The round goby (Neogobius melanostomus), a fish species originating from the Ponto-Caspian region, is considered one of the “100 worst” invasive species in Europe (Vilà et al. 2009) due to its ability to successfully colonize many European and North American basins (Corkum et al. 2004; Kornis et al. 2012). It often becomes a dominant species at invaded sites (Sapota and Skóra 2005; Kornis et al. 2013; Jůza et al. 2018). Negative impacts of N. melanostomus include displacement of native fishes mainly via diet overlap (Dubs and Corkum 1996; French III and Jude 2001; Ustups et al. 2016), modification of benthic communities (Lederer et al. 2008; Pagnucco et al. 2016; Henseler et al. 2021), new trophic pathways for pollutants (Hogan et al. 2007; Almqvist et al. 2010; Hares et al. 2015), and disturbance of food webs (Kuhns and Berg 1999; Pagnucco et al. 2016; Pennuto et al. 2018), often leading to ecosystem instability (Pagnucco et al. 2016; Pennuto et al. 2018).

High consumption rates, multiple spawning events during the season, and non-selective foraging of N. melanostomus represent key drivers of its negative impact on invaded environments (Corkum et al. 2004; Laverty et al. 2017; Dashinov and Uzunova 2020; Franta et al. 2021). Moreover, human-made changes to stream morphology (Winter et al. 2006; Oeberst et al. 2012), interconnectivity of inland waters through canals (Bij de Vaate et al. 2002; Panov et al. 2009), and general biodiversity loss (Olden et al. 2004) lead to decreased abundance of predators at higher trophic levels (Weis 2011) and increased ecosystem invasibility, which facilitate the impact of invaders, including N. melanostomus (Wallach et al. 2010; Weis 2011).

In ecosystems with a stable trophic pyramid, predation risk may significantly reduce the impact of mesopredators on lower trophic levels (Preisser et al. 2005; Creel and Christianson 2008; Mitchell and Harborne 2020) through the consumptive effect (CE), i.e., direct predation, and non-consumptive effects (NCEs; Sih et al. 2010) caused by visual and chemical cues of predation risk. NCEs can include increased use of shelters, decreased foraging activity, and increased vulnerability to other predators (Rahel and Stein 1988; Preisser et al. 2005; Pūtys et al. 2015; Mitchell and Harborne 2020). These phenomena can reduce the overall impact of invasive species (Aquiloni et al. 2010; Musseau et al. 2015). Aquatic systems represent a unique habitat in which chemical cues serve as key transmitters of information, particularly under conditions where visual sensing is often limited (Burks and Lodge 2002). Hence, fish olfactory systems are among the most developed among vertebrates (Kasumyan 2004). Moreover, predator-derived chemical cues can be transported over long distances in lotic systems and act as early warning signals for prey before visual contact, thereby evoking NCEs (Brown et al. 1995).

Neogobius melanostomus has a wide range of potential predators. Shortly after invasion and establishment, N. melanostomus was recorded in the diets of many native predators (Reyjol et al. 2010; Roseman et al. 2014; Mikl et al. 2017a), sometimes even as a dominant dietary component (Dietrich et al. 2006; Reyjol et al. 2010; Hares et al. 2015; Crane and Einhouse 2016; Hempel et al. 2016; Herlevi et al. 2023). For example, the burbot (Lota lota) is considered a voracious predator of N. melanostomus (Madenjian et al. 2011; Mikl et al. 2017a). However, knowledge of the NCEs of native predators on N. melanostomus is scarce and mostly limited to behavioral responses, without evaluating whether and how these NCEs alter its CE on native prey (Pūtys et al. 2015; Michels et al. 2021; Galli et al. 2023). This represents a critical gap in understanding whether restocking native predators can mitigate the ecological impact of N. melanostomus.

To elucidate the effect of NCEs on the ecological impact of N. melanostomus as an invasive mesopredator, the functional response of N. melanostomus and maximum food intake were compared in the presence and absence of chemical cues released by dead conspecific prey, a native predator, or both. Functional responses represent a fundamental tool for analyzing trophic interactions (DeLong 2021), including the effects of environmental conditions and NCEs on predation pressure imposed by native and invasive mesopredators on native prey and its consequences for predator-prey dynamics (Sentis et al. 2017; Faria et al. 2023a, 2023b). Functional responses quantify how predation rates change with prey density. They can be broadly described by three parameters: attack rate, which characterizes the increase in predation rate at low prey densities; handling time, which determines the maximum predation rate at high prey densities; and a scaling exponent, which characterizes the overall shape of the relationship. For example, a change in the shape of the functional response from a saturating Holling type II to a sigmoidal Holling type III and consequently low predation pressure at low prey densities can stabilize predator-prey dynamics (DeLong 2021). Functional responses have already been used to quantify how trophic interactions of N. melanostomus vary with temperature (Gebauer et al. 2018), habitat complexity (Gebauer et al. 2019), and body size (Franta et al. 2023).

The European eel (Anguilla anguilla) was chosen as a representative native predator. Several aspects identify A. anguilla as a successful predator of N. melanostomus: i) both species share a benthic lifestyle (Emde et al. 2014) and similar temperature optima (Seymour 1989; Lee and Johnson 2005); ii) both occur sympatrically in the Elbe River; iii) the morphology of A. anguilla allows for effective foraging in riprap habitats, frequently used as shelters by N. melanostomus (Janáč et al. 2019); iv) gobiids frequently occur in A. anguilla stomachs (Doornbos and Twisk 1987), and unpublished records document N. melanostomus in A. anguilla stomachs from the Baltic Sea; and v) both fish species share the same parasite species (Emde et al. 2014; Janáč et al. 2019). Neogobius melanostomus was exposed to conspecific and heterospecific chemical cues under laboratory conditions to determine whether predator-derived cues originating from the consumption of conspecific prey are required to elicit NCEs in N. melanostomus or whether heterospecific cues alone are sufficient. It was hypothesized that these cues would reduce the feeding of N. melanostomus by altering the functional response of individuals exposed to conspecific or heterospecific cues through a lower attack rate or longer handling time compared with individuals under the control condition without chemical cues.

Net weight of N. melanostomus individuals (mean ± SD = 2.30 ± 0.38 g, range = 1.52–2.99 g) did not differ significantly between treatments (Type II Wald test:χ22 = 2.14, P = 0.34) or initial prey densities (Type II Wald test: χ52 = 5.59, P = 0.35) but differed between experimental rounds (Type II Wald test: χ12 = 16.7, P < 10^-3).

The proportion of dead prey in the controls, P_dead, was low (0.015 ± 0.038, range = 0–0.289) and exceeded 0.10 in only two replicates (details not shown). According to the best model, P_dead increased significantly with initial prey biomass W₀ (Suppl. material 1: table S1), but the slope was small in magnitude (Suppl. material 1: table S4), and even at the highest initial biomass, W₀ = 650 mg, the predicted proportion of dead prey was only 0.022 (95% CI = 0.013–0.038). Thus, the data were not corrected for background prey mortality in the functional response trials.

Predation by N. melanostomus on chironomid larvae was characterized by a Type II functional response in each treatment, as the proportion of eaten prey declined with initial prey biomass (Fig. 3; logistic regression: control, mean slope ± SE = –0.0052 ± 0.0001, z = –33.8, P < 10^-3; heterospecific cues, –0.0048 ± 0.0001, z = –34.2, P < 10^-3; conspecific cues, –0.0048 ± 0.0001, z = –34.4, P < 10^-3). Model selection identified only one plausible model (FR8 in Suppl. material 1: table S2), with an attack rate that differed significantly between the conspecific cues treatment and the other two treatments (a₁₂ ≠ a₃) and treatment-specific handling time (h₁ ≠ h₂ ≠ h₃), although the handling times in the control and conspecific cues treatments were very similar (Fig. 2; Suppl. material 1: table S5). No evidence was found for a size-dependent attack rate a, as the allometric exponent b of the best model was not significantly different from 0, but handling time h declined significantly with body size (Fig. 2; Suppl. material 1: table S5). The other competing models also yielded similar estimates of both allometric exponents (details not shown).

The amount of prey consumed by individual fish during the 24 h trial at the two highest initial prey densities varied between 6.2% and 21.4% of net fish weight (13.1 ± 3.7%). Model selection identified two plausible models (Suppl. material 1: table S3); both showed that the amount of ingested prey increased with fish weight, but food intake did not depend on the treatment according to the best model (Fig. 4; Suppl. material 1: table S6).

Biological invasions are considered a major driver of ecological degradation (Pyšek et al. 2020) and are associated with substantial annual economic costs (Nentwig 2007; Cuthbert et al. 2021). Moreover, once established, invasive species are often almost impossible to eradicate (Harris 1988). Restoration of aquatic ecosystems, including native predator populations, can mitigate the ecological impact of non-native hydrobionts (Wallach et al. 2010; Weis 2011), including highly invasive fish such as the round goby (N. melanostomus), which is listed among the 100 worst invasive species in Europe (Vilà et al. 2009). However, the overall impact of native predators on N. melanostomus, including their CEs and NCEs, should be investigated in detail to assess their role in suppressing the negative effects of N. melanostomus on native freshwater biota. While records of CEs are relatively well documented across the regions colonized by N. melanostomus, evidence for NCEs remains ambiguous.

Chironomid larvae are one of the most important diet components of N. melanostomus (Janssen and Jude 2001; Phillips et al. 2003; Števove and Kováč 2016). Consistent with previous studies (Fernandez-Declerck et al. 2023), a Type II functional response toward this prey was observed in the present study. A Type II functional response was also previously reported in N. melanostomus feeding on other prey, including isopods, fish larvae, and amphipods (Gebauer et al. 2018, 2019; Paton et al. 2019; Franta et al. 2023). Generally, strong interactions correspond to higher resource utilization by N. melanostomus compared with analogous native species, e.g., the bullheads Cottus gobio and Cottus bairdi (Laverty et al. 2017) or the logperch Percina caprodes (Paton et al. 2019), resulting in substantial dietary overlap and consequent displacement of native species. The resulting effect of a predator on its prey combines CEs and NCEs, with NCEs potentially being equal to or stronger than CEs (Sih et al. 2010). No significant effect of A. anguilla-derived cues on the gastric evacuation rate of N. melanostomus was detected by Szydłowska et al. (2024). In the present study, evidence was found for a change in the functional response of N. melanostomus depending on predator-derived chemical cues with or without conspecific alarm cues. Functional response parameters, i.e., handling time and attack rate, have been shown to vary with anthropogenic noise, body size, habitat complexity, and population origin in N. melanostomus (Gebauer et al. 2019; Paton et al. 2019; Fernandez-Declerck et al. 2023; Franta et al. 2023). The present results confirm chemical cues as an additional influential factor.

Fish olfaction is the most developed olfactory system among vertebrates and plays an important role in reproductive, parental, feeding, defensive, territorial, schooling, and migration behaviors (Kasumyan 2004). Generally, gobies respond equally to visual and chemical cues (Utne and Bacchi 1997; McCormick and Manassa 2008). In N. melanostomus, chemical cues (Gammon et al. 2005) and visual signals play an important role in reproductive strategy (Yavno and Corkum 2010), and behavioral changes in response to predator risk have been documented. Previous studies mostly focused on shelter use or other antipredator responses, such as physiological and behavioral changes (Michels et al. 2021; Augustyniak et al. 2022; Kłosiński et al. 2022; Galli et al. 2023). For instance, operculum movement and blood cortisol levels in N. melanostomus increased in response to conspecific alarm cues (Pūtys et al. 2015; Galli et al. 2023). However, despite the cortisol increase, Galli et al. (2023) found no behavioral changes in N. melanostomus. Moreover, another non-native Ponto-Caspian goby, Neogobius fluviatilis, responded less intensively to conspecific alarm cues than the native Gobio gobio in terms of vertical movement, immobility time, and social distance response (Kłosiński et al. 2022).

However, no previous study has focused on how predation risk affects trophic interaction strength in N. melanostomus. In the present study, the trophic interaction strength of N. melanostomus at low prey densities, reflected by attack rate values, declined with predation risk only when the fish were exposed to a combination of predator-derived chemical cues and conspecific alarm cues. This likely indicates avoidance behavior by N. melanostomus and reduced foraging effort in habitats with low prey availability and very high predator risk. However, the handling time of N. melanostomus increased in the presence of heterospecific chemical cues of predation risk, whereas the handling times of N. melanostomus in the control and conspecific predation risk treatments reached comparable values. It cannot be determined whether this effect was caused by alarm cues originating from the heterospecific prey, chemical cues released by A. anguilla after consuming heterospecific prey, or a combination of both. Additionally, earthworm-derived chemical cues may have acted as a feeding stimulus for N. melanostomus.

Both attack rate and handling time have been shown to depend on body size in N. melanostomus (Franta et al. 2023). Handling time is driven by morphological traits that covary with body size, such as gape size (Kislalioglu and Gibson 1976; Aljetlawi et al. 2004) and digestion capacity (Aljetlawi et al. 2004; Li et al. 2018). Despite a significantly narrower size range of N. melanostomus in this study compared with that in Franta et al. (2023), handling time declined significantly with body size. Unlike handling time, no significant differences were found in maximum intake rates among the experimental treatments. Although handling time and maximum intake rate both reflect predator foraging capacity, they are derived from different data and approaches. Finally, the results were consistent with size-independent attack rates. This result likely reflects the high ability of N. melanostomus to effectively search for prey in the present experimental setting regardless of body size.

In addition to body size, individual boldness may also influence responses to predation risk. The results are based on N. melanostomus originating from an established population. As invasive fronts are often characterized by a greater proportion of bold individuals (Brandner et al. 2013; Myles-Gonzalez et al. 2015), NCEs may be weaker in such populations than in long-established populations. However, strong NCEs usually trade off against weaker CEs and vice versa (Creel and Christianson 2008; Sih et al. 2010). The higher relative metabolic rate of bold individuals is associated with higher energetic demands and increased foraging activity, which expose them to greater predation risk (Bell and Sih 2007; Brandner et al. 2013). Therefore, the total impact of predators on N. melanostomus, evaluated as the sum of CEs and NCEs, may remain similar regardless of whether individuals originate from invasive-front or established populations (Sih et al. 2010).

Anguilla anguilla occurs sympatrically with N. melanostomus in its native area, although at lower densities (Apostolou et al. 2013). Moreover, both species coexist and share the same habitats at invaded sites in the Elbe River, where the study fish originated, at least since 2015 (Buřič et al. 2015). Although the abundance of A. anguilla in the experiment exceeded its current typical abundance in European rivers (Prchalová et al. 2013), the results provide the first experimental evidence of a decrease in the foraging activity of N. melanostomus under predation risk. Moreover, N. melanostomus is an important diet component of piscivorous fish at invaded sites (Dietrich et al. 2006; Reyjol et al. 2010; Hares et al. 2015; Crane and Einhouse 2016; Mikl et al. 2017a; Herlevi et al. 2023), which may ultimately impose NCEs on the trophic interactions of N. melanostomus similar to those imposed by A. anguilla. Thus, the cumulative NCEs of all predators on N. melanostomus could be similar to or even greater in the wild than under the experimental conditions. Records of predator-induced increases in NCEs in non-native crayfish (Faria et al. 2023b), effective control of an invasive crayfish population by A. anguilla (Aquiloni et al. 2010; Musseau et al. 2015), and the high CE of L. lota on N. melanostomus (Madenjian et al. 2011; Mikl et al. 2017a), together with the present findings, suggest a potential role of native predators in mitigating the impacts of N. melanostomus in invaded European freshwater ecosystems. However, current abundances of both native predators are low in Central Europe. River modification and homogenization of freshwater ecosystems, high pressure from angling, commercial fisheries, and piscivorous birds are important factors contributing to decreases in their populations (Winter et al. 2006; Stapanian et al. 2010; Oeberst et al. 2012).

Even stable, pristine ecosystems cannot fully prevent invasion by non-native species (Sih et al. 2010). However, the invasibility and impact of invasive species can be reduced by restoration of freshwater habitats (Stachowicz et al. 2002; Olden et al. 2004; Wallach et al. 2010), as many invasive species benefit from predator-free environments (Weis 2011). Based on the present results, N. melanostomus is no exception. Indeed, a reduction in the consumption rate of N. melanostomus due to predation risk could lower its predation pressure on macrozoobenthos (Lederer et al. 2008; Mikl et al. 2017b; Henseler et al. 2021), competition with native benthic fish species (Dubs and Corkum 1996; French III and Jude 2001; Ustups et al. 2016), and transfer of energy, nutrients, and pollutants to higher trophic levels (Hogan et al. 2007; Almqvist et al. 2010; Hares et al. 2015). Although predatory fish populations are currently under high pressure (Stapanian et al. 2010; Oeberst et al. 2012), they have probably already affected the trophic interactions of N. melanostomus populations through a combination of NCEs (this study) and CEs (Madenjian et al. 2011; Hempel et al. 2016; Mikl et al. 2017a). Despite the limitations of the study, including short-term exposure and the use of a single prey species and a single higher-order predator species, the findings are consistent with previous evidence of native predator CEs (Dietrich et al. 2006; Reyjol et al. 2010; Madenjian et al. 2011; Hares et al. 2015; Crane and Einhouse 2016; Hempel et al. 2016; Mikl et al. 2017a; Herlevi et al. 2023), which collectively underscore the importance of protecting and managing predator populations to control the invasive N. melanostomus. Liversage et al. (2017) suggested modifying management practices to increase the CE of percids on N. melanostomus in the northeastern Baltic Sea. A similar approach could be applied to enhance predator NCEs on invasive fish species. Stabilizing native predatory fish populations to control other non-native fish depends mainly on sufficient freshwater ecosystem management, including habitat restoration and appropriate fisheries regulation (Winter et al. 2006; Stapanian et al. 2010; Oeberst et al. 2012).

In summary, this study reveals that predator-derived chemical cues can significantly alter the functional response of N. melanostomus. The decline in attack rate observed in the presence of conspecific chemical cues combined with alarm cues indicates avoidance behavior and reduced foraging effort in habitats with low prey availability and very high predator risk. In contrast, exposure to heterospecific chemical cues led to a longer handling time in habitats with high prey availability. Although further research is needed to disentangle the specific roles of individual prey- and predator-derived cues, the findings provide the first evidence that predation risk alone can reduce the foraging activity of N. melanostomus. Together with previous evidence of CEs exerted by native predators, the results support the idea that native predator communities can provide predator-driven biotic resistance against invasive species. These findings highlight the ecological importance of maintaining robust native predator populations as a means of enhancing both CEs and NCEs on this invasive species. Strengthening predator communities through effective freshwater ecosystem management may therefore help limit the predation capacity and ecological impact of N. melanostomus on native macrozoobenthos.