The round gobies were collected in late May and at the end of August 2019 using electrofishing (a backpack-pulsed-DC electrofishing unit ELT60II GI HONDA GXV 50; Hans Grassl GmbH, Germany) in the river Elbe near Dolní Žleb (North Bohemia, Czech Republic GPS: 50°50'39.9"N, 14°13'01.5"E) from a recently established population (Buřič et al. 2015). Specimens representing the most common size cohort at the invaded locality (total length, TL = 55–75 mm) were transported to the experimental facilities of the Institute of Aquaculture and Protection of Waters FFPW USB in České Budějovice and acclimated in a recirculating aquaculture system (RAS, total volume of 1600 l) for at least one week prior to the experiments. Physical-chemical water parameters such as, temperature (20.22 ± 0.07 and 14.19 ± 0.04 °C), oxygen saturation (94.05 ± 0.25 and 92.05 ± 0.33%) and pH (7.67 ± 0.02 and 7.68 ± 0.02), were controlled on a daily basis using an HQ40d digital multimeter (Hach Lange GmbH, Düsseldorf, Germany). The experimental temperature was gradually adjusted by approximately 1 °C every four days during the initial acclimatization period.

In accordance with our observations of the invaded localities along the river Elbe, as well as findings from other river systems (Ghedotti et al. 1995; Ray and Corkum 1997; French and Jude 2001; Diggins et al. 2002; Barton et al. 2005), three different types of prey (molluscs, amphipods and insect larvae) were chosen to represent the predominant macrozoobenthic groups in the round goby’s diet (e.g. Phillips et al. 2003; Borza et al. 2009; Števove and Kováč 2016; Všetičková et al. 2018). Specifically, the killer shrimp Dikerogammarus villosus (Sowinsky, 1894), pill clam (Pisidium sp.) and chironomid larvae (Chironomus sp.) were used. Specimens of D. villosus were collected at the same localities as the round gobies were captured. Pisidium sp. were sampled in a tributary pond (Točník, South Bohemia, Czech Republic GPS: 48°56'57.8"N, 14°55'39.3"E), while live chironomid larvae were purchased from a pet store.

Based on the modified methodology proposed by Richter et al. (2018), we performed a series of six experimental trials to test gut evacuation rates in round gobies under different prey availabilities (two different feeding modes), prey types (three different prey items) and seasons. The latter were represented by two temperatures corresponding to the common/actual thermal conditions of the water at invaded localities on the river Elbe (taken from long-term data gathered during field surveys) in summer (T = 20 ± 0.035 °C; mean ± SE; season A; series of experiments 1–3 performed in July 2019) and spring/autumn (T = 14 ± 0.027 °C; mean ± SE; season B; series of experiments 4–6 performed in October 2019). Prey availability was varied by providing (i) food only at the beginning of the experiment (non-continual feeding) or (ii) continuously throughout the experiment (continual feeding).

Fish were starved for 48 h prior to each experimental trial to ensure the complete emptying of their alimentary tracts. To avoid possible cannibalism provoked by starvation and to ensure that a particular individual consumed a given amount of food, all fish were placed separately in experimental arenas (plastic boxes, each with a total volume of 2.4 l and water volume of 2.2 l) equipped with aeration and shelter. All experimental arenas had opaque walls to prevent visual contact between tested fish or any other visual disturbance that might represent an additional stressor. The light regime was set to 14 h light (intensity of 1000 lux.m2): 10 h dark, corresponding to the light regime during the vegetation season in the field, with simulations of dusk and dawn. The water parameters (T, O₂ and conductivity) in each experimental arena were monitored daily.

After the 48-hour starvation period, the round goby were fed with the tested food (food A) consisting of one of three different types of prey depending on the trial number (Table 1a, Fig. 1). Each type of prey was provided in quantities corresponding to similar weights to standardise the initial mass between treatments, regardless of the prey species (Table 1b). Subsequently, two hours later, any remains of tested prey (food A) were removed from all experimental arenas, which signalled the start of the main experiment phase and sampling. Fish whose tested prey (food A) intake was not observed were excluded from the experiment. Then, experimental arenas were randomly marked with colour clips to distinguish between non-continuous and continuous feeding modes. Following this, half of the individuals were not fed with any additional prey (the non-continuous feeding mode) (Fig. 1a), while the other half of the fish were continuously fed with a different type of prey (food B) (Table 1a, Fig. 1b), which was easily distinguishable to ensure the correct identification of the tested type of prey (food A) during subsequent fish dissection. To maintain constant food availability and experimental arenas were checked every four hours (Fig. 1b).

Figure 1. 

Gut evacuation rate experimental design. (a) Non-continual feeding mode; (b) Continual feeding mode. Solid red arrows represent the application of tested food (Food A) and additional food (Food B); dashed lines indicate the removal of tested food (food A) remains. Blue arrows represent the checking times (every 4 hours) when additional food (Food B) was provided if necessary to ensure continuous feeding. Green arrows represent the sampling time points.

In total, we tested 420 fish, of which 70 were sampled during each experimental trial (which included one tested prey type, one temperature and two feeding modes) over 36 h at seven time points: 0, 2, 5, 9, 16, 24 and 36 h after the start of the main part of the experiment trial. At each time point, five randomly chosen fish from each feeding mode (non-continuous and continuous) were sampled. The collected fish were killed by overdosing with an MS-222 anaesthetic solution before being immediately frozen in dry ice to inhibit all metabolic processes in the digestive system. Specimens were then stored in a deep freezer (-80 °C) until processing, which took place within maximum of two weeks.

Before the gut determination of the round gobies was performed, measurements of total length (TL; mm), standard length (SL; mm) and body weight (W; g) were taken, and their sex determined (Kornis et al. 2012). Subsequently, fish were individually dissected. The gut extraction was conducted following Manko (2016). The removed complete gut (together with the food content) was weighed (EX224M analytical balance, OHaus Corporation, USA), its length measured, and the position of each food type (food A and B) observed. Then, the gut content (foods A and B separately) was carefully placed on previously heated (30 minutes at 500 °C in a muffle furnace) and weighed glass-microfibre discs (grade MGF, diameter 50 mm, Sartorius, Goettingen, Germany) to determine its wet weight (WW) with an analytical microbalance (M3P, Sartorius, Goettingen, Germany). Finally, samples of the gut content (food A and B), the fish carcasses and the empty dissected gut were dried at 105 °C overnight and stored in a desiccator to determine their dry weight (DW) with a microbalance.

The evacuation rate (R), represented by the regression coefficient, was determined through statistical modelling of the relative gut content weight over time. The relative gut content (GI) was defined as the proportion of the dry weight (DW) of the tested food (food A) remaining in the gut at a given time vs. the fish’s dry body weight. The evacuation rate, expressed as the proportion lost from the gut per hour, can be described by several models, the most frequently used of which are linear and exponential functions (Elliott and Persson 1978; Bromley 1994). We found that the exponential function was the most suitable for our data as it had the fewest residual variations. Specifically, we applied the model represented by

GI _t = GI ₀ e ^-Rt

where GI_t is the relative gut content at time t, GI₀ is initial intestine content at time t₀ after feeding (h), and R is the evacuation rate (Elliott and Persson 1978).

We fitted the equation using generalised linear models (GLMs) with a Gaussian distribution (the log link function applied) and tested the relationship between the relative gut content GI used as a response variable and explanatory variables: (I) ‘Time’ (continuous numerical variable), (II) ‘Prey availability’ (two-level factor), and (III) ‘Prey type’ (three-level factor) separately for each season (see below). The improvement of the fitted relationships by the addition of interaction terms – one- and two-order interactions between all given explanatory variables – was tested using likelihood ratio tests. In order to avoid errors with log-transforming observations of empty guts at specific times, all zero values of relative gut content were replaced by 0.871 mg to represent half of the overall minimum gut content mass. The use of a Gaussian distribution was appropriate given that the histogram of model residuals was found to be close to the Gaussian; the assumption of homoscedasticity was also fulfilled. The significances of all relationships were tested by partial F-tests.

The independence of the initial consumption of prey and size of fish – used as explanatory variables – during the experimental treatments was tested given that this issue may potentially affect the obtained conclusions. Using GLMs with the binomial distribution (logistic link function applied), the consumption probability was detected as significantly greater when round gobies were fed with chironomids than when fed with either Dikerogammarus villosus or Pisidium sp. (Suppl. material 1; p < 0.05). Therefore, individual values of GI were weighted by the proportions between prey consumed by the round goby and the total amount of offered prey, i.e. a weighted GLM was used to fit a shift in GI. In addition, significantly different consumption probabilities and significant differences in SL (tested using GLM with gamma distribution with the log link function applied) were also detected between the two seasons (Suppl. material 3; p < 0.05). Thus, two separate models with GI as a response were fitted for each ‘season’ to yield the regression coefficients representing the evacuation rate. Nevertheless, the fitted round goby SL in season B was only 2.06 mm higher than in season A, whilst SL variance was very low and similar in both seasons (Suppl. material 2). Based on these findings and given that it was the focus of our study, the factor ‘season’ (which represents different temperatures and temporarily distinct round goby stocks) was also left as the explanatory variable in a single GI model to test the significance of its additive effect. No other significant differences in the consumption probability or SL range for any explanatory variables of the interactions were detected (p > 0.05; sporadically marginal interactions detected). All data analyses and graphical procedures were performed using R software (CRAN, version 4.1.1).

The second-order interaction of ‘Time × Prey availability × Prey type’ in modelling effects on the relative gut content (GI) was significant (p < 0.01, Table 3) and both ‘Prey type’ and ‘Time’ had significant effects on the relative gut content itself (Table 3).

The analyses revealed that in continual feeding mode, the highest evacuation rate was for the Pisidium sp. (R = 0.2694 ± 0.044 g·g^-1 fish·h^-1; mean i.e. grams of prey per gram of fish per hour ± SE; further h^-1; Table 4a) as complete food depletion took place 16 h after food intake (Fig. 2a), while chironomid larvae were completely evacuated within 24 hours, representing the second highest value in terms of evacuation rate (R = 0.2043 ± 0.029 h^-1). Dikerogammarus villosus had the lowest evacuation rate of all prey types (R = 0.1330 ± 0.013 h^-1), with total food depletion taking place only after 36 hours (Fig. 2a). Conversely, the lowest evacuation rate in non-continual feeding mode was observed for Pisidium sp. (R = 0.02407 ± 0.002 h^-1; Table 4a) and the fastest for Dikerogammarus villosus (R = 0.04516 ± 0.005 h^-1; Table 4a), the latter represented by the lowest relative gut content 36 h after food intake. None of the tested ‘Prey types’ were completely digested within 36 h in the non-continual feeding mode (Fig. 2a).

Figure 2. 

Change in the relative gut content (GI) over time in the round goby (Neogobius melanostomus) in two different seasons (season A represented by T = 20 °C and season B represented by T = 14 °C) fed with three different prey species: 1. Pisidium sp. (red); 2. Chironomus sp. larvae (green); 3. Dikerogammarus villosus (blue), and in two different feeding modes: 1. Non-con = non-continual feeding mode (shown by dashed lines) and 2. Con = continual feeding mode (shown by solid lines). Relative gut content expressed as a ratio of the dry weight (DW) of the tested food (food A) for each treatment at the time of sampling vs. fish dry body weight [g.g^-1]. Whiskers represent 95% confidence intervals around mean estimates.

The second-order interaction of ‘Time × Prey availability × Prey type’ in modelling effects on the relative gut content was significant (p < 0.001, Table 3) and the effect of ‘Prey type’ on the relative gut content was also significant (p < 0.001).

The evacuation rates in the continual feeding mode varied between (R = 0.1091 ± 0.009 h^-1) for Dikerogammarus villosus and the highest evacuation rate (R = 0.2746 ± 0.032 h^-1) for Pisidium sp. (Table 4b), which significantly differed from the evacuation rate of chironomid larvae under the same feeding mode (Fig. 3b). The round gobies tested in continual feeding mode evacuated all the experimental food within 36 h (D. villosus), with complete gut evacuation after 24 h for chironomid larvae and after 16 h for Pisidium sp. (Fig. 2b). The lowest evacuation rate in the non-continual feeding mode was observed for chironomid larvae (R = 0.01408 ± 0.002 h^-1) and the highest evacuation rate for Pisidium sp. (R = 0.02676 ± 0.002 h^-1) (Table 4b). None of the tested ‘Prey types’ were completely digested within 36 h in non-continual feeding mode (Fig. 2b).

Figure 3. 

Overlap between the 95% confidence intervals of the gastric evacuation rates (R; n = 420) for all experimental treatments (three different prey types and two feeding modes) in both experimental seasons (T = 14 °C and T = 20 °C) in the round goby (Neogobius melanostomus).

We observed prey-type-specific differences in evacuation rates between seasons. For Dikerogammarus villosus, the highest evacuation rate was observed in season A (20 °C) for both continual and non-continual prey availability (Fig. 2, Table 4). Conversely, all individuals fed with Pisidium sp. had faster evacuation rates in season B (14 °C) in both feeding modes (Table 4). Overall, the fastest evacuation rate for the Chironomus sp. larvae was observed in the continual prey availability at the lower temperature, i.e. in season B (14 °C) (Table 4). For the non-continual feeding mode, the evacuation rate of chironomid larvae observed in season A (20 °C) (R = 0.02873 ± 0.003 h^-1) was faster than for the lower temperature conditions (Table 4).

According to the 95% confidence intervals, there was no observed overlap between the two feeding modes in the mean values of the evacuation rates in each season (Fig. 3), which indicates that there were significant differences between them. All prey types were digested and evacuated significantly faster in round gobies fed continuously (Table 4). By contrast, in the non-continual feeding mode, no round goby had completely evacuated its gut content 36 hours after feeding, unlike the continuously fed fish (Fig. 2).

Our study demonstrates differences in evacuation rates in the round goby, with the continual feeding mode resulting in the higher evacuation rates compared to the non-continual feeding mode. Multiple-food application during the continual feeding mode accelerated the process, causing fish to evacuate particular food types 5–10 times faster than those fed only once (Table 4). These observations are consistent with previous studies (Noble 1973; Jones 1974; Elliott 1975; Persson 1981, 1982) demonstrating that the presence of additional food leads to higher evacuation rates (Tekinay et al. 2003). This can be seen by the steeper slope in Fig. 2, which indicates more rapid relative gut-content loss over time. This can be attributed to mechanical factors whereby the tested prey is pushed along the digestive tract by subsequently ingested food (Riche et al. 2004), possibly aided by the distribution of mucus-secreting goblet cells (Hur et al. 2016). However, the evacuation process is preceded by muscular and enzymatic activities (Nilsson and Brönmark 2000). Thus, one possible explanation for our findings could be intensified and accelerated metabolism involving greater enzymatic activity due to increased gut content (O’Connor et al. 2000), where the higher quantity of food, compared to fish fed only once, may affect digestive enzyme activity (Peres et al. 1998). Moreover, using two types of prey (Food A, Food B) can induce a potential synergistic metabolic response, where the digestion and metabolism of one prey type influence the processing of another, especially considering that different diet compositions (varied prey types) can selectively activate specific enzymes (German et al. 2004). Therefore, the biochemical composition appears to be a crucial factor (Caviedes-Vidal et al. 2000), affecting overall gut transition and digestion processes, which may be heightened in response to temporary increases in specific nutritional prey content (German et al. 2004). This effect is particularly significant for species known for their diverse diet, indicating modulation of overall digestive enzyme activities on a larger scale (Karasov 1992).

At the same time, differences in food digestion and evacuation rates according to food availability coincide with field observations since, in the early phases of invasions, individuals of a non-native species take advantage of low population densities and abundant food supply to exploit a wide range of accessible resources (Iacarella et al. 2015). This leads to higher feeding rates and rapid growth of the invading population (Phillips 2009; Burton et al. 2010; Raby et al. 2010). By contrast, in mature and established populations with higher conspecific densities resource availability is more restricted and leads to intensified intraspecific competition (Brownscombe and Fox 2012; Azour et al. 2015).

We expected to observe differences between fish fed with different prey species since prey body structure influences evacuation rates (Jones 1974; Singh-Renton 1990; dos Santos and Jobling 1992; Andersen et al. 1999). In our study, round gobies exhibited the fastest evacuation rates when fed on Pisidium sp. in both seasons (under different thermal conditions) and when food was continuously available. Those observations contradict the pattern observed in the majority of the studies whereby soft-bodied prey is digested and evacuated faster than hard-bodied prey (Jones 1974; Singh-Renton and Bromley 1999; Temming and Herrmann 2003). Typically, an external protective body structure such as a shell or exoskeleton acts as a barrier for digestive juices and slows down the digestion process (Jones 1974; dos Santos and Jobling 1992; Bromley 1994). One of the possible explanations behind our observations is that hard-shelled food resembles ‘high-fibre’ food or roughage and thus is transported faster through the digestive tract (Hilton et al. 1983). Moreover, it is important to consider species’ digestive capabilities, which depend on differences between fish species in morphological features and levels of organisation in the digestive tract structure, in which the primary concern is to ensure optimum use of dietary nutrients (Ray and Ringø 2014). Therefore, depending on diet preferences and feeding behaviour, particular species adapt (Olsen and Ringø 1997; Ringø et al. 2003), which enhances the efficiency of their food processing and thus determines digestive tract structure (Jiao et al. 2023). The omnivorous round goby is functionally classified as a stomachless fish due to its relatively simple, straight tubular-segmented digestive tract (Geevarghese 1983; Barton 2007), in which morphological adaptations are closely correlated to ontogenetic changes in the morphology of its robust molariform teeth (Ghedotti et al. 1995; Andraso et al. 2011a). Given that bivalves are the primary prey in round goby diets in both their native range (Demchenko and Tkachenko 2017) and in invaded localities (Diggins et al. 2002; Adámek et al. 2007; Kornis et al. 2012; van Deurs et al. 2021; Behrens et al. 2022), these adaptations enable it to exploit hard-shelled resources and feed on molluscs by crushing (Angradi 2018). This allows digestive enzymes to access soft tissues (Aarnio and Bonsdorff 1997) and ensure rapid gut transit through a short digestive tract that maximises food processing efficiency (Horn and Messer 1992), as observed in other stomachless species. This trait, which involves the mechanical breakdown of food before it reaches the intestines and is recognized as compensation for the lack of a stomach (Fänge and Grove 1979), likely explains the observed faster evacuation rates of Pisidium sp. in our study. Based on our post-experimental gut content analysis, it was evident that all Pisidium sp. individuals were thoroughly minced and crushed by the round gobies and, in almost all cases, after 16 hours it was impossible to discern the prey structure. In our opinion, this foraging adjustment almost certainly contributes to much faster evacuation rates.

In addition to the observed faster evacuation rates of Pisidium sp., it is crucial to consider the predator-prey size relationship. Although round gobies feed on a range of mussel sizes, experimental data reveal a preference for smaller individuals, most probably due to their ease of swallowing and thinner shells (Schwartzbach et al. 2020). Particularly interesting is the observation that some hard-bodied prey items such as Dreissenidae under 6 mm in size (Andraso et al. 2011b) remain intact after passing through the round goby intestine or like Pisidium sp. (Mack and Andraso 2015; Andraso et al. 2017) even survive gut passage. However, our observations differed as we failed to detect any persistent intact forms despite the individuals used in our experiments being smaller than 6 mm. Nonetheless, it is important to bear these findings in mind as they suggest that the actual predatory efficiency of round gobies on molluscs is unclear and could be overestimated in some cases. A similar pattern has been observed in Ancylus fluviatilis Müller, 1774 in the river Elbe, which appears to be a preferred food resource for the round goby (Richter et al. 2022). In this case, shell remnants persist in the rear sections of the digestive tract for an extended period and form stacked structures. This can lead to overestimations when comprehensively examining the full contents of the entire gut.

By contrast, the round gobies fed on another hard-bodied prey – Dikerogammarus villosus – exhibited the slowest gut evacuation rates when continuously fed, which was accompanied by notable mechanical food processing in the sampled digestive tracts. The observed delay in the complete evacuation of crustacean prey (36 h) can be attributed to the resistance of the cuticular exoskeleton to enzymatic breakdown (Bromley 1991; Andersen 1999; Temming and Herrmann 2003), while the subsequent increase in evacuation duration time is correlated with rising ash content within the prey (Couturier et al. 2013).This finding concurs with studies conducted in other fish species that report slower evacuation of prey with structural support than of prey without (Singh-Renton and Bromley 1999; Temming and Herrmann 2003; Gillum et al. 2012). Furthermore, the observed decrease in the evacuation rate of amphipod prey in our studies may suggest greater use of consumed nutrients (Le et al. 2019) since previous research has revealed a reduction in the efficiency of digestion and absorption processes if the gut transit time is shorter (Garber 1983; Lee et al. 2000). Therefore, the prolonged gut evacuation rate coupled with evident food breakdown observed in our study suggests that Dikerogammarus villosus – alongside chironomids – can also serve as a nutritionally valuable food source and prey item providing fatty acid (Sushchik et al. 2006; Maazouzi et al. 2007). These potentially superior nutritional benefits are also supported by studies demonstrating the strong preference of round gobies for amphipod prey, which results in faster growth rates and better overall body condition than in diets based on bivalves (Polačik et al. 2009). This preference may be attributable to prolonged satiety, in which a slower evacuation rate could be an adaptation used during the species’ transport and active dispersal to new localities, as the digestive tract capacity of gobies and their temporary food storage function are diminished by their secondary loss of an anatomically distinct stomach (Barton 2007), and lack of intestinal bulb (Trzeciak et al. 2012)—a simple dilatation at the anterior part of the intestine— commonly recognized as a substitute for the stomach in agastric fish (Kapoor et al. 1976; Mokhtar et al. 2021).

In some areas, round goby feed almost exclusively on amphipods, which can constitute up to 70–80% of their diets (Brandner et al. 2013). This finding and our results together suggest that even when round gobies exhibit preferences and prey exclusively on amphipods (Diggins et al. 2002; Polačik et al. 2009; Emde et al. 2014), they may still exert less pressure on this prey group in terms of evacuated biomass quantity per unit of time than on other prey, e.g. the tested chironomid larvae. This implies potentially less predatory pressure on amphipods (even when they are present throughout all seasons in the benthic community) than on other groups of macrozoobenthic organisms whose availability is more seasonally limited (Zinchenko et al. 2017; Pander et al. 2022). However, our suppositions are only supported by our laboratory findings regarding evacuation rates. Since the round goby is recognised as an omnivorous species that feeds on the most readily available prey (Kornis et al. 2012; Nurkse et al. 2016), we believe that – rather than food preferences – local prey abundances together with evacuation rates primarily govern the potential predatory pressure exercised by this species. Hence, dietary analysis and macrozoobenthic availability in local populations will provide more accurate and comprehensive information regarding potential predatory pressure in any particular locality.

We cannot exclude that the differences in gastric evacuation rates between prey types (Food A) were affected also by using different types of second prey (Food B) as our methods included invariable pairs of prey combinations that may occur in nature and the round goby diet (Szydłowska et al. 2024). For example, studies on N. melanostomus (Trzeciak et al. 2012) and other gobiid species (Jaroszewska et al. 2008; Wołczuk et al. 2015) have confirmed the presence of an oesogaster with alveolar glands and functions for enzyme secretion and mucus production. These features serve as an effective functional replacement for the conventional stomach, enhancing digestion, particularly evident in the utilization of protein-rich food sources (Wołczuk et al. 2015). Based on this, chironomid larvae, known for their high protein content and used as a subsequent meal (Food B; De la Noüe and Choubert 1985; Nath et al. 2015), could potentially contribute to the rapid evacuation of Pisidium sp., as observed in continual feeding mode. However, when considering the analysis of results from the non-continuous feeding mode, which can be regarded as a ‘baseline’ where each prey type was tested separately, it appears that chironomid larvae were digested and evacuated at different rates depending on the season, while the evacuation rate for Pisidium sp. followed by Diptera larvae in continual feeding mode remained the highest regardless of these conditions. Thus, other indicators, such as the size of prey—for instance, even when meal portions are of the same mass, smaller prey are likely to be evacuated and metabolized faster (Nilsson and Brönmark 2000)—are also possibly affecting the evacuation rate of Pisidium sp. when tested as Food A. Following that, similar justifications for the possible effects could be applied to the other prey combinations tested within our study, depending on the specific example. Thus, the limited findings highlight the necessity for further investigation involving additional prey species, such as Food B, consistent across all three prey types examined in our study. This would be essential for gaining more robust insights into potential field scenarios. Moreover, factors such as surface area to volume ratio (Andersen 1999), shape (Karlsen and Andersen 2012), and energy density might also play a significant role in explaining our results, and their potential effects should be investigated in separate studies.

According to the literature, temperature is probably one of the most studied environmental factors (Shrable et al. 1969; Brett and Higgs 1970; Persson 1979; Bascinar et al. 2016; Dürrani and Seyhan 2019; Horstman 2020) since it has a major effect on metabolism, biochemical reactions and digestion in fish, and so naturally affects evacuation rates (Pääkkönen and Marjomäki 1997; Gillum et al. 2012). We predicted faster evacuation rates at 20 °C than at 14 °C given that this rate is commonly observed to speed up as temperatures increase (He and Wurtsbaugh 1993; De et al. 2016; Mazumder et al. 2020). Indeed, based on comparisons of the yielded coefficients, our study suggests a potential effect of temperature on the round goby evacuation rate. However, contrary to our expectations, no consistent pattern emerged across different prey types between seasons, thereby indicating the effects of temperature as a prey-type dependent. Following that, results confirmed anticipated thermal dependency in round gobies fed on Dikerogammarus villosus, with faster evacuation rates observed at higher temperatures in both feeding modes. This agree with the findings reported by Stehlik et al. (2015) for the clearnose skate Rostroraja eglanteria (Bosc, 1800), where similar ranges of evacuation rates, observed at almost the same temperatures as in our experiment, also reflect the typical thermal dependency (Wu et al. 2015; Horstman 2020). Conversely, round gobies fed on Pisidium sp. in both feeding modes exhibited faster evacuation at lower temperatures. This observation can possibly find its rationale in the significant seasonal differences in observed consumption probabilities (evident from the graphical representation of the initial relative gut content; Fig. 2). Specifically, round gobies tested in season B (14 °C; spring/autumn conditions) consumed more Pisidium sp. than those tested in season A (20 °C, summer conditions), as shown by the observed initial relative gut content. This matches field observations whereby larger round gobies in spring and autumn chiefly base their diet on molluscs (Borcherding et al. 2013). Since the round gobies tested in season B were also slightly bigger (2.06 mm on average) than those tested in season A and the predator and prey body size is regarded as an important variable affecting evacuation rates (Beamish 1972; Fänge and Grove 1979; Andersen 1984; Bromley 1994; dos Santos and Jobling 1995; Seyhan et al. 2020), we cannot completely exclude their possible impact on the effects observed. However, we believe that the detected difference in the size (SL) of the fish used in the two seasons – by a magnitude of just a few mm – although statistically significant, is biologically irrelevant and likely did not affect the actual result of our study in any way.

As the continual feeding mode reflects conditions observed in the wild by simulating different prey type combinations, we can expect similar outcomes in nature. Following that, the inconsistency observed in temperature-related patterns in the evacuation rate of each prey type might be explained not only by the higher relevance of other tested variables but also by the round goby’s wide environmental tolerance, as previously documented by several studies (Charlebois et al. 2001; Christensen et al. 2021), which is reflected in its broad temperature range (Moskal’kova 1996). Laboratory research has further corroborated these findings and demonstrated metabolic adaptability and ability of the species to thrive under disparate conditions, including temperatures ranging from 2.3 to 5.3 °C (Walsh et al. 2007; Silva et al. 2019) and its active feeding year-round even at low temperatures under natural conditions (Dashinov and Uzunova 2020). However, it is important to remember that seasonal changes in the effectiveness of prey use and diet are not only temperature-dependent but are also affected by particular prey types and their seasonal dynamics within the macroinvertebrate assemblages and thus by variations in the availability of food resources (Didenko et al. 2017). Therefore, the outcomes of our analysis underscore the importance of considering the interaction between all the abovementioned and tested conditions.

Compared to local native species such as ruffe Gymnocephalus cernua (Linnaeus, 1758) in the estuary of the river Elbe (Hölker and Temming 1996) and the river St. Louis (Henson and Newman 2000), our study found that, under the same thermal conditions, the round goby evacuates chironomid larvae up to three times faster. While the abovementioned studies primarily focussed on gastric evacuation rates (which assess stomachs), our findings relate to the gut evacuation rate that encompasses the whole alimentary tract due to its different structure. This variance in digestive anatomy may affect the comparability between species of evacuation rates. Nevertheless, this disparity might still indicate the superior efficiency of the round goby in acquiring food and competing for resources, consistent with findings from the experimental studies conducted by Bergstrom and Mesinger (2009), as the evacuation rates were found to regulate food intake rates (Aas et al. 2013; Hedden et al. 2020). This efficiency, combined with its feeding plasticity and ability to feed on the most abundant local prey (Carman et al. 2006), could be linked to the observed depletion (Lauer et al. 2004; Rakauskas et al. 2013; Ramler and Keckeis 2019) – or even in some particular cases collapse – of indigenous fish populations (Jůza et al. 2018) and negative trends in macrozoobenthic assemblages (Pennuto et al. 2010; van Deurs et al. 2021).

On the other hand, research by Lagrue and Bollache (2006) into the persistence of gammarids in European bullhead Cottus gobio Linnaeus, 1758 stomachs over 24–30 hours at 14 °C suggests that the round goby may in fact be less efficient at food processing than other analogous native fish species that prey on amphipods. This implies a potential lack of superiority of a non-native predator over its native counterparts in competing for food resources in localities where they occur sympatrically (Janáč et al. 2018; Roje et al. 2021). Our study corroborates this observation given that, among all the offered prey types, the slowest evacuation rate while continuously feeding at the 14 °C exhibited by round gobies was for crustaceans. However, Lagrue and Bollache (2006) noted that the persistence time was strongly influenced by temperature and prey type (e.g. the invasive Gammarus roeseli Gervais, 1835 had significantly greater persistence at 14 °C than its native counterparts), which highlights potential variations in outcomes if Dikerogammarus villosus would be included given its more robust and thicker exoskeleton and resistance (Błońska et al. 2015). This raises questions regarding our previous statement, as the inclusion of D. villosus could influence the persistence time in native fish predators. However, it is important to emphasise that while the gut evacuation rate is a good predictor and plays a primary role in regulating feeding rates (Hedden et al. 2020), the overall efficiency and impact of a predator depend on a series of factors including fish capture effectiveness and foraging behaviour (Gebauer et al. 2018; Franta et al. 2021), as well as the abundance of local prey populations.