosure and at the entrance to, and on top of, the refuges. and on the first tongue flick from the subject a 10 min timer was started on the video camera recording the trial. Subsequent tongue flicks were tallied according to the treatment side in which they occurred. After the 10 min trial, test subjects were returned to the housing vivaria and were only used in one trial a day. Fresh swabs were used for each trial, and the liner, experimental enclosure and refuges thoroughly washed with warm water and wiped with alcohol to eliminate residual chemical traces before air drying prior to next use. Only British P. muralis took part in multiple trials (i.e. each was tested against Z. vivipara and L. a. argus). All other lizards were only involved in one trial, with no replicates, to avoid habituation to scent cues.
Cowlog 2.0 software was used to retrospectively analyse video recordings and quantify time spent between the enclosure halves (Hanninen & Pastell, 2009). We limited behaviour classifications to either the time spent in each half or the time spent exhibiting escape behaviour in each half of the enclosure. We defined escape behaviour as time spent standing in an upright position against the wall of the enclosure performing scratching movements with the forelegs. During escape behaviour the lizards were not engaged in tongue flicking or assessing their surroundings. The duration of escape behaviour in each half was thus subtracted from the total time spent in the half enabling quantification of only exploratory or stationary behaviour. Where variation in the time spent between treatment halves was observed we tested the rate of tongue flicks occurring in each half. We also recorded the number of times individuals fully entered (entire body under a refuge) either control or treatment refuges.

Ethical Note
Capture, husbandry, humane euthanasia of P. muralis and experimental procedure were carried out under licence from Natural England 2016-21938-SPM-NNR and approved by the ethics committee of BC Ministry of Forests, Lands and Natural Resource Operations and Rural Development (NA18-288615). All P. muralis were euthanized (by pithing and decapitation) following anaesthesia with 25% benzocaine gel via oral administration and stored in 90% ethanol after the experimental trials for further study. Currently accepted best practice for euthanasia of small lizards involves either straight blunt force trauma to the cranium, decapitation and/or pithing without prior anaesthesia (Leary, 2013). Recent discussions within the research community have recommended the efficiency of benzocaine as a general anaesthetic for small reptiles and amphibians. As such, its use prior to existing methods exceeds protocols previously considered humane for dispatching small lizards.

Data Analysis
All analyses were performed in R version 3.4.2 (R Core Team, 2017).

Podarcis muralis response to scent of native species
We tested two key hypotheses concerning the response of P. muralis to native species. First, we tested whether there was a difference in the response between the three native species using generalized linear models (GLM) with binomial errors. Response variables were the number of tongue flicks and the time spent exploring in treatment and control halves. We used the D 2 value (percentage deviance explained) to evaluate model fit, calculated as 1-deviance/null deviance. Post hoc Tukey tests via the 'glht' function in the package 'multcomp' (Hothorn, Bretz, & Westfall, 2008) were then applied to explore pairwise differences between species. Second, we tested whether there was evidence of scent recognition or avoidance behaviour by P. muralis of each species individually. For each of the native species, we compared scent recognition (counts of tongue flicks in control versus treatment halves) and avoidance behaviour (time spent in each treatment half) by P. muralis using individual paired t tests. As all P. muralis individuals were male, sex was not included as a predictor in any models.
Native species' response to P. muralis For native species' responses, we first tested whether there was a difference in the responses by the different native species to P. muralis. Since both male and female animals were used, we ran individual GLMs with binomial errors to determine the effects of sex on the response variables (ratios of number of tongue flicks and time spent exploring in treatment and control halves). Second, we tested for evidence of scent recognition and avoidance behaviour towards P. muralis scent using paired t tests for each native species to compare behaviours in the control and treatment halves of the enclosure (as above).

P. muralis Response to Native Species
There was a significant difference in the response of P. muralis towards the three native species in terms of ratio of tongue flicks in each treatment half (Table 1).
Post hoc Tukey's HSD tests showed that P. muralis responded with more tongue flicks to the E. coerulea treatment compared to both Z. vivipara and L. a. agilis treatments. There was no difference in the P. muralis tongue flick response towards Z. vivipara and L. a. argus (z ¼ -1.18, P ¼ 0.46). Analysis of the P. muralis discriminatory response (tongue flicks) to control and treatment scents showed a significantly greater response to E. coerulea scent versus the control (paired t test: t 20 ¼ 2.63, P ¼ 0.01), no difference in response towards Z. vivipara scent versus the control (t 18 ¼ 0.14, P ¼ 0.88) and no difference between L. a. argus scent versus the control (t 14 ¼ 1.20, P ¼ 0.24; Fig. 2a).
There was also a significant difference in the response of P. muralis towards the other three species in terms of the ratio of time spent exploring each treatment half (Table 1). Post hoc Tukey's HSD tests showed that P. muralis spent significantly more time in the Z. vivipara treatment half of the arena than it did E. coerulea (z ¼ 10.65, P < 0.001) and L. a. argus (z ¼ -7.46, P < 0.001) treatments. There was no difference in the time spent by P. muralis near E. coerulea and L. a. argus treatments (z ¼ 2.28, P ¼ 0.05). Analysis of the aversion response (time spent near stimuli) of P. muralis to  Fig. 2b).

Native Species' Responses to P. muralis
Sex had a significant effect only on the time spent by L. a. argus and E. coerulea in the treatment halves (Table 2). However, owing to small sample sizes (E. coerulea male:female ¼ 7:10, L. a. argus 2:3, Z. vivipara 7:11), the pattern of response being similar between the sexes and our interest being primarily with species' response, we pooled the data across sexes for analysis.
Only L. a. argus showed a significant recognition response to P. muralis scent, with a mean ± SE ¼ 103 ± 28.0 tongue flicks in the treatment end of the enclosure with P. muralis scent and 42 ± 8.0 tongue flicks in the control end (t 4 ¼ 2.99, P ¼ 0.04). There was no significant difference in the mean number of tongue flicks between P. muralis scent and control ends of the enclosure for Z. vivipara (40.2 ± 6.1 scent versus 46.1 ± 8.6 control; t 17 ¼ -1.08, P ¼ 0.29) or E. coerulea (16.2 ± 2.4 scent versus 16.9 ± 3.9 control; t 16 ¼ -0.23, P ¼ 0.81; Fig. 3a).
Neither group showed a significant aversion/attraction response to P. muralis scent as indicated by the time spent in each treatment half. Although L. a. argus spent longer on average (66%) in the scented half than in the control (t 4 ¼ 1.93, P ¼ 0.12), that time can be attributed to four instances of direct attack (biting) of a scented swab. Conversely, Z. vivipara spent less time on average (38%) in the scent treatment half (t 17 ¼ -1.88, P ¼ 0.07). Despite less time being spent in the scent treatment half, the rate of tongue flicks by Z. vivipara was greater in this half (mean ± SE ¼ 0.24 ± 0.13) than in the control (0.16 ± 0.12; t 17 ¼ 2.10, P ¼ 0.05). The average time spent in each treatment half was relatively even for E. coerulea (55% scented; t 16 ¼ 1.22, P ¼ 0.23; Fig. 3b).

Refuge Use
Only E. coerulea used the scented refuge more often than the control refuge, whereas L. a. argus did not use either refuge. In all other experiments the control refuge received more visits than the scented refuge. The difference in visits between control and treatment refuge was greatest in the response of Z. vivipara to P. muralis, with four and nine visits, respectively (Fig. 4).

DISCUSSION
The results of this study highlight the potential for varied recognition and behavioural responses towards chemical cues within a native/non-native species model. In accordance with the study hypothesis, the results provide evidence of differing responses with taxonomic distance that are demonstrative of naïvety to scent cues of novel competitors and threat sensitivity between more closely related species.

P. muralis/E. coerulea Interaction
The responses of P. muralis and E. coerulea to each other's scent in our experiments complement the results of the only other behavioural study on interaction between these two species (Bertram, 2004). The previous study also found P. muralis scent to have no detectable effect on the behaviour of E. coerulea. The lack of discriminatory response of E. coerulea in the two studies suggests that the species is either unable to detect the odour of P. muralis, despite the highly developed chemosensory abilities of the species (Cooper, 1990), or does not respond behaviourally to the stimulus. Possible explanations for the observed lack of E. coerulea response is that, besides four snake species, there are no other terrestrial native reptile species occurring in sympatry with E. coerulea on  Vancouver Island (Gregory & Campbell, 1984;Matsuda, Green, & Gregory, 2006), and therefore a complete naïvety of E. coerulea to scent of a phylogenetically distant lacertid lizard species is not unexpected. Similarly, a response is not to be expected if chemoreception is limited to foraging and intraspecific communication in this species. Our observations of E. coerulea readily using scented P. muralis refuges, seemingly in preference to unscented ones, do, however, warrant further investigation into the species' ability to recognize the scent of P. muralis, and support the hypothesis that the physical presence of P. muralis, not its scent alone, may deter E. coerulea from cohabiting refuges with the non-native (Bertram, 2004). If chemosensory recognition were influencing selection for the P. muralis-scented refuges, this could indicate that in the absence of visual cues and any previous negative experience, the scented refuge was perceived to be safe in an otherwise novel and unfamiliar environment. A similar outcome has been observed in refuge selection trials involving a native and an invasive gecko, where both species preferred a refuge previously occupied by the other (Williams et al., 2016). Identification with common compounds in species' scents may have driven refuge choice in such cases (Garcia-Roa, Cabido, Lopez, & Martin, 2016;Mason & Parker, 2010), or there may be a benefit in using heterospecific cues as a measure of refuge quality (Parejo, Danchin, & Avil es, 2004). Naïvety was not symmetrical in this pairing, however, and the discrimination and heightened response of P. muralis towards the scent of E. coerulea, above that shown towards the scent of the other species in our experiment, is unexpected considering the taxonomic distance between the two. There are numerous examples across lizard species, including Podarcis spp., of an ability to discriminate between closely related species based on chemical cues alone (Barbosa et al., 2006;Cooper & Perez-Mellado, 2002;Cooper & Vitt, 1986;Gabirot et al., 2010;Labra, 2011). There is, however, little evidence for scent recognition of more taxonomically distant species outside predator/prey systems (e.g. snake predator/lizard prey interactions; Amo, Lopez, & Martin, 2004;Cabido, Gonzalo, Galan, Martin, & Lopez, 2004;Labra & Hoare, 2015;Zagar, Bitenc, Vrezec, & Carretero, 2015;Ortega et al., 2018). For example, Blanus cinereus, a fossorial amphisbaenian, reacted strongly to scent stimuli of sympatric snake and centipede predators, yet showed no difference in reaction towards a water control and an innocuous, sympatric skink (Lopez & Martin, 2001). Male Podarcis hispanica are capable of discriminating conspecific scent from that of Psammodromus algirus; however, no variation in tongue flick response between an odourless control and P. algirus scent suggests a lack of behavioural response towards, or inability to detect, the latter (Gomez, Font, & Desfilis, 1993). Regardless of the context in which P. muralis explored the scent of E. coerulea (i.e. inquisitiveness towards a novel or biologically relevant scent) the fact that this discrimination of scent did not elicit a behavioural response in P. muralis (e.g. attraction, avoidance or aggression) suggests that the stimuli (alone) may have been regarded as benign. A lack of behavioural response having distinguished the odour is to be expected if fitness costs associated with avoidance behaviour outweigh those of any naturally occurring direct interaction, such as limited agonistic behaviour between the two species (Langkilde et al., 2005). This response of P. muralis can also be interpreted as a heightened boldness and willingness to explore unfamiliar stimuli (neophilia), coupled with high levels of behavioural plasticity to mediate the response. Such facilitative traits may be common among species that become invasive (Damas-Moreira et al., 2019) and may be fundamental to the expansion of the species' range on an intercontinental scale. These results and reasoning are in keeping with observations of a greater propensity for P. muralis to make the first approach in controlled direct encounters with E. coerulea, but ultimate lack of aggression arising between the two (Bertram, 2004).  Podarcis muralis (PM) individuals in the laboratory. Initials beneath bars denote species pairings, such that the first of each pair is the focal species and the second is the scent treatment (e.g. PM vs EC is P. muralis exposed to E. coerulea scent). An asterisk indicates zero.

P. muralis/L. agilis/Z. vivipara Interaction
In light of the recognition of E. coerulea scent by P. muralis, the lack of similar response towards the more closely related lacertids L. a. argus and Z. vivipara strongly suggests a diminished response by P. muralis rather than an inability to detect their chemical cues. This lack of response could indicate that P. muralis has no evolved or learned aversion to the stimuli (at least in the absence of other cues), in accordance with the threat sensitivity hypothesis. A threat-sensitive response is apparent in the different antipredator responses of P. muralis in relation to visual and scent cues (Amo et al., 2006). Conversely, the responses of Z. vivipara and L. a. argus to scent cues of P. muralis appear to have been based on a threat-sensitive perception of potential for a costly encounter, eliciting definitive avoidance and aggressive behaviour in Z. vivipara and L. a. argus, respectively.
If the aggressive reaction of L. a. argus represents an innate antagonistic response to P. muralis then an appropriate reciprocal behaviour might be expected in the reverse treatment. More likely, however, is that the sensory naïvety of the captive-born L. agilis individuals used in our study led to misinterpretation of the novel scent of the swabs as a potential prey item. Similar attacks on swabs have been observed in experiments specifically testing lizard discrimination of prey odours (Cooper, 1990(Cooper, , 1991(Cooper, , 1992. These results must therefore be considered with caution without further tests (e.g. including visual cues and recognition tests between L. agilis and Z. vivipara).
Taken on its own, there is a certain amount of ambiguity in the tongue flick response of Z. vivipara towards P. muralis. The greater amount of time spent in the control half (although not statistically significant), paired with more than twice the number of visits to the control refuge and an increased rate of exploratory tongue flicking in the scented half, is, however, indicative of an aversion response by Z. vivipara towards the non-native scent cues. This is in keeping with field observations of population declines and displacement of Z. vivipara in areas where introduced P. muralis are thriving (Mole, 2010;R. Williams, personal observation). Avoidance behaviour elicited by indirect cues alone has obvious advantages to mediating potential fitness costs arising from direct encounters, particularly when the costs of avoidance are low (Langkilde et al., 2005).
Overall, our results suggest that non-native P. muralis are unlikely to alter their behaviour in response to indirect chemical cues from native lizards with which they may potentially compete. Continued range expansion of non-native populations and greater overlap with native species' ranges is therefore likely to increase the incidence of direct interaction and possible aggressive encounters that may have fitness costs for native lizards (and P. muralis). In the case of Z. vivipara, an aversion response to indirect cues from P. muralis may mitigate the chance of direct encounters but ultimately lead to displacement of the native from previously occupied areas. Conversely, interactions between P. muralis and E. coerulea and L. agilis have potential to be more direct, the outcomes of which are likely to be context dependent and further research is needed to determine the nature and outcome of interactions when resources are limited (i.e. refuge, basking spot).
Juveniles of all the species tested here are vulnerable to predation from larger adult lizards, and therefore further experiments on juveniles may also reveal ontogenetic differences in responses.
These results highlight how responses to indirect cues might act to shape the competitive interaction between invading and native species, interactions that will ultimately determine invasion success and the impact on native communities.

Declaration of Interest
None.