Studies using artificial streams were conducted in an indoor glass-house located in the Gyeongnam Department of Environmental Toxicology & Chemistry, Korea Institute of Toxicology, Jinju, Korea (Fig. S1). Each stream consisted of five glass sections, beginning with a tile section (2 m length × 0.3 m width × 0.1 m height; slope of approximately 4.3%). The tile section included shallow, fast-moving water disrupted by the presence of a rough substrate, causing ripples and waves without breaking the surface tension. The tile section was followed by a gravel section (2 m length × 0.3 m width × 0.1 m height), comprising sections of moving water with distinct riffles and pools, a water velocity of < 20 cm/s, and no surface turbulence. The gravel section was followed by a pool section (1 m length × 0.3 m width × 0.4 m height), a sediment section (2 m length × 0.4 m width × 0.2 m height), and a tail tank (1 m length × 0.4 m width × 0.5 m height). The bottom of the tile section was entirely lined with unglazed ceramic tiles (8.7 × 8.7 cm). The gravel obtained from the Nakdong River was placed in the gravel section. The bottom of the pool section was covered with standard sand (Joo Mun Jin Silicasand Co., Republic of Korea). The standard sand and sediment obtained from the Nam River were placed in the sediment section. The overall volume of the streams was ca. 0.84 m³ and dechlorinated tap water (5 L/min) was continuously introduced into a tile section. The top of each section was open and the recirculation system was not used. The water discharged from the tail tank was transferred to a wastewater treatment plant. This structure, with various widths and depths, realistically reflects the characteristics of actual streams and enables the use of various biological groups constituting the aquatic ecosystem for experiments.

Among the aquatic organisms inhabiting domestic rivers, periphytic algae, three types of invertebrate fauna, and two kinds of fish, constituting the ecosystem food chain, were selected as experimental organisms for each section of the artificial stream: tile section – periphytic algae; gravel section – M. macrocopa; pool section – Z. platypus; sediment section – G. tokunagai and L. hoffmeisteri, and tail tank – A. chinensis.

To stabilize the artificial stream, the substrate in each section was transplanted two months before the exposure experiment, and dechlorinated tap water was flowed at a rate of 5 L/min. For biological stabilization, the inoculation and stabilization times for each organism were determined based on their individual life-cycle.

Two months before the exposure experiment, sampled gravel with algae from a regional stream with similar water quality (pH: 7.54 ± 0.58; water temperature: 19.5 ± 4.0°C; nitrogen: 1.00 ± 0.32 mg-N/L; phosphorus: 0.01 ± 0.01 mg-P/L; TOC: 1.37 ± 0.63 mg/L) [26] to that of the dechlorinated tap water (pH: 7.42 ± 0.10; water temperature: 20.6 ± 0.6°C; nitrogen: 1.27 ± 0.78 mg-N/L; phosphorus: 0.01 ± 0.01 mg-P/L; TOC: 1.1 ± 0.04 mg/L), was placed on the top of the tile section to attach the periphytic algae to the sixty three unglazed ceramic tiles (7.9 cm length × 7.9 cm width). Over 70% of the tile surface in all streams was colonized by the algal community at the beginning of the experiment. The dominant algal species observed using a Leica MDG41 stereomicroscope (Leica Microsystems, Wetzlar, Germany) were Melosira varians, Microcystis viridis, and Volvox sp., respectively. These are commonly found in Korean domestic rivers [27]. M. macrocopa were cultured in the laboratory, and four animals aged < 24 h at the beginning of the experiment were introduced into each of the five cylindrical glass containers (3 cm diameter × 7 cm height), which were covered with a stainless steel mesh (80 mesh). Z. platypus and A. chinensis, 3–4 cm in length, were obtained from the Institute for Biodiversity Research and acclimated in square glass tanks for at least four weeks before the exposure experiment. On −14 d, 50 animals each of Z. platypus and A. chinensis were transferred to the pool and tail tanks. On −4 d, 10 second instar larvae of G. tokunagai, which were cultured in the laboratory, were introduced into each of a twenty five sand layer (contained in a circular stainless-steel container (9 cm diameter × 4 cm height) of sediment section. On −7 d, 10 L. hoffmeisteri obtained from an aquatic animal supplier (Alpha Fish, Yeosu, Korea), were introduced into the sediment layer (contained in a circular stainless-steel container (9 cm diameter × 4 cm height) of the sediment section. The system stabilization was confirmed by pH, dissolved oxygen (DO), conductivity, and water temperature measurements using a WTW Multi 340i portable meter with the following electrodes: For pH and temperature SenTix^® 41, for DO DIN plug CellOx^® 325, and for conductivity TetraCon^® 325; and the biological stabilization was confirmed by monitoring the number of species in each section of the artificial stream before the exposure experiment.

The experimental conditions for benzene exposure used a stabilized artificial stream that reflected an actual structure and biological group of the river. The exposure scenario assumes that benzene enters a river directly for a period of time through a hole in a storage tank caused by a tank truck accident and then flows downstream. The concentration and time of exposure to benzene were calculated by referring to the physicochemical properties of benzene, the results of the preliminary acute toxicity test (Table S1), and the technical guidelines for accident scenario selection presented by the National Institute of Chemical Safety [28]. The detail procedures for determining the exposure scenario were described in supplementary materials. Initial exposure concentrations were 150, 330, and 726 mg/L as low, medium, and high concentrations, respectively, and the total exposure time was determined to be 4 h (Table S2). Stock solution (1.5 g/L) for exposure experiments was prepared using benzene (for HPLC analysis, purity: > 99.9%, Sigma Aldrich, USA) with an aqueous solubility of 1.88 g/L at 23.5°C. For low, medium, and high concentrations, the stock solutions of 0.5, 1.1, and 2.42 L/min were mixed with 4.5, 3.9, and 2.58 L/min of dechlorinated tap water, respectively. The mixed solution (5 L/min) was then introduced to the top of the tile section and allowed to flow downstream. Water samples were collected in triplicate at 1, 3, 5, and 24 h after start of benzene exposure and analyzed for benzene concentration using an Agilent 1200 HPLC-DAD equipped with an XBridge^TM C18 (4.6 mm inner diameter, 150 mm length, 5 μm stationary phase particle size). In the operation mobile phase: acetonitrile/water (70:30) was used at a flowrate of 1 mL/min. Water temperature, pH, DO, and conductivity (WTW Multi 304i with SenTix^® 41, CellOx^® 325, and TetraCon^® 325) were periodically measured during the experiment.

The detail experimental conditions and biological parameters of aquatic organisms monitored in the artificial stream during the experimental period are summarized in Table S3. To identify the effect on the algal biomass, three replicate tiles were randomly sampled from the tile section on 0, 1, 7, 14, and 21 d and transferred to the laboratory. Chlorophyll-a (Chl-a) of periphytic algae as an indicator of algal biomass was analyzed using the following procedure. The periphytic algae were separated by rubbing the tile with a brush and transferred to 50 mL of distilled water. Approximately 25 mL of sample was centrifuged at 4000 × g for 20 min, and the supernatant was removed. The sample was filled with 15 mL of 90% acetone and homogenized. Then, the sample was covered with aluminum foil and dark-acclimated in a refrigerator for 24 h. After dark-acclimation, the sample was centrifuged at 4000 × g for 20 min and the supernatant was transferred to a glass vial, and the optical density (OD) at 750 nm, 664 nm, 647 nm, and 630 nm was measured using an Epoch microplate spectrophotometer (BioTek Instruments, Winooski, VT, USA). Finally, Chl-a was calculated through the following formula modified from [29]: Chl-a (mg/m²) = 11.85 × (OD₆₆₄ – OD₇₅₀) − 1.54 × (OD₆₄₇ – OD₇₅₀) − 0.08 × (OD₆₃₀ – OD₇₅₀) × volume of acetone (0.015 L) / volume of sample (0.025 L) × volume of total sample (0.05 L) / surface area of tile (0.087 m × 0.087 m). Mortality of adult M. macrocopa and newborn offspring was observed daily in each of the five glass containers, and the dead organisms and offspring were immediately removed. The daily mortality of Z. platypus in the pool section and A. chinensis in the tail tank was checked, and the total length and wet weight of the two fish species were measured after 30 d of the experiment to calculate the condition factor (K). Condition factor was calculated based on the following formula [30]: K = 100 × W × L³, where K = condition factor, L = fork length of fish (cm), W = body weight of fish (g). Four replicate sand and sediment samples (9 cm in diameter and 4 cm in depth) were taken from each stream at the different sampling positions. Live larvae of G. tokunagai were identified by sand sample at 0, 1, 7, 14, 28, and 36 d. The number and dry weight of L. hoffmeisteri were measured by sediment sample at 0, 1, 7, 14, 21, and 28 d.

Growth, mortality, and fecundity and water quality were assessed using statistical analyses. Normality and homogeneity of variance tests were determined using the Shapiro-Wilk test and Levene tests, respectively. If the assumptions of normality and homogeneity of variance were met, One Way Analysis of Variance (ANOVA) and post hoc test (Dunnett's test) were performed. If the data did not meet these assumptions, the data were log-transformed and retested for normality and homogeneity of variance. Finally, if these assumptions were not met, non-parametric test (Kruskal-Wallis ANOVA on Ranks test) and post hoc test (Dunn’s test) were performed. All analyses were conducted using the SigmaPlot software ver. 15 (Grafiti LLC., CA, USA). Toxicity values for no observed effect concentration (NOEC), low observed effect concentration (LOEC), median lethal concentration (LC50), and median effective concentration (EC50) were calculated using CETIS v.1.8.7.15 (Tidepool Scientific software, CA, USA). To calculate these toxicity metrics, the time-weighted average concentrations for low, medium, and high concentrations were calculated by multiplying the measured peak concentration by 4 h divided by 7 d [31].

The water quality parameters and benzene concentrations during the exposure experiments at low, medium, and high concentrations are summarized in Table S4. Throughout the experiment, temperature, dissolved oxygen (DO), pH, and conductivity remained stable across all treatments. The mean water temperatures for the control and the low, medium, and high exposure conditions were 20.9 ± 0.7°C, 20.9 ± 0.7°C, 21.0 ± 0.7°C, and 20.9 ± 0.7°C, respectively, with no statistically significant differences between control and treatments (p > 0.05). Similarly, DO levels across treatments (low 9.56 ± 0.81 mg/L, medium 9.62 ± 0.94 mg/L, high 9.60 ± 0.92 mg/L) did not significantly differ from the control (9.47 ± 0.77 mg/L) (p > 0.05). The pH and conductivity values showed no significant variations during the experiment, with mean pH values of 7.64 ± 0.32 (control), 7.71 ± 0.37 (low), 7.71 ± 0.36 (medium), and 7.73 ± 0.36 (high), and conductivity remaining constant at 125 ± 2 μS/cm across all treatments (p > 0.05).

Benzene concentrations showed a clear upward trend as the exposure concentration increased. During the exposure period, benzene concentrations ranged from 15.1 ± 1.3 to 16.7 ± 0.1 mg/L, 17.2 ± 1.0 to 22.7 ± 1.1 mg/L, and 25.0 ± 0.6 to 26.5 ± 0.8 mg/L for the low, medium, and high concentration treatments, respectively. At 5 h post-exposure, concentrations had decreased to 4.5 ± 0.1 mg/L (low), 5.8 ± 0.3 mg/L (medium), and 6.3 ± 0.2 mg/L (high), and were undetectable at 24 h. The peak benzene concentrations observed were 16.8 mg/L (low), 23.8 mg/L (medium), and 27.0 mg/L (high). The benzene concentrations identified in this study exhibited a gradual increase over time following the initial spill, reaching a peak concentration and subsequently declining after the termination of the spill. This trend is consistent with the observed changes in benzene concentrations in monitoring results and model simulations following actual benzene spills [32–34]. Based on these peaks, the time-weighted average concentrations for low, medium, and high exposure were calculated as 0.40, 0.57, and 0.64 mg/L, respectively. These values were subsequently used to calculate the toxicity metrics, including NOEC, LOEC, EC50, and LC50, which are discussed in detail in later sections.

The variation in Chl-a concentration of the periphytic algae is shown in Fig. 1. Initial Chl-a concentrations at low, medium, and high exposure levels (8.4 ± 2.8 mg/m², 7.7 ± 2.4 mg/m², and 4.0 ± 0.6 mg/m², respectively) were similar to or higher than those in the control group (4.3 ± 1.8 mg/m²) (p > 0.05) (Table 1). Following benzene exposure, Chl-a concentrations at 1 d showed a slight increase at the low exposure level (6.9 ± 3.1 mg/m²), while the medium and high levels (8.4 ± 0.5 mg/m² and 4.5 ± 1.7 mg/m², respectively) were comparable to the control (5.3 ± 0.7 mg/m²). However, no significant differences were observed between the control and any of the treated groups (p > 0.05) (Table 1). Throughout the remainder of the experiment, Chl-a concentrations exhibited fluctuations across all treatments, but no statistically significant changes were detected compared to the control (p > 0.05).

These findings suggest that benzene did not exert any discernible impact on the growth of the periphytic algae. Due to its lower density (0.876 g/cm³ at 20°C) relative to water, benzene is able to float easily in an aqueous environment. Furthermore, the benzene quadrupole results in the formation of a partial negative charge on both faces of the π system and a partial positive charge around the aromatic ring [35, 36]. These properties reduce the probability of contact with algae that adhere to the tile surface, thereby limiting the potential for toxic effects on periphytic algae.

Previous studies have also demonstrated relatively high tolerance of algae to benzene exposure in both short-and long-term tests. The 3 h EC50 for photosynthesis in Chlamydomonas angulosa and Chlorella vulgaris was 460.9 mg/L and 312.4 mg/L, respectively [37], while the 4 h EC50 for Ankistrodesmus falcatus was 310.1 mg/L [38]. The 1 d EC50 for growth in Chlorella vulgaris was 525 mg/L [39], and the LOEC for Scenedesmus quadricauda exceeded 1,400 mg/L [40]. Additionally, in Raphidocelis subcapitata, the EC50 following an 8 d exposure was determined to be 41 mg/L [41]. Even at lower benzene concentrations (50 and 100 mg/L), slight increases in algal density and diversity were observed in small ponds [42].

The survival and condition factors of Zacco platypus and Aphyocypris chinensis after benzene exposure are presented in Fig. 2 and Table 2. The short-term exposure to benzene resulted in significant changes in the survival of Z. platypus. Within the first 4 h, survival at medium and high exposure levels dropped to 92% and 62%, respectively, indicating acute toxicity. Over the 30 d experimental period, the survival rates stabilized, with final survival at the control, low, medium, and high exposure levels being 100%, 100%, 86%, and 62%, respectively. Although initial mortality was observed, no significant differences were seen in the long-term survival rates after 1 d, suggesting that short-term exposure of benzene had an immediate but not sustained lethal effect on Z. platypus.

The condition factor of Z. platypus after 30 d showed no major variations between exposure levels, with values of 0.74 ± 0.01, 0.70 ± 0.02, 0.76 ± 0.02, and 0.74 ± 0.03 for the control, low, medium, and high exposure groups, respectively. A statistically significant reduction in condition factor was noted at the low exposure level compared to the control (p < 0.05) (Table 1), but overall, short-term benzene exposure did not appear to affect the growth of Z. platypus. This suggests that while short-term exposure led to acute mortality in Z. platypus, its growth remained unaffected by benzene.

In contrast, A. chinensis exhibited a higher tolerance to benzene exposure. No acute toxicity effects were observed within the first 24 h, with survival remaining at 100% across all exposure levels for 1 d. Survival rates at 30 d were still high, at 92%, 96%, 100%, and 94% for the control, low, medium, and high exposure groups, respectively. Condition factors at 30 d were 1.04 ± 0.04, 1.09 ± 0.07, 1.16 ± 0.09, and 1.17 ± 0.05, with a significant increase noted at the medium and high exposure levels (p < 0.05) (Table 1). However, despite the increase in condition factor, the growth of A. chinensis was not inhibited by benzene exposure, indicating that this species is less sensitive to short-term benzene toxicity compared to Z. platypus.

In this study, the 1 d NOEC and LC50 for survival in Z. platypus were determined to be 0.40 mg/L and 0.66 mg/L, respectively, showing that this species is sensitive to benzene exposure. While there was an initial mortality in response to benzene, long-term growth was not impacted, with 30 d NOEC and LOEC values for growth being 0.64 mg/L and > 0.64 mg/L, respectively. In contrast, A. chinensis exhibited greater resistance to short-term exposure in the artificial stream system, with 1 d NOEC and LOEC for survival at 0.64 mg/L and >0.64 mg/L, respectively. When comparing the results of laboratory acute toxicity tests (96 h LC50) conducted on continuous exposure to benzene for 96 h, A. chinensis (11.9 mg/L) showed a similar toxic sensitivity to that observed in Z, platypus (14.9 mg/L). However, a comparison of the toxic responses observed at 1 d revealed that mortality occurred after 2 h of benzene exposure in Z. platypus, whereas mortality occurred after 1 d of benzene exposure in A. chinensis. This suggests that A. chinensis exhibits a relatively higher resistance to the initial toxicity caused by benzene. In addition, growth inhibition of A. chinensis in the artificial stream experiment was not observed, with NOEC and LOEC for growth also at 0.64 mg/L and > 0.64 mg/L, respectively.

Previous studies have shown variability in benzene sensitivity across fish species. For example, the 1 d LC50 for Carassius auratus and Ictalurus punctatus were reported as 46 mg/L and 425 mg/L, respectively [43, 44], while for Lepomis macrochirus the values ranged from 20 to 400 mg/L [44, 45]. Oryzias latipes showed an LC50 of 70 mg/L [46], whereas Oncorhynchus mykiss was found to be the most sensitive species, with a 1 d LC50 of 9.2 mg/L [44]. The 4 d LC50 of Oncorhynchus mykiss, Pimephales promelas, and Poecilia reticulata was 5.9 mg/L [47], 24.6 mg/L [48], and 28.6 mg/L [47], respectively. These values suggest that the sensitivity of Z. platypus to benzene lies at the lower end compared to other species, while A. chinensis demonstrated a higher tolerance.

Although the bioaccumulation of benzene in fish is generally regarded as relatively low, significant variations in benzene accumulation have been documented among different fish species and organs. The bioconcentration factor (BCF) of Leuciscus idus melanotus was determined to be less than 10 when exposed to 50 μg/L of 14C-labeled benzene in a closed system for 3 d [49]. The saltwater species Clupea harrengus has the highest BCF, which was measured to be approximately 11 [50]. In contrast, exposure to 14C-labeled benzene in the marine fish species Engraulis mordax was reported to be organ-specific, with relatively high BCFs in the gallbladder (8450), intestine (505), and liver (309) [51]. Additionally, the BCFs in Morone saxatilis were observed to be lower than those in E. mordax, with the highest concentration in the gallbladder (53.4) [51]. The BCF in Clupe harengus, a marine species exposed to 14C-labeled benzene, was considerably lower than in the aforementioned two species and higher in the following organs: gallbladder (31), gills (7), intestine, pyloric cecum, brain (6), liver (5), muscle, kidney (4), and gonads (2) [52]. Furthermore, the documented mechanisms of benzene toxicity are as follows: Following the absorption of benzene into the body, it is activated by cytochrome P450 enzymes, with the toxicity-related effects of benzene being mediated by its metabolites [53]. Benzene accumulates primarily in lipid-rich tissues, where it causes hematologic toxicities such as leukopenia, anemia, and thrombocytopenia [54]. In particular, in fish, benzene is absorbed across the gill surface and directly into the blood, where it attaches to red blood cells, some of which attach to lipoproteins [55]. Subsequently, it is transported through the blood and accumulates in tissues or undergoes biooxidation in the liver, kidneys, and muscle tissue to form phenol, with some benzene excreted through the gills in an unconverted form [56,57].

While the mechanism of benzene toxicity is not consistent for all species, a potential toxicological mechanism for the acute effects observed in Z. platypus involves the rapid accumulation of benzene in the bloodstream through respiration, which leads to central nervous system depression, narcosis, decreased respiratory rate, acute anemia, and eventually heart failure [58, 59]. These symptoms are consistent with the observations made in this study, and similar responses have been reported in other species, including Chinook salmon (Oncorhynchus tshawytscha) and striped bass (Morone saxatilis) [59]. In contrast, A. chinensis’s resilience to benzene toxicity may be due to more efficient detoxification mechanisms. Studies have demonstrated that tolerant species tend to accumulate detoxified metabolites more effectively than sensitive species [60]. For instance, Acipenser transmontanus, Oncorhynchus tshawytscha, and Salmo salar have been shown to have higher levels of detoxified metabolites compared to more sensitive species like Oncorhynchus kisutch and Salvelinus fontinalis [60]. This suggests that differences in the basal expression of biotransformation enzymes may play a role in the varying levels of benzene sensitivity observed between fish species. Further research is necessary to investigate the specific detoxification pathways in A. chinensis that may contribute to its observed tolerance to benzene exposure in this study.

The survivorship and fecundity of M. macrocopa under varying benzene exposure levels are presented in Fig. S2 and S3. Mortality was not observed until 3 d (Fig. S2), but survival rates decreased significantly from 4 d to 14 d under medium and high benzene exposure, both resulting in survival rates below 60% (Fig. S2). The LC50 values for survival decreased from 1.22 mg/L at 7 d to 0.71 mg/L at 12 d, remaining constant up to 14 d (Table 3). Initial fecundity was lower in the treatment groups compared to the control (Fig. S3), with high benzene exposure significantly reducing fecundity (p < 0.05) (Table 1). The cumulative offspring per live female at 5 d in the control, low, medium, and high exposure groups were 4.4 ± 1.9, 2.0 ± 1.7, 1.7 ± 1.3, and 1.0 ± 0.7, respectively (Fig. S3). Interestingly, fecundity recovered after 8 d at the high exposure level, showing no significant difference compared to the control (p > 0.05) (Table 1). The EC50 for fecundity increased from 0.44 mg/L at 5 d to 0.66 mg/L at 8 d, and remained stable up to 14 d (Table 3). The cumulative offspring per live female at 14 d in the control, low, medium, and high exposure levels were 11.0 ± 2.4, 10.5 ± 1.3, 15.3 ± 5.0, and 10.3 ± 2.0, respectively (Fig. S3).

The mean number of G. tokunagai larvae per sample is shown in Fig. 3. The mean number of individuals per sample at 0 d in the control, low, medium, and high exposure treatments were 14.3 ± 2.2, 13.3 ± 1.5, 16.0 ± 3.8 and 12.0 ± 2.0 larvae/sample, respectively (Fig. 3). After 4 h of exposure to benzene, G. tokunagai larval numbers decreased in a concentration-dependent way. The mean number of individuals per sample at 1 d in the control, low, medium, and high exposure levels were 16.8 ± 5.6, 13.8 ± 3.8, 10.0 ± 1.2, and 9.0 ± 3.3 larvae/sample. In particular, the high exposure level resulted in a significant decrease in the mean number of individuals per sample at 1 d, 7 d, and 14 d compared to the control (p < 0.05) (Table 1). The 7 d and 14 d EC50s for mean abundance were 0.59 mg/L and 0.54 mg/L, respectively. After 14 d, the number of larvae was not significantly different from that of the control (p > 0.05) (Table 1).

The mean number of individuals per sample of L. hoffmeisteri and the dry weight of L. hoffmeisteri are shown in Fig. 4. The initial number of L. hoffmeisteri at control, low, medium, and high exposure levels were 9.5 ± 1.0, 9.8 ± 0.5, 9.3 ± 1.5, 9.5 ± 1.0 individuals/sample, respectively (Fig. 4). The number of L. hoffmeisteri under all conditions did not change up to 21 d of the experiment, and there was no significant difference between the control and treatment groups (p > 0.05) (Table 1). The mean number of individuals per sample at 28 d in the control, low, medium, and high exposure level were 9.3 ± 1.5, 13.0 ± 2.9, 9.5 ± 1.7 and 15.5 ± 1.7 individuals/sample, respectively (Fig. 4). Even though the number of L. hoffmeisteri at high exposure level was significantly higher than that of the control (p < 0.05) (Table 1), benzene did not affect the survival of L. hoffmeisteri. The initial dry weights of L. hoffmeisteri at the control, low, medium, and high exposure level were 0.81 ± 0.28, 1.07 ± 0.11, 0.68 ± 0.06, 0.73 ± 0.21 mg/individual/sample, respectively. An increase in dry weight under all conditions was observed after exposure to benzene. The dry weights at control, low, medium and high exposure level at 28 d were 0.85 ± 0.21, 1.11 ± 0.14, 0.81 ± 0.14 and 1.12 ± 0.21 mg/individual/sample, respectively (Fig. 4) and the dry weight of L. hoffmeisteri under all conditions were similar to those of the control (p > 0.05) (Table 1).

Benzene, with a low to moderate bioaccumulation potential [61], accumulates differently across trophic levels. Water fleas have a relatively high bioconcentration factor (BCF: 153–225) [62] compared to that of algae (BCF: 30) and fish (goldfish BCF: 4.3) [63]. Benzene can be accumulated in the body of Daphnia pulex, and after 72 h, 15% of the accumulated benzene still remains in the body [62]. PAHs, including benzene, are narcotic chemicals that accumulate in the lipid bilayer of cells [64] and can cause toxicity if certain concentration thresholds are met. The observed decrease in survivorship and fecundity of M. macrocopa can be attributed to the bioaccumulation of benzene and its metabolites, which likely led to the activation of cytochrome P450 enzymes, resulting in DNA mutations and eventual mortality [64,65]. The recovery of fecundity in later stages suggests that benzene and its metabolites were metabolized and eliminated from the body over time, reducing the toxic effects. Similar toxic effects have been observed in G. tokunagai, where the absence of cytochrome P450 CYP6EV11 gene expression has been linked to increased mortality [66]. L. hoffmeisteri's tolerance to benzene, with 48 h LC50 of 6638 mg/L, suggests a higher resistance compared to other species [67,68]. These findings demonstrate that benzene's toxic effects on invertebrates are species-dependent, influenced by bioaccumulation potential and metabolic activity.

Episodic benzene pollution in freshwater environments poses a significant concern for both ecosystem and human health, though its effects are not yet fully understood. Most existing studies focus on continuous benzene exposure in laboratory conditions [12,43–45,48,58], while field studies investigating the impacts of episodic pollution events on aquatic organisms are scarce. In Korea, field assessments following episodic pollution typically rely on the identification of dead fish and the effects of water samples on daphnids [69], leaving a gap in understanding the broader effects of temporary benzene exposure on aquatic life.

In this study, artificial stream experiments were conducted with minimal variation in environmental conditions to simulate real-stream scenarios. This approach allowed for the observation of benzene's fate in freshwater ecosystems and its toxic effects on aquatic species across different trophic levels. Our results revealed differences in susceptibility among species, providing insight into their potential as sentinel species for detecting benzene pollution. Z. platypus was the most sensitive species during the initial exposure period, showing high vulnerability to benzene. Delayed toxicity effects were observed in increased mortality and decreased fecundity in M. macrocopa and decreased abundance in G. tokunagai. Notably, M. macrocopa showed signs of reproductive recovery following toxic effects.

These findings suggest that Z. platypus could serve as an early sentinel species for detecting benzene spills, while M. macrocopa and G. tokunagai could be effective in monitoring delayed toxic effects.

This study utilized an artificial stream system to assess the toxic effects of short-term benzene exposure on various aquatic species. However, several limitations exist, and future research directions are suggested based on these limitations.

This study offers baseline data for acute benzene toxicity and highlights species with potential as environmental indicators, laying groundwork for further research and monitoring strategies.