Behavioural responses to environmental novelty in demersal, shelter-associated invasive ﬁ sh and their native analogues

Invasive species may differ from native species in terms of behavioural responses to the stress of encountering a novel environment. Learning about the nature of these differences can help us understand the mechanisms of dispersal and success of the alien species in colonized environments. Here, we investigated this topic using two Ponto-Caspian gobies as model species. They constitute a speci ﬁ c group of invasive ﬁ sh spreading in North America and Europe. They are benthic, territorial ﬁ sh of low mobility; they are poor swimmers and are strongly associated with shelters. We compared the behaviour of two invasive goby ﬁ shes (the racer goby, Babka gymnotrachelus, and the monkey goby, Neogobius ﬂ uviatilis ) to that of their native counterparts (the European bullhead, Cottus gobio , and the gudgeon, Gobio gobio , respectively). We used three laboratory tests to measure boldness e shyness traits: shelter occupancy test, novel object test and open ﬁ eld test. The European bullhead left the shelter later and was less active, and avoided the open ﬁ eld to a greater extent than the racer goby. The gudgeon was more associated with the shelter and novel object than the monkey gobyand, in contrast to the monkey goby, decreased its activity in the presence of the novel object and in the open ﬁ eld. All the species were attracted to the vicinity of the novel object. Our study suggests that the invasive Ponto-Caspian gobies are bolder when confronted with structural changes in their environment and have a greater potential to spread across the open bottom, devoid of hiding places, compared to their native analogue species

Freshwater fish are among the largest groups of animals involved in biological invasions Bernery et al., 2022;Genovesi et al., 2015).There is a growing number of papers on the mechanisms of the spread of invasive freshwater fish species (Bernery et al., 2022;Hayes & Barry, 2008), but many aspects of the process are still not fully understood, including the role of behavioural traits.After establishment, invasive organisms expand their ranges in new environments, which, in the case of mobile animals such as fish, is commonly related to active habitat exploration.The tendency to take risks and explore unfamiliar environments is often related to the organism's dispersal ability (Chapple et al., 2012;Myles-Gonzalez et al., 2015).In general, in their introduced range, invasive animals are associated with a higher level of boldness and habitat exploration, and greater dispersal capabilities compared to both conspecifics remaining in the native range (Myles-Gonzalez et al., 2015;Pintor et al., 2008), and native species encountered in the invaded areas (Juette et al., 2014 and references therein).It is unclear, however, how these behavioural traits are manifested in a specific group of demersal, shelter-associated invasive freshwater fish with burst-and-hold swimming modes, such as invasive gobies (Egger et al., 2021;Tierney et al., 2011).A fish exhibiting this specific swimming behaviour maintains its position on the substrate using its modified pelvic fins (pelvic sucker) and moves forward using brief bursts, after which the fish holds still on the substrate again.Thus, the fish characterized by this type of swimming spend most of their time motionless and therefore we can expect that their behavioural reaction to novelty will be different from that of the fish continuously present in the water column.For burst-andhold swimmers, behavioural responses to novelty in the environment can be varied and complex depending on their association with a shelter.For example, for organisms less associated with shelters, a structural change in the environment may pose a threat, whereas for shelter users, such a change may indicate an attractive hiding opportunity.On the other hand, shelter-associated burst-and-hold swimmers may respond more negatively to unfamiliar locations.Therefore, the interpretation of boldness and dispersal abilities in these fish is not easy and cannot be inferred from studies on invasive pelagic fish species showing prolonged swimming (e.g.Ashenden et al., 2017;Lukas et al., 2021;Rehage & Sih, 2004).Importantly, it is unclear here how such invasive species differ from their native counterparts in this respect.
The Ponto-Caspian gobies (Gobiidae) provide a good example of typically benthic invasive fish that spread in freshwater systems.Their native Ponto-Caspian region, consisting of the Black, Azov and Caspian Seas, lower sections of rivers flowing into these seas and their deltas, is a rich hotspot of invasive species that have successfully established themselves in many temperate regions of the world.This has been facilitated by their evolution under conditions of high fluctuations of abiotic factors (e.g.salinity, water level, temperature, dissolved oxygen) leading to the formation of a specific assemblage of eurytopic species tolerant to sudden changes in the environment (Bij de Vaate et al., 2002;Rewicz et al., 2014).The Ponto-Caspian species spread successfully within European waters (Roche et al., 2013), facilitated by interconnecting river basins and shipping, including the use of ship ballast waters (Bij de Vaate et al., 2002).Via ballast waters, they have also colonized the Laurentian Great Lakes of North America (Kornis et al., 2012).Invasion of the gobiids has caused changes in biocenoses, as these animals displace or threaten native fish species (Kornis et al., 2012;van Kessel et al., 2016) by outcompeting them for food (Kakareko et al., 2013) or shelter (Bło nska et al., 2016;Jermacz et al., 2015), alter food webs as predators (Barton et al., 2005;Janssen & Jude, 2001) and prey (Almqvist et al., 2010) and affect fish parasite populations (Ondra ckov a et al., 2021).The gobies are habitat generalists associated with diverse types of bottom habitats, ranging from homogeneous open bottom areas (Pła ˛chocki et al., 2020;Sapota, 2004) to structurally complex locations (Borcherding et al., 2013;Jan a c et al., 2018).
Our goal was to determine whether the behavioural reactions of the invasive gobies to novel environments and risky situations differ from those exhibited by the native co-occurring fish.We conducted laboratory tests to assess a range of behaviours associated with exploration and reactions to novelty in the environment.Three types of exploratory behaviour are worth mentioning here.The first is novelty-adjustive behaviour, in which the animal is passively confronted with a novel environment (an open field) or a novel environmental feature (a new object) and must adapt to the new situation.The second type is novelty-seeking behaviour.In this case, the individual searches for novelty on its own initiative (e.g. by leaving the shelter and exploring the area outside).The last type is goal-oriented, novelty-seeking behaviour.In this case, active exploration is only a means to achieve a goal, which, in itself, may be familiar (e.g.exploring an unfamiliar area to find a familiar food; Mc Reynolds, 1962).In our study, we used behavioural tests to explore both forced and spontaneous exploration, and our laboratory conditions allowed the use of completely novel stimuli (Berlyne, 1960).Following the classical approach of measuring animal boldnesseshyness, we performed (1) a shelter occupancy test, (2) a novel object test and (3) an open field test.
We hypothesized that the invasive gobies would react differently to the novel environment compared to their native counterparts.Specifically, we predicted the gobies would be bolder, that is, spend less time in the shelter, approach the novel object sooner and spend more time in the open field, and more explorative, thus covering a greater area in the open field test and leaving the shelter more often.Testing these hypotheses would help identify behavioural traits of strictly benthic fish allowing a better understanding of their dispersal mechanisms, which is particularly important in the case of invasive species.

Animals
We tested two goby species of Ponto-Caspian origin, pairing them with their native counterparts, which were local benthic fish, co-occurring with the gobies in the field and with a similar lifestyle: (1) the invasive racer goby, Babka gymnotrachelus, and the native European bullhead, Cottus gobio and (2) the invasive monkey goby, Neogobius fluviatilis and the native gudgeon, Gobio gobio (Kakareko et al., 2016;Pła ˛chocki et al., 2020).
We collected ca.120 juvenile fish (ca. 5 cm in total length) from the wild in summer (JulyeAugust).The European bullhead and the racer goby were caught from the Brda River (central Poland, 53 08 0 52.5 0 N, 17 58 0 10.5 0 E) by a diver using an aquarium net.The gudgeon and the monkey goby were obtained from the Pilica River (eastern Poland, 51 45 0 49.0 00 N, 21 08 0 56.7 00 E) by electrofishing (EFGI 650, BSE Bretschneider Spezialelektronik, Germany).As we compared the results only between the co-occurring pair members, different capture methods did not influence the results of the comparisons.After capture, we transported the fish to the laboratory and held them in 420-litre single-species stock tanks (50 individuals per tank).The stock tanks were filled with conditioned (24 h aged, aerated) tap water (pH 8.12 ± 0.19, electrical conductivity 606.2 ± 3.7 mS/cm, oxygen level 8.2 ± 0.4 mg/litre and 82.4 ± 3.6%; mean ± SD; measured with a Multi 340i Meter, WTW, Weilheim, Germany) at a temperature of 16.5 ± 0.5 C, maintained by air conditioning.The stock tanks were equipped with aquarium filters, aerators and ceramic and stony shelters, but had no bottom substrate.The photoperiod was set at a 12:12 h light:dark cycle with lights on at 0700.We fed the fish daily with frozen chironomid larvae ad libitum and exchanged water in the stock tanks once a week (ca.30% of water volume).

General Experimental Conditions
We performed three laboratory tests to assess behavioural responses of fish to a novel environment: (1) a shelter occupancy test, (2) a novel object test and ( 3) an open field test.In the shelter occupancy test, the main measure of boldness is the latency to leave the shelter: the shorter the time, the bolder the fish (Brown et al., 2007).In the novel object test, bolder fish tend to inspect the object sooner, more often and spend more time near the object (Wright et al., 2006).In the open field test, boldness is indicated by fish activity, time spent in the central part and the number of entries to the central part of the open field (Collier et al., 2017).
We tested single fish in 33-litre plastic experimental tanks (39 Â 30 cm and 28 cm high) with opaque white walls.The tanks were oval to avoid fish using corners as shelters.They were filled with 24 h aged and aerated (with an air stone) tap water.Above the experimental tank, we placed an IP video camera (SNB-6004P, Samsung, Changwon, South Korea) and an infrared lamp (MFL-I/ LED5-12 850 nm, eneo, R€ odermark, Germany).The whole set-up was covered by Styrofoam screens to prevent fish being disturbed by external visual stimuli.All the test species are nocturnal (Er} os et al., 2005;Grabowska et al., 2016;Nowak et al., 2019), spending most of the day hiding against predators.Exploration of a novel environment may put an animal at risk of encountering a predator regardless of the time of the day, but during a dark night the risk may be lower because of the worse prey visibility.Thus, we checked whether the studied species would show different propensities to take this risk during their activity and resting periods, to get a full picture of their behaviour.Therefore, experimental trials were performed both during the day and at night.The trial timing was adjusted to the photoperiod in which the fish were kept and acclimatized to avoid disturbing their circadian rhythm.Accordingly, the experiments were started 1 h after the light was turned on or off for the day and night trials, respectively.At the end of the trial, the fish was moved to a postexperimental tank to guarantee that each fish was used only once in the experiment.The water in the experimental tank was exchanged after each trial.Fish were not fed for 12 h before the trial to standardize hunger levels.
Videos were analysed using Noldus Ethovision XT 10.1 (Noldus Information Technology, Wageningen, The Netherlands).The behaviours determined in the three experiments are listed in Table A1.As the tested species are associated with the bottom of water bodies, we did not take their vertical movements into account.We measured the total length of the fish using the video frames and the ImageJ (Schneider et al., 2012) program to check for natural differences in size between sympatric species (see Results for details).

Experimental Procedures Shelter occupancy test
The experimental tank was filled with water to the level of 8 cm and a shelter was provided for the fish.This was made of two PVC tubes of different diameters.The smaller tube was glued to the bottom of the tank with aquarium silicone glue, whereas the bigger tube was movable and allowed us to open and close the entrance to the shelter.Additionally, we used two square PVC plates (10 Â 10 cm) to prevent the fish from entering the space behind the shelter (Fig. 1a).The shelter was equipped with a removable cap to isolate fish from external stimuli.At the start of the trial, a single fish was placed inside the closed shelter and the cap was placed on the top.After 2 min, the experimenter gently turned the bigger tube to open the shelter and allow the tested fish to swim outside.Each trial lasted for 30 min.
For the video analysis, we defined four zones: (1) the Shelter itself; (2) Inspection Zone, a 2 cm wide ring around the shelter entry; in this zone only the fish's head was outside the shelter; (3) Near Shelter Zone, a 3 cm wide ring around the Inspection Zone; when the fish entered this zone, its whole body was outside the shelter; (4) Dangerous Zone, the rest of the tank bottom (Fig. 2a); a fish was defined as having entered this zone when its head and pelvic fins were inside (here and for the other tests).

Novel object test
The experimental tank was filled with conditioned tap water to the level of 10 cm.A single fish was placed in the experimental tank for 12 h of acclimation.We divided the fish into two treatments: the control group, which had no contact with the novel object (to determine the baseline behaviour of tested fish) and the group experiencing the novel object during the test (hereafter: object group).In the latter group, after the acclimation period, the novel object (made of brown PVC, spherical with bevelled bottom and top; Fig. 1b) was gently dropped from the surface as close to the Plates preventing the fish from swimming behind the shelter centre of the tank as possible.The novel object was filled with plasticine to make it sink faster and to stabilize it on the bottom.Each trial lasted 20 min after the introduction of the novel object.
The first zone set for video analysis was the Object Zone.This was the part of the bottom covered directly by the object.In the control group, it was a randomly selected place near the centre of the experimental tank, that is, within the area where the actual novel object sank to the bottom.We used this approach to compare fish behaviour after the object introduction to their normal behaviour in the tank.Around the object, we set a 2.5 cm wide (approximately half of the average body length of the tested fish) Inspection Zone.The rest of the tank bottom was the Safe Zone (Fig. 2b).

Open field test
In the open field test, the experimental tank was filled with water to the level of 10 cm.A grey PVC tube (10 cm in diameter, 15 cm in height) was placed in the centre of the experimental tank (Fig. 1c) to provide the same start point for each trial.At the start of the trial, a single fish was placed into the tube for a 5 min acclimation period.Then, the tube was gently removed, and the trial started.Each trial was recorded for 3 h.We selected two test periods from each trial for further analysis.The first period (early response period) was the initial 20 min of the test, which corresponded to the most stressful period for the fish (i.e.just after placing it in the novel environment).The second period (late response period) was between 160 and 180 min after the start of the test.It was set after preliminary observations as a period when the fish became better acclimatized to the novel conditions.The differences between these periods indicate a stress response of the fish to the novel environment.
For the video analysis, we divided the bottom of the tank into 32 segments of equal areas, grouped in two zones.The Dangerous Zone was the central part of the bottom (24 segments).The Safe Zone was the 2.5 cm wide ring along the walls (eight segments; Fig. 2c).We used segments to check whether the fish explored the whole tank bottom or was active only within its limited area.

Ethical Note
The present study adheres to the ASAB/ABS guidelines for the use of animals in research (ASAB Ethical Committee & ABS Animal Care Committee, 2019).The fish were collected and used under the permit of the Local Committee for Ethics in Animal Research in Bydgoszcz, Poland, statement no.53/2022.In addition, the capture and use of the European bullhead, which is protected by law in Poland, was approved by the Regional Directorate for Environmental Protection in Poland (approval numbers: WOP.6401.4.5.2017.MO, WOP.6401.4.19.2018.MO).All procedures carried out within the study met the European Union guidelines on the protection of animals used for scientific purposes (Directive 2010/63/UE).
Fish were handled carefully during all procedures, including collection from the wild, transport, holding in tanks, testing, etc., to avoid affecting their welfare and behaviour.The catch was made as quickly and gently as possible.The fish were transported (over 2e3 h) in sealed plastic bags filled with oxygenated water, placed inside a Styrofoam cooler to maintain a constant temperature.The housing conditions guaranteed good animal welfare, which was manifested by the overall activity and food intake by the fish throughout the research period.We also did not notice any external signs of stress or disease (e.g.unnatural body shape, skin changes, swimming problems).After the experiments, the European bullhead and gudgeon were released where they were caught.Invasive gobies were euthanized by an overdose of tricaine methane sulphonate (MS-222).We kept fish in the anaesthetic solution up to The novel object test.The Object Zone (OZ) was the area of the object itself.In the control group (novel object absent), the Object Zone was set randomly near the centre of the tank, i.e. within the area where the actual novel object sank to the bottom.The Inspection Zone (IZ) was set directly around the OZ.The rest of the bottom was the Safe Zone (SZ).( c) The open field test.The Safe Zone (SZ) contained 25% of the total experimental area (bottom of the tank).The rest of the bottom was the Dangerous Zone (DZ).Both zones were divided into smaller segments of equal areas.
10 min after cessation of opercular movement, which is enough to cause brain death (Neiffer & Stamper, 2009).

Statistical Analysis
The differences in fish total length within each pair were tested using paired t tests.
For the European bullhead/racer goby pair we performed 16 trials of the shelter occupancy test with the racer goby and 20 trials with the European bullhead for each time of day, and 16 trials for each species Â time of day combination in the open field test.In the novel object test, we performed 10 trials with the racer goby for each treatment Â time of day combination.For the European bullhead, this experiment was replicated 12 times for each treatment Â time of day combination except the treatment with the novel object present during the day, for which 16 trials were performed.For the gudgeon/monkey goby pair, we performed 16 trials of the shelter occupancy test for each species Â time of day combination, 12 trials of the novel object test for each species Âtreatment Â time of day combination and 16 trials of the open field test for each species Â time of day combination.We compared results only within the above-mentioned pairs, as they were composed of species co-existing and interacting with each other in the same environments.
We performed a principal component analysis (PCA) on the correlation matrix separately for each test and each pair of fish species (six analyses in total) using behavioural variables from Table A1 (except for latency variables in the open field test, see below).The PCA reduces problems with multiple statistical comparisons and avoids the multicollinearity of multiple independent variables by producing orthogonal components based on sets of intercorrelated raw variables.Principal components (PC) were extracted based on their eigenvalues greater than 1.We took the original variables with absolute values of their loadings greater than 0.5 after KaisereVarimax rotation into account when explaining the meanings of the particular PCs.
The normality and homoscedasticity assumptions of determined PCs were not violated based on visual inspection of residual plots; thus, we decided to use parametric tests for their further analysis.For the shelter occupancy test, we used a two-way general linear model (GLM), with Species (one of the two species in the pair) and Time of day (day or night) as between-subject factors.Principal components from the novel object test were analysed using a three-way GLM with Species, Time of day and Treatment (control/object) as between-subject effects.For this test, we only interpreted terms involving the novel object presence effect as indicating responses of fish to the introduction of the novel object.For the open field test, we used a three-way general linear mixed model (GLMM), where Species and Time of day were betweensubject fixed effects, Period (early or late response) was a withinsubject fixed effect, and an individual ID was a random effect.For the open field test, we analysed the latency variables (time to the first occurrence of an event) separately, using the Cox proportional hazards model, as censored observations were present in the data set.Moreover, these variables were valid only for the early response period.For the novel object and shelter occupancy tests, we included latency variables to the PCA.For each analysis, we started with the full factorial model, then we simplified it by dropping consecutively the highest order nonsignificant interaction terms.Based on the Akaike information criterion (AIC), we retained a more complex model when its AIC was lower than that of the simpler model by 2 or more (Burham & Anderson, 2002).All models were then followed by Tukey HSD post hoc tests.For the novel object test, we were interested only in fish responses to the introduction of the novel object to the environment; thus, we only interpreted the main effect of Treatment and its interactions with the other factors.In the case of interactions, we tested all comparisons involving treatment.

RESULTS
The descriptive statistics (means, SDs, sample sizes) for behaviours measured during each experiment are available in the Supplementary material.
The total length of the fish was (mean ± SD) 5.24 ± 0.75 cm for the European bullhead, 5.30 ± 0.78 cm for the racer goby, 4.93 ± 0.93 cm for the gudgeon and 4. As the fish in both pairs lived in sympatry, we considered that the size differences between the gudgeon and the monkey goby reflected those found in nature.
The PCA extracted in total nine PCs for the racer goby/European bullhead pair (two PCs for the shelter occupancy test, four PCs for the novel object test, three PCs for the open field test) and 10 PCs for the gudgeon/monkey goby pair (three PCs for the shelter occupancy test, four PCs for the novel object test, three PCs for the open field test).The details of the PCA results are presented in Tables A2  and A3.

The European Bullhead/Racer Goby Pair
In the shelter occupancy test, high PC1 scores indicated low exploration of the area outside the shelter (shyness; Table A2).The GLM for PC1 showed a significant effect of Time of day (Table A4).Both species explored the area outside the shelter more at night than during the day (Fig. 3a).For PC2, high scores indicated fewer shelter exits and a higher association with the shelter (fear; Table A2).The GLM showed a significant effect of a Species*Time of day interaction (Table A4).The European bullhead tested during the day was more associated with the shelter than at night, as well as more associated with the shelter compared to the racer goby (Fig. 3b).
In the novel object test, higher PC1 scores indicated more time spent in the Inspection Zone, that is, at some distance to the object, rather than in its close proximity (Table A2).The GLM for PC1 showed a significant effect of the Species*Time of day*Treatment interaction (Table A5).When tested during the day, both species spent time in the Inspection Zone experiencing the novel object.At night, only the racer goby exhibited this response (Fig. 4a).For PC2, higher scores indicated low general activity (inactivity: Table A2).The GLM showed a significant Species*Time of day*Treatment interaction (Table A5).When tested at night, both species decreased their activity after the appearance of the novel object compared to their baseline behaviour.This activity reduction shown by the European bullhead was more pronounced than that shown by the racer goby as indicated by the significant difference in activity level between species at night when the novel object was present.Moreover, in the presence of the novel object, the European bullhead was less active at night than during the day (Fig. 4b).
For PC3, higher values indicated fewer freezing events by the fish (Table A2).The GLM showed significant main effects of Species and Treatment and a Species*Time of day interaction (Table A5).Both species exhibited more freezing events in the absence of the novel object than in its presence (Fig. 4c).For PC4, higher scores indicated avoidance of the zone directly associated with the object (Table A2).The GLM showed a significant Species*Time of day*Treatment interaction (Table A5).The racer goby avoided the novel object at night.(Fig. 4d).
For PC1 in the open field test, higher scores indicated open field exploration (Table A2).The GLMM showed a significant effect of Species, as well as Species*Period and Time of day*Period interactions (Table A6).The European bullhead decreased its open field use in the late response period compared to the early response  A2).The presented values are estimates (means ± 95% confidence interval) predicted for significant terms of the general linear models (Table A4)  A2).The presented values are estimates (means ± 95% confidence interval) predicted for significant terms of the general linear models (Table A5): (a, b, d) Species (bullhead/racer goby)*Treatment (novel object absent/present)*Time of day (day/night) interaction and (c) a main effect of Treatment.Asterisks indicate differences at P < 0.05.
period.In contrast, the racer goby increased its open field use in the late response period compared to the early response period.During the late response period, the racer goby used the open field more often than the European bullhead (Fig. 5a).The racer goby used the open field more at night than during the day.Moreover, at night, it used the open field more than the European bullhead (Fig. 5b).For PC2, higher scores indicated high general activity level (Table A2).
The GLMM showed significant main effects of Species and Period, as well as of their interaction, and of a Species*Time of day interaction (Table A6).The European bullhead was more active than the racer goby during the day (Fig. 5c).In the early response period, the European bullhead was more active than in its late response period and than the racer goby (Fig. 5d).For PC3, higher scores indicated exploration of the safe, peripheral part of the arena (Table A2).The GLMM showed significant main effects of Species, Period and their interaction (Table A6).The European bullhead explored the Safe Zone to a greater extent than the racer goby during the early response period.The racer goby explored the Safe Zone to a lesser extent during the early than in the late response period (Fig. 5e).
The Cox proportional hazards model showed a significant effect of Species on the latency to the first movement and the first transition from the Safe Zone to the Dangerous Zone (Table A7).The European bullhead started both activities sooner than the racer goby (Fig. 6).

The Gudgeon/Monkey Goby Pair
In the shelter occupancy test, the higher scores of PC1 indicated low exploration of the area outside the shelter (shyness; Table A3).The GLM showed significant main effects of Species and Time of day (Table A8).The gudgeon explored the area outside the shelter to a lesser extent than the monkey goby (Fig. 7a).Both species explored this part of the arena less during the day than at night (Fig. 7b).On the other hand, the GLM showed no significant effects of the tested factors on PC2 (Table A8), for which higher scores indicated fewer shelter exits and a higher association with the shelter (fear; Table A3).For PC3, higher scores indicated more time outside the shelter (Table A3).The GLM showed a significant effect of Time of day (Table A8): both species spent more time outside the shelter at night than during the day (Fig. 7c).
In the novel object test, higher scores of PC1 indicated high general activity of the fish (Table A3).The GLM showed a significant main effect of Time of day and of a Species*Treatment interaction (Table A9).The gudgeon decreased its activity after the novel object appearance.Moreover, when the object was present in the arena, the gudgeon was less active than the monkey goby (Fig. 8a).For PC2, higher scores indicated object avoidance (Table A3).The GLM showed a significant effect of Species*Treatment and Time of day*Treatment interactions (Table A9).Both species spent more time in zones related to the object after the novel object appearance compared to the control.This reaction was more pronounced in the gudgeon, as indicated by the significant difference between the species when the novel object was present (Fig. 8b).Both species preferred the object more during the day than at night (Fig. 8c).For PC3, higher scores indicated few object inspections (visits in zones around the Novel Object; Table A3).The GLM showed significant Species*Treatment and Time of day*Treatment interactions (Table A9).The monkey goby entered the areas near the object less often than in the control and less often than the gudgeon (Fig. 8d).Moreover, at night, the fish visited the novel object zones less compared to their baseline behaviour (Fig. 8e).For PC4, higher scores indicated more freezing events and more time spent in the Safe Zone (fear; Table A3).The GLM showed significant effects of  A2).The presented values are estimates (means ± 95% confidence interval) predicted for significant terms of the general linear mixed models (Table A6): (a, d, e) Period (early/late response)*Species (bullhead/racer goby) interaction, (b) Time of day (day/night)*Period interaction and (c) Time of day*Species interaction.Asterisks indicate differences at P < 0.05.
Species*Time of day, Species*Treatment and Time of day*Treatment interactions (Table A9); however, the post hoc tests revealed the differences only for the last interaction.During the day, both species decreased their fear level after the novel object appearance compared to their baseline behaviour.At night, we observed the opposite response: in the presence of the novel object, both species increased their fear level compared to the baseline behaviour.In general, both species exhibited more fear responses at night than during the day when the novel object was present (Fig. 8f).
For PC1 in the open field test, higher scores indicated low activity and more freezing events (inactivity; Table A3).The GLMM for PC1 showed a significant effect of a Species*Time of day*Period interaction (Table A10).In the early response period, the gudgeon was less active than the monkey goby regardless of the time of day.In the late response period, the gudgeon was also less active during the day than at night.Moreover, at night, the gudgeon was less active in the early than in the late response period (Fig. 9a).For PC2, higher scores indicated high explorative behaviour (Table A3).The GLMM showed a significant effect of a Species*Period interaction (Table A10).The monkey goby was more explorative in the early than in the late response period (Fig. 9b).For PC3, higher scores indicated more time spent in and more visits to the Dangerous Zone (boldness; Table A3).The GLMM showed a significant effect of a Species*Time of day*Period interaction (Table A10), resulting from the higher boldness of the gudgeon compared to the racer goby in the late response period during the day (Fig. 9c).The Cox proportional hazards model showed a significant effect of Species on the latency to the first movement (Table A11), whereas there were no effects of the tested factors on the time of the first transition from the Safe Zone to the Dangerous Zone (Table A11).The gudgeon showed the first movement earlier than the monkey goby (Fig. 10).

DISCUSSION
We confirmed our hypothesis that, at least in some contexts shaped by additional factors, the invasive gobies presented different behaviours from their native counterparts when facing novel situations.The results of the three tests, used to assess boldnesseshyness of the fish, suggest that the invasive gobies were relatively bolder than the natives.
In the shelter occupancy test, the European bullhead left the shelter later and was less active outside the shelter than the racer goby.The gudgeon also showed higher affinity with the shelter and explored the arena outside the shelter to a lesser extent than the monkey goby.We therefore observed a similar pattern of behavioural differences in the pairs of species studied.In both cases, the gobies were less shelter-oriented (although for the racer goby this was visible only during the day) and more active than the natives.A3).The presented values are estimates (means ± 95% confidence interval) predicted for significant terms of the general linear models (Table A8): (a) a main effect of Species (gudgeon/monkey goby) and (b, c) Time of day (day/night).Asterisks indicate differences at P < 0.05.For PC2, no significant effects of the tested factors were found in the model.
These behaviours in the shelter occupancy test are typically considered boldness indicators (Brydges & Braithwaite, 2009;McCormick et al., 2017;Mustafa et al., 2019).In this context, the two gobies could be considered bolder, as they were more likely to exhibit active novelty-seeking behaviour and take risks in an area outside their safe hiding place.
In the novel object test, all the species were attracted to the object's vicinity, which suggests that they seemed to treat it as a potential shelter.On the other hand, they tended to decrease their activity in the presence of the object.A decrease in activity in the form of an active search for stimuli is one of the most common reactions of animals to a threat and the observed behaviour fits more to an attempt to adapt to the prevailing conditions.This reaction was generally more pronounced in the native species than in the invaders.Additionally, for the European bullhead and racer goby pair there were differences in fish reaction to the novel object depending on the time of day.This dependence may be explained by the lower visibility of the object at night (i.e. in darkness) compared to the day, and, in consequence, a limited capability of assessment of the real threat.As the novel object in our test was an artificial, inanimate object, the only source of information about it was the visual cue.When the object appeared during the day, both species could see and determine it as not dangerous.When the same situation took place at night, low visibility combined with the absence of other signals (i.e.chemicals) resulted in a higher level of uncertainty about danger for the fish.The propensity to inspect a novel object shown by the racer goby combined with its higher activity, despite its avoidance of the novel object itself, could be associated with the higher boldness (White et al., 2013) of this species at night compared to the European bullhead.The native gudgeon spent more time close to the object and visited the zones around it more often than the monkey goby.These results interpreted alone may indicate that the gudgeon is bolder than the monkey goby as it took a risk inspecting the novel object (White et al., 2013;Wright et al., 2006).However, the higher affinity of the gudgeon with the shelter (in the shelter occupancy test) and its lower general activity after the novel object's appearance compared to the monkey goby suggest that the gudgeon perceived the object as a potential shelter, and that is why it was associated with its vicinity.At the same time, the bolder monkey goby explored the environment.
The results of the open field test, considered in the light of the results of the previous two tests, indicated lower boldness of the native fish compared to the invasive gobies.In the first pair, the European bullhead started its activity faster, was more active and explored the peripheral zone to a greater extent than the racer goby in the initial period after emerging into the open field, suggesting higher boldness (Forsatkar et al., 2016).However, this activity can also be interpreted as an attempt to find an escape route to a more friendly environment in a sheltered location.The results of the shelter occupancy and novel object tests showed that the bullhead was more associated with the shelter and less active after the novel object's appearance than the racer goby.Taking the results of the three tests together, the high levels of activity and exploration expressed by the bullhead during the initial period of the open field test might have resulted from the hyperactivity caused by the high anxiety experienced by the tested fish (Jarrold et al., 2020).As the bullhead was active mostly in the peripheral part of the arena, this  A3).The presented values are estimates (means ± 95% confidence interval) predicted for significant terms of the general linear models (Table A9): (a, b, d) Species (gudgeon/monkey goby)*Treatment (novel object absent/present) interaction and (c, e, f) Time of day*Treatment interaction.Asterisks indicate differences at P < 0.05.
was likely related to looking for a refuge and/or possibility to escape.The results from the late response period, when the fish became more acclimated to the experimental arena, support the idea that, apart from an initial period of hyperactivity, the European bullhead was less explorative (which could be interpreted as more cautious) than the racer goby.The exploration of the open field by the European bullhead decreased, while the exploration of the whole arena by the racer goby increased compared to the early response period.As a result, the racer goby during the late response period explored the whole arena to a greater extent than the bullhead, while during the initial period, facing the uncertainty about the environment, the racer goby reduced its activity to avoid detection by a potential predator.To sum up, we are of the opinion that behavioural reactions to stress related to the novel environment were more pronounced in the European bullhead, which is more associated with shelter and less explorative than the racer goby.In the second pair of fish tested, we do not associate the higher activity of fish with hyperactivity, and we therefore interpret it classically as a sign of boldness.This is because there is no contrast between the elevated activity of the fish in the initial phase of the open field test and the generally low activity in the previous tests, as it was in the case of the European bullhead.The native gudgeon showed lower activity in the early response period compared to the late response period at night.Such lower activity in the initial period in the novel environment is a typical behavioural response to danger (Lima & Dill, 1990;Teplitsky & Laurila, 2007), indicating that the gudgeon was less bold than the monkey goby.Moreover, the bolder behaviour of the monkey goby was also confirmed by its higher activity in the initial test period compared to the gudgeon, regardless of the time of day.We observed the lower boldness of the monkey goby compared to the gudgeon in the late response period at night and the lower explorative behaviour of the former in the late response period compared to the initial period regardless of the time of day.In light of the previous, shorter tests, we may assume that the goby collected information about the environment during the initial period in the open field test, so that in the second period of this test it could reduce its activity, as there were no further changes in the environment (no novel stimuli).To summarize, the behavioural responses to novelty expressed by the gudgeon were more pronounced than those of the monkey goby.A3).The presented values are estimates (means ± 95% confidence interval) predicted for significant terms of the general linear mixed models (Table A10):  A11).
Our findings on the activity of these fish and their association with the hiding places, are supported by the other studies that have been carried out on these species.Augustyniak et al. (2022) showed that the European bullhead, facing a predator danger (a direct stressor affecting behaviour), spent more time in the shelter and was less active than the racer goby.The greater association of the European bullhead with the shelter compared to the racer goby was also shown by Grabowska et al. (2016).In the wild, the European bullhead is commonly associated with areas with high water velocities (Carter et al., 2004;Roje et al., 2021) and stony bottoms (Kakareko et al., 2016).Thus, the shelter, besides providing protection against predators, allows them to save energy resources when facing higher water currents.The racer goby, in turn, occurs in more lentic habitats, including soft (mud, sand) sediments (Kakareko et al., 2016;Pła ˛chocki et al., 2020), which demand lower energy expenditure to keep the position against the water current.The more pronounced behavioural responses of the gudgeon than the monkey goby to predation cues was shown by Kłosi nski et al. (2022).They found that the gudgeon decreased its activity to a greater extent relative to the monkey goby.Importantly, the gudgeon individuals showed thigmotaxis in the presence of predation cues.The authors concluded that this probably increased the sense of safety in stressed fish.These results can be related to our observations in the novel object test, where the gudgeon spent more time near the object and treated it as a potential shelter.Combining the results by Kłosi nski et al. ( 2022) and our current study, it seems that the gudgeon facing a stressful situation first tries to find a shelter.This seems to be contrast with the results of Augustyniak et al. (2022), where the gudgeon facing a direct predator danger exhibited escape responses instead of staying inside a provided shelter.However, we could assume that the risk of a direct predator attack is a more stressful stimulus than the novel environment or the presence of an alarm substance alone; thus, the fish exhibit different behavioural strategies to survive.The monkey goby, in turn, relies on the shelter to a lesser extent, as in dangerous situations it may burrow in the sandy bottom and stay there motionless (e.g.C apov a et al., 2008;Er} os et al., 2005).To summarize, in the present work, we showed for the first time that the invasive Ponto-Caspian gobies faced with stressful situations related to sudden changes in the environment containing a new stimulus of an absolute nature (new object, open field) may be less dependent on the availability of hiding places than native species of the same guild.Instead, they explore the area to a greater extent than their native counterparts.Although these differences seem to appear in specific ecological contexts shaped by additional factors, for example, the time of day, they are likely to occur in the wild and affect behaviour.It is also worth noting that the gudgeon was the only gregarious species in our study; thus, the presence of conspecifics could also affect its behavioural responses to novelty.Nevertheless, behaviours exhibited by invasive gobies may enhance their dispersal abilities over long distances, as active exploration of the area may increase the probability of finding the transport vector (e.g.entering the ballast water tank of a ship).This would allow the invaders to enter novel areas (Chapple et al., 2012), whereas their boldness may be related to the short-distance (local) dispersal after their release in the field (Fraser et al., 2001).The same set of behaviours may facilitate goby establishment in the invaded areas, as their greater activity may lead to more efficient resource finding and utilization compared to the native fish species present in the environment.Additionally, the gobies, which are less shelterdependent than the native species encountered in the invaded areas, may be capable of colonizing bare, open bottom areas, which was not considered earlier.We must keep in mind that the greater boldness of the gobies compared to the natives may also increase the risk of encountering a predator (McGlade et al., 2022), although this is not always the rule as Blake et al. (2018) showed a positive predator-dependent role of boldness in the prey's survival.However, the net effect of such a set of features can significantly increase the chances of a successful invasion of these species in aquatic environments where human interference leads to uniformity of the bottom topography, such as the offshore areas of regulated, channelized rivers.The results of our study also bring new information about the methodology used to study the boldness-related behaviours of shelter-associated species.As the gudgeon in our study treated the novel object as a shelter rather than as a source of novelty, we recommend introducing a familiar hiding place in the experimental set-up of future novel object tests with this species.Moreover, to avoid the hyperactivity of the tested fish in the open field, as shown by the European bullhead in our study, the methodology of the open field test for this specific group of shelterassociated fish species might be improved by providing more realistic conditions inside the arena, by, for example, adding a substrate to the bottom of the tank.The Species and Time of day were set as between-subject factors.Bold type indicates significant effects (P < 0.05).The analyses were run for principal components obtained in the PCA based on fish behavioural variables (Table A2).

Figure 1 .
Figure 1.Experimental set-ups.(a) The shelter occupancy test.The fish was placed inside a closed shelter.After 2 min of acclimation, the shelter was opened (the bigger tube was gently turned to match the entrance position with the smaller tube) and the observation started.(b) The novel object test.The fish was placed in the tank for a 12 h acclimation period.Then, in the novel object treatment, the novel object was gently placed near the middle of the tank and the observation started.(c) The open field test.The fish was placed in the insertion cylinder in the tank for a 5 min acclimation.Then, the cylinder was gently removed and the observation started.All the dimensions are given in cm.

Figure 2 .
Figure2.The zones set for the video analysis.(a) The shelter occupancy test.In the Inspection Zone (IZ) only the fish's head was outside the shelter.The fish present in the Near Shelter Zone (NSZ) left the Shelter with its whole body.The rest of the bottom was the Dangerous Zone (DZ).(b) The novel object test.The Object Zone (OZ) was the area of the object itself.In the control group (novel object absent), the Object Zone was set randomly near the centre of the tank, i.e. within the area where the actual novel object sank to the bottom.The Inspection Zone (IZ) was set directly around the OZ.The rest of the bottom was the Safe Zone (SZ).(c) The open field test.The Safe Zone (SZ) contained 25% of the total experimental area (bottom of the tank).The rest of the bottom was the Dangerous Zone (DZ).Both zones were divided into smaller segments of equal areas.

Figure 3 .
Figure 3. Shelter occupancy by the European bullhead and racer goby depending on the time of day (day, night): (a) principal component 1 (PC1) indicates a low level of exploration of the area outside the shelter and (b) PC2 indicates a high association with the shelter (TableA2).The presented values are estimates (means ± 95% confidence interval) predicted for significant terms of the general linear models (TableA4): a main effect of Time of day (day/night) (a) and Time of day*Species (bullhead/racer goby) interaction (b).Asterisks indicate differences at P < 0.05.

Figure 4 .
Figure 3. Shelter occupancy by the European bullhead and racer goby depending on the time of day (day, night): (a) principal component 1 (PC1) indicates a low level of exploration of the area outside the shelter and (b) PC2 indicates a high association with the shelter (TableA2).The presented values are estimates (means ± 95% confidence interval) predicted for significant terms of the general linear models (TableA4): a main effect of Time of day (day/night) (a) and Time of day*Species (bullhead/racer goby) interaction (b).Asterisks indicate differences at P < 0.05.

Figure 5 .
Figure 5. Behaviours of the European bullhead and racer goby in the open field test: (a, b) principal component 1 (PC1) indicates open field exploration, (c, d) PC2 indicates activity and (e) PC3 indicates exploration of the peripheral part of the arena (TableA2).The presented values are estimates (means ± 95% confidence interval) predicted for significant terms of the general linear mixed models (TableA6): (a, d, e) Period (early/late response)*Species (bullhead/racer goby) interaction, (b) Time of day (day/night)*Period interaction and (c) Time of day*Species interaction.Asterisks indicate differences at P < 0.05.

Figure 6 .Figure 7 .
Figure 6.The latency to (a) the first movement and (b) the first transition from the Safe Zone (SZ) to the Dangerous Zone (DZ) of the European bullhead and racer goby after being introduced to the open field tank.The presented values are predicted by the Cox proportional hazards model for species differing significantly from each other (as shown by asterisks), after pooling day and night responses (nonsignificant effect of Time of day; TableA7).

Figure 8 .
Figure 8. Responses of the gudgeon and monkey goby to the novel object: (a) principal component 1 (PC1) indicates activity, (b, c) PC2 indicates object avoidance, (d, e) PC3 indicates a low number of object inspection events and (f) PC4 indicates fear (TableA3).The presented values are estimates (means ± 95% confidence interval) predicted for significant terms of the general linear models (TableA9): (a, b, d) Species (gudgeon/monkey goby)*Treatment (novel object absent/present) interaction and (c, e, f) Time of day*Treatment interaction.Asterisks indicate differences at P < 0.05.

Figure 9 .
Figure 9. Behaviours of the gudgeon and monkey goby in the open field test: (a) principal component 1 (PC1) indicates inactivity, (b) PC2 indicates explorative behaviour and (c) PC3 indicates boldness (TableA3).The presented values are estimates (means ± 95% confidence interval) predicted for significant terms of the general linear mixed models (TableA10): (a, c) Species (gudgeon/monkey goby)*Time of day (day/night)*Period (early/late response) interaction and (b) Species*Period interaction.Asterisks indicate differences at P < 0.05.

Figure 10 .
Figure 9. Behaviours of the gudgeon and monkey goby in the open field test: (a) principal component 1 (PC1) indicates inactivity, (b) PC2 indicates explorative behaviour and (c) PC3 indicates boldness (TableA3).The presented values are estimates (means ± 95% confidence interval) predicted for significant terms of the general linear mixed models (TableA10): (a, c) Species (gudgeon/monkey goby)*Time of day (day/night)*Period (early/late response) interaction and (b) Species*Period interaction.Asterisks indicate differences at P < 0.05.

Table A2
Results of the principal component analyses (PCA) on behavioural variables of the European bullhead and racer goby in various experiments.Latency to the first IZ visit (0.530), Number of the IZ visits (À0.851),Latency to the first OZ visit (0.849), Time in the OZ (À0.709),Number of the OZ visits (À0.524)Dangerous Zone; SZ, Safe Zone; IZ, Inspection Zone; NSZ, Near Shelter Zone; OZ, Object Zone.See Fig. 2 and Table A1 for details.Correlations of measured variables with the principal component after KaisereVarimax rotation (loadings with absolute values higher than 0.5 shown).Results of the principal component analyses (PCA) on behavioural variables of the gudgeon and monkey goby in various experiments Distance to the Shelter entry (À0.889),Latency to the first IZ visit (0.802), Latency to the first NSZ visit (0.883), Latency to the first DZ visit (0.954), Time in the DZ (0.829), Number of the DZ visits (0.670) DZ, Dangerous Zone; SZ, Safe Zone; IZ, Inspection Zone; NSZ, Near Shelter Zone; OZ, Object Zone.See Fig. 2 and Table A1 for details.
a Principal components discriminated by the PCA.b Eigenvalue of the principal component.c Percentage of variance explained by the principal component.d a Principal components discriminated by the PCA.b Eigenvalue of the principal component.c Percentage of variance explained by the principal component.d Correlations of measured variables with the principal component after KaisereVarimax rotation (loadings with absolute values higher than 0.5 shown).

Table A4
Two-way general linear models to test the shelter occupancy of the European bullhead and racer goby depending on the time of day