Brown algae were collected from the East Coast of Pangandaran, Pangandaran Regency, West Java, Indonesia, at coordinates 7°42'22.5"S, 108°39'50.3"E. The collection site and morphology of the material used in this study are shown in Fig. 1. The material was authenticated as Sargassum polycystum (Sargassaceae) by the Herbarium Jatinangor, Laboratory of Plant Taxonomy, Department of Biology, Universitas Padjadjaran, Indonesia, under certificate No. 113/HB/03/2025. A common raw-material batch was used for both extraction experiments. The material was cleaned, dried at 40 °C for 24 h, ground, and sieved. The reflux experiment used 60-mesh powder for all runs, whereas the maceration experiment evaluated 40-, 60-, and 80-mesh powder.

Gallic acid, quercetin, 2,2-diphenyl-1-picrylhydrazyl (DPPH), ascorbic acid, and Folin–Ciocalteu reagent were purchased from Sigma-Aldrich (St. Louis, MO, USA). Ethanol, methanol, sodium carbonate, aluminum chloride, sodium acetate, and ferric chloride were obtained from Merck (Darmstadt, Germany). Mayer’s, Wagner’s, and Liebermann–Burchard reagents were freshly prepared using analytical-grade reagents. Spectrophotometric measurements were performed using a UV–Vis spectrophotometer (Metas, China).

Two independent three-factor, three-level BBD experiments were conducted. The reflux experiment evaluated extraction temperature (50, 60, and 70 °C), extraction time (1, 2, and 3 h), and solid-to-solvent ratio (1:5, 1:10, and 1:15). Particle size was held constant at 60 mesh. The maceration experiment evaluated extraction time (24, 48, and 72 h), particle size (40, 60, and 80 mesh), and solid-to-solvent ratio (1:5, 1:10, and 1:15) at 25 ± 2 °C. Each design comprised 15 independent extraction runs, including three independently prepared center points to estimate pure experimental error. The factor levels are summarized in Table 1.

For reflux extraction, 50 g of 60-mesh dried powder was combined with 96% ethanol at the BBD-specified solid-to-solvent ratio and heated under a condenser at the assigned temperature and time. For maceration, 50 g of dried powder of the assigned particle size was immersed in 96% ethanol at the specified solid-to-solvent ratio in closed vessels at 25 ± 2 °C, protected from light, for the assigned duration. After extraction, the mixtures were filtered using Whatman No. 42 filter paper. Filtrates were concentrated using a rotary evaporator (HEM-01, China) and subsequently evaporated on a water bath until a viscous, concentrated extract with no visibly free-flowing solvent was obtained. Residual solvent content was not quantified in the original experiment. Extraction yield was calculated as the mass of concentrated extract divided by the initial mass of dried material and expressed as a percentage.

The common raw-material batch was evaluated for dry-material recovery, loss on drying, and moisture content. Preliminary qualitative phytochemical screening was performed on one reflux extract and one maceration extract selected as representative method-specific extracts for group-level characterization; these screening results were not used to identify individual compounds or to infer the composition of all 15 runs.

For alkaloids, 1 mL of extract solution was treated with ammonia and acidified with 2 N sulfuric acid; aliquots of the acid phase were then tested separately with Mayer’s, Wagner’s, and Dragendorff’s reagents. A white precipitate with Mayer’s reagent, a brown/orange precipitate with Wagner’s reagent, or an orange-to-reddish precipitate with Dragendorff’s reagent was interpreted as a positive reaction. For flavonoids, 3 mL of sample was treated with approximately 0.1 g of magnesium powder and concentrated hydrochloric acid; development of a red to reddish-black color was considered positive. Phenolics and tannins were screened by adding 1% ferric chloride to the sample and observing green, blue-black, or dark coloration. Saponins were assessed by vigorous shaking; formation of a persistent foam layer that remained after acidification with 2 N hydrochloric acid was interpreted as positive. Steroidal/triterpenoid constituents were screened using the Liebermann–Burchard reaction, in which characteristic green or red-purple color development was interpreted as a positive group-level reaction. Because these color and precipitation reactions are nonspecific, the results were interpreted only as preliminary indications of metabolite groups.

Total phenolic content (TPC) was determined using the Folin–Ciocalteu method with gallic acid as the standard, with modifications from Susanti et al. (2025). Gallic acid standards were prepared at 20, 40, 60, 80, and 100 µg/mL. A 0.5 mL aliquot of standard or sample was mixed with 2.5 mL of Folin–Ciocalteu reagent and 2 mL of 7.5% sodium carbonate. After incubation for 45 min at room temperature in the dark, absorbance was measured at 763 nm. TPC was expressed as mg gallic acid equivalents per gram of dry extract (mg GAE/g). Each extract was measured in triplicate; the reported variation represents analytical repeatability.

Total flavonoid content (TFC) was determined using the aluminum chloride colorimetric method with quercetin as the standard (Husna et al. 2025; Susanti et al. 2025). Quercetin standards were prepared at 20, 40, 60, 80, and 100 µg/mL. A 1 mL aliquot of standard or sample was mixed with 1 mL of 10% aluminum chloride and 8 mL of sodium acetate solution. The mixture was incubated for 30 min at room temperature in the dark, and absorbance was measured at 433 nm. TFC was expressed as mg quercetin equivalents per gram of dry extract (mg QE/g). Measurements were performed in triplicate.

DPPH radical-scavenging activity was determined with modifications from Herdini et al. (2026) and Riwanti et al. (2021). DPPH was prepared in methanol and diluted to a working concentration of 50 µg/mL. Extract stock solutions were prepared in methanol and diluted to a concentration series selected for each extract to cover the response range used for IC₅₀ estimation. The final extract concentrations in the 5 mL assay mixture were 20, 40, 60, 80, and 100 µg/mL. When the 50% inhibition region was not bracketed, the concentration series was extended to 110, 120, 130, 140, and 150 µg/mL. For each measurement, the appropriate aliquot of extract solution was transferred to a 5 mL volumetric flask, 3 mL of 50 µg/mL DPPH solution was added, and methanol was added only to bring the total assay volume to 5 mL. Thus, methanol served as the assay solvent and volume adjuster rather than as an additional test concentration. The mixtures were incubated at 37 °C for 20 min in the dark, and absorbance was measured at 518 nm. Percentage inhibition was calculated relative to the DPPH control, and IC₅₀ was estimated from the concentration-inhibition relationship for each extract.

Ascorbic acid (2, 4, 6, 8, and 10 µg/mL) was included as a positive-control response. The measured inhibition at the lowest tested concentration was already above 50%, so the experimental range did not bracket the 50% response. For this reason, an IC₅₀ obtained by extrapolating the fitted line outside the tested concentration range was not considered methodologically valid and was not reported. The positive-control data are therefore presented only as evidence that the assay responded to a reference antioxidant. A lower ascorbic acid concentration series would be required in a future experiment to determine an interpolated positive-control IC₅₀.

The reflux and maceration experiments were conducted and analyzed as two independent datasets. The reflux dataset was analyzed in Minitab Statistical Software version 19 and the maceration dataset in version 21 because these were the software installations available when the respective datasets were processed. The same BBD/RSM modeling framework, term assessment, adequacy criteria, and desirability-based optimization logic were applied to both datasets; therefore, the version difference reflects software availability rather than a difference in statistical design.

For each response, linear, two-factor interaction, and quadratic terms were assessed using the general second-order model:

Y=b0+b1x1+b2x2+b3x3+b4x1x2+b5x1x3+b6x2x3+b7x12+b8x22+b9x32,

where Y is the predicted response, b0 is the intercept, bi are linear coefficients, bii are quadratic coefficients, and bij are interaction coefficients. Model adequacy was evaluated using analysis of variance, model p-values, lack of fit, R², adjusted R², and predicted R² (Myers et al. 2016; Montgomery 2017; Czyrski and Jarzębski 2020). Multi-response optimization used desirability functions to maximize yield, TPC, and TFC and minimize IC₅₀. The predicted optimum for each method was validated using three independently prepared extraction experiments, and validation results were compared with the reported 95% confidence and prediction intervals.

A descriptive matched center-point assessment used the three center-point runs from each design. The center points for both designs used 60-mesh powder, 50 g of material, 96% ethanol, and a 1:10 solid-to-solvent ratio, while temperature and duration remained method-specific. The assessment was therefore interpreted as a comparison of process configurations rather than an isolated causal effect of the extraction method.

Phenolic-equivalent recovery (mg GAE/100 g dry algae) was calculated as:

Phenolic recovery = extraction yield (%)×TPC(mg GAE g extract ).

Flavonoid-equivalent recovery (mg QE/100 g dry algae) was calculated as:

Flavonoid recovery = extraction yield (%)×TFC(mgQEg extract ).

Because extraction yield (%) represents grams of extract recovered per 100 g of dry starting material, multiplication by TPC or TFC yields the corresponding equivalent recovery per 100 g of dry algae.

The collected material was authenticated as Sargassum polycystum, family Sargassaceae. Processing 40 kg of fresh algae produced 2.6 kg of dried powdered material, corresponding to a dry-material recovery of 6.50%. The powder showed a loss on drying of 7.97 ± 0.24% and a moisture content of 11.83 ± 0.08% (Table 2).

The method-specific representative reflux and maceration extracts used for qualitative screening showed positive reactions for phenolics, flavonoids, tannins, saponins, and Liebermann–Burchard-reactive steroidal/triterpenoid constituents. Alkaloid reactions were reagent-dependent: Mayer’s reagent was negative, whereas Wagner’s and Dragendorff’s reactions were positive (Table 3). These results are preliminary group-level observations and do not establish the presence of individual compounds.

Reflux extraction produced yields of 5.46–15.38%. The highest observed yield was 15.38% at 60 °C, 2 h, and 1:10, whereas the lowest yield was 5.46% at 50 °C, 2 h, and 1:5. Reflux TPC ranged from 12.61 to 53.05 mg GAE/g, TFC from 15.29 to 42.41 mg QE/g, and DPPH IC₅₀ from 98.66 to 155.48 µg/mL. The highest TPC and TFC values and the lowest IC₅₀ were observed at 50 °C, 1 h, and 1:10.

Maceration produced yields of 2.01–15.30%. The highest observed yield was 15.30% at 72 h, 60 mesh, and 1:15, whereas the lowest yield was 2.01% at 24 h, 40 mesh, and 1:10. Maceration TPC ranged from 12.62 to 62.98 mg GAE/g, TFC from 14.60 to 40.77 mg QE/g, and DPPH IC₅₀ from 36.19 to 171.91 µg/mL. The highest TPC and TFC values and the lowest IC₅₀ were observed at 24 h, 80 mesh, and 1:10. In both designs, the condition producing the highest yield differed from the condition producing the strongest antioxidant-related quality responses (Table 4).

All reflux response models were statistically significant, with model p-values of 0.001 for yield, 0.005 for TPC, 0.005 for TFC, and 0.006 for IC₅₀. Extraction temperature had the largest reported F-values among the reflux factors. Reflux R² values ranged from 85.55% to 97.96%, and lack of fit was non-significant for all responses. Predicted R² was substantially lower than adjusted R² for TPC, TFC, and IC₅₀, indicating limited out-of-sample precision for these bioactive responses.

All maceration response models were also significant, with p < 0.001 for yield and TPC, p = 0.002 for TFC, and p = 0.003 for IC₅₀. Extraction time had the largest reported F-value for every maceration response. Maceration R² values ranged from 81.78% to 99.16%, and lack of fit was non-significant. Predicted R² values were lower for TPC, TFC, and IC₅₀ than for yield, although they were higher than the corresponding reflux predicted R² values (Table 5).

The center points of both designs used 60-mesh powder and a 1:10 solid-to-solvent ratio. Reflux center-point runs were performed at 60 °C for 2 h, whereas maceration center-point runs were performed at 25 ± 2 °C for 48 h. Reflux produced a higher mean yield, while maceration produced a higher mean TPC and a lower mean DPPH IC₅₀. Mean TFC values were similar. Because time and thermal conditions differed, these values compare process configurations rather than isolate the effect of the extraction method (Table 6).

The reflux multi-response optimum was predicted at 50.404 °C, 1.0404 h, and a solid-to-solvent ratio of approximately 1:9.55, with a composite desirability of 0.663. Predicted responses were 9.78% yield, 41.61 mg GAE/g TPC, 38.60 mg QE/g TFC, and 113.16 µg/mL IC₅₀. Validation produced 10.32%, 39.68 mg GAE/g, 33.24 mg QE/g, and 109.25 µg/mL, respectively. All validation values fell within the corresponding 95% confidence and prediction intervals (Table 7).

The maceration optimum was predicted at 24 h, 80 mesh, and a solid-to-solvent ratio of approximately 1:5.30, with a composite desirability of 0.704. Predicted responses were 5.35% yield, 56.08 mg GAE/g TPC, 37.39 mg QE/g TFC, and 55.20 µg/mL IC₅₀. Validation produced 5.54%, 59.39 mg GAE/g, 38.03 mg QE/g, and 55.66 µg/mL, respectively. These values were within the reported intervals, although the IC₅₀ prediction interval was wide. Three-dimensional response surfaces illustrate the method-specific effects of the main extraction variables on TPC and DPPH IC₅₀ under the optimized solvent-ratio conditions (Fig. 2).

Normalization to the initial dry algal mass produced a different interpretation from concentration per gram of extract. Based on the validated optimum values, reflux recovered an estimated 409.50 mg GAE and 343.04 mg QE per 100 g of dry algae, whereas maceration recovered 329.02 mg GAE and 210.69 mg QE per 100 g of dry algae. Thus, the optimized maceration extract was more concentrated in phenolic equivalents and showed a lower IC₅₀, while optimized reflux produced greater total phenolic- and flavonoid-equivalent recovery from the starting biomass (Table 8).

The two BBD experiments showed distinct response patterns because reflux and maceration operate under different process conditions. Reflux combines controlled heating with solvent condensation and can increase mass transfer within a shorter processing time. However, higher temperatures or prolonged exposure can reduce phenolic and flavonoid stability (Zhang et al. 2018; Antony and Farid 2022). The highest reflux TPC and TFC values and the lowest reflux IC₅₀ occurred at the lowest investigated temperature and shortest extraction time, indicating that mild heating was more favorable for antioxidant-related quality than more intensive reflux conditions. Maceration showed a different time-dependent profile: the highest yield occurred after 72 h, whereas the highest TPC and TFC values and the lowest IC₅₀ occurred after 24 h with 80-mesh powder. Smaller particles increase contact area and shorten diffusion pathways (Prasedya et al. 2021). Extending extraction beyond the quality optimum may increase total soluble mass without proportionally increasing phenolic or flavonoid constituents; prolonged contact can also be accompanied by equilibrium, oxidation, or degradation of released compounds (Cheng et al. 2025).

A central finding was that maximum extraction yield did not coincide with maximum antioxidant-related quality. Yield measures total soluble material, whereas TPC and TFC represent equivalent concentrations per gram of extract. A high-yield extract may contain polysaccharides, salts, pigments, and other non-target material, while a lower-yield extract may contain a higher concentration of phenolic- and flavonoid-reactive constituents (El Mannoubi 2023). This distinction is particularly relevant to brown algae because their cell walls contain complex polysaccharide and protein matrices that can release bulk material under conditions that do not proportionally increase antioxidant-related constituents (Beuder and Braybrook 2023; Badruddin et al. 2024). The yield-normalized analysis adds an important qualification: although optimized maceration had higher TPC per gram and lower DPPH IC₅₀, optimized reflux recovered more phenolic- and flavonoid-equivalent material per 100 g of starting algae. Process selection therefore depends on whether the application prioritizes extract concentration or total recovery from biomass.

The descriptive center-point assessment compared conditions sharing 60-mesh powder, 50 g of material, 96% ethanol, and a 1:10 solid-to-solvent ratio. Reflux yielded more extract, whereas maceration produced higher TPC and lower IC₅₀. Nevertheless, the configurations differed in temperature and duration; they cannot be interpreted as a fully controlled head-to-head test isolating the extraction method. The analysis supports only the narrower conclusion that the two practical configurations produced different yield-quality profiles.

All response models were significant and had a non-significant lack of fit, indicating adequate representation of the observed design-space data (Myers et al. 2016; Czyrski and Jarzębski 2020). However, high R² does not guarantee strong prediction. Predicted R² values for reflux TPC, TFC, and IC₅₀ were substantially lower than the adjusted R² values. Maceration models showed higher predicted R² values, but prediction was still less stable for bioactive responses than for yield. Validation values were close to predictions and fell within the reported confidence and prediction intervals; however, the wide prediction interval for maceration IC₅₀ indicates substantial uncertainty. The models should therefore be used within the investigated factor ranges and not extrapolated to untested conditions.

Qualitative phytochemical tests indicated reagent-reactive phenolic, flavonoid, tannin, saponin, and steroidal/triterpenoid groups in the method-specific representative extracts. These reactions do not identify individual compounds, and Folin–Ciocalteu and aluminum chloride assays are group-level measurements rather than selective quantification of phlorotannins or specific flavonoids. The antioxidant assessment was limited to DPPH, which measures chemical radical-scavenging under defined solvent conditions and does not establish cellular antioxidant activity. Additional assays such as ABTS, FRAP, ORAC, or cellular oxidative-stress models would broaden interpretation (Apak et al. 2016). Chromatographic profiling by high-performance liquid chromatography with diode-array detection (HPLC-DAD) or liquid chromatography-mass spectrometry (LC-MS) would also be required to relate process conditions to individual phlorotannin, flavonoid, or pigment constituents (Lee et al. 2024). Residual solvent content of the concentrated extracts was not quantified, and this limitation should be considered in future process standardization.

Reflux and maceration should be considered complementary process options. Reflux is relevant when shorter processing time and greater total recovery are prioritized, provided that temperature and duration are controlled. Maceration is relevant when an extract more concentrated in phenolics and a stronger DPPH response are prioritized and a longer processing time is acceptable. Future development should evaluate assisted extraction technologies and green-process metrics, including solvent consumption and energy demand (Chemat et al. 2019; Lee et al. 2024).