500 g of Gmelina arborea fruits (see Figure 1) were washed under tap and rinsed with distilled H₂O. The fruits were separated from its seeds; ground and 400 cm³ of distilled H₂O were added to the extract. The ground fruits were filtered using Whatman filter paper No. 1. The supernatant was concentrated to 300 cm³ and diluted with 100 cm³ conc. HCl and refluxed for 2 hours. The solution was allowed to cool until crystals began settling out. The solution was later put in the refrigerator to allow for more crystal formation. The crystals were filtered, recrystallized from hot methanol and air dried in the laboratory at room temperature. The crystals were stored in a dark sample bottle to avoid auto oxidation.

Buffer solutions of pH 1 to 8 were prepared using KCl, HCl, KHC₈H₄O₄, KH₂PO₄.

5% solution of the purified crystals were prepared by dissolving 5 g of the crystals in methanol containing 0.01% conc. HCl and made up to 100 cm³ in a standard flask. 1 cm³ of this solution was diluted to 10 cm³ and its λmax was determined spectrophotometrically in a 1 cm³ cuvette by scanning from 200 - 700 nm.

5 cm³ of 5% crystal solution and 5 cm³ of standard solutions of a metal under study were introduced into 8 beakers and their pH values varied in aqueous phase from 1 to 8 using freshly prepared buffer solutions. The absorbance of each analyte in solution was read at the adjusted pH. The λmax of the analyte metal was obtained from plot between pH and the absorbance of the analyte.

Fresh standard solutions of Cu²⁺, Zn²⁺, Ca²⁺, and Mg²⁺ were prepared at various concentrations ranging from 1.0 to 3.0 ppm. 5 cm³ of each standard solution were introduced into five sample bottles containing 5 cm³ of crystal solution and kept at pH5. After equilibration of the solutions, the absorbances of the crystal-metal complexes were recorded at maximum wavelength after scanning from 200 to 700 nm.

Five tilapia fish samples were dried at 105˚C until constant weight was maintained. The dried fish samples were pulverized and weighed into porcelain crucibles. 1 cm³ conc. HNO₃ was added to each crucible and ashed at 450˚C. The ashes were dissolved in 5 cm³ 1:1 HCl and samples transferred into 250 cm³ flask containing 1 cm³ 5% crystal solution at optimum pH. The crystal-sample solutions were made up to mark with distilled water and analysed using UV-Vis spectrophotometer and AAS (210CGP bulk scientific).

Paired T-test was used to determine significant variation between the developed method using the Gmelina arborea crystals and AAS. The null hypothesis (H₀) suggested that AAS and the developed method both produce comparable results for the heavy metals under study. This method was tested against the alternative hypothesis (H₁) which suggested otherwise therefore dismissing the suitability of the crystal as a metallochromic agent for heavy metals determination. The H₀ was accepted when p ≥ 0.05 and t values were within critical values i.e. −2.5706 ≤ t ≥ 2.5706. The paired T-test was done with SPSS 21.0.

The spectral data obtained from the analysis of the extract in comparison with literature values are shown in Table 1. The values of λmax of the extract matched with reported values that the crystals from Gmelina arborea extract were of cyanidin. The experimental values were not significantly variable when matched with those reported [22].

The effect of pH on the formation of cyanidin-metal complexes was observed from a calibration curve. At various pH values, these cyanidin-metal complexes showed reproducible results. However, all the metal complexes showed prominent absorbances at pH = 5. In consideration of selectivity and sensitivity of the method, pH value of 5 was chosen as the optimum pH for the simultaneous determination of these metals complexes with cyanidin in aqueous media.

The analyses results of metals from fish samples analysed with AAS and UV-Vis is shown in Table 2. The metal distribution was Ca > Mg > Cu > Zn for AAs and Cu > Mg > Zn > Ca when cyaniding method was used. These distribution patterns in the fish samples suggest different levels of the metals in the different environments the fishes habited. Table 3 shows the paired T-test of the significant differences in the methods and how they affect the H_o.

The mean concentration of Zn was (1.5160 ± 0.7071) ppm from AAS and (1.4814 ± 0.7167) ppm when analysed using cyanidin. Their mean difference was (0.0342 ± 0.0687). The difference was not significant (p > 0.05) and T < 2.5706. H₀ was accepted as summarised in Table 3. Therefore Zn can be analysed using both AAS and cyaniding method. This may be said for Cu since there mean differences supported the H₀ (p > 0.05, −2.5706 ≤ t ≥ 2.5706) as shown in Table 3. This can be attributed to the stable complexes they formed with cyanidin extracted from Gmelina arborea.

Metals with (*) are those analysed using AAS.

The difference in mean of concentration of Ca in fish sample using cyanidin method and AAS was quite large (6.5440 ± 0.9698) compared to differences in means of Zn and Cu. The difference was significant (p < 0.05) and T value > 2.5706. Therefore it did not support H₀. This analogy applies to Mg whose difference in means was appreciably significant, p < 0.05. This in turn produced T-test values greater than 2.5706 (7.428 for Ca and 3.104 for Mg). Clearly statistics of Ca and Mg did not support H₀. There large differences in mean may be suggesting a large shift in the UV region which explains absorption of light by unstable complexes of cyanidin with Ca and Mg. Therefore the result indicates that cyanidin is not a suitable metallochromic agent for the estimation of Ca and Mg in the fish samples.

From the results obtained and the statistical evaluation of the data, cyanidin method and AAS were both good alternatives to analysing Zn and Cu. Conversely, cyanidin method seems not a very good alternative to AAS in the quantitative determination of Ca and Mg. The method has shown good sensitivity, reliability and easy preparation of the reagent compared with other existing extractive spectrophotometric determination methods.

This work recommends the following: firstly the authors have added Gmelina arborea as a source of cyanidin to published list of plants that contain cyanidin. There is plethora of plant sources yet to be identified and analysed for cyanidin. Such works should be done on plants that are yet to be explored to determine their analytical importance. This work has demonstrated the usefulness this method has on aquatic samples like fish. It is recommended that this method be used for metal analyses on other environmental samples such as soil, water, dust and plant tissues. As a work in progress, the authors are isolating other ligands contained in Gmelina arborea to determine their suitability in metal analyses.