The wastewater used for the study was obtained as raw wastewater before entering the nearby sedimentation pond from a typical SSGM site in Tarkwa, Ghana. The physicochemical properties ( Table 1 ): pH, Turbidity, TSS, TDS, and EC were measured using a Crison pH meter, HACH DR/2000 spectrophotometer, Hydro Test HT 1000, Hanna RODI EC/TDS meter and Crison Conductometer Basic C30, respectively. The metal ions (As, Cd, Cr, Cu, Fe, Hg) concentration of the wastewater, as highlighted in Table 2 , was analysed using a Varian AA240FS Fast Sequential Atomic Absorption Spectrometer. Notably, the same analytical methods were used to obtain effluent results before and after coagulation.

Dry Cassia fistula seeds ( Figure 1(a) ) used in this study were collected from the University of Mines and Technology (UMaT), Tarkwa campus. The seeds were washed and air-dried for two days. The dried seeds were milled and then sieved using a 500 µm aperture size screen (undersize of < 500 µm) ( Figure 1(b) ). The ground powder was mixed with distilled water to make a 1% suspension, which was then vigorously shaken for 45 min using a magnetic stirrer to promote water extraction of the coagulant proteins and to achieve solubilization of active ingredients in the seed. The solution was then passed through a 0.45 µm filter paper (Whatman no. 42, 125 mm dia.). The filtrate portions were used for the required dose of natural coagulants. Fresh solutions were prepared daily and kept refrigerated to prevent any aging effects (such as changes in pH, viscosity, and coagulation activity). Solutions were shaken vigorously before use. Notably, the methodology used in this study is similar to that presented by Gandiwa et al. (2020) .

For the batch coagulation test, a graduated cylinder (1 L) was employed (as shown in Figure 2 ), filled with the wastewater sample, and dosed with the prepared CFSE coagulant at varying dosages of 5, 10, and 15 g/L. The mixture of wastewater and the coagulant was vigorously shaken to ensure homogenisation. The shaking intensity was calibrated to 180 rpm (rapid) and 20 rpm (slow) using a digital tachometer. This approach aligns with screening protocols for natural coagulants in field conditions with limited resources, although it is less precise than mechanical stirring ( Saritha et al., 2017 ; Getahun et al., 2024 ).

A timer was started to keep track of the duration of the experiment. The colloidal particles were allowed to settle, and the position of the suspension-liquid interface was measured at different time intervals. This methodology is illustrated in Figure 2 , where the position of the suspension-liquid interface is indicated by the red arrow. The measurement times for a batch settling curve were 0, 5, 10, 15, 20, 25, and 30 min, respectively. Notably, at the start of the test, the suspension-liquid interface was typically measured more frequently, as the colloidal particles were settling at a relatively fast pace. Later in the test, the frequency of the measurements was decreased because the interface was moving more slowly. After 30 min of settling, the levels of physicochemical parameters and heavy metals of the supernatants formed were measured and compared with the initial parameters ( Table 1 and Table 2 ). Notably, the results presented are the mean/average (the accuracy was ± 5%) of triplicate experiments conducted for each condition to ensure reproducibility. This methodology is similar to that reported in Najm et al. (1998) and Ndabigengesere et al. (1995) . To ensure the reliability of the results, a comparative study between jar test (mechanical stirring at 150 rpm rapid/30 rpm slow) and the method used for this work was conducted. The result showed a <10% deviation (especially in turbidity removal efficiency). Again, all the tests in this work were performed by a single trained operator, hence reducing the inter-operator variability.

The settling behaviour and kinetics of suspensions coagulated with CFSE at varying dosages (5, 10, 15 g/L) were analysed from the height-time relationships shown in Figure 3 . Tests were conducted in 1 L graduated cylinders following manual homogenisation (vigorous shaking for 2 min) at 25 ± 2˚C, with interface height measurements at 0, 5, 10, 15, 20, 25, and 30 min. The blue-line plot in Figure 3 is the settling behaviour of the suspension without CFSE.

Figure 3 shows the height-time relationship for suspensions coagulated with CFSE at varying dosages. The settling behaviour for all dosages exhibited a characteristic three-phase settling pattern similar to those described by Zodi et al. (2009) , Bernard et al. (2019) , Font et al. (1999) , and Fitch (1979) . The first phase (initial linear settling zone) is characterised by the regular height reduction of the solid/liquid interface (0 - 10 min) with a constant settling velocity ( Font et al., 1999 ), followed by the transition-settling period (10 - 20 min), with decreasing settling velocity ( Bernard et al., 2019 ) and third phase (the compression settling zone) (>20 min), which is characterised by a very low variation of the interface (minimal interface movement). The three-phase settling pattern aligns with charge-neutralisation mechanisms typical of plant-based coagulants ( Benalia et al., 2024 ). The rapid initial settling (0 - 10 min) corresponds to microfloc formation through cationic protein adsorption ( Ndabigengesere et al., 1995 ), while the velocity decline in the transition phase (10 - 20 min) reflects hindered settling due to increased solid concentration ( Guibai & Gregory, 1991 ). The compression phase (>20 min) indicates sludge network formation where further settling requires structural rearrangement. In terms of dosage-dependent performance, the 10 g/L curve (curve below 5/15 g/L) showed the lowest interface position at 10 - 20 min, indicating optimal floc formation and/or optimal charge neutralisation. Contrarily, the lower dosage (5 g/L) provides insufficient active sites, while overdosing (15 g/L) causes colloidal re-stabilisation through steric effects/hinderance ( Saxena & Brighu, 2023 ).

The initial settling rates, calculated from the slope of the linear portion (0 - 10 min) of the settling curves ( Bernard et al., 2019 ), were 1.98, 3.03, and 3.05 cm/min for 5, 10, and 15 g/L dosages, respectively. The highest initial settling rate at 10 g/L (3.03 cm/min) indicates optimal coagulation performance, exceeding values reported for Moringa oleifera (1.8 cm/min at 20 g/L) by Gandiwa et al. 2020 and aligning with optimized tannin-based coagulants (2.5 - 3.1 cm/min) reported by Ibrahim et al. (2021) . Additionally, the overall average settling rates (calculated as total height change over 30 min) were 0.136, 0.157, and 0.129 cm/min for 5, 10, and 15 g/L, respectively. The peak at 10 g/L further confirms its efficiency in forming dense, rapidly settling flocs. The inverse relationship between initial and overall rates reflects compression phase dominance at higher dosages.

The effect of the CFSE as a coagulant on reducing the levels of selected physicochemical parameters (pH, turbidity, TDS, EC, TSS) was also examined.

Figure 4 shows the effect of the CFSE coagulant on the pH of the wastewater at varying dosages. The initial pH of the wastewater received was 6.35 (as indicated by the brown line in Figure 4 ). Generally, the addition of the CFSE to the wastewater increased the pH of the wastewater regardless of the dosage rates. Thus, the pH increased: from 6.35 to 6.44 for 5 g/L, 6.35 to 6.57 for 10 g/L, and from 6.35 to 6.89 for 15 g/L. The pH increase, though seemingly modest in absolute terms, was statistically significant (p < 0.05) and represents a 1.2 to 3.5-fold reduction in H⁺ ion concentration due to the logarithmic nature of pH. This alkaline shift is important as it enhances cationic protein activity in CFSE for charge neutralisation ( Ndabigengesere et al., 1995 ), and promotes hydrolysis of metal ions that improve floc nucleation ( Font et al., 1999 ; Amagloh & Benang, 2009 ). Comparatively, similar pH changes (ΔpH = 0.3 - 0.8) in other plant coagulants like Moringa oleifera have been shown to significantly improve turbidity removal ( Gandiwa et al., 2020 ), confirming the operational relevance of even numerically small pH changes.

Figure 5 shows the turbidity level reducing efficiency (TRE) of the CFSE at various dosage rates. The initial turbidity of the wastewater as received was 3500 (as indicated with the red bar in Figure 5 ). The results show that the addition of the CFSE significantly reduced the turbidity at various dosages. Particularly, the 5 g/L dosage rate reduced the turbidity from 3500 NTU to 237 NTU (93.2% TRE) whilst the 10 g/L dosage rate reduced the turbidity levels from 3500 NTU to 237 NTU (96.4% TRE). Additionally, a dosage rate of 15 g/L led to a reduction of turbidity level from 3500 NTU to 89 NTU (97.4% TRE). The turbidity-reducing efficiency of the CFSE observed can be linked to the coagulating proteins and/or polysaccharides present in the seed as reported by Oladoja et al. (2017) and Ribeiro et al. (2019) . These proteins and/or polysaccharides are noted to have charged ions. Hence when released in water, the charged ions cause neutralisation and settling of oppositely charged colloids, which are responsible for turbidity in water/wastewater ( Crapper et al., 1973 ). Moreover, the direct relations observed between the TRE and the dosage rates, thus where TRE increased with a higher dosage rate, could be linked to the amount of active charged ions present in coagulants added.

Figure 6 highlights the effect of using CFSE as a plant-based coagulant on treating the levels of TDS of typical SSGM wastewater. The initial TSD level of the wastewater as received was 2400 mg/L (as indicated with the red bar in Figure 6 ). Generally, the results showed that the TDS level of the wastewater reduced drastically when contacted with the CFSE, regardless of the dosage. Particularly, it was observed that at 5, 10, and 15 g/L dosages, the TDS level decreased from 2400 mg/L to 168 mg/L (representing 93% reduction of TDS level), 236 (representing 90.2% reduction) and 471 mg/L (representing 80.4%. reduction), respectively. Notably, it was also observed that the reduction efficiency decreased with increasing dosage rate.

The TDS reduction observed can be attributed primarily to multiple mechanisms facilitated by bioactive compounds (proteins and polysaccharides) present in plant-based coagulants such as CFSE. Three of these probable mechanisms are: (i) Charge neutralisation (Cationic proteins adsorb onto negatively charged colloidal particles, reducing electrostatic repulsion and enabling aggregation) ( Ndabigengesere et al., 1995 ; Benalia et al., 2024 ), (ii) adsorption complexation (functional groups (e.g.., Polysaccharide hydroxyl/carboxyl) form complexes with dissolved metal ions through ligand exchange) ( Alazaiza et al., 2022 ; Wagh et al., 2022 ) and (iii) enmeshment (precipitated metal hydroxides incorporate dissolved ions during sweep flocculation) ( Saxena & Brighu, 2023 ). Notably, the significant removal of heavy metals ( Table 3 ) supports the role of adsorption and enmeshment, while turbidity reduction ( Figure 5 ) indicates charge neutralisation. The relative contribution of these mechanisms varies with contaminant type.

Moreover, the observed inverse relationship between TDS and CFSE dosage, may be attributed to over-coagulation. According to Benalia et al. (2024) and Saxena & Brighu (2023) , at higher coagulant doses, re-stabilisation of colloidal particles occurs due to charge reversal (where particles become positively charged) or steric hindrance from excess polymeric chains. Additionally, high ionic strength from the excess plant extract can compress electrical double layers, reducing electrostatic interactions necessary for aggregation ( Igwegbe & Onukwuli, 2019 ). This phenomenon has been similarly reported for other natural coagulants like Moringa oleifera ( Gandiwa et al., 2020 ).

Figure 7 highlights the impact of using CFSE as a plant-based coagulant in treating TSS of SSGM wastewater. Notably, the initial TSS level of the wastewater was 6780 mg/L (as indicated with the red bar in Figure 7 ). In general, the results indicated that the TSS level of the wastewater reduced drastically when contacted with the CFSE, regardless of the dosage rate. Particularly, the level of TSS in the wastewater was reduced from 6780 mg/L to 276 mg/L (95% reduction efficiency) at a 5 g/L CFSE dosage rate. The dosage rates of 10 and 15 g/L CFSE yielded reduction efficiencies of 97.5% (reduced the level from 6780 mg/L to 199 mg/L) and

99.2% (reduced the level from 6780 mg/L to 172 mg/L), respectively. Notably, this behaviour is similar to that observed for the TDS removal, hence it is expected that the same coagulation mechanism underpins this phenomenon as discussed in section 3.2.3.

Figure 8 shows the effect of CFSE on the EC of wastewater. The results show that the EC of the wastewater initially at 4790 µS/cm was reduced to 335 µS/cm (93% decrease) at a CFSE dosage rate of 5g/L. However, the results further showed that the reduction efficiencies decreased with increasing CFSE dosage rate. Particularly, CFSE dosage rates of 10 and 15 g/L yielded reduction efficiencies of 90.15% (i.e., a decrease from the initial EC of 4790 µS/cm to 472 µS/cm) and 80.33% (i.e., a decrease from the initial EC of 4790 µS/cm to 942 µS/cm), respectively. Notably, this trend is like the behaviour observed in the work published by Gandiwa et al. (2020) , where plant-based (moringa oleifera and cactus opuntia) was used for the treatment of water. The observed behaviour may be due to the dissociation of the coagulant-producing ions that raise the EC of the wastewater ( Marobhe et al., 2007 ; Gandiwa et al., 2020 ). Additionally, the observed inverse relationship between EC reduction and dosage mirrors observations for Sesamum indicum coagulants ( Igwegbe & Onukwuli, 2019 ), attributed to ion leaching from plant extracts at high concentrations.

Heavy metals are often difficult to remove from wastewater/water due to their low concentration and susceptibility to biodegradation ( Priya et al., 2022 ; Renu et al., 2017 ), hence heavy metal removal efficiency of the CFSE was also explored. Table 3 shows the heavy metal removal efficiency of the CFSE at 5, 10, and 15 g/L dosage. Generally, it was observed that the CFSE drastically reduced the heavy metals concentrations from 0.02 ppm to 0.005 ppm for As, 0.004 ppm to 0.001 ppm for Cd, 0.004 ppm to 0.001 for Cr, 1.749 to 0.076 ppm for Cu, 22.24 to 10.4 ppm for Fe, and 0.026 to 0.020 ppm for Hg. The corresponding reducing efficiencies for the heavy metals from 0 to 15 g/l dosages were 45 - 75% for As, 25%- 75% for Cd, 6.98% - 65.12% for Cr, 59.35% - 95.65% for Cu, 19.06% - 53.24% for Fe, and 11.54% - 23.08% for Hg.

The mechanisms reported to underpin the efficiency of removing heavy metals by plant-based coagulants are primarily driven by mechanisms such as adsorption and polymer bridging ( Alazaiza et al., 2022 ). However, the adsorption mechanism predominates the heavy metal removal efficiency of plant-based coagulants such as CFSE due to the presence of functional groups ( Aziz et al., 2021 ). FTIR analyses of Cassia fistula seeds revealed key functional groups (Carboxyl (-COOH) and hydroxyl (-OH) groups, Amine groups (-NH₂), Sulfhydryl groups (-SH)) involved in metal adsorption ( Ngan et al., 2017 ; Wagh et al., 2022 ). Whilst the Carboxyl and hydroxyl groups of the CFSE participate in ion exchange and complexation with metal cations (such as Cu²⁺, Cd²⁺) through electrostatic attraction and sharing of electron pairs ( Alazaiza et al., 2022 ), the Sulfhydryl groups exhibit high affinity for soft metals like Hg²⁺ ( Wagh et al., 2022 ). Contrarily, the Amine groups get protonated at wastewater pH (6.35 - 6.89) to form - ${\text{NH}}_3^{+}$, enabling anion exchange for oxyanions like arsenate ( ${\text{HAsO}}_4^{2-}$) ( Nimesha et al., 2022 ). This implies that polymer bridging plays a secondary role in this study, primarily enhancing floc aggregation post-adsorption ( Saxena & Brighu, 2023 ).

* WHO and Parsons, 2004 .