Introductions of non-native species (NNS) are major drivers of biodiversity loss. Gammarids (Crustacea, Gammaroidea) have been particularly successful in establishing and spreading in their non-native range, especially in Europe. While their impacts are wide-ranging, interference competition with native species has received limited study to date. Here, we assessed the competitive abilities of the successful North American NNS Gammarus tigrinus relative to the European native Gammarus duebeni, over a chironomid larva as a single food resource. We staged four types of dyadic contest encounters, with individuals of the native or NNS added to the experimental arena containing the food resource, and inter- or intraspecific competitor individuals added upon the first individual taking possession of the resource, or after 20 minutes. Gammarus tigrinus were more likely to take hold of the bloodworm in the opening 20 minutes, and did so more quickly than G. duebeni. During this period, they were also less thigmotactic than the native, being more explorative and spending a smaller proportion of time in the outer zone of the arena. They exhibited more aggressive interactions and activity with increasing size and mass, whereas larger G. duebeni were shown to be less aggressive and less active. Gammarus tigrinus were found to be significantly less likely to lose possession to G. duebeni than they were to conspecifics, whereas G. duebeni were similarly likely to lose possession to G. tigrinus as to conspecifics. Overall, our findings indicate that the behaviour and competitive ability of G. tigrinus demonstrated here add to a list of traits that facilitate its invasion success. In addition, our method offers potential as an effective, standardisable means of assessing the competitive abilities of gammarid NNS. We encourage future studies to develop it further, incorporating alternative resources, such as habitat, and to assess the role of ecologically relevant abiotic stressors in determining contest outcomes.

Animal behaviour, Baltic Sea, contests, exploitative competition, impact assessment, interference competition, non-native species

Global translocations of non-native species (NNS) are major drivers of biodiversity loss, affecting ecosystem services, human health and welfare, food security, and economic costs (Pyšek et al. 2020, Cuthbert et al. 2021, IPBES 2023). Amphipod crustaceans of the family Gammaroidea are one group of particularly damaging NNS, which have been highly successful at displacing natives, especially in Europe (Grabowski et al. 2007, Cuthbert et al. 2020). Gammarids are significant drivers of disturbance through competition for food and habitat resources, as well as predation, with major implications for native biodiversity in recipient freshwater and brackish ecosystems (Conlan, 1994). One such example is the North American Gammarus tigrinus, which has been shown to have negative impacts on native amphipod assemblages in introduced ecosystems (Jänes et al. 2015, Reisalu et al. 2016), as well as socio-economic costs to fishermen through damaged fishing gear and injured catches (Pinkster et al. 1977). Native to the brackish waters of tidal estuaries along North America’s Atlantic coast, the first European occurrence was in England in 1931, likely arriving in ship ballast water (Sexton and Cooper 1939). From here, it was introduced to Northern Ireland in the 1950s before spreading southwards, replacing the native G. lacustris and G. duebeni celticus as the dominant gammarid in Lough Conn’s sublittoral zone (O’Grady and Holmes 1983). It arrived and established in mainland Europe when the English stock was deliberately introduced to the polluted Werra river in Germany in 1957, and subsequently when the Lough Neagh population was introduced to IJsselmeer in the Netherlands in 1964. Since then, its spread has led to cases of replacement or population decreases for the native G. pulex, G. zaddachi and G. d. duebeni (Nijssen and Stock 1966, Pinkster et al. 1992, Kazanavičiūtė et al. 2024). In 1975, it reached the German part of the Baltic Sea, and from there its coastal range expansion has continued (see Rewicz et al. 2019 and references therein).

The Baltic Sea is the world’s largest brackish-water basin and a highly unique ecosystem, with a low number of native species, many of which are postglacial immigrants, and at least 132 non-native and cryptogenic species (Casties et al. 2016, Ojaveer et al. 2017). Here, the combination of few native species, environmental instability and high anthropogenic pressure means the ecosystem is deemed sensitive to biological invasions (Reisalu et al. 2016, Rewicz et al. 2019). Indeed, G. tigrinus is joined by another North American gammarid NNS in Melita nitida, as well as Lake Baikal’s Gmelinoides fasciatus, and seven NNS from the Ponto-Caspian region: Chaetogammarus warpachowskyi, Chelicorophium curvispinum, Dikerogammarus villosus, D. haemobaphes, Chaetogammarus ischnus (formerly Echinogammarus ischnus: Copilaș-Ciocianu et al. 2023), Spirogammarus major (formerly E. trichiatus: Copilaș-Ciocianu et al. 2023), Obesogammarus crassus, and Pontogammarus robustoides (Rewicz et al. 2019). Of these, G. tigrinus is viewed as one of the most euryhaline (Grabowski et al. 2006). While various studies have looked at the environmental tolerance of the species (Casties et al. 2019, Paiva et al. 2020), its life history traits (Pinkster et al. 1977, Grabowski et al. 2007) and even its ability to facilitate the consumption of congeneric native species by predators (Kotta et al. 2010), there remains a gap in the literature with regards to interference and exploitative competition outcomes over food resources with native gammarids. More broadly, there has been little study of contests to understand interspecific interactions, and its role on niche partitioning, species coexistence and biodiversity (Paijmans and Wong 2017), with even less in an invasion ecology context (but see, for example, Zeng et al. 2019).

Heeding recent calls for more studies to help decipher the competitive mechanisms at play between NNS and native species (Damas-Moreira et al. 2020), we paired G. tigrinus with G. duebeni, which has been outnumbered or replaced by the NNS in Ireland, the Netherlands and in the Baltic Sea (e.g. Vistula Lagoon, Poland: Grabowski et al. 2006; Dassower See, Germany: Kazanavičiūtė et al. 2024). Staging a series of dyadic contest experiments over a finite food resource, we used an experimental setup that involved staging four types of contest pairings: G. tigrinus vs G. tigrinus, G. duebeni vs G. duebeni, G. tigrinus vs G. duebeni and G. duebeni vs G. tigrinus. We sought to investigate species behaviour and determine the role of species identity, as well as body size and mass (traditional proxies for fighting ability, termed “resource holding potential”: Arnott and Elwood 2009) in determining contest outcomes. Specifically, we assessed boldness, activity and competitive ability using eight key tests. Boldness was assessed by examining the latency to approach the resource (1) and thigmotactic behaviour (2). Next, we assessed activity via the number of line crosses (see Fig. 1) per time in the experimental arenas (3), and finally, we assessed competitive ability (4–8). Boldness, i.e. how individuals behave in potentially risky situations (Réale et al. 2007), has been suggested to be a determinant of whether individuals are likely to disperse or remain sedentary, or whether they are short or long-distance dispersers (Fraser et al. 2001). This, alongside activity, has been highlighted as a beneficial behavioural trait across multiple stages of the introduction process (Chapple and Wong 2016), and they have been positively correlated in a number of invasive NNS (Brodin and Drotz 2014, Lukas et al. 2021). With G. duebeni having no known non-native populations (Paiva et al. 2018, Cuthbert et al. 2020), we propose these traits will be less obvious for the native species. As a result, we hypothesised that: (1) expecting it to be bolder, the non-native G. tigrinus is more likely than the native G. duebeni to take hold of the food resource in the opening 20 minutes, and of the individuals from both species taking hold of the food resource during that period, G. tigrinus is faster to take possession of the resource; (2) for the period of time that the individual added first is alone in the arena, G. tigrinus is less thigmotactic, i.e. spends less time in the outer ring of the arena; and (3) G. tigrinus is more active than G. duebeni. In terms of their competitive ability over food resources, our hypotheses were informed by previous studies which have shown G. tigrinus to be successful at outcompeting native species for preferred habitat (Kotta et al. 2011, Reisalu et al. 2016), while exhibiting high feeding rates (Dick 1996, Dickey et al. 2021) and demonstrating aggression towards native gammarids (Orav-Kotta et al. 2009, Kotta et al. 2010). As a result, we hypothesise that: 4) the trial type (i.e. species match-up) and size disparity will have significant effects on the number of aggressive interactions (i.e. bumps, approaches: see Table 1); 5) contests involving the “aggressive” G. tigrinus will be more likely to escalate into “wrestles” over the food resource 6) the number and duration of wrestling bouts will be significantly affected by trial type and contestant size disparity; 7) for the gammarid added first, larger G. tigrinus will spend greater proportions of time in possession; and 8) Gammarus tigrinus will be more likely to dispossess G. duebeni, and larger individuals will be more likely to dispossess smaller individuals.

Figure 1.

A diagram of markings on the experimental arena, broken down into the constituent zones (B) and lines (C). The zones were used as a determinant of thigmotactic behaviour (hypothesis 2), whereas the lines were used to determine line crosses, a measure of activity (hypothesis 3).

Gammarus tigrinus were more likely to take hold of the bloodworm in the opening 20 minutes, and those that did, did so more quickly than G. duebeni (hypothesis 1). Twenty-one of the individuals added first took hold of the bloodworm in the opening 20 minutes, the majority of which were G. tigrinus (G. tigrinus: n = 15, 71.4%; G. duebeni: n = 6, 28.6%). Of the individuals that took the bloodworm in the opening 20 minutes, G. tigrinus had a significantly lower latency (mean +/- SE: G. tigrinus 270.00 +/- 55.00 seconds; G. duebeni 343.23 +/- 77.36 seconds; Wilcoxon rank sum exact test: W = 74, p = 0.023).

There was a significant effect of species on the proportion of time spent in the outer zone (hypothesis 2), with G. tigrinus spending more time in the middle and inner zones than the native G. duebeni (Beta regression: z = 5.858, p < 0.001; Fig. 2; Suppl. material 1: table S1). There was also a significant two-way interaction of species and length on activity (F_1,67 = 5.914, p = 0.019; Suppl. material 1: table S2) as measured by the number of line crosses per time spent in the arena (hypothesis 3), with G. tigrinus more active with increasing length, but G. duebeni less active with increasing length (Fig. 3A). A similar pattern was shown when mass was accounted for, with a significant two-way species and mass interaction on activity (F_1,67 = 6.646, p = 0.016; Fig. 3B; Suppl. material 1: table S3).

Figure 2.

Boxplots outlining the relative proportional time spent in each of the three zones and how this differed between G. tigrinus and G. duebeni A outer zone B middle zone C inner zone.

Figure 3.

Scatter plots showing the effects of A mass and B head-to-telson length on the number of line crosses per second in the arena, and how this differed between G. tigrinus and G. duebeni. Lines shown with 95% confidence intervals.

With regards to hypothesis 4, there was no significant effect of trial type and head-to-telson length disparity or mass disparity on the combined number of bumps and approaches between the participant gammarids (Quasipoisson GLM: trial type the sole independent variable in both minimum adequate models, p = 0.249). However, at an individual level, a significant two-way interaction effect between species and mass on the number of approaches and bumps (Quasipoisson GLM: F_1,68 = 5.202, p = 0.040; Suppl. material 1: table S4), with larger G. tigrinus committing more bumps and approaches, but larger G. duebeni committing fewer (Fig. 4).

Figure 4.

Scatterplot outlining the effect of mass on the number of aggressive interactions, and how this differed between G. tigrinus and G. duebeni. Lines shown with 95% confidence intervals.

There was no significant effect of trial type on proportion of contests that led to wrestles (4-sample test for equality of proportions without continuity correction: p = 0.517; hypothesis 5). Similarly, there were no significant effects for trial type and head-to-telson length disparity (Gaussian GLM: length disparity the sole independent variable in minimum adequate model, p = 0.165) or mass disparity (Gaussian GLM: trial type the sole independent variable in minimum adequate model, p = 0.153) on the log-transformed time spent wrestling or the number of wrestling bouts (Quasipoisson GLM: trial type the sole independent variable in both minimum adequate models, p = 0.115; hypothesis 6).

There were no significant effects of trial type and size disparity or mass disparity on the proportion of time in possession found for the gammarid added first (Quasibinomial GLM: trial type the sole independent variable in both minimum adequate models, p = 0.230; hypothesis 7). However, takeover success when G. tigrinus was in possession was significantly affected by the species out of possession (Binomial GLM: z = 2.249, p = 0.025; Suppl. material 1: table S5), with G. tigrinus taking possession significantly more than G. duebeni (Table 3; hypothesis 8). Takeover success when G. duebeni was in possession was not significantly affected by the species out of possession (Binomial GLM: the species previously not in possession was the sole independent variable in minimum adequate model, p = 0.205).

The introductions of NNS with overlapping ecological niches and functional similarity can lead to displacement of native species, something that has repeatedly been documented for those faced with the arrival of G. tigrinus (Pinkster et al. 1992, Jänes et al. 2015, Reisalu et al. 2016). In this study, we staged dyadic contests, using four combinations featuring the NNS and the native G. duebeni, over a single food resource, whereby one contestant was allowed up to twenty minutes to take possession of the resource undisturbed, with a second contestant added once the first contestant took possession, or after twenty minutes had elapsed. We found G. tigrinus to exhibit behaviours deemed to be beneficial across multiple stages of the invasion process and to exert a competitive advantage over the native gammarid which, through limiting access to resources, could affect population growth and survival.

During the initial period, we found G. tigrinus to be more likely to take possession of the resource, and of the individuals from both species that took possession, G. tigrinus exhibited a lower latency to do so than G. duebeni. The ability to efficiently identify, locate, take and retain possession of food resources is a valuable way of overcoming competitors in a novel ecosystem. Lower latencies to commence foraging have also been linked to greater boldness (Short and Petren 2008), and indeed, during this same period (i.e. before the addition of the second individual), G. tigrinus spent a lower proportion of time in the outer zone, indicating less centrophobic or thigmotactic behaviour than the native G. duebeni. Thigmotactic behaviour is a common indicator of where an individual falls along a boldness-shyness axis (dos Santos et al. 2023, Augustyniak et al. 2024), and boldness is a trait offering benefits at different stages of the introduction process. While some behaviours can have mixed effects depending on the stage (e.g. exploratory behaviour might enhance the likelihood of uptake into transport vectors but increase the likelihood of detection by biosecurity checks during transit: Chapple et al. 2011), boldness is thought to provide benefits during uptake, introduction, establishment and spread (Chapple et al. 2012, Gruber et al. 2018). Boldness is also often correlated with other “dispersal-enhancing traits” (Gruber et al. 2018, McGlade et al. 2022) such as activity and aggression within a behavioural syndrome (Sih et al. 2004), and the combination of high boldness and activity levels has been shown to enhance feeding opportunities (Brownscombe and Fox 2013) and survival in the presence of predators (Blake et al. 2018). We also discovered size-dependent activity and aggression differences between the species, with larger G. tigrinus found to be more active and aggressive, but larger G. duebeni found to be less so. While this aggression did not equate to increased success of dispossessing the native G. duebeni, which were equally likely to concede possession to conspecifics as they were to G. tigrinus, it may have offered some sort of deterrent, with G. tigrinus being better at resisting takeover attempts from G. duebeni than from other G. tigrinus individuals.

The inability of G. duebeni to dispossess G. tigrinus when in possession could be explained by competitive naiveté, something that remains relatively unstudied (Heavener et al. 2014). It may be that G. duebeni adapt to G. tigrinus as a competitor over time, but this is hard to predict, with some behavioural changes instant, some occurring over an individual’s lifetime, and some over multiple generations (Ruland and Jeschke 2020). Furthermore, considering the rates at which native gammarids have been replaced by G. tigrinus (e.g. Kazanavičiūtė et al. 2024), such adaptive processes may be too lengthy. For example, Heavener et al. (2014) showed that despite a close taxonomic relationship, native bush rats (Rattus fuscipes) failed to recognise the chemical cues of non-native black rats (Rattus rattus) in Australia, despite the two species interacting competitively for over 200 years. However, adaptation periods can be much shorter. The NNS American mink (Neovison vison) established in the UK at a time that otters (Lutra lutra) and polecats (Mustela putorius), both native competitors, were largely absent. Since then, populations of the native species have recovered and expanded, and mink have changed from being nocturnal to diurnal over the course of a decade, which is theorised as an adaptation to the rebounding natives (Harrington et al. 2009). It remains to be seen if G. duebeni can adapt competitively and quickly enough, or whether they can expand the niche differentiation with G. tigrinus through resource partitioning or avoidance, either in space or time.

In the present study, we did not see any significant effect of body mass or length on contest outcome or aggressive behaviour between G. tigrinus and G. duebeni. Those size measures are commonly used determinants of competitive ability or “resource holding potential” (Arnott and Elwood 2009, Zeng et al. 2019), and would be expected to influence contest dynamics. It could be that the size disparities and sample sizes between our competing species might have been too small to find such an effect, and it is an area that warrants further investigation in future studies. During contests, individuals are thought to assess their own and their opponent’s resource holding potential as well as the value of the contested resource, and this can dictate whether individuals risk injurious and potentially deadly fights (Arnott and Elwood 2008). In addition to morphological adaptations, which we propose are less relevant for gammarids, it may be that some species attach a naturally higher value to a contested resource than others, potentially driven by metabolism, or having more specialist demands in terms of diet or habitat. Other studies have also found species to be the most important predictor, rather than size. For example, this was shown for sympatric salamanders (Anthony et al. , 1997) and rockpool fishes (Paijmans and Wong 2017).

While the purpose of this study was to assess species-level differences, the methods employed could also be used to compare the competitive abilities of age-groups, sexes or populations. Indeed, there have been a number of calls to assess population-level differences within invasion ecology of late (Haubrock et al. 2024) and such differences have been noted in our study species. For example, individuals from the Dutch, Lough Neagh-originating G. tigrinus population are deemed less adapted to freshwater conditions than those from the German, England-originating population which subsequently invaded the eastern areas of the Netherlands (Pinkster et al. 1992). Therefore, applying dyadic contests across a salinity spectrum could reveal the environmental tipping point where one population is at a competitive advantage. Another avenue for study could be to look at the North American native range of G. tigrinus, where it is represented by six genetically distinct lineages grouped in two main clades (the “northern species” and “southern species”: Kelly et al. 2006). With the European populations descended from the “northern species” (Rewicz et al. 2019), it would be of interest to see how these clades compare, and whether differences in behaviour make one clade more prior-adapted to exert potential impacts. More generally, such a population focus can allow our method to help test invasion ecology hypotheses like the invasion front hypothesis (Lopez et al. 2012, Iacarella et al. 2015) or anthropogenically induced adaptation to invade (Hufbauer et al. 2012, Briski et al. 2018).

An important next step is to ground-truth the methods applied in this study and to use more real-world examples. While size seemed to play a minor role in our study, we note that adult G. duebeni can reach almost twice the size of adult G. tigrinus (Kolding 1981, Ward 1985, Grabowski et al. 2007), and we recommend that future studies account for larger individuals to determine their competitive performance and if the trend of decreasing activity and aggression holds. While size disparity can lead to different habitat utilisation and in turn segregation and coexistence (e.g. Platvoet et al. 2009), there are many case studies where much larger gammarid NNS have displaced smaller natives (however, conversely, native G. lacustris has been displaced by the physically smaller G. fasciatus in Lake Peipsi in Estonia and Russia: Panov et al. 2000, Panov and Berezina 2002). One such example is the larger Ponto-Caspian NNS D. villosus (adult males can reach 30 mm body length: Nesemann et al. 1995), which has had negative impacts on G. tigrinus and G. duebeni populations in the Netherlands (Dick and Platvoet 2000). Size disparity may also heighten the risk of contests descending into intraguild predation (Polis et al. 1989), and asymmetrical mutual predation has been shown to be a key driver of NNS replacing natives (Dick and Platvoet 1996, Nakata and Goshima 2006). Staging dyadic contests with large interspecific size disparities could help reveal the prevalence of this phenomenon.

Going forward, we propose that the findings of this study, and the methods implemented to derive them, offer applied potential in the form of NNS impact assessment. Indeed, recent impact assessment measures comparing resource consumption rates between NNS and trophically analogous natives have become incredibly popular within invasion ecology (Dick et al. 2014, Faria et al. 2023, 2025). Nevertheless, the per capita nature of the method, with study individuals left to feed in the absence of conspecific or interspecific competitors (Dickey et al. 2022), is an obvious shortcoming that could be addressed using the method at hand. Indeed, while the functional response-derived Relative Impact Potential metric (Dick et al. 2017, Dickey et al. 2020) has incorporated proxies for the numerical response to compare the population impacts of NNS relative to native species, it could be further developed to include a measure of competitive ability (or conversely, the degree of biotic resistance posed by native species within an ecosystem: Twardochleb et al. 2012, MacNeil et al. 2013), leading to a measure of potential impact across three axes. This would determine: 1) the “undisturbed” maximum feeding rate relative to a native, 2) the potential population size or reproductive rate of a NNS relative to a native, and 3) a measure of interspecific competitive strength to determine the probability of the two prior axes being realised. All three measures will likely change depending on abiotic conditions (e.g. salinity, temperature, noise pollution), and these can in turn be accounted for, allowing, for example, the prediction of future impacts with climate change. These three axes would account for many of the myriad mechanisms and traits that have been suggested as explaining the invasion success and impacts of G. tigrinus to date. For example, the NNS exhibits a broad reproduction period, known to be reproductively active all year round in the Gulf of Riga (Kotta et al. 2010), as well as a high reproduction rate, a short development time, high feeding rates with a propensity for “surplus killing” of prey (Dickey et al. 2021), and broad ecological tolerances in terms of salinity, temperature, and pollution (Grabowski et al. 2007, Reisalu et al. 2016).

In summary, we propose that this method, used here to demonstrate the competitive ability of a widespread NNS that has led to population declines of native gammarids across Europe and is expanding its range further (Rewicz et al. 2019), can offer a standardised, effective means of assessing inter- and intraspecific contests over limited resources. Bridging the gap between animal behaviour studies and applied NNS impact assessment, we propose it can be easily tailored depending on the hypotheses and study systems in question. While more ground-truthing is required, we believe it has the potential to become a useful component of horizon scans, and thus facilitate the impact assessment, prediction, and prioritisation of NNS as required by EU legislation and global biodiversity targets.

Thanks to the three anonymous reviewers and Prof. Tammy Robinson-Smythe for the constructive feedback during the review process which undoubtedly improved the manuscript.